Skip to main content

Full text of "Transactions"

See other formats 


ocUtu 


^^£^?mca/ .Szwm^ 


MM 


Digitized  by  the  Internet  Archive 

in  2010  with  funding  from 

University  of  Toronto 


http://www.archive.org/details/transactions43elec 


TRANSACTIONS 


OF   THE 


Amfrtran 
lEUttrorhi^mtral  Sorirtti 


VOLUME   XLIll 


FORTY-THIRD   GENERAL   MEETING 

NEW    YORK    CITY 
MAY   3,  4    AND    5,   1923 


PUBLISHED  BY 

Ollir  AmrrUan  €lfrtrori|fmiral  ^orWy 

AT  THE  OFFICE  OF  THE  SECRETARY 

COLUMBIA  UNIVERSITY,  NEW  YORK  ClTY 

1923 


TRANSACTIONS 


OF   THE 


Ammran 


VOLUME   XLIII 


FORTY-THIRD   GENERAL   MEETING 

NEW    YORK    CITY 
MAY    3,  4    AND    5,   1923 


T 


t..  ^' 


PUBLISHED   BY 

®l|p  Amrrtran  ^Itttrotiftmxcui  ^orirty 

at  the  office  of  the  secretary 
Columbia  University,  New  York  City 

1923 


TV 

Copyright  1923  bj-  the  Amefiean  Electrochemical  Society,  w,  iL,0 


Permission  to  reprint  parts  of  the  Transactions  is  herehj-  granted 
to  current  periodicals,  provided  due  credit  is  given. 


The   Society   is  not   responsible   for   the   statements   and   opinions 
advanced  in  papers  or  in  discussion  thereon. 


Prices  of  Volumes  ;I  to  XLIII  (excepting  Vols.  I,  II,  III,  VII, 
XXIX  and  XXXI),  to  non-members,  $6.50  per  copy;  to  members, 
$2.50,  excepting  Vols.  XLI  and  XLII,  which  are  $4.00  each;  to 
public  libraries,  colleges,  scientific  societies  and  journals,  $4.00. 
Volumes  I,  II,  III,  VII,  XXIX  and  XXXI  are  double  above  prices. 
Prices  are  for  volumes  bound  in  cloth,  and  include  delivery  within 
the  postal  union. 


When  the  stock  of  anj-  volume  has  been,  reduced  to  25  copies 
(as  Vol.  Ill),  the  sale  is  limited  to  those  purchasing  the  given 
volume  as  part  of  a  complete  set. 


Complete  sets  will  be  sold  to  anybody  at  25  per  cent  discount 
on  above  prices ;  members  may  obtain  the  volumes  necessary  to 
complete  their  sets  at  25  per  cent  reduction  on  above  prices. 


WARE  BROS.   COMPANY.   PRINTERS 
1010   ARCH    ST..    PHILA. 


OFFICERS  OF  THE  SOCIETY 


PRESIDENT 

A.  T.  HINCKLEY 

Term  expires  1924 
PAST-PRESIDENT 

C.  G.  SCHLUEDERBERG 

Term  expires  1924 
VICE-PRESIDENTS 

H.  C.  PARMELEE  LAWRENCE   ADDICKS 

A.  H    HOOKER  G.  K.  ELLIOTT 

W.  S.  LANDIS  HENRY  HOWARD 

Terms    expire     1924  Terms    expire    1925 

MANAGERS 

CARL  HERING  E.  F.  CONE  F.  M.  BECKET 

J.  V.  N.  DORR  W.  M.  CORSE  C.  B.  GIBSON 

F.  A.  J.  FITZ  GERALD  WM.  BLUM  R.   A.   WITHERSPOON 

lerms    expire    1924  Terms    expire     1925  Terms    expire    1926 

TREASURER 

F.  A.  LIDBURY 

Term  expires  1924 
SECRETARY 

COLIN  G.  FINK,  Columbia  University,  New  York  City 

Term  expires   1924 


COMMITTEES 


PUBLICATION    COMMITTEE 

F.  A.  J.  FiTz  Gerald,  Chairman 
CouN  G.  Fink,  Ex  Officio  A.  T.  Hinckley,  Ex  Officio 

H.  M.  Goodwin  Wm.  Blum 

H.  C.  Parmelee  E.  F.  Cone 

Terms    expire    1924  Terms   expire    1925 


COMMITTEES — ContinuMl 

COMMIITEE   ON   I,OCAL    SECTIONS 

F.  J.  Tone,  Chairman  Colin  G.  Fink  C.  G.  Schluederberc 


PATENT   COM>UTTEE 

L.  H.  BaEkeland,  Chairman       F.  G.  CoTTRELL 


E.  J.  Prindle 


PUBLIC  KELATION8  COM3IITTEE 


A.  T.  Hinckley,  Chairman 


W.  D.  Bancroft 
Carl  Hering 
C.  G.  Burgess 
E.  G.  AcHESoN 
L.  H.  Baekeland 

W.   H.   W.\LKER 

W.  R.  Whitney 


F.   A.    LiDBURY 

Lawrence    Addicks 
F.  A.  J.  FitzGerald 
Colin  G.  Fink 
F.  J.  Tone 
W.  S.  Landis 
AcHESON  Smith 


C.  G.  Schluederberg 


MEMBERSHIP    COMMITTEE 

E.  L.  Crosby,  Chairman 

M.  deK.   THOMPSON,  Cambridge,  Mass.   J.    W.   BECKMAN,   Oakland,    Calif. 


L.  E.  SAUNDERS,  Worcester,  Mass. 
G.  B.  HOG.\BOOM,  Waterbury,  Conn. 
H.  C.  PARMELEE,  New  York,  N.  Y. 

C.  F.   ROTH,  New  York,  N.  Y. 
E.   F.   KERN.  New  York,  N.  Y. 

J.     A.   SEEDE,   Schenectady,  N.  Y. 
H.   W.   GILLETT.  Ithaca,   N.   Y. 
G.    H.   CLAMER,   Philadelphia,   Pa. 
R.  E.   ZIMMERMAN,   Pittsburgh,   Pa. 
J.   E.  ISENBERG,  State  College,  Pa. 

D.  A.  LYON,   Washington,  D.  C. 
P.  J.  KRUESI,   Chattanooga,  Tenn. 
THEODORE    SWANN,    Anniston,    Ala. 
L.    A.   DREFAHL,   Cleveland.  O. 

T.   F.   BAILY,   Alliance,   O. 
G.   K.   ELLIOTT,  Cincinnati,   O. 
W.   K.    BOOTH,   Chicago,   111. 
O.    P.    W.\TTS.    Madison,   Wis. 


C.   E.   WILLIAMS,  Seattle,  Wash. 

F.   T.   KAELIN,   Montreal,   Canada. 

J.  T.  BURT-GERRANS,  Toronto,  Canada. 

J.   H.   BUTTERS,  Hobart,   Tasmania 

HANS  LANDOLT,   Turgi,   Switzerland 

BIRGER  CARLSON,   Stockholm,   Sweden 

B.  BERG-HANSEN,  Christiania,  Norway. 
K.    B.    QUIN.XN,   W.    Somerset,   Cape 

Province,   S.  Africa. 

C.  A.  KELLER,   Paris,  France. 

H.   F.  ETCHELLS,   Sheffield,  England. 
W.    E.    HUGHES,    Seaford,   England. 
F.    GIOLITTI,   Torino.  Italy. 
F.    FOERSTER,    Dresden,    Germany. 
J.    BABOROVSKY,    Brno,    Crecho- 
Slovakia. 

VOGORO   KATO,  Tokio,  Japan. 

JEN  CHOW.  Shanghai,  China. 

A.  H.  ATEN,  Amsterdam,   Netherlands. 


WAYS  AND  MEANS  COMMITTEE 

H.  B.  CoHO.  Chairman 
Lawrence  Addicks  Carl  Hering 

Colin  G.  Fink  F.  A.  Lidbury 

H   C.  Parmelee 


DIVISIONS  OF  THE  SOCIETY 

BXECTROTHERMIC  DIVISION 

G.  K.  Elliott,  Chairman 
DoRSEY  A.  Lyox,  Vice-Chairman  L.  C.  Judson,  Secretary-Treasurer 

Membera-at-L.arg'e 

James  H.  Parker  F.  M.  Becket 

W.  J.  Priestley  Bradley    Stoughtox 

Terms    expire     1924  Terms    expire    1925 

ELECTBODEPOSITION    DIVISION 

S.  Skowronski,  Chairman 
Chas.  a.  Mann,  Vice-Chairman        Wm.  Blum,  Secretary-Treasurer 

Meinber8-at-LiarK« 

Lawrence  Aduicks  F.  R.  Pyne 

F.  C.  Mathers  M.   R.   Thompson 

Terms    expire     1924  Terms    expire    192S 

Foreign  Representatives 

W.  E.  Hughes,  Seaford,  England  Bertram  Wood,  Hobart.  Tasmania 


TECHNICAL  COMMITTEES 

I'RIMARV   BATTERIES 

C.  F.  Burgess,  Chairman  D.  L.  Ordway 

SECONDAKT   BATTERIES 

P.  G.  Salom,  Chairman  O.  W.  Brown 

ELECTRO   ANAI.Y8I8 

H.  S.  LuKENS,  Chairman  Geo.  S.  Forbes 

RADIO  ACTIVITY 

S.  C.  LiND,  Chairman  H.  S.  Miner 

CHLORINE   AND   CAISTIC 

E.  M.  Sergeant,  Chairman  Hugh  K.  Moore  L.  D.  Vorce 

CORROSION 

Colin  G.  Fink,  Chairman  W.  M.  Corse  W.  D.  Richardson 

FIXED   NITROGEN 

W.  S.  Laxdis,  Chairman  Arthur  B.  Lamb  Frank  S.  McGregor 

WATER  POWER 

J.  H.  Harper,  Chairman  J.  L.  YardlEY  H.  J.  Pierce 

ORGANIC    ELECTROCHEMISTY 

C.  J.  Thatcher,  Chairman  Alexander  Lowy 

ELECTROCHEMISTRY  OF  GASEOUS  CON'DrCTION 

Duncan  MacRae,  Chairman  Wm.  R.  Morr 

INSUT^ATING  OILS  AND  VAR^^SHES 

H.  C.  p.  Weber,  Chairman 
C.  D.  HocKEs  C.  J.  Rodman  Christian  Dantsizex 

RADIO   B.4TTERJES 

C.  F.  Burgess,  Chairman  Geo.  W.  Vixal  C.  A.  Gillixgham 


LOCAL  SECTIONS  OF  THE  SOCIETY 

Philadelphia  Section 
Carl  Herixg,  Chairman,  Philadelphia,  Pa. 
S.   S.  Sadtler,  Secretary,   Philadelphia,  Pa. 

New  York  Section 

P.  D.  Merica,  Chairman 

P.  D.  V.  ManxXing,  Secretary-Treasurer,  50  E.  41st  St. 

Pittsburgh  Section 

C.  B.  GiBSOX,  Chairman,  E.  Pittsburgh,  Pa. 

S.  L.  GooDALE,  Secretary-Treasurer,  Pittsburgh,  Pa. 

Niagara  Falls  Section 

E.  AI.  Sergeant,  Chairman,  Niagara  Falls,  N.  Y. 

L.  C.  JuDSox,  Secretary-Treasurer,  Niagara  Falls,  N.  Y. 


TABLE  OF  CONTENTS. 


PAGE 

Portrait  of  President  A.  T.  Hinckley Frontispiece 

Proceedings  of  the  Forty-third  General  Meeting 1 

Portrait  of  Dr.  Edward  G.  Acheson.  (Honorary  Member) 5 

Dr.  Edward  G.  Acheson  and  His  Work— F.  A.  J.  FitzGerald 5 

Members  and  Guests  Registered  at  the  Forty-third  General  Meeting 18 

PAPERS. 

Presidential   Address — Opportunities   for   the   American   Electrochemist 

Abroad — C.  G.  Schluederberg   21 

PAPERS  ON  "ELECTRODE  POTENTIALS."  . 

Newer  Aspects  of  Ionization  Problems — Hugh  S.  Taylor 31 

Oxygen    Overvoltage   of   Artificial    Magnetite   in   Chlorate    Solutions — 

H.  C.  Howard   51 

The  Effect  of  Current  Density  on  Overvoltage — M.  Knobel,  P.  Caplan 

and  M.  Eiseman  55 

Electrotitration  wMth  the  Aid  of  the  Air  Electrode — N.  Howell  Furman.  79 

The  Hydrogen  Electrode  in  Alkaline  Solutions — A.  H.  W.  Aten 89 

Electrolytic  and   Chemical   Chlorination  of  Benzene — Alexander  Lowy 

and  Henry  S.  Frank   107 

The  Reactions  of  the  Lead  Storage  Battery — M.  Knobel 99 

Notes  on  the  Electrodeposition  of  Iron — Harris  D.  Hineline 119 

The  Influence  of  the  Base  Metal  on  the  Structure  of  Electrodeposits — 

W.  Blum  and  H.  S.  Rawdon  See  Vol.  44 

Current  Distribution  and  Throwing  Power  in  Electrodeposition — H.  E. 

Haring  and  W.  Blum  See.  Vol.  44 

The  Electrodeposition  of  Nickel  on  Zinc— A.  Kenneth  Graham. See  Vol.  44 
The    Effect    of    Iron    on    the    Electrodeposition    of    Nickel  —  M.    R. 

Thompson    See  Vol.  44 

Heat    Insulating    Materials    for    Electrically    Heated    Apparatus — J.    C. 

Woodson     127 

Methods  of  Handling  Materials  in  the  Electric  Furnace  and  the  Best 

Type  of  Furnace  to  Use — Frank  W.  Brooke 149 

vii 


PAGE 

The    Conversion   of    Diamonds   to    Graphite   at   High    Temperatures — 

M.  deKay  Thompson  and  Per  K.  Frolich 161 

The    Relation   between    Current,    Voltage   and    the   Length    of    Carbon 

Arcs— A.  E.  R.  W'estman   171 

Electric  Furnace  Detinning  and  the  Production  of  Synthetic  Gray  Iron 

from  Tin-Plate   Scrap — C.   E.  Williams,   C.   E.   Sims  and  C.  A. 

Newhall     191 


PAPERS  ON  "THE  PRODUCTION  AND  APPLICATION 
OF  THE  RARER  METALS.'" 

Present  Status  of  the  Production  of  Rarer  Metals — C.  James 203 

The  Preparation  of  Fused  Zirconium — Hugh  S.  Cooper 215 

Experiments  with  Uranium,  Boron,  Titanium,  Cerium  and  Molybdenum 

in  Steel— H.  W.  Gillett  and  E.  L.  Mack 231 

Some  Effects  of  Zirconium  in  Steel — F.  M.  Becket 261 

Inherent  Effect  of  Alloying  Elements  in  Steel — B.  D.  Saklatwalla 271 

Notes  on  the  Metallurgy-  of  Lead  Vanadates — Will  Baughman 281 

Preparation  of  Metallic  Uranium — R.  W.  Moore 317 

The    Reduction   of    Some    Rarer    Metal    Chlorides   by    Sodium — M.    A. 

Hunter  and  A.  Jones   See  Vol.  44 

Experiments  Relative  to  the  Determination  of  Uranium  by  Means  of 

Cupferron — Jas.  A.  HoUaday  and  Thos.  R.  Cunningham 329 

Cobalt— Its  Production  and  Uses— C.  W.  Drury 341 

Chromizing — F.   C.  Kelley    351 

The    Preparation    of    Platinum   and   of    Platinum-Rhodium    Alloy    for 

Thermocouples — Robert  P.  Neville   371 

Investigations  on  Platinum  Metals  at  the  Bureau  of  Standards — Edward 

Wichers  and  Louis  Jordan    385 

Some  Notes  on  the  Metals  of  the  Platinum  Group — Fred  E.  Carter. . .  .397 


Volume  XLIII  1923 


TRANSACTIONS 

OF  THE 

Autmran  l£UrtrorI|^ittiral   ^nmtg 


PROCEEDINGS 

CONDENSED   MINUTES  AND  RECORD  OF  THE   FORTY-THIRD  GENERAL 

MEETING  OF  THE  SOCIETY,  HELD  AT  THE  COMMODORE  HOTEL, 

NEW  YORK  CITY,  MAY  3,  4  AND  5,   1923. 

The  total  registration  at  this  meeting  was  254,  of  whom  168 
were  members  and  86  guests. 

PROCEEDINGS  OF  WEDNESDAY,  MAY  2,   1923 

The  registration  of  Society  members  and  guests  began  at  6.00 
P.  M.  on  the  mezzanine  floor  of  the  Commodore  Hotel.  At 
7.00  P.  M.  the  Board  of  Directors  met  at  dinner  for  the  purpose 
of  conducting  its  annual  business  meeting. 

PROCEEDINGS  OF  THURSDAY,  MAY  3,  1923 

The  meeting  convened  at  9.30  A.  M.  with  President  C.  G. 
Schluederberg  in  the  Chair.  Having  but  recently  returned  from 
a  trip  to  the  Far  East,  the  President  expressed  his  gratitude  at 
being  back  in  the  United  States  again,  and,  on  behalf  of  the 
Society,  heartily  welcomed  the  members  and  guests  present.  He 
then  called  upon  Dr.  Wm.  G.  Horsch,  who  had  arranged  for 
the  Symposium  on  "Electrode  Potentials,"  to  assume  the  Chair. 
Dr.  Horsch  briefly  mentioned  that  the  subject  of  electrode  poten- 
tials is  one  which  lies  particularly  within  the  field  of  the  electro- 
chemist,  and  that  the  papers  and  discussions  of  this  session  should 
give  rise  to  a  rather  comprehensive  cross-section  of  the  line  of 


2  PROCEEDINGS. 

progress  at  the  present  time.  Papers  by  the  following  authors 
were  presented  and  are  printed,  together  with  discussions,  in 
these  Transactions :  Hugh  S.  Taylor;  H.  C.  Howard;  M.  Knobel, 
P.  Caplain  and  M,  Eiseman ;  N.  Howell  Furman ;  A.  H,  W. 
Aten ;  M.  Knobel ;  Alexander  Lowy  and  H.  S.  Frank. 

At  11.30  the  meeting  adjourned.  Within  a  quarter  of  an  hour, 
members  and  guests  were  conveyed  by  special  buses  to  the  plant 
of  the  McGraw-Hill  Co.,  Inc.,  where  a  complimentary  luncheon 
was  served.  This  was  followed  by  an  inspection  trip  through 
the  various  departments  of  the  printing  and  publishing  plant. 
Moving  pictures  depicting  the  construction  and  operation  of  the 
Diesel  engine  concluded  the  visit,  the  members  returning  to  the 
Commodore  Hotel  to  attend  the  afternoon  session. 

At  2.30  P.  M.  President  Schluederberg  opened  the  annual 
business  meeting  of  the  Society.  Secretary  Fink  presented  the 
reports  of  the  Board  of  Directors  and  of  the  Secretary.  In  the 
latter  report  it  was  pointed  out  that  since  the  Baltimore  meeting 
the  Society  had  published  and  distributed  four  volumes  of  the 
Transactions  (Vol.  39,  40,  41  and  42),  thus  bringing  the  publi- 
cation work  up  to  date.  Following  the  presentation  of  this 
report,  Dr.  Hering  offered  the  following  resolution :  That  a 
vote  of  thanks  be  expressed  to  the  Secretary  and  his  office  for 
having  brought  the  volume  publications  up  to  date.  The  motion 
was  unanimously  carried.  The  above  reports,  together  with  that 
of  the  Treasurer  are  included  in  subsequent  pages  of  the  pro- 
ceedings. 

The  next  order  of  business  included  reports  of  Standing 
Committees.  Dr.  C.  F.  Burgess  presented  the  report  of  his 
committee  on  dry  cells,  which  indicated  that  in  1914,  17,092,438 
dry  cells  were  sold  in  this  country  at  a  valuation  of  $8,719,164, 
while  in  1919  the  number  had  increased  to  173.754,676  at  a 
valuation  of  over  $25,000,000.  To  promote  rapid  work  and  to 
avoid  unnecessary  du])lication  of  effort  this  committee  made  the 
following  recommendations : 

1.  That  a  committee  on  dry  cells  be  appointed  for  the  coming 
year,  with  instructions  to  oft'er  its  co-operation  to  the  Bureau  of 
Standards  in  standardization  of  tests  for  dry  cells  in  various  kinds 
of  radio  service. 


PROCEEDINGS.  3 

2.  That  members  of  the  American  Electrochemical  Society  who 
are  engaged  in  dry  cell  manufacture,  and  who  are  desirous  of 
contributing  to  and  taking  part  in  this  work,  indicate  their  desires 
to  the  Bureau  of  Standards. 

Dr.  S.  C.  Lind  presented  the  report  of  the  Radio  Activity  Com- 
mittee, in  which  it  was  especially  pointed  out  that  in  the  produc- 
tion of  radium  in  the  United  States  there  has  been  a  marked 
cessation  of  activity.  This  is  directly  due  to  the  discovery  and 
development  of  a  large  deposit  of  high  grade  radium-bearing 
uranium  ore  in  the  Belgian  Congo  by  the  Katanga  Copper  Co. 
It  was  the  recommendation  of  Dr.  Lind's  committee  that  the 
Society  undertake  to  concentrate  in  its  Transactions  the  results 
of  numerous  investigations,  in  the  field  of  radio  activity,  which 
are  published  in  such  a  widely  scattered  range  of  journals. 

The  report  of  the  Organic  Electrochemistry  Committee  was 
presented  by  its  Chairman.  Dr.  C.  J.  Thatcher.  The  progress 
being  made  in  this  field  will  be  discussed  extensively  at  a  sym- 
posium to  be  held  at  the  Spring  meeting,  1924. 

^Ir.  A.  T.  Hinckley  presented  the  report  of  his  committee  on 
membership.  The  essential  data  of  this  report  are  published  in 
following  pages. 

Chairman  FitzGerald  of  the  Publication  Committee  presented 
a  report  which  in  part  is  as  follows : 

To  the  Board  of  Directors.  American  Electrochemical  Society: 
During  the  year  1922-23  your  committee  has  received  and 
examined  98  papers.  Of  these  20  were  rejected,  27  were  returned 
to  the  authors  for  revision,  48  were  accepted  without  change  and 
3  were  withdrawn. 

Every  paper  is  sent  to  at  least  2  examiners.  Some  papers 
during  the  past  year  have  been  reported  on  by  5  examiners,  and 
several  papers  have  been  reported  on  by  3  examiners,  before 
final  action  by  the  committee.  This  change  in  the  routine  ex- 
amination of  papers  has  made  it  necessary  to  insist  on  an  earlier 
date  for  the  submission  of  papers  than  has  formerly  been  the 
practice,  and  the  ruling  of  the  Publication  Committee  as  to  the 
latest  date  for  receiving  papers  will  in  future  be  rigidly  enforced. 
A  new  rule  of  the  Publication  Committee  in  relation  to  papers 


4  PROCEEDINGS. 

submitted  for  symposia  is  that  one  of  the  two  examiners  to  whom 
a  symposium  paper  is  submitted  shall  be  the  member  in  charge  of 
the  symposium.  The  object  of  this  rule  is  to  avoid  embarrass- 
ments which  are  apt  to  arise  when  the  member  in  charge  of  the 
symposium  has  invited  papers  which  the  Publication  Committee 
is  disposed  to  reject. 

Prior  to  presenting  the  report  of  the  Tellers  of  Election,  the 
Chair  read  a  communication  from  H.  C.  Parmelee  in  which  he 
withdrew  his  name  from  the  report  in  favor  of  A.  T.  Hinckley. 
The  Chair  then  read 

THE  REPORT  OF  THE  TELLERS  OF  ELECTION 

The  following  is  a  list  of  votes  cast  in  the  election  of  officers 
for  the  year  1923-1924: 

President:   A.  T.  Hinckley,  246. 

Vice-Presidents:  Lawrence  Addicks,  212 ;  G.  K.  Elliott,  148 ; 
Henry  Howard,  117;  Dorsey  Lyon,  103;  W.  Lash  Miller,  71; 
W.  R.  Mott,  60 ;  A.  T.  Hinckley,  2. 

Managers:  F.  M.  Becket,  208;  C.  B.  Gibson,  185 ;  R.  A.  With- 
erspoon,  149;  G.  B.  Hogaboom,  99;  D.  B.  Rushmore,  51;  Law- 
rence Addicks,  1. 

Treasurer:    F.  A.  Lidbury,  240. 

Secretary:   Colin  G.  Fink,  244. 

Void  Ballots,  33.  ^  ata    ttt 

Lincoln  T.  Work 

^    ^     '        Arthur  K.  Doolittle 

The  President  announced  the  following  elections,  as  the  result 
of  the  Tellers'  report: 

President:   A.  T.  Hinckley. 

Vice-Presidents:  Lawrence  Addicks,  G.  K.  Elliott.  Henry 
1  If)ward. 

Managers:  F.  M.  Becket,  C.  B.  Gibson,  R.  A.  Witherspoon. 
Treasurer:   F.  A.  Lidbury. 
Secretary:  Colin  G.  Fink. 

Following  this  announcement  the  Chir  requested  Mr.  Acheson 
Smith  to  escort  President-elect  Hinckley  to  the  platform. 


(5ti:c<.a^r^      ;^ /^^X^^^iO^^^ 


PROCEEDINGS.  5 

DR.  EDWARD  G.  ACHESON  MADE  HONORARY  MEMBER 

President  Schluederberg  announced  that,  at  the  meeting  of  the 
Board  of  Directors  held  Wednesday  evening,  Dr.  Edward  G. 
Acheson  was  elected  to  Honorary  Membership  upon  official  rec- 
ommendation from  15  members  of  the  Society.  Dr.  Hering 
escorted  Dr.  Acheson  to  the  platform,  whereupon  Mr.  F.  A.  J. 
FitzGerald  delivered  the  following  introductory  address  on  E.  G, 
Acheson  and  his  work. 


DR.  EDWARD  G.  ACHESON  AND  HIS  WORK 

By  F.   A.  J.   FitzGerald.' 

The  twenty-first  anniversary  of  the  American  Electrochemical 
Society  has  a  special  significance.  It  is  fitting  that  its  coming 
of  age  should  be  marked  by  the  conferring  of  its  Honorary 
Membership  on  one  whose  name  is  universally  known  for  what 
he  has  done  in  advancing  electrochemical  industry.  Inasmuch  as 
I  had  the  good  fortune  to  serve  an  eight  years'  apprenticeship 
with  Dr.  Edward  G.  Acheson  when  some  of  his  inventions  were 
made,  I  can  perhaps  contribute  something  in  the  way  of  an  appre- 
ciation of  his  work  in  electrochemistry. 

You  have  no  doubt  observed  that  some  naive  amateurs  inter- 
ested in  industrial  research  seem  to  consider  the  objective  of  the 
work  as  something  of  minor  importance  compared  with  the  pro- 
cess of  reaching  it;  but  actually  the  conception  of  an  invention 
often  demands  a  rarer  gift  than  the  working  out  of  its  details. 
Dr.  Acheson's  work  is  characterized  particularly  by  this  gift  of 
choosing  objectives  to  which  he  devotes  his  inventive  genius. 

One  of  his  early  objectives  was  the  production  of  an  abrasive 
material,  the  properties  of  which  would  surpass  anything  that 
could  be  obtained  from  natural  sources,  for  he  visualized  clearly 
the  vast  industrial  importance  of  such  a  material.  So  he  made 
his  well-known  experiment  with  the  arc  light  electrode  and  the 
plumber's  soldering  bowl  filled  with  clay  and  powdered  coke, 
which  resulted  in  the  discovery  of  carborundum  and  the  building 
of  a  miniature  electric  furnace  for  its  manufacture.    This,  I  think, 

»  FitzGerald  Labs.,  Niagara  Falls,  N.  Y. 


6  PROCIIEDINGS. 

illustrates  clearly  what  I  mean  by  Achesoii's  faculty  of  seeing  a 
valuable  objective,  the  direction  of  experimental  work  which 
would  have  a  great  industrial  future. 

Acheson's  subsequent  work  in  creating  the  artificial  abrasive 
industry  illustrates  another  characteristic  of  his  which  is  so  fre- 
quently lacking  in  those  who  may  perhaps  equal  him  in  original 
ideas;  that  characteristic  is  his  ability  to  concentrate  on  his  sub- 
ject. The  late  William  De  Morgan,  probably  most  widely  known 
as  the  author  of  those  remarkable  series  of  novels  which  began 
with  "Joseph  Vance,"  devoted  most  of  his  life  to  the  manufacture 
of  pottery,  and  did  wonderful  work  in  that  field  with  his  inventive 
genius ;  but  according  to  one  who  knew  him  well,  "his  mind  was 
ever  full  of  original  methods  and  ideas  on  all  sorts  of  subjects," 
and  "it  was  perhaps,  to  some  extent,  the  wide  range  of  William 
De  Morgan's  inventive  and  creative  ability  which  tended  in  a 
measure  to  hamper  the  success  of  the  pottery."^ 

Like  De  Morgan,  Acheson's  mind  is  full  of  original  methods 
and  ideas ;  but  these  are  not  allowed  to  interfere  with  the  develop- 
ment of  any  particular  objective  he  has  in  mind.  Note  how  in 
the  early  days  of  carborundum  manufacture  he  observed  and 
recognized  the  value  of  the  artificial  graphite  produced  by  the 
decomposition  of  silicon  carbide ;  how  he  realized  the  remarkable 
refractory  qualities  of  silicon  carbide ;  how  he  made  calcium 
carbide  in  his  electric  furnace ;  but  he  did  not  allow  these  things 
to  divert  his  attention  from  the  great  objective,  the  production 
of  carborundum,  which  would  revolutionize  the  abrasive  industry, 
on  a  large  scale. 

Observe  also  that  Acheson  fully  recognized  the  importance  of 
basing  his  work  on  fundamental  scientific  principles.  This  may 
not  appear  a  surprising  thing  to  us  at  the  present  day,  for  we 
would  certainly  be  astonished  at  an  electrochemical  plant  that 
attempted  to  run  without  scientific  control.  But  when  Acheson 
made  his  first  little  vial  of  carborundum  he  sold  it  at  40  cents 
a  carat  and  devoted  the  proceeds  to  the  purchase  of  a  microscope, 
and  when  he  organized  the  Carborundum  Company  with  its  little 
100-kilowatt  plant  in  Monongahela  he  at  once  established  a  chem- 
ical laboratory   in  charge  of  a  German  chemist,  in  those  days 

2  "William  De  Morgan  anH  His  wife"  by  A.  M.  W.  Stirling.  Henry  Holt  and  Co., 
1922. 


PROCEEDINGS.  7 

believed  to  be  the  best  variety.  This  was  an  extraordinary  thing 
for  an  abrasive  manufacturer  in  those  days,  and  led  to  the  desig- 
nation of  the  Carborundum  Company  as  a  plant  "run  by  educated 
blockheads." 

This  was  the  beginning  of  the  manufacture  of  artificial  abra- 
sives and  it  is  not  necessary  to  tell  you  what  that  electrochemical 
industry  has  become;  but  I  may  note  that  the  world  production 
of  electric  furnace  abrasives  in  1895  came  from  Acheson's  100- 
kilowatt  furnace  in  Monongahela,  while  25  years  later  at  Niagara 
Falls  alone  the  power  used  in  this  industry  amounted  to  20,000 
kilowatts. 

In  the  development  of  Acheson  Graphite,  we  again  have  an 
excellent  example  of  his  faculty  for  selecting  an  objective  that 
would  develop  into  an  important  industry.  Acheson  observed 
the  formation  of  graphite  in  his  early  carborundum  furnace ;  but 
it  was  not  until  several  years  later,  after  carborundum  had  already 
become  an  important  industry  and  when  he  undertook  to  graph- 
itize  carbon  anodes  at  the  Carborundum  Company's  plant  for  use 
in  the  Castner  caustic  soda  cell,  that  he  concentrated  his  energy 
on  the  building  up  of  the  graphite  industry.  It  is  needless  to 
dwell  on  the  value  of  Acheson  graphite  in  electrolytic  and  electro- 
thermic  processes ;  but  it  will  be  interesting  to  consider  some  other 
work  relating  to  its  development. 

Persistent  refusal  to  accept  defeat  is  a  quality  of  the  highest 
value  in  war  and  in  pioneer  work  in  a  new  process  or  industry. 
There  is  another  quality,  however,  that  is  perhaps  rarer  and 
equally  valuable  and  this  is  the  faculty  of  recognizing  a  tactical 
error  and  effecting  a  strategic  retreat.  In  the  development  of  the 
graphitizing  industry,  after  much  experimental  work  was  done 
on  the  production  of  bulk  graphite  from  anthracite  coal,  Dr. 
Acheson  designed  a  furnace  for  its  production  that  has  since 
figured  prominently  in  text  books  both  in  this  country  and  abroad, 
and  yet  that  furnace  was  never  run  commercially.  On  paper  it 
looked  excellent;  it  was  built  and  the  special  electrical  apparatus 
required  for  it  obtained  and  set  up;  but  about  six  hours'  trial 
convinced  Dr.  Acheson  that  it  was  useless  as  a  commercial  ap- 
paratus and  it  was  then  and  there  sentenced  to  the  scrap  pile. 

The  thing  I  wish  to  emphasize  is  that  Acheson  saw  at  once  that 


8  PROCEEDINGS. 

the  design  was  faulty ;  that  it  would  be  throwing  away  money  to 
try  to  get  it  to  work,  and  that  complete  abandonment  of  the  scheme 
was  the  proper  course  in  spite  of  the  large  expenditure  already 
incurred.  It  is  worth  noting  that  subsequent  experience  has 
completely  demonstrated  the  soundness  of  Acheson's  judgment ; 
but  there  can  be  no  question  that  it  required  a  high  order  of 
courage  on  his  part  to  resist  the  temptation  of  trying  to  make  an 
apparatus  work  on  which  so  much  had  been  spent. 

A  large  and  important  use  for  graphite  is  in  the  manufacture 
of  crucibles.  When  Acheson  started  experimental  work  on  these 
he  found  that  the  clay  bond  used  in  making  them  was  imported, 
the  reason  given  for  this  being  that,  while  in  this  country  we  had 
plenty  of  refractory  clays,  none  of  these  combined  sufficiently 
high  plasticity  with  refractory  qualities.  Acheson  reacted  char- 
acteristically to  this,  and  determined  to  investigate  the  plasticity 
of  clays.  He  apparently  felt,  as  I  think  he  always  does,  that 
it  is  not  necessary  to  depend  on  natural  sources  for  things  of 
this  sort,  but  that  they  can  be  better  made  by  man. 

He  immediately  began  experimental  work  based  on  the  effect 
of  organic  substances  on  clay,  and  two  or  three  days  later,  on  a 
Monday  morning,  he  came  into  the  laboratory  carrying  a  big 
load  of  straw,  which  he  deposited  on  the  bench.  He  then  told  me 
that,  on  the  day  before,  his  children  had  been  having  at  their 
Sunday  School  the  story  of  the  Egyptian  bondage  of  the  Children 
of  Israel,  and  the  difficulties  they  met  in  making  brick  without 
straw.  Why,  he  asked,  did  they  need  straw  for  their  brick? 
Surely  not  as  a  mere  re-enforcement,  but  because  the  aqueous 
extract  of  the  straw  gave  to  the  clay  the  plasticity  and  mechanical 
strength  which  he  was  seeking.  He  began  the  straw  experiments 
that  day,  and  I  worked  on  big  wash  tubs  of  clay  and  aqueous 
extracts  of  straw  nearly  all  night  with  Dr.  Acheson,  whose  health 
was  bad  at  the  time,  following  up  and  directing  the  experiments 
through  the  long  distance  telephone.  Thus  Egyptianized  clay  was 
discovered. 

It  was  about  this  time  also  that  Acheson  was  working  on 
Siloxicon  ;  but  I  can  not  do  more  than  mention  it  by  name  so  as 
to  leave  a  little  time  for  the  consideration  of  one  more  of  his 
inventions. 


PROCEEDINGS.  9 

I  have  already  called  attention  to  Dr.  Acheson's  keen  appre- 
ciation of  what  is  needed.  At  the  present  day  we  have  all  sorts 
of  problems,  the  handling  of  which  will  profoundly  affect  the 
future  of  civilization.  We  have  reparations,  the  disastrous  effects 
of  phrases  like  "self-determination,"  German  marks,  etc.,  but 
there  is  none  of  the  problems  more  important  than  that  of  the 
mineral  oil  fields.  Twenty  years  ag'o  the  average  man  was  not 
worrying  about  mineral  oil,  today  he  is  thinking  of  it  seriously. 
But  even  now  he  is  only  thinking  of  it  as  a  source  of  fuel,  more 
particularly  as  supplying  the  wants  of  his  motor  car  or  his  Ford. 

But  fuel  is  not  the  most  important  point ;  we  have  other  sources 
of  fuel — even  our  powerful  army  of  moral  uplifters  can  not 
amend  the  constitution  of  nature.  Careful  study  of  the  subject 
shows  that  the  great  value  of  mineral  oil  lies  in  its  lubricating 
qualities,  and  it  is  becoming  clearer  and  clearer  now  that  the 
strong  argument  for  its  conservation  is  found  in  its  importance 
as  a  lubricant.  Twenty  years  ago,  Acheson  saw  this  clearly,  and 
determined  to  turn  his  inventive  genius  towards  finding  a  sub- 
stitute for  mineral  lubricating  oils,  or  at  least  something  that 
would  lead  to  their  economical  use. 

It  would  take  too  long  to  follow  the  development  of  methods 
of  producing  a  nearly  chemically  pure  non-coalescing  graphite 
for  lubrication,  nor  is  this  necessary  because  the  technical  details 
of  this  work  are  probably  well  known  to  most  of  you.  It  is  suffi- 
cient to  note  that  the  inception  of  the  work  on  deflocculated 
Acheson  graphite  was  similar  to  that  which  characterizes  other 
fields  in  which  Acheson  worked,  and  that  in  its  development  we 
find  the  same  ingenuity,  resourcefulness  and  persistence  which 
distinguish  his  other  work. 

Those  of  you  who  are  familiar  with  Acheson's  work  will  under- 
stand how  inadequate  this  review  is.  I  hope,  however,  that  I 
have  said  enough  to  show  to  those  unacquainted  with  the  details 
of  his  work  why,  in  the  history  of  electrochemical  industry,  one 
of  the  great  names  is  that  of  Edward  Goodrich  Acheson. 


Following  the  above  address.  President  Schluederberg  pre- 
sented Dr.  Acheson  with  a  certificate  of  Honorary  Membership. 
In  response.  Dr.  Acheson  spoke,  in  part,  as  follows : 

2 


lO  PROCEEDINGS. 

"I  appreciate  most  highly  the  honor  you  are  conferring  upon 
me.  It  is  a  matter  of  much  gratification  to  me  to  know  that  I 
assisted  in  the  organization  and  the  early  work  of  our  Society. 
I  am  gratified  to  know  that  it  has  become  a  national  society  of 
considerable  magnitude,  with  a  foreign  membership  of  which  we 
can  be  proud.  It  has  done  good  work  in  the  past,  and  I  hope  and 
believe  it  will  do  more  valuable  work  in  the  future.  I  hope  that 
it  will  continue  to  hold  its  place  among  the  national  societies,  and 
that  we  will  all  have  good  reason  to  be  proud  of  having  been 
enrolled  in  its  membership." 

President  Schluederberg  then  invited  President-elect  Hinckley 
to  assume  the  Chair  during  the  presentation  of  his  presidential 
address,  entitled,  "Opportunities  for  the  American  Electrochemist 
Abroad."'    This  address  is  printed  in  full  in  this  volume. 

Thereafter  papers  by  the  following  were  presented  for  dis- 
cussion :  W.  Blum  and  H.  S.  Rawdon ;  H.  E.  Haring  and  W. 
Blum ;  A.  Kenneth  Graham ;  M.  R.  Thompson ;  H.  D.  Hineline. 
These  papers,  with  the  exception  of  that  of  H.  D.  Hineline,  will 
be  published  in  Volume  44  of  the  Transactions.  The  final  paper 
of  this  session  was  by  Will  Baughman,  on  lead  vanadates.  It  is 
printed  in  this  volume. 

At  6.00  P.  M.  the  Council  of  the  Electrothermic  Division  held 
a  dinner-meeting.  This  was  followed  by  a  meeting  of  the  Advi- 
sory Committee  to  the  Bureau  of  Mines  on  electrometallurgical 
work. 

PROCEEDINGS  OF  FRIDAY,  MAY  4,  1923 

The  session  was  called  to  order  at  9.30  A.  M.  by  President 
Schluederberg,  and  the  results  of  the  election  of  officers  to  the 
Electrothermic  Division  and  the  Electrodeposition  Division  for 
1923-1924  were  announced.  The  officers  of  these  divisions  are 
printed  on  the  first  pages  of  this  volume. 

The  technical  session  began  with  the  presentation,  by  title,  of 
papers  by  J.  C.  Woodson  and  Frank  W.  Brooke.  Both  these 
papers  had  been  read,  but  not  preprinted,  at  the  42nd  meeting  of 
the  Society.  Then  followed  the  presentation  for  discussion  of 
papers  by  the  following  authors :  M.  deKay  Thompson  and  Per 
K.  Frolich;  A.  E.  R.  Westman ;  C.  E.  Williams.  C.  E.  Sims  and 
C.  A.  Newhall ;  C.  W.  Drury  ;  F.  C.  Kelley.    All  the  above  papers, 


PROCEEDINGS.  1 1 

with  discussions,  are  printed  in  this  volume.     The  meeting  ad- 
journed at  11.30. 

At  12  o'clock  members  and  guests  left  by  train  for  Westport, 
Conn.  At  the  kind  invitation  of  Dr.  J.  V.  N.  Dorr  and  Mr.  H.  N. 
Spicer,  a  visit  was  made  to  the  Westport  Mill  of  the  Dorr  Co. 
The  members  were  also  guests  at  an  enjoyable  luncheon  served 
amid  the  beautiful  and  idyllic  surroundings  of  the  mill.  Later 
in  the  afternoon,  members  and  guests  went  on  to  the  Westport 
Country  Club,  where  a  golf  tournament  was  staged  by  the  men, 
while  the  ladies  enjoyed  bridge  and  walks.  During  the  dinner, 
which  was  served  under  the  auspices  of  the  Xew  York  Section 
of  the  Society,  and  for  the  successful  arrangement  of  which 
Mr.  Irving  Fellner  was  responsible,  the  following  golf  prizes 
were  awarded :  An  engraved  silver  loving  cup,  donated  by  Dr. 
Dorr,  to  Frank  J.  Vosburgh ;  a  niblick,  as  booby  prize,  to -Robert 
Burns.  This  was  followed  by  the  clever  rendition  of  a  funny 
song,  by  Messrs.  Lidbury  and  Hinckley,  which,  as  a  parody  of 
"Mr.  Gallagher  and  Mr.  Shean,"  characterized  numerous  mem- 
bers of  the  Society.  Thereafter  dancing  was  enjoyed  by  many, 
and  it  was  with  great  reluctance  that  the  party  returned  to  New 
York  later  in  the  evening. 

PROCEEDINGS  OF  SATURDAY,  MAY  5,  1923 

On  Saturday,  at  9.15  A.  M.,  President  Schluederberg  opened 
the  meeting  by  introducing  Dr.  F.  M.  Becket,  who  had  arranged 
for  an  interesting  and  comprehensive  session  on  the  "Production 
and  Application  of  the  Rarer  ^letals."  Dr.  Becket  assumed  the 
Chair  and  papers  were  presented  by  the  following  authors : 
C.  James;  H.  S.  Cooper;  H.  W.  Gillett  and  E.  L.  Mack;  F.  M. 
Becket;  B.  D.  Saklatwalla;  R.  W.  Moore;  J-.  A.  Holladay  and 
T.  R.  Cunningham ;  R.  P.  Neville ;  Edward  Wichers  and  Louis 
Jordan  ;  F.  E.  Carter.  These  papers,  with  discussions,  are  printed 
in  these  Transactions.  Another  paper  which  had  been  contributed 
toward  this  symposium,  but  arrived  too  late  to  permit  of  its 
presentation  at  this  meeting,  will  be  printed  in  the  subsequent 
volume,  vis.,  44.  This  paper  is  entitled  "The  Reduction  of  Some 
Rarer  Metal  Chlorides  by  Sodium,"  by  ^M.  A.  Hunter  and  A. 
Jones.    In  concluding  the  technical  program,  President  Schlueder- 


12  PROCEEDINGS. 

berg,  on  behalf  of  the  Society,  thanked  Dr.  Becket  for  his  suc- 
cessful efforts  in  procuring  the  many  excellent  papers  for  this 
session.  He  also  expressed  thanks  to  those  who  had  contributed 
papers  and  discussion  to  the  meeting. 

Prior  to  adjourning  the  meeting,  Dr.  Hering  offered  the  fol- 
lowing 

RESOLUTION   OF  THANKS. 

Resolved:  That  a  vote  of  thanks  be  given  to  the  following  for 
having  made  this  forty-third  meeting  of  the  American  Electro- 
chemical Society  such  a  success : 

The  Dorr  Co.,  and  especially  to  Dr.  J.  V.  N.  Dorr  and  Mr. 
H.  N.  Spicer. 

The  Westport  Countr}^  Club. 

The  McGraw-Hill  Co. 

The  New  York  Local  Section. 

The  Local  Committee,  and  especially  to  Mr.  Irving  Fellner,  its 
active  chairman. 


ANNUAL  REPORT  OF  THE  BOARD  OF  DIRECTORS 

To  the  Members  of  the  American  Electrochemical  Society: 

The  following  are  some  of  the  important  items  of  business 
transacted  by  your  Board  of  Directors  during  the  past  year:  The 
following  constitutional  amendment,  effective  January  1,  1923. 
submitted  over  the  signatures  of  15  members,  was  adopted:  That 
Article  4,  Section  2.  "The  annvial  dues  shall  be  five  dollars",  be 
changed  to  read  "The  annual  dues  shall  be  eight  dollars."  It  was 
further  moved  and  passed  that  commencing  January  1,  1923, 
bound  volumes  be  charged  to  members  at  the  rate  of  $5.00  per 
year,  to  non-members  at  $6.50  per  volume,  and  to  public  libraries 
and  scientific  societies  at  $4.00  per  volume.  The  proposed  by-laws 
for  the  Electrodeposition  Division  were  adopted,  the  result  of 
the  vote  being  115  in  favor,  1  opposed.  Mr.  Acheson  Smith  was 
appointed  to  represent  the  Society  on  the  National  Research 
Council,  June  30,  1922.  to  June  30,  1925.  Acting  on  a  resolution 
submitted  and  signed  by  seventeen  members  of  the  Society,  Dr. 
Carl  Hering  was  unanimously  elected  to  Honorary  IMembersbip 
at  the  .Annual  Meeting  of  the  Board.     See  Volume  41.  2  (1922) 


PROCEEDINGS.  13 

It  was  adopted  that  the  price  of  our  Transactions  to  members  of 
the  Faraday  Society  be  the  same  as  to  our  members.    The  reloca- 
tion  of    the    Society's    headquarters    from    Bethlehem,    Pa.,    to 
Columbia  University,  New  York,  was  adopted  by  a  majority  two- 
thirds  vote  of  the  Board  of  Directors.    The  change  was  accord- 
ingly made  August  1,  1923.     At  the  July  Directors'  Meetuig  it 
was  resolved  that  the  Publication  Committee  hereafter  be  guided 
by  a  limit  of  about  400  pages  per  volume  of  the  Transactions. 
In  August  of  last  year  the  following  measure  was  adopted : 
That  each  Board  of  Directors  of  the  Society  prepare  a  tentative 
program  for  the  two  meetings  of  the  subsequent  year  and  that  it 
appoint,  not  later  than  the  fall  meeting,  a  committee  from  among 
its  members  to  carry  such  programs  into  effect,  subject  to  the 
approval  of  the  new  Board. 

The  Board  approved  that  any  person  whose  membership  was 
suspended  during  the  war  on  account  of  nationality,  may  upon 
written  application  to  the  secretary,  be  reinstated  without  election 
or  payment  of  the  initiation  fee. 


SECRETARY'S  ANNUAL  REPORT 
To   the   Board   of   Directors   of   the  American   Electrochemical 
Society: 

Gentlemen  :  The  Society  held  two  General  ^leetings  during 
1922— one  in  Baltimore,  Md.,  April  27,  28  and  29,  at  which  the 
attendance  was  125  members  and  77  guests,  total,  202 ;  the  second 
in  Montreal,  Que.,  September  21,  22  and  23,  at  which  the  regis- 
tration was  75  members  and  135  guests,  total  210.  The  Trans- 
actions of  the  spring  meeting,  the  feature  of  which  was  the  session 
devoted  to  the  reading  and  discussion  of  papers  on  "Electric  Fur- 
nace Cast  Iron,"  include  24  papers,  and  those  of  the  fall  meeting, 
embodying  a  symposium  on  "Industrial  Heating,"  22  papers. 

The  following  bound  Transactions  of  the  Society  have  been 
mailed  to  the  membership  since  the  last  Annual  Meeting  of  the 

Society : 

Volume  XXXIX,  Atlantic  City  Meeting,  in  June,  1922. 
Volume  XL,  Lake  Placid  Meeting,  in  November,  1922. 
Volume  XLL  Baltimore  Meeting,  in  February.  1923. 
Volume  XLII,  Montreal  Meeting,  in  April,  1923. 


14  PROCEEDINGS. 

This  brings  the  distribution  of  volumes  up  to  date,  the  next 
volume  to  be  issued  being  the  one  which  will  cover  the  transac- 
tions of  this  meeting.  The  edition  of  the  above  mentioned  vol- 
umes of  the  Transactions  was  as  follows : 

o     .                        Copies  bound  r-        r^ 

Copies  com-            •       ,„„,  r„,  r^     ■  „r              Free  Copies 

N^olume  No.                   plete  bound            -  JXy  SoC  ^St'orLe '         °^  each  paper 

in  cloth                   ietysub  ^^orage              to  authors 

XXXIX              1,650              350  250              10 

XL              1,550              350  200              10 

XLI               1,400              300  200               10 

XLII              1,400              150  200              10 


The  stock  of  volumes  on  hand  April  1,  1923,  was  as  follows: 

Volume  I,  66;  II,  88;  III,  11 ;  IV,  187;  V,  210;  VI,  208;  VII, 
167 ;  VIII,  301 ;  IX,  307 ;  X,  242 ;  XI,  262 ;  XII,  252 ;  XIII,  202 
XIV,  Z77;  XV,  345;  XVI.  408;  XVII,  436;  XVIII,  589;  XIX 
387;  XX,  371 ;  XXI,  429;  XXII,  372;  XXIII,  346;  XXIV,  487 
XXV,  491 ;  XXVI,  479 ;  XXVII,  230 ;  XXVIII,  458 ;  XXIX,  96 
XXX,  421;  XXXI,  87;  XXXII,  322;  XXXIII,  281;  XXXIV 
262 ;  XXXV,  4(H ;  XXXVI.  523  ;  XXXVII,  358 ;  XXXVIII,  514 
XXXIX,  972 ;  XL,  813  ;  XLI,  787 ;  XLII,  989.    Index  1-20,  512. 

Condition  of  Membership  of  the  Society  in  1922. 

Members  January  1,  1922  2,172 

Qualified  as  members  in  1922 108 

2,280 


Deaths  in  1922 14 

Resignations  in  1922 Ill 

Dropped  for  non-payment  of  1921  dues 194 


319 


Members,  December  31,  1922 1,961 

Net  decrease  for  calendar  year 211 


PROCEEDINGS.  '5 

Condition  May  1,  1922. 

Members  January  1,  1923 1'961 

Qualified  as  members  to  May  1,  1923 28 


Deaths  ^ 


1.989 
Dropped  for  non-payment  of  1922  dues 204 

Members,  May  1,  1923 J'784 

Members,  April  27,  1922 ^'^^ 

211 
Net  decrease 

Financial  Statement. 
The  following  is  a  statement  of  receipts  and  expenditures,  as 
of  December  31,  1922: 

.   Receipts  in  1922 

Cash  Balance-January  1.  1922 ■-■•^^  3.537.13 

Entrance    Fees    ■.:::;...     6.74575 

Current   Dues    _  7gQ95 

Back  i^"$!  ••■iQ2i ■.:::■.:■.:■.:...  3,088.00 

Advance  Dues,   iy^.3    ^^ 

Advance  Dues,  1924   355  00 

Volumes — 1921    ^  851  02 

Volumes — 1922    1117  00 

Volumes— 1923 ;VV 79A\^c\ 

Sale   of   Publications— non-Members 'oficin 

Sale  of  Reprints   ^°^-i" 

Sale  of   Preprints ^^Z:'J. 

Sale  of  Membership  Certificates   o-^V 

Sale  of  Society  Pins   ■•••:■-■-••••  y   ■  \;-  :  \ 4^47 

Payment  of  1920  Transactions  by  Faraday  Society  ....  W^.4/ 

Subscription  to  Faraday  Society  Transactions    184.50 

Advance  Subscriptions  to  Ten-Year  Index  •..••••;•••  ^^'^ 

Sale  of  U.  S.  Victory  Bond  and  other  Liberty  Bonds. . .  8,910.67 

Sale  of  Phila.  Electric  Bonds  with  accrued  interest....  ^']^-f 

Interest  on  Liberty  Bonds VA(i\ 

Interest  on  Philadelphia  Electric  Bonds  ^^^-^ 

Interest  on  Bank  Balances  ^^-^V 

Miscellaneous— Refunds  on  Insurance,  etc J^.oi 

Electrothermic   Division    

Total  Receipts,  January  1  to  December  31,  1922  34.061.27 

^  ,  ,  $37,598.40 

lotal    -^1545  42 

Total    Disbursements    "^^'^^^ 

Cash  Balance-December  31,  1922  $  6,052.98 


1 6  PROCEEDINGS. 

EXPEXDITURES  IX   1922 

Publication  Expenses: 

Printing  of  Volume  38   $  3,127.64 

Printing  of  Discussion — Volume  38   431.40 

Printing  of  Volume  39   3,038.94 

Printing  of  Discussion — Volume  39   597.26 

Printing  of  Volume  40   2,528.38 

Printing  of  Discussion — Volume  40 389.07 

Preprints  for  Volume  39   834.20 

Preprints   for  Volume  40    2.855.72 

Preprints  for  Volume  41 2.525.60 

Preprints  for  Volume  42  1,916.82 

Engraving 677.71 

Extra    Reprints     312.75 

Directory  of  Members    (1921)    1,204.46 

Printing  of  Discussion — Volume  41    325.22 

Constitution  and  By-Laws    72.50 

Printing  of    Discussion — Volume   42    313.68 

Total   Publication   Expenses    $21,151.35 

Office  and  General  Expenses: 

Secretarial    Appropriation    $  3.900.00 

Office   Printing    803.75 

Office    Postage 16.95 

Office  Expense — Stationery  and  Supplies    1,107.13 

Postage  on  Preprints  and  Bulletins    817.95 

Postage  on   Volumes    410.59 

Freight  and  Express  on  Volumes  and  Preprints   81.98 

Expenses  of  Meetings    1.017.36 

Membership    Certificates    4.42 

Membership    Committee    102.85 

Publication    Committee    40.10 

Booth    Committee    75.00 

Local   Sections    235.00 

Electrothermic   Division    2.75 

Electrodeposition    Division    28.00 

Moving  Expense   (Bethlehem  to  New  York)    112.50 

Contribution  to  Annual  Tables  of  Constants    75.00 

Storage   and   Insurance    279.38 

Auditing  and  Accounting  Expenses   170.26 

Total   Office  and  General  Expenses    $  9.280.97 

Total  Expenditures.  January  1  to  December  31,  1922  $30,432.32 

Refund: 
Return    of    Loan    (with    interest)    to    J.    W.    Richards' 

Estate    $  1.018.46 

Walter  Dalton,  for  overpayment  of  Dues   5.00 

Advance  Subscription  to  Ten-Year  Index    64.00 

Collection  Charge  on  Canadian  Checks   .39 

$  1.087.85 

Bad  Debts  Charged  Off: 
J.  B.  Grenagle   (check  uncollectible)    $       25.25 

Total   Disbursements    $31,545.42 


PROCEEDINGS.  17 

TREASURER'S  ANNUAL  REPORT,   1922 

January  1,  1922,  Cash  Balance   $  3,537.13 

Total   Receipts,    1922    34,061.27 

$37,598.40 

Total    Expenditures    31,545.42 

Balance,  December  31,  1922  $  6,052.98 

Balance  in  Power  City  Bank,  12-31-22  $  7,160.57 

Deposits  not  included  in  Bank  Statement  360.85 

Balance  retained  as  petty  cash  by  Secretary's  Office...  50.00 

7,571.42 

Less  December,  1922,  checks  not  in  1,518.44 

Balance,  December  31,  1922,  as  above  $  6,052.98 

We  have  examined  the  above  statement  of  accounts,  receipts, 
and  expenditures  for  the  year  1922,  and  find  the  same  to  be 
correct. 

(Signed)     H.  B.  Coho, 
(Signed)     Harry    J.    WoivF, 

Auditors. 


i8 


PROCEEDINGS, 


MEMBERS  AND  GUESTS  REGISTERED  AT  THE  FORTY-THIRD 
GENERAL  MEETING 


Franz  D.  Abbott 

E.  G.  Acheson 
Lawrence  Addicks 
A.  N.  Anderson 
William  C.  Arsem 

D.  K.  Bachofer 
R.  O.  Bailey 
A.  T.  Baldwin 

F.  M.  Becket 

E.  O.  Benjamin 
M.  H.  Bennett 
Geo.  M.  Berry 
Edw.  L.  Blossom 
Wm.  Blum 

W.  H.  Boynton 
Robert  H.  Buckie 
C.  F.  Burgess 

C.  O.  Burgess 
R.  M.  Burns 

D.  C.  Burroughs 
P.  Caplain 

D.  C.  Carpenter 

F.  E.  Carter 
H.  Casselberry 
N.  K.  Chaney 

G.  W.  Coggeshall 
H.  B.  Coho 

S.  J.  Colvin 

E.  F.  Cone 
H.  S.  Cooper 
W.  M.  Corse 
J.  H.  Critchett 
Ed.  L.  Crosby 
Thomas  S.  Curtis 
C.  Dantsizen 

F.  W.  Davis 
Wm.  Delage 
P.  K.  Devers 
Arthur   K.  Doolittle 
E.  F.  Doom 

J.  V.  N.  Dorr 


Members 

Wm.  Dreyfus 
W.  F.  Edwards 
C.  H.  Eldridge 
W.  H.  Falck 
F.  F.  Farnsworth 
Alex  L.  Feild 
Colin  G.  Fink 

F.  A.  J.  FitzGerald 
J.  A.  Fogarty 
Oscar  R.  Foster 
Gay  N.  Freeman 
N.   H.  Furman 

A.  J.  Gailey 
Richard  H.  Gaines 
W.  H.  Gesell 
A.  E.  Gibbs 
C.  B.  Gibson 
H.  W.  Gillett 

G.  C.  Given 
J.  B.  Glaze 

A.  Kenneth  Graham 
Carl  Hambuechen 
H.  E.  Haring 
L.  O.  Hart 
W.  G.  Harvey 
Carl  Hering 
Chas.  H.  Herty 
A.  T.  Hinckley 
C.  D.  Hocker 
Geo.  B.  Hogaboom 

E.  M.  Honan 
A.  H.  Hooker 
W.  G.  Horsch 
L.  E.  Howard 
O.  Hutchins 
W.  C.  Hyatt 
John  Johnston 
Louis  Jordan 

F.  R.  Kemmer 
E.  F.  Kern 

R.  H.  Kienle 


D.  H.  Killefer 
Max  Knobel 

V.  R.  Kokatnur 

C.  G.  Koppitz 
W.  S.  Landis 
Harry  R.  Lee 
F.  A.  Lidbury 
W.  T.  Little 

E.  A.  Lof 
J.  M.  Lohr 
Russell  Lowe 
Dorsey  A.  Lyon 
Paul  McAllister 
J.  Y.   McConnell 
Robert  J.  McKay 
Duncan   MacRae 
Chas.  P.  Madsen 
Paul  D.  V.  Manning 
J.  W.  Marden 

A.  L.  Marshall 
M.  W.  Merrill 
H.  S.  Miner 
R.  B.  Moore 
W.  C.  Moore 
W.  R.  Mott 
Martha  E.  Munzer 

D.  L.  Ordway 
N.  Petinot 

E.  C.  Pitman 
H.  W.  Forth 
R.   Prefontaine 
W.  J.  Priestley 
O.  C.  Ralston 

J.  W.  H.  Randall 
W.  C.  Read 
H.  T    Reeve 
C.  H.  M.  Roberts 

F.  W.  Robinson 
C.  J.  Rodman 
Chas.  F.  Roth 

B.  D.  Saklatwalla 


PROCEEDINGS. 


19 


L.  E.  Saunders 
C.  G.  Schluederberg 
Louis  Schneider 
J.  A.  Seede 
R.  L.  Shepard 
Acheson  Smith 
W.  S.  Smith 
J.  S.  Speer 
H.  N.  Spicer 
A.  D.  Spillman 
E.  C.  Sprague 
Reston   Stevenson 
M.  E.  Stewart 
Bradley  Stoughton 
Haakon  Styri 
Henry  P.  Taber 


E.  Takagi 
Fioyd  D.  Taylor 
Hugh  S.  Taylor 
Sterling  Temple 
C.  J.  Thatcher 
M.  R.  Thompson 

F.  J.  Tone 

A.  E.  Thurber 
L.  S.  Thurston 
Henry  A.  Tobelmann 
R.  Turnbull 
F.  M.  Turner,  Jr. 
C.  H.  Tyler 
M.  A.  Ulbrich 
Mary  Upshur  Von 
Isakovics 


L.  D.  Vorce 
Frank  J.  Vosburgh 

E.  A.  Vuilleumier 
Helen    Gillette    Weir 
C.  J.  Wernlund 

A.  E.  R.  Westman 
Clyde  E.  Williams 
Roger  Williams 
A.  M.  Williamson 
Charles  Wirt 
W.  A.  Wissler 
Wm.  J.  Wooldridge 
L.  T.  Work 

F.  Zimmerman 


Guests 

Mrs.    E.    G.    Acheson,    New    York 

City 
Robert   Aiken,    Washington.   D.    C. 
Jerome  Alexander,  New  York  City 
H.  A.  Anderson,  New  York  City 
R.  W.  Baldwin,  Milwaukee,  Wis. 
Mrs.    E.     O.     Benjamin,.  Newark, 

N.  J. 
P.  H.  Brace,  Pittsburgh,  Pa. 
Robert  E.  Brown,  New  York  City 
R.  C.  Burner,  Bayside,  N.  Y. 
Joseph   T.    Butterfield,    New    York 

City 
Mrs.     Fred     E.     Carter,     Newark, 

N.  J. 
Mrs.  G.  W.  Childs,  New  York  City 
W.  H.  Coy,  New  York  City 
Helen  E.  Bailing,  New  York  City 
Edmund  S.  Davenport,  Bloomfield, 

N.  J. 
A.  W.  Davison,  Troy,  N.  Y. 
Mrs.  Maude  T.  Doolittle,  New 

York  City 
R.  W.  Erwin,  Flushing,  L.  I., 

N.  Y. 
Mrs.  Colin  G.  Fink,  Yonkers,  N.  Y. 
Charles    FitzGerald,    Malba,    L.    I., 

N.  Y. 
Mrs.  F.  A.  J.  FitzGerald,   Niagara 

Falls,  N.  Y. 


Mrs.   Oscar  R.  Foster,  New  York 
City 

Mrs.  W.  H.  Gesell,  Montclair, 
N.  J. 

F.  R.  Glenner,  New  York  City 
Max  Greeff,  East  Orange,  N.  J. 

E.  T.  Gushee,  Detroit,  Mich. 

Mrs.  Henry  K.  Hardon,   New 
York  City 

J.  E.  Harris,  New  York  City 

Henry  S.  Haupson,  New  York  City 

George  W.  Heise,  Bayside,  N.  Y. 

R.  E.  Hickman,  Maplewood,  N.  J. 

O.  K.  Holderman 

H.  D.  Holler,  New  York  City 

Mrs.  A.  H.  Hooker,  Niagara  Falls, 
N.  Y. 

Mrs.  W.  G.  Horsch,  New  York 
City 

G.  P.  Houghland,  Parlin,  N.  J. 
H.  C.  Howard,  Jr.,  Princeton,  N.  J. 
N.  Iseki,  New  York  City 

C.  James,  Durham,  N.  H. 

Mrs.  John  Johnston,  New  Haven, 
Conn. 

F.  C.  Kelley,  Schenectady,  N.  Y. 

D.  B.  Keyes,  New  York  City 
W.   P.  Kierman,  Bloomfield,  N.  J. 


20 


PROCEEDINGS. 


Mrs.    D.    H.    Killefer.    Xew    York 

City 
H.  W.  Langzettel,  Westport,  Conn. 
Mrs.   H.  W.  Langzettel.   Westport. 

Conn. 
H.  H.  Lowry,  Xew  York  City 
W.  A.  Linch,  New  York  Cit}- 
C.  E.  MacQuigg.  Xew  York  Cit}- 
Wm.  A.  Moore,  Waterbury,  Conn. 
Edward  G.  Nellis,  Xew  York  Cit}- 
Keizo  Xishimura.  Xew  York  City 
W.  B.  Xottingham,  Xew  York  City 
K.  L.  Page,  Boston,  Mass. 
P.  G.  Paris,  Westport,  Conn. 
F.  Peters,  Westport,  Conn. 
Mrs.  F.  Peters,  Westport,  Conn. 
J.  M.  Price,  Xew  York  Citj- 
M.  B.  Rascovich,  Xew  York  City 
H.  C.  Rentschler,  Bloomfield,  X.  J. 
H.  K.  Richardson,  Xewark,  X.  J. 
Mrs.  F.  W.  Robinson,  Maplewood, 

N.  J. 
Ancel  St.  John,  Brooklyn.  X.  Y. 
John  R.  Sheffield.  Jr.,  Brooklvn, 

X.  Y. 
George  Smith,  Xew  York  Citj- 
Mrs.  C.  W.  Spicer.  Plainfield,  X.  J. 


Mrs.  H.  X.  Spicer,  X'ew  York  Cit>- 
Mrs.  E.  C.  Sprague,  Buffalo,  X.  Y. 
Mrs.  W.  A.  Stedman,  Westport, 

Conn. 
T.  A.  Schwartz,  Prince  Bav,  L.  I., 

X.  Y. 
Theodore  M.   Switz,  East  Orange, 

N.  J. 
Stem  Tiberg,  X'ew  York  Cit}' 
Magnus  Tigershield,  Soderfors, 

Sweden 
R.  J.  Traill,  Ottawa,  Canada 
Miss  Bervle  Van  Allen,  Xew  York 

City 
H.  X.  \'an  Dansen,  X'ew  York  City 
G.  A.  Vaughn,  Jr.,  New  York  City 
Alois    von    Isakovics,    Monticello, 

X.  Y. 
Miss  B.  von  Isakovics,  Monticello, 

X.  Y. 
W.  B.  Wallis,  Pittsburgh,  Pa. 
W.  B.  Williams,  Xew  York  City 
Mrs.  A.   M.  Williamson,   Xiagara 

Falls,  N.  Y. 
Mrs.  Charles  Wirt,  Philadelphia,  Pa. 
J.  C.  Woodson.  East  Pittsburgh,    Pa. 
Mrs.  L.  T.  \A'ork,  Yonkers,  N.  Y. 
L.  F.  Yutema,  New  Haven,  Conn. 


The  Presidential  Address  presented   at  the 

Forty-third      General     Meeting      of  the 

American     Electrochemical     Society.  i>» 
Sew  York  City  May  3.  1923. 


OPPORTUNITIES  FOR  THE  AMERICAN  ELECTROCHEMIST 
ABROAD 

By    C.    G.    SCHLUEDERBERG.' 

There  have  appeared  in  the  journals  from  time  to  time  fairly 
complete  reports  on  the  development  of  electrochemistry  ui 
Europe  and  what  has  been  accomplished  in  our  own  country  is, 
of  course,  a  matter  of  general  information.  Visits  to  bouth 
America  and  the  Far  East  during  the  past  year  have  aftorded 
opportunity  for  first-hand  information  and  personal  observation, 
and  it  is  therefore  felt  that  a  brief  summary  of  what  has  been 
done  in  these  two  sections,  or  of  what  the  indications  are  for  the 
future,  may  help  to  round  out  our  fund  of  information  oii  electro- 
chemical development  and  on  the  opportunities  abroad  for  the 

electrochemist. 

Electrochemical  or  electric- furnace  development  on  any  com- 
mercially appreciable  scale  inherently  requires  large  amounts  of 
electric  power  at  low  cost;  therefore,  in  considering  opportunities 
for  the  electrochemist  it  is  perforce  necessary  to  give  thought  to 
the  power  resources  of  the  locality  under  observation,  as  these 
are  so  intimately  allied  with  the  possibilities  for  the  successful 
development  of  the   industries   for  which   the  electrochemist  is 

"^^InTouth  America,  the  west  coast  comitries  of  Peru  and  Bolivia, 
with  their  large  mineral  wealth  and  mining  operations  extending 
back  over  hundreds  of  years  to  the  time  of  the  Incas,  naturally 
appeal  to  the  imagination  as  fertile  fields  for  electrochemical 
activities.  Copper,  silver,  tin,  vanadium  and  other  ores  are  mined 
m  quantities  and,  in  the  case  of  the  copper,  refined  locally  to  a 
high  degree  of  purity  in  large  smelters  of  the  most  modern  type; 
the  final  purification  by  electrolysis  is,  however,  not  carried  out 
on  the  ground,  but  usually  at  some  of  the  large  refineries  m  the 

1  Westinghouse  Elec.  &  Mfg.  Co.,  East  Pittsburgh,  Pa. 


22  C.  G.  SCHLUEDERBERG. 

vicinity  of  New  York,  the  metal  as  shipped  containing  upwards 
of  96  per  cent  copper  or  copper  and  silver. 

In  view  of  the  fact  that  this  metal  is  shipped  in  such  a  pure 
state,  and  that  it  can  receive  final  purification  in  existing  refineries 
close  to  the  markets,  there  is  at  the  present  time  no  necessity  for 
the  investment  of  the  additional  capital  which  would  be  required 
for  the  building  of  an  electrolytic  plant  on  the  ground.  A  decided 
change  in  labor  or  power  rates  of  existing  electrolytic  refineries, 
or  in  market  conditions,  might  possibly  justify  such  an  electrolytic 
plant  in  the  future. 

Water-power,  while  not  over-abundant  on  the  Pacific  side  of 
the  Andes,  is  available  on  the  eastern  slopes  in  quantities.  Pres- 
ent transportation  facilities  to  the  sites  of  such  power,  as  well 
as  conditions  inherent  to  the  tropical  climate  of  that  region,  while 
bad,  cannot  be  considered  as  insuperable  obstacles.  Many  sur- 
veys have  been  made  and  it  is  quite  certain  that  power  develop- 
ments will  take  place. 

Considerable  experimental  work  along  electrochemical  lines  has 
been  done  by  one  of  the  larger  companies  in  Peru,  with  a  view 
to  the  working  out  of  a  satisfactory  process  for  extracting  the 
silver  from  certain  of  the  local  complex  ores,  which  so  far  it  has 
not  been  possible  to  work  on  a  commercial  scale.  Should  the 
results  of  this  research  work  prove  satisfactory,  it  is  quite  likely 
that  an  electrochemical  plant  of  size  would  be  erected  in  the 
Cordilleras  of  the  Peruvian  Andes  in  the  neighborhood  of 
La  Oroya  or  Cerro  de  Pasco.  The  opportunities  for  the  electro- 
chemist  in  connection  with  the  complex  silver  ores  of  Ii*eru  loom 
large  indeed.  Undoubtedly  ores  of  many  of  the  other  less  com- 
mon metals  will  afford  equally  attractive  possibilities  to  the 
electrochemist  with  enough  pioneering  spirit  in  his  make-up  not 
to  be  deterred  by  primitive  living  conditions  and  the  discomforts 
of  working  at  the  high  altitudes  which  surround  the  deposits  of 
the  precious  metals  in  this  country,  the  fabulous  wealth  of  which 
was  first  revealed  to  the  then  civilized  world  by  the  indomitable 
Pizarro  almost  exactly  400  years  ago. 

In  Bolivia,  renowned  for  its  large  deposits  of  rich  tin  ore  as 
well  as  of  copper,  silver,  and  other  useful  and  precious  metals, 
electric  tin  reduction  furnaces  have  been  tried,  but,  at  least  up  to 
the  time  of  my  visit  a  few  months  ago  had  failed  to  prove  com- 


OPPORTUNITIES   FOR  AMERICAN  ELECTROCHEMISTS.  23 

mercially  successful,  the  cost  of  carbon  in  the  form  of  coal 
required  for  reduction  purposes  being  one  of  the  contributing 
factors.  Here  again  the  opportunity  for  the  electrochemist  is 
great.  Just  as  in  Peru,  water-power  is  available  in  the  tropical 
sections  of  Bolivia,  and  the  promise  of  large  oil  developments  in 
the  central  and  eastern  parts  offers  the  chance  of  cheap  fuel  for 
steam  stations,  so  that  from  the  standpoint  of  power  the  estab- 
lishment of  electrochemical  or  electrothermal  processes  in  this 
country  so  rich  in  natural  resources  is  entirely  feasible. 

Farther  south  along  the  west  coast,  the  northern  half  of  Chile 
is  another  country  richly  endowed  with  minerals,  and  containing 
what  is  probably  the  largest  copper  mine  in  the  world,  as  well  as 
large  deposits  of  iron  ore,  saltpeter,  etc.,  but  with  an  almost  entire 
absence  of  water-power  or  even  rain,  while  the  southern  portion, 
not  so  richly  endowed  with  metal-bearing  ores,  is  blessed  with  an 
abundant  rainfall,  water-power,  and  coal.  However,  the  dis- 
tances are  great  and  present  indications  for  long-distance  electric 
power  transmission  not  promising.  In  spite  of  this  handicap, 
electrolytic  refining  of  copper  is  carried  out  on  a  large  scale  at 
the  Chile  Exploration  Company's  copper  mine  in  northern  Chile, 
but  the  copper  ore  here  is  in  the  form  of  salts  readily  soluble, 
from  the  solution  of  which  the  metal  can  be  obtained  more 
readily  and  economically  by  electrolytic  means  than  otherwise, 
in  spite  of  the  necessity  of  generating  electricity  at  an  oil-fired 
steam  plant  on  the  sea  coast  many  miles  distant  from  the  mine, 
firing  the  boilers  with  oil  transported  by  ship  from  Mexico  and 
the  transmitting  of  energy  over  high-tension  lines  at  110,000 
volts.  Even  the  water  for  lixiviation  of  the  ore,  as  well  as  for 
all  other  purposes  at  this  mine,  has  to  be  carried  for  many  miles 
through  large  pipe  lines  from  distant  mountain  sources. 

The  plant  of  this  company  represents  the  one  outstanding 
electrochemical  development  on  the  west  coast  of  South  America. 
It  is  a  monument  to  the  American  electrochemists,  through  whose 
efforts  the  many  details  incident  to  the  successful  development 
of  a  commercially  successful  process  for  the  extraction  of  ore  on 
a  large-tonnage  basis,  not  the  least  important  of  which  was  the 
production  of  an  insoluble  anode,  have  been  satisfactorily 
worked  out. 

Indications  of  oil  resources  near  the  eastern  boundary  of  Chile, 


24  t:.  G.  SCHLUEDERBKRG. 

as  well  as  further  developments  in  long-distance  power  trans- 
mission, give  promise  of  additional  opportunities  for  the  electro- 
chemist  in  this  progressive  South  American  republic  so  far- 
famed  for  its  mineral  resources. 

So  much  for  the  west  coast  of  South  America, 

In  the  front  rank  of  those  countries  bordering  on  the  east  coast 
and  readily  reached  by  a  two-day  journey  from  Chile  over  the 
famous  Transandine  Railway  is  the  republic  of  Argentina,  for 
whose  renowned  wealth,  however,  cattle  and  cereals  and  not 
minerals  are  responsible.  Argentina  is  almost  devoid  of  water- 
powers  of  any  size.  Even  in  mineral  resources  she  is  almost 
totally  lacking.  It  is  true  that  near  the  northeastern  border  are  the 
Falls  of  the  Iguassu,  reputed  to  be  capable  of  delivering  many 
hundreds  of  thousands  of  horsepower,  and  on  the  western  border 
the  waterfalls  of  the  Andes,  but  these  are  so  far  removed  from 
present  centers  of  civilization  or  human  activity  of  any  kind  that 
even  modern  electric  transmission  developments,  using  220,000 
volts,  do  not  indicate  that  it  is  yet  advisable  to  attempt  the  har- 
nessing of  these  waterfalls.  As  a  matter  of  fact,  the  Falls  of  the 
Iguassu,  located  almost  at  the  point  where  Brazil,  Argentina,  and 
Paraguay  touch,  are  nearer  the  center  of  industrial  activities  in 
Brazil  than  in  Argentina. 

The  progressive  republic  of  Uruguay  very  much  resembles 
Argentina  both  in  resources  and  in  that  at  the  present  time  there 
are  no  electrochemical  or  electric  furnace  developments,  with 
the  possible  exception  of  one  or  two  small  steel  furnaces  for  use 
in  foundries,  so  that  in  neither  of  these  countries  does  any  imme- 
diate opportunity  exist  for  the  electrochemist. 

Brazil,  rich  in  mineral  resources  and  with  great  quantities  of 
water-power  distributed  over  her  vast  area,  offers  much  in  the 
way  of  opportunity  to  the  electrochemist.  He  will  find  here 
great  beds  of  rich  iron  ore,  immense  deposits  of  manganese,  vast 
stores  of  the  rarer  metals  and  elements  so  widely  used  in  the 
industries,  fluxes,  and  reducing  agents  in  the  form  of  charcoal 
from  the  rapidly  maturing  eucalyptus  tree,  and  water-powers  in 
abundance.  These  are  near  the  sea  coast  and  existing  centers 
of  civilization,  many  of  them  already  developed  with  power  lines 
extending  over  wide  stretches  of  territory. 

The  most  important  electrochemical  development  is  that  of  the 


OPPORTUNITIES   FOR  AMERICAN  ELECTROCHEMISTS.  25 

Brazilian  Electrometallurgical  Company  at  Ribeirao  Preto.  where 
two  30-ton  electric  pig-iron  furnaces  have  been  erected,  together 
with  two  6-ton  Bessemer  converters  for  the  direct  conversion  of 
the  hot  iron  ore  into  steel,  as  well  as  a  Ludlum  6-ton  electric 
steel  furnace  for  the  treatment  of  such  steel  as  may  be  received 
from  the  Bessemer  converters  and  require  special  doctoring  in 
order  to  bring  it  up  to  the  desired  composition.  In  addition,  there 
are  rolling  mills  for  plates  and  shapes,  reheating  furnaces,  and 
the  necessary  auxiliaries.  Recent  reports  from  this  operation 
indicate  that  so  far  the  plant  has  worked  only  on  scrap  metal, 
with  some  pig  iron,  which  is  melted  in  the  Ludlum  steel  furnace. 
They  have  rolled  as  much  as  20  tons  of  round  and  square  bars 
per  day,  which  have  been  offered  at  prices  10  per  cent  below  quo- 
tations on  similar  foreign  material.  Owing  to  the  railroad  not 
having  been  completed  to  the  iron  ore  mine,  no  ore  has  yet  come 
in,  and  hence  reduction  operations  have  not  commenced.  It  is 
reported  that  the  company  has  been  able  to  book  enough  business 
to  keep  the  plant  busy  for  the  next  year  or  more. 

Whereas  on  the  west  coast  of  South  America  practically  all 
electrochemical  and  electrometallurgical  processes  and  operations 
are  carried  on  by  Americans  or  Europeans,  on  the  east  coast, 
as  in  Brazil,  this  work  is  being  carried  on  and  financed  in  a  large 
measure  by  Brazilians,  although  even  here  the  apparatus,  of 
American  or  European  manufacture,  so  far  has  generally  been 
installed  by  American  or  European  engineers. 

A  second  plant  for  the  electric-cupola  reduction  of  the  local 
deposits  of  iron  ore  on  a  much  larger  scale  is  under  active  con- 
sideration in  this  same  district,  and  it  is  reported  that  the  rather 
large  financing  required  is  being  carried  out  successfully  and  that 
steel  rails  will  be  the  principal  product. 

With  abundant  cheap  power  readily  available  from  the  numer- 
ous waterfalls,  and  plentiful  deposits  of  iron  ore  of  an  excellent 
grade,  as  well  as  manganese  and  other  necessary  alloys  and 
fluxes,  and  a  local  market  for  pig  iron  and  steel  products,  the 
only  material  thing  which  seems  to  stand  in  the  way  of  Brazil 
becoming  a  considerable  producer  of  iron  and  steel  products  seems 
to  be  the  question  of  a  suitable  reducing  agent,  such  as  coal  or 
coke.  Here  it  becomes  necessary  to  substitute  charcoal  usually 
obtained  from  eucalyptus  trees,  which  mature  within  five  years  in 


26  C.  G.  SCHLUEDERBERG. 

this  tropical  climate,  and  the  wood  of  which,  planted  in  large 
numbers,  regularly  serves  as  fuel  for  railways  and  industrial 
plants.  The  fact  that  in  the  electric  cupola  carbon  is  consumed 
only  in  proportion  to  the  amount  of  ore  reduced,  and  that  it  does 
not  have  to  serve  the  dual  purpose  of  both  fuel  and  reducing 
agent,  is  a  factor  of  no  mean  importance  where  only  such  an 
expensive  form  of  carbon  is  readily  obtainable. 

Outside  of  possibly  a  few  electric  furnaces  for  foundry  use, 
the  above  summary  covers  electrolytic  and  electrothermal  activi- 
ties in  six  of  the  principal  countries  of  the  South  American  con- 
tinent. Reports  from  the  other  countries,  not  visited,  do  not  give 
any  immediate  encouragement  to  the  electrochemist,  but  undoubt- 
edly certain  of  the  northern  countries,  when  further  developed, 
will  offer  opportunities  similar  to  those  of  Peru  and  Bolivia. 

Turning  to  the  Far  East,  we  find  one  country  at  least  with  a 
development  along  lines  approaching  our  own  or  that  of  Europe — 
Japan — rich  in  water-powers,  many  of  them  already  developed  on 
a  large  scale,  high-tension  transmission  lines  everywhere,  and 
the  one  idea  in  the  minds  of  all  her  70,000,000  people  of  emulating 
western  civilization  and  making  industrial  progress  as  rapidly  as 
possible,  and  willing  to  sacrifice  almost  every  other  consideration 
to  this  end.  This  assimilation  of  western  civilization  started 
not  more  than  two  generations  ago,  and  has  attained  results  to 
date  which  must  command  our  admiration. 

The  war  gave  an  impetus  to  the  industries  of  Japan  as  to 
those  of  other  countries ;  the  ones  already  in  existence  increased 
and  many  others,  electrochemcial  and  electrothermal,  came  into 
being.  These  include  manufactures  of  soda,  chlorate,  carbide, 
ferro-alloys,  pure  pig  iron,  electrolytic  zinc,  copper,  etc.,  along 
with  others.  As  in  other  countries,  some  of  them  since  the  war 
had  difficulty  in  maintaining  their  existence.  Today  general 
business  in  Japan,  while  quiet,  is  improving,  but  many  of  her 
electrochemical  industries  are  working  only  part  time  or  are 
shut  down. 

One  plant  located  at  Odera,  150  miles  north  of  Tokyo,  almost 
on  the  shores  of  Lake  Inawashiro,  close  to  the  immense  power 
plants  of  the  power  company  bearing  the  name  of  that  lake,  one 
of  the  first  large  high-tension  systems  in  Japan  was  established 
in  the  fall  of  1916.    It  was  completed  in  less  than  six  months,  and 


OPPORTUNITIES   FOR  AMERICAN   ELECTROCHEMISTS. 


27 


is  devoted  largely  to  electrolytic  extraction  of  zinc  and  the  produc- 
tion of  ferro-alloys. 

The  sulfide  ore  for  this  plant  is  brought  from  a  mine  some 
fifty  miles  distant,  and  after  being  crushed  and  roasted  is  leached 
with  sulfuric  acid  made  locally  in  a  chamber  process  plant,  puri- 
fied with  zinc  shavings,  and  deposited  on  prepared  cathodes  in 
cells  much  resembling  those  used  in  an  ordinary  copper  refinery. 
Costs  are,  however,  higher  than  for  imported  electrolytic  zinc,  in 
spite  of  low  power  and  labor  charges,  but  the  purity  of  the 
product  seems  to  be  higher,  as  evidenced  by  the  following 
analysis : 


Japanese 

Electrolytic 

Zinc 

Electrolytic     Zinc 
from  U.  S.  A.  (1918) 

Zinc    

99.96721 
0.01749 
0.00319 

0.00462 
0.00749 

99.88205 

Lead    

0.075468 

Copper    

0.00796 

Iron     

0.00518 

Cadmium       

0.0O48 

The  costs  of  the  local  product  have,  however,  been  consider- 
ably higher  than  those  of  the  products  imported.  They  were 
given  as  230  yen  for  the  zinc  ore,  which  is  higher  on  account  of 
the  extensive  freight  rates,  70  yen  for  electricity,  70  yen  for 
labor,  and  150  yen  for  overhead.  (The  yen  equals  approximately 
50  cents  in  U.  S.  gold).  Labor  in  this  locality  is  very  cheap,  run- 
ning from  160  yen  to  180  yen  (80  to  90  cents  in  U.  S.  gold)  per 
day.  The  capacity  of  the  plant  is  reported  at  300  tons  of  zinc 
per  month. 

This  plant  also  turns  out  a  very  pure  grade  of  pig  iron  made  in 
an  open  type  electric  furnace,  with  phosphorus  of  0.021  per  cent, 
sulfur  0.005  per  cent,  and  copper  0.018  per  cent;  ferro-silicon 
25,  50  and  75  per  cent,  and  silicon  90  per  cent ;  60  per  cent  f  erro- 
chrome,  80  per  cent  ferro-manganese,  18  per  cent  ferro-phos- 
phorus,  and  cadmium  of  99.5  per  cent  purity,  obtained  from  the 
zinc  shavings  used  in  purifying  the  zinc  sulfate  solution  in  connec- 
tion with  the  production  of  electrolytic  zinc.  Calcium  carbide  is 
being  manufactured  in  fair  quantities  at  present ;  the  extensive 
fishing  industries  make  considerable  use  of  acetylene  torches  in 
night  fishing  and  absorb  an  appreciable  portion  of  this  product. 


28  C,  G.  SCHLUEDERBERG. 

The  Cottrell  process  is  being  applied  and  at  the  present  time 
nitrogen  fixation  and  fertiHzer  manufacture  is  receiving  much 
consideration. 

That  Japan  is  determined  to  keep  to  the  fore  in  electrochemical 
development  is  evidenced  not  only  by  what  she  has  already  done 
in  her  industries,  but  also  by  the  training  given  her  young  electro- 
chemists  in  the  various  courses  on  electrochemistry  forming  part 
of  the  regular  curriculum  of  her  universities,  by  the  amount  of 
attention  given  electrochemical  subjects  in  the  local  engineering 
and  scientific  journals,  and  also  by  the  fact  that  of  all  foreign 
countries  Japan  is  best  represented  in  the  American  Electro- 
chemical Society,  which  numbers  in  its  membership  about  75 
Japanese  members  residing  in  Japan,  as  well  as  many  others  resid- 
ing in  this  and  other  countries. 

China,  in  spite  of  her  400,000,000  people  and  most  ancient  of 
civilizations,  at  the  present  time  ofifers  but  little,  if  any,  oppor- 
tunity to  the  electrochemist.  Although  both  rich  coal  and  iron 
deposits  exist,  and  fairly  modern  blast  furnaces  and  steel  plants 
are  available  for  turning  out  pig  iron  and  finished  shapes  and 
rails,  there  has  been  practically  no  electric  furnace  development, 
and  electrochemistry,  except  as  applied  in  a  few  small  plating 
shops  and  possibly  in  the  new  mint  in  Shanghai,  is  an  unknown 
quantity.  There  is  but  little  water-power  available  or  developed 
throughout  the  vast  alluvial  plain  forming  the  eastern  central  part, 
or  Great  Middle  Kingdom  of  China.  In  one  or  two  of  the  more 
important  coast  cities  electric  furnaces  have  begun  to  be  used  in 
the  foundries  and  shops  of  the  larger  companies.  At  Hongkong, 
in  a  steel  foundry,  a  two-ton  furnace  of  one  of  the  better-known 
makes  has  given  an  excellent  account  of  itself,  albeit  hand  regula- 
tion of  electrodes  by  Chinese  labor  melting  from  cold  scrap  has 
caused  considerable  conversation  between  the  central  station  and 
foundry  managers,  with  instructions  from  the  former  to  the  latter 
to  keep  ofif  the  line  during  peak  hours. 

Disturbed  political  conditions,  instability  of  the  republican  form 
of  government  due  to  the  absolute  unpreparedness  of  the  mass  of 
the  people  for  self-government,  maintenance  of  separate  armies  by 
the  various  provincial  governors  in  their  endeavor  to  hold  indi- 
vidual power,  as  well  as  a  few  other  disturbing  factors,  are  at 
present  militating  against  industrial,  economic,  scientific,  and  all 


OPPORTUNITIES   FOR   AMERICAN    ELECTR0CHEMIST5.  2g 

Other  progress.  This  condition  applies  even  in  South  China  in 
the  province  of  Yunnan,  far-famed  for  its  wealth  of  tin,  copper, 
zinc,  and  other  ores  and  water-power.  The  immediate  prospects 
of  either  electrochemical  or  electric  furnace  developments  in  China 
are  not  encouraging. 

Xor  in  Indo-China,  the  northern  part  of  which,  bordering  as  it 
does  on  Yunnan,  is  rich  in  both  coal  and  mineral  deposits,  are 
there  any  evidences  of  electrochemical  developments  having  been 
undertaken  by  the  French,  although  here,  as  in  the  ]\Ialay  Penin- 
sula, the  problem  of  the  reduction  of  tin  is  ever  present. 

Certainly  the  Yunnan  Province  of  China,  Northern  Indo-China, 
and  the  Malay  Peninsula  offer  an  interesting  field  for  the  electro- 
chemist  who  is  not  afraid  to  go  and  remain  abroad  amid  living 
conditions  radically  different  from  those  existing  in  the  United 
States. 

Philippines:  What  should  be  the  outpost  of  American  enter- 
prise and  business  in  the  Far  East  and  what  undoubtedly  will  be, 
provided  the  uncertainty  regarding  self-government  is  eliminated 
and  the  United  States  protectorate  maintained  for  a  definitely 
stated  period  of  years,  the  Philippine  Islands,  although  blessed 
with  some  water-power  and  mineral  deposits  and  possibly  with 
oil  resources,  so  far  offer  practically  no  field  for  the  electro- 
chemist.  However,  once  political  conditions  are  sufficiently 
stabilized  to  justify  entry  of  big  business  interests  on  a  worthwhile 
scale,  there  is  promise  for  the  development  of  many  industries,  in 
some  of  which  undoubtedly  electrochemistry  will  apply.  There 
is  every  indication  that  were  the  United  States  to  guarantee 
definitely  that  present  supervision  over  the  Philippine  Islands 
would  apply  for  a  certain  period  of  years,  say  thirty,  fifty,  or 
more,  before  a  local  government  would  be  considered,  there  is  no 
doubt  in  the  minds  of  those  most  familiar  with  the  situation  that 
big  business  would  come  into  the  Philippines,  which  are  so  ideally 
suited  for  the  raising  of  many  products,  and  that  these  islands 
would  become  the  important  center  of  American  activities  in  the 
Far  East,  which  they  deserve  to  l^e. 

SUMMARY. 

Summarizing  the  possibilities  for  electrochemical  activity  in 
South  American  and  Far  Eastern  countries,  it  would  seem  that 


30  C.  G.  SCHLUEDERBERG. 

greatest  immediate  progress  will  be  made  in  Japan,  the  west 
coast  of  South  America,  and  Brazil.  There  will  be  perhaps  more 
general  research  work  done  in  Japan  than  elsewhere,  although 
naturally  the  large  companies  operating  on  the  west  coast  of 
South  America  will  continue  research  along  the  lines  pertaining 
to  their  particular  operations.  Certain  countries  with  ample  power 
but  relatively  few  minerals,  such  as  Japan,  are  especially  suited 
to  the  conversion  of  the  raw  products  of  their  neighbors  into 
finished  materials  of  world-wide  application.  The  future  un- 
questionably holds  much  in  store  for  them. 

Unquestionably  the  aggressive  industrial  activity  of  Japan  will 
result  in  the  establishment  of  many  industries  throughout  that 
country,  in  w^hich  electrochemistry  and  the  electrochemist  will 
play  a  large  part.  There  can  be  no  question  that  the  Japanese  are 
alive  to  the  possibilities  of  electrochemical  development,  and  that 
their  activities  along  this  line  will  be  just  as  great  as  the  industrial 
and  financial  prosperity  of  the  country  will  permit.  However,  it 
must  be  kept  in  mind  that  the  opportunities  for  the  electrochemist 
in  Japan  will  loom  especially  large  for  the  Japanese  electro- 
chemist, while  in  the  other  countries  considered  it  is  likely  that 
Americans  and  Europeans  will  predominate. 

Richly  endowed,  as  many  of  these  countries  are,  with  both 
water-power  and  minerals,  one  cannot  travel  through  them  with- 
out being  impressed  by  the  tremendous  possibilities  for  the  devel- 
opment of  both  natural  and  economic  resources. 

It  must  be  remembered  that  things  move  more  slowly  than 
with  us,  and  that  it  takes  a  long  time  for  ideas  to  take  hold  and 
a  still  longer  time  for  definite  results  to  follow.  The  electrochem- 
ist working  in  these  countries  must  be  of  the  pioneering  type  and 
possessed  of  infinite  patience  and  perseverance,  but  once  progress 
is  initiated  along  sound  lines  the  opportunities  for  profit  present 
themselves  in  far  more  glowing  colors  than  is  usually  the  case 
in  our  own  or  European  countries.  The  way  is  difficult  and 
progress  slow,  but  the  possibility  of  reward  is  fully  commensurate 
with  the  efifort  involved. 


A  paper  presented  at  the  Forty-third 
General  Meeting  of  the  American  Elec- 
trochemical Society  held  in  New  York 
City,  May  3,  1923.  Dr.  Wm.  G.  Horsch 
in   the   Chair. 


NEWER  ASPECTS  OF  IONIZATION  PROBLEMS.' 

By  Hugh   S.   Taylor.'-' 

Abstract. 
A  resume  is  presented  of  some  recent  work  by  Born,  Fajans, 
Haber  and  others  on  the  problem  of  energy  changes  accompany- 
ing the  conversion  of  some  solid  crystalline  substances  and  of  the 
hydrogen  halides  into  dissolved  ions.  An  outline  is  given  of  the 
concepts  of  lattice  energy  and  of  the  electron  affinity  of  halogens, 
the  quantitative  side  of  the  problem  receiving  detailed  considera- 
tion. In  the  latter,  small  modifications  of  the  earlier  calculations 
have  been  made  whenever  newer  material  of  a  more  reliable 
nature  seemed  to  be  available.  Some  of  the  lines  along  which 
progress  may  be  anticipated  have  been  indicated. 


The  modern  electrochemist  must  survey  with  pride  the  back- 
ground which  his  predecessors  in  the  science  provided  for  the 
more  recent  advances  which  the  examination  of  the  physics  of 
the  atom  and  the  X-ray  spectra  of  elements  and  compounds  have 
achieved.  The  concept  of  ions  as  current  carriers,  of  the  existence 
of  free  ions  in  solution,  of  the  mobility  and  hydration  of  ions,  of 
potentials  due  to  ions  and  the  tendency  to  form  ions  were  all 
familiar  to  the  electrochemist  before  the  newer  ideas  of  electronics 
were  formulated,  and  they  materially  aided  the  rapidity  with  which 
the  newer  developments  found  ready  acceptance.  The  indebted- 
ness is,  however,  mutual.  In  the  light  of  modern  ideas  as  to  the 
structure  of  atoms  and  ions,  with  the  aid  of  the  quantitative  rela- 
tionships which  the  physicist  has  developed  between  the  various 
electrical  states  which  a  substance  may  achieve,  the  electrochemist 

'  Manuscript  received  February  14,   1923. 

*  Laboratory  of  Physical   Chemistry,   Princeton  University,   Princeton,   N.   J. 

31 


32  HUGH   S.   TAYLOR. 

can  recast  anew  his  own  ideas,  can  seek  new  methods  of  attack  on 
older  problems  and,  mayhap,  can  find  a  clearer  method  of  pre- 
sentation of  the  fundamentals  of  this  science. 

THE    CONCEPT    OF   LATTICE    ENERGY    AND   ITS   THERMOCHEMICAI. 

APPLICATIONS. 

The  pioneer  work  of  Lane  and  the  Braggs  on  the  X-ray  spectra 
of  elements  and  compounds  served  to  focus  attention  on  the  atoms 
as  the  essential  units  of  the  crystal  structure,  even  in  the  case  of 
compounds.  A  cubical  crystal  of  rock  salt  was  shown  to  consist 
of  alternate  sodium  and  chlorine,  in  three  dimensional  space  at 
the  corners  of  elementary  cubical  units,  each  sodium  spaced  2.81 
X  10"*  cm.  from  its  six  neighboring  chlorines,  each  chlorine  simi- 
larly spaced  from  six  sodiums.  The  distance  between  two  similar 
atoms,  twice  the  above  magnitude,  may  be  designated  as  the  "lattice 
constant,"  8. 

By  a  more  refined  X-ray  analysis,  Debye  and  Scheerer^  were 
able  to  show  that  only  a  fraction  of  this  distance  between  atoms 
was  actually  occupied  by  the  atoms,  that  the  electrons  surrounding 
the  nucleus  of  an  atom  were  concentrated  in  a  relatively  small 
space  around  the  nucleus,  of  radius  approximately  one  tenth  that 
of  the  lattice  constant.  Further  investigation  by  Born*  revealed 
that  the  units  of  such  a  crystal  were  electrically  charged,  carrying 
each  a  single  charge.  The  units  were,  in  fact  not  atoms  but  the 
respective  ions. 

A  detailed  analysis  of  the  attraction  existing  between  oppositely 
charged  ions  so  situated  in  space  and  repulsion  between  the  elec- 
trons comprising  the  outer  shells  of  such  ions,  led  to  the  conclu- 
sion that  the  attraction  varies,  normally,  inversely  as  the  square 
of  the  distances.  The  repulsion,  however,  in  the  case  of  simple 
cubic  structures,  such  as  sodium  and  potassium  chlorides,  was 
shown  to  vary  as  the  inverse  tenth  power,  or  the  potential  of  the 
repulsive  force  varies  as  the  inverse  ninth  power  of  the  distance 
between  the  ions. 

The  connection  between  such  attractive  and  repulsive  forces  on 
the  one  hand  and  the  compressibility  of  the  crystal  on  the  other, 
established  the  approximate  validity  of  the  conclusions  reached : 
the  cohesive  force  of  such  regular  crystals  is  purely  electrical  in 

»Phy-;ikal,  Zeitsch.   19,  474   0918). 

«  Ber.   Deut.  physikal.  Ges..  20,  210   (1918):   Ann.   Physik,    (T\')    61,  87    (1920). 


NEWER   ASPECTS    OF   IONIZATION    PROBLE^IS.  33 

its  origin.  It  should  be  observed  that  the  inverse  ninth  power 
relations  in  the  case  of  such  crystals  lead  directly  to  the  assump- 
tion of  a  cubical  atom  model,  such  as  is  now  familiar  from  the 
publications  of  Lewis, ^  Langmuir,®  and  Kossel." 

On  the  basis  of  these  assumptions  as  to  attraction  and  repulsion, 
Born  proceeded  to  the  calculation  of  the  electrostatic  work  neces- 
sary to  evaporate  one  mol.  of  the  crystal  into  free  gaseous  ions, 
that  is  to  say,  the  work  necessary  to  remove  the  ions  from  the 
positions  they  occupied  in  the  crystal  to  an  infinite  distance  from 
one  another.  Born  found  that  this  energ}^  was  expressible  by  an 
equation 

Na     n  —  1 
A  = 

S  n 

where 

N  is  the  Avogadro  number,  6.06  x  10-^ 

S  is  the  lattice  constant. 

a,  is  a  constant  characteristic  of  the  lattice  type  and  in  the  case 
of  the  alkali  halides  of  the  cubic  system  =  13.94  e-. 

e  =:  4.774  X  10'^^  electrostatic  units,  the  charge  of  the  electron. 

n  r=  9  for  the  alkali  halides  with  the  exception  of  lithium  salts, 
for  which  n  =  5. 

It  is  to  be  observed  that  the  cubical  arrangement  is  impossible 
for  the  lithium  ion,  since  it  has  only  two  electrons  in  the  outer 
shell.  Born  consequently  adopted  at  the  outset  the  lower  value 
n  =  5,  demanded  by  the  Bohr  electron  ring  system.  Typical 
results  obtained  in  this  way  for  a  variety  of  salts  are  set  forth 
in  the  following  table,  in  which  the  unit  of  energ}'  is  the  kilogram 
calorie. 

Salt 

A 


LiCl 

XaCl 

KCl 

XaBr 

KBr 

Xal 

KI 

179 

182 

163 

171 

155 

158 

144 

An  alternative  method  of  stating  these  energy  quantities  is  to 
regard  them  as  the  magnitude  of  the  free  energy  decrease,  when 
one  gram  molecule  of  the  crystalline  solid  is  formed  from  the 
gaseous  ions. 

(Li)^  -f-  (CI)-  =  [LiCl]  -f  Aj.^/^) 

=  J.  Am.   Chem.   Soc,  38,   762   (1916). 
•J.  Am.  Chem.   Soc,  41,  868.   1543    (1919). 
'Ann.   d.   Phys.,  49,   229   (1916). 

•Here   and   in   following   equations   parentheses,    (  ),    refer    to   gaseous    components, 
brackets,    f  1 ,  to  solid  substances. 


34  HUGH   S.   TAYLOR. 

An  approximate  test  of  the  accuracy  of  these  values  was  at- 
tempted by  Born  by  correlating  the  above  energy  quantities  with 
the  thermal  magnitudes  involved  in  a  reaction  of  the  type, 

[NaCl]  +  [KI]  =  [KCl]  +  [Nal] 

Fajans  ^  has  shown  that  in  the  case  of  all  these  halides  the  change 
in  total  energy  at  ordinary  temperatures,  U300,  is  approximately  = 
1.003 A.  For  all  practical  purposes,  therefore,  the  differences  in 
the  above  magnitudes.  A,  will  be  essentially  equal  to  the  differ- 
ences of  the  thermal  magnitudes,  U.  The  net  change  in  the  A 
values  for  the  above  compounds  can  therefore  be  equated  to  the 
net  heat  effect  of  the  above  reaction,  i.  e., 

A    =    ANaCl    +    AkI    ANal   AkcI    = 

QNaCl     +    QkI   QNal  —   QkCI 

where  Q  values  are  the  heats  of  formation  of  the  solid  salts 
from  the  elementary  components.  A  brief  test  will  show,  how- 
ever, that  such  heat  effects  are  small,  of  the  order  of  a  few 
kilogram  calories,  while  the  individual  values  are  of  the  order  of 
100  Cal. ;  so  far  as  the  test  went  it  was  favorable  to  the  A  values 
obtained  by  Born. 

A  more  accurate  test  was  devised  by  Fajans.®  By  substi- 
tuting the  heats  of  solution  of  the  several  salts  in  very  dilute 
solution  for  the  heats  of  formation  used  by  Born,  the  net  effect 
of  the  several  heats  of  solution  could  be  equated  to  the  net  value 
of  A.     Thus  in  the  reaction  given  above, 

[NaCl]   4-  aq  =  Na+    +  Cf     -f   H, 


aq 

+ 


[KI]       +  aq  -  K-      +  F       -f  H 


aq  '         aq 


[KCl]     +  aq  =  K+     -f  Cl"^  +  H 


[Nal]     -f  aq  =  Na,;  -f  F^     -f  H^ 

where  H  refers  in  each  case  to  the  heat  of  solution  of  the  solid 
salt,  to  yield  a  very  dilute  solution  in  which  it  may  be  assumed 
all  of  the  salt  is  dissociated.  Now,  since  the  net  result  as  regards 
ionic  content  is  the  same,  whether  the  solutions  be  made  up  from 
NaCl  and  KI  or  from  KCl  and  Nal,  it  follows  that 

H=:Hi-f-H2  —  H3—  H,  =  AA=rAUrrAQ 

»Ber.   Deut.  physikal.  Ges..   21,  542   (1919). 


NEWER  ASPECTS   OE   IONIZATION    PROBLEMS. 


35 


111  this  case  the  results  were  of  a  higher  order  of  accuracy,  since 
the  individual  H  values  were  small  as  compared  with  the  Q 
values  used  by  Born.  They  were  known  to  a  higher  degree  of 
accuracy,  and  hence  the  test  yielded  by  the  use  of  such  figures 
was  more  reliable.  Table  I  illustrates  the  agreement  obtained  in 
a  large  number  of  examples  studied. 

Table  I. 


Reaction 

A 

H 

KCl  +  LiBr    =  KBr  +  LiCl 
KCl  +  Lil       =  KI     +  LiCl 
KCl  +  NaBr  =  KBr  +  NaCl 
KCl  +  Nal      =  KI      +  NaCl 

+4 

+7 
+3 
+5 

+3.6 
+7.2 
+2 
+3.4 

The  Concept  of  Heat  of  Hydration  of  Gas  Ions. 
Fajans^"  further  pointed  out  that  the  heat  of  solution  of  such 
salts  may  be  regarded  as  composed  of  two  effects,  (a)  the  heat 
energy  required  to  convert  the  solid  salt  completely  into  free  gas 
ions  (i.  e.,  the  heat  equivalent  of  the  lattice  energy,  or  U  =: 
1.003A),  and  (b)  the  heat  of  solution  of  these  gas  ions  in  water. 
As  is  readily  seen  from  the  following  equation, 

(K)^  +  (CI)-  =  [KCl]  +  U 


[KCl]  +  aq  =  K;, 


4-  Cl;^  +  L 


(K)^  +  (CI)-  +  aq  =  Klq  +  Cl;^  +  (U  +-  L) 

this  heat  of  hydration  of  the  gaseous  ions  is  the  quantity 
(Q  +  L)  =  W.  Table  II  gives  a  summary  of  the  values 
obtained  for  the  heat  of  solution  W  of  the  gaseous  ions  of  a 
variety  of  salts,  as  compiled  by  Fajans  from  Born's  lattice 
energies  and  available  heats  of    solution  of  alkali  halide  salts. 


Table  II. 


Salt 

WCat  +  An 

Salt 

WCat  +  An 

Salt 

Wcat  +  An 

LiCl 
NaCl 
KCl 
RbCl 

187 

180.5 

159 

150 

LiBr 
NaBr 
KBr 
CsCI 

178 
171 

150 

151 

Lil 
Nal 
KI 
TlCl 

168 
159 
139 
159 

1"  Ber.   Deut.   physikal.  Ges.,   21,   549   (1919). 


36 


HUGH   S.   TAYLOR. 


A  check  on  these  results  was  readily  obtained,  since  it  is 
apparent  that  these  values  should  be  strictly  additive  quantities, 
dependent  on  the  values  for  the  individual  cations  and  anions  of 
the  several  salts.     Thus — 

(Na)+  +   (CI)-  +  Aq  =  Na+    +  CF    +  W^^+  ^,- 

(K)+     +   (CI)-  +  Aq  =  K+      +   Cr    +  W^+  c,- 


or     (Na)+  —  (K)+    +  Aq 


Na+  —  K+    +  W 

an  an  ' 


Na"^,  cr 


W  +      - 


W     +  —  W   + 


Now,  since  Wjj^+  —  ^k+  must  naturally  be  independent  of 

the  anions  associated  with  them  in  the  salts,  it  follows  that 

WNa+.  Br-   "   '^K+,  Br"    =    ^tC. 

In  this  way,  Table  III  was  obtained. 

Table  III. 


1              Cl 

Br 

I 

Mean 

Wl- 

-  Wj,^ 

+28 

+28 

+29 

+28 

WNa^ 

-  Wk- 

+21.5 

+21 

+20 

+21 

WRb^ 

-  Wk- 

—9 

—9 

Li 

Na 

K 

Wci- 

-    WBr- 

+9 

+9.5 

+9 

+9 

Wfir- 

-  Wj- 

+10 

+12 

+11 

+11 

As  is  evident,  the  differences  between  the  heats  of  hydration 
of  two  cations  or  two  anions  are  quite  definite  quantities,  and  are 
independent  of  the  ion  with  which  a  given  ion  is  associated.  The 
additivity  of  heats  of  hydration  is  thus  established,  and  the 
plausibility  of  the  lattice  energ}'  calculations  enhanced. 

Further  Refinements  in  Lattice  Energy  Calculations  and  a  Test 
of  the  More  Accurate  Data. 
Bom's  original  calculations  had  shown  that  the  calculation  of 
the  exponent  for  the  repulsive  force,  on  the  basis  of  compressi- 


NEWER  ASPECTS    OF    IONIZATION    PROBLEMS. 


37 


bility  data,  led  to  a  value  somewhat  smaller  for  sodium  than 
for  potassium  salts.  For  lithium  a  much  lower  exponent,  n  ==  5, 
was  used,  the  low  value  being  attributed  to  the  lack  of 
cubical  structure  in  the  lithium  ion.  Fajans  and  Herzfeld," 
accordingly,  have  recalculated  the  lattice  energies  of  a  series  of 
alkali  halides,  assuming  in  addition  to  a  repulsive  force  varying 
as  the  9th  power  other  terms,  involving  the  5th  and  7th  powers 
when  the  cations  and  anions  are  of  different  size.  The  lattice 
energies  so  obtained  are  set  forth  in  Table  IV,  the  older  values 
of  Born  being  enclosed  in  parentheses. 


Table  IV. 


F 

Cl 

Br 

I 

Xa 

210.4 
(220.3) 

170.0 
(181.6) 

159.7 
(171.6) 

146.7 
(158.3) 

K 

192.2 
(190.7) 

159.0 
(163) 

150.4 
(155.3) 

139.1 
(145.1) 

Rb 

154.6 
(155.5) 

146.5 
(148.7) 

135.8 
(139.5) 

The  corrections  throughout  are  greater  with  the  sodium  salts 
than  with  the  potassium  salts.  The  smaller  value  for  repulsive 
force  of  sodium,  as  found  by  Born,  receives  a  satisfactory 
explanation  in  the  newer  work.  As  before,  the  thermochemical 
test  of  these  newer  values  can  be  made  on  the  basis  of  additivity 
of  the  heats  of  hydration  of  the  gas  ions.  Table  V  gives  the 
results  of  such  a  test. 

Table  V. 


F 

Wp-  — 

Wci- 

Cl 

Wcr — 
Wsr- 

Br 
159.5 

Wer-  — 
Wi 

I 

Na  .  .  . 

209.8 

41.3 

168.5 

9.0 

11.6 

147.9 

WNa  + 

—  Wk+ 

14.0 

13.9 

142 

13.9 

K  .... 

195.8 

41.2 

154.6 

9.3 

145.3 

11.3 

134.0 

Wk+ 

-    WRb  + 

.  .  •  • 

.... 

4.8 

.... 

Rb  .. 

.... 

149.8 

.... 





I'Z.    Physik.   2,   309    (1920;. 


38  HUGH   S.  TAYLOR. 

The  means  of   the   several   differences  may  be  therefore  ex- 
pressed thus — 

W^^.    -  Wk+    =  14.0  Wp-      —  Wcr    =  41.2 

Wk+    -  WRb+  =    4.8  Wcr     -  Wer-    =    9.1 

Wer    -   Wj-    =    11.4 

Agreement  within  1  Cal.  is  obtained  in  each  case.  The  values 
obtained  by  Fajans  and  Herzfeld  by  this  refined  calculation  are, 
however,  regarded  by  Born  and  Gerlach^-  as  somewhat  too  low. 
Before  their  reasoning  can  be  adduced,  we  must  apply  the  con- 
cept of  lattice  energy  to  the  determination  of  the  electron  affinity 
of  halogen  atoms.  Before  passing  to  this  problem,  however,  we 
may  indicate  an  alternative  method,  independent  of  the  lattice 
theory,  of  testing  the  values  obtained  for  the  differences  of  the 
heats  of  hydration  of  the  various  gas  ions.  The  method  is  due  to 
Fajans^^  and  makes  use  of  various  thermochemical  data,  the  ioniza- 
tion potentials,  heats  of  sublimation  and  of  dissociation,  respec- 
tively of  potassium  and  hydrogen,  in  a  determination  of  the  heat 
hydration  difference  W  ~  —  Wk+.  The  method  is  in 
reality  the  application  of  Hess'  law  of  constant  heat  summation, 
employed  with  the  aid  of  the  following  equations: 

[K]  +     Aq  =  K-    -r  OH;^  -f    >^(H,)   +  48.1  (1) 

H:,  +  OH-^  -  Aq  +   13.6  (2) 

whence 

[K]  -I-  H;^  =    K^^  +   >^(H2)   +  62  Cal.  (3) 

now 

(K)  =  [K]  -I-  21  (4) 

(K^)    -f    0    ==  (K)   -f  99.  (5) 

>^(Ho)   =  (H)  -  45.  (6) 

(H)  =  (H")   +   0-312  (7) 

Whence  by  addition,  (3)  +  (4)  -f  (5)  -f  (6)  +  (7) 

(K-)   4-  H:^  =  (H+)   +  K;^  -  175  Cal.  (8) 

or  since 

(H+)   +  Aq  =  H-    +  Wj,. 


aq 


«Z.  Physik,  5,  435    (1921). 

'=  Eer.  Deut.  physikal,  Ges.,  20,  712   (1918). 


NEWER  ASPECTS   OF  IONIZATION    PROBLEMS.  39 

and 

(K+)   +  Aq  =  K;  +   W^+ 

the  equation   (8)   may  be  written 

Wjj+  —  Wg-+  =  175  Cal. 

Equations  (1)  and  (2)  are  the  ordinary  thermochemical  equa- 
tions. Equation  (4)  represents  the  heat  of  subHmation  of 
potassium,  (5)  the  ionization  potential  of  gaseous  potassium. 
Equation  (6)  gives  the  heat  of  dissociation  of  hydrogen  and  (7) 
the  ionization  potential  of  atomic  hydrogen.  All  these  several 
quantities  may  be  experimentally  determined,  though  the  order 
of  accuracy  is  not  as  yet  high  in  the  case  of  several.  Nevertheless 
the  calculation  goes  to  show  a  pronounced  energy  difference 
between  the  heat  of  hydration  of  gaseous  hydrogen  ions,  and 
that  of  the  gaseous  potassium  ion.  We  shall  return  to  a  dis- 
cussion of  this  magnitude  at  a  later  stage.  In  a  similar  manner 
and  similarly  independent  of  the  concept  and  calculations  of 
lattice  energy  Fajans  and  Sachtleben^*  obtained 


and 


WNa+  —  Wk+  =    16  ±  4   Cal. 


Wk+   —  WRb+  =  6  ±  4   Cal. 


These  values  stand  in  good  agreement  with  those  noted  previously 
as  derived  from  lattice  energy  calculations.  These  latter  there- 
fore may  be  given  a  reasonable  measure  of  confidence. 

THE  CONCEPT  OF  ELECTRON  AFFINITY  AND  ITS  MEASUREMENT. 

Theories  of  atomic  structure  have  familiarized  us  recently  with 
the  tendency  of  atoms  to  approach  the  rare  gas  type  of  structure, 
by  the  loss  or  gain  of  an  electron.  The  energy  changes  involved 
have  been  less  prominently  put  forward.  The  loss  of  a  valence 
electron  by  a  sodium  atom,  yields  a  sodium  ion  whose  outer 
system  of  electrons  is  that  of  the  neon  atom.  This  loss  of  an 
electron  is,  however,  an  energy  consuming  process,  the  energy 
involved  being  given  by  the  ionizing  potential  of  sodium  vapor, 
or  5.1  volts,  equivalent  to  a  heat  energy  input  of  118  Cal.  per 
gram  atom  of  sodium  vapor. 

"Cited  by  Fajans,  Z.  Physik.  2,  328   (1920). 


40  HUGH   S.   TAYLOR. 

In  a  similar  manner,  potassium  reverts  to  the  argon  type  with 
an  energy  expenditure  of  4.3  volts  or  a  heat  equivalent  of  99.6 
Cal.  At  the  other  end  of  the  groups  in  the  periodic  system  the 
halogen  atoms  display  a  tendency  to  add  an  electron  and  assume 
the  rare  gas  type  of  structure,  a  chloride  ion  being  similar  to 
argon,  bromide  ion  to  krypton,  iodide  ion  to  xenon,  fluoride  ion 
to  neon.  \\^hat  the  energy  change  involved  in  such  a  process  is, 
whether  positive  or  negative,  are  questions  to  which  no  direct 
method  of  determination  has  as  yet  been  able  to  provide  an  answer. 
Several  indirect  methods,  however,  serve  to  show  that  the  affinity 
of  a  chlorine  atom  for  an  electron  is  positive,  that  energy  is 
yielded  in  the  process  of  formation  of  a  negative  halide  ion  from 
a  neutral  halogen  atom,  or  conversely  that  energy  is  expended 
in  removing  an  electron  from  a  halide  ion. 

Born^^  and  Fajans^**  have  indicated  one  method  of  solution  of 
the  problem,  making  use  of  the  lattice  energy  calculations  previ- 
ously considered.  The  magnitude  of  the  electron  affinity  of 
chlorine  atoms  for  electrons  may  be  deduced  by  the  consider- 
ation of  two  methods,  whereby  solid  potassium  chloride  may  be 
converted  into  free  gaseous  ions,  potassium  and  chloride  ions. 
The  one  way  obviously  is  that  involving  the  lattice  energy  previ- 
ously discussed.  Let  us  assume  Bom's  first  calculation  of  this 
magnitude,  163  Cal.  The  alternative  method  consists  in  decom- 
posing the  solid  salt  into  its  electrically  neutral  constituents, 
metallic  potassium  and  gaseous  molecular  chlorine,  whereby  106 
Cal.  are  absorbed,  equal  to  the  heat  produced  when  solid  potas- 
sium chloride  is  produced  from  its  elements. 

To  obtain  the  gaseous  ions  from  the  elements  it  is  further 
necessary  to  vaporize  the  metal,  the  heat  absorbed  being  21  Cal., 
and  to  ionize  the  vapor  whereby  as  we  have  already  seen  a  further 
99  Cal.  are  required.  Similarly  the  molecular  chlorine  must  be 
dissociated  into  atoms,  the  heat  absorbed  being  31  Cal.  per  gram 
atom,  and  then  each  of  the  chlorine  atoms  attaches  itself  to  one 
of  the  electrons  set  free  by  ionization  of  the  potassium  vapor. 
This  last  step  involves  the  unknown  electron  affinity,  E,  of  the 
halogen  atom.  Since,  however,  we  have  arrived  at  the  same  end 
point  by  two  independent  paths  this  unknown  quantity  E  can  be 

"Ber.   Deut.  physikal.  Ges.   21,   679   (1919). 
"Ibid.,  21,  714  (1919J. 


NEWER   ASPECTS   OF   IONIZATION    PROBLEMS. 


41 


obtained  by  equating  the  energy  quantities  involved  in  the  two 
steps, 

(_163)  =  (-106)  +  (-21)  +  (-99)  +  (-31)  +  E 

whence  E  =  94  Cal. 

The  method  of  calculation  of  the  electron  affinities  of  bromine 
and  iodine  atoms  may  be  similarly  deduced  as  Table  VI  shows. 

Table  VI. 


[KX]  =  (K  )  -^  (X-) 

[K]  +  54(XO  =  [KX]    

(X)  =  H(XO    

(K)  =  [K]   

(K^)  -  0  =  (K)     

whence 

(X)  +  0  =  (X-) 


CI 


—163 

+106 
+31 
+21 
+99 

+94 


—155 
+99 
+23 
+21 
+99 

+87 


—144 

+87 
+18 

+21 
+99 

+81 


Note: — The  first  line  of  the  table  gives  Born's  original  values  for  lat- 
tice energies.  The  newer  data  of  Fajans  would  reduce  these  from  4  to 
6  units.  The  second  line  gives  the  normal  heats  of  reaction,  probably  with 
an  accuracy  of  ^  per  cent.  The  third  line  gives  the  heats  of  dissociation 
of  the  halogens.  These  are  less  certain.  The  value  chosen  by  the  writer 
for  CI2  is  a  mean  of  two  recent  determinations  (Trautz  Z.  anorg.  Chem., 
122,  81,  (1922),  Q  =  70  Cal.  per  mol. ;  Henglein,  Z.  anorg.  Chem.,  123, 
137,  (1922),  Q  =  54  Cal.)  The  mean  62  Cal.,  is  concordant  with  the 
value  calculated  by  Trautz  from  the  absorption  band  of  chlorine.  See, 
however,  V.  Halban  and  Siedentopf,  Z.  physik.  Chem.,  103,  85,  (1922), 
who  dispute  the  existence  of  such  a  band.  The  bromine  value  is  due  to 
Bodenstein,  Z.  Elektrochem.  22,  317.  (1916).  The  iodine  value  is  from 
Starck  and  Bodenstein,  Z.  Elektrochem.,  16.  961,  (1910).  The  fourth 
and  fifth  lines  are  respectively  the  heat  of  vaporization  and  the  ionization 
potential  of  potassium. 


Born  and  Fajans  both  tested  their  calculations  by  a  method, 
independent  of  the  lattice  energy  concept,  based  on  the  ionization 
potentials  of  the  hydrogen  halides.  Born  assumed  that  HCl 
ionizes  to  give  H*  +  CI,'  an  assumption  later  confirmed  experi- 
mentally by  Foote  and  Mohler,^^  who  determined  the  ionization 
potential  Juci^o  be  approximately    14.0  volts.     The   method  of 

IT  J.  .^m.  Chem.   Soc.  42,  1832   (1920). 

4 


42  HUGH   S.  TAYLOR. 

deriving  the  electron  affinity  of  the  halogen  is  conveniently  dem- 
onstrated by  the  following  diagrammatic  outline  due  to  Haber/^ 


(HCl)     < 

>^                     Qhci 

(H),  (CI) 

A 

Jhci 

Jh 

)+, 

Ecl 
(C1-)     < 

(H+),   (■ 

-).  (CI) 

(H) 

The  direction  of  the  arrows  corresponds  with  the  evolution  of 
heat.  Similar  diagrams  may  be  set  up  for  hydrogen  bromide, 
hydrogen  iodide  and  hydrogen  cyanide.  Now  Qhcp  which  is 
the  heat  of  formation  of  hydrogen  chloride  from  hydrogen 
and  chlorine  atoms,  also  involves  the  heats  of  dissociation  of 
corresponding  molecules.  For  the  halogens  we  shall  use  the 
values  previously  given.  For  hydrogen  we  shall  use  the  mean 
of  Langmuir's  value  and  that  of  Herzfeld/^  namely  90  Cal.  per 
mol.  For  the  ionization  of  hydrogen  atoms  the  generally  accepted 
value  is  310  Cal.  per  gram  atom  or  13.4  volts.  For  the  ionization 
potentials  of  the  halides  we  shall  use  the  measurements  of  P. 
Knipping,-°  as  amended  recently  by  Franck-^  and  Grimm.^^  The 
several  relationships  may  be  expressed  in  the  following  equations : 

5^(H,)   +  'AiCl)  =   (HCl)   +  22  Cal. 

(H)  =  y^   (H,)   +  45 

(CI)  =  y^   (CU)   -f  31 

(HCl)  =   (H^  +   (Cr)  —  313 

H*  +  0  =   (H)   +  310 

Whence,  by  summation  (CI)   -|-  0  =  (Cf)   +  Eci 

Where  E^,    r:r  22  -f  45  +  31  —  313  +  310 
or  Eci    :=  95  Cal. 

This  value  is  in  good  agreement  with  that  for  E^,j  obtained 
from  the  lattice  energy  calculations  of  Born.    The  errors  of  the 

"  Ber.  Deut.  pliysikal.   Ges.,  21,  754   (1919). 

i»Z.  Elektrochem.   25,  302   (1919). 

WZ.  Physik.  7,  328  (1921). 

"Z.  Physik.  11,  155   (1922). 

»Z.  phys.  Chem.   102,  504  (1922). 


NEWER  ASPECTS   OF  IONIZATION    PROBLEMS. 


43 


above  data  are  therefore  of  the  same  order  as  the  uncertainties 
of  the  lattice  energy  data.    Similarly 

Eg,  =  12.1  +  45+23  —  300  +310  =  90  Cal. 
Ei     =  1.5  +45  +18  —  290.5  +  310  =  84  Cal. 

The  accuracy  of  these  determinations  amounts  to  ±  10  Cal. 
since  there  is  an  uncertainty  in  the  ionization  potential  data,  esti- 
mated by  Knipping  at  ±  7  Cal.,  uncertainty  in  the  dissociation 
values  for  hydrogen  and  chlorine,  which  may  easily  amount  to  rt 
3  Cal.  Nevertheless,  both  the  lattice  energy  calculations  and 
those  based  on  ionization  potentials  serve  to  show  the  affinity  of 
halogen  atoms  for  electrons  is  a  large  positive  quantity. 

Fajans'-'"'  has  shown  that  the  evaluation  of  this  electron  affinity 
for  various  gases  has  a  definite  utility,  in  indicating  the  probable 
effect  of  electron  impacts  with  molecules  of  the  various  gases. 
Thus  in  the  case  of  the  halogens,  the  quantitative  data  already 
given  lead  to  the  following  equations. 


(X,)  =  (X)  +  (X) 

(X)  +  0  =  fx-) 

whence   (X,)  +  0  =  (X)  +  (X") 

similarly  (X.)  =  2(X) 

2X  +  20  :-  2(X-)   

whence  (XJ  +  20  =  2(X-)   


CI 

Br 

—62 

-^16     i 

+95 

+88 

^33 

+42 

—62 

—46     1 

+190 

+176    : 

+138 

+130     j 

—36  Cal. 

+82  Cal. 

+46  Cal. 

—36 
+164 
+128 


It  is  evident  therefore  that  the  production  either  of  atom  plus 
ion  or  two  ions  by  collision  of  electrons  with  chlorine  molecules 
are  both  strongly  exothermic  processes.  The  collision  of  a  slow 
moving  electron  with  a  chlorine  molecule  will  therefore  probably 
give  a  negative  ion  and  a  neutral  atom  in  the  sense  of  the  first 
set  of  equations.  If  two  electrons  collide  simultaneously,  two 
halogen  ions  will  result,  with  still  larger  evolution  of  heat.  The 
affinity  of  halogen  atoms  for  electrons  is  large  enough  to  split  the 
atomic  linkage  in  the  molecule.  The  existence  of  negative  halogen 
molecules  is  therefore  not  probable. 

One  further  method  of  measurement  of  electron  affinity  has 

"Ber.  Deut.   physikal.  Ges.,  21,   724   (1919). 


44-  HUGH   S.   TAYLOR. 

been  suggested  by  Franck-*  in  a  very  suggestive  paper  connect- 
ing affinity  with  spectroscopic  phenomena.  This  is  not  the  place 
to  ampHfy  the  subject.  Interested  readers  may  be  referred  to  its 
treatment  in  the  recent  monograph  by  Foote  and  Mohler/^  where 
additional  details  on  all  three  methods  of  computation  may  be 
obtained.  The  most  recent  spectroscopic  data  give-^  for  E^,,  , 
89.3  Cal.,  Eg^  =  67.5  Cal.  and  for  Ej,  59.2  Cal.,  values  therefore 
somewhat  lower  than  in  the  preceding. 

Electrode  Processes  and  the  Newer  Concepts. 

The  importance  attaching  to  the  heats  of  hydration  of  individual 
gas  ions  prompts  a  closer  inquiry  into  the  determination  of  these 
quantities.  Only  one  method  of  obtaining  these  seems  to  have 
been  indicated  in  the  literature.  This  method  is  based  on  the 
value  for  the  absolute  potentials  of  electrodes.  The  commonly 
accepted  value  for  the  single  potential  of  the  normal  calomel 
electrode  is  -}-  0.56  volt.  This  value  is  based  on  measurements 
involving  the  dropping  mercury  electrode.  Estimates  of  the 
accuracy  of  this  value  vary  widely.  By  some  it  is  regarded  as 
accurate  to  within  a  few  hundredths  of  a  volt.  Others,-^  however, 
claim  a  much  greater  error  than  this  amounting  to  some  tenths 
of  a  volt. 

Ostwald,  accepting  the  higher  accuracy  of  the  value  0.56  volt 
has  shown-^  that  with  the  use  of  this  value  the  heats  of  ionization 
of  various  elements  may  be  determined  approximately.  With  the 
aid  of  the  equation 

d    TT 


nF  TT  —  U  =  nFT 


dt 


using  for  tt  the  value  for  the  single  potential  of  a  given  electrode 

(Jtt 

and    for  its   temperature   coefficient,   Ostwald   and   Jahn-" 

determined  U  for  a  number  of  elementary  electrode  processes. 

"Z.  Physik.  5,  428    (1921). 

**  The    Origin    of    Spectra,    A.    C.    S.    Monograph    Series,    Chera".    Catalog    Company. 
1922,  Chap.  VIII. 

"Angerer,   Z.    Physik.    11,    169    (1922). 

»^Cf.  Garrison,  J.  Am.  Chem.  Soc.,  45,  37   (1923). 

i»Z.   physikal  Chem.,   11.  506   (1893). 

29  Z.  physikal.  Chem.,   18,  421    (1895). 


NEWER   ASPECTS   OF    IONIZATION    PROBLEMS.  .45 

M  =^  W 

aq 

For  the  reaction  at  the  hydrogen  electrode 

(H^j  pt  -  h:^, 

the  heat  of  reaction  was  shown  to  be  very  small,  and  of  the 
order  of  0  ±  1  Cal.  An  error  of  0.25  volt  in  the  determination 
of  the  absolute  potential  would  involve  a  corresponding  varia- 
tion of  0.25  X  23  =  ±6  Cal.  in  this  value  for  the  electrode  pro- 
cess, which  is  not  a  higher  order  of  error  than  is  inherent  in  the 
calculations  recorded  in  earlier  sections  of  this  paper.  Accepting 
Ostwald's  value  for  the  hydrogen  electrode  process,  Fajans  has 
associated  it  with  the  material  accumulated  by  him  with  reference 
to  the  combined  heats  of  hydration  of  cation  and  anion.  The  con- 
version of  molecular  hydrogen  to  hydrogen  ions  in  solution  takes 
place  in  a  cell  in  which  the  gas  is  bubbled  over  a  platinum  elec- 
trode. The  hydrogen  ions  pass  into  the  solution,  the  electrons 
remain  on  the  platinum  side  of  the  double  layer  at  the  junction  of 
electrode  and  electrolyte.  The  net  heat  change  of  this  electrode 
process,  0  ±  1  Cal.  is  composite  of  two  thermal  magnitudes,  that 
of  the  change  from  molecular  hydrogen  to  dissolved  hydrogen 
ions  and  that  associated  with  the  presence  of  electrons  in  the 
platinum  metal.  This  latter  is  equal  and  opposite  to  the  energy 
required  to  evaporate  electrons  from  the  metal,  a  quantity  which 
amounts,  according  to  Born,-^"  to  approximately  100  Cal.'' 

We  may  therefore  obtain  the  heat  of  hydration  of  gaseous 
hydrogen  ions  assuming  this  value  by  the  following  set  of 
equations : 

(H^)   -f   0  =    (H)   +  312  Cal. 

(H)  =   ^   (HJ   +  45  Cal. 

(H3)   -f  Pt    4-  aq  =  h;^  +  Pt(-)   +   0  Cal. 

Ft  (— )  =  Ft  +   0  —  100  Cal. 

Hence,  by  addition  (H+)  +  Aq  ^H^^  +  257  Cal. 
Now,  as  was  shown  in  an  earlier  section 


Wjj,  -  Wj,,  =   175  Cal. 


*•  Loc.   cit. 


3'  Most  varied  values  are  to  be  found  in  the  literature  for  this  magnitude  varying 
between  2.5  volts  for  platinum  containing  hydrogen  to  6.6  volts  (Langmuir)  in  which 
special  precautions  were  taken  to  ensure  the  absence  of  this  gas.  No  high  order  of 
accuracy   can   therefore   be   assigned   to   this   quantity. 


46.  HUGH   S.  TAYLOR. 


Hence  it  follows  that 

Wk. 

=  257  —  175  =  82  Cal, 

And,  since 

Wg.    +  W^i-    =  159 

Wj,,-  =  77  Cal. 

On  the  basis  of  such  a  procedure  Born'-  has  compiled  the 
following  table  of  individual  heats  of  hydration  of  gas  ions. 


H- 

Li^ 

Na^ 

K* 

Rb^ 

Cs- 

CI- 

Br 

r 

262 

110 

103 

82 

71 

74 

77 

68 

57 

The  hydrogen  value  in  this  table  is  slightly  different  from  that 
given  above  due  to  small  variations  in  the  values  used  for  the  heat 
of  dissociation  of  molecular  hydrogen.  Especially  noteworthy  is 
the  decrease  in  value  for  the  heat  of  hydration  with  increase  in 
the  size  of  the  ion. 

It  is  apposite  at  this  point  to  define  more  precisely  the  signifi- 
cance of  the  concept  of  hydration  of  gas  ions.  According  to 
Fajans  it  is  not  to  be  regarded  as  involving  the  solution  of  a 
gaseous  ion  in  the  water  with  the  formation  of  ion-hydrates  of 
definite  stoicheiometric  composition.  Rather  has  one  to  assume 
that,  through  the  charge  which  the  ion  carried,  the  oppositely 
charged  parts  of  the  polar  water  molecules  in  its  immediate 
neighborhood  are  oriented  towards  the  ion,  whereas  the  similarly 
charged  portions  of  the  water  molecules  are  turned  away  from 
the  ion,  and,  in  their  turn,  act  electrically  upon  the  molecules  in 
their  immediate  environment.  There  results,  therefore,  a  kind  of 
electric  polarization  in  the  solution.  In  the  preceding  table  there 
is  obviously  a  greater  heat  of  hydration  the  smaller  the  gas  ion. 
The  electrical  forces  are  operating  at  smaller  distances.  Born 
concludes  that  in  every  case,  irrespective  of  the  nature  of  the  gas 
ion  the  heat  of  hydration  of  the  gas  ion  will  be  a  positive  quantity. 

In  the  preceding  calculation  of  the  individual  value  for  the 
heat  of  hydration  of  the  hydrogen  gas  ion,  it  is  apparent  that  the 
calculation  involves  the  magnitude  of  the  heat  change  associated 
with  the  electron  emission  from  the  metal  used  as  the 
hydrogen   gas   electrode,   which   was    platinum   in    the   example 

^-  Loc.  cit. 


NEWER  ASPECTS   OF  IONIZATION    PROBLEMS.  47 

considered.  It  was  pointed  out  that  a  considerable  degree  of 
uncertainty  attaches  to  this  value.  It  would  therefore  be  well 
worthy  of  experimental  investigation  how  or  whether  the  char- 
acteristics of  the  hydrogen  electrode  change,  as  a  consequence  of 
alteration  in  the  metal  used  as  electrode  material.  Of  many  sub- 
stitutes considered,  tantalum  appears  to  us  to  offer  possibilities  of 
usefulness.  It  is  known  to  take  up  many  times  its  own  value  of 
hydrogen.  Furthermore,  its  thermionic  emission  has  been  most 
carefully  studied.  The  data  for  an  independent  check  on  the 
magnitude  of  the  individual  heats  of  hydration  should  therefore 
be  easily  obtained.     We  plan  to  obtain  these  if  possible. 

Meanwhile  we  must  regard  the  data  already  given  in  the  pre- 
ceding table  as  tentative.  The  corresponding  calculation  for  a 
potassium  electrode  does  not  yield  the  same  value  for  W^.^.  as  is 
obtained  indirectly  from  the  above  calculations,  as  the  following 
equations  demonstrate. 


[K] 

=  (K)    -  21 

(K) 

=  (KO  -f    e  -  99 

(K^) 

+  Aq      =  K;^      -f  W^. 

0 

-t-  K        =  K(— )  +  50.6 

Now  we  have  shown  (p.  290)  that 


And  since 


[K]  -F  h;^  =  k;^  +  y,(K,)  +  ez 


Hia  =   V2    (H,)   +  Aq  +  0 


it  follows  that  the  electrode  process  potassium  —  K*  or  alter- 
natively 2  [K]  =  k;^  +  K  (— )  +  62  Cal.  This  would  yield 
for  Wg^  a  value  ^132  Cal.,  deviating  most  markedly  from  that 
given  by  the  hydrogen  calculation.  The  diversity  between  the 
two  has  its  origin  in  the  two  equations 

Pt  +   e   =  Pt  (— )   -f  100 
and 

K  +   0   =  K  (— )   +  50.6 

This  latter  is  the  most  probable  value  from  determinations  of  the 
photoelectric  effect,  and  seems  equally  as  well  founded  as  the 
platinum  value  used  by  Born  and  Fajans. 


48  DISCUSSION. 

It  is  evident,  therefore,  that  the  problem  is  only  in  its  initial 
stages.  Much  work  remains  to  be  done,  much  progress  to  be 
made.  In  the  present  communication,  the  argument  has  been 
confined  solely  to  the  hydrogen  and  alkali  halides,  because  with 
the  aid  of  direct  measurements  of  ionization  potentials  of  the 
hydrogen  halideS,  and  with  the  readily  verified  calculations  of  lat- 
tice energy,  for  the  relatively  simple  alkali  halide  crystal  lattices 
a  surer  basis  for  calculations  existed.  The  treatment  is  being 
extended  to  other  compounds,  as  a  recent  attempt  by  Grimm^' 
indicates.  There  opens  up  a  new  field  of  investigative  work  which 
cannot  fail  to  have  its  influence  on  the  development  of  electio- 
chemical  science  in  general. 


DISCUSSION. 


vS.  C.  LiND^ :  Prof,  Taylor's  paper  is  extremely  interesting.  I 
had  frequently  been  tempted  to  undertake  a  similar  analysis  to 
that  of  Prof.  Taylor,  and  therefore  it  interests  me  all  the  more. 
At  the  time  that  many  of  us  became  members  of  the  Society,  we 
were  entirely  satisfied  with  the  electrolytic  pressure  theory,  or  the 
Nernst  theory,  of  what  happens  at  an  electrode.  It  has  been 
evident  to  many  of  us  for  some  time  that  it  would  be  extremely 
important  to  study  electrode  phenomena  from  the  standpoint  of 
gaseous  ionization.  One  of  the  difficulties  has  been  to  know 
whether  the  elctrolytic  ion  is  exactly  the  same  as  the  gaseous 
ion ;  and  that  is  one  of  the  assumptions  Prof.  Taylor  has  had  to 
make,  about  which  there  might  possibly  be  some  question. 

It  will  be  useful  in  the  future  to  use  these  conceptions  that  Prof. 
Taylor  has  brought  to  our  attention,  whether  they  ultimately 
prove  to  be  correct  or  not. 

John  Johnston^:  I  would  like  to  support  the  idea  Dr.  Taylor 
is  emphasizing,  namely,  that  the  usual  picture  of  the  process  of 
ionization  in  solution  is  not  satisfactory,  and  that,  so  far  as  I 
know,  no  one  has  outlined  a  satisfactory  picture  of  the  process  of 
forming  an  ion  at  the  electrode  in  solution. 

"Z.  physikal.  Chem.,  102,   113  and  504   (1922). 

'  Chief  Chemist,  U.   S.   Bureau  of  Mines,  Washington,  D.  C. 

'  Yale  University,  New  Haven,  Conn. 


NEWER  ASPECTS  OF  IONIZATION   PROBLEMS.  49 

S.  C.  LiND :  Those  of  us  who  have  believed  for  a  long  time  that 
gaseous  ions  might  be  chemically  active  have  met  opposition  on 
the  part  of  physicists,  who  have  pointed  out  that  in  the  case  of  the 
electrolytic  ions,  the  sodium  ion  and  the  chlorine  ion,  we  have  a 
special  case.  That,  for  example,  in  the  case  of  the  sodium  ion  with 
one  positive  charge,  and  the  chlorine  ion  with  one  negative  charge, 
under  the  Lewis- Langmuir  theory  leads  to  the  rare  gas  configura- 
tion, which  we  all  admit  to  be  inert.  In  other  words,  if  the  as- 
sumed electrolytic  sodium  ion  were  present  as  a  gaseous  ion,  you 
would  not  expect  it  to  be  chemically  active,  except  through  an 
electrical  attraction  for  the  electrical  opposite.  Therefore,  if  that 
is  true,  we  would  not  expect  the  sodium  ion  with  one  negative 
charge  to  be  active  toward  electrically  neutral  water,  nor  would 
we  expect  it  to  lead  to  a  reaction  with  high  heat  of  reaction.  I 
merely  want  to  ask  Prof,  Taylor  what  explanation  he  would  give 
of  that  objection  of  the  physicist.  It  is  not  one  that  I  am  raising 
at  all,  but  one  that  has  been  raised  to  theories  that  I  hold. 

H.  S.  Taylor  :  I  do  not  know  what  the  answer  to  such  an  ob- 
jection is.  The  chlorine  ion  is  certainly  a  peculiarly  stable  system. 
At  one  time  I  (along  with  Dr.  Lind  and  some  others  who  had  been 
working  in  the  field  before)  thought  that  if  I  could  succeed  in 
getting  the  chlorine  ion  in  a  hydrogen  chlorine  mixture  I  could  get 
a  reaction.  In  view  of  what  has  accumulated  with  regard  to  the 
nature  of  chlorine  ion,  I  myself  am  skeptical  now.  I  would  agree 
with  the  physicist  in  saying  that  the  chlorine  ion  is  something 
akin  to  a  noble  gas,  except  in  so  far  as  you  have  an  excess  nega- 
tive charge,  and  thereby  can  have  electrical  attraction  such  as  is 
present  in  solid  sodium  chloride.  This  is  one  of  the  problems 
that  certainly  needs  the  intense  co-operation  of  the  physicist  and 
the  electrochemist. 

W.  C.  Moore'  :  A  number  of  years  ago  I  was  interested  in 
gaseous  conduction,  particularly  with  reference  to  the  flaming  arc. 
I  looked  up  the  literature  on  the  subject  and  found  that  Prof.  H. 
A.  Wilson,  who  at  that  time  was  doing  considerable  work  on  con- 
duction in  gas  flames,  had  discovered  that  potassium  ion  in  po- 
tassium chloride  vapor  carried  three  positive  charges ;  whereas, 

*  Research    Chemist,    U.    S.    Industrial    Alcohol    Co.,    Baltimore,    Md. 

5 


50  DISCUSSION. 

we  suppose  we  know  that  it  carries  one  charge  only,  in  solutions 
in   water. 

There  is  an  interesting  discrepancy  here ;  we  need  some  means 
of  determining  how  the  number  of  charges  on  a  potassium  ion 
vary  in  raising  the  temperature  from  that  of  a  bunsen  burner  to 
that  of  the  flaming  arc. 

H.  C.  Howard*  :  The  same  objection  could  be  urged  against 
the  activity  of  the  electrolytic  potassium  ion,  because  that  also 
reverts  to  the  rare  gas  type. 

S.  C.  Lind:  It  is  correct  that  many  of  the  electrolytic  ions 
follow  a  special  class,  and  fall  into  the  rare  gas  series.  On  the 
other  hand,  it  does  not  follow  that  we  can  not  have  some  kind  of 
a  chlorine  negative  ion  as,  for  instance,  CI 2,  with  one  negative 
charge,  which  will  not  fall  into  that  class.  We  have  in  the 
gaseous  ions  a  much  wider  variety  in  nature  than  in  electrolytic 
ones,  and  we  should  not  be  hasty  to  conclude  that  there  is  no  such 
thing  as  a  chemically  active  gaseous  ion. 

H.  S.  Taylor  :  The  little  section  on  page  43  tends  to  show  that 
the  existence  of  negative  halogen  molecules  is  not  probable.  I 
think  the  evidence  on  that  point  is  fairly  conclusive,  since  the 
magnitudes  of  the  heat  quantities  involved  are  so  tremendously 
large. 

*  Princeton,  N.  J. 


A  paper  presented  at  the  Forty-third 
General  Meeting  of  the  American  Elec- 
trochemical Society  held  in  Nezv  York 
City,  May  3,  1923,  Dr.  Wm.  G.  Horsch 
in   the   Chair. 


OXYGEN  OVERVOLTAGE  OF  ARTIFICIAL  MAGNETITE  IN 
CHLORATE  SOLUTIONS.' 

By  H.  C.  Howard." 

Abstract. 

Attempts  were  made  to  oxidize  sodium  chlorate  to  perchlorate 
electrolytically  at  a  magnetite  anode.  Negative  results  were 
obtained.  The  oxygen  overvoltage  of  a  magnetite  anode  in  N 
sodium  chlorate  was  measured  and  found  to  be  from  0.4  to  0.6 
volt  lower  than  that  of  smooth  platinum. 


Several  years  ago,  in  the  course  of  a  study  of  the  electrolytic 
oxidation  of  sodium  chlorate  to  perchlorate,  some  time  was 
devoted  to  an  attempt  to  find  a  substitute  for  the  expensive 
platinum  anodes  usually  employed. 

All  of  the  common  and  many  of  the  rarer  metals  were  tried 
and  all,  except  those  of  the  platinum  group,  were  found  to 
corrode  very  rapidly  when  used  as  anodes  in  a  sodium  chlorate 
electrolyte.  Carborundum  and  the  various  high  silicon  alloys 
were  shown  to  be  valueless  and  the  oxide  electrodes,  such  as 
lead  peroxide  and  manganese  dioxide,  which  have  been  used 
effectively  as  insoluble  anodes  in  certain  cases,  decomposed  very 
quickly  under  the  conditions  present  in  this   electrolysis. 

It  was  known  that  an  artificial  magnetite  had  been  used  with 
success  in  some  of  the  German  alkali-chlorine  cells,  and  we  were 
anxious  to  test  this  material.  Finally  we  obtained  samples  of 
such  electrodes,  through  the  courtesy  of  the  Chile  Exploration 
Co.      These    artificial    magnetite    electrodes    proved    to   be    very 

'  Manuscript  received  February  2,   1923. 

^  Contribution   from  the  Chemical  Laboratory  of  Princeton   University. 

51 


52 


H.    C.    HOWARD. 


resistant  to  corrosion,  and  in  this  respect  appeared  to  offer  a  good 
substitute  for  platinum.  Analysis  of  the  electrode  in  which  these 
anodes  had  been  tested  showed,  however,  that  no  perchlorate  had 
been  formed  during  the  electrolysis,  and  this  was  found  to  be  the 
case  in  all  later  experiments,  even  under  the  most  favorable  con- 
ditions for  perchlorate  formation,  such  as  low  temperature  and 
high  current  density. 

At  the  time,  this  was  explained  by  assuming  that  the  over- 
voltage  of  oxygen  at  magnetite  is  very  much  lower  than  at  smooth 

Table  I. 
Oxygen  Overvoltage  of  Magnetite  in  Sodium  Chlorate. 

Electrolj'te,  N  sodium  chlorate.     Temperature,  20°  C.     Area  of  the  anode 

was  36  sq.  mm.  in  each  case.     The  potentials  are  referred  to 

-V  calomel  electrode  as  zero,  and  are  all  positive. 


Smooth    P] 

atinum  Anode 

Magnetite    Anode 

anode 

c.    d. 
amp./sq.    dm. 

potential 

V. 

c.  d. 
amp./sq.    dm. 

potential 

1                 ^• 

magnetite 
potential* 

3.0 

2.04 

1.4 

1.58 

1.58 

S.S 

2.10 

2.8 

1.61 

1.61 

8.3 

2.17 

5.0 

i            1.66 

1.65 

11.0 

2.23 

11.0 

1           1.72 

1.69 

22.3 

2.46 

16.6 

1.79 

1.75 

22.8 

1.86 

1.80 

•  Potential  of  the  magnetite,  corrected  for  the  voltage  drop  in  the  electrode  itself. 
The  resistance  of  the  electrode  and  contact  was  0.7S  ohm.  The  resistance  of  the  plati- 
num electrode  was  negligible. 

platinum,  and  hence,  at  anodes  of  the  former  material,  oxygen 
evolution  takes  place  in  preference  to  the  oxidation  of  the  chlorate 
ion  to  the  perchlorate. 

A  search  of  the  literature  revealed  no  data  on  the  oxygen  over- 
voltage  of  magnetite,  and  since  lack  of  time  prevented  further 
experimental  work,  a  test  of  the  explanation  offered  was  not  then 
possible. 

Recently  a  few  measurements  of  the  oxygen  overvoltage  of 
magnetite  in  sodium  chlorate  have  been  made  in  this  laboratory. 

The  results  of  these  measurements  are  presented  in  Table  I 
and  Fig.  1. 


OXYGEN    OVERVOLTAGE    OF    MAGNETITE. 


53 


These  data  and  curves  show  clearly  that  the  oxygen  overvoltage 
of  magnetite  is  much  lower  than  that  of  smooth  platmum. 


;♦       .b        16        2<»       ^^        2f 


Fig. 


1.    Potentials  of  platir.um  and  magnetite  in  N  sodium  chlorate. 


CONCLUSION. 

The  failure  to  oxidize  chlorates  to  perchlorates  at  a  magnetite 
anode  together  with  the  fact  that  such  an  anode  has  been  shown 
to  hav'e  a  much  lower  oxygen  overvoltage  than  a  smooth  platmum 
one  at  which  such  an  oxidation  takes  place  readily,  afford  further 
conkrmation  of  the  hypothesis  that  there  is  a  direct  relationship 
between  the  overvoltage  of  an  electrode  and  its  oxidizing  or 
reducing  power. 

DISCUSSION. 
Colin  G  Fink^  :  Mr.  Howard's  paper  is  interesting  and  brings 
up  the  general  subject  of  the  insoluble  anode.    In  electrolytes  such 
as  we  have  studied,  particularly  with  SO,  ions  present,  a  number 

1  Consulting  Metallurgist,  New  York   City. 


54  DISCUSSION. 

of  reactions  occur.  The  ultimate  anode  reaction  is  the  liberation 
of  oxygen  gas.  Now  anything  that  will  hasten  the  evolution  of 
oxygen  gas  in  preference  to  the  dissolution  of  the  metal  of  the 
anode,  will  cut  down  the  corrosion  of  the  anode  under  investiga- 
tion. In  other  words,  you  finally  come  to  a  point  where  you  can 
use  a  very  soluble  anode,  providing  you  have  on  the  surface  a 
thin  film  of  a  catalyzer,  which  will  hasten  the  discharge  of  the 
SO4  ions  and  the  formation  of  oxygen  gas  in  preference  to  the 
formation  of  metal  compounds. 

In  other  words,  the  overvoltage  phase  of  the  insoluble  anode 
is  a  phase  which  has  not  always  been  taken  into  account,  because 
primarily  the  metals  have  been  studied  from  the  purely  chemical 
solubility  point  of  view. 

M.  KnobeL"  :  Regarding  the  relation  between  oxidizing  or 
reducing  power  and  overvoltage,  while  one  can  find  a  good  many 
cases  in  the  literature  where  a  high  overvoltage  metal  does  give  a 
greater  oxidation  or  reduction  than  a  low  overvoltage  metal,  one 
can  also  find  as  many  cases  where  this  relation  does  not  hold. 

W.  G.  HoRSCH" :  Since  Dr.  Bancroft  is  not  here,  it  may  be 
safe  to  quote  him  as  stating  that  oxidation  at  an  anode  may  be 
linked  up  with  high  overvoltage,  but  is  not  necessarily  a  conse- 
quence thereof. 

I  took  the  liberty  to  subtract  the  reversible  potential  of  the 
calomel-oxygen  cell  from  Mr.  Howard's  results  and  compare  the 
overvoltages  thus  obtained  with  those  of  Dr.  Knobel  in  the  next 
paper  on  our  program,  and  I  get  0.3  to  0.4  of  a  volt  difference  at 
practically  all  points  in  the  curve.  The  curves  as  determined  by 
these  two  authors  thus  show  good  agreement  as  to  shape. 

H.  C.  Howard:  My  results  are  referred  to  the  calomel  elec- 
trode, as  zero.  The  potential  of  the  calomel  electrode  has  already 
been  subtracted.  Whereas,  Dr.  Knobel's  results  represent  real 
overvoltages.  You  would  obtain  more  nearly  comparative  values 
if  you  subtracted  the  reversible  potential  of  oxygen  from  my 
results. 

W.  G.  Horsch:  What  I  meant  was  platinum  in  terms  of 
oxygen. 

*  Mass.  Inst,  of  Technology,  Cambridge,  Mass. 
«  Chile  Exploration  Labs.,  New  York  City. 


A  paper  presented  at  the  Forty-third 
General  Meeting  of  the  American  Elec- 
trochemical Society  held  in  New  York 
City,  May  3,  1923,  Dr.  JVm.  G.  Horsch 
in   thi   Chair. 


THE  EFFECT  OF  CURRENT  DENSITY  ON  OVERVOLTAGE/ 

By  M.  Knobel.  P.  Caplan,  and  M.  Eiseman' 
INTRODUCTION. 

There  are  numerous  references^  in  the  literature  on  the  effect  of 
current  density  on  overvoltage,  but  they  are  in  general  more  or  less 
isolated  values  and  for  comparatively  small  current  densities. 
The  experimental  conditions  are  so  different  also  that  it  would  be 
difficult  to  compile  a  comparable  set  of  data.  On  account  of  the 
great  technical  importance  of  this  phase  of  overvoltage,  and  also 
for  the  theoretical  interpretation  of  overvoltage,  it  was  thought 
desirable  to  have  extensive  and  consistent  data  in  this  field.  In 
the  following  work  we  have  attempted  to  include  all  the  more 
common  metals  and  alloys  as  cathodes,  and  to  determine  oxygen 
and  halogen  overvoltages  on  as  many  electrodes  as  possible. 
While  the  overvoltage  values  obtained  may  not  be  acceptable  as 
absolute  values,  they  should  at  least  be  comparable  as  the  experi- 
mental conditions  were  maintained  the  same  in  all  cases. 

METHOD  OF   MEASUREMENT. 

W'e  have  accepted  as  our  definition  of  overvoltage  "the  poten- 
tial necessary  in  excess  of  the  reversible  potential  to  discharge  the 
product  in  question,  both  potentials  being  measured  under  identical 
conditions  as  external  hydrogen  pressure,  temperature  and  con- 
centration of  solution."  Thus  the  hydrogen  overvoltage  on  a  lead 

'  Manuscript  received  November  4,  1922. 

^  Contribution  from  the  Rogers  Laboratory  of  Physics,  Electrochemical  Laboratory, 
Massachusetts  Institute  of  Technology. 

'Tafel.  Z.  Physik.  Chem.  50,  641  (1904);  Ghosh,  J.  Am.  Cham.  Soc.  36,  2333 
(1914);  37,  733  (191S);  Rideal,  J.  Am.  Chem.  Soc.  42,  94  (1920);  Newbery,  J.  Am. 
Chem.  Soc.  109,  1051,  1066  (1916);  Sacerdotti,  Z.  Elektrochem.  17,  473  (1911); 
Tainton,  Trans.  Am.  Electrochem.  Soc.  41,  389  (1922);  Reichinstein,  Z.  Elektrochem. 
17,  85  (1911);  Coehn  &  Osaka,  Z.  anorg.  Chem.  34,  86-102  (1903);  Foerster  & 
Yamasaki,  Z.  Elektrochem.  16,  321  (1910);  Bennewitz,  Z.  Physik.  Chem.  72,  202 
0910);  Lewis  &  Jackson,  Z.  Physik.  Chem.  56,  193  (1906);  Coehn  &  Dannenberg,  Z. 
Physik.  Chem.  38,  609  (1901);  Gockel,  Z.  Physik.  Chem.  32,  607  (1900);  Niitton  & 
Law,   Trans.   Far.    Soc.   3,   50    (1907). 

55 


56  M.  KNOBEL,  P.  CAPLAN,  AND  M.  EISEMAN. 

cathode  at  a  given  current  density  will  be  the  potential  difference 
between  that  lead  cathode  and  the  solution,  minus  the  potential 
difference  between  a  reversible  electrode  (practically  platinized 
platinum  with  no  current  flowing)  and  the  same  solution  at  the 
same  temperature  and  pressure.  We  believe  this  to  be  the 
generally  accepted  definition. 

There  are  two  methods  of  measuring  overvoltage,  and  the  ques- 
tion arises  as  to  which  gives  the  overvoltage  just  defined.  The 
first  is  to  insert  a  reference  electrode  with  the  tip  against  the 
cathode  and  measure  the  electrode  potential  while  the  current  is 
flowing.  The  second  or  commutator  method,  which  has  been 
championed  principally  by  Newbery*  on  the  other  hand,  allows 
for  shutting  off  the  electrolyzing  current  while  the  electrode 
potential  measurement  is  being  made;  it  alternately  allows  the 
electrolyzing  current  to  pass  and  then  connects  the  cell  to  the 
potentionmeter.  The  main  argument  for  use  of  the  commutator 
is  that  all  ohmic  resistance  drops  are  eliminated ;  but  let  us  defer 
the  discussion  of  this  point  until  we  have  analyzed  the  commu- 
tator method  to  see  whether  it  gives  correct  values. 

The  potential  measured  in  the  commutator  method  depends  on 
the  concentration  of  the  electrode  products  stored  up  during  the 
period  of  electrolysis.  The  curves  determined  by  LeBlanc^  with 
an  oscillograph  and  a  commutator  throw  light  on  this  point.  The 
t}'pical  curves  obtained  by  him  for  the  variation  of  electromotive 
force  across  the  cell  (ordinates)  with  time  (abscissae)  are  shown 
in  Fig.  1,  2  and  3.  Fig.  1  is  for  the  electrolysis  of  a  normal  iodine 
and  potassium  iodide  solution  between  platinum  electrodes ;  Fig. 
2  is  for  0.05N  iodine  and  potassium  iodide  in  one  normal  sul- 
furic acid,  and  Fig.  3  for  one  normal  sulfuric  acid.  In  all  of 
these  curves  the  portions  A  are  for  the  time  when  the  electrolyzing 
current  is  on,  portions  B  when  the  current  is  shut  off  and  the 
oscillograph  only  is  connected  to  the  cell;  and  portions  C  when 
the  electrolyzing  current  is  passing  in  the  reverse  direction.  A 
difference  in  LeBlanc's  procedure  and  Newbery's  must  be  noted 
in  that  the  electrode  products  must  supply  current  to  operate  the 
oscillograph  in  LeBlanc's  arrangement  when  the  outside  current 

<  Trans.  Far.   Soc.   15   (1919") ;  J.  Am.   Chem.   Soc.  42,   2007    (1920). 
»  "Die    Elektromotorischen    Kraitte   der    Polarization    und    ihre    Messungen    mit    Hilfe 
Jes  Oszillographen"  Hall,  1910. 


EFFECT  OF   CURRENT  DENSITY   ON   OVERVOLTAGE. 


57 


is  shut  off,  while  Newbery  takes  no  current  from  the  cell  in  this 
interval. 

Fig.  1  indicates  that  no  polarization  has  occurred,  neither  in  the 
nature  of  overvoltage  nor  concentration  polarization  in  the  solu- 
tion. The  fact  that  portion  B  is  directly  on  the  zero  potential  axis 
means  that  the  two  electrodes  are  in  the  same  condition,  that  is, 
have  the  same  potential  difference  with  respect  to  the  solution, 
and  the  flatness  of  portion  A  shows  that  the  electrolysis  is 
occurring  at  constant  potential.  The  distance  of  A  and  C  from 
the  axis  is  presumably  due  to  ohmic  resistance  drop  in  the  whole 
cell.  This  electrolysis  is  therefore  reversible  as  far  as  the  elec- 
trodes are  concerned. 


A 

B 

B 

C 

Fig.  1 


Fig.  2 


Fig.  3 


Fig.  2  indicates  the  existence  of  some  polarization  due  to 
accumulation  of  electrode  products,  as  hydrogen  and  oxygen,  or 
concentration  differences  in  the  electrodes.  This  curve  is  typical  of 
all  LeBlanc's  measurements  on  oxidation  reduction  cells  such 
as  the  ferri-ferro  ion  electrode,  etc.  The  form  of  Fig.  3  is 
without  doubt  caused  by  the  accumulation  of  Hg  and  O,  on  the 
electrodes.  During  the  time  represented  by  A  the  electromotive 
force  gradually  increases  as  the  gas  concentration  increases.  The 
maintenance  of  the  potential  at  B  has  its  source  in  the  gases 
at  the  electrodes  yielding  current  by  going  back  into  solution. 


eg  M.   KNOBEL,  P.  CAPLAN,  AND  M.  EISEMAN. 

It  is  the  electromotive  force  represented  by  B  that  the  com- 
mutator method  should  measure.  For  the  low  current  densities 
used  by  LeBlanc  in  his  work  (about  0.0045  ampere  per  sq.  cm.) 
the  commutator  should  give  essentially  correct  results.  If  no 
current  were  taken  from  the  cell  during  the  time  B,  that  portion 
of  the  curve  would  probably  be  more  nearly  horizontal,  which 
of  course  it  should  be,  to  allow  accurate  measurement  on  a  poten- 
tiometer, and  to  have  a  definite  meaning. 

In  consideration  of  the  very  small  time  interval  (about  0.019 
second)  when  the  electrolyzing  current  is  on  and  ofif,  and  of  the 
small  current  used,  the  gases  liberated  at  the  electrodes  cannot 
attain  high  pressures  and  will  not  tend  to  diffuse  away 
appreciably.  They  should  then  have  essentially  the  same  concen- 
tration as  when  the  current  is  passing,  and  B  should  give  the 
back  electromotive  force  or  the  overvoltage  (of  both  electrodes) 
according  to  the  definition  previously  given.  In  support  of  this 
reasoning  is  the  fact  that  at  low  current  densities  Newbery's 
values  obtained  by  the  commutator,  are  the  same  within  the  limits 
of  reproducibility  of  overvoltage,  as  those  obtained  by  the  direct 
method. 

However,  at  high  current  densities  and  with  electrodes  other 
than  platinum  the  above  relations  cannot  hold.  It  is  well  known 
that  platinum  has  a  much  greater  power  to  occlude  or  adsorb 
gases  than  other  metals.  At  the  opposite  extreme  is  mercury 
which  probably  adsorbs  only  extremely  small  quantities  of  gas. 
The  gas  accumulation  at  a  mercury  electrode  must  then  occur  in 
a  laver  of  solution  under  which  conditions  the  gas  may  easily  be 
carried  away  by  convection  or  diffusion.  A  curve  analogous  to 
Fig.  3  for  mercury  electrodes  would  show  a  sharp  drop  in  the 
portion  B  and  the  overvoltage  measured  by  the  potentiometer 
would  be  much  lower  than  the  back  electromotive  force  when 
the  current  was  passing.  Newbery  in  fact  gives  values  for  mer- 
cury overvoltage  at  low  current  densities,  obtained  by  the  commu- 
tator method,  considerably  lower  than  those  obtained  while  the 
current  is  passing. 

At  high  current  densities  the  stirring  effect  of  the  evolved  gases 
will  also  cause  portion  B  to  drop  sharply  from  its  maximum.  For 
large  currents  the  gas   pressure  is  comparatively  much   larger, 


EFFECT   OF   CURRENT  DENSITY   ON   OVERVOLTAGE.  59 

which  in  itself  will  tend  to  increase  the  loss  of  gas  by  diffusion. 
Of  probably  much  greater  importance  however  is  the  violent 
stirring  of  the  solution  directly  at  the  electrode  surface  by  the 
evolved  gas.  Most  of  the  gas  which  is  in  the  solution,  possibly 
in  a  super-saturated  state,  will  be  swept  away  so  that  it  is  no 
longer  in  contact  with  the  electrode.  Again  this  explanation  is 
supported  by  the  very  low  values  obtained  by  Newbery  at  high 
current  densities,  at  mercury  as  well  as  at  other  electrodes. 

The  time  interval  in  which  this  must  occur  is  small.  For  the 
speed  of  2,500  revolutions  per  minute  of  the  commutator  as  used 
by  Newbery  the  current  is  broken  only  0.012  second,  but 
LeBlanc's  time  interval  is  0.019  second,  and  an  appreciable  drop 
has  occurred  in  the  case  in  Fig.  3  where  the  drop  is  least  to  be 
expected.  Some  of  these  points  have  been  tested  experimentally 
recently  by  Tartar  &  Keyes"  and  all  their  results  directly  confirm 
the  conclusions  drawn  here.  Other  investigators^  have  criticized 
this  commutator  method,  but  we  will  not  attempt  to  discuss  their 
criticisms  here.  We  believe  the  method  can  fairly  be  rejected. 
The  extensive  tables  of  Newber}'^  are  of  little  value  if  the  objec- 
tions to  the  commutator  method  are  valid. 

The  question  of  eliminating  ohmic  resistance  drop  in  the 
closed  circuit  method  is  a  serious  one.  Obviously  the  reference 
electrode  tip  cannot  be  situated  any  large  distance  from  the 
electrode  surface,  or  the  potential  drop  in  the  solution  will  be 
measured  with  the  over  voltage.  One  method,  which  unfor- 
tunately has  been  rather  widely  used  in  an  attempt  to  obviate 
this  difficulty®,  is  to  place  the  reference  electrode  behind  the 
cathode,  that  is,  on  the  opposite  side  from  the  anode.  This  is 
obviously  in  error  for  the  current  density  is  indefinite  and  much 
smaller  on  the  back  face  of  the  cathode  and  the  potential  so  meas- 
ured bears  no  relation  to  the  potential  difference  between  the 
electrode  and  the  solution  in  contact  with  the  front  face.  While 
the  electric  potential  of  the  whole  electrode  is  the  same,  the  solu- 
tion in  front  of  and  behind  the  electrode  need  by  no  means  have 

•J.  Am.  Chem.  Soc.  44,  557  (1922). 

•  Tainton,  Trans.  Am.  Electrochem.  Soc.  41,  389  (1922);  Maclnnes,  T.  Am.  Chem. 
Soc.  42,   2233    (1920). 

«J.    Chem.   Soc.    109,   1051,    1066    (1916)    111,   470    (1917). 

*  See  for  example  Nutton  and  Law,  Trans.  Far.  Soc.  3,  SO  (1917);  Pring  and  Curzon 
Ibid.  7,  237   (1911). 


6o 


M.   KNOBEL,  P.  CAPLAN,  AND  M.  EISEMAN. 


the  same  electric  potential,  and  therefore  the  potential  difference 
between  electrode  and  electrolyte  will  be  different  on  the  two  sides. 
While  seemingly  unnecessary  we  have  tested  this  point  experi- 
mentally and  confirmed  the  statement  made. 

If  the  reference  electrode  tip  is  placed  on  the  front  side  it  dis- 
turbs the  current  flow  lines  in  the  small  region  near  the  tip.     A 


1  to 

1 

tea 

^ 

^ 

^ 

^ 

-^ 

^ 

> 
3 

,--- 

-4 

•^ 

^ 

^+"» 
1^^^ 

!> 

^ 

^ 

'^ 

Y^ 

^ 

■» 

S 

/" 

yy 

^ 

^^ 

V 
2.^, 

/ 

y 

^ 

9 

1 

1  r 

/ 

y 

1 

/ 

r 

/ 

/ 

1 7a 

V 

i 

2  76 
/•75 

as  to  ;-s  t-o  i-&  Od 

Distance    from    electrode    surface    in   mm. 

Fig.    4.     Showing   effect   of  varying   size   of  electrode   tips. 


large  tip  or  one  pressed  too  closely  to  the  surface  would  cause 
an  appreciable  decrease  in  current  density  in  the  electrolyte 
immediately  between  the  tip  and  the  electrode,  and  too  low  over- 
voltage  values  will  result.  It  would  appear  that  the  smallest 
possible  tip  would  be  desirable.  This  was  tested  out  experi- 
mentally by  a  series  of  tips  of  different  diameters.  They  were 
moved  up  to  an  electrode  from  some  distance  out  in  the  solution 


EFFECT  OF  CURRENT  DENSITY  ON  OVERVOLTAGE.       6 1 

and  the  potential  plotted  as  a  function  of  the  distance.  As 
uniform  a  current  density  as  possible  of  0.5  ampere  per  square 
centimeter  was  maintained. 

An  auxiliary  reference  electrode  held  at  a  constant  position 
with  respect  to  the  lead  cathode  used,  indicated  the  constancy 
of  the  latter.  The  results  are  shown  in  Fig.  4.  The  ordinate 
scale  is  the  electrode  potential  in  volts  with  an  arbitrary  zero. 
The  abscissae  indicate  the  distance  of  the  tip  from  the  electrode 
surface,  measured  in  mm.  The  tip  sizes  corresponding  to  the 
numbers  on  the  curves  are  as  follows: — No.  1,  4.6  mm.  in 
diameter ;  No.  2,  3.1 ;  No.  3,  2.3  ;  No.  4,  1.3 ;  No.  5,  0.08.  The  slope 
of  the  curves  is  of  course  due  to  the  resistance  of  the  electrolyte. 
For  the  larger  tips  the  drop  in  potential  in  excess  of  the  ohmic 
resistance  drop,  as  the  tip  approaches  the  electrode,  is  marked. 
The  use  of  a  tip  as  large  as  the  first,  pressed  against  the  electrode 
would  introduce  an  error  as  large  as  0.04.  If  the  tip  is  one  milli- 
meter or  less  in  diameter,  however,  we  concluded  it  would  give 
essentially  correct  values  practically  independent  of  the  tip 
diameter. 

Another  experiment  with  a  mercury  cathode  proves  conclu- 
sively the  lowering  of  the  potential  measured  by  the  lowering  of 
the  current  density  in  the  above  manner.  The  tip  was  lowered 
from  a  point  out  in  the  solution  until  it  was  some  distance  under 
the  mercury  surface.  The  electrode  potential  decreased  at  a 
constant  rate,  due  to  the  change  in  the  ohmic  resistance  in  the 
electrolyte  until  the  tip  had  made  a  slight  depression  in  the 
mercury.  When  the  tip  was  pushed  still  further  into  the  mer- 
cury, the  electrode  potential  dropped  very  quickly  to  a  value  not 
far  from  the  hydrogen  electrode  potential,  and  was  not  influenced 
at  all  by  changes  in  the  current  density  on  the  remainder  of  the 
mercury  surface.  A  small  electrode  tip  was  also  pushed  into  soft 
lead  sufficiently  to  cause  a  marked  lowering  in  potential. 

We  therefore  have  used  tips  of  approximately  one  millimeter  or 
less  in  diameter  and  have  pressed  them  directly  but  lightly  against 
the  active  electrode  surface  during  measurements.  A  small  piece 
of  cotton  is  inserted  in  the  end  of  the  tip  to  prevent  bubbles  enter- 
ing the  tube  and  breaking  the  electrical  circuit. 


62  M.  KNOBEL,  P.  CAPLAN,  AND  M.  EISEMAN 

APPARATUS. 

The  electrolyzing  vessel  was  a  U-tube  of  3.8  cm.  (1.5  in.) 
tubing,  the  anode  and  cathode  being  in  opposite  arms,  and  the 
cross  tube  plugged  with  cotton  wool  to  prevent  mixing  of  the 
solutions.  The  electrode  under  investigation  was  placed  directly 
opposite  the  cross  arm  and  was  sufficiently  small  (usually  one 
centimeter  square)  so  that  a  uniform  current  density  was 
obtained.  Three  ammeters  of  dififerent  ranges  were  placed  in  the 
electrolyzing  current  line  to  measure  accurately  small  and  large 
currents.  The  potential  measuring  apparatus  was  an  ordinary 
potentiometer  sensitive  to  0.1  millivolt.  The  reference  electrodes 
used  were  Hg,  HgoSO^,  H2SO4  (2A/')"  with  sulfuric  acid  solu- 
tions Hg,  HgO,  KOH  (IN)  with  alkaline  solutions^^  and  the 
normal  calomel  electrode  with  the  salt  solutions,  each  being 
checked  against  the  standard  hydrogen  electrode  occasionally. 

The  electrodes  were  always,  when  possible,  made  of  square 
sheets,  exactly  one  centimeter  on  each  side.  A  projection  left 
on  this  sheet  or  a  stout  wire  soldered  to  the  back,  passed  up 
through  a  glass  tube.  The  back  and  connecting  strip  up  to  the 
glass  were  heavily  coated  with  asphalt.  This  was  found  to  be 
very  satisfactory,  the  asphalt  being  unattacked  in  all  the  solutions 
used  and  having  no  tendency  to  peel  off  as  paraffine  does.  In 
every  case  the  actual  resistance  of  the  electrode  and  lead  itself  was 
determined  and  corrections  made  for  the  ohmic  resistance  drop 
if  it  were  appreciable. 

The  surface  wherever  possible,  was  polished  with  No.  0000 
emery  paper.  Any  further  polishing  seemed  useless  as  the  sur- 
face is  so  soon  roughened  after  passing  the  electrolyzing  current. 
All  measurements  were  made  with  the  apparatus  in  a  thermostat 
at  25°C.,  regulated  to  within  0.2°C. 

MATERIALS. 

All  hydrogen  overvoltages  were  measured  in  pure  two-normal 
sulfuric  acid,  care  being  always  taken  to  saturate  the  solution  first 

"98  g.  HjSO,  per  1,000  g.  of  water. 

"  In  calculating  tlie  oxygen  overvoltages  the  electromotive  force  of  the  hydrogen- 
oxygen  cell  was  taken  as  1.227  volts.  See  L,ewis  and  Randall  J.  Am.  Chem.  See. 
36,   1969   (1914). 


EFFEICT  OF   CURRENT  DENSITY   ON   OVERVOLTAGE.  63 

with  hydrogen.  Oxygen  overvoltages  were  determined  in  one 
normal  potassium  hydroxide.  No  particular  effort  was  made  to 
saturate  this  solution  with  oxygen  as  the  results  were  too 
unsteady  to  warrant  it  and  were  not  improved  by  preliminary 
saturation.  For  the  halogen  overvoltages,  saturated  solutions  of 
the  sodium  or  potassium  halide  were  used,  saturated  further  with 
the  pure  halogen.  The  strong  solutions  were  used  to  give  a  good 
conducting  solution  and  to  avoid  depletion  of  ions  at  the  electrode. 
The  saturation  with  the  halogen  is  obviously  necessary  since  the 
equilibrium  potential  with  no  current  will  only  be  obtained 
under  that  condition. 

Wherever  possible  the  pure  metals  were  obtained  for  electrodes. 
No  extraordinary  care  was  exercised  however  as  the  measure- 
ments could  not  be  made  with  a  precision  to  necessitate  it. 

The  procedure  in  making  a  run  was  to  set  the  current  at  the 
desired  value  and  make  the  electrode  potential  determination  within 
one  minute,  then  raise  the  current  and  make  the  next  measure- 
ment, etc.  The  objection  will  immediately  be  raised  against  this 
procedure  that  insufficient  time  is  allowed  for  the  electrode  to 
come  to  a  constant  value,  but  it  was  found  without  question  that 
more  nearly  reproducible  values  could  be  obtained  on  the  increase 
and  decrease  of  current  in  this  way.  If  five  or  ten  minutes  were 
allowed  at  each  step  the  electrode  had  so  changed  by  the  time  a 
complete  run  had  been  made,  from  the  lowest  current  to  the 
highest  and  back,  that  the  last  value  was  several  tenths  of  a  volt 
dift'erent  in  some  cases  from  what  it  had  been  for  the  same  current 
at  the  start.  Particularly  with  the  higher  currents  the  measure- 
ments were  made  quickly  to  avoid  destroying  the  surface  and  to 
prevent  excessive  heating.  In  general  measurements  were  made 
both  on  the  increase  and  decrease  of  current,  but  only  the  values 
for  increasing  current  are  listed.  Check  runs  on  newly  made 
electrodes  were  made  in  every  case ;  these  were  considered  satis- 
factory if  the  forms  of  the  curves  were  similar  and  the  deviation 
of  the  two  not  more  than  0.1  volt.  One  of  the  runs  only  is 
listed  and  not  an  average  of  the  two  as  the  form  of  the  curve 
is  believed  to  be  more  important  than  absolute  values. 


64 


M.  KNOBEL,  P.  CAPLAN,  AND  M.  EISEMAN. 


RESULTS. 

The  results  are  listed  in  Tables  I  to  V  and  shown  graphically 
in  Fig.  5  to  9.    The  current  density  is  given  in  milliamperes  per 

Table  I. 
Hydrogen  Overvoltages  at  25° C. 


Current 

Overvoltage  in  Volts. 

density 

milL 

amp. 

per 

sq.  cm. 

Au 

Cd 

Cu 

Plat- 
inized 
Pt 

Smooth 
Pt 

A! 

Graph- 
ite 

0 

0.466 

0.000 

0.0022 

0.1 

o.m 

0.651 

0.35  i 

0.0034 

0.499 

0.3166 

1 

0.241 

0.981 

0.479 

0.0154 

0.024 

0.565 

0.5995 

2 

•  •  •  • 

0.0208 

0.034 

0.625 

0.6520 

5 

0.332 

1.086 

0.548 

0.0272 

0.051 

0.745 

0.7250 

10 

0.390 

1.134 

0.584 

0.0300 

0.068 

0.826 

0.7788 

50 

0.507 

1.211 

•  ■  ■  • 

0.0376 

0.186 

0.968 

0.9032 

100 

0.588 

1.216 

0.801 

0.0405 

0,288 

0.996 

0.9774 

200 

0.668 

1.228 

0.988 

0.0420 

0.355 

1.176 

1.0794 

500 

0.770 

1.246 

1.186 

0.0448 

0.573 

1.237 

1.1710 

1000 

0.798 

1.254 

1.254 

0.0483 

0.676 

1.286 

1.2200 

1500 

0.807 

1.257 

1.269 

0.0495 

0.768 

1.292 

1.2208 

Current 

Overvo 

tage  in   Vc 

)ltS. 

density 

mill. 

1 

amp. 

per 

sq.  cm. 

Ag 

Sn 

Fe 
electrode 

Chem- 
metal 

Brass 

Monel 
metal 

Dur- 
iron 

0 

0.2411 

0.2026 

0.2824 

0.1680 

0.1 

0.2981 

0.3995 

0.2183 

0.3160 

0.3832 

0.1911 

0.1710 

1 

0.4751 

0.8561 

0.4036 

0.6592 

0.4967 

0.2754 

0.1970 

2 

0.5787 

0.9469 

0.4474 

0.7249 

0.5346 

0.3022 

0.2136 

5 

0.6922 

1.0258 

0.5024 

0.7885 

0.5960 

0.3387 

0.2443 

10 

07618 

1.0767 

0.5571 

0.8349 

0.6459 

0.3832 

0.2856 

50 

0.8300 

1.1851 

0.7000 

0.9322 

0.8011 

0.5345 

0.5096 

100 

0.8749 

1.2230 

0.8184 

0.9696 

0.9104 

0.6244 

0.6129 

200 

0.9379 

1.2342 

0.9854 

0.9989 

1.1088 

0.7108 

0.7240 

500 

1.0300 

1.2380 

1.2561 

1.0407 

1.2318 

0.8619 

0.8591 

1000 

1.0890 

1.2306 

1.2915 

1.0682 

1.2544 

1.0716 

1.0205 

1500 

1.0841 

1.2286 

1.2908 

1.0859 

1.2491 

1.2095 

1.1400 

square  centimeter  and  the  overvoltage  in  volts.  The  overvoltage 
is  given  to  tenths  of  a  millivolt  where  the  steadiness  of  the  indi- 
vidual values  seemed  to  warrant  it,  although  it  is  unlikely  that 
the  measurements  could  be  reproduced  to  better  than  two  to  three 


EFFECT  OF   CURRENT  DENSITY   ON   OVERVOLTAGE. 


65 


si|o.\     a\    jaB)|0:ija.\o 


66 


M.  KNOBEL,  P.  CAPLAN,  AND  M.  EISEMAN. 


millivolts.  At  the  higher  current  densities  the  potential  often 
became  rather  variable  and  was  recorded  therefore  in  some  cases 
only  to  hundredths  of  a  volt. 

The  values   for  hydrogen   overvoltage   on   zinc,   bismuth  and 


TabIvE  I. — Continued 
Hydrogen  Overvoltages  at  25°  C. 


Current 

Overvoltage  in  Volts 

density 

m 

mill. 

amp. 

Zn 

Carbon 

Bi 

Ni 

Pb 

per    . 

sq.   cm. 

1 

0.716 

0.78 

0.563 

0.52 

2 

0.726 



0.633 

5 

0.726 

0.64 

0.98 

0.705 

1.060 

10 

0.746 

0.70 

1.05 

0.747 

1.090 

50 

0.926 

0.82 

1.15 

0.890 

1.168 

100 

1.064 

0.89 

1.14 

1.048 

1.179 

300 

1.168 

1.04 

1.20 

1.130 

1.217 

500 

1.201 

1.10 

1.21 

1.208 

1.235 

1000 

1.229 

1.17 

1.23 

1.241 

1.262 

1500 

1.243 

1.23 

1.29 

1.254 

1.290 

Current 

Overvoltage 

Current 

Overvoltage 

Current 

Overvoltage 

density 

Hg 

density 

Te 

density 

Pd 

0.00 

0.2805 

0.000 

0.000 

0.0769 

0.5562 

0.416 

0.6564 

0.227 

0.0546 

0.769 

0.8488 

0.832 

0.3505 

1.135 

0.1392 

1.54 

0.9295 

1.667 

0.4162 

2.27 

0.1820 

3.87 

1.0060 

4.16 

0.4405 

4.54 

0.2349 

7.69 

1.0361 

8.32 

0.1530 

11.35 

0.3165 

38.7 

1.0634 

41.6 

0.4705 

22.7 

0.4034 

76.9 

1.0665 

83.2 

0.4733 

113.5 

0.7205 

154 

1.0751 

166.7 

0.4986 

227 

0.8607 

387 

1.1053 

416 

0.5370 

454 

0.9521 

769 

1.108 

832 

0.5940 

1135 

1.0513 

1153 

1.126 

1250 

0.6590 

2270 
3400 

1.1168 
1.1570 

nickel  are  not  plotted,  but  the  form  of  the  curves  for  zinc  and 
nickel  is  not  dissimilar  to  that  of  graphite,  and  the  form  of  the 
bismuth  curve  resembles  that  of  lead. 

The  following  notes  supplement  the  table  of  hydrogen  over- 
voltages  : 


EFFECT  OF  CURRENT  DENSITY  ON  OVERVOLTAGE. 


67 


s;[ov    ni    9aB;iOAJ3AO 


68 


M.  KNOBEL,  P.  CAPLAN,  AND  M.  EISEMAN. 


The  graphite  was  a  very  soft  variety,  the  exposed  surface  of 
which  was  rubbed  with  No.  0  emery  paper. 

The  tellurium  was  badly  attacked  by  the  acid  so  that  the 
results  are  probably  of  no  value. 


Table  IL 
Chlorine  Overvoltages  at  25° C. 


Platinized  Pt 

Smooth   Pt                                 Grapiiite 

Current 
density 
mili.  amp. 
per   sq.   cm. 

Overvoltage 

Current 
density 
mili.    amp. 
per  sq.  cm. 

Current       ! 
Overvoltage            ?^ensity        |       Over- 
^           mill.  amp.           voltage 
per  sq.  cm. 

1.1 

5.7 
14.5 
21.7 
38.8 
60 
100 
200 
520 
1340 
1490 

0.0060 

0.0140 

0.0180 

0.0190 

0.0210 

0.024 

0.026 

0.035 

0.050 

0.089 

0.103 

1.1 

5.7 
11.4 
22.8 
43.0 
100 
200 
500 
750 
1000 
1350 

0.008 

0.0199 

0.0299 

0.0378 

0.0457 

0.0540 

0.0870 

0.161 

0.212 

0.236 

0.263 

40 
70 
100 
200 
500 
740 
980 
1131 

0.186 
0.193 
0.251 
0.298 
0.417 
0.466 
0.489 
0.535 

Table  III. 
Bromine  Overvoltages  at  25°  C. 


Platinized 

Platinum 

Smooth 

Platinum 

Graphite 

Current 

Current 

Current 

density 
mili.  amp. 

Overvoltage 

density 
mili.  amp. 

Overvoltage 

density 
mili.  amp. 

Overvol- 
tage 

per  sq.  cm. 

per  sq.  cm. 

per  sq.  cm. 

10 

0.002 

20 

0.002 

10 

0.002 

30 

0.005 

30 

0.004 

30 

0.008 

50 

0.007 

50 

0.006 

50 

0.016 

100 

0.012 

230 

0.033 

100 

0.27 

200 

0.025 

300 

0.357 

200 

0.54 

300 

0.041 

360 

0.113 

300 

0.81 

420 

0.056 

400 

0.156 

390 

0.108 

500 

0.069 

420 

0.164 

550 

0.163 

590 

0.082 

440 

0.178 

740 

0.218 

760 

0.130 

520 

0.266 

840 

0.253 

940 

0.202 

720 

0.379 

990 
1110 
1210 

0.329 
0.356 
0.400 

EFFECT  OF  CURRENT  DENSITY  ON  OVERVOETAGE. 


69 


1 

\ 

\ 

\ 

1 

1 

1 

* 

r 

\ 

L 

\ 

\ 

i       1 

\ 

\ 

\ 

\ 

'T 

a 

\ 

\ 

\ 

V 

\ 

^ 

i 

\  ^' 

\ 

n 

\ 

\i 

\ 

s, 

\ 

\ 

\v 

\ 

V    1 

^ 

■^ 

-o 

«    o 


r-.     ? 


j;       c 
_     ^      I 


sjio^    ni    38b)ioaj9ao 


70 


M.  KNOBEL,  P.  CAPLAN,  AND  M.  EISEMAN. 


The  zinc  was  obtained  from  a  dry  cell  casing.  It  was  suffi- 
ciently pure  so  that  it  would  not  dissolve  in  the  acid. 

The  bismuth  sample  was  a  piece  of  the  crystalline  metal,  coated 
with  asphalt  except  for  approximately  one  square  centimeter  of 
its  surface.    No  attempt  was  made  to  smooth  the  surface. 

Table  IV. 
Iodine  Overvoltages  at  25° C. 


Platinized  Platinum 

Smooth 

Platinum 

Graphite 

Current 

Current 

Current 

density 
mili.  amp. 

Overvoltage 

density 
mili.  amp. 

Overvoltage 

density 
mili.   amp. 

Over- 
voltage 

per  sq.  cm. 

per   sq.   cm. 

per  sq.  cm. 

10 

0.006 

12.3 

0.0039 

1.2 

0.002 

20 

0.012 

23 

0.0070 

5.7 

0.007 

40 

0.022 

50 

0.0127 

11.7 

0.0139 

110 

0.032 

90 

0.0216 

19.7 

0.0239 

220 

0.050 

130 

0.0353 

34.8 

0.0348 

400 

0.070 

200 

0.0510 

50 

0.0538 

710 

0.118 

310 

0.0744 

100 

0.0974 

810 

0.130 

520 

0.120 

200 

0.175 

1000 

0.196 

690 

0.150 

400 

0.315 

1300 

0.216 

1030 

0.220 

590 

0.451 

1460 

0.266 

1160 
1330 
1500 

0.245 
0.277 
0.292 

840 

0.645 

Table  V. 
Oxygen  Overvoltages  at  25° C. 


Current 

density 

mili. 

Overvoltage   in 

Volts 

amp. 
per 

Soft 
Graph- 

Au 

Cu 

Ag 

Chem- 
metal 

Smooth 
Pt 

Plat'z'd 
Pt 

Smooth 

Ni 

Spongy 
Ni 

sq.  cm. 

ite 

0.673 

0.422 

0.398 

1 

0.525 

0.580 

0.55 

0.721 

0.353 

0.414 

5 

0.705 

0.927 

0.546 

0.674 

0.90 

0.80 

0.480 

0.461 

0.511 

10 

0.896 

0.963 

0.580 

0.729 

1.02 

0.85 

0.521 

0.519 

0.563 

20 

0.963 

0.996 

0.605 

0.813 

0.92 

0.561 

.... 

50 

1.064 

0.637 

0.912 

1.10 

1.16 

0.605 

0.676 

0.653 

100 

1.091 

1.244 

0.660 

0.984 

1.084 

1.28 

0.638 

0.726 

0.687 

200 

1.142 

0.687 

1.038 

1.101 

1.34 

•  •  •  • 

.0.775 

0.714 

500 

1.186 

1.527 

0.735 

1.080 

1.127 

1.43 

0.705 

0.821 

0.740 

1000 

1.240 

1.63 

0.793 

1.131 

1.154 

1.49 

0.766 

0.853 

0.762 

1500 

1.282 

1.68 

0.836 

1.14 

1.175 

1.38 

0.786 

0.871 

0.759 

EFFECT  OF   CURRENT  DENSITY   ON   OVERVOLTAGE. 


71 


1\ 

\ 

1 

A 

\ 

\ 

\ 

\ 

\ 

\ 

\ 

\ 

\ 

\ 

*  \ 

l\ 

\ 

\ 

1 

\ 

^ 

1 

\\ 

\ 

\ 

Y 

°' 

\  \ 

\ 

\, 

\ 

\ 

,\ 

\ 

\ 

\ 

,  1 

\, 

\ 

\, 

\ 

\ 

■v 

\ 

\ 

\ 

y 

V 

\ 

\ 

\ 

\ 

V 

■\ 

\ 

\ 

,\ 

\\ 

\' 

\, 

\ 

\ 

^ 

K 

yw' 

\ 

\ 

^ 

\ 

\ 

1 

A 

\ 

^ 

\. 

\ 

i 

\ 

\ 

\ 

\ 

\ 

\ 

\ 

A 

V 

\ 

\ 

11 

^ 

^ 

-I 


IN*         ^ 


u 

c 

w 

"E 

V 

•  - 

c 

p 

to 

1 

c 
0 

c 

c 

74 

c 

l-H 

•0 

, 

w 

u 

l-l 

•— 

S5 

.. 

F 

a 

ii 

0 

E 

"E 

r 

U 

"c. 

0 
0 

c 

< 

•c 

UJ 

fe 

CO 

« 

"c 

c. 

^' 

c 
c 

- 

> 

c 

0 

u 

s 

^c 

c 

a 

u 

■a 

.c 

c 

CO 

c 
re 
u 

« 

u 

c 

S^IOA   'n   aSB^iOAaaAO 


72  M.  KNOBEL,  P.  CAPLAN,  AND  M.  EISEMAN. 

The  nickel  was  electrolytically  deposited  on  a  platinum  sheet 
from  a  pure  nickel  sulfate  solution. 

The  brass  sample  was  a  piece  which  contained  60  per  cent 
copper  and  40  per  cent  zinc. 

The  palladium  became  coated  with  a  dense  black  layer  which 
could  be  wiped  off,  or  which  cleared  up  entirely  on  standing  a 
few  minutes,  restoring  the  original  bright  surface. 

The  tin  sample  also  became  covered  with  a  black  layer,  similarly 
to  palladium,  but  this  layer  was  adherent  and  did  not  disappear 
on  standing. 

The  gold,  cadmium,  copper,  aluminum,  silver,  mercury,  pal- 
ladium, platinum  and  tin  electrodes  were  of  metal  which  was 
"chemically  pure." 

The  following  note  applies  to  the  oxygen  overvoltages : 

The  gold  was  strongly  attacked  by  the  oxygen.  After  the  run 
it  was  a  bright  copper  red  color.  On  account  of  the  extraordi- 
narily high  overvoltage  on  gold,  these  values  were  checked  at 
three  separate  times  and  the  figures  given  appear  to  be  correct.  It 
is  possible  that  the  oxide  coating  (which  appears  to  be  very 
adherent)  introduces  an  ohmic  resistance,  and  that  there  is  a 
partial  valve  action  as  on  aluminum. 

We  will  leave  the  theoretical  discussion  of  these  results  for 
a  later  article,  in  which  a  theory  of  overvoltage  will  be  outlined. 
The  following  general  observations  may  be  made  on  the  hydrogen 
overvoltages : 

1.  The  general  form  of  the  current  density  overvoltage  curve 
is  similar  to  that  of  a  logarithm  curve.  Except  in  a  few  cases, 
however,  a  simple  logarithm  equation  cannot  be  fitted  to  the 
entire  curve.  Such  an  equation  which  is  valid  for  very  low  and 
very  high  currents  would  show  a  much  sharper  bend  in  the  current 
density  range  from  10  to  200  milliamperes  per  square  centimeter, 
than  the  observed  curves. 

2.  Metals  generally  specified  as  having  a  high  overvoltage, 
as  lead,  mercury  and  cadmium,  rise  sharply  to  a  high  overvoltage 
at  low  current  densities  and  then  increase  but  little  with 
increasing  current. 

3.  Metals  of  "low"  overvoltage,  as  copper  and  gold,  show  a 
more    gradual    increase    of    overvoltage    with    current,    but    in 


EFFECT  OF  CURRENT  DENSITY  ON  OVERVOETAGE. 


73 


! 

1 

\ 

.1 

V 

\ 

\ 

1 

\  \ 

i 

w  \ 

\ 

w 

K 

\ 

w 

c« 

o                  a 

^ 

! 

a    O 


cc     CJ 


SflOA    "!    3Sb:)[oaj3ao 


74  M.   KNOBEL,  P.  CAPLAN,  AND  M.  EISEMAN. 

general  with  the  exception  of  platinum  and  gold,  finally  attain  as 
high  an  overvoltage  as  "high  overvoltage"  metals. 

4.  No  hydrogen  overvoltages  measured  by  us  exceed  the  value 
of  about  1.30  volts^-,  but  the  trend  of  most  of  the  curves  is 
toward  this  value. 

5.  Platinized  platinum  holds  a  unique  position  among  these 
other  smooth  metals  in  that  it  maintains  its  low  overvoltage  even 
at  exceedingly  high  current  densities.  In  another  experiment,  not 
listed,  the  current  through  a  well  platinized  electrode  was  increased 
until  a  spark  passed  from  electrode  to  solution  (at  a  current 
density  of  14  amperes  per  square  centimeter)  and  the  overvoltage 
just  before  this  point  was  reached  was  only  0.50  volt. 

No  generalizations  of  importance  are  apparent  in  regard  to  the 
halogen  overvoltages,  except  that  platinized  platinum  shows  the 
lowest  and  graphite  the  highest  values.  The  forms  of  the  curves 
are  widely  different,  some  being  nearly  linear. 

The  oxygen  overvoltages  are  rather  less  reliable  than  either  the 
hydrogen  or  halogen  overvoltages,  due  to  unaccountable  varia- 
tions with  time.  Even  after  long  polarization  at  a  given  current, 
the  overvoltage  may  vary  by  a  tenth  of  a  volt  or  more.  The 
shape  of  the  curves  is  in  general  logarithmic. 

SUMMARY. 

Values  of  the  hydrogen  overvoltage  at  twenty-two  cathodes ; 
of  the  chlorine,  bromine  and  iodine  overvoltages  at  three  anodes ; 
and  of  the  oxygen  overvoltage  at  nine  anodes,  have  been  deter- 
mined and  tabulated  at  various  current  densities  from  one  milli- 
ampere  to  one  and  one-half  amperes  per  square  centimeter.  All 
measurements  were  made  at  25 °C.  ±  0.2° C. 

An  investigation  of  the  method  of  measuring  overvoltage  has 
led  to  the  conclusion  that  the  use  of  a  small  glass  tip  less  than 
one  millimeter  in  diameter,  pressed  against  the  active  electrode 
surface  while  the  current  is  passing,  will  give  correct  results. 

"  While  higher  values  may  be  found  in  the  literature,  we  believe  they  are  due  to 
imperfect   elimination   of   ohmic   resistances. 


EFFECT  OF  CURRENT  DENSITY  ON  OVERVOLTAGE.  75 

DISCUSSION. 

W.  G.  HoRSCH^ :  In  connection  with  this  overvoltage  work, 
has  Dr.  Knobel  ever  attempted  to  analyze  the  commutator  method 
by  applying  equations  similar  to  those  that  have  been  developed 
for  expressing  the  rise  and  decay  of  current  in  the  ordinary 
copper  wire  circuits? 

M.  Knobee:  I  have  done  just  that  with  the  curves  of  over- 
voltage  as  a  function  of  the  time.  The  course  of  these  curves 
is  a  function  of  the  concentrations  of  the  gas ;  and  deducting  the 
theoretical  equation  and  making  the  equation  fit  that  curve,  con- 
stants are  obtained  which  involve  the  concentration  of  hydrogen 
on  the  electrode.    I  hope  soon  to  publish  the  results. 

P.  Caplain  :  I  think  it  important  to  emphasize  something  that 
Dr.  Knobel  has  pointed  out.  Ultimately,  it  appears,  there  is  no 
such  thing  as  a  high  or  low  overvoltage  electrode.  If  you  in- 
crease the  current  sufficiently,  and  permit  a  sufficient  lapse  of 
time,  the  hydrogen  overvoltages  of  the  metals  investigated  ap- 
parently tend  to  rise  to  the  same  maximum  value,  provided  no 
secondary  reactions  take  place.  The  literature  of  the  past  has 
considered  low  and  high  overvoltage  metals,  but  the  results,  in 
the  last  analysis,  seem  to  indicate  that  this  is  a  fallacy. 

M,  Knobel  :  I  have  tried  all  possible  methods  and  have  spent 
considerable  time  trying  to  get  reproducible  overvoltage  measure- 
ments, but  it  seems  almost  impossible.  There  are  unaccountable 
variations  with  time  which  apparently  can  not  be  eliminated. 

E.  O.  Benjamin-:  I  would  like  to  ask  ]\Ir.  Caplain  whether 
the  condition  that  he  approaches,  in  saying  that  the  overvoltage  of 
all  these  metals  would  be  nearly  alike,  is  not  an  approach  to  a 
true  gas  electrode,  in  eliminating  the  characteristics  of  the  metallic 
elements  ? 

P.  Caplain  :  That  is  what  the  results  seem  to  indicate.  I 
should  not  go  much  further,  because  Dr.  Knobel  is  personally 
working  on  the  problem.  Assuming  that  the  electrode  does  be- 
come saturated  with  gas,  and  applying  the  gas  laws,  a  value  of 
overvoltage  may  be  calculated  which  corresponds  to  that  obtained 
experimentally. 

E.  O.  Benjamin:     In  carrying  on  the  work  of  electrolysis 

1  Chile   Exploration  Labs.,   New  York  City. 

-  Consulting  Engr.  and  Chemist,  Newark,  N.  J. 


76 


DISCUSSION. 


of  water  on  a  large  scale,  in  dealing  with  a  square  meter 
electrode,  taking  a  characteristic  volt-ampere  curve,  we  obtain 
at  zero  current  flow  about  1.5  volts  as  the  potential  between 
a  nickel  anode  and  an  iron  cathode  in  a  sodium  hydroxide  solution. 
The  volt-ampere  curve  generally  assumes  a  form  similar  to  the 

3000- 


2500- 


2000- 


1500- 


1000- 


600- 


curve  shown  in  Fig.  1 .  We  reach  point  "e"  where  the  curve  nearly 
assumes  the  form  of  a  straight  line,  and  it  seems  that  from  the 
point  "e"  to  the  point  "f,"  we  have  full  saturation  of  the  elec- 
trodes. Below  "e"  we  have  a  partial  saturation  of  the  electrodes, 
and  are  gradually  building  up  the  gas  film.  Above  "f"  the  resist- 
ance seems  to  increase,  due  to  the  formation  of  molecular  gas  on 
the  electrodes  and  the  accumulation  of  gas  bubbles  in  the  elec- 


EFFECT  OF   CURRENT  DENSITY  ON   OVERVOLTAGE.  77 

trolyte,  thereby  reducing  the  effective  cross-section  of  the  electro- 
lyte. 

This  may  have  some  bearing  on  the  fact  that  a  gas  electrode 
is  actvialiy  formed  at  a  point  above  "e,"  which  we  may  assume  to 
be  the  saturation  point  of  the  electrode.  In  many  cases  this  curve 
has  been  referred  to  as  the  decomposition  characteristic  of  a  cell ; 
and  with  the  same  conditions  as  to  electrolyte,  temperature,  etc., 
but  regardless  of  pressure,  we  never  vary  more  than  about  0.02 
volt. 

In  a  large  electrode  we  do  not  necessarily  have  a  complete  film 
of  the  gas,  and  below  that  point  when  a  portion  of  the  metallic 
electrode  is  exposed,  it  is  what  I  refer  to  as  an  unsaturated  con- 
dition, meaning  that  there  is  metallic  surface  which  can  be  coated 
or  that  will  hold  a  gas  film. 

M.  Knobel  :  That  is  quite  in  accord  with  some  results  which 
I  am  getting.  I  think  it  will  appear  that  the  maximum  overvolt- 
age  occurs  when  there  is  a  single  molecular  layer  of  gas  on  the 
electrodes ;  when  you  start  to  build  up  a  second  layer  you  are 
then  forming  practically  free  monatomic  hydrogen  at  the  maxi- 
mum overvoltage. 

Carl  Hering''  :  When  the  electrode  is  covered  with  a  molar 
film  of  gas,  the  voltage  rises  fifty  to  a  hundred  fold,  and  produces 
an  arc  over  the  whole  surface,  as  I  showed  in  a  paper  some  years 
ago.  If  the  electrode  is  covered  completely  with  a  film  of  gas, 
this  arc  heats  the  electrode  so  quickly  that  one  can  melt  steel 
under  water. 

A.  H.  W.  Aten*  (Communicated)  :  From  experiments  on  the 
scattering  of  lead  cathodes  I  concluded  in  1916'^  that  the  over- 
voltage  for  hydrogen  evolution  might  be  ascribed  to  a  slow  com- 
bination of  hydrogen  atoms  to  hydrogen  molecules.  If  the  result 
of  the  present  authors,  that  the  overvoltage  at  high  current  den- 
sities approaches  a  limiting  value  of  1.3  volt,  independent  of  the 
nature  of  the  metal  (except  in  the  case  of  plantinized  platinum) 
is  considered  from  this  point  of  view,  this  would  mean  that  1.3 
volt  possibly  corresponds  to  the  potential  of  atomic  hydrogen  of 
one  atmosphere.  Now  the  potential  of  atomic  hydrogen  is  given 
by  the  formula: 

2  Consulting   Electrical   Engr.,   Philadelphia,   Pa. 

*  Prof,   of  Chemistry,   University   of  Amsterdam,   Holland. 

=  Proc.  Roy.  Acad.  Sci..  Amsterdam,  18,  1379  (1916). 


78  DISCUSSION. 

Vh  =  Eh   -   0.058  logio  Ch-  —  0.058  login  Ph         (1) 
and  the  potential  of  molecular  hydrogen  by 

Vh,  =  Eh,   +   0.058  logio  Ch^  —  0.029  logic  Ph,        (2) 

where  Eh,  —  Eh  is,  according   to  the  assumption   made   above, 
equal  to  1,3  volt. 

If  one  mol  of  hydrogen,  at  a  pressure  of  one  atmosphere,  is 
dissolved  into  atoms  at  constant  volume,  the  pressure  will  be  two 
atmospheres.  The  decrease  in  free  energ}%  in  transforming  this 
atomic  hydrogen  galvanically  into  molecular  hydrogen,  will, 
according  to  (1)  and  (2)  be  given  by 

2F  (Eh,  —  Eh)+  2F  X  0.058  logio  2.  (3) 

Ph,  being  =   1  and  Ph  =  2. 

The  value  of  the  second  term  lies  within  the  experimental  un- 
certainty, and  we  can  put  the  decrease  of  the  free  energy-  equal  to 
2  X  96,500  X  1.3  joules  =  250,000  joules  =  60,000  cal. 

The  value  of  60,000  cal.  for  the  molal  dissociation  energy  of 
hydrogen  is  rather  low.  This  figure  is  calculated  from  the  Bohr- 
Debyes  model  of  the  hydrogen  molecule,  but  the  experimental 
values  are  higher. 

Langmuir'^  finds  84.000  cal.,  and  Isuardi'  95.000  cal..  whereas 
Franck,  Knipping  and  Kruger*  calculate  81,300  ±  5,700  cal.  from 
ionization  potentials.  Assuming  these  latter  values  to  be  correct, 
it  should  be  concluded  that  even  at  the  highest  current  densities 
the  concentration  of  H  atoms  at  the  cathode  remains  less  than 
that  corresponding  to  one  atmosphere. 

M.  Knobel  (Coiinuiiiiicatcd)  :  The  discussion  by  Prof.  Aten 
brings  up  a  point  of  which  we  were  aware,  but  had  excluded  from 
the  present  article,  planning  to  discuss  it  with  other  theoretical 
interpretations  of  the  data  in  connection  with  a  theory  of  over- 
voltage.  In  view  of  the  uncertainty  in  the  experimental  and  cal- 
culated values  of  the  free  energy  of  formation  of  Ho  from  Hj, 
we  believe  one  can  conclude  that  the  monatomic  hydrogen  is  at 
a  pressure  of  approximately  one  atmosphere,  when  the  maximum 
overvoltage  is  reached. 

"Z.  f.  Elektrochemie,  23,  217  (1917). 
•Z.  f.  Elektrochemie,  21,  405  (1915). 
"Ber.  Deut.  physik.  Ges.  21,  728  (1919). 


A  paper  presented  at  the  Forty-third 
General  Meeting  of  the  American  Elec- 
trochemical Society  held  in  New  York 
City,  May  3,  1923,  Dr.  Wm.  G.  Horsch 
in   the   Chair. 


ELECTROTITRATION  WITH  THE  AID  OF  THE  AIR  ELECTRODE.' 

By  N.  Howell  Furman.* 

Abstract. 
A  brief  report  of  further  progress  in  the  study  of  some  uses 
of  the  air  electrode  is  given.  The  results  indicate  clearly  that 
the  air  electrode  is  capable  of  giving  satisfactory  results  in 
electrotitration,  either  in  the  presence  or  absence  of  oxidizing 
agents. 

INTRODUCTION. 

The  results  of  a  preliminary  study  of  some  applications  of 
cells  composed  of  a  AT  calomel  electrode  in  conjunction  with  some 
one  of  the  electrodes: 

(A)  Oxygen  Electrode, 

(B)  Air  Electrode, 

(C)  Platinized  Platinum  Electrode, 

(D)  Burnished  Platinum  Electrode, 

have  been  presented  in  a  recent  communication.^ 

It  was  shown  in  a  general  way  that  the  oxygen  electrode-iV 
calomel  electrode  cell  may  be  used  to  construct  titration  curves 
which  are  in  large  measure  analogous  to  those  which  are  obtained 
in  the  familiar  hydrogen  electrode  titrations.  Special  emphasis 
was  placed  upon  the  use  of  the  oxygen  electrode  in  the  acidimetry 
and  alkalimetry  of  solutions  which  contained  strong  oxidizing 

agents.  , 

The  majority  of  the  results  which  were  there  presented  were 
obtained  with  the  aid  of  the  oxygen  electrode.     A  number  of 

1  Manuscript  received  January  31,   1923. 

»  Cor^tribution  from  the  Chemical  Laboratory  of  Prmceton  Unn-ers.ty. 
3  Furman.  J.  Am.  Chem.   Soc.  44.  2,685    (1922). 

79 


8o  N.    HOWELL    FURMAN. 

results  which  were  obtained  by  means  of  the  other  electrodes 
(B,  C  and  D)  were  included.  After  a  brief  preliminary  study, 
the  burnished  platinum  electrode  was  found  to  be  extremely 
sensitive  to  minute  variations  in  the  details  of  handling.  Recently 
van  der  Meulen  and  Wilcoxon*  have  described  the  conditions 
under  which  the  burnished  platinum  electrode  may  be  employed 
successfully. 

The  air  electrode  is  very  slightly  influenced  by  minor  changes 
in  the  mode  of  manipulation,  in  which  respect  it  is  superior  to 
the  platinized  and  burnished  platinum  electrodes.  The  potential 
of  an  air  electrode  toward  any  given  solution  is  subject  to  the 
well-known  drift  in  value  which  is  ordinarily  ascribed  to  the 
process  of  oxide  formation.^  Nevertheless  satisfactory  titration 
curves  may  be  obtained  either  in  the  presence  or  absence  of 
oxidizing  agents.  Furthermore,  a  fair  approximation  of  the 
hydrogen  ion  concentration  may  be  obtained  by  making  an  em- 
pirical calibration  of  the  electromotive  force  of  the  air  electrode-AT 
calomel  electrode  cell  with  the  aid  of  a  series  of  solutions  of 
known  hydrogen  ion  concentration. 

Arthur  and  Keeler^  have  described  a  continuous  recording 
apparatus  for  the  measure  and  control  of  the  alkalinity  of  boiler 
feed  water  by  means  of  the  air  electrode-0.1  A''  calomel  electrode 
combination.  The  electromotive  force  readings  of  this  cell  were 
calibrated  in  terms  of  grains  of  alkalinity  per  gallon  of  water. 
They  state  that  this  method  involving  the  air  electrode  was  more 
reliable  under  continuous  operating  conditions  that  the  colon- 
metric  method  in  the  hands  of  unskilled  operators. 

EXPERIMENTAL. 

The  apparatus,  mode  of  procedure,  details  of  standardizing 
reagents,  and  general  technique  were  described  in  the  previous 
paper.  It  is  perhaps  well  to  repeat  that  the  electromotive  force 
values  were  repeatedly  referred  to  the  value  of  a  Weston  satu- 
rated type  standard  cell  by  means  of  a  potentiometer  set  of 
moderate  precision.^ 

*  \'an   der   Meulen  and  Wilcoxon,  Ind.   and   Eng.   Chem.,    15,   62    (1923). 

°  A  brief  review  of  the  literature  relating  to  the  oxygen  and  air  electrodes  was  given 
in  the  previous  paper  (ref.  3)  p.  2686.  Data  were  given  relating  to  the  magnitude 
of  the  drift  to  be  expected. 

«  Arthur  and  Keeler,  Power,  55,  768  (1922). 

'  Leeds  and  Northrup  students'  type  potentiometer,  and  a  galvanometer  of  current 
sensitivity  2  mm.  per  micro-ampere. 


ELECTROTITRATION  BY  THE  AIR  ELECTRODE.         8l 

A  number  of  curves  typical  of  those  obtained  are  plotted  in 
Fig.  1.  It  should  be  noted  that  the  form  of  the  apparatus  made 
it  necessary  to  dilute  the  solutions  somewhat  before  the  titrations 
were  commenced.  The  volume  was  ordinarily  75  cc.  at  the 
start  of  each  titration. 


.6 


.5 


.3 


f 


.1 


f 

4' 

/^  a 

/ 

7 

/T^ 

^i 

^ 

w 

^ 

/ 

a 

i 
t 

I 

>     (1 

-<- 

t 

^ 

^ 

Jj 

> < 

> ' 

r= ' 

y 

5  /O  /5 

CC.   AC/D  - 


20 


2.5 


30 


Fig.    1.    Air    Electrode   Titrations. 


(1)  Titration    of    15.87    cc.    of    0.09123    N    alkali    and    8.33    cc.    0.1000    N 
NazCrOi  with  0.09986  A'  HCl. 

Cal.   14.50,   found   14.55   cc.   acid  to  neutralize   free  alkali. 
Calc.    8.34,    found    8.43    cc.    acid    to    transform    chromate    into    dichromate 
(distance  o  to  b  curve   1). 

(2)  25  cc.  of  0.5000  A^  NaaCOs  titrated  with  0.5176  A^  H2SO4. 
Point  c   (bicarbonate   point)    calc.    12.07,   found   12.19  cc.   of  acid. 

Second    inflection     (complete    neutralization)     calc.    24.15,    found    24.20    cc. 
of  acid. 

(3)  Titration  of  25  cc.  0.09123  N  NaOH  with  0.09986  N  HCl.  Calc.  22.84, 
found  22.93  cc. 

(4)  Titration  of  25  cc.   0.4937  N  NaOH  with  0.5050  A^  acetic  acid.     Calc. 
24.44,  found  24.40  cc. 


82 


N.    HOWELL    FURMAN. 


A  qualitative  measure  of  the  difference  in  hydrogen  ion  con- 
centration of  0.1  A^  hydrochloric  acid,  as  contrasted  with  0.5  N 
acetic  acid,  is  given  by  the  relative  positions  on  the  voltage  scale 
of  the  end  portions  of  curves  (3)  and  (4)  respectively.  Point 
c  curve  (2)  represents  the  completion  of  the  conversion  of 
sodium  carbonate  into  bicarbonate.  Point  a  curve  ( 1 )  represents 
the  neutralization  of  free  alkali  in  the  presence  of  chromate; 
while  point  b  represents  the  completion  of  the  conversion  of 
chromate  into  dichromate. 

Fig.  2  contains  titration  curves  for  nitric  (1)  and  perchloric 


5  10  15         20 

C  C.  OF   nZAGENT-^ 

Fig.   2.    Air   Electrode  Titrations. 

(1)  25  ec.  0.5285  .V  nitric  acid  titrated  with  0.4541  A^  NaOH.   Calc.  29.10, 
found  29.15  cc. 

(2)  24.  cc.   of  0.09986  N  HCl  titrated  with  0.09123  .V  NaOH.   Calc.   26.27, 
found  26.25  cc. 

(3)  Titration  of  25  cc.  of  0.09497  N   HClGi  with  0.09123  N  NaOH.   Calc. 
26.02,   found   25.98  cc. 

Data    showing   the   results    of    No.    (4)    and    (5)    will    be    found    in    exp.    7, 
Table  I. 


ELECTROTITRATIOX  BY  THE  AIR  ELECTRODE.  83 

(3)  acids,  together  with  a  curve  for  hydrochloric  acid  (2)  for 
purposes  of  comparison.  A  salt  bridge  of  approximately  0.1  N 
sodium  nitrate  was  interposed  between  the  solution  and  the  calo- 
mel electrode  during  the  titration  of  the  nitric  acid.  Curve  (4) 
represents  the-  titration  of  dichromate,  in  a  mixture  of  chromate 
and  dichromate,  with  alkali.  At  the  end  of  this  titration  all  of 
the  chromate  was  converted  into  dichromate  by  means  of  standard 
acid.  The  distance  a  to  &  (curve  5)  represents  the  amount  of 
acid  required. 

In  the  previous  paper*  it  was  shown  that  an  accurate  deter- 
mination of  free  alkali  in  the  presence  of  chromate  (providing 
the  carbonates  were  absent),  or  of  free  acid  in  the  presence  of 
dichromate,  could  be  made  with  the  aid  of  the  oxygen  or  air 
electrodes.  The  method  may  be  extended  to  the  analysis  of 
mixtures  of  chromate  and  dichromate  as  the  following  results 
will  serve  to  show. 

Solutions  of  chromate  and  dichromate  were  prepared.  Each 
solution  was  standardized  against  freshly  standardized  ferrous 
sulfate  by  the  electrometric  method  of  Forbes  and  Bartlett.^ 
Known  portions  of  the  solutions  were  mixed.  The  mixture  was 
then  analyzed  by  one  of  the  following  methods. 

(A)  The  amount  of  acid  necessary  to  convert  the  chromate, 
which  was  present  in  the  mixture,  into  dichromate,  was  deter- 
mined electrometrically.  The  total  amount  of  dichromate  was 
then  determined  electrometrically  either  (1)  by  means  of  standard 
alkali,  or  (2)  by  means  of  freshly  standardized  ferrous  sulfate, 
after  the  addition  of  a  large  excess  of  acid.  Total  alkali  re- 
quirement minus  acid  equivalent  to  chromate  equals  alkali 
equivalent  to   dichromate  present. 

(B)  The  amount  of  alkali  necessary  to  convert  dichromate  into 
chromate  was  determined.  The  total  acid  requirement  was  then 
found  (1)  by  direct  titration  with  standard  acid  (distance  a  to  & 
curve  5,  Fig.  2)  or  (2)  by  reduction  with  standard  ferrous  sul- 
fate after  strongly  acidifying  the  solution.  Then,  total  acid 
requirement  minus  alkali  equivalent  to  dichromate  equals  acid 
equivalent  to  chromate. 

spurman,  J.  Am.   Chem.    Soc,   44,  2,685    (1922). 

9  Forbes   and  Eartlett,  J.  Am.   Chem.   Soc.   35,   1,327   (1913). 


84 


N.    HOWELIv    FURMAN. 


AH  of  the  electrometric  methods,  as  well  as  a  number  of  other 
physico-chemical  methods,  agree  in  finding  two  sharply  defined 
changes  in  the  neutralization  curves  of  chromic  acid  or  of 
acidified  chromate  solutions.  Similar  changes  appear  in  the 
curves  for  the  acidification  of  alkaline  chromate  solutions.  Mar- 
gaillan^"  investigated  the  neutralization  of  M/30  solutions  of 
chromic  acid,  both  by  means  of  conductance  titration  and  hydro- 
gen electrode  titration.  By  both  methods  a  sharp  change  was  re- 
vealed when  one  mole  of  sodium  hydroxide  per  mole  of  chromic 
anhydride  (CrOs)  had  been  added;  a  second  sharp  change  ap- 
peared when  two  moles  of  alkali  per  mole  of  chromic  anhydride 

Table  I. 

Electrometric  Analysis  of  Mixtures  of  Chromate  and  Dichromate. 
Results  No.  1  to  3  are  calculated  to  0.1  N ;  No.  4  to  8  to  0.5  N. 


1 

2 

3 

4 

5 

No. 

Bichromate 

Dichromate 

Chromate 

Chromate 

Method 

Taken 

Found 

Taken 

Found 

Used 

cc. 

cc. 

cc. 

cc. 

(see   above) 

1 

8.33 

8.44 

18.05 

17.93 

B  2 

2 

16.67 

16.72 

18.05 

17.96 

B  2 

3 

8.33 

8.44 

7.19 

7.24 

B  1 

4 

16.85 

16.89 

7.19 

7.20 

A  1 

5 

16.85 

16.79 

14  38 

14.45 

A  1 

6 

33.70 

33.78 

14.38 

14.44 

A  1 

7 

16.85 

16.80 

1438 

14.39 

B  1 

8 

33.70 

33.71 

7.19 

7.24 

B  1 

had  been  added.  Hughes"  who  has  recently  investigated  the 
glass  cell  (glaskette),  with  an  improved  apparatus  similar  to 
that  of  Haber  and  Klemensiewicz^^,  presents  an  interesting  curve 
for  the  neutralization  of  chromic  acid.  The  two  inflections  which 
the  author  obtained  with  the  aid  of  the  oxygen  or  air  electrodes 
appear  at  hydrogen  ion  concentrations  (empirically  estimated) 
which  are  in  fair  agreement  with  those  obtained  by  Hughes.  His 
method  seems  to  be  the  most  reliable  which  has  thus  far  been 
devised  for  measuring  hydrogen  ion  concentrations  in  solutions 
of  highly  colored  oxidizing  agents. 

«>  Margaillan,   Compt.   rend.,   157,   994    (1913). 

1'  Hughes,  J.  Am.  Chem.   Soc,  44,   2,860    (1922). 

'2  Haber  and  Klcmensiewicz,  Z.  physik.  Chem.,  67,  385   (1909). 


ELECTROTITRATION  BY  THE  AIR  ELECTRODE. 


85 


These  physico-chemical  resuhs  have  been  used  as  arguments 
for  or  against  (depending  upon  the  view-point  of  the  individual) 
one  or  the  other  of  the  two  stoichiometrically  equivalent  sets  of 
reactions : 


I 


{ 


(a)  H^CrO.  +  NaOH 

(b)  NaHCrO^  +  NaOH 


=  XaHCrO^ 
=  Xa,Cr04  - 


-  H,0 
H.O 


(  (a)   H(H.Cr.OT  + 
Ub)   ^^(Na^Cr.O:  -f 


2NaOH         =  Xa^Cr^O;  +  2H2O) 
2XaOH       =  ZXa.CrOi  +  H,0) 


In  this  work  the  second  set  (II)  has  been  adopted  as  being 
more  probable  and  convenient  in  picturing  the  relations  at  the 
two  points  of  inflection. ^^ 

Table  II. 

Variation  of  E.  M.  F.  of  Air-N  Calomel  Cell  with  Changes  in 
Hydrogen  Ion  Concentration. 


Approx. 

Normality 

Approx. 

Normality 

Time 

E.  M.  F. 
Volt 

Time 
Min. 

E.    M.    F. 

Min. 

Volt 

Acidic 

Basic 

25. 

Acidic 

Basic 

0.004 

0. 

0.007 

0.640 

0.077 

2.5 

0.007 

0.643 

27.5 

*0.004 

0.086 

5. 

0.007 

0  644 

30. 

0.004 

0.088 

6. 

0.006 

0.037 

30.5 

6.667 

0.630 

7.5 

0.006 

0.029 

35. 

0.007 

0.632 

10. 

0.006 

0.038 

36. 

0.003 

0.118 

10.5 

0.667 

0615 

38. 

.  ■  .  • 

0.003 

0.101 

12.5 

0.007 

0.618 

40. 

0.003 

0.101 

14. 

0.007 

0.617 

41. 

0.007 

. , , , 

0.634 

17. 

0.007 

0.616 

45. 

0.007 

> .  •  • 

0.633 

20. 

0.007 

0.615 

50. 

0.007 

. . .  • 

0.630 

21. 

0.004 

0.076 

185. 

0.007 

.... 

0.625 

22.5 

0.004 

0.073 

*  Change  caused  by   one   drop  of   approx.   0.5   A^  acid. 

The  experience  of  Arthur  and  Keeler,  as  well  as  numerous 
observations  made  in  the  course  of  this  work,  point  to  a  field  of 
usefulness  of  the  air  electrode  in  approximate  hydrogen  ion 
concentration  measurement  and  control.  Some  idea  of  the  readi- 
ness of  response  of  the  air  electrode  to  repeated  variations  in 

'8  A  comprehensive  discussion  of  the  nature  of  chromic  acid,  and  the  equilibrium 
relations  in  chrjmate  and  dichromate  solutions,  together  with  abundant  literature 
references  are  to  be  found  in  Abesg's  Handbuch  der  anorg.  Chem.,  IV,  1,  2nd  half, 
pp.   306-311,  Pub.  by   S.   Hirzel,   Leipzig,    1921. 


86  DISCUSSION. 

hydrog^en  ion  concentration  may  be  obtained  from  the  results  in 
Table  11.  Known  quantities  of  acid  and  alkali  were  alternately 
added  to  a  given  volume  of  solution. 

It  should  be  noted  that  the  readings  in  alkaline  solution  are 
more  sluggish  in  coming  to  a  state  of  gradual  drift  than  are 
those  in  acid  solution. 


DISCUSSION. 


]\I.  R.  Thompson^  :  Prof.  Furman's  paper  supplies  information 
on  an  interesting  phase  of  electrode  potential  measurements,  and 
is  an  important  contribution  to  this  subject. 

In  his  previous  paper  mentioned,  a  more  extended  discussion 
was  given  of  the  irreversibility  of  the  electrode.  If  the  oxygen 
(or  air)  electrode  were  really  reversible,  indicating  a  definite 
equilibrium  for  oxygen  (or  equivalent  hydroxyl)  ions,  the  meas- 
urements obtained  should  be  complementary  to  those  of  hydrogen 
ion  concentrations  by  the  reversible  hydrogen  electrode  and  would 
readily  serve  to  calculate  the  latter  concentrations.  This  depends, 
of  course,  upon  the  well-known  equilibrium  of  hydrogen  and 
hydroxyl  ions  in  water,  giving  at  about  25°   C.  the  relationship 

^  10-1^ 


CoH  — 


Or  in  Sorensen  units,  pH  r=  14  —  pOH.  Physico-chemical 
neutrality  exists  at  pH  =  pOH  ^  7  (which  is  the  end  point 
only  when  strong  acids  and  bases  are  combined),  and  in  the 
diagrams  this  point  is  represented  roughly  by  an  e.  m.  f.  not 
far  from  0.3  volt. 

Actually,  Prof.  Furman  and  others  have  shown  that  the  oxygen 
(or  air)  electrode  is  not  quite  reversible  and  that  measurements 
by  means  of  it  only  serve,  at  best,  to  calculate  approximate  hy- 
droxyl and  hydrogen  ion  concentrations,  if  the  latter  are  desired. 
This  condition  does  not  interfere,  however,  with  extensive  appli- 
cations of  the  electrode  for  relative  determinations,  such  as  the 
accurate  establishment  of  the  end  points  in  certain  classes  of 

1  Assoc.    Chemist,   Bureau   of    Standards,    Washington,    D.    C. 


ELECTROTITRATIOX  BY  THE  AIR  ELECTRODE.         87 

titrations  and  the  paper  has  demonstrated  this  fact  satisfactorily. 
We  may  expect  a  rapidly  increasing  field  of  usefulness  for  the 
air  electrode. 

N.  H.  FuRMAN :  I  hope  that  the  accuracy  of  this  method  of 
measuring  hydrogen  and  hydroxyl  ion  concentrations  will  not  be 
taken  too  seriously.  The  experience  of  the  research  department 
of  Leeds  &  Northrup  Co.-  and  my  own  seem  to  point  out  that 
about  all  you  can  hope  for  is  an  accuracy  of  one-half  of  a  Soren- 
sen  (pH)  unit. 

I  did  not  wish  to  give  the  impression  that  this  electrode  will 
give  values  for  hydrogen  or  hydroxyl  concentration  of  the  same 
order  of  accuracy  as  the  hydrogen  electrode.  It  does  serve  as  a 
rough  substitute  for  the  hydrogen  electrode  in  some  cases,  where 
the  solution  is  exposed  to  air  and  must  remain  so.  It  has  some 
usefulness  in  rough  control  work. 

O.  C.  Ralston^  :  For  over  a  year  we  have  been  using  an  air 
electrode  one  centimeter  square  and  heavily  platinized  at  the 
Pacific  Experiment  Station  of  the  Bureau  of  Mines  (Berkeley, 
Calif.)  for  the  purpose  of  following  the  course  of  hydrolytic 
reactions,  either  on  the  large  scale  or  during  titration  in  the  la- 
boratory. It  is  of  great  use  in  following  hydrolytic  purification 
of  electrolytes  of  zinc  sulfate,  copper  sulfate,  or  in  preparing 
iron-free  aluminum  sulfate  solutions. 

We  found  it  necessary  to  study  the  air  electrode  in  much  the 
same  way  that  Prof.  Furman  has  done  and  we  agree,  I  think, 
almost  entirely  with  his  conclusions.  The  most  important  thing 
about  it  is  that  the  air  electrode  can  not  be  used  as  an  exact 
measure  of  hydrogen  ion  (or  hydroxyl  ion)  concentration,  but 
as  an  indicator  of  the  end  of  certain  reactions,  due  to  changes  in 
direction  of  the  voltage-titration  curve  or  voltage-time  curve,  it 
is  very  satisfactory.  I  had  hoped  to  present  a  paper  at  this  meet- 
ing on  the  more  practical  applications  of  the  air  electrode,  but 
will  have  to  postpone  its  presentation. 

For  such  a  reaction  as  the  separation  of  ferric  iron  from  a 
copper  sulfate  solution,  using  powdered  limestone  or  copper  oxide 
for  hydrolyzing  the  ferric  sulfate,  the  voltage-titration  curve  is 

=  Private  communication    from    Dr.   I.    B.    Smith    of   Eeeds   &   Northrup   Co. 
3  U.   S.   Bureau  of  Mines,   Berkeley,   Calif. 


88  DISCUSSION. 

almost  horizontal  till  nearly  all  the  iron  has  been  precipitated  as 
a  basic  salt,  and  then  the  voltage  suddenly  drops,  indicating  the 
end  of  the  reaction.  Chemical  control  of  this  hydrolysis  is  diffi- 
cult, because  if  a  sample  of  the  pulp  is  filtered  at  this  point  the 
iron  stays  in  the  filtrate  as  a  colloidal  compound  which  makes  the 
solution  look  like  coffee.  Under  these  conditions  it  is  difficult 
for  the  chemist  to  determine  if  all  the  iron  in  true  solution  has 
been  hydrolyzed. 

The  air  electrode,  of  course,  functions  in  oxidizing  solutions 
where  the  hydrogen  electrode  fails,  especially  solutions  containing 
ferric  iron.  Since  most  technical  operations  with  inorganic  com- 
pounds in  solution  are  complicated  with  the  presence  of  iron  in 
the  solution,  the  hydrogen  electrode  has  previously  found  little 
use  in  this  field.  On  the  other  hand,  the  air  electrode  is  not  only 
satisfactory  but  more  easily  manipulated,  because  the  solutions  in 
practice  are  usually  saturated  with  air  or  are  stirred  or  agitated 
with  air  so  that  the  form  of  air  electrode  that  can  be  used  is  ex- 
tremely simple. 

W.  G.  HoRSCH* :  In  studying  methods  of  this  sort,  starting 
out  possibly  with  pure  solutions  and  then  trying  to  apply  the 
results  to  solutions  that  contain  other  constituents,  we  must  be 
careful  that  the  method  is  peculiar  to  the  reaction  that  we  are 
studying  or  to  the  endpoint  that  we  wish  to  obtain. 

*  Chile  Exploration  Labs.,  New  York   City. 


A  paper  presented  at  the  Forty-third 
General  Meeting  of  the  American  Elec- 
trochemical Society  held  in  New  York 
City,  May  3,  1923,  Dr.  Wm.  G.  Horsch 
in   the   Chair. 


THE  HYDROGEN  ELECTRODE  IN  ALKALINE  SOLUTIONS.' 

By  A.  H.  W.   Aten.« 

Abstract. 
When  a  hydrogen  electrode,  saturated  with  hydrogen,  is  in 
equihbrium  with  0.1  A/"  HCl,  it  is  in  the  same  state  of  equihbrium 
with  1.0  A''  HCl,  and  vice  versa.  This  is  not  the  case,  however, 
when  the  solution  of  an  alkali  is  used  in  place  of  an  acid.  When 
a  hydrogen  electrode  in  equilibrium  with  1.0  A^  NaOH  is  put  in 
0.1  A''  NaOH,  or  the  reverse,  a  considerable  time  period  is  required 
to  reach  a  new  equilibrium.  The  same  phenomenon  is  observed 
in  a  more  marked  degree  when  the  electrode  is  changed  from 
0.1  N  NaOH  to  0.1  A''  HCl,  or  the  reverse.  The  explanation 
suggested  is  that  the  electrode  must  absorb  Na  or  give  it  off,  as 
the  case  may  be,  in  order  to  reach  an  equilibrium  with  the  final 
solution.  [C.  H.  E.] 


In  the  course  of  an  investigation,  in  which  a  hydrogen  electrode 
was  brought  into  contact  with  solutions  of  varying  alkalinity,  it 
was  found  that  the  potential  in  a  given  solution  was  markedly 
afifected  by  the  alkalinity  of  the  preceding  solution.  This  phenom- 
enon was  further  examined,  and  the  following  is  an  account 
of  part  of  the  results. 

In  the  apparatus  shown  in  Fig.  1  the  tube  A,  containing  the 
hydrogen  electrode,  is  filled  with  a  solution  of  a  given  alkalinity, 
say  0.1  A'',  and  the  tube  B  with  a  solution  of  diff"erent  alkalinity, 
say  1.0  N.  After  the  hydrogen  electrode  has  reached  equilibrium 
in  the  0.1  A^  solution,  this  solution  is  removed  through  the  tap  C, 
and  the  \.0  N  solution,  which  has  been  saturated  with  hydrogen,. 
is  introduced  through  the  tap  D.     The  potential  of  the  hydrogen 

1  Manuscript  received  January  15,  1923. 

'  Prof,  of  Chemistry  at  the  Univ.  of  Amsterdam. 

89 


90 


A.    H.   W.    ATEN. 


electrode  is  measured  against  a  decinormal  calomel  electrode.  The 
liquid  junction  is  made  by  a  saturated  solution  of  potassium 
chloride.  The  values  given  are  those  immediately  measured, 
without  attempting  to  correct  them  further  for  liquid  potentials. 
The  temperature  was  room  temperature,  about  18°  C.  The 
electrodes  consisted  of  gold  sheet,  0.5  x  3  cm.,  covered  with  pal- 
ladium black  by  electrolyzing  a  solution  of  palladium  chloride  (0.3 
per  cent  palladium)  with  0.1  amp.  for  5  min.  Then  these 
electrodes  were  cathodically  polarized  in  a  solution  of  sulfuric 
acid,  in  order  to  reduce  the  absorbed  palladium  chloride,  and  to 
charge  them  with  hydrogen. 


When  an  electrode  thus  treated  was  brought  into  contact  with 
0.1  A''  HCl,  contained  in  tube  A,  the  potential  rose  in  8  min. 
from  +0.340  to  +0.400.  In  contact  with  1.0  A^  HCl  it  rose 
from  +0.288  to  +0.343  in  14  min.  Under  these  circumstances 
the  equilibrium  was  reached  in  a  short  time.  It  should  be 
observed  that,  in  bringing  the  electrode  into  the  tube  A,  the 
entering  of  some  air  could  not  be  avoided.  Hence  the  potential 
is  initially  too  negative. 


THE  HYDROGEN  ELECTRODE.  9I 

When  now,  after  equilibrium  was  reached  in  0.1  A''  HCl,  this 
solution  was  replaced  by  1.0  TV  HCl,  which  had  been  saturated 
with  hydrogen  in  the  tube  B,  the  potential  was  immediately 
-f- 0.343.  In  the  same  way,  when  1.0  N  HCl  was  replaced  by 
0.1  A''  HCl,  the  potential  was  immediately  +0.401.  These  are 
sensibly  the  equilibrium  potentials. 

From  this  observation  it  follows  that,  when  a  hydrogen  elec- 
trode, saturated  with  hydrogen,  is  in  equilibrium  with  0.1  A''  HCl, 
it  is  in  the  same  state  in  equilibrium  with  1.0  A?"  HCl,  and  vice 
versa.  This  is  no  longer  the  case  when  a  solution  of  an  alkali  is 
used  in  place  of  an  acid. 

If  a  hydrogen  electrode,  which  is  in  equilibrium  with  1.0  A^ 
NaOH,  is  brought  into  0.1  AT  NaOH,  the  potential  is  at  the 
beginning  some  20  millivolts  too  positive,  and  reaches  after  some 
time  the  equilibrium  potential  for  0.1  N  NaOH. 

On  the  other  hand,  when  an  electrode,  which  is  in  equilibrium 
with  0.1  A''  NaOH,  is  brought  into  contact  with  a  solution  of 
1.0  A''  NaOH,  its  potential  is  at  first  some  20  millivolts  too  nega- 
tive and  falls  more  or  less  slowly  to  the  equilibrium  potential. 
The  same  phenomenon  is  observed,  and  in  a  more  marked  degree, 
when  a  hydrogen  electrode  is  brought  from  0.1  A^  NaOH  into 
0.1  N  HCl. 

It  is  evident  that  a  hydrogen  electrode  in  an  acid  should  behave 
otherwise  than  in  an  alkali,  since  in  an  acid  the  only  active  sub- 
stance on  the  electrode  is  hydrogen,  while  in  sodium  hydroxide 
sodium,  also  may  be  electromotively  active. 

If  metallic  sodium  forms  a  solid  solution  with  palladium,  the 
electrode  cannot  be  in  equilibrium  with  a  solution  of  sodium 
hydroxide,  unless  it  contains  metallic  sodium  at  a  certain  concen- 
tration, which  is  determined  by  the  hydrogen  potential,  i.  e.,  by 
the  hydroxyl  ion  concentration,  and  also  by  the  sodium  ion  concen- 
tration. The  electrode  is  therefore  a  sodium  electrode  as  well  as 
a  hydrogen  electrode. 

Let  us  suppose  that  the  potential  of  the  sodium-palladium 
electrode  is  a  logarithmic  function  of  the  sodium  content,  then  the 
potential  will  be  given  by  an  equation  of  the  form : 

Ej,,  =  e^,  —  0.058  log.o  C^.,   +  0.058  log.o  C^.*  (1) 


92  A.    H.    W.    ATEX. 

where  C^^  denotes  the  concentration  of  the  metalHc  sodium  in 
the  palladium,  and  C^^+  the  concentration  of  sodium  ions  in  the 
solutions.     e,sa  is  a  constant. 

The  potential  of  the  hydrogen  electrode  is  given  by 

Eh  =  ^oH  —  0.058  log.o  CoH-  (2) 

When  we  put  C^j  =  1  for  an  electrode  which,  both  with 
respect  to  hydrogen  and  to  sodium,  is  in  equilibrium  with  a 
solution  for  which  Cx^*  =  1  and  Cqh"  =  1.  then  it  follows, 
while  E.N-a  must  be  equal  to  Eh) 


Hence  the  sodium  concentration  of  an  electrode  in  equilibrium 
with  a  solution  of  sodium  hydroxide  of  the  concentration  Cqh- 
and  Cx-+  must  be  : 

Cxa    ^==    Cx-j     •    CoH"  (3) 

If  now  an  electrode,  which  is  in  equilibrium  with  a  solution  of 
the  concentration  C^/  •  C^,j^-  is  brought  into  contact  with  a 
solution  of  the  concentration  C^^-"  •  Cqh-  its  hydrogen  potential 
will  be  : 

Eh  =  e  —  0  058  log.o  C;h-  (4) 

and  its  sodium  potential 

Esa  =  e  —  0.058  log.o  Csa-  CoH-    +  0.058  log.o  C;.,^       (5) 

and  the  difference: 

E,.,  -  Eh  -  0.058  log.o  -^^^^^""r  (6) 

If  thus  an  electrode,  charged  with  hydrogen  at  one  atmosphere, 
is  brought  from  0.1  A^  HCl  into  0.1  N  XaOH,  the  electrode  is  not 
in  equilibrium  with  this  latter  solution,  because  it  contains  no 
sodium.  It  will  therefore,  in  an  alkaline  solution,  lose  hydrogen 
and  take  up  sodium,  according  to  the  equation : 

H  +  Na+  -   H+  +  Na 

If  this  reaction  takes  a  certain  time,  the  electrode  will  at  first 
be  too  negative,  and  approach  the  equilibrium  potential,  as  the 
above  given  reaction  proceeds. 


THE  HYDROGEN   ELECTRODE. 


93 


In  the  same  way,  if  an  electrode  is  brought  from  1.0  A''  NaOH 
into  0.1  A''  NaOH,  it  will  be,  according  to  the  equations  (4)  and 
(5)  0.058  volt  more  positive,  if  it  behaves  fully  as  a  sodium 
electrode,  and  0.058  volt  more  negative,  if  it  acts  fully  as  a 
hydrogen  electrode.  Now  neither  of  these  is  probable,  so  one  will 
find  a  value  that  lies  between  the  potential  of  the  sodium  electrode, 
and  that  of  the  hydrogen  electrode.  In  any  case,  however,  the 
potential  must  at  first  be  more  positive  than  the  equilibrium 
potential. 


0  40 


.0 


O  4i 


-■  ■   I       1 
riG.  2 

^^^ 

5^ 

A 

f 

/ 

' 

0-4. 

J 

/ 

1 

1 

r 

/ 

/ 

SO 


/20 
Time  in  Minutes 


/eo 


Fig.  2  shows  two  curves  which  give  the  potential  as  a  function 
of  time  for  two  electrodes  which  had  been  in  contact  with  a 
solution  of  sodium  hydroxide,  and  were  then  brought  into  0.1  A/' 
HCl.  "A"  relates  to  an  electrode  covered  with  platinum  black, 
that  had  been  in  contact  with  1  A''  NaOH  for  three  days,  during 
which  time  a  slow  current  of  hydrogen  was  passed  through  the 
apparatus.  "B"  is  the  curve  for  an  electrode,  covered  with  palla- 
dium black,  which  was  left  in  0.1  A''  NaOH  for  thirty  hours. 
Both  electrodes  were  completely  immersed  in  the  liquid.  It  is  seen 
that  the  potential  is  at  first  some  40  or  50  millivolts  too  positive 
and  that,  after  an  hour,  the  equilibrium  potential  is  approached, 
though  not  yet  fully  reached. 

The  same  behavior  is  found  when  an  electrode  is  brought  from 
a  stronger  alkaline  solution  into  a  weaker  one.  In  Fig.  3  the  curve 


94 


A.    H.    W.    ATEN. 


"C"  represents  the  potential  for  an  electrode  which  has  been  in 
contact  with  0.1  A^  NaOH  and  is  then  put  into  a  0.01  A'^  solution. 
Curve  "D"  gives  the  potential,  when  a  1.0  A^  solution  of  NaOH 
is  replaced  by  a  0.1  iV  solution.  In  the  latter  case  the  potential 
was,  after  an  hour,  still  0.008  volt  too  positive.  Next  morning 
the  equilibrium  potential  was  reached.  The  total  difference 
between    the    potential    immediately    found   and    the    equilibrium 


I.  OS 


I 


$ 


i.lO 


1 

riG.  3. 

/ 

V—      C 

f 

/ 

I 

^ 

41 



.Z)_. 

'-"""" 

/ 

60  /go 

Time    /n   Minutes 


/eo 


potential  is  here  less  than  for  the  curve  "C,"  because  this  electrode 
had  been  in  contact  with  the  1.0  A/"  NaOH  for  two  hours  only, 
while  the  electrode  C  remained  thirty  hours  in  the  0.1  A^  solution. 
The  reverse  is  observed  when  an  electrode  is  brought  from 
an  acid  solution  into  an  alkaline  solution,  or  from  a  weaker  alka- 
line solution  into  a  stronger  one.  This  is  shown  by  the  curves  in 
Fig.  4,  of  which  "E"  relates  to  an  electrode  which  was  brought 
from  0.1  N  HCl  into  0.1  N  NaOH.  "F"  gives  the  potential, 
when  passing  from  a  0.01  A^  solution  of  NaOH  to  a  1  A^  solution. 


THE  HYDROGEN  ELECTRODE. 


95 


and  "G"  when  a  solution  of  0.01  A^  NaOH  is  replaced  by  a  0.1  iV 
solution. 

The  same  phenomena  are  observed  when  the  electrode  has  been 
charged  with  hydrogen  and  sodium  by  cathodic  polarization.  The 
apparatus,  shown  in  Fig.  5,  permits  polarizing  an  electrode  "A" 
in  a  solution  of  NaOH,  while  a  current  of  hydrogen  is  passed 
through  this  solution,  and  through  a  second  solution,  contained 


t.os 

V 

\ 

nQ.<i 

\ 

•— 

^ 

,  v 

■^ 

* 

•  Q 

1.  lO 

5 

\ 

i 

\ 

-_ 

"• * 

r 

1.15 

o 

io 

/ 

zo 

Timv    in  Minuter 


in  the  tube  B,  which  is  in  this  way  freed  from  oxygen.  When 
the  electrode  "A"  is  polarized  for  10  min.  with  a  current  of  10 
milliamperes  and  the  current  is  then  broken,  the  potential,  immedi- 
ately after  polarization,  is  found  too  positive. 

The   deviation   from  the  equilibrium   potential   increases   with 
increasing  dilution  of  the  alkali  as  follows: 


Concentration  of  NaOH 

1.0  A^ 
0.49  A^ 
0.21  A^ 
0.083  N 


Deviation     millivolts 

0 

4 

6 

13 


96 


A.    H.    W.    ATEX. 


This  may  be  explained  by  observing  that  during  the  electro- 
lysis the  liquid  in  contact  with  the  electrode  is  more  strongly 
alkaline  than  the  bulk  of  the  liquid,  because  of  the  discharge  of 
hydrogen  ions  at  the  cathode.  This  concentration  polarization 
gives  the  cathode  a  too  positive  potential.  In  consequence  the 
electrode  takes  up  more  sodium  from  the  solution  than  corre- 
sponds to  the  equilibrium  potential.  The  electrode  remains, 
therefore,  some  time  after  polarization,  too  positive. 


^\^': 


'^ 


As  the  relative  increase  in  alkalinity  during  electrolysis  is 
greatest  for  the  more  dilute  solutions,  the  effect  on  the  potential 
will  be  greatest  in  solutions  with  small  alkali  content. 

When  now,  a  short  time  after  polarization,  the  solution  is 
diluted  with  its  own  volume  of  water,  the  potential  is  again 
found  too  positive.  This  can  be  ascribed  to  the  fact  that  the 
electrode  is  still  too  positive  as  a  result  of  the  polarization. 

If,  on  the  other  hand,  the  electrode  was  polarized  in  0.005  A^ 


THE  HYDROGEN  ELECTRODE.  97 

NaOH,  and  then,  immediately  after  polarization,  a  strong  solu- 
tion of  NaOH  was  introduced,  the  potential  was  found  22  milli- 
volts too  negative,  when  the  resulting  solution  was  0.3  A^.  This 
experiment  is  more  decisive,  because  it  shows  that  the  electrode, 
though  at  first  too  positive,  becomes  too  negative  by  greatly 
increasing  the  alkali  content.  The  most  certain  proof  that  a 
hydrogen  electrode  in  alkali  acts  partly  as  a  sodium  electrode 
would  be  given  if  it  were  possible  to  show  that  in  increasing  the 
alkali  content  the  electrode  takes  temporarily  a  more  negative 
potential.  This  would  mean  that  by  increasing  the  alkali  content 
ten  times,  the  electrode  should  be  more  than  58  millivolts  too 
negative.    So  great  a  value  was  never  found. 

One  can  however  leave  the  hydroxyl  ion  concentration  almost 
unchanged,  and  increase  the  sodium  ion  concentration  by  adding 
a  strong  solution  of  NaCl  to  a  weak  solution  of  alkali. 

In  this  case  the  potential  of  an  electrode  must,  according  to 
equation  (5)  become  more  negative  if  it  behaves  as  a  sodium 
electrode,  and  remain  approximately  constant  if  it  acts  as  a 
hydrogen  electrode.  Now  it  was  found  that  the  potential  of  an 
electrode,  covered  with  smooth  palladium,  after  polarization  in 
0.01  N  NaOH  fell  as  much  as  30  millivolts,  when  a  strong 
solution  of  sodium  chloride  was  added,  so  as  to  make  the  liquid 
0.2  A^  with  respect  to  sodium  chloride. 

When  the  electrode  is  polarized  in  a  solution  containing 
NaCl  -|-  NaOH,  the  potential  after  polarization  is  about  20 
millivolts  more  positive  than  in  a  solution  which  is  only  0.01  A^ 
to  NaOH,  corresponding  to  a  greater  sodium  content  of  the 
electrode.  Now  it  is  evident  that  the  electrode  should  take  up 
more  sodium  in  this  latter  experiment,  because  polarization  in  a 
solution  of  NaOH  with  a  great  excess  of  NaCl  gives  rise  to  a 
stronger  alkalinity  at  the  cathode  than  in  NaOH  alone.  In  the 
first  case  the  hydroxyl  ions  are  removed  by  diffusion  only,  in  the 
second  case  by  diffusion  and  by  the  current. 

The  same  experiment  was  repeated  with  an  electrode  covered 
with  palladium  black,  because  smooth  palladium  contains  a  rela- 
tively small  quantity  of  hydrogen,  and  the  electrode  is  therefore 
very  sensitive  to  oxygen.  So  a  trace  of  oxygen,  contained  in 
the  solution  of  sodium  chloride,  might  give  the  electrode  a  too 


^8  A.    H.   W.  ATEN. 

negative  potential.  Here  a  saturated  solution  of  NaCl  was  used, 
which  had  been  boiled,  and  saturated  with  hydrogen  for  two 
hours.  Only  a  small  quantity  of  this  liquid  was  introduced  into 
0.01  N  NaOH.  The  potential  fell  immediately  from  +1.043 
to  -1-1.012  volt  and  rose  afterwards  very  slowly  to  the  equi- 
librium potential. 


A  paper  presented  at  the  Forty-third 
General  Meeting  of  the  American  Elec- 
trochemical Society  held  in  New  York 
City,  May  3.  1923,  Dr.  Wm.  C.  Horsch 
in   the   Chair. 


THE  REACTIONS  OF  THE  LEAD  STORAGE  BATTERY' 

By   M.   Knobel- 

Abstract, 

The  theories  of  the  lead  storage  battery  are  discussed.  The 
results  of  the  author's  investigations,  together  with  the  work 
of  IMacInnes,  Adler  and  Joubert,  complete  the  evidence  in 
favor  of  the  Gladstone  and  Tribe  theory  of  the  reactions  in  the 
lead  storage  battery.  The  claim  of  Fery  that  only  one  mol  of 
sulfuric  acid  is  used  per  two  faradays  on  discharge  is  not  sup- 
ported, which  tends  to  disprove  the  theory  he  proposes. 


The  theory  of  the  reactions  of  the  lead  storage  battery  proposed 
by  Fery^  is  represented  by  the  following  equation  for  discharge: 

Pb  +  H,SO,  +  Pb,0,  =  PbSO,  +  H.O  +  3PbO,      (1) 

or  possibly 

Pb  +  H3SO,  +  PboOa  =  PbSO,  +  H2O  +  PbO,    (2) 

It  differs  from  the  generally  accepted  theory  of  Gladstone  and 
Tribe*  represented  by  the  equation 

Pb  +  PbO,  +  2H,SO,  3=  2PbSO,  +  2H2O  (3) 

in  the  supposition  that  there  is  a  higher  oxide  of  lead  than  the 
peroxide  on  the  anode  which  changes  over  to  the  peroxide  on 
discharge.     A  consequence  of  this  supposition  is  that  only  one 

^  Manuscript  received  November  22,  1922. 

3  Contribution  from  the  Rogers  Laboratory  of  Physics,  Electrochemical  Labora- 
tory.  Massachusetts  Institute   of  Technology. 

»Lumiere  Elec,  34,  305  (1916);  J.  Physique,  6,  21  (1916);  Bull.  Soc.  d'en.  Ind. 
Nat.,    118,   92    (1919). 

*  "The  Chemistry  of  the  Secondary  Batteries  of  Plante  and  Faure,"  MacMillan,  1883. 

For  a  discussion  of  this  reaction  see  also  Dolezalek  "The  Theory  of  the  Lead 
Accumulator,"   Translated  by   Von    Ende. 

99 


loo 


M.    KNOBEL. 


mol  of  sulfuric  acid  should  be  used  for  every  two  faradays  on 
discharge,  instead  of  two  mols,  as  Gladstone  and  Tribe's  theory 
requires.  Fery  supports  his  theory  by  the  claims  that  the  higher 
oxide  of  lead  can  be  shown  to  exist  on  the  fully  charged  plate 
by  chemical  analysis ;  that  lead  peroxide  will  not  give  the  potential 
of  the  storage  battery  anode ;  and  that  only  one  mol  of  sulfuric 
acid  per  two  faradays  is  actually  used  on  discharge.  The  careful 
experiments  of  Maclnnes,  Adler  and  Joubert^  cast  considerable 
doubt  on  the  first  two  of  these  claims.  They  found  by  analysis 
that  the  material  on  the  positive  plate  coincided  closely  in  composi- 
tion with  lead  peroxide.  They  also  found  that  PbO,  formed 
chemically  and  electrochemically  on  a  platinum  sheet  gave  the 
same  electromotive  force  in  sulfuric  acid  as  the  storage  battery 
anode,  contrary  to  the  experiments  of  Fery. 


Table  I. 


Amp. 

Time  of  discharge  in  hr. 

* 

5 

6 

1.40 

4 

9 

1.48 

3 

14 

1.56 

2 

24 

1.69 

1 

59 

1.79 

In  regard  to  the  last  mentioned  claim  of  Fery,  existing  data  are 
at  variance  and  are  quite  inconclusive.  Maclnnes,  Adler  and 
Joubert  found  that  a  quantity  of  sulfuric  acid  between  one  and 
two  mols  per  two  faradays  is  used  on  discharge,  the  values  vary- 
ing from  1.34  to  1.79  mols.  W.  Kohlrausch  and  Heim"  conclude 
that  2  mols  per  two  faradays  are  used,  but  their  work  is  of  doubt- 
ful accuracy.  Scheneck  and  Farbaky^  find  on  the  average  1.23  mols 
for  the  same  quantity.     The  work  of  Pfai¥^  in  this  connection  is 

=    Trans.  Am.  Electrochem.  Soc,  37,  641   (1920). 

•  Electrotechnische    Zeit.,    10,    327    (1889).      These    authors    used    a    hydrometer    for 
the  specific  gravity  determinations   and   no  mention   is  made   of  temperature  control. 
^  Dingler's   Polytech.  Jour.,  257,   357   (188S). 
8  Centralblatt  fur.  Accumulatoren,   11,  73,   173   (1901). 


THE  REACTION'S   OF   THE   LEAD  STORAGE  BATTERY.  lOI 

the  most  useful,  in  that  he  has  taken  into  account  the  other 
variables  which  might  affect  the  quantity  of  HoSO^  used.  He 
determined  the  quantity  of  acid  used  per  two  faradays  (which 
will  hereafter  be  designated  by  (/>)  at  successive  intervals  during 
the  course  of  a  single  discharge,  and  also  during  several  discharges 
at  different  current  densities.  The  latter  data  are  reproduced  in 
Table  I. 

These  figures  are  significant,  showing  as  they  do  how  greatly  <^ 
varies  with  the  current.  The  tendency  of  <^,  as  seen  best  from  a 
curve,  is  to  approach,  at  sufficiently  small  current  densities,  the 
theoretical  value  two,  required  by  the  Gladstone  and  Tribe  equa- 
tion. On  account  of  the  importance  of  this  point  in  determining 
which  reaction  takes  place  in  the  battery  it  was  thought  desirable 
to  ascertain  whether  ^  would  not  be  found  equal  to  2  at  very  small 
current  densities. 

EXPERIMENTAL. 

The  first  experiments  were  made  with  a  positive  and  two  nega- 
tive pasted  plates^  of  the  following  specifications : 

Dimensions,  14.5  cm.  wide,  12.5  cm.  in  height  and  0.25  cm. 
thick ;  weight  of  unpasted  grid  225  g. ;  weight  of  dry  pasted  posi- 
tive 366  g. ;  and  of  dry  negatives  347  g. ;  rated  capacity  20  amp.-hr. 
They  were  put  in  a  jar  of  such  a  size  that  about  1,000  g.  of  elec- 
trolyte just  covered  the  plates.  A  rubber  cover  was  used  on  the 
cell  to  prevent  loss  of  electrolyte.  A  copper  coulometer  was  used 
to  determine  the  number  of  faradays  passed.  The  specific  gravity 
determinations  were  made  at  25°  C.  with  a  10  cc.  pycnometer  and 
the  percentages  of  sulfuric  acid  were  interpolated  from  the  tables 
of  Landolt  and  Bornstein.  From  one-half  to  two  hours,  with 
occasional  stirring,  was  allowed  for  the  electrolyte  to  become 
uniform  in  composition  after  a  run,  the  longer  time  being  for 
rims  at  higher  currents.  All  measurements  were  made  at  a 
temperature  of  26° C.  ±  2°C. 

The  results  on  these  pasted  plates  are  listed  in  the  first  six  runs 
of  Table  II,  the  headings  of  which  are  self-explanatory.  In  every 

»  These  plates  vere  obtained  through  the  kindness  of  the  .American  Storage  Battery 
Co.,  manufacturers  of  the  "Harvard"  battery  and  are  designated  by  them  as  Type  A. 


I02 


M.    KNOBEL. 


case  the  values  of  <f>  were  found  approximately  equal  to  two/° 
including  the  run  at  10  amperes,  which  is  five  times  the  rated 
current  for  the  cell.  The  high  values  of  (f>  in  the  first  two  runs 
are  probably  due  to  local  action  of  the  acid  on  the  new  grids.  The 
deviations  in  the  calculated  value  of  cj>  may  be  as  large  as  3  per 
cent,  for  although  the  density  determinations  are  comparatively 
precise,  the  quantity  of  acid  consumed  is  determined  as  the 
dift'erence  in  two  large  numbers  and  the  percentage  error  in  the 
difference  becomes  large. 

Table  II. 
Consutnption  of  H^SO^  in  Storage  Battery  Discharge 


Time 

Weight 

of  dis- 

Density 

of  acid 

of 

Weight 

Weight 

Run 

Current 
amp. 

charge 

hr. 

(ap- 
prox). 

begin- 
ning 

end 

trolyte 

g. 
(begin- 
ning) 

of 

HaSO* 

used 

Faradays 

H2SO4 

theo- 
retical 

* 

1 

0.10 

53.0 

1.1245 

1.1106 

986 

21.1 

0.2018 

19.79 

2.13 

2 

0.50 

20.0 

1.1299 

1.1031 

950 

39.0 

0.3719 

36.47 

2.14 

3 

1.0 

16.0 

1.1394 

1.1000 

967 

58.3 

0.5905 

57.91 

2.01 

4 

2.0 

4.0 

1.1440 

1.1230 

957 

31.2 

0.3159 

30.98 

2.01 

5 

5.0 

2.5 

1.1465 

1.1187 

949 

41.8 

0.4218 

41.37 

2.02 

6 

10.0 

0.7 

1.1872 

1.1772 

1400 

21.0 

0.2093 

20.55 

2.04 

7 

0.5 

10.0 

1.1775 

1.1679 

920 

14.5 

0.1760 

17.26 

1.68 

8 

0.5 

12.5 

1.1679 

1.1561 

906 

18.2 

0.2319 

22.75 

1.60 

9 

0.5 

5.0 

1.1561 

1.1537 

888 

3.5 

0.0724 

7.11 

0.98 

10 

0.5  to 
0.1 
0.1 

16.0 

1.1537 

1.1522 

882 

2.1 

0.0695 

6.80 

0.62 

11 

65.0 

1.1893 

1.1741 

941 

23.0 

0.2445 

24.00 

1.92 

Run  8  is  a  continuation  of  run  7;  run  9,  of  run  8;  run  10,  of  run  9. 


Experiments  were  made  next  with  a  Plante  type  positive^^ 
(Manchester  plate)  of  the  same  superficial  area,  with  the  same  two 
pasted  negatives.  The  results  on  this  plate  are  listed  in  Runs  7  to 
11  of  Table  II.  Runs  7  to  10  are  for  successive  periods  in  a  single 
discharge,  and  it  is  seen  that  the  quantity  of  acid  used  decreases 

•"  <<>  was  calculated  from  the  equation  : 

_    WP  (a  —  b) 
*  ~  C  (49  -  40b) 
in    which    W    is    the    weight    in    grams    of    the    electrolyte    before    the    run,    F    equals 
96,500,    C   is   the  number   of   coulombs    passed,    a   and   b   are   percentages    (x   0.01)    of 
HjSOi  before  and  after  the  run  respectively. 

'1  Maclnnes,  Adler  and  Joubert  used  a  Manchester  plate. 


THE  REACTIONS  OF  THE  LEAD  STORAGE  BATTERY.  I05 

constantly  as  the  discharge  continues.  Run  11  is  at  a  lower  cur- 
rent density  than  Runs  7  to  10  and  for  the  same  part  of  the  dis- 
charge as  Run  8.  It  is  seen  that  the  value  of  </>  (1.92)  at  the 
lower  current  is  much  nearer  to  2  than  at  the  higher  current,  which 
confirms  the  results  of  Pfaff  previously  given.  It  was  thought 
unnecessary  actually  to  discharge  the  Manchester  plate  at  a 
current  low  enough  to  make  <f>  equal  to  2  since  the  experiments  on 
the  pasted  plate  showed  that  2  is  the  correct  value. 

DISCUSSION  OF  RESULTS. 

The  explanations  which  have  been  suggested  for  the  small 
amount  of  acid  used  on  discharge,  with  the  exception  of  the 
theory  of  Fery,  are  based  primarily  on  the  supposition  that  there 
exists  a  lack  of  sulfate  ions  in  the  pores  of  the  positive  plate.  The 
present  results  confirm  this  hypothesis.  On  the  pasted  plate  the 
material  is  so  porous  and  the  reaction  proceeds  to  such  a  small 
depth  that  a  large  concentration  difference  of  acid  cannot  exist. 
While  the  sulfate  ions  migrate  out  of  the  plate,  they  are  easily 
replaced  by  dilifusion. 

The  construction  of  the  ^Manchester  plate,  with  the  inserted 
lead  ribbon  buttons,  however,  is  such  as  to  produce  ideal  condi- 
tions for  the  depletion  of  sulfate  ions  in  the  inner  parts  of  the  plate. 
The  acid  can  diffuse  only  very  slowly  into  the  narrow  and  deep 
channels  of  the  button.  This  effect  should  become  greater  as  the 
discharge  continues  and  the  reaction  proceeds  farther  into  the 
plate  and  the  experimental  values  of  </>  do  in  fact  decrease  as  the 
discharge  continues  as  seen  in  Runs  7  to  10.  At  lower  current 
densities  the  migration  outward  of  the  ions  is  less  and  the  time 
for  diffusion  of  acid  back  into  the  plate  is  greater  so  that  the  con- 
centration decrease  in  the  pores  should  be  less  than  at  higher  cur- 
rents. The  high  value  of  <f>  in  Run  11  confirms  the  above  state- 
ment. It  is  probably  only  on  the  Manchester  or  similar  plate 
which  contains  such  deep  channels  that  the  above  effect  will  be 
observed. 

Just  what  reaction  takes  place  when  less  than  the  theoretical 
amount  of  acid  is  used  is  still  somewhat  uncertain.    It  is  probable 


I04 


DISCUSSION. 


that  a  mixture  of  lead  oxide  and  lead  sulfate  is  formed,  varying 
in  proportions  as  the  lack  of  sulfate  ions  becomes  greater.  When 
this  condition  exists  the  free  energy  of  the  reaction  will  be  less 
and  the  voltage  of  the  cell  should  become  less.  On  the  pasted 
anode,  the  voltage  was  constant  until  near  the  end  of  the  discharge, 
when  it  dropped  rapidly.  With  the  IManchester  plate,  however,  the 
decrease  in  voltage  during  discharge  was  much  more  gradual,  and 
a  large  fraction  of  the  discharge  (Run  10)  was  obtained  after  the 
voltage  had  fallen  below  1.7.  A  part  of  this  decrease  in  voltage 
is  due,  of  course,  to  the  decrease  in  acid  concentration,  and  part  to 
the  increasing  resistance  of  the  cell  on  discharge,  but  a  part  may 
be  due  to  the  decreased  free  energy  of  the  reaction. 

SUMMARY. 

It  has  been  found  that  under  proper  conditions  the  amount  of 
acid  used  on  discharge  of  the  lead  storage  battery  corresponds  to 
that  required  by  the  Gladstone  and  Tribe  theory.  The  evidence  in 
favor  of  the  latter  theory  is  thus  completed  and  the  theory  pro- 
posed by  Fery  appears  untenable. 


DISCUSSION. 


Helen  Weir^  :  May  I  ask  Dr.  Knobel  what  evidence  he  has 
that  any  other  compound  is  formed  than  lead  sulfate,  as  a  part 
in  the  discharge? 

M.  Knobel:  From  the  number  of  faradays  passed  through  the 
battery,  the  amount  of  lead  peroxide  reduced  is  known,  but  less 
acid  is  used  than  there  would  be  if  only  lead  sulfate  were  formed. 

Helen  Weir:  As  I  understand  no  analyses  have  been  made; 
is  there  not  a  possibility  of  an  experimental  error  there? 

M.  Knobel:  Since  the  value  of  <f>  came  out  as  low  as  0.6, 
the  percentage  error  would  have  to  be  as  large  as  70  per  cent, 
which  is  improbable. 

1  Union  Carbide  and  Carbon  Res.  Lab..  Long  Island  City,  N.  Y. 


THE  REACTIONS  OF  THE  LEAD  STORAGE   BATTERY.  1 05 

Helen  Weir:  You  will  remember  Dolzalek  substantiated  the 
double  sulfate  theory  by  thermo-dynamic  relations,  and  those 
relations  need  no  modification  to  support  the  theory. 

M,  Knobel:  I  do  not  see  that  thermodynamic  calculations 
have  anything  to  do  with  the  present  question;  it  is  a  matter  of 
Faraday's  Law  and  the  quantity  of  materials  used. 

Helen  Weir:  But  in  the  slow  discharges  where  you  have 
the  most  ideal  condition  for  diffusion  and  can  get  the  most 
accurate  determination,  you  come  nearest  to  finding  that  you  do 
use  two  mols  per  faraday.  Do  you  not  think  that  this  is  evidence 
that  a  lead  sulfate  is  formed  rather  than  an  intermediate  or 
additional  compound? 

On  your  lower  rates  of  discharge  your  sulfate  is  formed 
throughout  the  pores  of  the  plate  and  into  the  interior,  so  that 
you  can  get  an  ideal  condition  for  your  determination,  but  on  a 
high  rate  where  you  have  it  formed  on  the  surface,  it  is  possible 
that  you  may  have  sulfuric  acid  trapped  in  the  pores  which 
diffuses  out  very  slowly. 

M.  Knobel  :  In  answer  to  that  point  I  may  say  that  when  the 
density  determination  was  made  one  hour  after  the  run  and  then 
again  two  hours  later,  there  was  found  to  be  no  appreciable 
difference. 

Helen  Weir:  How  do  you  explain  recuperation  then?  At 
any  high  rate  of  discharge  you  can  get  remarkable  recuperation 
in  two  hours,  due  presumably  to  diffusion  of  acid. 

M.  Knobel:   That  is  in  the  voltage. 

Helen  Weir:  Ampere-hour  capacity  also.  It  is  an  inter- 
esting phase  of  the  storage  battery  subject,  and  I  would  like  to 
see  some  more  work  on  it. 

M.  Knobel  (Communicated)  :  A  further  point  may  be  men- 
tioned in  regard  to  Mrs.  Weir's  question  of  the  possibility  of 
experimental  error.  If  acid  were  trapped  in  the  pores  of  the 
plate  and  came  out  slowly,  a  density  determination  made  a  long 


I06  DISCUSSION. 

time  after  the  run  should  be  larger  than  one  made  just  after  the 
run  and  would  therefore  result  in  a  still  smaller  value  oi  eft . 

Recuperation  may  be  explained  by  diffusion  of  the  acid  back 
into  the  plates.  The  first  result  of  this  would  be  to  build  up  the 
voltage.  The  capacity  would  then  increase,  but  only  because  the 
voltage  has  increased,  the  capacity  being  limited  by  an  arbitrary 
low  voltage. 


A  patter  presented  at  the  Forty-third 
General  Meeting  of  the  American  Elec- 
trochemical Society  held  in  New  York 
City,  May  3,  1923,  Dr.  Wm.  G.  Horsch 
in    the   Chair. 


ELECTROLYTIC  AND  CHEMICAL  CHLORINATION  OF  BENZENE.' 

By  Alexander  Lowy  and  Henry  S.  Frank. ^ 

Abstract. 

One  of  the  processes  to  which  the  electrolytic  method  applies  is 
the  chlorination  of  benzene,  and  it  seemed  advisable  to  investigate 
this  reaction  because  (1)  only  a  limited  amount  of  work  has  been 
done  on  this  subject,  (2)  reports  in  the  literature  are  often 
contradictory,  and  (3)  investigation  of  this  kind  might  throw 
additional  light  on  the  mechanism  of  chlorination. 


HISTORICAL. 

A  considerable  amount  of  research  work  has  been  done  on  the 
chlorination  of  benzene.  A  number  of  important  references  for 
the  chlorination  of  benzene  by  the  chemical  method,  under  variable 
conditions,  are  given  below.^ 

Concerning  chlorination  of  benzene  by  the  electrolytic  method, 
Miihlhofer*  stated  that  benzene  is  not  appreciably  affected  by  elec- 
trolytic chlorine.  Miihlhofer  found,  however,  that  when  20  g. 
toluene  was  stirred  vigorously  into  250  cc.  cone.  HCl,  and  the 
mixture  electrolyzed,  a  good  current  yield  of  chlorotoluene  was 
obtained  (30  per  cent  ortho,  70  per  cent  pasa).    The  addition  of 

'  Manuscript  received  January  26,   1923. 

*  Contribution  from  the  Dept.  of  Chemistry,  University  of  Pittsburgh,  Pittsburgh,  Pa. 
»Pogg.    Ann.    29,   231    (1833);    31,   283    (1834);    Ann.    chim.   26,   59    (1834);    63,    41 

(1836);  Comptes  rendus  des  travaux  de  chimie  (1849)  429;  Proc.  Royal  Soc.  7,  18.  94 
(1854);  J.  C.  S.  15,  41  (1862);  16,  76  (1863);  Ann.  chim.  (4)  15,  186  (1868);  Ber.  8, 
1400  (1875);  Ann.  225,  199  (1884);  Compt.  rend.  127,  1,026  (1898);  126,  1212  (1898); 
170,  1319,  (1920);  Bull.  soc.  chim.  29,  283  (1903);  J.  Soc.  Chem.  Ind.  35,  1130  (1916); 
Chem.  Soc.  Proc.  24,  15-16  (1908);  Germ.  pat.  219,242;  U.  S.  Pat.  1,180,964;  U.  S. 
Pat.  1,189,736,  and  corresponding  foreign  patents;  Compt.  rend.  170,  319  (1920);  J. 
Am.  Chem.  Soc.  36,  1007-11   (1914). 

♦  Dissertation    "Uber   die    Einwirkung   elektrolytisch    erzeugter    Halogene    auf    Organ- 
ischen  Verbindungen."  Technische  Hochschule,   Munich,   (1905). 

I07 


I08  ALEXANDER  IvOWY  AND  HENRY  S.  FRANK. 

iodine  did  not  influence  the  products  in  nature  or  amount.   Benzyl 
chloride  was  not  formed. 

Schluederberg^  electrolyzed  benzene  in  ether  saturated  with 
ZnClo,  removing  76  to  80  per  cent  of  the  chlorine  from  inorganic 
combination.  When  benzene  was  floated  on  a  layer  of  Oettel's 
solution  (160  g.  NaCl,  140  g.  H0SO4  made  up  to  1  liter  with  water) 
and  electrolyzed,  the  rotating  anode  being  entirely  in  the  benzene 
layer,  an  efficiency  of  only  2.29  per  cent  was  obtained.  Analogous 
experiments  with  toluene  gave  efficiencies  ranging  from  4.07  to 
38.25  per  cent.  Of  the  chlorine  acting,  76  to  96  per  cent  substituted 
in  the  ring  and  the  rest  in  the  side  chain. 

Van  Name  and  Maryott^  electrolyzed  benzene  in  a  glacial  acetic 
acid  solution  of  LiCl,  and  found  that  chlorination  took  place.  It 
also  took  place,  however,  when  chlorine  was  bubbled  through  the 
solution  and  electrolysis  in  addition  to  this  bubbling  produced  no 
added  effect. 

Fichter  and  Glanzstein^  used  glacial  acetic  acid  to  prepare  a 
homogeneous  solution  containing  benzene  and  concentrated  aque- 
ous HCl.  Electrolysis  of  this  solution  gave,  under  various  condi- 
tions, chlorobenzene,  p-dichlorobenzene,  sym-tetrachloro  and  hexa- 
chloro  benzene  and  in  addition  pentachlorophenol  and  chloranil. 
The  relative  amounts  of  these  substances  present  in  the  product 
depended  regularly  upon  the  current  density  employed,  (in  addi- 
tion to  other  determining  conditions,  such  as  temperature)  which 
indicates  that  an  electrochemical  reaction  was  taking  place. 

Neminski  and  Plotnikow*  electrolyzed  the  molecular  compound 
AlBrg  .  SCeHg,  and  observed  that  the  hydrocarbon  separated  at 
the  cathode,  and  bromination  took  place  at  the  anode. 

Some  additional  light  on  whether  or  not  the  chlorination  is  elec- 
trochemical is  shown  by  the  work  of  Cohen,  Dawson  and  Cros- 
land.^  They  electrolyzed  toluene  with  a  carbon  anode  over  a  layer 
of  cone.  HCl.  As  in  Schluederberg's  later  work,  substitution  was 
largely  in  the  ring.  When,  however,  chlorine  was  bubbled 
through  the  same  mixture  under  the  same  conditions,  substitution 
was  almost  entirely  in  the  side  chain. 

»J.  Pliys.  Chem.  12,  595  (1908). 

«Am.  J.   Sci.   35,    130-70   (1913). 

'Ber.  49,  2473-89    (1916). 

«J.  Russ.  Phys.  Chem.  Ges.  40,  391-96  (1908). 

*J.  Chem.  Soc.  87,   1034   (1905). 


THE  CHLORINATION  OF  BENZENE.  I09 

THEORETICAI,. 

The  work  described  in  the  references  cited  above  seems  to  indi- 
cate that  in  the  absence  of  carriers,  chlorine  does  not  substitute 
for  hydrogen  in  benzene,  and  most  text  books  make  this,  or  an 
equivalent  statement.  They  then  define  a  carrier  as  a  substance 
which  catalyzes  the  substitution  of  halogens  into  the  ring  in  aro- 
matic hydrocarbons.  Most  lists  of  carriers  include  1,  Fe,  FeClg, 
SbCl3,  SbCIs,  M0CI5,  Al  or  AICI3,  PCl^,  S,  ZnCl^,  and  Sn.  For 
other  halogens  the  corresponding  halides  are  used. 

The  mechanism  of  the  action  of  carriers  seems  to  be  fairly 
well  agreed  upon;  the  carrier  forms  an  addition  compound  with 
the  halogen,  which  it  then  liberates  in  a  more  active  state  than 
before.  This  view  is  supported  by  the  fact  that  benzene  can  be 
chlorinated,  for  instance,  by  ICl,  ICI3,  FeCL,  SbClg,  M0CI5,  etc., 
in  the  absence  of  any  free  chlorine.  The  function  of  the  chlorine 
in  these  processes  seems  therefore  to  be  regeneration  of  the  original 
compound  (ICI3,  etc.). 

This  concept  of  the  mechanism  of  chlorination  has  been 
extended  by  Schluederberg^"  to  include  electrolytic  chlorination. 
as  follows:  Benzene  is  chlorinated  by  the  action  of  negatively 
charged  CI',  and  any  source  of  negatively  charged  chlorine  will 
therefore  act  as  a  chlorinating  agent.  He  points  out  that  each  of 
the  carriers  mentioned  dissociates,  furnishing  a  negative  chlorine, 
as: 

ICl  ->  P  -f  CI- 
FeClg  -»  FeCla  +  Cl- 
aud so  on.     Moreover,  the  fact  that  electrolytic  chlorine  may  be 
considered  negatively  charged  explains  why  benzene  can  be  chlori- 
nated by  electrolytic  methods. 

According  to  another  theory,  however,  halogens  that  substitute 
in  the  nucleus  in  aromatic  compounds  bear  a  positive  charge.  This 
is  derived  by  Fry"  from  his  electronic  conception  of  the  structure 
of  benzene,  and  his  argument  is  that  each  of  the  substances  men- 
tioned as  carriers  is  a  possible  source  of  positive  halogen.  He 
also  points  out  that  in  the  case  of  toluene,  the  presence  of  mois- 
ture promotes  substitution  in  the  ring,  anhydrous  halogens,  in  the 

"•J.  Phys.  Chem.  12,  S93,  (1908). 

11  Fry  "Electronic  Conception  of  Valence." 


no  ALEXANDER  LOWY  AND   HENRY   S.   FRANK. 

absence  of  carriers,  tending  to  substitute  in  the  side-chain.  He 
suggested  the  following  mechanism  to  explain  this:  the  chlorine 
(for  instance)  first  reacts  with  the  water 

CI2  +H,0  ±5  HCl  +  HOCl 

The  HOC^  then  acts  as  a  source  of  positive  chlorine  which 
replaces  a  positive  hydrogen  in  the  benzene : 

H" 


C 


+ 


H^C  C= 


H^  +   HO' 


-CI 
Cl^  >  I         I         +  H,0 


H^C^    /C^  H" 
C^ 


H' 


In  this  as  in  the  ordinary  interpretation  of  the  reaction,  half  the 
original  chlorine  is  converted  to  HCl. 

Work  done  in  the  course  of  this  investigation  confirms  Fry's 
hypothesis  that  water  may  act  as  a  halogen  carrier,  as  far  as 
chlorine  is  concerned.  With  chlorine  and  benzene  in  the  presence 
of  water,  substitution  was  obtained.  Under  identical  con- 
ditions, with  anhydrous  materials,  no  substitution  took 
place,  but  a  considerable  amount  of  benzene  hexachloride  was 
produced.  As  the  results  with  water  were  duplicated  when  light 
was  excluded  from  the  system,  there  was  obviously  no  photochemi- 
cal action,  and  the  eftect  was  due  entirely  to  the  water,  which 
probably  acts  as  suggested  by  Fry. 

A  point  in  favor  of  this  theory  is  the  fact  that  the  yield  was 
nearly  doubled  by  allowing  the  charge  to  stand  overnight  at  the 
conclusion  of  the  experiment.  The  additional  chlorination  here 
was  undoubtedly  effected  by  the  chlorine  that  had  been  in  solu- 
tion. 

According  to  the  above,  the  electrolytic  chlorination  cannot  be 
an  electrochemical  phenomenon  since  electrolytic  chlorine  at  the 
moment  of  liberation  is  negative.  According  to  Fry's  hypothesis, 
the  chlorination  in  the  electrolytic  cell  would  have  to  be  called  a 
secondary   reaction   depending   upon   the   previous   liberation   of 


THE   CHLORIXATIOX   OF   BENZENE.  Ill 

molecular  chlorine,  Clj,  and  possible  only  on  account  of  the  water 
present,  the  action  being, 

CI2   +  H2O  ~"^  HCl  ^  HOCl 


+   HOCl  ^   i      j    +  H2O 


If,  however,  the  chlorination  is  not  electrochemical,  it  is  difficult 
to  explain  the  fact  that  the  electrolytic  reaction  furnishes  some 
more  highly  chlorinated  products,  which  the  non-electrolytic  reac- 
tion does  not  give  under  the  same  conditions.  Indeed,  the  latter 
fact  points  strongly  to  direct  anodic  depolarization  as  the 
explanation  of  the  electrolytic  process. 

It  would  appear  that  the  true  explanation  of  what  takes  place 
in  the  chlorination  of  benzene  must  be  sought  further,  and  it  is 
not  hard  to  understand  the  disagreement  among  the  men  who 
have  chlorinated  benzene  electrolytically,  as  to  whether  the  phe- 
nomenon is  chemical  or  electrochemical :  the  results  obtained  point 
to  both  conclusions. 

The  addition  of  iodine  as  a  carrier  to  the  system  where  aqueous 
chlorination  was  taking  place  increased  the  extent  of  the  chlorina- 
tion, but  not  to  anything  like  the  degree  in  which  it  would  have 
done  so,  had  the  system  been  anhydrous. 

EXPERIMENTAL     PART. 

(A)     Electrolytic  Methods. 

The  electrolytic  experiments  were  conducted  in  the  apparatus 
represented  in  Fig.  1.  Particular  attention  is  called  to  the  water 
seal  arrangement,  and  to  the  glass  tube  by  which  the  porous  cup 
cathode  cell  was  suspended.  The  former  made  it  possible  to  intro- 
duce or  remove  a  charge  easily  and  quickly  without  interfering 
with  the  stopper,  which  was  permanently  set  up  and  insured  an 
entirely  gas-tight  system.  The  glass  tube  conducted  the  hydro- 
gen from  the  cathode  directly  to  the  outside  atmosphere,  prevent- 
ing the  formation  of  an  explosive  mixture  inside  the  cell. 

An  experiment  was  made  as  follows :  750  cc.  of  12  per  cent 


112 


ALEXANDER  LOWY  AND  HENRY  S.  FRANK. 


HCl  (sp.  gr.  1.06)  and  75  cc.  of  pure  benzene  were  placed  in  the 
beaker  (4  on  Fig.  1),  and  fitted  into  the  cell.  The  mixture  was 
agitated  vigorously  by  means  of  a  bell-type  stirrer,  and  current 


K)  4-  W'  >o'  r>'  a) 


0  > 


i  "^J 
^ 


V 


°-    *    t;  "^  "0  *: 


<  ^j  CD         ^ 

— •    /U    K)    't^    <0 


d  t 


was   passed.      Under    the   conditions    of    agitation   the    mixture 
resembled  an  emulsion. 

A  stream  of  air  was  maintained  through  the  system  by  means 
of  an  aspirator  bottle.    The  air  inlet  in  the  stopper  is  not  shown. 


THE  CHLORINATIOX   OF   BENZENE.  I  l3 

but  was  equipped  with  a  soda-lime  tube  through  which  the  air 
passed  before  entering.  The  possible  constituents  of  the  vapor 
passing  up  the  condenser  were  chlorine,  hydrogen  chloride,  ben- 
zene vapor,  oxygen,  nitrogen,  hydrogen  (perhaps  by  diffusion 
through  the  porous  cup),  water,  CO2,  CO.  The  two  latter  were 
formed  by  anodic  oxidation  of  the  benzene.  The  oil-tube,  (8), 
contained  heavy  lubricating  oil  cooled  to  0°  C,  the  function  of 
which  was  to  absorb  the  benzene  vapor.  Tube  (9)  contained 
aqueous  KI.  The  water  removed  any  HCl  in  the  vapors,  and  the 
KI  completely  removed  the  chlorine.  As  the  experiment  con- 
tinued, this  solution  gradually  colored  up  to  a  deep  red.  The  U 
tube  (10)  contained  NaOH  solution,  which  absorbed  COg.  The 
NaOH  was  about  0.5  A'',  of  undetermined  strength,  and  was  always 
present  in  excess.  At  the  conclusion  of  the  experiment  the  con- 
tents of  the  U  tube  was  washed  into  a  beaker,  and  the  solution 
neutralized  to  phenolphthalein  with  HCl  of  undetermined 
strength.  Two  drops  of  methyl  orange  were  then  added,  and  the 
solution  titrated  with  standard  HCl  to  a  fairly  strong  end-point. 
The  hydrochloric  acid  used  here  was  equivalent  to  half  the  CO2 
evolved  during  the  electrolysis. 

The  remaining  constituents  of  the  gas,  CO,  O2,  H,  and  Nj  were 
passed  over  hot  CuO  in  tube  (11),  where  the  CO  was  oxidized 
to  COj.  This  was  absorbed  by  an  excess  of  standard  Ba(OH)2 
in  the  wash-bottle  (12),  and  the  excess  back-titrated  to  phenol- 
phthalein with  standard  oxalic  acid.  This  gave  a  figure  for  the  CO 
produced  by  the  electrolysis.  The  use  of  two  oil-absorbing  tubes 
did  not  always  prevent  the  escape  of  some  benzene  vapor,  which 
was  then  burned  over  the  CuO  and  determined  as  CO.  For  this 
reason  the  CO  figure  is  not  always  trustworthy.  It  will  be  noted 
that  a  possible  escape  of  HCl  or  Clg  into  the  NaOH  in  tube  (10) 
could  not  harm  the  CO2  determination  as  long  as  sufHcient  NaOH 
was  present  to  be  alkaline  to  phenolphthalein  at  the  end  of  the 
experiment,  which  was  always  the  case.  It  is  reasonably  certain, 
however,  that  no  Clj  or  HCl  did  escape  in  this  way. 

The  cell  was  kept  at  the  desired  temperature  by  circulating 
water  from  the  constant  temperature  bath  through  the  glass  spiral 
coil  (6).  Control  within  one  degree  was  obtained  in  this  way, 
and  it  was  possible  to  run  at  any  temperature  between  12°  and 
70°  C. 


114  ALEXANDER  LOWY  AND  HENRY  S.  FRANK. 

After  an  experiment  was  completed,  the  charge  was  allowed 
to  stand  over  night  in  place,  after  which  the  layers  w^ere  separated 
and  the  aqueous  layer  extracted  once  with  benzene.  The  benzene 
extract  was  added  to  the  oily  layer  and  rendered  alkaline,  made 
up  to  about  500  cc.  with  water,  and  steam  distilled.  The  distillate 
was  separated,  and  the  lighter  oily  layer,  consisting  of  benzene 
and  substituted  chlorobenzenes  subjected  to  fractional  distillations 
three  times,  the  fractions  collected  being  below  90°,  90-120°,  120- 
135°.  The  latter  fraction  contained  the  chlorobenzene.  A  trace 
of  charred  residue  was  left  in  the  distilling  flask  although  the  dis- 
tillation was  made  from  an  oil  bath. 

The  residue  from  steam-distillation  was  extracted  twice  with 
ether,  and  the  residue  left  on  evaporating  off  the  ether,  weighed 
as  the  alkali-insoluble  product.  It  consisted  of  more  highly  chlori- 
nated products.  The  residue  from  the  ether  extraction  was  acidi- 
fied, and  again  extracted  twice  with  ether.  The  residue  remain- 
ing this  time  on  evaporating  ofif  the  ether  was  weighed  as  the 
alkali-soluble  product.  It  was  a  tarry  mass  with  a  strongly 
phenolic  odor. 

A  number  of  electrolytic  experiments  were  performed  under 
difterent  conditions,  giving  different  results.  The  purpose  of 
each  experiment,  its  results  and  conditions  are  given  in  Table  I. 

(B)     Non-Electrolytic. 

The  first  non-electrolytic  experiments  were  conducted  in  the 
same  apparatus,  the  chlorine  being  bubbled  in  through  a  capillary 
in  contact  with  the  platinum  anode.  In  order  to  discover  whether 
the  results  were  in  any  way  due  to  the  presence  of  the  platinum, 
or  of  the  alundum  porous  cup,  etc.,  the  rest  of  the  purely  chemi- 
cal experiments  were  made  in  a  glass  bottle  containing  nothing 
but  a  glass  inlet  tube,  a  glass  outlet  tube,  and  a  glass  bell  stirrer. 
The  latter  was  also  equipped  with  a  mercury  seal,  but  precautions 
were  taken  to  prevent  contamination  of  the  charge  with  mercury. 
The  chlorine  was  generated  by  dropping  an  excess  of  HCl  on  a 
weighed  amount  of  KMnCj.  It  was  bubbled  through  water  to 
remove  HCl,  and  in  one  case  through  cone.  H^SO^  to  remove 
moisture. 

The  charge  was  the  same  as  in  the  other  series,  except  that  in 


THE  CHLORINATION   OF   BENZENE. 


115 


Table  I. 
Experiments  in  the  Electrolytic  Chlorination  of  Benaene. 


Expt. 
No. 


V 

2- 

3 

43 

5 

7 

8 

95 

10 
11 
12 

13" 
14^ 
15' 
16" 


Factor 

Studied 

Amp. 

5.2 

Temp. 

5.2 

Temp. 

5.2 

Temp. 

5.2 

Temp. 

5.2 

Temp. 

5.2 

Stir'g 

5.2 

Stir'g 

5.2 

Temp. 

5.2 

Carrier 

5.2 

Stir'g 

5.2 

Carrier 

5.2 

Anolyte 

5.2 

CD. 

2.6 

CD. 

8.1 

Stir'g 

5.2 

Anode 

amp.   c.  d. 

per 

sq.    dm. 


Volts 


116-120 
116-120 
116-120 
116-120 
116-120 
116-120 
116-120 
116-120 
116-120 
116-120 
116-120 
116-120 
116-120 
58-60 
171-180 
116-120 


8.0 
8.7 
9.4 
7.4 
8.0 
10-11 
7.6 
8.0 
6.6 
6.4 
6.4 
8.0 
12-12.4 
7.0 
13.2 


i   Curent 

passed 
amp.  hr. 


16.2 
16.2 
16.2 
16.2 
16.2 
16.2 
16.2 
16.2 
16.2 
16.2 
16.2 
16.2 
16.2 
16.2 
16.2 
16.2 


'  Anolyte 

Temp. 

per    cent. 

°C. 

12HC1 

30 

12HC1 

15 

12HC1 

20 

12HC1 

55-60 

12HC1 

40 

12HC1 

12-15 

12HC1 

50 

12HC1 

30 

12HC1 

70 

12HC1 

70 

12HC1 

70 

12HC1 

20      j 

20  NaCl 

50       ' 

12HC1 

20      i 

12HC1 

20       1 

12HC1 

20       ' 

Stirring 


500  r.p.m. 
500  r.p.m. 
500  r.p.m. 
500  r.p.m. 
500  r.p.m. 
500  r.p.m. 
300  r.p.m. 
300  r.p.m. 
500  r.p.m. 
500  r.p.m. 
300  r.p.m. 
300  r.p.m. 
300  r.p.m. 
300  r.p.m. 
300  r.p.m. 
300  r.p.m. 


Expt. 
No. 

Benzene 

Distillate 

Alkali 

Alkali 

Carrier 

used 

lac-iss" 

insol. 

sol. 

CO2 

CO 

g. 

g. 

g. 

g. 

g. 

g. 

r 

None 

65 

14.0 

0.04 

0.16 

0.232 

0.056 

2- 

None 

65 

13.0 

0.01 

0.19 

0.123 

0.048 

3 

None 

65 

13.5 

0.04 

0.14 

0.13 

0.074 

4' 

None 

65 

17.5 

0.20 

0.18 

0.23 

0.10 

5 

None 

65 

16.3 

0.02 

0.07 

0.20 

0.15 

6* 

None 

65 

11.8 

0.01 

0.07 

0.119 

7 

None 

65 

19.2 

0.13 

0.22 

0.21 

0.12 

8 

None 

65 

17.5 

0.03 

0.16 

0.176 

0.12 

9^ 

None 

65 

9.6 

1.34 

1.87 

0.25 

10 

1    g.l2 

65 

11.8 

0.53 

0.38 

0.29 

0.16 

11 

None 

65 

12.4 

0.09 

0.35 

0.31 

0.15 

12 

1    g.l2 

65 

18.6 

0.16 

0.18 

0.13 

0.12 

13' 

None 

65 

7.Z 

0.66 

0.66 

0.21 

0.19 

14^ 

1    g.l2 

65 

10.7 

0.14 

0.13 

0.209 

15' 

1    g.l2 

65 

15.82 

3.28 

0.50 

0.10 

0.146 

16" 

None 

65 

15.8 

... 

1  Oily   layer   was   orange   color. 
^  Oily  layer  lemon  yellow. 
'  Oily  layer  orange  red. 
■*  Yellow  oily  layer. 
°  Red  oily  layer. 


"  Color  was  pale  green. 

^  Expt.  conducted  twice  as  long  as  others. 

*  Run  two-thirds  as  long  as  others. 

*  Worked  only  for   Chlorobenzene. 


The  anode  was  in  every  case  a  platinum  wire  loop  of  4.5  sq.  cm.  area. 
Remarks  as  to  color,  etc.,  are  made  only  in  typical  instances,  and  are 
indicative  of  a  general  trend. 


II6 


ALEXANDER  LOWY  AND   HENRY   S.   FRANK. 


most  cases  distilled  water  replaced  the  hydrochloric  acid.  The 
chlorine  was  bubbled  through  at  a  uniform  rate  and  the  charge 
was  well  stirred.  The  product  of  an  experiment  after  standing 
over  night,  was  separated,  and  the  oily  layer  washed  free  of 
chlorine  with  XaOH.    It  was  then  fractionated  for  chlorobenzene 


Table  II. 
Non-Electrolytic  Chlorination  of  Benzene. 


Ben- 

Other 

Distillate 

Expt. 
No. 

Factor 
studied 

zene 

used 

g- 

com- 
ponent 

120-135' 

Remarks 

18 

Chemical 

65 

12  per 

22.6 

Experiment     conducted     in 

action 

cent  HCl 

electrolytic  cell.  Excess 
CI2  used.  Qualitative  ex- 
periment. 

19 

Quantitative 

6,S 

12  per 

19.8 

Same  as  No.  18  except  CU 

relations 

cent  HCl 

generated  from  19  g. 
KMn04  (equivalent  to 
16.2  amp.  hr.) 

20 

Foreign 

65 

12  per 

20.6 

New  apparatus  used.  Noth- 

material 

cent  HCl 

ing  present  except  react- 
ants.  Same  amt.  CU  as 
in  No.  19. 

21 

Function 
of  HCl 

65 

Dist.  H2O 

12.7 

Worked  up  immediately  in- 
stead of  standing  overnight 
as  in  all  other  expt.  -f- 
same  CI2. 

22 

Effect  of 
standing 

65 

Dist.  H=0 

20.4 

Same  as  No.  21,  but  stood 
overnight. 

23 

Effect  of 
light 

65 

Dist.  H2O 

21.4 

Same  as  No.  22,  except  that 
light  was  excluded. 

24 

Effect  of 

65 

Materials 

. , 

No  substitution  took  place. 

moisture 

were  an- 
hydrous 

Benzene  hexachloride  was 
obtamed.  Same  amount 
of  CI,  used. 

just  as  before.  There  were  no  higher  chlorination  products, 
either  alkali-soluble  or  alkali-insoluble  formed,  or  at  most,  mere 
traces. 

The  results  of  the  various  experiments  as  well  as  their  pur- 
poses, and  the  conditions  under  which  they  were  made,  appear  in 
Table  II. 


THE  CHLORINATION   OF  BENZENE.  II7 

DISCUSSION    OF    RESULTS    AND    SUMMARY. 

1.  It  is  possible  to  chlorinate  benzene  by  stirring  it  in  with 
aqueous  HCl  and  electrolyzing.    Aqueous  NaCl  can  also  be  used. 

2.  The  yield  of  chlorobenzene  increases  with  increase  of  tem- 
perature up  to  60°. 

3.  The  yield  of  chlorobenzene  is  affected  by  the  rate  of  stirring. 

4.  The  introduction  of  iodine  as  a  carrier  increases  the  yield  of 
chlorobenzene. 

5.  The  amount  of  higher  chlorinated  products  formed  increases 
in  general  with  rise  in  temperature. 

6.  The  amount  of  benzene  decomposed  to  COj  by  anodic  oxida- 
tion increases  with  the  temperature. 

7.  Increase  in  current  density  rapidly  increases  the  alkali-in- 
soluble product. 

8.  Water  acts  as  a  carrier  in  the  chemical  chlorination  of  ben- 
zene. 

9.  No  substitution  takes  place  when  dry  chlorine  is  passed 
into  dry  benzene.  However,  chlorine  forms  addition  products  of 
the  type  of  benzene  hexachloride. 

10.  Miihlhofer^^  states  that  the  addition  of  iodine  as  a  cata- 
lyzer does  not  alter  the  course  of  the  electrolysis.  The  above 
experiments  show  that  under  the  conditions  cited,  iodine  seems 
to  catalyze  the  chlorination. 

11.  A  new  form  of  apparatus  was  devised  for  this  type  of 
electrolytic  work. 

12.  A  preliminary  series  of  experiments  was  conducted  to 
study  the  electrolytic  chlorination  of  benzene.  A  more  detailed 
study  of  this  process  under  variable  conditions,  as  well  as  elec- 
trolytic bromination  and  iodination,  will  be  reported  in  subsequent 
papers. 

"  Dissertation  cited,  quoted  in  Haber  and  Moser  "Die  elektrolytischen  Prozesse  der 
organischen  Chemie."  p.  97. 


A  paper  presented  at  the  Forty-third 
General  Meeting  of  the  American  Elec- 
trochemical Society  held  in  New  York 
City,  May  3,  1923,  G.  B.  Hogaboom  in 
the   Chair. 


NOTES  ON  THE  ELECTRODEPOSITION  OF  lRON\ 

By   Harris   D.    Hineli.ve.^ 

Abstract. 
Experiments  were  carried  out  to  determine  the  type  of  plating 
bath  that  would  give  good  deposits  of  iron  on  rubber.  Particular 
attention  was  given  to  baths  of  high  "throwing"  power.  \''arious 
formulas  were  tried  out.  A  saturated  bath  of  ferrous  and  calcium 
chlorides,  containing  chromous  chloride  and  hydroquinone  as 
reducing  agents,  gave  the  best  results.  Further  investigation  is 
encouraged.  [A.  D.  S.] 


The  problem  presented  was  that  of  depositing  a  substantial 
thickness  of  iron  onto  rather  irregularly  shaped  rubber  articles,  this 
involving  a  process  for  preparing  a  conducting  coating,  a  plating 
bath  which  would  give  good  heavy  deposits,  in  thicknesses  up 
to  12.5  mm.  (y^  in.),  and  have  a  high  throwing  power  to  ensure 
filling  the  crevices.  The  conducting  coating  on  the  rubber  was 
easily  obtained  by  varnishing  it,  and  then  brushing  in  graphite, 
repeating  the  application  of  graphite  at  intervals  until  the  varnish 
was  too  dry  to  take  on  any  more. 

The    ferrous    ammonium    sulfate    bath    suggested    by    D.    R. 

Kellogg^  was  tried  out,  but  found  to  be  unsatisfactory,  as  it  is  too 

easily  rendered  useless  by  organic  extractives  from  the  rubber. 

The  deposit  was  badly  pitted.     Kellogg,  too,  records  the  failure 

of  his  bath  when  organic  compounds  entered  it. 

The  best  summary  of  work  done  on  iron  plating  baths  is  that 

by  Mr.  W.  E.  Hughes.*     From  this  summary  we  concluded  that 

'  Original   Manuscript  received    Sept.    3,   1922. 

2  Pittsburgh  Park,  Pittsburgh,  Pa. 

3  Trans.  Am.   Inst.  Min.  and  Met.   Eng.,  Feb.,   1922. 

*  Trans.  Am.  Electrochem.  Soc,  40,  185,  et  seq.,  (1921). 

119 


I20  HARRIS   D.    HINELINE. 

the  Fisher-Langbein,  ferrous  calcium  chloride  bath  would 
probably  be  the  most  promising  for  our  problem.  However,  this 
bath  caused  serious  pitting  and  corrosion  on  both  iron  and  rubber 
cathodes.  A  simple  ferrous  chloride  bath,  (150  g./L.)  was 
equally  bad,  suggesting  that  the  trouble  might  be 
due  to  the  presence  of  ferric  chloride.  A  bath  made  up  of 
150  g./L.  of  ferrous  chloride  and  100  g./L.  of  sodium  acid  sul- 
fite to  insure  complete  reduction  of  all  the  iron,  gave  a  very 
good  deposit  for  several  days,  reaching  a  thickness  of  about  0.24 
mm.  (3/32  in.).  Corrosion  then  began,  due  partly  to  exhaustion 
of  the  sulfite,  and  partly  to  high  acidity. 

No  mention  is  made  in  Mr.  Hughes'  paper  of  any  trials  of  a 
bath  made  up  of  ferrous  sulfite.  A  solution  was  therefore  pre- 
pared by  treating  a  suspension  of  ferrous  carbonate  in  water  with 
sulfur  dioxide.  A  solution  containing  about  50  g./L.  of  iron  was 
obtained,  probably  present  as  ferrous  acid  sulfite.  On  electrolysis, 
no  iron  deposit  was  obtained,  but  instead,  a  thick  mass  of  mate- 
rial at  the  cathode,  which  seemed  to  be  a  mixture  of  ferrous  sul- 
fite and  sulfide.  The  mass,  on  treatment  with  dilute  acid,  first 
evolving  sulfur  dioxide  and  then  hydrogen  sulfide,  dissolved 
completely. 

Since  the  addition  of  the  sulfite  to  the  ferrous  chloride  solu- 
tion gave  a  good  deposit  it  did  seem  that  a  reducing  agent  in 
the  bath  would  improve  conditions.  A  canvass  of  available 
reducing  agents  suggested  hydroquinone  as  a  possibility.  A  bath 
containing  250  g./L.  of  ferrous  chloride,  a  little  ferrous  carbonate 
and  5  g./L.  of  hydroquinone  gave  a  deposit  free  from  corrosion 
and  pitting.  A  solution  of  150  g./L.  of  ferric  chloride  was  then 
reduced  with  an  excess  of  hydroquinone.  The  lower  solubility  of 
the  resulting  quinone  caused  it  to  crystallize  out. 

The  solution  with  its  suspended  crystals  was  then  electrolyzed 
at  a  c.  d.  of  about  2.7  amp./sq.  dm.  (25  amp./sq.  ft.).  It  yielded  an 
excellent  deposit,  free  from  corrosion  and  treeing,  but  very  brittle, 
due  probably  to  high  acidity  from  the  reduction  of  the  ferric  salt. 
Large  gas  bubbles  were  evolved  at  the  cathode,  but  they  did  not 
produce  gas  pits.  A  solution  containing  200  g./L.  of  ferrous 
chloride,  200  g./L.  calcium  chloride  and  10  g./L.  hydroquinone 
gave  an  excellent  deposit,  and  after  electrolysis  over  a  period  of  a 


NOTES    ON    THE    EIvECTRODEPOSlTlON    OF    IRON.  12 1 

month,  during  which  a  deposit  3  mm.  (^  in.)  in  thickness  was 
made,  the  bath  was  still  working  well.  Anode  corrosion  had 
liberated  considerable  sludge,  so  the  bath  was  filtered.  The  fil- 
tered bath  again  gave  much  trouble  due  to  corrosion,  but  the  addi- 
tion of  10  g./L.  of  hydroquinone  restored  the  bath  to  good 
working  condition. 

In  plating  baths  a  wide  variety  of  addition  agents,  both  organic 
and  inorganic,  is  used.  Glue,  glycerine,  gum  arabic  and  dextrose 
were  tried  in  a  Fisher-Langbein  bath  containing  hydroquinone, 
but  all  were  rapidly  destroyed,  yielding  a  sludge  and  breakdown 
products  which  ruined  the  bath.  In  one  case  20  g./L.  of  glue  were 
added,  giving  a  bath  of  such  high  viscosity  that  the  hydrogen 
liberated  at  the  cathode  was  held  in  place.  The  iron  was  deposited 
between  the  bubbles,  resulting  in  a  bulky  deposit  of  fine  iron 
crystals. 

Chromous  chloride  being  a  good  reducing  agent,  a  bath  was 
made  up  containing  10  g./L.  of  it,  and  200  g./L.  each  of  ferrous 
chloride  and  calcium  chloride.  This  gave  a  very  good  deposit  at 
a  c.  d.  as  high  as  8  amp./sq.  dm.  (75  amp./sq.  ft.)  even  when 
continued  over  a  period  of  a  month,  although  some  trouble  was 
experienced  with  treeing.  Manganous  chloride  gave  a  similar 
result,  but  required  much  larger  quantities.  Antimony,  added  as 
chloride,  plated  out  before  the  iron;  zinc  was  without  eflfect. 

A  bath  containing,  per  L.,  350  g.  of  ferrous  chloride,  225  g.  of 
calcium  chloride,  20  g.  of  chromous  chloride,  and  5  g.  of  hydro- 
quinone, was  chosen  as  having  about  the  best  proportion  among 
its  various  constituents.  This  was  used  at  a  temperature  between 
60  and  70°  C.  It  showed  no  corrosion,  no  pitting,  only  a  little 
treeing  and  fair  metal  quality.  Deposits  as  thick  as  12.5  mm. 
(1/2  in.)  were  made  which  had  about  the  strength  of  a  mediocre 
grade  of  cast  iron.  The  individual  crystals  of  metal  were  very 
large,  some  extending  entirely  through  the  deposit. 

In  most  of  the  trials  made  in  search  of  a  suitable  bath  flat  rubber 
or  iron  sheets  were  used  as  cathodes.  It  was  soon  noticed  that 
although  rubber  is  a  good  insulator,  a  "strike"  of  deposited  metal 
over  its  surface  could  be  obtained  without  special  preparation.  A 
sheet  of  rubber  simply  suspended  in  the  bath  by  a  metal  clamp, 
so  that  part  of  the  clamp  was  submerged,  would  take  a  "strike'* 


122  HARRIS  D.    HINELINE. 

over  its  entire  surface  in  a  very  short  time.  This  fact  seriously 
compHcated  the  problem  of  falling  the  grooves  and  crevices  in  the 
specimens.  The  plating  bath  we  developed  does  not  "throw"  its 
deposit  at  all  well,  in  fact  none  of  the  iron  baths  will  "throw"  as 
well  as  zinc  and  copper  cyanide  baths  will. 

We  were  not  able  to  prepare  a  conducting  line  with  varnish  and 
graphite  at  the  bottoms  of  the  crevices  and  keep  the  deposit  on  it. 
In  a  very  short  time  the  metal  would  strike  over  the  entire  rubber 
surface,  whereupon  it  ceased  to  deposit  in  the  crevices.  Rubber 
being  a  good  insulator,  this  was  most  unexpected.  The  physical 
character  of  the  surface  of  the  rubber  or  perhaps  the  interfacial 
tension  between  rubber  and  solution  may  account  for  the 
phenomenon. 

A  variety  of  expedients  were  tried  to  overcome  the  difficulty. 
Small  anodes  placed  within  the  crevices,  so  as  to  shorten  the  cur- 
rent path,  were  ineffective  because  of  lack  of  anode  area.  Strongly 
charged  shields,  covering  the  projecting  portion,  showed  possi- 
bilities, but  an  adequate  insulating  covering  for  the  shields  was 
not  available.  Anodes  in  contact  with  the  projection  forced  the 
deposit  into  the  grooves,  but  did  not  fill  them  to  the  bottom.  Hard 
rubber  shields  to  control  the  path  of  current  flow  worked,  until 
the  deposit  was  thick  enough  at  some  point  to  touch  the  shield,  a 
strike  then  took  place  over  the  shield,  and  it  got  all  the  deposit. 
A  rapid  stream  of  electrolyte,  impinging  on  the  surface  of  the 
projection  would  keep  the  deposit  oft  of  it,  but  the  stirring  of 
the  solution  was  too  great,  and  a  knobby  deposit  resulted. 

It  finally  became  evident  that  it  would  be  necessary  to  cover  the 
ejitire  rubber  surface  with  stopping  oft*  material ;  a  tung  oil  bak- 
ing varnish,  lightly  baked,  was  found  to  be  adequate.  Conducting 
lines  were  then  put  in  the  bottom  of  the  crevices,  and  it  was  found 
that  if  the  applied  voltage  did  not  exceed  0.45  volt  the  deposit  did 
not  creep,  and  the  crevices  could  be  filled  completely  before  any 
metal  deposited  on  the  tops  of  the  projections.  With  the  crevices 
filled,  a  slightly  higher  voltage  would  cause  the  deposit  to  strike 
over  the  entire  surface.  An  observation  that  should  be  recorded 
is  the  effect  of  various  hydro-carbons  on  the  bath.  Saturated 
petroleum  products,  such  as  gasoline,  kerosene,  or  machine  oil, 
produce  bad  treeing  of  the  deposit,  even  when  present  in  exceed- 


NOTES   OX    THE   ELECTRODEPOSITION    OF   IRON,  123 

ingly  small  quantities,  while  turpentine,  in  considerable  quantity, 
is  without  effect  on  the  deposit. 

The  bath  made  up  as  first  indicated  worked  reasonably  well,  but 
proved  to  be  a  little  too  dilute.  Better  results  were  obtained  from 
a  bath  made  up  with  equal  parts  of  ferrous  chloride  and  calcium 
chloride  in  such  a  quantity  as  to  make  the  bath  saturated  at  a 
temperature  of  about  30°  C,  to  which  was  added  about  20  g./L. 
of  chromous  chloride,  and  about  5  g./L.  of  hydroquinone.  This 
bath  worked  best  at  a  temperature  of  60°  to  70°  C.  Lower  tem- 
peratures gave  a  poorer  deposit,  less  strength  and  more  inclusions, 
while  higher  temperatures  showed  rather  too  much  evaporation. 
The  material  in  the  bath  appears  to  be  present  in  the  form  of  a 
double  salt  of  iron  and  calcium  chloride  in  which  the  ferrous  ion 
is  much  reduced  in  concentration.  The  solution  is  much  lighter 
in  color  than  the  equivalent  solution  of  ferrous  chloride  alone, 
and  the  salts  crystallizing  out  are  also  much  lighter  in  color  than 
ferrous  chloride  crystals  and  different  in  crystal  habit. 

It  is  of  primary  importance  that  the  iron  in  the  bath  be  kept 
in  the  ferrous  condition.  It  is  difficult  to  determine  the 
amount  of  ferric  iron  in  any  given  bath,  but  the  permissible 
amount  in  a  satisfactory  bath  is  certainly  below  1  g./L.  and 
probably  below  0.1  g./L.  This  low  concentration  of  ferric 
iron  may  be  maintained  by  the  use  of  hydroquinone  in  the  solution, 
in  spite  of  the  high  partial  pressure  of  oxygen  at  the  surface  of 
the  solution,  and  in  spite  of  oxygen  liberated  at  the  anode, 
when  anode  corrosion  is  less  than  100  per  cent.  The  hydro- 
quinone has  a  higher  reduction  potential  than  ferrous  chloride, 
and  a  comparatively  small  concentration  of  it  will  keep  the  ferric 
iron  content  sufficiently  low.  The  hydroquinone  is  oxidized  to 
quinone,  which,  although  not  as  soluble  as  the  hydroquinone  itself, 
is  still  somewhat  soluble,  so  as  to  bring  a  low  concentration  to 
the  cathode  surface.  The  hydrogen,  which  was  always  liberated  at 
the  cathode  to  a  certain  extent,  is  taken  up  by  the  quinone  to 
reform  hydroquinone,  thereby  maintaining  the  hydroquinone 
content. 

The  hydroquinone  may  be  considered  as  a  carrier  of  hydrogen 
from  the  cathode  to  the  solution,  thereby  taking  care  of  the  hydro- 
gen which  otherwise  might  form  gas  pits  on  the  cathode.  Likewise 


124  HARRIS   D.    HINEUNE;. 

the  quinone  may  be  considered  as  a  carrier  of  oxygen  from  the 
anode,  to  obviate  the  difficulties  due  to  less  than  100  per  cent  anode 
corrosion.  The  hydroquinone  and  its  oxidation  product  will 
remain  in  the  solution  without  further  destruction  for  a  period 
of  months.  There  appeared  to  be  no  electrode  reactions  which 
were  sufficiently  powerful  to  cause  further  reaction  with  the 
ring  nucleus  of  the  compound.  A  concentration  of  hydroquinone 
of  5  to  20  g./L.  appears  to  be  ample  as  long  as  anode  corrosion 
efficiency  does  not  get  too  low.  It  is  possible  that  an  anode  corro- 
sion efficiency  of  less  than  95  per  cent  will  ruin  the  solution 
regardless  of  any  treatment. 

The  action  of  the  small  percentage  of  chromous  chloride  is  not 
so  readily  explained  nor  is  it  as  conspicuous.  The  chromium 
appears  to  be  plated  out  slowly.  Analysis  of  a  typical  deposit 
showed  0.3  per  cent  chromium,  which  is  probably  in  the  metallic 
condition,  since  the  amount  of  inclusions  of  electrolyte  in  the 
deposit  were  far  too  small  to  account  for  such  a  quantity  of 
chromium.  It  was  first  considered  that  the  chromium  served  as  a 
carrier  of  oxygen  and  hydrogen  in  the  same  way  as  the  hydro- 
quinone. However  the  evidence  in  support  of  this  view  is  not 
strong,  since  oxidation  of  the  chromium  seems  to  proceed  through 
an  intermediate  step  when  the  chromium  precipitates. 

It  is  more  probable  that  the  chromium  plates  out  slowly  to  give 
a  slight  breaking  up  of  the  iron  crystal  structure  somewhat  after 
the  idea  of  interleaved  nickel  in  copper  deposits.^  This  point  should 
be  checked  up,  for  if  it  proves  to  be  correct  it  is  possible  that 
the  addition  of  other  metal  salts,  perhaps  nickel  chloride,  cobalt 
chloride,  or  some  similar  salt  might  yield  a  deposit  which  would 
have  finer  crystal  structure  and  better  strength  than  the  iron 
deposits  so  far  obtained. 

Control  of  the  acidity  of  the  plating  solution  is  not  of  the 
critical  importance  that  is  required  in  the  control  of  the  reduction 
of  the  bath.  The  solution  should  be  slightly  acid,  sufficiently  so  to 
prevent  the  precipitation  of  ferrous  hydroxide  or  ferrous  carbon- 
ate. The  minimum  satisfactory  acidity  is  probably  about  0.01  per 
cent  and  it  may  go  to  about  0.5  per  cent  of  HCl.  The  acidity  of 
the  bath  is  satisfactory  when  the  bath  is  made  up  from  good 

6  Trans.  Am.   Electrochem.   Soc.  40,  307,   (1921). 


NOTES    ON    THE    ELECTRODEPOSITION    OF    IRON.  1 25 

grades  of  ferrous  chloride  and  calcium  chloride,  and  it  will  stand 
the  addition  of  about  2  g./L.  of  concentrated  HCl  without  produc- 
ing a  deposit  which  is  excessively  brittle.  The  addition  of  ferrous 
carbonate,  calcium  carbonate  or  caustic  is  permissible  to  reduce 
the  acidit}--  of  the  bath,  but  if  continued  to  the  point  where  iron 
precipitates  as  a  carbonate  or  hydroxide,  the  bath  immediately 
gives  treeing  deposits. 

Another  method  of  keeping  the  bath  reduced  to  the  ferrous 
condition  was  under  trial,  but  without  conclusive  results.  Early 
experiments  showed  that  the  presence  of  sulfur  dioxide  in  the 
plating  bath  was  not  harmful.  It  did  not  seem  possible  to  add  the 
gas  directly,  since  the  oxidation  product,  being  sulfate,  would 
precipitate  a  portion  of  the  calcium  from  the  bath  and  liberate  an 
excess  of  hydrochloric  acid.  It  does  seem  possible,  however,  to 
add  to  the  bath  small  quantities  of  normal  calcium  sulfite  made  by 
suspending  calcium  hydroxide  in  water  and  passing  in  the 
weighed  quantity  of  sulfur  dioxide  gas.  This  would  yield  a 
precipitate  of  calcium  sulfate,  which  is  harmless,  and  would  not 
change  either  the  calcium  content  or  the  acidity  of  the  plating 
bath.  Results  at  this  time  have  not  been  continued  far  enough  to 
show  whether  this  is  practical  or  not. 

The  foregoing  experiments  are  somewhat  desultory  in  character 
and  do  not  follow,  as  rigorously  as  might  be  desired,  a  definite 
line  of  logical  research.  It  is  hoped,  however,  that  other  workers 
interested  in  similar  work  will  find  suggestions  of  value  in  this 
paper. 


A  paper  presented  at  the  Forty-second 
General  Meeting  of  the  American  Elec- 
trochemical Society,  held  in  Montreal, 
and  brought  up  for  discussion  at  the 
Forty-third  meeting  in  New  York  City, 
May  4,  1923,  President  Schluederberg 
in  the  Chair. 


HEAT  INSULATING  MATERIALS  FOR  ELECTRICALLY  HEATED 
APPARATUS'. 

By  J.  C.  Woodson-. 
IXTRODL'CTION, 

Heat  and  heat  processes  enter  into  practically  every  form  of 
manufacture  and  the  industry  is  indeed  scarce  that  does  not 
somewhere  in  its  organization,  utilize  this  form  of  energy  to 
fashion  or  perfect  its  product.  This  has  been  true  of  indus- 
try since  its  inception,  yet  only  within  the  last  two  decades  has 
there  been  any  real  effort  to  conserve  or  reduce  the  heat  lost  in 
these  processes.  Even  today,  there  is  very  limited  data  available 
on  the  subject  of  heat  insulating  material,  except  for  certain 
specific  temperatures  and  under  conditions  which  do  not  neces- 
sarily hold  for  other  conditions. 

While  the  attempt  will  be  made  in  this  paper  to  be  as  general 
as  possible  on  this  subject,  attention  is  called  to  the  fact  that  most 
of  the  data  and  curves  given  refer  to  heat  insulating  material  used 
in  connection  with  electrically  heated  apparatus.  It  is  vital  and 
absolutely  necessary  to  conserve  all  the  heat  possible  with  such 
apparatus,  which  also  requires  careful  attention  to  other  char- 
acteristics of  insulating  material  ordinarily  considered  unimpor- 
tant. The  rapid  and  almost  phenomenal  increase  in  the  commer- 
cial use  of  electrically  heated  apparatus,  ovens,  furnaces  and 
machines,  indicates  that  all  other  forms  of  heat  and  heat  treat- 
ment will  sooner  or  later  be  supplanted,  to  a  large  extent,  by  elec- 
tric heat.  This  change  is  now  and  will  continue  to  be,  dependent, 
to  a  greater  or  less  degree,  upon  the  available  heat  insulating 
mediums  and  the  ability  of  engineers  and  manufacturers  to  apply 
them  properly. 

*  Original  manuscript  received  August  8,   1922. 

'Electric     Heating     Engineering     Dept.,     Westinghouse     Elec.     and     Mfg.     Co.,     E. 
Pittsburgh,  Pa. 

127 


128 


J.  C.  WOODSON. 


TEMPERATURE  RANGES  CONSIDERED. 

Low  temperatures,  such  as  80° F.  (27°C.)  or  lower  will  be  con- 
sidered only  briefly,  and  for  convenience  we  will  divide  our  tem- 
perature ranges  into  the  5  divisions  shown  in  Table  I. 

Table  I. 


Division 

Range 

'Application 

1 

0° 
—  18° 

to 
to 

200° 
93° 

F. 
C. 

Refrigeration,  cooling,  water  heating, 
drying,  presses,  air  heating,  various 
liquids. 

2 

200° 
93° 

to 
to 

350° 
177° 

F. 
C. 

Steam  pipes,  drying,  color  enamel, 
presses,  baking. 

3 

350° 
177° 

to 
to 

600° 
315° 

F. 
C. 

Japanning,  core  baking,  bread  baking, 
presses,  appliances,  liquids. 

4 

600° 
315° 

to 
to 

1,000° 
538° 

F. 
C. 

Tempering,  annealing,  solder,  babbitt, 
tin  melting. 

5 

1,000° 
538° 

to  2.000° 
to  1,093° 

F. 
C 

Heat  treating,  drawing,  forging, 
melting,    enameling. 

To  cover  these  five  ranges,  there  are  numerous  commercial 
grades  of  insulating  material  of  various  trade  names  and  ratings ; 
a  great  many  for  divisions  one  and  two  and  tapering  off  to  only 
two  or  three  reliable  grades  for  division  five.  Practically  all  of 
these  commercial  grades  can  be  located  in  three  classes  by  funda- 
mental composition  as  stated  in  Table  II. 

9 

Table  II. 


Class 

Division 

Composition 

A 

1 

Hair,  wool,  felt,  wood  pulp,  animal 
and  vegetable  fiber,  asbestos  paper, 
cork. 

B 

2,  3,  4 

Asbestos,  magnesia,  sponge,  earths, 
mineral  wool. 

C 

4,  5 

Diatomaceous  earth,  mineral  wool, 
earths,  silicates. 

From  this  Table,  it  is  evident  that  there  is  no  clear  or  definite 
dividing  line  between  either  the  temperature  division  or  the  classes 


HEAT  INSULATING  MATERIALS.  1 29 

by  composition,  as  there  is  a  certain  amount  of  overlapping.  Cer- 
tain combinations  of  these  fundamental  ingredients  also  produce 
distinct  grades  of  insulation,  entirely  different  from  any  of  the 
component  parts.  Also,  certain  ingredients  are  used  in  one  class 
as  insulating  material,  and  in  another  class  as  a  mechanical  binder 
or  strengthener  of  the  true  insulation,  such  as  asbestos  in  classes 
B  and  C  and  mineral  wool  in  class  C. 

There  are  numerous  qualities  desired  in  heat  insulating  mate- 
rials and  different  applications  require  different  qualities,  but  in 
general  a  good  heat  insulating  material  should  have  the  following 
characteristics. 

1.  Low  heat  conductivity. 

2.  Low  specific  heat. 

3.  Low  specific  gravity. 

4.  Non-inflammable. 

5.  Strong  and  durable  mechanically. 

Low  conductivity  to  reduce  radiation  losses;  low  specific  heat 
to  save  as  much  power  in  heating  up  period  as  possible  and  make 
apparatus  faster;  low  specific  gravity  to  keep  down  unnecessary 
weight  and  save  heating  up  power  as  No.  2 ;  non-inflammable  as 
most  insulations  are  subjected  to  periodic  or  locally  high  tempera- 
tures ;  No.  5  for  length  of  life  and  reliability. 

Other  attributes  to  be  desired  are : 

6.  Electrical  non-conduction. 

7.  Have  no  chemical  action  on  metals. 

8.  Easily  shaped  or  formed. 

9.  Permanent  in  setting  (no  shifting  or  settling). 

10.  Impervious  to  action  of  liquids,  (water,  acids,  oil)^. 

Practically  all  commercial  insulations  have  most  of  these  quali- 
ties in  some  degree,  the  two  last  being  the  ones  most  often  left  out. 
In  the  writer's  experience.  No.  10  is  not  attained  by  any  present 
day  insulations ;  though  several  grades  will  stand  drenching  in 
water  and  after  being  thoroughly  dried  prove  to  be  practically 
as  good  as  ever.  However,  while  still  wet,  this  insulation  is  almost 
useless^. 

'Weidlein,  Chem.  and  Met.  Eng.,  24,  295,   (1921). 


I30  J.  C.  WOODSON. 

An  evacuated  space  is  the  best  thermal  insulator  of  conducted 
heat  known,  while  gases  under  certain  conditions  are  probably 
next.  Air  is  a  good  insulator  if  it  can  be  entrapped  in  small 
enough  spaces  to  prevent  convection  currents,  and  to  this  fact  and 
arrangement  most  present  day  heat  insulators  owe  their  value  as 
such.  This  minute  honey-combing  of  the  structure  places  multi- 
tudes of  confined  dead  air  spaces  in  series  opposing  the  heat  flow, 
with  only  minute  point  contact  of  the  material  fibers  or  crystals 
for  direct  conduction. 

Heat  transfer  by  radiation  through  insulating  material  is  prob- 
lematical, as  these  radiations  are  stopped  by  the  insulation  and  the 
heat  carried  by  conduction ;  or  with  some  insulations  the  rays  are 
to  a  certain  extent  refracted  so  that  the  penetration  is  relatively 
shallow.  At  temperatures  beginning  with  300°  C.  this  character- 
istic is  important. 

The  law  of  heat  flow  through  resisting  materials  is  analogous 
to  Ohm's  law  for  electrical  circuits,  expressed  as  I  =  E/R  where 
I  is  the  current,  R  the  resistance  and  E  the  voltage  pressure  or 
difference  between  two  points.  Likewise  the  amount  of  heat  flow- 
ing between  two  points  of  dififerent  temperatures  can  be  expressed 
as 

w  =  f  a) 

where  W  is  watts  flowing  as  heat,  Td  is  temperature  difference 
and  R  is  the  thermal  resistance  of  the  path  of  flow.  This  means 
that  the  rate  of  heat  flow  is  directly  proportional  to  the  tempera- 
ture pressure  or  dift'erence,  and  inversely  proportional  to  the 
resistance  of  the  path  or  material  composing  the  thermal  circuit. 
From  the  above,  it  follows  that 

R   =  ^  (2) 

w 

In  formulas  1  and  2,  Td  is  expressed  in  °C.   R  is  the  total  thermal 
resistance  of  the  circuit.    Therefore 

R=:kr=^.±  (3) 

A  A         c 


HEAT  INSULATING  MATERIALS.  13I 

Where 
R  :=  total  resistance  of  circuit  in  thermal  ohms 
h  =  length  of  circuit  in  inches 
A  =  area  of  path  in  sq.  in. 
r  =:  specific  resistance  of  circuit  in  thermal  ohms  per  inch 

cube 
c  =  thermal  conductivity  in  watts  per  inch  cube  per  °C. 
(r  =  1/c) 

By  substituting  in  formula  No.  1,  we  have 

=  —  .  =  —  .  c  .  TdC)  (4) 

L  r  L 

Where  W  is  watts  flowing  per  unit  of  time.  Tables  III,  IV,  and 
V,  give  the  values  of  r  for  a  number  of  building  and  insulating 
materials. 

The  above  simple  formulae  are  little  recognized  and  seldom 
used,  due  to  the  many  awkward  and  arbitrary  units  ordinarily  used 
by  engineers,  so  that  while  the  rule  remains  simple,  the  means  of 
applying  and  using  it  are  often  complicated  and  involved.  In  this 
country,  the  usual  unit  used  is  the  British  thermal  unit,  and  the 
method  of  expressing  heat  flow  is  given  by  the  equation 

Q   =  KAt   (^^^1^^)  (5) 

Where  Q  is  the  quantity  of  heat  flowing  through  a  path  of  area 
A  in  time  "t"  the  length  of  the  path  is  "th"  with  a  temperature 
difference  of  T^  —  T..  K  is  the  coefficient  of  thermal  conduc- 
tivity of  the  material  of  the  circuit.  These  units  are  ordinarily 
expressed  as  follows. 

Q  =  B.  t.  u.  transmitted 

A  =  sq.  ft. 

t  =:  hours 

th  =  inches 

T,  —  T,  =r  °F. 

K  =  B.  t.  u.  per  sq.  ft.,  per  inch  of  thickness,  per  hr.,  per  °F. 
temperature  difference 

*  C.   p.  Randolph,  Trans.  Am.  Electrochem.   Soc.  21,  543.   (1912). 


132 


J.  C.  WOODSON. 


Table  III. 


Material 


Air     

Air-cel  asbestos  

Balsa  wood  

Cabot  quilt  

Calorox    

Cork  board   

Cotton  wool    

CjT)ress  wood  

Eiderdown     

Eiderdown     

Fibrofelt 

Gimco  thermalite   . . . 

Ground  cork  

Hair  felt   

Hard  maple   (wood). 

Insulite     

Kapok     

Keystone  hair  felt  . . . 

Linof elt    

Lith  board 

Mahogany  wood    . . . . 

Nonpareil  corkboard. 

Oak    wood    

Pulp  board   

Remanit  (charred  silk) 

Sheep's   wool    

Tar-paper  roofing  . . . 

Vacuum   

Virginia  pine    

White  pine    

Wool   felt    


Density 
lb.  per 
cu.  ft. 


0.08 
8.8 

7.5 
16.0 

4.0 

6.9 

7.0 
29.0 

6.77 

0.134 
11.3 
17.0 

9.4 
17.0 
44.0 
11.9 

0.88 
19.0 
11.3 
12.5 
10.2 

34.0 
38.0 


6.9 

55.0 

34.0 
32.0 
21.0 


I        K 

Spec.         E-  t-  "• 

Heat  P^^, 

sq.    ft. 

!        etc. 


0.240 
0.281 


0.44 
0.362 


0.20 

0.48 
0.40 


0.40 

0.32 
0.50 


0.57 


0.67 
0.39 


0.175 

0.500 

0.350 

0.321 

0.221 

0.279 

0.291 

0.666 

0.1345 

0.438 

0.329 

0.272 

0.296 
0.246 
1.124 
0.296 
0.237 
0.271 
0.300 
0.379 
0.304 

0.916 
1.000 
0.458 
0.274 
0.246 
0.708 
0.041 
0.958 
0.792 
0.363 


r. 

thermal 

ohms 

per 

cu.  in. 


1560.0 
546.0 
780.0 
851.0 

1235.0 
979.0 
938.0 
410.0 

2030.0 
623.0 
830.0 

1013.0 

923.0 

1110.0 

242.3 

923.0 

1151.0 

1008.0 

910.0 

721.0 

898.0 

298.0 
273.0 
596.0 
996.0 

1110.0 
386.0 

6666.0 
285.0 
345.0 
752.5 


At 

temp. 

"F. 


n 

77 
77 
77 
77 
77 
77 
77 
212 
212 
77 
93 

77 
77 
77 
77 
77 
77 
77 
77 
150 

77 
77 
77 
300 
300 
300 
300 
300 
300 
300 


Authority 


Van  Dusen 
Van  Dusen 
Van  Dusen 
Van  Dusen 
Van  Dusen 
Van  Dusen 
Van  Dusen 
Van  Dusen 
Randolph 
Randolph 
Van  Dusen 
General  Ins.  and 

Mfg.  Co. 
Van  Dusen 
Van  Dusen 
Van  Dusen 
Van  Dusen 
Van  Dusen 
Van  Dusen 
Van  Dusen 
Van  Dusen 
Armstrong  Cork  and 

Insulation  Co. 
Van  Dusen 
Van  Dusen 
Van  Dusen 
Stott 

Van  Dusen 
Van  Dusen 
Van  Dusen 
Van  Dusen 
Van  Dusen 
Van  Dusen 


HEAT 

INSULATING  MATERIALS 

133 

Table  IV. 

Density 

Spec. 
Heat 

K 

B.  t.  u. 

r. 

thermal 

At 

Material 

lb.   per 
cu.  ft. 

per 

sq.   ft. 

ohms 
per 

temp. 

Authority 

etc. 

cu.   in. 

Air-cel  asbestos  

8.8 

0.292 

0.500 

546.0 

17 

Van  Dusen 

Air-cel  asbestos  

15.6 

0.683 

399.0 

0  to  392 

Randolph 

Asbestos  felt   

30  to  40 

0.549 

497.0 

400 

Franklin  Mfg.  Co. 

Asbestos  fiber 

12.5  to  18.7 

0.608    to 

448.0    to 

0.497 

548.0 

932 

Randolph 

Asbestos  fire  felt   . . . 

27.6 

1.093 

249.0 

370 

McMillan 

Asbestos  lumber    

123.0 

2.710 

100.5 

Van  Dusen 

Asbestos  mill  board.. 

61.0 

0.833 

328.0 

Van  Dusen 

Asbestos  paper    

50  to  70 

0.300 

1.250 

218.5 

150 

Marks 

Asbestos  sponge  felted 

0.509 

537.0 

400 

Stott 

Asbestos  sponge  felted 

34.4 

0.329 

830.0 

392 

Randolph 

Carey  carocel  

0.540 

506.0 

370 

McMillan 

Carey  duplex   

0.636 

429.0 

370 

McMillan 

Carey  85%  magnesia. 

18.0 

0.546 

500.0 

370 

McMillan 

Carey  85%  magnesia. 

18  to  24 

0.312 

0.500 

546.0 

300 

Wiedlien 

Carey  85%  magnesia. 

18  to  24 

0.312 

0.585 

467.0 

600 

Wiedlien 

Carey  serrated  

0.682 

401.0 

370 

McMillan 

Celite  powder 

10.6 

0.289 

0.309 

883.0 

n 

Van  Dusen 

Diatomaceous  earth 

and  asbestos  

20.7 

0.497 

549.0 

0  to  750 

Randolph 

85%  magnesia    

13.5 

0.455 

600.0 

0  to  7.50 

Randolph 

Fire  felt  roll 

43.0 

0.624 

438.0 

n 

Van  Dusen 

Fire  felt  sheets   

26.0 

0.583 

468.0 

77 

Van  Dusen 

Fullers    earth  

33.0 

0.708 

386.0 

77 

Van  Dusen 

Gypsum  plaster  

56.0 

0.26 

2.250 

121.4 

77 

Van  Dusen 

Insulex  

29.0 

0.281 

0.916 
0.549 

298.0 
497.0 

77 

400 

Van  Dusen 

J.  M.  asbestocel  

McMillan 

J.  M.  asbestos  sponge 

12.0 

felted    

42.0 
13  to  16 

0.312 

0.468 
0.507 

583.0 
538.0 

370 
370 

McMillan 

J.  M.  85%  magnesia.. 

McMillan 

J.  M.  85%  magnesia.. 

16.8 

0.444 

615.0 

470 

J.  M.  Co. 

J.   M.  fine  corrugated 

asbestos   

15.6 

0.538 
0.666 

507.0 
409.0 

470 
370 

J.  M.  Co. 

J.  M.  indented  

McMillan 

J.  M.  moulded  asbes- 

tos     

21.6 

0.778 
1.087 

351.0 
251.0 

370 
370 

McMillan 

J.  M.  vitrobestos  .... 

McMillan 

K.  &  M.  air-cel  asbes- 

tos     

12.5 

0.48 

0.680 
0.433 

402.0 
631.0 

370 

400 

Stott 

Laminated  cork  

Stott 

Mineral  wool  

12.5 

0.198 

0.275 

993.0 

77 

Van  Dusen 

Mineral  wool  

26.6 

0.198 

0.479 

570.0 

932 

Randolph 

Nonpareil  H.  P 

22.56 

0.20 

0.470 

581.0 

370 

McMillan 

Nonpareil  H.  P.  block 

27.0 

0.20 

0.543 

502.5 

370 

McMillan 

Plastic  85%  magnesia 

0.587 

465.0 

370 

McMillan 

Poplox    

1.43 
5.80 

0.384 
0.463 
0.350 
0.510 

712.0 
589.0 
780.0 
536.0 

572 

932 

77 

370 

Randolph 

Poplox    

Randolph 

Rock  cork   

Van  Dusen 

Sallmo  wool  felt 

McMillan 

Silica  

106.0 
10.0 

0.45 
0.20 

1.775 
0.418 
0.320 

153.8 
653.0 
853.0 

932 
400 
200 

Randolph 

Solid  cork  

Stott 

Thermo  fiber' 

F.  D.  Farnum  and  Co 

35%  magnesia   

29.8 

0.569 

480.0 

0  to  750 

Randolph 

Vitrified  Monarch 

block  

40  to  45 

0.842 

324.0 

400 

Franklin  Mfg.  Co. 

134 


J.  C.  WOODSON. 


Table  V. 


Material 


Density- 
lb.  per 
cu.  ft. 


Alundum  brick 127  to  149 


Bauxite  brick  . 
Carborundum  . 
Chromite  brick 


Concrete  , 

Feldspar   

Fire  brick   !  Ill  to  178 

Gas  retort  brick ', 


118.0 
128.0 
128.0 

170  to  180 


Spec. 
Heat 


Glass    

Glass  

Graphite  brick  

Infusorial  earth 

Insulbrix    

Iron   

Lime  stone    

Magnesia  brick 

Nonpareil  brick 

Nonpareil  brick 

Retort  brick   

Sand    

Silica  brick   

Silo-cel  brick 

Silo-cel  brick 

Silo-cel  powder  

White  buildings  brick. 


150  to  170 

112.0 

43.0 

36.0 
480.0 
170.0 

125.0 

27.0 

25.8 
116.0 
110.0 

98.5 

30.0 

31.0 

12  to  15 

118.0 


0.174 

0.20 

0.253 

0.18 


0.19 

0.118 

0.217 

0.324 

0.20 

0.295 

0.195 

0.29 

0.225 

0.2089 

0.2089 


K 
B.  t.  u, 

per 
sq.   ft. 

etc. 


7.26  to 
4.03 
9.41 

40.8 

7.19  to 

19.5 
6.38 

16.05 

10.1  to 

12.4 

11.03 
7.00 
4.33 

71.9 
0.583 
0.84 
420.0 

15.0 

17.05 
1.10 
0.477 

10.95 
2.70 
5.81 
0.67 
0.745 
0.300 

10.90 


r. 

thermal 

ohms 

per 

cu.  in. 


37.5  to 

67.7 

29.0 
6.69 
38.0  to 

14.0 

42.8 

17.0 
27.0  to 

22.0 

24.7 

39.0 

63.0 

3.8 

468.0 

325.0 

0.65 

18.2 

16.0 
248.0 
572.0 

24.9 

101.0 

47.0 
407.0 
366.0 
910.0 

25.0 


At 

temp. 

Authority 

1112 

Randolph 

1832 

Randolph 

2072 

Randolph 

2072 

Randolph 

McMillan 

212 

Randolph 

2072 

Randolph 

Marks 

McMillan 

78 

Randolph 

Randolph 

77 

Van  Dusen 

1000 

Quigley  Fur.  Spec.  Co. 

Marks 

McMillan 

2072 

Randolph 

1600 

Armstrong  Cork  Co. 

470 

McMillan 

2072 

Randolph 

McMillan 

1832 

Randolph 

470 

McMillan 

1600 

Celite  Prod.  Co. 

77 

Celite  Prod.  Co. 

1832 

Randolph 

Many  of  the  materials  given  in  the  Tables  III,  IV  and  V  are  not  heat- 
insulating  materials  in  the  ordinary  sense  of  the  term,  but  are  given  only 
for  purposes  of  comparison.  The  authorities  given  refer  to  the  value 
of  K.  K  is  expressed  as  B.  t.  u.  per  hour,  per  square  foot,  per  inch  of 
thickness,  per  °  F.  difference. 


HEAT  INSULATING   MATERIALS.  1 35 

For  flat  surfaces  of  sufficient  area  so  that  the  end  or  edge  effect 
is  relatively  small,  this  formula  can  be  used  as  given,  though  only 
approximately  correct.    AIcAIillan  gives  this  formula  as 

Q  =      *^~*t  (6) 

^  X  1  ^     ^ 

k  a 

Where 
Q  =  B.  t.  u.  per  sq.  ft.,  per  hr.  transmitted 
ts  =  temperature  of  hot  surface,  °F. 
ta  =:  temperature  of  surrounding  air  °F. 
X  z=  thickness  of  insulation  in  inches 

a  =:  surface  transmission  factor  (1/a  =^  surface  resistance) 
k  =  conductivity  of  material. 

This  takes  into  account,  not  only  the  absolute  mean  conductivity 
of  the  insulation,  but  also  the  resistance  that  is  offered  by  the 
surface  of  the  material  to  the  transmission  of  heat.  This  factor 
1/a  varies  between  wide  limits,  and  has  been  determined  for  only 
a  few  materials,  so  that  for  ordinary  calculations  0.5  is  taken  as 
the  value  of  1/a  for  still  air  conditions  and  a  good  grade  of  insulat- 
ing material  at  medium  temperatures. 

f  f 

From  formula  No.  5  it  is  evident  that  the  factor  —^ ^  is  the 

th 

determining  variable,  and  expresses  the  rate  of  temperature  drop 
with  distance  through  the  material,  and  its  limiting  value  or 
dT/dth  is  the  "temperature  gradient"  of  any  point  in  the  path  of 
flow,  assuming  that  K  is  a  true  constant  for  the  full  thickness  of 
the  materials. 

For  cylindrical  surfaces  such  as  steam  pipes,  tanks,  boilers,  etc.,. 
it  can  be  shown  that  the  heat  loss  is  equal  to 

Where 

Rj  is  inside  radius  of  covering  in  inches 

R,  is  radius  of  outside  of  covering  (or  insulation)  in  inches 


136 


J.  C*  WOODSON. 


R    is  outside  radius  of  pipe  in  inches  (usually  taken  equal  to 

Ri  in  above  equation) 
Q    is  rate  of  heat  flow  per  in.  B.  t.  u.  per  sq.  ft.,  per  hr. 
Tj  is  temperature  of  inside  of  pipe  in  °F. 
T,  is  temperature  of  outside  of  insulation  °F. 

This  is  the  formula  generally  used  for  all  cylindrical  surfaces 
and  Table  IV  gives  the  value  of  K  for  a  number  of  different 


,' 


/ 


X 


/ 


Tfrsrma/  CorK^ucityify 


A/o'  7 pare  I 


'  Mfh 


^res^u  rs  ff/oi.  vf 


zoo       300       4CC        soo        bco        700        aoo 
Te/rrperature  Difference  -Ue^rees^ 

Fig.    1. 


insulations  commonly  used  for  such  surfaces.  T,  is  ordinarily 
taken  as  the  temperature  on  the  outside  surface  of  the  covering 
or  even  room  temperature,  whereas  it  actually  refers  to  the  tem- 
perature of  the  outside  of  the  insulation,  which  for  steam  pipes 
would  be  under  the  canvas  sheathing. 

In  the  above  formulae,  numbers  4,  5,  7,  etc.,  two  assumptions 
are  made  which  are  not  strictly  correct ;  first,  that  K  is  constant  in 


HEAT  INSULATING   MATERIALS. 


137 


value  throughout  the  thickness  of  the  insulation,  and  second,  that 
the  value  of  K  varies  inversely  with  the  thickness.  The  value  of 
K  varies  with  the  temperature  as  shown  in  Fig.  1,  so  that  it  pre- 
sents a  curve  between  T^  and  T,.  It  is  a  matter  of  common  knowl- 


— \\ 

Terrrp  . 

7,ff  Si 
3C 
10 

o'r 

Ir 

V 

\ 

\ 

\ 

04 

\^ 

\ 

\\\ 

^ 

\\ 

:^===~ 

T/7/C Alness  -  //7c//es-8S  VoAfefy^es/a- 
yanaf/on  0/  /7eo/  ^rtf/7Srr7isscon  for  \^artO(/s 
t/7icf^nesses  of  mo^erio^  orr  fiai  surfaces 
Fig.  2. 


edge  that  the  insulating  value  does  not  increase  directly  with  the 
thickness,  but  so  far  no  general  law  has  been  worked  out.  Stott'' 
attempted  this  and  states  that  for  85  per  cent,  magnesia,  the  law  is 


K2 


V  th, 


(8) 


»  Power,  1902. 
10 


138 


J.  C.  WOODSON. 


Where  K^  and  K^  are  the  coefficients  of  conductivity  and  th^  and 
tho  are  the  thicknesses,  while  for  every  other  material  a  different 
constant  is  required.  These  have  not  been  accurately  determined 
as  yet.  Fig.  2  shows  this  general  relation  for  85  per  cent  magnesia 
on  flat  surfaces.     Stott's  law  will  not  hold  for  fiat  surfaces  as  it 


T/?iC/Cr>ess  of  /ls/>es/os  fo/yer  ts  C^S t/7c/> 


??(i 

?.in 

\ 

?no 

\ 

f^n 

\ 

^ 

fflO 

^ 

\ 

r-7r 

He, 

■.-/  ceo 

^ 

1 

l.i.C 

\ 

\ 

1  so 

\ 

140 

< 

\ 

f=tn 

^ 

\ 

■v 

1?0 

■    \ 

K 

^cr  br, 

r^t  rc 

tin 

N 

- 

1  2  3  4  .5  <i>  7  8 

/Vi/mber  of  tfricKnesses  cT  /ishestos  /''o/>er 

Curk'e  s/iCkV{f70  if7e/^/'ec/tt'enes3  c/"  (Tc^rrercio/  As^s/es  roper  /or 
inSi/Ja/ic/7    o/"  Brif/ti  Tin  Piptr 

Fig.  3. 


takes  into  account  the  increased  radiating  surface  on  a  pipe  or 
cylinder. 

From  the  above,  it  will  be  seen  that  these  two  conditions  tend 
to  counteract  each  other,  so  that  the  result  is  a  curve  that  will 
vary  for  each  temperature  and  each  insulating  material.  Common 
practice  is  to  follow  the  inverse  square  root  law  for  cylinders,  and 


HEAT  INSULATING   MATERIALS.  I39 

use  a  multiplier  for  flat  surfaces,  such  as  ovens,  which  really 
depend  more  upon  the  mechanical  construction  of  the  oven  than 
upon  the  characteristics  of  the  insulation.  This  multiplier  for 
formula  No.  5  varies  from  1.2  to  2.5,  depending  on  conditions. 

For  many  applications,  such  as  medium  temperature  ovens  and 
high  temperature  furnaces,  it  is  customary  to  construct  the  v^ralls 
of  layers  of  different  materials  having  different  internal  resist- 
ances. The  heat  loss  from  such  a  flat  wall  can  be  calculated  by 
the  following  formula ;  using  the  notation  and  form  of  formula 
No.  5  we  have 

ki  kj  kn 

where  th  is  the  thickness  of  the  various  layers  and  K  the  conduc- 
tivity ;  or  more  accurately  this  is  given  by  McMillan  as 

Q   = t^^ta ^^^^ 

using  the  same  notations  as  formula  No.  6.  For  cylindrical  sur- 
faces this  becomes 

Q  =  — T , '"^f^^i r     (11) 

fs  loge  tg/ri  ts  lege  tJt^  4-    -^ 

ki  kj  a. 

in  which  is  is  radius  of  outside  surface  of  insulation  r^  is  outside 
radius  of  cylinder  and  r,  equals  r^  plus  thickness  of  first  layer  of 
insulation,  t^  equals  rj  plus  thickness  of  second  layer,  etc. 

APPUCATION    TO    APPARATUS. 

The  materials  in  class  A  Table  II  are  used  successfully  only  for 
quite  low  temperature  work,  and  due  to  this  fact  the  heat  loss  is 
generally  low  regardless  of  insulation  used.  For  this  reason,  little 
attention  is  paid  to  the  proper  selection  and  too  often  a  few  layers 
of  asbestos  paper  is  used,  as  this  is  easy  to  obtain  almost  anywhere. 
It  has  been  shown  that  the  heat  loss  from  a  bare  bright  tin  pipe 
is  less  than  from  the  same  pipe  covered  with  7  layers  of  0.025- 


140 


J.  C.  WOODSON, 


"inch  (0.64  mm.)  asbestos  paper  at  approximately  180°  F.  (82°C.) 
in  the  pipe  (Fig.  3)®.  So  it  is  obvious  that  it  would  be  better 
economy  to  use  some  of  the  fibrous  or  spongy  insulations  given 
in  Table  III  even  though  the  first  cost  and  cost  of  installation  was 
higher  than  for  the  asbestos  paper. 

Class  B,  Table  II,  is  by  far  the  most  important  class,  as  most 
commercial  and  industrial  applications  fall  within  it.  To  meet  this 


Fig.  4. 


demand  there  are  dozens  of  grades  and  brands  of  commercial  insu- 
lations on  the  market.  Table  IV  gives  only  a  few  representative 
grades  of  this  class.  Much  care  should  be  exercised  in  the  selec- 
tion of  an  insulation  in  this  class,  as  many  are  good  under  certain 
conditions  and  poor  under  other  conditions  at  the  same  tempera- 
tures. For  instance  some  will  stand  soaking  in  water  and  when 
dried  out  are  apparently  as  good  as  ever.  Others  disintegrate  and 
fall  to  pieces  under  the  action  of  water  or  any  other  liquid.  Some 
grades  will  stand  up  and  hold  their  place  and  position  under  con- 

«  University  of  111.  Bulletin  No.    117. 


HEAT  INSULATING   MATERIALS. 


141 


tinual  jarring  and  vibration,  others  settle  down  and  leak  out  of 
their  retaining  walls  and  leave  an  air  space.  So  other  considera- 
tions besides  thermal  characteristics  are  important,  depending 
upon  the  particular  application^. 

In  the  application  of  these  insulations  to  electrical  apparatus, 
the  largest  per  cent  will  go  on  tanks,  boilers,  etc.,  and  on  ovens. 


Fig.  5. 


drying  cabinets,  etc.  These  are  shown  in  figures  4  and  5.  On  the 
former  the  insulation  is  usually  applied  exactly  as  pipe  covering, 
with  an  outer  surface  of  canvas,  while  with  ovens  the  insulation 
is  ordinarily  confined  between  two  thin  sheet  metal  walls.  In 
building  such  ovens,  care  should  be  exercised  so  to  construct 
them  that  there  is  a  minimum  of  continuous  through  metal  from 
inside  to  outside  of  the  wall ;  that  all  joints  are  tight  and  well 

TE.  R.  Weidlein,  Chem.  and  Met.  Eng.  24,  295,   (1921). 


142 


T.  C.  WOODSON. 


^ 

Q^ 

/ 

/ 

/ 

k 

^ 

^ 

/ 

/ 

> 

^ 

} 

/ 

/ 

^ 

1 

/ 

/ 

o           0/ 

/ 

s 

/ 

V 

Y 

Ten 

7prro^ 

yrp  -  // 

7/?^/  C 

jrt^e 

^ 

t?: 

/ 

'/ 

/ 

> 

•A/o. 
A'o.. 

•Vool  // 

r  Inst/ 1 

// 

/ 

^ 

/ 

/ 

/ 

^ocm 

Tamp. 

' 0- 

/npu 

'  t'n  Wc 

rts 

1 

tA/c/r 


JSOO 

Fig.  6. 


^ 

^ 

8 

^ 

^^ 

^ 



— 

? 

^ 

^ 

^ 

^ 

^^ 

3- 

Tt/r> 

fre.^^ 

ers'ui 

fCur^ 

? 

1 

y 

.\c.-2  '^'-> 

^ 

/ 

X 

.\c2- 

Curn 

1>7 

R 

/ 

V 

1 

r,^ 

t  m  /^ 

'>^u*»t 

Fic.  7. 


HEAT  INSULATING  MATERIALS. 


143 


packed ;  and  that  the  outer  surface  of  the  oven  is  one  that  does  not 
radiate  the  conducted  heat  readily.  Cases  are  on  record  of  similar 
ovens  in  which  one  was  finished  in  black  iron  and  one  in  bright 
galvanized  iron.  At  500°F.  (260°C.)  the  black  oven  showed  a 
radiation  loss  30  per  cent  greater  than  the  galvanized  oven.  Other 
conditions  may  have  contributed  to  this  difference,  but  it  is  believed 


2O0  3te  400  xu 

Heai  los^s  from  Bars  /ro/7  F'tpe  efsc^eferminetJ  S^  i/ar40t/3  fnifa.^ii gators 
The  figures  Yz" ,   H",  etc.,   indicate  the  diameter  of  the  pipes. 
Pig.  8. 

the  different  character  of  surface  was  the  main  cause.  In  ovens  of 
several  hundred  square  feet  radiating  surface,  this  is  a  feature  to 
be  watched  closely. 

As  brought  out  previously,  it  is  essential  that  the  specific  heat 
or  heat  absorbing  power  of  an  insulation  be  taken  into  considera- 
tions as  well  as  its  conductivity.  Fig.  6  and  7  show  curves  of 
identical  ovens,  one  with  a  commercial  grade  of  mineral  wool,  the 


144 


J,  C.  WOODSON. 


Other  with  a  commercial  grade  of  aircel  asbestos  insulation.  It  will 
be  noted  that  the  former  not  only  has  a  lower  constant  loss,  but 
comes  up  to  temperature  more  rapidly,  thus  storing  less  power  to 
be  lost  when  the  oven  is  shut  down  at  night. 

Some  of  the  insulating  materials  in  Table  V  can  be,  and'  often 
are,  used  for  temperatures  as  low  as  300° F.  (149°C.)  but  their 
real  field  lies  in  furnace  work,  where  temperatures  of  1,000  to 
3,000° F.  (538  to  1,650°C.)  are  encountered. 


Fjg.  9. 


While  these  insulators  will  stand  direct  contact  with  the  heating 
elements  and  temperatures  of  2,000°  F.,  it  is  better  practice  to  line 
the  inside  of  the  furnace  with  a  good  grade  of  refractory  fire  brick, 
and  place  the  insulating  brick  outside  of  these.  As  these  insulating 
brick  are  not  strong  mechanically,  a  layer  of  building  brick  or  red 
brick  outside  of  them  will  protect  them  and  insure  permanent 
insulating  value.  Fig.  9  shows  one  of  the  large  electrical  furnaces 
insulated  in  this  manner. 

Due  to  the  fact  that  the  absolute  mean  conductivity  of  air  is 


HEAT  INSULATING  MATERIALS.  I45 

considerably  lower  than  any  present  day  commercial  insulation, 
industrial  plant  engineers  often  try  to  increase  the  efficiency  of 
furnaces  and  boiler  settings  by  including  air  spaces  in  the  walls. 
The  results  are  invariably  the  opposite  from  those  desired.  This  is 
due  to  the  fact  that  even  thin  air  spaces  readily  set  up  convection 
currents,  and  that  the  radiant  heat  leaps  across  the  air  space  with 
little  opposition,  especially  if  the  air  space  is  close  to  the  inside 
of  the  furnace. 

Tests  by  the  U.  S.  Bureau  of  Mines,  proved  that  a  wall  of  solid 
fire  brick  or  building  brick  lost  less  heat  than  a  similar  wall  with  a 
2-inch  (51  mm.)  air  space  enclosed  in  it^.  Therefore  this  practice 
is  poor  and  should  be  abandoned  entirely  where  medium  and  high 
temperatures  are  involved. 

CONCLUSIONS. 

k 

While  there  are  numerous  grades  of  heat  insulations  on  the 
market,  there  are  none  that  can  compare  with  electrical  insulators. 
Of  all  the  different  grades,  there  are  only  a  few  fundamentally 
different  sorts,  as  some  half  dozen  items  will  cover  the  raw  mate- 
rials successfully  used.  In  all  these  materials  the  true  insulation 
value  lies  almost  entirely  in  the  entrapped  dead  air  spaces  of  their 
structure.  The  difference  between  grades  then  really  goes  back 
to  the  physical  structure  of  the  crystals  or  cells.  This  fact  leads 
many  engineers  astray  in  the  use,  in  furnace  and  oven  walls,  of  air 
spaces,  which  actually  increase  rather  than  decrease  the  heat  loss. 

The  application  of  poor  insulation  can  have  the  same  effect  as 
the  air  spaces  mentioned  above,  as  shown  by  the  University  of 
Illinois  in  tests  of  asbestos  paper  on  hot  air  pipes.    See  Fig.  2. 

While  the  conductivity  of  an  insulation  is  of  primary  importance, 
other  thermal  characteristics  must  be  considered,  such  as  specific 
heat  and  specific  weight.  The  application  also  has  to  be  considered 
with  regard  to  the  physical  properties  of  the  material. 

The  laws  of  heat  flow  are  simple  and  follow  closely  those  for 
electrical  energy,  but  are  little  used  or  understood.  This  probably 
is  due  in  part  to  the  fact  that  there  are  few  reliable  data  available 
on  the  subject,  and  of  these  the  values  given  by  different  authors 
vary  over  wide  limits. 

'  Bureau  of  Mines  Bulletin  No.  8. 
11 


146  DISCUSSION. 

It  is  the  writer's  opinion  that  a  great  deal  more  research  and 
development  work  should  be  done  along  the  lines  of  heat  insula- 
tion engineering,  as  we  have  about  come  to  a  stop  and  have 
accepted  our  present  standards  by  saying  "there  is  bound  to  be 
a  certain  amount  of  heat  lost,  and  this  is  as  good  as  we  can  do." 

I  believe  that  if  there  was  a  wider  distribution  of  available  data 
and  a  broader  dissemination  of  the  laws  and  character  of  heat 
flow  and  its  prevention,  it  would  help  to  conserve  the  national 
coal  supply  and  result  in  better  insulation  methods  being  developed. 
The  progress  of  electrically  heated  apparatus  is  dependent  to  a 
large  extent  upon  the  efficiency  of  its  insulation,  and  warrants  the 
keenest  attention  of  electrical,  chemical  and  mechanical  engineers, 
as  well  as  of  heating  and  ventilating  engineers. 


DISCUSSION. 


Carl  Hering^  :  Mr.  Woodson  spoke  of  the  thermal  ohm. 
This  is  decidedly  the  best  unit  to  use  for  electrical  engineers,  who 
deal  with  energy  in  electrical  units  but  it  is  not  the  most  conve- 
nient unit  to  use  when  you  are  dealing  with  B.  t.  u.'s  and  calories, 
as  in  the  case  of  steam  pipes.  It  is  something  Hke  using  the  cir- 
cular mil  and  the  square  mil ;  each  of  them  is  the  best  unit  to  use 
under  particular  circumstances,  because  the  conversion  factor  is 
then  unity. 

I  do  not  know  whether  Mr.  Woodson  called  attention  to 
another  point,  namely  the  effect  of  joints  in  the  insulation,  which 
is  quite  important.  For  instance,  if  in  the  wall  of  a  furnace  the 
bricks  are  placed  on  edge,  you  get  a  much  better  insulation  than 
if  placed  flat,  because  there  is  an  extra  joint,  and  a  joint  is  a  very 
important  heat  insulator.  I  have  seen  the  material  on  one  side 
of  a  joint  red  hot,  while  on  the  other  side  it  was  black.  There  is 
great  heat  insulation  in  a  joint. 

He  refers  to  the  spaces  in  finely  divided  material.  The  late  Mr. 
Stanley ,2  of  the  General  Electric  Company,  made  researches  with 
such  material  several  years  ago,  and  found  the  interesting  results 

'Consulting  Electrical  Engr.,   Philadelphia,  Pa. 
^  Personal  communication. 


HEAT  INSULATING   MATERIALS.  147 

that  as  such  material  is  compressed,  which  means  that  the  air 
spaces  become  smaller,  the  heat  insulation  at  first  improves,  but 
after  reaching  a  certain  point,  if  you  compress  it  still  more,  the 
heat  insulation  again  diminishes.  There  is  a  maximum  point  to 
which  one  should  compress  such  granular  or  fluffy  material. 

Foundrymen  have  discovered  that  if  they  whitewash  the  out- 
side of  a  furnace,  it  makes  them  feel  more  comfortable,  which 
to  us  means  that  whitewashing  the  outside  of  a  furnace  adds  quite 
a  little  to  the  heat  insulation ;  it  is  hotter  to  the  touch,  but  emits 
less  heat. 

F.  A.  J.  FitzGerald^:  Joint  heat  insulation  referred  to  by 
Dr.  Hering  is  particularly  noticeable  in  carbon  electrodes.  For 
example,  in  an  Acheson  graphite  electrode,  while  a  well-made 
joint  may  have  a  very  low  electrical  resistance,  the  resistance  of 
the  flow  of  heat  is  extremely  high.  In  furnaces  with  metallic 
resistors,  where  the  terminals  are  made  of  some  metal,  it  would 
be  interesting  to  find  out  if  one  could  get  a  low  electrical  resistance 
with  a  high  heat  resistance. 

J.  C.  Woodson  {Communicated)  :  The  matter  of  surface  or  joint 
resistance  to  heat  flow  is  gone  into  on  page  135  on  this  paper,  and 
is  largely  responsible  for  increased  insulating  value  of  insulation 
when  this  insulation  has  several  distinct  joints  or  parallel  sur- 
faces. This  is  also  a  partial  explanation  of  the  increased  effi- 
ciency found  in  insulated  walls  using  more  than  one  kind  of 
insulating  medium  in  the  same  wall.  IMore  definite  data  should 
be  made  available  on  the  true  value  of  the  surface  resistance  of 
insulating  materials  under  varying  conditions. 

*FitzGerald  Labs.,  Niagara  Falls,  N.  Y. 


A  paper  presented  at  the  Forty-second 
General  Meeting  of  the  American  Elec-  ■ 
trochemical  Society  held  in  Montreal, 
and  brought  up  for  discussion  at  the 
Forty-third  meeting  in  New  York  City, 
May  4,  1923,  President  Schluederberg  in 
the   Chair. 


METHODS    OF    HANDLING    MATERIALS    IN    THE   ELECTRIC 
FURNACE  AND  THE  BEST  TYPE  OF  FURNACE  TO  USE^ 

By    Frank   W.   Brooke* 

Abstract. 
The  author  discusses,  in  general,  the  design  of  various  electric 
furnaces,  such  as  the  plain  box  type,  the  special  box  type,  the  car 
type,  the  recuperative  and  continuous  furnaces,  and  refers  to  their 
advantages  and  disadvantages.  Attention  is  drawn  to  the  method 
of  handling  materials  for  these  furnaces,  so  that  a  uniform  tem- 
oerature  and  high  furnace  efficiency  may  be  maintained. 
^  [A.  D.  S.] 


The  outstanding  engineering  features  that  have  made  the  mod- 
ern electric  furnace  for  temperatures  up  to  980°  C.  (1800°  F.)  so 
successful  are  the  drawn  nickel-chromium  resistance  elements  and 
the  high  standard  of  thermal  insulation.  Those  who  have  read 
the  many  interesting  papers  presented  at  various  times,  especially 
by  such  authors  as  Mr.  E.  F.  Collins,  will  realize  the  careful  study 
and  pioneer  work  that  has  been  given  to  these  subjects,  and  the 
accurate  data  that  have  been  compiled  by  those  interests  that  are 
successfully  pushing  the  electric  heat-treating  furnace. 

One  problem,  however,  that  will  not  be  standardized  for  many 
years  to  come  is  the  method  of  handhng  the  material  to  be 
treated  both  while  it  is  in  the  furnace  and  out  of  the  furnace. 
The  efficiency  of  the  best  design  of  furnace  can  be  entirely  ruined 
by  poor  handling  methods.  On  the  other  hand,  given  a  particular 
method  of  handling  the  material,  it  becomes  a  problem  for  an 
experienced  furnace  engineer  to  design  the  furnace  to  meet  this 
method. 

>  Original  manuscript  received  September  20,  1922. 

»  Chief  Engineer,  Wm.  Swindell  &  Bros.,  Pittsburgh,  Pa. 

149 


I50 


FRANK   W.  BROOKE. 


Take,  for  instance,  the  handling  of  very  hght  materials  in 
many  of  the  existing  fuel-fired  furnaces.  Strength  and  resistance, 
as  long  as  possible  to  scaling,  has  necessitated  a  high  ratio  of  the 
weight  of  the  handling  medium  in  the  furnace  to  the  work  being 
treated.  It  is  quite  common  to  meet  cases  where  this  ratio  is  2 
to  1,  or  more,  which  means  that  twice  as  much  fuel  is  expended 
in  heating  the  holding  or  carrying  device  as  in  heating  the  work. 


2J> 


Fig.  1 


The  electric  furnace  designer  has  met  these  cases  in  a  proper 
engineering  manner.  The  whole  of  his  furnace  engineering  is 
taken  from  accurate  mathematical  data.  The  heat  input  is  exact 
and  constant.  He  studies  specific  heats  not  only  at  room  tem- 
peratures but  along  the  range  of  his  working  temperatures.  He 
calculates  the  heat  losses  and  so  forth.  Besides,  he  is  not  con- 
fronted with  the  eating  away  of  his  holding  devices  by  oxidation 
to  the  same  extent. 

It  is  the  object  of  this  paper  to  point  out  some  of  the  advantages 
and  disadvantages  of  various  furnace  designs  and  methods  em- 


HANDLING  MATERIALS  IN  ELECTRIC  FURNACES. 


151 


ployed  in  the  handling  of  materials  for  electric  furnaces  at  the 
higher  temperatures. 

The  plain  box  type  of  furnace  having,  say,  one  door:  This  design 
is  simple  and  the  many  methods  of  handling  material  for  it  are  too 
well  known  to  require  description.  It  allows  for  high  grade  thermal 
insulation,  and  care  need  only  be  taken  to  provide  against  the 
door  losses.  This  is  done  by  making  a  well-sealed  door,  and  by 
doubling  up  resistance  elements  at  the  door  ends.    See  Fig.  1. 

Special  box  type:  In  electric  furnaces  the  volume  of  the  furnace 
chamber  is  not  restricted  to  the  same  degree  as  in  the  fuel-fired 
furnace,  and  therefore  lends  itself  to  the  use  of  special  design, 
such  as  is  shown  in  Fig.  2.  This  furnace  is  designed  for  the 
accurate  heat  treatment  of  steel  products  having  varied  lengths, 
where  the  output  does  not  warrant  the  use  of  a  different  length 
of 'furnace  for  each  different  length  of  product. 


Fig.  2 


It  is  fitted  with  three  partition  doors,  each  having  a  special 
seal,  and  the  heating  units  are  so  placed  that  there  is  a  uniform 
heat  distribution  under  all  conditions  of  operation.  The  end 
doors  face  into  two  different  shops  and  give  an  extremely  flexible 
arrangement  for  the  class  of  work  for  which  it  was  designed,  at  a 
firgt  cost  and  an  operating  cost  both  much  lower  than  if  separate 
furnaces  were  used. 

Another  interesting  variation  of  box  type  furnace  is  one  in 
which  a  preheating  chamber  is  put  back  to  back  with  the  high  tem- 
perature chamber,  having  in  this  case  only  one  partition  wall.  This 
is  used  for  vitreous  enameling  work,  or  for  heat  treating  fine  tools. 
In  the  latter  case  the  low  temperature  chamber  can  also  be  used 
for  drawing. 


152 


FRANK   W.  BROOKE. 


For  vitreous  enameling  a  special  form  of  hearth  is  used,  either 
for  facilitating  the  handling  of  the  work,  as  in  thin  sheet  work, 
or  for  allowing  a  large  amount  of  the  heat  to  be  applied  at  the 
bottom  of  the  work  to  be  treated.  It  is  essential,  as  in  bath  tub 
work,  for  the  enameling  to  be  done  "through  the  work"  as  well  as 
from  the  upper  surface. 

Car  type:  This  gives  a  much  better  handling  arrangement,, 
especially  for  large  pieces  handled  by  the  crane.  In  considering  its 
use,  it  should  be  borne  in  mind  that  when  the  car  is  withdrawn 
from  the  furnace  the  entire  hearth  bottom  is  exposed  to  rapid  heat 
loss.  In  order  to  give  high  thermal  efficiency  and  uniformity  of 
temperature,  an  electric  furnace  car  bottom  is  much  more  massive 
than  that  used  in  the  ordinary  fuel-fired  furnace,  and  when; 
exposed  to  direct  radiation  loss  it  loses  a  greater  quantity  of  heat. 


Fig.  3. 

For  this  reason  the  time  cycle  of  electric  car  type  furnaces 
should  be  arranged  to  ensure  the  car  being  rapidly  unloaded  and 
reloaded,  or  else  a  dummy  furnace  should  be  provided,  as  shown 
in  Fig.  3,  whereby  there  is  an  exchange  of  heat  not  only  between 
the  cooling  of  a  hot  charge  to  a  cold  charge  but  also  from  car  to 
car.  For  heat  treatments  requiring  a  long  time  of  heating,  hold- 
ing and  cooling  in  the  furnace,  it  is  advantageous  to  build  the  door 
directly  on  to  the  car. 

Recuperative  furnaces:  Fig.  3,  already  referred  to,  is  perhaps 
the  simplest  form  of  this  type  of  furnace.  Providing  the  time 
between  heats  is  not  too  long  and  the  output  warrants  two  fur- 
naces, its  use  invariably  pays.  Where  the  heats  are  of  long  dura- 
tion, the  steady  radiation  loss  of  the  dummy  furnace  defeats  its 
economy. 

A  better  form  of  recuperative  furnace  is  shown  in  Fig.  4.    This 


HANDLING  MATERIALS  IN  ELECTRIC  FURNACES. 


153 


arrangement  allows  for  recuperation  from  one  heating  chamber, 
but  on  the  other  hand  requires  three  chambers  per  unit.  It 
requires  also  considerable  rail  switching,  but  has  given  excellent 
fuel  economy. 

Still  another  form  of  recuperation  is  shown  in  the  counter-flow 
type  of  furnace,  the  various  designs  of  which  are  too  numerous 
to  illustrate.  Fig.  5  shows  a  car  type  of  counter-flow  furnace,  now 
used  in  the  annealing  of  gray  iron  castings. 

The  furnace  is  divided  into  seven  sections,  each  corresponding 
to  a  car  length.  Only  the  middle  section  is  equipped  with  heating 
units  and  has  a  short  dividing  wall.     The  heating  units  are  sus- 


FiG.  s. 

pended  on  the  four  walls  so  that  each  preheated  car  is  heated  from 
both  sides.  On  either  side  of  the  heating  section  are  two  cooling 
and  two  preheating  sections. 

The  trains  of  parallel  cars  move  in  opposite  directions,  and  each 
moves  one  car  length  at  equal  intervals  of  time.  Therefore  a 
heated  car  and  its  charge  leaving  a  heating  section  is  placed 
directly  beside  a  car  and  its  charge  partially  preheated,  and  is 
given  a  period  of  interchange  of  heat.  It  then  moves  forward 
another  car  length,  and  some  of  its  remaining  heat  is  given  up  to  a 
cold  car  and  its  charge  coming  from  the  transfer  chamber.  There 
is  also  a  transfer  chamber  at  either  end,  enclosed  in  lightly  insu- 
lated walls,  as  the  cars  after  being  reloaded  still  retain  a  consider- 
able amount  of  heat  well  worth  conservinsf. 


154 


FRANK  W.  BROOKE. 


It  is  interesting  to  note  that  the  first  electric  furnace  installed 
of  this  design  had  a  partition  wall  running  the  entire  length  of  the 
furnace,  in  which  port  holes  had  been  left  at  the  top  and  bottom. 
The  design  of  this  furnace  was  taken  from  a  previous  fuel  type 
design,  but  the  engineers  built  the  partition  wall  in  such  a  way 
that  the  portions  in  the  recuperative  chambers  could  be  readily 
removed  if  necessary,  They  soon  found  this  to  be  necessary. 
It  was  also  found  that  instead  of  longitudinal  partition  being 
necessary,  transverse  partitions  between  each  section  were  abso- 


Charging  y. 


end 


i+ 


^ 


Chain^ 


Drive 
shaFt 


Fig.  6. 


^/ 

^ 

'////////////////////////////////////////////, 

V//A 

Charging 
end 

1 

i 

'Gv' 

I 

k 

J  yv////////////////Zvy/y//y////y////////Ai(^  ^ 

X 

^ 
^ 

'\p\ 

1 

^jjgin  \. ~~' 

^ 

Fig.  7. 


lutely  essential  for  temperature  uniformity  and  efficient  heat 
exchange.  The  work  which  is  small  and  of  very  thin  gray  iron 
castings  is  placed  on  superimposed  trays,  and  one  end  of  each  tray 
forms  part  of  the  transverse  partition. 

Continuous  furnaces:  In  electric  furnace  salesmanship  it  has 
been  the  author's  experience  that  the  prospective  electric  furnace 
user  thinks  first  of  all  of  a  continuous  furnace  to  do  the  work, 
feeling  that  a  continuous  furnace  is  a  labor-saving  furnace,  a  fuel 
saver,  routes  his  work  better  and  is  more  modern.  There  are, 
however,  many  points  of  electric  furnace  engineering  to  be  con- 
sidered before  a  complete  knowledge  of  these  points  can  be  given, 
and  it  is  surprising  how  often  it  can  be  shown  (excepting  in  such 


HANDLING  MATERIALS  IN  ELECTRIC  FURNACES, 


155 


cases  as  producers  of  large  quantity  of  uniform  products  as  in  the 
motor  car  industry)  that  a  continuous  furnace  is  not  the  best  all- 
round  furnace  to  install,  often  proving  a  disappointment  to  the 
prospective  customer. 

Fig.  6  shows  a  continuous  furnace  used  for  heat  treating  light 
flat  discs,  and  working  satisfactorily.  The  limiting  feature  of  such 
a  design  is  the  temperature  of  the  conveyor.  If  carried  above 
650°  C.  (1200"  F.)  the  stretch  becomes  a  serious  consideration. 


Fig.  8. 


A  surprising  feature  in  such  a  design  is  the  loss  of  heat  caused 
by  the  exposed  ends  of  the  chain.  It  is  easy  to  see  that  the  loss 
would  be  material,  but  actual  experiments  in  one  particular  design 
show  a  thermal  efficiency  of  about  18  per  cent.  This  can  be  greatly 
improved  by  the  proper  boxing-in  of  the  ends,  but  where  chain 
area  must  be  available  for  loading  and  unloading  this  enclosing  is 
limited. 

A  better  arrangement  for  carrying  light  work  through  the  fur- 
nace is  shown  in  Fig.  7.  This  consists  of  three  chain  systems,  only 
the  middle  one  of  which  is  always  inside  the  highly  insulated  walls 


156 


FRANK   W.  BROOKE. 


of  the  furnace.  The  charging  and  discharging  systems  do  not 
attain  a  temperature  sufficiently  high  to  cause  a  serious  heat  loss. 
When  a  chain  system  is  totally  enclosed  in  the  way  shown,  it  must 
be  remembered  that  this  chain  attains  the  temperature  of  heat  bal- 
ance of  the  furnace,  which  is  decidedly  higher  than  the  chain  shown 
in  Fig.  6.  It  is  also  more  difficult  to  take  care  of  heat  losses 
through  the  journals  when  higher  shaft  temperatures  and  stretch 
adjustment  must  be  taken  care  of. 


Fig.  9. 


Fig.   10. 


The  author  is  at  the  present  time  engaged  in  the  designing  of 
three  different  types  of  electric  furnaces,  in  each  of  which  the 
mechanical  handling  of  the  material  is  of  vital  importance,  as  the 
material  must  not  be  marked  and  furnace  efficiency  and  tempera- 
ture uniformity  is  of  utmost  importance.  He  is  not  at  liberty  to 
publish  these  designs  now,  but  hopes  at  some  future  occasion  to 
give  a  paper  on  this  subject. 

For  the  continuous  conveying  of  work  through  an  electric  fur- 
nace at  the  higher  temperatures,  the  so-called  "doughnut"  furnace 
offers  an  excellent  method  of  carrying  out  the  operation.  A  plan 
diagram  of  this  is  shown  in  Fig.  8.  It  has  many  advantages.  The 
conveying  hearth  is  made  of  refractory  materials,  and  can  there- 
fore handle  materials  at  the  limiting  temperatures  of  electric  fur- 


HANDLING  MATERIALS  IN  ELECTRIC  FURNACES.  157 

nace  heat  treatment.  The  charging  and  discharging  doors  are  adja- 
cent, and  in  many  operations  one  man  can  attend  to  both.  The 
thermal  efficiency  is  high,  as  only  the  hearth  seals  under  the  fur- 
nace offer  any  insulating  difficulty.  It  is  used  to  special  advantage 
in  the  heat  treatment  of  gears  and  small  machine  parts.  When  the 
work  to  be  treated  is  of  a  uniform  character,  automatic  loading 
chutes  can  be  adapted,  and  the  work  can  be  "swept-out"  at  the 
discharge  end. 

The  "push"  type  of  furnace  is  perhaps  one  of  the  most  efficient 
types  of  continuous  furnaces  used,  and  is  shown  in  Fig.  9.     It 


Fig.  11. 

is  restricted  to  work  of  a  uniform  shape,  and  work  which  will 
push  in  a  long  column  without  bridging.  This  tendency  to  bridge 
can  be  lessened  by  inclining  the  furnace  and  thus  lessening  the 
friction  to  push.  It  is  an  excellent  method  of  conveying  such  parts 
as  small  connecting  rods,  push  rods,  small  cylindrical  pieces,  etc., 
and  fits  in  with  production  heat  treatment  for  small  parts.  The 
speed  of  travel  can  be  varied  through  a  wide  range  and  the  dis- 
charge end  can  be  sealed  in  the  quenching  tank. 

The  "gravity  roll"  type  of  furnace  has  the  same  degree  of  effi- 
ciency and  usefulness  as  the  "push"  type,  but  is  still  further 
restricted  to  products  that  will  roll  by  gravity.    The  feed  through 


158 


FRANK,  W.  BROOKE. 


the  furnace  can  be  regulated  by  a  discharge  timing  gear,  shown  in 
its  simplest  form  in  Fig.  10. 

The  "push"  furnace  is  simple  in  construction,  has  a  low 
operating  cost,  and  has  a  lower  first  cost  than  the  many  other 
types  of  furnace.    See  Fig.  11. 

The  "walking  beam"  type  of  furnace  forms  one  of  the  fasci- 
nating means  of  handling  materials  through  an  electric  furnace.  It 
is  restricted  to  uniform  shapes  and  sizes,  such  as  automobile 
crankshaft,  connecting  rods,  bars  of  steel,  etc.  In  the  more  simple 
type  of  walking  beam,  shown  in  Fig.  12,  the  beams  are  hned  with 


Fig.   12. 


~~i  nnnnnnnnnnnn,xnk\nn\\nnnnnj^nnn.nns'^.^^ 


^ 


^ 


Fig.   13. 


Fig.  14. 


refractory  material,  but  the  continuous  top  surfaces  of  the  beams 
must  be  kept  in  a  true  horizontal  plane,  so  that  the  points  of  con- 
tact with  the  work  are  made  with  each  beam  simultaneously  ;  other- 
Av-ise  the  work  will  creep  more  along  one  side  of  the  furnace  than 
the  other.  A  good  beam  mechanism  which  ensures  work  tempera- 
ture uniformity  and  furnace  efficiency  adds  considerably  to  the 
first  cost  of  the  furnace. 

An  interesting  variation  of  the  "walking  beam"  furnace  is  the 
type  shown  in  Fig.  13,  designed  to  treat  shells,  crankshafts,  short 
shafts,  axles,  etc.,  in  such  a  way  that  the  axes  of  the  work  are 
always  parallel  and  there  can  be  no  jamming  in  the  furnace.    The 


HANDLING  MATERIALS  IN  ELECTRIC  FURNACES.  1 59 

way  the  work  progresses  through  the  furnace  is  clearly  shown  in 
Fig.  14. 

For  the  heat  treatment  of  steel  balls  and  similar  materials,  a 
design,  such  as  is  shown  in  Fig.  15,  offers  an  excellent  method. 
The  author  does  not  know  of  any  furnace  of  this  type  in  which 
electricity  is  used  as  a  fuel,  but  there  is  no  reason  why  electric 


Fig.  is. 

resistance  units  cannot  be  applied  to  give  all  the  inherent  values  of 
the  electric  furnace. 

The  figures  shown  and  types  referred  to  are  very  general.  It 
would  be  difficult  to  give  references  that  would  be  complete  and 
fair. 


A  paper  presented  at  the  Forty-third 
General  Meeting  of  the  American  Elec- 
trochemical Society  held  in  New  York 
City,  May  4,  1923,  President  Schlueder- 
berg  in  the  Chair. 


THE  CONVERSION  OF  DIAMONDS  TO  GRAPHITE  AT  HIGH 
TEMPERATURES.' 

By  M.  DeKay  Thompson  and  Per  K    Frouch.^ 
Microscopic  Work  By  J.  L.  Gillson. 

Abstract. 

It  is  shown  conclusively  that  diamonds  change  slowly  at  1650° 
C.  to  a  substance  that  gives  the  Brodie  test  for  graphite,  and 
that  the  velocity  of  this  reaction  is  increased  about  26  times  by 
an  increase  of  100°  above  this  temperature. 


INTRODUCTION, 


In  looking  over  the  previous  work  on  this  subject,  there  is 
found  quite  a  lack  of  agreement  in  the  results  obtained  by 
different  investigators.  This  may  be  accounted  for  by  inaccuracy 
in  the  temperature  measurements,  as  in  many  cases  it  is  not 
clearly  explained  how  these  were  made,  and  in  others  it  is  stated 
that  temperatures  were  only  estimated.  The  object  of  the  present 
investigations  was  to  determine  at  what  temperature  diamonds 
change  to  some  other  form  of  carbon  with  appreciable  velocity, 
and  to  determine  whether  or  not  this  other  form  is  graphite. 
It  would  be  hopeless  to  attempt  to  verify  experimentally  the 
calculations  of  Weigert^  and  of  Pollitzer*,  according  to  whom 
the  temperature  below  which  diamond  is  stable  is  respectively 
372°  and  340°   C,  because  the  reaction  velocity  is  too  slow  at 

*  Manuscript    received    February    1,    1923. 

*  Contribution  from  the  Electrochemical  Laboratory  of  the  Rogers  Laboratory  of 
Physics,  and  Geological  Laboratory,  Massachusetts  Institute  of  Technology,  Cambridge. 

^Abegg's  Handbuch   der  anorg.    Chem.   Ill,   2,  p.   47    (1909). 

*  Die  Berechnung  chemisher  Affinitaten  nach  dem  Nernstschen  Warmetheorem. 
p.    136    (1912). 

i6i 


l62  M.  DE  KAY  TH0MP50X  AND  PER  K.  FROUCH. 

such  low  temperatures.  According  to  Boeke^,  if  the  heats  of 
combustion  determined  by  Roth  and  Wallasch®  are  used  in  this 
calculation  in  place  of  those  of  Berthelot,  used  by  Weigert  and 
by  Pollitzer,  the  result  is  that  at  atmospheric  pressure  diamond 
is  unstable  at  all  temperatures  down  to  the  absolute  zero. 

All  of  these  calculations  were  made  by  the  Nernst  heat  theorem, 
the  data  for  which  are  the  specific  heats  of  diamond  and  graphite 
down  to  temperatures  near  the  absolute  zero,  and  the  total  energ}'' 
change  of  the  reaction :  diamond  — >  graphite. 

PREVIOUS   WORK. 

The  references  in  the  foot  note^  are  to  the  most  important 
previous  work  on  this  subject,  but  in  order  to  save  space  they 
will  not  be  considered  at  length.  They  may  be  briefly  summarized 
as  follows: 

Diamond  changes  to  graphite  in  the  arc,  but,  according  to 
Moissan,  not  at  2000°  C. ;  according  to  Doelter,  diamonds  are 
only  blackened  in  the  surface  up  to  2500°  C,  while  Vogel  and 
Tammann  say  diamond  changes  to  graphite  at  1200°.  It  was 
on  account  of  these  discordant  results  that  the  following  work 
was  undertaken. 

EXPERIMENTAL. 

A  small  Arsem^  vacuum  furnace  was  used  for  heating  the 
diamonds.  These  were  placed  in  the  center  of  a  carbon  crucible 
on  a  small  carbon  plate  as  shown  in  Fig.  1.  In  the  first  experi- 
ment the  cover  over  the  diamond  was  omitted.  The  crucible 
had  a  cover  with  a  hole  through  which  the  temperature  was 
taken  with  a  Leeds  and  Northrup  optical  pyrometer,  which  was 
compared  with  a  standard  optical  pyrometer,  and  the  small  cor- 
rection applied.  The  temperature  had  to  be  taken  through  a 
mica  window  in  the  top  of  the  furnace.    The  effect  of  the  mica 

"Centralbl.   Min.  Geol.   und  Palaontologie  321    (1914). 

«B.  d.d,  chem.  Ges.  46,  896  (1913). 

'  Moissan,  Le  Four  Electrique,  157  (1897);  Parson  and  Swinton.  Proc.  Roy.  Soc. 
80,  784  (1907-8);  Vogel  and  Tammann,  Z.  phvs.  Chem.  69,  600  (1910);  Doelter,  Mona- 
thaft  f.  Chemie  32,  280   (19)1). 

*  Trans.  Am.  Electrochem.  Soc.  9,  153  (1906). 


C0NVER5I0X   OF   DIAMONDS    TO   GRAPHITE.  I63 

window  on  the  temperature  indication  was  determined  by  read- 
ing the  temperature  of  another  furnace  with  the  window  between 
the  hot  body  and  the  pyrometer,  and  without  the  window.  It 
was  found  that  at  1260°  with  the  window  in  place  the  correction 
to  be  added  is  40',  at  1550'  the  correction  is  60',  and  at  1710' 
it  is  75'.  Corresponding  corrections  were  applied  to  the  readings 
through  the  window.  The  temperature  of  the  plate  covering  the 
diamond  was  taken.  This  must  have  been  very  nearly  under 
black  body  conditions.  It  is  believed  that  temperatures  are  cor- 
rect to  within  30'  C. 

The  furnace  was  evacuated  to  only  5  mm.  in  the  first  four  runs, 
after  this  two  pumps  were  connected  in  series,  and  the  vacuum 
was  reduced  to  less  than  1  mm.  In  any  case  there  was  little 
chance  of  any  oxygen  getting  at  the  diamond  with  so  much  other 
carbon  present,  and  all  of  the  oxygen  present  could  not  have 
burnt  more  than  a  small  fraction  of  the  diamond  if  it  burnt 
nothing  else. 

The  power  in  the  furnace  was  kept  constant  by  means  of  a 
carbon  plate  resistance  and  was  read  by  a  wattmeter.  With  con- 
stant power  the  temperature  remained  constant. 

The  method  of  procedure  decided  on  was  to  heat  for  a  given 
time  to  different  temperatures  and  examine  the  product  micro- 
scopically and  chemically.  The  chemical  test  consisted  in  the 
Brodie  test  for  graphite,  by  oxidizing  in  a  solution  of  nitric  acid 
and  potassium  chlorate.^  For  microscopic  examination  a  small 
piece  was  cracked  oft  and  immersed  in  a  solution  of  sulfur 
in  methylene  iodide,  which  has  a  high  index  of  refraction. 

The  method  of  carrying  out  the  Brodie  test  was  to  digest 
the  sample  at  60°  C.  with  finely  ground  potassium  chlorate  and 
concentrated  nitric  acid  for  24  hours.  It  was  washed,  dried  and 
the  treatment  repeated.  Three  treatments  changed  a  sample  of 
graphite  to  yellow  graphitic  oxide.  Acheson  graphite  was  tested 
and  gave  a  yellow  product,  while  coke  dissolved  completely  giv- 
ing the  solution  a  yellow  color,  probably  due  to  iron. 

The  results  are  contained  in  Table  I. 

'  For  the  description  of  the  Brodie  test  see  Moissan,  The  Electric  Furnace,  SO 
(1904);  Donath,  Der  Graphit,  IS  (1904);  Selvig  and  Ratliff,  Trans.  Am.  Electrochem. 
See.  37,  121   (1920). 


164 


M.  DE  KAY  THOMPSON  AND  PER   K.   FROLICH. 


Table  I. 
Effects  of  Heating  Diamonds  to  High  Temperature. 


No. 

Time   required  to   reach 
highest     temperature 

Duration    of 

heating   at   highest 

temp. 

Highest 
temp. 

Vacuum 

1 

3  hr. 
2  hr. 

2060-2090 
1000-1015 

5  mm. 

2 

30  min.  to  800° 

5  mm. 

1  hr.  at  800° 

3 

30  min. 

2.5  hr. 

1150-1200 

5  mm. 

4 

30  min. 

5.5  hr. 

1135-1155 

5  mm. 

5 

30  min. 

6.5  hr. 

1250 

less  than 
1  mm. 

6 

In  20  min.  raised  to  1250° 
40  min.,   1250-1350 

6  hr. 

1350 

less  than 
1  mm. 

7 

30  min. 

4.5  hr. 

1680 

less  than 
1  mm. 

8 

25  min.  to  1530 

25  min. 

1535 

less  than 
1  mm. 

9 

30  min.  to  1520 

5  hr. 

1535 

less  than 
1  mm. 

10 

9.5  hr. 

1650 

Less  than 

1  mm. 

11 

12  hr. 

1650 

Less  than 

1  mm. 

12 

30  min. 

2.5  hr. 

1865 

Less  than 
1  mm. 

13 

25  min. 

2  hr. 

1760 

Less  than 
1  mm. 

14 

25  min. 

1  hr. 

1760 

Less  than 
1  mm. 

REMARKS. 

Exp.  1.  Diamond  completely  destroyed.     Product  gave  Brodie  test. 

Exp.  2.  Diamond  No.  2.  Light  grey  color.  Transparent ;  dark  spot 
appeared  in  center.     No  superficial  change. 

Exp.  3.  Diamond  No.  2.  Dark  grey.  Filled  with  small  dark  spots. 
Small  piece  cracked  off  showed  double  refraction.  This  small  sample 
contained  a  dark  spot. 

Exp.  4.  Diamond  No.  2.     No  appreciable  change. 

Exp.  5.  Diamond  No.  2.  No  superficial  change.  Looked  darker.  Alore 
internal  dark  spots. 

Exp.  6.  Diamond  No.  2.  More  spots  and  larger.  Under  microscope 
they  had  a  brown  color  and  spongy  appearance.  The  dark  color  of  tiie 
diamond  seemed  to  be  due  to  many  cracks  which  totally  reflected  the 
light  in  air,  but  in  the  sulfur  solution  in  methylene  iodide  the  diamond 
was  clear  except   for  the  spots. 

Exp.  7.  Diamond  No.  2.  Completely  black.  Blackened  paper  slightly. 
The  spots  inside  now  black,  not  brown  as  at  first.     The  black  spots  have 


CONVERSION    OF   DIAMONDS    TO   GRAPHITE. 


165 


a  metallic  luster  when  observed  in  reflected  light  under  the  microscope 
identical  with  that  of  graphite. 

Exp.  8.  Diamond   No.   3.     Slightly  grey. 

Exp.  9.  Diamond  No.  4.     Turned  black. 

Exp.  10.  Diamond  No.  2.  Brittle  and  easily  broken.  Part  of  surface 
shiny,  rest  dull  black.  Blackened  paper  like  a  pencil.  One  small  corner 
still  transparent  in  small  piece  examined  under  microscope.  Still  hard 
enough  to  scratch  steel. 

Exp.  11.  Diamond  No.  2.  Diamond  was  cracked  into  a  number  of 
pieces.  All  treated  with  HNO3  and  KCIO3.  Small  ones  changed  to 
graphitic  acid.  The  large  pieces  did  not  dissolve,  but  were  of  lighter 
color.     Total  duration  at  1650°,  26  hr. 

Exp.  12.  Diamond  No.  5.     Black  residue.     Gave  Brodie  test. 

Exp.  13.  Diamond  No.  6.     Diamond  was  split  to  pieces. 

Exp.  14.  Diamond  No.  7.  Diamond  appeared  at  about  the  same  degree 
of  change  as  No.  2  after  26  hr.  at  1650°.  Parts  were  found  thrown  out 
from  the  central  hole  in  the  plate,  as  though  the  diamond  exploded. 


-i 

T| 

I      U-*— H 

-JJ 

F.-g.a 


Fi 


•9 


Fig.  1.    Diamond  (D)  in  Graphite  Crucible. 

Fig.  2.  Diamond  No.  7.  Fragment  of  diamond  lying  with  outside  face  uppermost. 
Heavy  coating  of  graphite  on  the  outside  with  numerous  specks  of  graphite  scattered 
through  the  still  transparent  but  cracked  interior.  Heavy  cross-hatching  represents 
the  graphite  coated  surface;  the  light  represents  the  interior  of  the  clear  diamond,  x  150. 

Fig.  3.  Diamond  No.  2  after  E-xp.  No.  7.  Cleavage  Face  broken  from  diamond 
shows  dendritic  development  of  graphite  on  the  face,  as  at  "a,"  and  the  development 
along  incipient  cleavage  planes,  as  at  "b."  Dots  on  edges  are  graphite  developing  in 
the  interior  of  the  diamond.     Somewhat  generalized,  x  300. 


l66  DISCUSSION. 

DISCUSSIOX  OF  RESULTS. 

The  above  experiments  show  conclusively  that  diamonds  change 
slowly  at  1650°  C.  to  a  substance  that  gives  the  Brodie  test. 
This  change  takes  place  about  26  times  as  rapidly  by  raising  the 
temperature  100°.  Diamonds  turn  dark  at  1000°,  but  this  is 
largely  due  to  numerous  cracks  causing  absorption  of  light  from 
total  reflection.  The  cracks  were  probably  produced  by  the  small 
black  spots.  These  spots  are  doubtless  the  beginnings  of  change 
to  graphite,  producing  strains  and  double  refraction.  Doelter 
also  found  double  refraction  in  diamonds  that  had  been  heated. 
Experiments  8  and  9  show  these  cracks  are  not  due  alone  to 
thermal  expansion  and  contraction,  for  if  they  were  the  two 
diamonds  would  have  had  the  same  appearance. 


DISCUSSION. 


W.  C.  Arsem^  :  There  seems  to  be  a  great  deal  of  confusion  in 
regard  to  the  nature  of  the  different  forms  of  carbon  that  we 
meet  with.  There  is  no  question  about  what  we  mean  by  diamond 
or  pure  graphite.  They  are  definite  crystalline  substances,  and 
X-ray  analysis  has  shown  them  to  have  definite  molecular  struc- 
ture and  definite  lattices.  When  we  consider  the  different  so- 
called  amorphous  carbons  and  so-called  graphites  of  indefinite 
character,  there  is  some  doubt. 

It  seems  to  me  that  when  carbon  is  set  free  from  an  organic 
compound  or  derived  from  some  different  form  by  heat,  if  the 
conditions  are  not  favorable  for  an  arrangement  or  rearrangement 
of  the  atoms  to  form  a  definite  crystalline  structure,  we  must 
have  a  mixed  lattice  structure. 

Amorphous  carbon  derived  from  sugar  will  not  have  a  definite 
crystalline  structure.  It  will  be  a  heterogeneous  arrangement  of 
atoms.  You  may  have  different  characteristic  groupings  here  and 
there  throughout  the  whole  mass  or  solid  particle,  but  no  definite 
repeated  chain  or  pattern  structure,  such  as  you  have  in  a  crys- 
talline substance.     In  the  same  way  we  would  expect,  on  heating 

'  Consulting    Chemical    Engr.,    Schenectady,    N.    Y. 


COXVERSIOX    OF  DIAMONDS    TO    GRAPHITE.  167 

the  diamond  to  temperatures  far  below  the  point  of  mobiHty,  that 
the  atoms  can  not  rearrange  themselves  to  form  a  definite  lattice 
structure.  So  that  we  w-ould  not  expect  to  get  pure  graphite  by 
heating  the  diamond  any  more  than  we  would  expect  to  get  pure 
graphite,  from  heating  certain  amorphous  carbon  to  a  hio-h 
temperature. 

Xow,  as  to  the  Brodie  test,  I  believe  that  when  pure  crystalline 
graphite  is  oxidized,  that  the  so-called  graphitic  acid,  the  yellow 
organic  substance  which  is  formed,  has  a  definite  chemical  struc- 
ture, and  can  be  identified,  but  when  obtained  from  a  carbon  with 
a  mixed  lattice  structure,  the  same  as  v.e  would  expect  to  get 
on  heating  most  amorphous  carbons  or  the  diamond,  the  yellow 
oxidation  product  does  not  have  a  definite  structure.  It  may 
be  a  mixture  of  substances,  or  it  may  be  a  complex  organic  com- 
pound with  a  mixed  lattice  structure  corresponding  to  the  struc- 
ture of  the  carbon  from  which  it  is  derived. 

The  Brodie  test,  in  the  light  of  our  present  knowledge  of 
molecular  structure  can  not  be  regarded  as  a  satisfactory  test 
for  graphite,  or  even  for  the  presence  of  graphitic  structure, 
until  more  is  known  of  the  chemical  nature  of  the  yellow  oxidation 
products  obtained  with  diflerent  carbons. 

I  presented  a  paper  before  the  Society  some  years  ago,  and 
mentioned  the  heating  of  a  diamond  to  3,CXX)°.  I  found  a  specific 
gravity  of  the  product  that  was  about  1.8,  whereas,  pure  graphite 
should  have  a  specific  gravity  of  2.25.  I  do  not  think  that  pure 
graphite  can  result  from  heating  the  diamond  under  these  con- 
ditions, and  I  do  not  believe  that  the  data  given  in  the  present 
paper  support  that  conclusion. 

CoLix  G.  FiXK- :  I  should  like  to  refer  briefly  to  experiments 
recently  carried  out  at  the  laboratories  of  the  Siemens-Halske 
Company.^  They  started  out  with  amorphous  carbon  in  fine  thread 
form,  and  heated  this  up  to  temperatures  of  3,000  to  3,600°. 
What  they  obtained  was  a  graphite  of  2.23  specific  gravity. 

They  tested  these  fine  filaments,  and  curiously  enough  they 
could  be  rolled,  and  their  length  extended  by  10  per  cent.  Fur- 
thermore, they  could  take  the  filaments  and  wind  them  into  a 

2  Consulting   Metallurgist,  New   York  City. 
3Z.   Elektrochem.   29,  171    (1923). 


1 68  DISCUSSION. 

very  small  coil,  and  then  straighten  them  out  again  just  as  though 
they  were  made  of  lead. 

F.  A.  J.  FitzGerald*  :  I  understand  from  Dr.  Fink  that  the 
Siemens-Halske  graphite  filaments  had  a  positive  temperature 
coefficient  of  electric  resistivity  like  the  Gem  filaments,  on  which 
Dr.  Whitney  was  working  some  years  ago  at  Schenectady. 

Does  Mr.  Arsem  consider  the  Brodie  test  a  definition  of  graph- 
ite as  Berthelot  suggested?  It  is  probable  that  the  best  test  for 
graphite  is  found  in  a  study  of  the  lattice  structure  by  X-ray 
examination. 

W.  C.  Arsem  :  On  some  of  the  work  on  Gem  filaments  we 
produced  a  graphite  with  a  specific  resistance  one-third  that  of 
mercury  and  a  pronounced  positive  temperature  coefficient.  Some 
of  this  was  produced  in  thin  sheets  that  had  the  characteristics 
of  tin  foil  and  could  be  rolled  up,  but  I  never  succeeded  in  rolling 
any  of  it  thinner  and  extending  its  length.  I  had  hopes  of  doing 
so,  but  about  that  time  we  lost  interest  in  it,  because  the  tungsten 
lamp  looked  more  promising,  and  all  efforts  were  devoted  to  that 
work. 

S.  C.  LiND^ :  It  seems  to  me  the  important  thing  in  connection 
with  this  paper  is,  whether  the  spots  that  Prof.  Thompson  has 
found  can  be  examined  by  the  X-ray  method.  We  agree  it  is 
perfectly  satisfactory  for  graphite  or  diamond,  but  if  it  can  not 
be  used  on  the  spots  as  they  exist  in  diamond,  I  do  not  see  that 
it  would  be  of  much  use  in  that  connection. 

I  would  like  to  mention  one  or  two  points  in  connection  with 
some  observations  we  have  recently  made.  I  do  not  wish  to  go 
fully  into  the  matter  because  we  are  publishing  the  results  in 
the  Journal  of  the  Franklin  Institute  next  month.  In  connection 
with  some  work  in  the  coloring  of  diamonds  with  alpha  rays, 
we  found  under  some  conditions  that  "carbon  spots"  were  pro- 
duced. I  will  not  call  these  spots  graphite,  because  we  have  not 
proved  they  are.  The  results  were  produced  at  ordinary  tem- 
perature, showing  that  high  temperature  is  not  necessary  to  con- 
vert diamond  into  some  other  form  of  carbon.  These  spots  were 
small,  round  spots  in  the  interior  of  the  diamond,  which  does  not 

«  FitzGerald  Labs.,  Niagara   Falls,  N.   Y. 

s  Chief  Chemist,  U.   S.   Bureau  of  Mines,  Washington.  D.  C. 


CONVERSION    OF  DIAMONDS   TO    GRAPHITE.  1 69 

support  the  idea  that  they  are  produced  directly  by  alpha  rays. 
In  some  instances  they  appeared  to  grow  from  a  center.  There 
was  a  halo  surrounding  the  round  spot,  which  made  us  think 
they  grew  from  a  center  acting  as  a  seeding  point  in  the  diamond. 
Furthermore,  at  the  temperature  of  a  blast  lamp  they  disap- 
peared and  reverted  to  diamond.  Whether  this  reversion  will 
satisfy  Dr.  Fink  or  not,  I  do  not  know,  but  it  satisfied  the  jewelers 
who  had  furnished  the  diamonds,  because  otherwise  they  would 
have  stood  a  considerable  financial  loss.  These  spots  disappeared 
absolutely  and  went  back  to  perfect  diamonds,  showing,  I  think 
for  the  first  time,  that  there  is  some  form  of  carbon  other  than 
diamond,  which  will  revert,  under  conditions  we  have  alwavs 
supposed  were  not  characteristic  of  diamond  in  the  stable  form, 
into  diamond.  Whether  or  not  this  controverts  the  phase  rule, 
I  will  leave  to  you  to  decide. 

Ancel  St.  John®:  Regarding  the  question  as  to  whether  the 
X-ray  would  be  able  to  tell  about  the  change  from  diamond  to 
graphite,  I  can  say  that  it  would.  But  I  do  not  know  how  soon 
the  work  on  this  will  be  completed. 

I  have  on  my  desk  a  few  rather  small  diamonds.  They  are 
unfortunately  not  as  large  nor  as  numerous  as  I  would  like  to 
play  with,  but  they  will  do.  For  with  a  very  small  quantity  of 
material  you  can  get  results  by  the  X-ray  method  which  are 
incontrovertible.  Thus,  in  a  photograph  of  the  pattern  from 
diamond,  A.  W.  HulF  has  obtained  twenty-five  of  the  twenty- 
seven  lines  calculated  for  the  structure  assigned  to  diamond,  and 
in  a  pattern  from  graphite,  nineteen  of  the  forty  lines  required 
by  its  proposed  structure.  By  mounting  specimens  in  a  dififerent 
way,  I  have  recorded  thirty  of  these  graphite  lines  in  a  single 
pattern. 

One  of  the  problems  I  want  to  attack  when  I  get  time  is  to 
find  out  just  what  happens  to  the  crystal  structure  when  a  dia- 
mond is  heated  until  it  goes  through  some  of  the  transformations 
mentioned. 

Colin  G.  Fink:  Dr.  Lind  says  you  can  get  the  spots  at  low 
temperature  by  the  action  of  alpha  rays  on  the  diamond.     Does 

"  Union   Carbide  and   Carbon   Research   Lab.,   Inc.,   Long   Island   City,   N.    Y. 
"Physical    Review,    10,   695    (1917). 

12 


1 70  DISCUSSION. 

he  mean  to  infer  that  in  the  Arsem  furnace  rays  from  the  in- 
candescent crucible  may  have  had  some  effect  ?  That  it  was  not 
the  heat  alone,  but  radiation  from  the  hot  crucible  that  may  have 
brought  about  the  spots? 

S.  C.  LiND :  I  did  not  intend  to  draw  such  an  inference.  All 
I  meant  to  point  out  was  that  it  is  evidently  possible,  under  some 
conditions  other  than  high  temperatures,  to  change  diamond  into 
some  other  form  of  carbon.  I  did  not  mean  to  question  Prof. 
Thompson's  results. 


A  paper  presented  at  the  Forty-third 
General  Meeting  of  the  American  Elec- 
trochemical Society  held  in  New  York 
City,  May  4,  1923,  President  Schlueder- 
berg    in   the    Chair, 


THE  RELATION   BETWEEN   CURRENT,   VOLTAGE  AND    THE 
LENGTH  OF  CARBON  ARCS' 

By  A.  E.  R.  Westman^ 

Abstract. 

An  account  of  the  early  work  on  the  electric  arc  has  been  given 
by  Mrs.  Ayrton^,  and  a  summary  of  present  knowledge  by  Stein- 
metz^.  Each  of  these  writers  is  the  author  of  a  formula  connect- 
ing current,  voltage,  and  arc  length ;  but  in  the  experiments  on 
which  these  formulas  were  based  the  currents  only  ran  up  to  30 
amperes  or  so,  and  it  seemed  desirable  to  add  to  the  experimental 
material.  The  present  paper  gives  an  account  of  measurements 
made  with  currents  up  to  770  amperes. 


I.     INSTRUMENTS  AND  APPARATUS. 

The  voltage  over' the  arc  was  measured  directly  by  a  Weston 
model  45  voltmeter  with  scale  0-150  volts  (Vj  Fig,  1).  In  addi- 
tion, in  order  to  get  closer  readings,  a  second  voltmeter  (Vo)  of 
the  same  type  but  with  scale  0-3  volts  was  used  to  measure  the 
difference  between  the  e.  m.  f.  of  a  storage  cell  (B)  and  the  p.  d. 
over  70  ohms,  forming  part  of  a  resistance  of  1,000  ohms  in  paral- 
lel with  the  arc;  in  the  run  of  March  8th  referred  to  later,  the 
voltage  over  the  arc  was  given  by  the  relation  : 

e  =  28.0  +  0.340  div.  (1) 

where  "div."  is  the  reading  Vg  in  fiftieths  of  a  volt. 

The  current  was  measured  by  a  Weston  millivoltmeter  model 
45  (M  Fig.  1)  in  conjunction  with  appropriate  shunts  (S).    The 

'  Original  Manuscript  received  August  21,  1922. 

2  Department   of   Electrochemistry,    Univ.    of   Toronto,   Toronto,    Ont. 

3  H.    Ayrton,    "The    Electric    Arc,"    Electrician    Pub.    Co.,    London,    1908 
*  Chem.  and  Met.  Eng.,  22,  458  (1920). 

171 


172 


A.  E.   R.   WESTMAN. 


millivoltmeter,  the  shunts,  and  the  voltmeters  mentioned  above, 
were  all  calibrated  by  comparison  with  Weston  standards. 

The  current  was  derived  from  a  100  kw.  110  v.  d.  c.  compound 
wound  generator  driven  by  a  steam  turbine  in  the  power  house  of 
the  University ;  during  my  experiments  no  other  load  was  carried 
on  this  generator.  Owing  to  a  breakdown  in  the  power  house, 
however,  niany  of  the  measurements  (including  those  of  March 
8th)  had  to  be  made  with  power  supplied  by  the  Toronto  Electric 
Light  Company. 

I    I  I  U  V. 

oooooooo 


Fig.    1 


The  rheostat  (R  Fig.  1)  in  series  with  the  arc  consisted  of  ten 
parallel  coils  of  No.  12  iron  (telegraph)  wire  immersed  in  run- 
ning water.  Each  coil  had  a  resistance  when  new  of  about  half 
an  ohm,  and  was  provided  with  its  own  switch ;  the  resistance  in 
series  with  the  arc  could  thus  be  varied  from  0.5  to  0.05  ohm.  The 
relation  between  current  /,  and  voltage  over  the  arc,  e,  is  given  by 


=  B  —  iR 


(2) 


where  E  is  the  voltage  at  the  source  of  supply,  and  R  the  resistance 
of  the  rheostat.  If  the  temperature  of  the  rheostat  wires,  and 
hence  their  resistance,  were  independent  of  the  current,  and  if 
there  were  no  voltage  drop  in  the  leads  carrying  in  the  current,  R 
and  E  would  be  constants,  and  the  relation  between  current  and 
voltage  over  the  arc  would  be  linear  so  long  as  the  setting  of  the 
rheostat  remained  unchanged ;  this  is  the  relation  assumed  by  Mrs. 
Ayrton  and  others  in  their  work  with  low  amperage.  Under  the 
conditions  of  my  own  experiments,  however,  the  temperature  of 
the  rheostat  wires  was  far  from  independent  of  the  current,  the 


THE  LENGTH  OF  CARBON  ARCS.  173 

water  flowing  hot  from  the  wire  box;  and  yet  a  graph  of  the 
200  current  and  voltage  readings  of  the  run  of  March  8th  referred 
to  below  shows  a  straight  line  from  ^  =  20  to  ^  =  55  volts,  with 
the  equation 

e  —  137.4  —  0.292  {  (3) 

There  were  no  greater  variations  (±3  per  cent  of  the  current) 
than  can  be  ascribed  to  fluctuations  in  the  line  voltage,  but  this 
was  in  the  neighborhood  of  110  volts,  not  137.4.  The  following 
discussion  may  serve  to  clear  this  matter  up,  although  the  simple 
conditions  postulated  below  are  of  course  not  strictly  fulfilled  in 
practice. 

Since  the  heat  generated  in  the  rheostat  was  carried  off  by  a 
constant  flow  of  water,  the  average  temperature  of  the  latter 
would  be : 

fw—  fo=  oRP  (4) 

where  t^  is  the  temperature  of  inflowing  water,  R  the  resistance 
of  the  wires,  and  a  a  constant  depending  on  the  rate  of  flow. 

When  a  steady  state  has  been  attained,  the  difiference  between  t 
(the  temperature  of  the  wires)  and  t_^  will  be  proportional  to  the 
flow  of  heat  through  the  stationary  film  of  water  at  their  surface, 
or 

t    -  t^  =  bRP  (5) 

Finally,  the  resistance  of  the  wires  may  be  expressed  by 

R  =  [I   +  c  a  —  to  )\  ,  Ro  (6) 

where  R  is  the  resistance  at  the  temperature  t  ,  and  c  the  tem- 
perature coefficient  of  the  resistance  of  iron  wire. 

From  equations  4,  5  and  6  may  be  found  an  expression  for  R 
as  a  function  of  the  current ;  this  when  introduced  into  equation 
(2)  gives:— 

e  =  E  -  Roi  /  {I—  c  (a  ^  b)  Roi-\  =  E  —  Ai;  (1  —  BP)   (7) 

Setting  £  r=  110,  and  choosing  A  and  B  so  that  for  ^  =  20  and 
^  =  55  the  values  of  i  are  those  given  by  the  empirical  relation  of 
equation  (3),  there  results: 

^  =  110  —  0.1720  //(I  —  1.403  X  10«  r)  (8) 


174 


A.  E.  R.  WESTMAN. 


The  following  table  gives  for  a  number  of  values,  of  e,  the  values 
of  i  calculated  from  the  rational  formula  (8)  and  those  from  the 
convenient  empirical  formula  (3).  Between  280  and  400  amperes, 
that  is,  while  the  p.  d.  over  the  cell  varies  from  55  to  20  volts,  the 
greatest  difference  is  5  amperes ;  thus  for  the  purpose  of  calcu- 
lating the  current,  Mrs.  Ayrton's  straight  line  construction  is  accu- 
rate enough  even  though  the  conditions  for  which  it  was  originally 
deducted  are  not  fulfilled ;  if  it  be  used  to  calculate  the  potential 
difference  from  the  current,  however,  an  error  of  1.4  volts  would 
be  introduced  at  c  :=  35. 

Table  I. 


p.   d.   over  arc 

Amp.    Eq.    8 

Amp.  Eq.  3 

Diff. 

110.00 

0.0 

94.0 

94.0. 

92.47 

100.0 

154.0 

54.0 

73.41 

200.0 

219.5 

19.5 

55.00 

282.5 

282.5 

0.0 

50.63 

300.0 

297.6 

—  2.4 

43.96 

325.0 

320.4 

—  4.6 

36.76 

350.0 

345.0 

-  5.0 

29.26 

375.0 

370.8 

—  4.2 

20.00 

402.5 

402.5 

0.0 

Assuming  c  =  0.0055  and  /q  =  10°  C,  the  values  and  A  and  B 
chosen  above  lead  to  a  temperature  of  53°  C.  for  the  wire  when  e 
=z  55,  and  to  a  temperature  just  below  100°  C.  when  e  =  20. 
This  corresponds  to  the  experimental  conditions ;  for  the  amount 
of  water  sent  through  the  rheostat  was  just  sufficient  to  prevent 
rumbling  with  the  heaviest  current  it  was  proposed  to  employ. 

The  arc  in  my  experiments  was  struck  between  a  stationary 
vertical  anode  and  a  vertical  cathode  carried  by  a  jib ;  thus  the 
cathode  was  always  above  the  anode.  Both  electrodes  were  four 
inches  in  diameter  and  were  manufactured  from  a  petroleum  coke 
base  by  the  National  Carbon  Company. 

The  jib  holding  the  cathode  was  raised  and  lowered  by  an  elec- 
tric winch  motor  set.  Each  revolution  of  the  armature  of  the 
motor  raised  the  jib  0.553  mm.,  and  by  attaching  a  Veeder  revolu- 
tion counter  to  the  armature  the  position  of  the  electrode  could 
be  known  to  half  a  millimeter;  owing  to  the  great  weight  of  the 


THE  LENGTH  OF  CARBON  ARCS.  175 

jib,  365  kg.  (800  lb.),  and  smooth  bearings,  there  was  very  Uttle 
slack.  After  trial,  however,  this  method  was  discarded  in  favor  of 
measurement  with  a  cathetometer.  This  was  an  upright  instru- 
ment provided  with  telescope  and  vernier,  manufactured  by  Becker, 
Hatton  Wall,  London;  it  was  sighted  on  a  very  fine  line  ruled 
with  india  ink  on  drawing  paper  attached  to  the  cathode  holder. 
Readings  were  made  to  0.01  mm.  which  was  closer  than  necessary, 
and  perhaps  rather  beyond  the  accuracy  of  the  instrument. 

II.    CONDITIONS  FOR  A  STEADY  ARC. 

By  far  the  greater  part  of  the  time  so  far  devoted  to  this  inves- 
tigation has  been  spent  in  finding  conditions  under  which  a  steady 
arc  can  be  maintained;  for  no  measurements  at  all  are  possible 
unless  the  voltmeter  readings  are  definite  and  show  no  variations 
except  the  steady  rise  due  to  the  electrodes  burning  away.  The 
principal  difficulties  met  with  and  overcome  were  those  due  to  air 
draughts,  magnetic  flux  from  the  cables,  eccentric  arcs,  "hum- 
ming," and  "groaning";  minor  difficulties,  due  to  spontaneous 
shifting  of  the  arc,  still  remain.  Owing  to  the  heavy  currents  used 
in  my  work,  the  disturbances  from  these  sources  were  incompar- 
ably greater  than  those  met  with  by  previous  experimenters. 

Air  draughts:  At  first,  a  circular  wall  of  sihca  firebrick  was 
built  around  the  arc,  no  "bond"  (mortar,  etc.)  being  used  in  the 
construction ;  then  a  top  of  firebrick  was  added,  and  as  it  appeared 
'  that  each  added  protection  improved  the  arc,  the  brick  wall  was 
replaced  by  a  steel  furnace  shell  76  cm.  in  diameter  and  91  cm. 
high  (30  X  36  in.)  lined  with  clay  firebrick  and  covered  with  large 
firebrick  blocks  and  sheet  asbestos.  To  increase  the  heat  insula- 
tion, the  space  between  lining  and  shell  was  filled  with  crushed 
firebrick.  Finally,  the  iron  column  described  below,  which  acted 
as  magnetic  shield,  was  almost  closed  at  the  bottom  with  brick. 
With  this  arrangement,  the  air  draught  was  reduced  to  a  slow, 
evenly  distributed  stream  moving  parallel  to  the  electrodes ;  any 
further  reduction  of  the  air  supply  resulted  in  irregular  currents, 
which  blew  the  arc  sideways. 

Magnetic  disturbances:  An  arc,  being  a  flexible  conductor 
carrying  a  current,  is  blown  to  one  side  by  any  magnetic  flux  at 
an  angle  to  the  direction  of  the  arc.  Such  a  flux  was  set  up  by  the 


176  A.  E.  R.   WESTMAN. 

loop  of  the  circuit  of  which  my  arc  was  a  part ;  and  as  no  conve- 
nient arrangement  of  the  cables  could  be  found  with  which  the  dis- 
turbance from  this  source  was  not  noticeable,  I  resorted  to  the  use 
of  iron  shields.  After  experimenting  with  various  arrangements 
of  iron  pipes  and  sheets,  satisfactory  results  were  obtained  by 
surrovmding  the  arc  with  a  laminated  iron  column  38  cm.  high 
(15  in.),  3.8  cm.  thick  (1.5  in.)  and  30.5  x  30.5  cm.  (12  x  12  in.) 
in  (rectangular)  cross  section,  built  up  of  fifteen  square  iron 
window  frames.  The  "magnetic  shield"  so  constructed  offers  a 
path  of  low  reluctance  in  the  horizontal  direction,  and  only  after 
it  was  installed  could  the  arc  be  steadied  sufficiently  for  accurate 
readings. 

Centering  the  arc:  If  new  electrodes  be  used,  with  ends  turned 
flat  in  the  lathe,  it  may  be  hours  before  the  arc  settles  down  to  burn 
steadily  in  one  place,  and  the  crater  thus  formed  is  seldom  in  the 
center  of  the  anode.  If  the  anode  surface  be  covered  with  pow- 
dered carbon,  the  arc  chooses  its  final  position  more  quickly,  but 
seldom  selects  the  center.  If  a  small  hole  (10  mm.  in  diameter, 
15  mm.  deep)  be  drilled  in  the  anode,  the  arc  will  stay  in  that 
spot  if  it  happens  to  run  into  it. 

Having  obtained  a  centered  arc  by  the  use  of  such  a  hole,  I 
allowed  it  to  burn  for  six  hours ;  and  using  the  resulting  cratered 
anode  and  pointed  cathode  as  patterns,  turned  a  pair  of  new 
electrodes  to  the  same  shape  and  size  in  a  lathe,  making  the  crater, 
however,  only  about  a  quarter  as  deep  as  that  of  the  used  electrode. 
The  anode  was  then  mounted  in  the  furnace  and  powdered  carbon 
poured  over  it,  filling  the  crater  and  covering  the  flat  top  to  the 
depth  of  5  or  6  mm. ;  the  carbon  powder  obtained  by  dusting  off 
the  electrodes  after  a  run  proved  best  for  this  purpose.  When  the 
arc  was  struck  (using  one  coil,  about  0.5  ohm,  in  the  rheostat)  it 
centered  at  once,  and  all  the  carbon  dust  burned  out  of  the 
crater  in  a  few  minutes.  In  all  subsequent  experiments,  artificially 
shaped  electrodes  were  employed ;  by  the  use  of  a  tool  made  for 
the  purpose  in  the  University  machine  shop,  the  anode  could  be 
reshaped  without  removing  it  from  the  furnace. 

Lozv  voltage  arcs:  When,  after  the  arc  had  been  struck  as 
described,  the  current  was  increased  by  lowering  the  resistance  in 


THE  LENGTH  OF  CARBON  ARCS.  177 

the  rheostat,  it  often  happened  that  the  arc  was  blown  to  one  side 
and  extinguished.  In  the  end  I  found  that  this  could  be  avoided 
by  lowering  the  cathode  until  the  potential  difference  over  the  arc 
was  only  about  twenty  volts  before  raising  the  current  to  the 
desired  amperage. 

The  general  belief,  in  which  I  shared,  that  an  arc  between  car- 
bons cannot  be  maintained  at  less  than  about  40  volts,  kept  me 
from  discovering  this  method  sooner ;  as  a  matter  of  fact,  there 
is  no  difficulty  in  maintaining  a  20- volt  arc  for  well  on  to  an  hour 
under  the  conditions  of  my  experiments.  Both  yirs.  Ayrton's 
f  ormula° : 

e  =  38.88  +  2.074  /  +  (11.66  +  10.54  l)/i  (9) 

and  Steinmetz's  formula" : — 

e  =  36  4- 52  {I  +  0.8)/\/T~  (10) 

contain  a  constant  term  (38.88  or  36  volts  respectively)  commonly 
referred  to  as  the  "counter  electromotive  force  of  the  arc."  These 
formulas  are  in  good  agreement  with  the  results  of  experiment  up 
to  currents  of  thirty  amperes,  but  it  is  obvious  that  if  this  inter- 
pretation is  to  be  adhered  to,  the  counter  e.  m.  f.  must  decrease 
with  increase  of  current. 

Humming  arcs:  When  the  arc  gave  out  a  humming  noise  (direct 
current,  not  alternating  current  was  used,)  inspection  through  col- 
ored glasses  showed  that  it  was  flickering,  /.  e..  that  the  luminous 
part  of  the  arc  was  enlarging  and  contracting  periodically.  A 
high  note  accompanies  rapid  pulsation  of  the  arc,  but  very  slow 
changes  in  volume  (4  or  5  times  per  second)  are  inaudible.  Once 
begun,  the  humming  usually  gets  louder  and  louder,  but  without 
much  change  in  note. 

In  a  humming  arc  the  voltage  over  the  electrodes  oscillates ;  if 
the  hum  is  loud,  the  needle  of  the  voltmeter  is  set  in  rapid  vibration 
and  may,  in  addition,  swing  over  a  range  of  10  to  15  volts  every 
couple  of  seconds.  If  the  hum  is  not  loud,  voltage  readings  may 
be  secured ;  but  instead  of  rising  at  the  usual  rate  as  the  electrodes 

'  Loc.  cit.,  p.   184;  /  gives  arc  length  in  mm. 

•Trans.   .\m.   Inst.   Elec.   Eng.   25.   803    (1906),   and   Chem.   and   Met.    Eng.   22.   462 
(1920).     In  equation  (10)   above  /  gives  arc  length  in  cm. 

13 


178  A.  E.  R.  WESTMAN. 

burn  away,  the  voltage  may  remain  almost  stationary  for  half  an 
hour  or  more.  Stationary  voltage  over  the  arc  may  be  accepted  as 
indicating  an  approaching  hum. 

A  humming  arc  always  left  the  cathode  with  a  very  pointed  tip, 
like  those  observed  by  ]\Irs.  Ayrton  in  her  "hissing  arcs."  The 
only  cure  is  to  shut  off  the  current,  cool  the  cathode,  and  rasp  off 
the  tip,  preferably  to  a  diameter  of  about  30  mm.  (see  below). 

Swinging  arc:  On  one  occasion  I  was  bothered  for  days  by  an 
arc  which  emitted  a  note  rising  in  pitch  when  the  cathode  was 
lowered  and  falling  when  it  was  raised.  At  first  I  took  this  for 
a  "humming"  arc,  but  reshaping  the  cathode  failed  to  remove  the 
trouble.  In  the  end,  a  vertical  saw  cut  through  the  anode,  which 
had  become  elliptical  in  horizontal  section,  revealed  a  crack  in  the 
carbon  across  the  bottom  of  the  crater;  the  arc  had  travelled 
backwards  and  forwards  along  this  crack,  and  dug  out  a  trench 
with  a  craterlet  at  each  end. 

Groaning  arcs:  When  the  arc  is  burning  normally  the  whole 
crater  is  so  hot  that  it  is  impossible  to  distinguish  a  specific  anode 
spot;  but  if  the  arc  groans,  inspection  shows  that  a  white  hot 
spot  is  jumping  from  the  bottom  to  the  edge  of  the  crater  and 
back  again ;  when  the  spot  is  on  the  edge,  the  arc  is  blown  out- 
wards (away  from  the  center  of  the  crater)  and  the  "groan"  is 
heard.  At  first  the  spot  stops  only  two  or  three  seconds  in  each 
position ;  but  this  period  soon  lengthens,  rising  in  four  or  five 
minutes  to  about  ten  seconds,  whereupon  the  groan  rises  to  a 
shriek,  and  the  arc  is  extinguished. 

In  a  groaning  arc  the  voltage  over  the  electrodes  falls  about 
thirty  volts  each  time  the  anode  spot  reaches  the  edge  of  the 
crater,  and  rises  again  when  it  returns  to  the  bottom. 

It  is  evident  that  an  arc  can  "groan"  only  if  the  cathode  is  too 
near  the  edge  of  the  crater  in  comparison  with  its  distance  from 
the  bottom.  The  groaning  can  be  stopped  (1)  by  making  the 
cathode  narrower  at  the  point,  (2)  by  widening  the  crater 
throughout,  (3)  by  bevelling  the  edge  of  the  crater,  thus  making 
the  latter  wider  at  the  top,  or  (4)  by  cutting  down  the  edge  of  the 
crater,  thus  making  the  latter  more  shallow.  All  four  methods 
have  their  limitations.     If  the  cathode  is  too  narrow  at  the  tip,  the 


THE  LENGTH  OF  CARBON  ARCS. 


179 


arc  will  hum ;  if  the  crater  is  too  wide, 
the  arc  will  burn  small  craterlets  in 
the  bottom  and  jump  from  one  to  the 
other;  if  its  walls  are  too  much  bev- 
elled, the  same  thing  will  occur;  and 
if  the  crater  is  too  shallow,  the  arc  will 
not  remain  in  it.  Electrodes  cut  to  the 
dimensions  of  the  accompanying  sec- 
tion (Fig.  2),  which  is  drawn  from 
the  templates  used  in  the  laboratory, 
will  give  a  good  arc  for  4  or  5  hours 
with  currents  up  to  300  or  400 
amperes. 


III. 


Fig.   2 


DIRECT  DETERMINATION  OF 
ARC   LENGTH. 

In  Mrs.  Ayrton's  work  the  "length 
of  the  arc"  was  defined  to  be  the  verti- 
cal distance  between  the  point  of  the 
cathode  and  the  horizontal  plane  drawn 
through  the  edge  of  the  crater ;  length 
zero  did  not  mean  that  the  carbons 
were  in  contact,  but  that  the  point  of 
the  cathode  was  just  entering  the 
crater.  It  is  obvious  that  in  an  arc 
between  unformed  carbons  burning  with  constant  current 
and  "constant  arc  length"  so  defined,  the  vertical  distance 
between  the  electrodes  will  not  be  constant  at  all,  but  will  keep 
increasing  until  the  crater  attains  its  final  dimensions.  This  did  not 
escape  Airs.  Ayrton,  who  studied  the  changes  in  voltage  accom- 
panying the  formation  of  the  crater ;  nevertheless,  she  adhered  to 
her  own  definition  of  "I"  and  measured  it  on  an  image  of  the 
arc  projected  by  a  lens.  In  my  own  work,  the  depth  of  the  crater 
was  such  a  large  fraction  of  the  whole  distance  betvveen  the  elec- 
trodes— it  often  reached  35  mm.,  while  the  total  length  never 
exceeded  60  mm. — and  the  voltage  so  evidently  depended  on  the 
total  length  and  not  on  that  above  the  crater's  edge,  which  more- 
over with  my  high  currents  was  too  irregular  to  afford  a  fixed 


l8o  A.  E.  R.  WESTMAN. 

point  for  the  measurements,  that  I  decided  to  measure  the  total 
vertical  length  as  the  "length  of  arc."  This  quantity,  of  course,  is 
greater  than  Mrs.  Ayrton's  "T  ;  to  avoid  confusion  I  indicate  it  by 
the  capital  letter  "L". 

Measurement  of  L:  When  a  steady  arc  had  been  secured  the 
height  of  the  cathode  was  determined  (by  revolution  counter  or 
cathetometer),  voltage  and  current  were  read,  and  the  circuit 
broken.  To  protect  crater  and  tip  from  burning  away,  powdered 
carbon  was  then  poured  down  a  pipe  into  the  crater,  and  the 
cathode  was  lowered  until  its  tip  was  also  protected  by  the  powder. 
The  furnace  was  then  allowed  to  cool  (about  ten  hours  was 
required),  the  carbon  was  blown  out,  the  height  of  the  cathode 
measured  again,  and  a  plastic  ball  of  moistened  fire  clay  was 
squeezed  into  place  so  as  to  take  an  impression  of  crater  and  tip. 
When  the  clay  had  hardened,  the  length  of  the  arc  when  the  cir- 
cuit was  broken  {L,  as  defined  above)  was  determined  by  caliper 
measurements  of  the  clay  model  plus  the  difference  between  the 
two  cathetometer  readings.  With  care,  such  measurements  can 
be  made  within  a  millimeter  or  two;  if  the  electrodes  are  too  hot, 
the  surface  of  the  model  will  be  rough ;  if  they  have  cooled  to 
room  temperature,  the  clay  takes  three  or  four  hours  to  harden, 
and  a  good  model  results. 

Every  such  measurement  involved  shutting  down  the  furnace 
for  at  least  twelve  hours ;  I  have  obtained,  so  far,  eight  fairly 
good  results,  besides  a  number  of  failures.  In  Table  II,  those 
marked  with  a  star  were  measured  with  the  cathetometer,  the 
other  two  with  the  revolution  counter. 

For  comparison,  the  values  of  /  from  Mrs.  Ayrton's  formula 
(Eq.  9)  ancl  from  Steinmetz's  formula  (Eq.  10)  are  given  in 
Table  II,  both  of  them  in  millimeters. 

IV.     DETERMINATION    OF    CHANGES    IN    VOLTAGE    CONSEQUENT    ON 
KNOWN   CHANGES  IN  ARC   LENGTH. 

With  a  view  of  obtaining  results  more  rapidly,  it  was  decided 
to  measure  the  changes  in  voltage  caused  by  lowering  or  raising 
the  cathode  small  measured  distances.  Such  determinations  should 
give  a  series  of  values  of  dc/dL  (or  dc/dl,  which  is  the  same  thing) 


THE  LENGTH  OF  CARBON  ARCS. 


I8I 


under  the  condition  that  di  =  —  k.de  (see  Eq.  3)  ;  from  these 
by  integration,  using  results  with  the  clay  model  as  integration 
constants,  the  relation  between  e  and  L  might  be  determined.  A 
number  of  fairly  good  runs  were  made  with  currents  from  300 
to  400  amperes,  of  which  that  of  March  8,  1922,  was  the  best, 
i.  e.,  showed  the  smoothest  burning  arc.  Four  others  were  made 
with  currents  from  700  to  800  amperes,  but  in  none  of  them 
was  the  arc  steady  enough  to  give  satisfactory  results  by  this 
method ;  I  hope  to  get  better  results  soon. 


Table  II. 


amp. 


280 
*311 
*350 
*385 

390 
♦712 
*732 
*770 


148  X  10' 
135  X  10" 
147  X  10' 
119x10^ 
168  xlO' 
356  xlO' 
315x10- 
262  x  lO' 


volts 


53.0 
43.5 
42.0 
31.0 
44.5 
50.0 
43.0 
34.0 


L 

mm. 


56 
42 
42 
30 
42 
55 
49 
43 


/   Ayrton 

/    Steinmetz 

mm. 

mm. 

6.7 

•      46.7 

2.2 

17.4 

1.5 

13.6 

-  3.8 

—10.9 

2.1 

24.2 

5.42 

63.8 

2.0 

28.4 

—  2.3 

2.7 

In  each  of  these  runs,  current,  voltage  over  the  arc,  and  "divi- 
sions" on  the  second  voltmeter  were  recorded  every  minute ;  Table 
III  gives  "divisions"  and  time  in  minutes  after  striking  the  arc 
for  the  run  of  March  8th ;  the  voltage  over  the  arc  can  be  obtained 
from  the  number  of  divisions  by  means  of  Eq.  (1),  and  the  cur- 
rent by  Eq.  (3).  Below  28  volts  the  second  voltmeter  could  not 
be  used ;  the  numbers  entered  in  Table  III  under  "volts"  give 
the  readings  of  the  first  voltmeter  (scale  0-150  volts). 

Table  IV  gives  the  cathetometer  readings  (in  millimeters)  for 
March  8th ;  when  the  cathode  was  raised  the  cathetometer  reading 
increased.  At  ^  =  57,  i.  e.,  57  minutes  after  striking  the  arc,  the 
cathode  was  moved  to  the  left  in  the  hope  of  steadying  the  arc; 
at  t  =  111,  it  was  moved  again  to  the  left  to  stop  humming;  at 
^  =  138  it  was  shifted  again  to  stop  groaning.  The  effect  on  arc 
length  caused  by  these  movements  can  only  be  guessed. 

For  ^  =  231,  a  clay  model  gave  L  =  42  mm. ;  values  of  L  for 
other  values  of  t  (above  t  =  138)  were  calculated  from  the  cathe- 


1 82 


A.  E.  R.  WESTMAN. 


tometer  readings,  C,  on  the  assumption  that  the  carbons  burn 
away  at  the  uniform  rate  of  9.0  mm.  per  hour  irrespective  of 
the  wattage.    Thus  for  138  <  f  ^  231  L  =:  C  —782.6  +  0.15  ^ 


Table  III. 


Time 

Time 

Time 

Time 

Time 

Time 

min. 

Volt 

min. 

Div. 

min. 

Div. 

min. 

Div. 

min. 

Div. 

min. 

Div. 

17 

21.2 

50 

14.8 

87 

52.0 

121 

60.0 

156 

70.0 

197 

38.0 

18 

21.2 

51 

15.0 

88 

53.0 

122 

42.5 

157 

70.5 

198 

38.0 

18 

21.2 

52 

34.2 

89 

54.0 

123 

43.5 

199 

38.0 

19 

21.2 

53 

33.5 

90 

40.2 

124 

43.8 

165 

71.5 

200 

37.5 

20 

21.3 

54 

34.0 

90 

35.0 

125 

45.5 

166 

73.0 

55 

34.0 

126 

47.0 

167 

73.5 

201 

17.0 

21 

21.7 

56 

44.0 

91 

380 

127 

48.2 

168 

74.5 

202 

17.0 

22 

21.9 

59 

44.5 

92 

40  0 

128 

4S.0 

169 

75.5 

203 

17.5 

23 

21.6 

60 

44.8 

93 

41.2 

129 

48.0 

170 

53.0 

204 

18.0 

24 

21.5 

94 

42.0 

130 

48.5 

205 

18.5 

25 

21.6 

61 

45.0 

95 

43  0 

171 

53.5 

206 

44.0 

26 

21.6 

62 

45.9 

96 

70.0 

131 

49.2 

172 

54.0 

207 

43.0 

27 

23.7 

63 

46.2 

97 

69.5 

132 

49.8 

173 

54.5 

208 

43.0 

28 

24.0 

64 

46.5 

98 

69.5 

133 

51.0 

174 

55.0 

209 

42.5 

29 

24.2 

65 

47.0 

99 

69.5 

134 

230 

175 

37.0 

210 

42.2 

30 

28.6 

66 

47.2 

135 

23.7 

176 

39.5 

67 

37.0 

leo 

69.5 
59.5 
60.5 
61.0 

136 

25.5 

177 

44.5 

211 

42.5 

min. 

Div. 

68 

36.5 

137 

27  0 

178 

45.0 

212 

42.5 

31 

3.0 

69 

37.5 

101 
102 
103 

139 

15.0 

179 

46.0 

213 

4.0 

32 
33 

3.5 
3.5 

70 

38.0 

140 

17.0 

180 

46.0 

214 
215 

3.5 
3.5 

34 

5.7 

71 

38.5 

104 

62.2 

181 

47.0 

216 

3.5 

35 

6.1 

72 

39.2 

105 

63.0 

141 

18.0 

182 

47.5 

217 

4.0 

36 

7.0 

73 

39.4 

106 

63.0 

142 

19.0 

183 

28.0 

218 

4.5 

37 

79 

74 

40.5 

107 

40.5 

143 

19.5 

184 

29.0 

219 

27.5 

38 

8.5 

75 

40.7 

108 

43.0 

144 

20.0 

185 

30.0 

220 

28.5 

39 

9.0 

76 

40.5 

109 

46.0 

145 

20.0 

186 

31.5 

77 

50.0 

110 

49.0 

146 

20.5 

187 

32.5 

221 

29.0 

40 

10.0 

78 

51.0 

147 

48.0 

188 

33.2 

222 

30.0 

41 

10.2 

79 

52.0 

112 

54.5 

148 

47.0 

189 

34.5 

223 

31.0 

42 

10.6 

80 

52.7 

113 

55.0 

149 

46.5 

190 

7.0 

224 

2.0 

43 

11.0 

114 

56.0 

150 

46.5 

225 

2.0 

44 

12.0 

81 

53.8 

115 

57.0 

191 

7.0 

226 

2.0 

45 

12.0 

82 

43.5 

116 

55.2 

151 

47.0 

192 

7.2 

227 

2.0 

46 

12.5 

83 

47.2 

117 

58.0 

152 

47.5 

193 

7.8 

228 

45.5 

47 

13.0 

84 

48.5 

118 

58.5 

153 

48.0 

194 

8.0 

229 

45.0 

48 

13.8 

85 

50.0 

119 

59.0 

154 

70.0 

195 

8.2 

49 

14.0 

86 

51.0 

120 

60.0 

155 

1 

70.0 

196 

37.5 

231 

44.5 

For  110  <  ^  <  139,  I  have  replaced  the  subtrahend  782.6  by 
780.4,  thus  making  an  allowance  of  2.2  mm.  for  the  adjustment  of 
the  cathode  at  t  —  138.  For  58  <  f  <  139,  the  subtrahend  783.1 
is  used,  which  is  within  half  a  millimeter  of  the  first.     Before 


THE  LENGTH  OF  CARBON  ARCS. 


183 


1 84 


A.  E.  R.  WESTMAN. 


t  =z  57,  the  subtrahend  is  788.1.  These  last  three  values  had  to  be 
chosen  arbitrarily,  and  there  is  no  independent  check,  as  the  adjust- 
ment consisted  in  moving  the  cathode  sideways  in  order  to  secure 
a  steady  arc ;  but  the  value  employed  for  calculating  the  last  hour 
and  a-half  of  the  run  was  obtained  from  the  direct  determination 
with  the  clay  model. 

Table  IV. 

Note:     The  cathode  was  moved  to  give  the  new  cathetometer  reading 
C  millimeters  about  half  a  minute  after  the  time  entered  under  t. 


t 

c 

t 

C 

t 

C 

16 

806.4 

107 

809.4 

183 

792.1 

27 

809.3 

111 

(moved) 

190 

783.8 

30 

813.0 

112 

819.4 

196 

791.3 

52 

812.8 

122 

804.6 

201 

785.6 

57 

(moved) 

134 

795.2 

206 

791.7 

59 

817.9 

138 

(moved) 

213 

780.5 

67 

812.4 

139 

794.3 

219 

786.8 

77 

816.7 

147 

802.7 

224 

778.8 

91 

811.0 

154 

810.9 

228 

790.0 

96 

820.9 

170 

801.8 

231 

(model) 

101 

815.4 

175 

797.3 

Fig.  3  reproduces  the  data  of  Table  III,  with  scales  of  voltage 
and  current.  The  lines  are  "calculated"  values,  based  on  the 
assumption  that  a  change  of  one  millimeter  in  L  causes  a  change 
of  3  divisions,  or  1.02  volts,  in  the  potential  difference  over  the 
arc.  In  most  cases  where  the  voltage  rises  or  falls  more  than  on 
this  assumption  should  correspond  to  the  movement  of  the  cathode, 
the  instantaneous  change  is  followed  by  a  slower  movement 
towards  the  calculated  value ;  the  obvious  explanation  is  that  the 
points  of  origin  of  the  arc,  or  one  of  them,  have  shifted  along 
the  electrodes.  At  /  =  90  there  is  direct  evidence  of  such  a  shift- 
ing of  the  arc ;  the  cathode  was  lowered  5.7  mm.  and  the  voltage 
dropped  4.8  volts  at  once,  but  within  half  a  minute  fell  another  1.7 
volts,  most  of  which  was  recovered  in  the  next  couple  of  minutes ; 
and  at  f  ^  82  there  was  a  sudden  drop  of  3  volts  without  any 
motion  of  the  cathode  at  all,  this  again  was  quickly  recovered. 
The  figure  also  gives  examples  of  the  stationary  or  falling  voltage 
which  accompanies  humming,  for  instance  at  t  =  140,  180,  and 
204;  between  ;  =  157  and  /  =  175  the  humming  was  so  loud  that 
no  voltage  readings  could  be  secured. 


THE  LENGTH  OF  CARBON  ARCS.  1 85 

Taking  the  results  as  a  whole,  there  can  be  no  doubt  that  the 
assumptions  made  are  justified  at  least  as  a  good  first  approxima- 
tion ;  and  that  for  currents  between  300  and  400  amperes,  where 
e  =  137 A  —  0.292  i,  de/dL  is  very  close  to  one  volt  per  milli- 
meter. Mrs.  Ayrton's  formula  (Eq.  9)  by  differentiating  and 
introducing  the  relation  between  current  and  voltage  given  in 
Eq.  3,  leads  to  de/dl  =:  2.1  volts  per  mm.,  and  although  L  is 
different  from  /,  the  changes  in  these  two  quantities  consequent  on 
raising  or  lowering  the  cathode  are  the  same.  Steinmetz's  formula 
(Eq.  10)  leads  to  dc/dl  =  0.33  volt  per  mm.  for  300  amperes, 
and  0.44  for  400  amperes.  Thus  these  two  formulas,  while  in 
accordance  with  the  experimental  results  up  to  30  amperes  or  so, 
can  evidently  not  be  relied  upon  for  currents  of  300  amperes  or 
more. 

SUMMARY. 

Conditions  have  been  found  under  which  a  steady  arc  can  be 
maintained  between  carbon  electrodes  with  currents  of  300  to 
400  amperes,  and  a  fairly  steady  arc  with  currents  up  to  800 
amperes.  Humming,  swinging  and  groaning  arcs  have  been 
described,  together  with  a  way  to  avoid  them. 

A  20-volt  arc  can  easily  be  maintained,  and  has  been  intro- 
duced as  part  of  the  routine  of  building  up  the  arc. 

A  straight-line  construction  may  be  used  to  represent  the  rela- 
tion between  current  and  potential  difference  over  the  arc  when 
the  rheostat  consists  of  water-cooled  iron  wire. 

Direct  determinations  of  the  distance  between  the  electrodes  for 
various  currents  and  voltages  have  been  made  by  the  use  of 
cathetometer  and  clay  models. 

Changes  in  the  voltage  caused  by  raising  or  lowering  the 
cathode  for  measured  distances  have  been  determined. 

For  currents  between  300  and  400  amperes  and  potential 
difference  over  the  arc  55  to  20  volts,  the  p.  d.  in  volts  is  approxi- 
mately equal  to  the  distance  between  the  electrodes  in  millimeters  ; 
for  currents  of  700  amperes  or  so  the  voltage  is  less  than  the 
distance. 

The  formulas  proposed  by  I\Irs.  Ayrton  and  by  Steinmetz  for 


1 86  DISCUSSION. 

low  currents  are  not  in  agreement  with  the  experimental  results 
for  high  currents. 

These  experiments  were  carried  out  in  the  Electrochemical 
Laboratory  of  the  University  of  Toronto  dtaring  the  winter  of 
1921-22 ;  my  thanks  are  due  to  Professor  W.  Lash  Miller  for  the 
interest  he  has  taken  in  the  work. 

University  of  Toronto, 
August,  1922. 


DISCUSSION. 

F.  G.  Dawson^  {Communicated)  :  It  would  appear  that  some 
variables  in  the  environment  of  an  arc  not  recorded  in  this  paper 
might  affect  its  characteristics.  The  temperature  of  the  enclosure 
and  the  constancy  of  conditions  of  the  gaseous  atmosphere  in 
the  enclosure  are  certainly  not  without  effect  on  an  alternating 
current  arc.  In  operating  the  experimental  indirect  arc  steel  fur- 
nace of  the  Bureau  of  Mines-  the  writer  was  impressed  by  the 
marked  efifect  of  these  two  factors  on  the  stability  and  length  of 
the  a.  c.  arc.  There  was  a  critical  temperature  in  the  preheating 
of  the  empty  furnace,  below  which  the  arc  would  not  hold  steadily 
without  constant  electrode  adjustment,  but  above  which  its  sta- 
bility was  high. 

If  the  furnace  was  luted  up  so  as  to  avoid  any  draft  at  all 
within  the  furnace,  and  consequently  to  build  up  a  slight  pressure 
within  the  furnace,  and  to  avoid  any  sudden  influx  of  air,  the 
temperature  at  which  the  arc  became  fully  steady  was  lowered. 
An  opening  the  size  of  a  pin  head  would  allow  change  of  pressure 
and  change  of  the  composition  of  the  atmosphere,  with  tiny  fluc- 
tuations of  the  arc  due  to  disintegration  of  the  graphite  electrode, 
which  would  notably  increase  the  voltage  necessary  to  hold  the 
arc.  It  is  not  certain  that  the  results  obtained  by  Mr.  Westman 
would  have  been  the  same  in  the  absence  of  a  positive  flow  of  air. 

At  operating  temperatures,  a  tiny  opening  had  no  appreciable 

»  Detroit  Electric  Furnace  Co.,   Detroit,  Mich. 

t  H  W.  Gillett,  and  E.  L.  Mack,  Experimental  production  of  certain  alloy  steels. 
Bur.   Mines  Bull.    199,   1922,  p.   14. 


THE  LENGTH   OE  CARBON  ARCS.  I87 

effect  on  the  Bureau  of  Mines'  steel  furnace,  and  when  the  fur- 
nace was  so  hot  that  ionizing  vapors  were  present  by  volatilization 
from  the  refractories,  or  at  a  much  lower  temperature  if  sodium 
silicate  had  been  used  in  repairing  the  lining,  the  voltage  required 
to  hold  the  arc  would  fall  off  to  certainly  well  under  20  volts. 

There  is  no  statement  in  the  paper  as  to  whether  or  not  the 
lining  glazed  or  showed  any  signs  of  decomposition,  but  the  empty 
furnace  used  must  have  been  very  hot  after  running  6  hr.  From 
Fig.  3,  one  would  calculate  that  some  80  Kw.-hr,  had  been  put 
into  the  furnace. 

The  current  density  in  the  electrodes,  and  their  composition, 
i.  e.,  whether  carbon  or  graphite,  must  affect  the  temperature  of 
the  electrode  tips.  With  a  positive  flow  of  air,  the  burning  away 
of  the  fip  must  have  altered  conditions  throughout  the  run. 

If  Steinmetz's  experiments  were  done  in  a  fairly  tight  enclo- 
sure, the  differences  between  his  formula  and  Mr.  Westman's 
results  may  be  at  least  partly  accounted  for. 

Electric  furnace  men  would  welcome  a  similar  study  of  a.  c. 
arcs  in  which  not  only  the  variables  studied  by  Mr.  Westman  on 
d.  c.  arcs,  but  also  the  others  mentioned  above,  are  included. 

A.  E.  R.  Westman  {Communicated)  :  The  work  presented 
was  undertaken  as  a  necessary  preliminary  to  the  study  of  arcs 
under  such  practical  conditions  as  high  current  densities,  graphite 
electrodes,  alternating  current,  etc.  An  experimental  study  of 
these  factors  is  now  under  way  in  this  laboratory. 

In  Dr.  Steinmetz's  experiments^  he  presented  his  equation  as 
an  approximation,  and  added  that  more  recent  and  extended 
investigation  seems  to  show  that  it  is  not  rigidly  correct.  My 
conditions  were  in  accordance  with  Steinmetz's  definition  of  a 
normal  arc,  in  which  no  mention  is  made  of  the  degree  of  ioniza- 
tion of  the  surrounding  gases.  There  seems  to  be  no  good  reason 
to  believe  that  such  ionization  would  affect  a  heavy  current  arc, 
or  that  it  would  be  materially  increased  by  partially  surrounding 
the  arc  with  screens  as  in  my  experiments. 

Mr.  Dawson  reports  that  with  a  furnace  thoroughly  sealed,  he 
found  no  trouble  from  air  draughts  in  his  apparatus.  The  top 
of  the  magnetic  shield  was  quite  open,  and  when  I  sealed  it  at 

'  Chem.   and   Met.   Eng.    22,   248,    (1920). 


1 88  DISCUSSION. 

the  bottom,  down  draughts  and  irregular  currents  interfered. 
Mr.  Dawson's  data  and  his  description  of  his  work  with  an  en- 
closed arc  are  welcome ;  there  is  next  to  no  information  of  this 
kind  in  the  literature. 

There  is  no  statement  in  miy  paper  to  the  effect  that  20  volts 
is  the  minimum  voltage  for  holding  an  arc.  With  electrodes 
shaped  as  shown  on  page  179,  it  is  evidently  impossible  for  L  (as 
defined  on  page  180)  to  be  less  than  about  15  mm.  For  this  reason 
I  have  not  been  able  to  run  below  17  volts. 

J.  Kelleher*  {Communicated)  :  On  pages  175  and  176,  Mr. 
Westman  describes  a  magnetic  shield  used  to  eliminate  the  effect 
of  magnetic  disturbances  caused  by  the  electrical  circuits  of  the 
furnace.  As  no  drawing  is  given  showing  the  position  of  this 
shield  with  regard  to  the  arc,  I  shall  suppose  that  the  arc  was 
formed  midway  between  the  top  and  bottom  of  the  shield.  It 
seems  to  me  that  if  this  were  the  case,  then  with  a  mean  furnace 
input  of  about  20  kw.,  (see  Table  II)  the  temperature  of  these 
laminations  would  soon  rise  and  reach  that  point  at  which  iron 
loses  its  magnetic  properties.  This  I  believe  to  be  about  780°  C. 
If  this  happened,  the  shield  would  be  of  little  value,  except  when 
the  furnace  was  cold.  It  would  be  of  interest  in  this  relation 
to  know  if  the  shielding  effect  decreased  as  the  furnace  tempera- 
ture increased.  In  my  own  work  on  arcs  I  found  great  difficulty 
in  maintaining  a  long  steady  arc  until  the  electrodes  between 
which  the  arc  was  formed  and  the  surrounding  furnace  walls, 
etc.,  had  reached  a  high  temperature. 

If  this  is  not  the  case  this  same  shield  which  I  imagine  consists 
of  cast  iron  might  be  responsible  for  the  humming  arcs  as  de- 
scribed on  page  177.  If  some  variation  in  the  current  occurred,  and 
a  certain  amount  of  residual  magnetism  was  present  in  the  shield, 
oscillations  in  the  current  might  be  set  up  causing  an  alternating 
potential  which  would  increase  or  decrease  the  volume  of  the  arc 
core,  this  again  increasing  or  decreasing  the  volume  of  the  gas 
surrounding  the  core.  If  an  oscillograph  were  connected  in  the 
furnace  circuit  to  indicate  both  current  flowing  and  the  potential 
across  the  arc,  the  humming  arc  and  perhaps  the  groaning  arc 
might  be  explained. 

*  Cliippawa,   Ont.,   Canada. 


THE    I.ENGTH    OF    CARBON    ARCS.  1 89 

The  use  of  a  cathetometer  seems  slightly  in  excess  of  the  re- 
quirements of  accuracy,  as  I  notice  no  corrections  have  been  made 
for  thermal  expansion  in  the  determination  of  "h"  on  page  180. 

This  interesting  work  I  hope  will  be  continued,  and  instead  of 
using  two  carbon  electrodes  a  bath  of  some  metal  such  as  iron 
might  be  substituted  for  the  anode  and  a  comparison  made  of  the 
behavior  of  the  arcs  as  described  and  those  occurring  where  the 
anode  is  a  metal. 

A.  E.  R.  Westman  (Communicated)  :  The  magnetic  shield 
did  not  reach  a  temperature  higher  than  800°  C.  in  most  of  these 
runs,  as  there  was  always  a  current  of  air  between  the  arc  and 
the  shield.  I  can  not  say  whether  the  shielding  effect  decreased 
•during  a  run,  as  other  circumstances  such  as  the  deepening  of  the 
crater  tended  to  make  the  arc  unsteady  near  the  end  of  a  run. 

The  cathetometer  was  used  more  especially  for  measuring  the 
movements  of  the  cathode,  which  were  sometimes  as  small  as 
4  mm. ;  these  movements  would  cause  no  appreciable  change  in 
the  temperature  of  the  electrodes,  and  so  no  error  from  thermal 
expansion  would  be  introduced.  However,  these  results  are  only 
a  first  approximation,  the  results  of  more  accurate  measurements 
will  be  reported  later. 


A  paper  presented  at  the  Forty-third 
General  Meeting  of  the  American  Elec- 
trochemical Society  held  in  New  York 
City,  May  4,  1923,  President  Schlueder- 
berg  in  the  Chair. 


ELECTRIC  FURNACE  DETINNING  AND  PRODUCTION  OF 
SYNTHETIC  GRAY  IRON  FROM  TIN-PLATE  SCRAP.' 

By  C.  E.  Williams, 2  C.  E.  Sims,^  and  C.  A.  Newhall> 

Abstract, 
Experiments  were  conducted  in  a  small  electric  furnace  in 
which  tin-plate  scrap  was  melted  with  various  addition  agents 
in  attempts  to  remove  the  tin  from  the  iron.  Sodium  chloride, 
iron  sulfide,  and  an  oxidizing  slag  were  used  under  various  con- 
ditions. The  conclusions  reached  were  that  in  the  electric  furnace 
complete  detinning  is  impossible,  and  any  detinning  impractical. 
Melting  tests  conducted  in  the  cupola  showed  that  the  amount 
of  detinning  was  dependent  upon  the  amount  of  surface  of 
metallic  tin  exposed  to  the  oxidizing  gases,  and  will  be  somewhere 
between  the  limits  of  0  and  50  per  cent.  Test  bars,  prepared  by 
melting  pig  iron  with  various  quantities  of  tin,  were  subjected  to 
physical  tests.  The  results  showed  that  a  tin  content  of  one  per 
cent  or  less  did  not  seriously  afifect  the  properties  of  gray  iron. 
Synthetic  cast  iron  made  from  tin-plate  scrap  was  used  success- 
fully in  making  commercial  castings  of  good  quality. 


INTRODUCTION. 

The  investigation  here  described  was  conducted  at  the  North- 
west Experiment  Station  of  the  U.  S.  Bureau  of  Mines,  in  co- 
operation with  the  College  of  Mines  of  the  University  of  Wash- 
ington, The  object  of  the  study  was  to  determine  the  possibilities 
of  converting  tin-plate  scrap  or  used  tin  cans  into  a  marketable 

*  Published  by  Permission  of  the  Director,  U.  S.  Bureau  of  Mines.  Manuscript 
received  February  1,   1923. 

2''  Metallurgist  and  Electrometallurgist,  respectively,  North-west  Experiment  Station, 
U.  S.  Bureau  of  Mines,  at  Seattle,  Wash,  in  cooperation  with  College  of  Mines, 
University  of  Washington, 

*  Washington  Electrochemical  Co.,  Seattle,  Wash, 

191 


192  C.  E.  WILLIAMS,  C.  E.  SIMS  AND  C.  A.   NEWHALL. 

Steel  or  iron  product  by  electrothermal  means.  ^Nlost  of  the 
tin-plate  scrap  produced  in  this  country  is  detinned  and  subse- 
quently melted  in  the  open-hearth  furnace,  producing  steel ;  a 
small  quantity  of  it  is  melted  with  other  iron  scrap  in  the  cupola 
for  the  manufacture  of  sash  weights  and  similar  low-priced 
castings ;  and  some  is  used  in  certain  hydrometallurgical  plants 
to  precipitate  metals  from  solutions.  A  small  quantity  of  used 
tin  cans  is  treated  similarly  to  tin-plate  scrap,  as  above  described. 
There  are  three  established  methods  of  detinning,  namely :  the 
chlorine,  electrolytic  alkali  and  the  alkali-saltpeter  processes,  pro- 
ducing respectively  tin  tetrachloride,  metallic  tin  and  tin  oxide. 

A  plant  utilizing  one  of  the  established  detinning  processes,  to 
be  profitable,  must  be  operated  on  a  comparatively  large  scale, 
the  minimum  capacity  having  been  variously  estimated  as  between 
50  to  100  tons  of  tin-plate  scrap  per  day.  In  some  localities 
the  quantity  of  tin-plate  scrap  or  old  cans  available  may  be  so 
small,  or  the  market  for  the  recovered  tin  so  limited,  that  another 
process  of  utilizing  these  waste  materials  would  be  required.  In 
districts  near  can  factories  there  may  be  an  oversupply  of  tin- 
plate  clippings  and  punchings,  and  in  cities  where  efficient  meth- 
ods of  collecting  old  cans  are  in  vogue,  such  materials  may  be 
available  at  much  lower  prices  than  the  cost  of  steel  scrap.  In 
such  cases  a  method  of  using  this  cheap  form  of  iron  in  a  more 
profitable  manner  than  for  conversion  into  sash  weights  would 
be  desirable. 

The  weight  of  the  tin  coating  on  tin  plate  varies  between  wide 
limits.  Results  of  the  analysis  by  the  National  Canners'  Asso- 
ciation^ of  many  thousands  of  cans  showed  weights  of  from  0.53 
to  6.37  lb.  of  tin  per  base  box  containing  112  sheets  and  weighing 
100  lb.,  the  grand  average  of  all  analyses  being  from  0.81  to 
2.94  lb.  per  base  box.  During  the  past  few  years,®  the  tin  re- 
covered by  detinning  clean  tin-plate  scrap  amounted  to  1.6  lb. 
for  each  100  lb.  treated.  Hence,  assuming  a  recovery  of  95  per 
cent,  the  average  tin  content  of  tin-plate  scrap  would  be  1.7 
per  cent,  which  probably  represents  the  average  content  fairly 
closely.     The  tin  content  of  used  cans  will  be  usually  found  a 

*  Relative  value  of  dtflFerent  weights  of  tin  coating  on  canned  food  containers. 
National   Canners'  Assn.,  Washington,  D.  C,   1917. 

"Secondary  Metals  in  1919,  1920,  and  1921.  J.  D.  Uunlap.  Mineral  Resources  of 
the  United   States.     U.   S.  Geological   Survey. 


DETIXNING  OF  TIN-PLATE  SCRAP.  I93 

few  tenths  less  than  this,  due  to  losses  by  mechanical  abrasion 
and  by  solution  in  the  foodstuff  contained  in  the  can,  although 
if  solder  were  used  in  sealing  the  can  the  tin  content  might  be 
above   1.7  per  cent. 

Not  much  information  is  available  regarding  the  effect  of  tin 
upon  the  properties  of  steel  or  cast  iron.  In  detinning,  the 
attempt  is  made  to  produce  a  product  containing  less  than  0.1  per 
cent  tin,  although  during  the  war  this  limit  was  not  insisted  upon 
by  purchasers  of  detinned  scrap.  In  the  present  investigation  no 
time  was  spent  on  chemical  or  electrolytic  detinning,  but  attempts 
were  made  to  remove  the  tin  by  some  action  during  the  process 
of  melting  the  scrap.  The  impracticability  of  removing  a  large 
proportion  of  the  tin  in  this  manner  was  soon  determined,  and 
a  study  was  then  made  to  determine  the  possibilities  of  using 
tin-plate  scrap  in  producing  gray  iron  without  removing  the  tin. 
With  this  in  view  a  study  was  made  of  the  effect  of  various 
quantities  of  tin  on  the  properties  of  gray  iron. 

EXPERIMENTS  ON  DETINNING. 

The  physical  and  chemical  properties  of  tin  and  its  compounds 
are  such  as  to  offer  little  encouragement  to  the  possibility  of 
detinning  iron  in  the  electric  furnace.  The  popular  belief,  that 
tin  is  volatilized  when  iron  containing  it  is  melted,  is  not  founded 
on  fact,  because  the  boiling  point  of  tin  is  2270''  C.  Tin  is  found 
in  gases  from  cupolas  in  which  tin-plate  scrap  is  being  melted, 
but  its  presence  is  probably  due  to  the  burning  of  the  tin  to 
oxide  which  is  then  carried  mechanically  through  the  stack  by 
the  escaping  gases.  Although  the  melting  point  of  tin  is  only 
232°  C.  and  that  of  iron  1500°  C,  the  tin  coating  on  most  tin 
plate  is  so  thin  that  the  tin,  although  above  its  melting  point,  will 
not  flow  off  and  thus  permit  separation. 

The  volatility  of  the  chlorides  of  tin  suggests  the  use  of  sodium 
chloride.  The  reaction  would  require  oxidizing  conditions  and 
would  undoubtedly  produce  stannous  chloride  whose  boiling  point 
is  603°  C.  The  most  obvious  time  to  conduct  this  reaction  would 
be  before  fusion  of  the  iron,  in  order  to  permit  the  maximum  con- 
tact of  salt  and  air  with  the  tin  coating.  The  facts  that  stannous 
sulfide  boils  at  1230°   C.  and  that  it  can  be  made  by  the  action 


194  C.  E.  WILUAMS,  C.  E.  SIMS  AND  C.  A.  NEWHALL. 

of  iron  sulfide  on  metallic  tin  offer  the  possibility  of  detinning 
with  pyrite.  The  reaction  would  have  to  be  complete  enough  to 
permit  the  use  of  only  a  slight  excess  of  pyrite  and  thus  avoid 
the  introduction  into  the  iron  of  too  much  sulfur. 

It  has  been  suggested  that  detinning  could  be  accomplished  by 
melting  under  an  oxidizing  slag,  thereby  oxidizing  the  tin  and 
slagging  it  off.  Complete  removal  of  the  tin,  however,  could 
not  be  expected  by  this  means,  because  tin  is  lower  in  the  electro- 
motive series  than  iron  and  would  be  kept  in  a  reduced  condition 
by  the  metallic  iron.  Tin  is  soluble  in  iron  up  to  19  per  cent,^ 
and  hence,  molten  tin-plate  scrap  would  contain  tin  in  a  dilute 
solution  (about  1.7  per  cent),  which  would  contribute  to  the 
difficulties  of  removing  it  by  a  chemical  reaction. 

A  preliminary  study  of  the  reaction  with  salt  at  a  temperature 
below  the  melting  point  of  iron  was  first  made.  Strips  of  tin 
plate  placed  in  fire  clay  roasting  dishes  were  heated  in  a  muffle 
and  treated  with  fumes  of  sodium  chloride.  Tin  was  volatilized 
at  temperatures  above  500°  C.  when  the  atmosphere  was  kept 
strongly  oxidizing,  but  the  resultant  iron  sheet  was  badly  oxidized 
and  unfit  for  conversion  into  steel  or  iron. 

The  subsequent  tests  were  carried  out  in  a  basic-lined  single- 
phase  series-arc  stationary  furnace.  The  hearth  was  23  x  38  cm. 
(9  x  15  in.)  in  cross  section  and  conveniently  held  the  50-lb. 
charges  used.  Test  No.  4  was  made  in  a  carbon-lined,  direct- 
heating,  single-arc,  stationary  furnace.  The  tin-pjate  scrap, 
which  consisted  of  clippings  and  rejected  can  ends,  varied  greatly 
in  tin  content  and  much  difficulty  was  had  in  obtaining  true  sam- 
ples of  the  charges  to  the  furnace.  A  fairly  uniform  feed  was 
obtained  by  using  only  the  can  ends  of  uniform  gauge. 

Numerous  analyses  showed  that  the  average  tin  content  was 
1.25  per  cent,  although  the  tin  content  of  some  charges  probably 
varied  as  much  as  10  or  15  per  cent  from  this  average  value. 
Hence,  great  accuracy  is  not  claimed  for  the  results  given  below, 
which  show  the  extraction  of  tin  obtained  by  the  dift'erent  treat- 
ments. However,  these  results  do  show  approximately  the  mag- 
nitude of  the  detinning  obtained,  and  the  relative  effectiveness  of 
the  various  methods  tried.   In  order  to  make  the  results  compara- 

'Tammann,  Z.   f.  anorgan.  Chem.,  53,  281-295   (1907). 


DETINNING  OF  TIN-PLATE  SCRAP. 


195 


ble,  the  conditions  were  kept  as  nearly  uniform  as  possible  in 
all  tests.  The  furnace  was  preheated  before  charging  and  the 
molten  charge  held  in  the  furnace  for  at  least  a  half  hour  in 
order  to  superheat  the  metal  and  permit  any  reactions  to  go  to 
completion. 

Tin-plate  scrap  was  first  melted  with  carbon  in  the  electric 
furnace.  The  results,  one  of  which  is  entered  in  Table  I,  show 
that  no  tin  was  removed  by  the  treatment.  A  series  of  experi- 
ments using  sodium  chloride  with  various  other  reagents  was 
conducted.  A  large  excess  of  salt,  amounting  to  10  per  cent  of 
the  weight  of  the  scrap  was  charged  with  the  scrap  into  the 
furnace.  In  some  cases  reducing  and  in  others  oxidizing  condi- 
tions were  maintained  during  the  test.  Table  I  shows  the  essen- 
tial data  of  these  experiments. 


Table  I. 
Tests  on  Chloride  Volatilization. 


Charge 

Tin 
in 
pig 
per 
cent 

Per 

cent 

tin 

removed 

Run 
No. 

Tin 
plate 
scrap 

lb. 

Salt 
lb. 

Iron 
ore 
lb. 

Sili- 
ca 
lb. 

Carbon 
lb. 

Remarks 

1 

2 

3 

4 

5 

6* 

7 

8 

50 
50 
50 
50 
50 
50 
50 
50 

5 
5 

5 

5 
5 
5 

is 

15 

15 

5 

3 
3 

4 
4 

4 

Carbon 
lining 

1.25 
1.20 
1.22 
0.96 
1.13 
0.74 
1.02 
1.02 

0 
4 
2 
23 
10 
41 
18 
18 

Reducing 
Reducing 
Reducing 

Slightly  oxidizing 

Oxidizing 

Oxidizing  slag 

Oxidizing  slag 

Oxidizing,  then 

reducing 

•  This  charge  forced  its  way  out  of  tap  hole  before  the  run  was  complete. 

Practically  no  tin  was  removed  by  melting  with  salt  and  carbon, 
the  reducing  atmosphere  caused  by  the  carbon  undoubtedly  pre- 
venting the  formation  of  tin  chloride.  About  23  per  cent  of 
the  tin  was  volatilized  by  melting  the  mixture  of  tin-plate  scrap 
and  salt  without  carbon,  and  about  10  per  cent  elimination  of 
the  tin  was  effected  using  an  oxidizing  slag.  In  two  tests,  using 
an  oxidizing  slag  with  sodium  chloride,  18  and  41  per  cent  of 
the  tin  was  removed,  but  the  larger  result  can  not  be  stressed 


196 


C.  E.   WILLIAMS,  C.  E.  SIMS  AND  C.  A.  NEWHALL. 


too  much  because  the  furnace  was  tapped  before  the  charge  was 
completely  melted.  In  no  case  was  the  elimination  of  tin  com- 
plete or  the  results  encouraging  enough  to  give  promise  of  success 
on  a  larger  scale. 

The  results  of  the  experiments  in  which  the  attempt  was 
made  to  volatilize  the  tin  as  sulfide  are  summarized  in  Table  II. 
^Mixtures  of  tin-plate  clippings  and  pyrites  in  various  ratios  were 
melted  with  carbon.  In  one  case  gypsum  was  substituted  for 
pyrite.  Runs  9  and  10  show  that  both  the  elimination  of  tin 
and  the  amount  of  sulfur  introduced  into  the  iron  are  propor- 
tional to  the  amount  of  pyrite  used.     The  removal  of  the  tin  was 


Table  II. 
Tests  on  Sid  fide   Volatilisation. 


Charge 

Analysis 

Per 
cent 
tin 

Run 

No. 

Tin 

plate 

Pyrite 

Gypsum 

Lime 

Silicon 

Fe-Si 

Carbon 

S 

Sn 

scrap 

lb. 

lb. 

lb. 

lb. 

lb. 

lb. 

per 

per 

lb. 

cent 

cent 

9 

25 

0.25 

1 

3 

0.21 

0.80 

36 

10 

25 

0.50 

, 

1 

3 

0.43 

0.67 

46 

11 

50 

1.50 

3 

1.5 

2 

6 

0.09 

1.20 

4 

12 

50 

5.5 

2 

6 

0.06 

1.07 

14 

13* 

50 

2.66 

2 

2.5 

0.36 

0.95 

24 

*  The  metal  in  this  test  was  treated  with   a  desulfurizing  slag  before  tapping. 

not  complete  in  any  test  and  became  less  rapid  as  the  concentra- 
tion of  the  tin  in  the  iron  became  less.  The  relatively  small  elim- 
ination obtained  in  Runs  11  and  12  was  due  to  the  basic  slag 
which  kept  the  sulfur  from  dissolving  in  the  iron.  As  a  result 
of  these  tests,  it  seems  that  although  tin  dissolved  in  molten 
iron  may  be  converted  to  sulfide  and  volatilized,  complete  elimina- 
tion is  i)robably  impossible  and  the  removal  of  even  small  amounts 
of  tin  by  this  means  introduces  a  large  amount  of  sulfur  into  the 
iron.  These  preliminary  experiments  were  not  sufficiently  en- 
couraging to  warrant  further  work  along  this  line. 

In  order  to  make  the  data  more  complete  tests  were  made  in 
which  iron,  coated  with  both  tin  and  lead  (terne  plate),  and 
galvanized  scrap  were  melted  with  carbon.     In  the  melt  using 


DETIXXING  OF  TIN-PLATE  SCRAP.  I97 

scrap  containing  1.9  per  cent  tin  and  2.5  per  cent  lead,  practically 
no  elimination  of  the  tin  and  complete  elimination  of  the  lead 
were  obtained.  Some  of  the  lead  was  vaporized  and  the  rest  of 
it,  on  tapping,  ran  out  of  the  furnace  ahead  of  the  molten  iron 
in  which  it  was  insoluble.  In  the  test  using  galvanized  scrap  con- 
taining 8.22  per  cent  zinc,  the  resultant  metal  contained  only  0.20 
per  cent  zinc.  Thus,  unlike  tin,  lead  and  zinc  are  both  readily 
removed  from  iron  by  melting  in  the  electric  furnace. 

Cupola  Tests. 

In  order  to  determine  the  degree  of  detinning  possible  in  the 
cupola,  the  following  experiments  were  conducted. 

Charges  consisting  of  25  lb.  of  tin-plate  scrap,  75  lb.  of  gray 
iron  and  a  large  excess  of  coke  were  melted  in  a  small  cupola 
45.5  cm.  (18  in.)  in  diameter.  In  one  case,  when  a  strong  blast 
was  used,  the  resultant  metal  contained  only  half  of  the  tin 
charged.  In  another  instance,  in  which  a  light  blast  was  used, 
the  temperature  of  the  metal  was  consequently  low  and  a  viscous 
melt  was  obtained  with  practically  no  elimination  of  tin.  A 
sample  obtained  from  the  castings  made  at  a  local  sash-weight 
foundry  by  melting  four  parts  gray  iron  and  one  part  baled  tin- 
plate  scrap  in  a  large  cupola  was  analyzed  and  found  to  contain 
0.37  per  cent  tin.  Assuming  that  the  tin-plate  scrap  contained 
1.7  per  cent  tin  (the  average  for  all  scrap  detinned  in  1921),  a 
recovery  of  practically  100  per  cent  of  the  tin  was  obtained. 

One  thousand  pounds  of  synthetic  cast  iron  was  made  in  the 
electric  furnace  from  tin-plate  scrap  and  used  as  the  iron  in  a 
regular  cupola  melt  at  a  local  foundry.  The  iron  before  melting 
in  the  cupola  contained  1.25  per  cent  tin,  and  after  melting  1.23 
per  cent  tin,  thus  showing  practically  no  elimination. 

It  is  apparent  from  the  above  study  that  some  elimination 
of  tin,  probably  up  to  50  per  cent,  may  be  effected  by  melting  thin 
sheets  in  a  strongly  oxidizing  atmosphere,  that  there  is  practically 
no  loss  of  tin  when  melting  large  pieces  of  iron  containing  tin 
in  solid  solution,  and  that  the  tin  removed  in  cupola  melting  is 
dependent  upon  the  conditions  of  melting  and  the  state  in  which 
the  tin  is  present. 


198  C.  E.  WILLIAMS,  C.  E.  SIMS  AND  C.  A.  NEWHALL. 

SYNTHETIC   CAST   IRON   FROM    TIN-PLATE  SCRAP. 

Effects  of  Tin  in  Cast  Iron. 

Believing  it  to  be  impractical  to  effect  detinning  in  the  electric 
furnace  or  the  cupola,  experiments  were  then  conducted  to  deter- 
mine what  effect  tin  had  on  cast  iron,  and  whether  suitable  cast- 
ings could  be  made  from  synthetic  cast  iron  made  from  tin-plate 
scrap.  Test  bars  containing  quantities  of  tin  varying  from  0.05 
per  cent  to  5.0  per  cent  were  cast  from  pig  iron  melted  with  the 
required  proportions  of  tin  in  an  electric  furnace.  The  tests  on 
these  bars  showed  that  tin  increases  the  hardness  and  decreases 
the  transverse,  compressive  and  tensile  strengths,  as  well  as  the 
resistance  to  impact.  Chemical  analyses  showed  a  decrease  in 
graphitic  carbon  as  the  tin  content  increased,  and  microscopic 
examination  gave  evidence  that  less  than  1  per  cent  of  tin  has  no 
effect  upon  the  size  and  shape  of  the  graphite.  These  effects  of 
tin  are  in  direct  proportion  to  the  amount  present,  and  roughly, 
1  per  cent  of  tin  will  reduce  the  strength  of  gray  iron  15  per 
cent.  The  effect  on  hardness  and  graphitic  carbon  will  not  be 
over  10  per  cent.  When  the  tin  content  is  2  per  cent  or  more, 
the  molten  iron  appears  dirty,  does  not  fill  the  mold  well,  and  the 
castings  are  rough  and  porous. 

It  seems,  therefore,  that  tin-plate  scrap  or  old  tin  cans  can  be 
used  in  the  production  of  synthetic  gray  iron  for  the  ordinary 
grade  of  castings  provided  the  tin  content  of  the  product  can  be 
kept  to  1  per  cent  or  less.  If  the  scrap  contains  more  than  1 
per  cent  of  tin  it  should  be  mixed  with  enough  tin-free  scrap  to 
bring  the  average  tin  content  to  about  this  figure. 

Commercial  Tests. 
In  order  to  obtain  more  data  on  the  value  of  tin-plate  scrap 
as  a  raw  material  in  the  manufacture  of  synthetic  cast  iron,  one 
thousand  pounds  of  it  were  melted  and  carburized  in  a  basic- 
lined,  single-phase,  roofed  Heroult  furnace.  The  composition  of 
the  product  is  shown  in  Table  III.  It  was  taken  to  a  local  foundry 
and  used  in  one  of  the  regular  cupola  melts.  Both  heavy  sections 
and  thin  ornamental  castings  were  made  from  it.  All  parts  of 
the  molds  were  well  filled,  and  the  castings  without  exception 
were  smooth  and  sound.     A  machining  test  was  made  by  the 


DETINNING   OF    TIN-PLATK    SCRAP. 


199 


manager,  who  was  well  satisfied  with  the  iron  and  gave  the  fol- 
lowing report: 

"Depth  of  chill,  nil.  Very  soft.  Cuts  readily  with  hack  saw. 
Drills  easily  with  ordinary  carbon  drill.  Turns  readily  in  lathe 
at  considerably  over  ordinary  speed.  On  facing  cut  run  at  36.6 
m.  per  min.  (120  ft.  per  min.),  1.6  mm.  (1/16  in.)  depth  of  cut, 
0.79  mm.  (1/32  in.)  feed  with  Rex  AA  high  speed  steel.  No 
difficulty  to  make  deep  cut  with  parting  tool.  Fracture  very  fine, 
dense,  close  grain,  rather  dark  in  color.    Elasticity  good." 

Five  hundred  pounds  more  of  synthetic  iron  of  the  same  com- 
position was  made  and  submitted  to  another  foundry  for  a  similar 
test ;  another  favorable  report  was  returned. 


Table  III. 

Composition  of  Synthetic  Iron  Made  from  Tin-Plate  Scrap  before 
and  after  Melting  in  a  Cupola. 


Before   melting 
per   cent 

After    melting 
per   cent 

c 

3.85 
1.34 
0.83 
e.45 
trace 
1.25 

3.78 
1.13 
0.60 
0.56 
trace 
123 

Si    

Mn     

P  

S   

Sn  

CONCLUSIONS. 

1.  It  is  impossible  to  remove  most  of  the  tin  in  tin-plate  scrap 
or  similar  material  by  any  of  the  electric  furnace  melting  pro- 
cesses tried;  moreover,  it  is  impractical  to  attempt  any  detinning 
by  these  means. 

2.  No  tin  is  volatilized,  ordinarily,  when  iron  scrap  contain-ing 
it  is  melted  in  the  electric  furnace. 

3.  The  amount  of  tin  volatilized  during  melting  in  the  cupola 
may  be  as  much  as  50  per  cent  in  some  cases,  whereas  in  others 
it  may  be  practically  nil,  depending  upon  the  amount  of  surface 
of  metallic  tin  exposed,  and  the  oxidizing  condition  of  the  blast. 

4.  Lead  can  be  removed  completely  from  iron  coated  with  lead, 


200  DISCUSSION. 

and  likewise,  zinc  can  be  largely  removed  from  galvanized  scrap 
by  melting  in  the  electric  furnace. 

5.  A  tin  content  of  1  per  cent  or  less  does  not  seriously  affect 
the  physical  properties  of  cast  iron. 

6.  Under  conditions  prevailing  in  many  parts  of  the  country, 
tin-plate  scrap  and  used  tin  cans  can  not  be  profitably  treated 
bv  any  of  the  established  detinning  processes.  This  potential 
waste  material  can  probably  be  recovered  most  usefully  and 
efficiently  by  treating  it  in  the  electric  furnace  to  produce  syn- 
thetic cast  iron,  using  low-grade,  tin-free  scrap  for  dilution  to 
reduce  the  tin  content  of  the  product  to  within  safe  limits. 

ACKNOWLEDGMENTS. 

The  authors  are  grateful  for  the  helpful  co-operation  of  the 
College  of  Mines,  University  of  Washington,  and  also  to  Mr. 
Lyall  Zickrick,  graduate  student  in  metallurg}'-  at  the  University 
of  Washington,  for  assistance  with  the  physical  examination  of 
the  cast  iron  test  bars ;  to  Messrs.  R.  J.  Anderson  and  G.  M. 
Enos,  of  the  Pittsburgh  Station  of  the  Bureau  of  Mines,  for  a 
microscopic  study  of  the  cast  iron  specimens  containing  tin,  and 
to  IMessrs.  E.  P.  Barrett  and  J.  D.  Sullivan,  of  the  Northwest 
Experiment  Station  of  the  Bureau  of  Mines,  for  the  large  amount 
of  analytical  work  done  in  connection  with  this  investigation. 


DISCUSSION. 


E.  L.  Crosby^  :  There  are  several  processes  which  look  feasible 
from  the  electric  furnace  operating  standpoint,  affording  possi- 
bilities of  using  sheet  scrap,  which  do  not  work  out  so  well  in 
practice.  The  instant  a  plant  starts  in  a  certain  community,  where 
a  cheap  supply  of  scrap  is  available,  the  law  of  supply  and  demand 
operates.  The  cost  of  the  material  goes  up,  and  it  does  not  allow 
for  any  commercial  margin. 

C.  G.  ScHLUEDERBERG- :  I  would  like  to  have  further  light  on 
just  what  the  market  is  for  some  of  this  low-grade  cast  iron.    Is 

>  \'ice-Pres.   and  Gen.   Mgr..   Detroit   Elec.    Furnace   Co.,   Detroit,    Mich. 
-  Westinghouse    Elec.    and    Mfg.    Co.,    East    Pittsburgh,    Pa. 


DETINNING  OF  TIN-PLATE  SCRAP.  20l 

there  enough  of  a  market  to  justify  large  operations  in  recovery 
of  this  material? 

H.  W.  GiLLETT" :  On  the  economic  end,  may  I  ask  whether  this 
refers  purely  to  selected  scrap,  clean  scrap,  or  whether  it  is  possible 
to  pick  up  old  tin  cans  and  shove  them  into  the  furnace  and  use 
them ;  in  other  words,  whether  somewhat  oxidized  scrap  would  be 
feasible  for  use  or  not  ?  When  this  proposition  came  up  years 
agOj  every  community  was  to  have  a  tin  can  wagon  to  collect 
them,  and  instead  of  having  them  go  to  the  garbage  man  they 
were  all  to  be  picked  up.  I  wonder  if  that  point  of  view  still 
olitains. 

C.  E.  Williams  :  In  regard  to  Mr.  Crosby's  point  regarding 
supply  and  demand,  and  that  brought  up  by  Dr.  Gillett,  we  are 
looking  to  the  future  in  this  case  just  as  much  perhaps  as  any 
question  in  which  we  are  involved. 

Much  of  the  tin-plate  scrap  is  treated  by  the  chlorine  detinning 
process,  which  produces  tin  tetrachloride  used  in  weighting  silk. 
We  are  not  sure  but  that  a  cheaper  or  better  substitute  for  tin 
tetrachloride  will  be  developed.  Such  a  development  would  liber- 
ate a  large  quantity  of  scrap.  IMoreover,  there  are  times  when 
the  spread  between  the  cost  of  tin-plate  scrap  and  of  detinned 
scrap  is  not  sufficiently  large  to  make  detinning  by  present  meth- 
ods profitable. 

Large  cities  are  developing  efficient  methods  of  garbage  collec- 
tion and  disposal,  in  which  large  picking  bands  are  operated,  old 
tinned  containers  being  segregated  and  baled  at  a  low  cost.  This 
practice  will  furnish  a  large  potential  supply  of  cheap  scrap  iron 
for  use  in  producing  foundry  iron  by  the  method  described  in  this 
paper.  A  foundry  in  Los  Angeles  is  now  operating  on  a  fairly 
large  scale  using  baled  cans  collected  in  this  manner  and  produces 
white  cast  iron  therefrom.  This  company  recently  put  into  opera- 
tion an  electric  furnace  for  producing  gray  iron,  but  I  do  not 
know  what  success  they  have  had. 

H.  W.  Gillett  :  Have  you  ever  tried  to  use  your  tin-plate  scrap 
as  part  of  the  iron  base  for  semi-steel  ?  If  your  gray  iron  is  not 
up  to  the  mark  on  account  of  the  presence  of  tin,  for  most  uses 

3  U.  S.  Bureau  of  Mines,   Ithaca,  X.  Y. 

14 


202-  DISCUSSION. 

yon  can  improve  the  quality  by  going  down  on  the  carbon  and 
making  semi-steel  out  of  it. 

C.  E.  Williams  :  We  have  not  investigated  the  properties  of 
semi-steel  containing  tin.  However,  it  would  probably  be  all 
right  unless  the  physical  properties  of  that  semi-steel  are  affected 
more  than  the  properties  of  gray  iron  are  affected  by  tin. 

W^e  know  that  tin  has  a  decided  action  upon  steel  when  present 
in  very  small  quantities,  and  that  it  does  not  have  much  eft"ect  on 
gray  iron  even  when  it  is  present  in  fairly  large  quantities.  So 
as  you  go  from  gray  iron  down  to  steel,  the  deleterious  effect  of 
tin  would  probably  increase.  Hence  it  might  be  that  semi-steel 
containing  a  few  tenths  per  cent  of  tin  would  be  affected  to  a 
greater  extent  than  is  gray  iron.  However,  this  is  something  that 
should  be  investigated. 


A  paper  presented  as  an  introduction  to 
the  session  devoted  to  the  reading  and 
discussion  of  papers  on  "Rarer  Metals," 
at  the  Forty-third  General  Meeting  of 
the  American  Electrochemical  Society, 
held  in  New  York  City,  May  5,  1923, 
Dr.  F.  M.  Becket  in   the  Chair. 


PRESENT  STATUS  OF  THE  PRODUCTION  OF  RARER  METALS.' 

By    C.   James^ 

Many  years  ago,  while  attending  a  lecture  in  London  Univer- 
sity, I  heard  of  the  work  of  Waldron  Shapleigh,  of  the  Welsbach 
Company.  The  lecturer  described  some  of  the  work  that  was 
done  in  the  days  before  the  modern  thorium  mantle  was  evolved. 
Shapleigh  had  separated  large  amounts  of  lanthanum,  praseody- 
mium, etc.,  in  very  pure  form.  The  enthusiastic  description  of 
the  beautiful  salts,  the  mystery  which  enshrouded  them,  and  the 
immense  opportunities  for  research  among  the  rare  metals  con- 
verted me  completely. 

Although  this  section  of  chemistry  has  appealed  to  many,  be- 
cause of  the  thought  that  there  must  be  something  unique  about 
these  substances,  yet  most  of  the  work  in  the  past  has  been 
devoted  to  their  chemical  characteristics.  Notwithstanding  the 
fact  that  much  time  has  been  spent  searching  for  methods  for 
detecting  and  for  the  quantitative  determination  of  these  elements, 
we  find  that  in  many  cases  good  methods  are  completely  lacking. 
The  separation  of  tantalum  and  columbium  is  such  an  one.  Even 
the  determination  of  the  mixed  oxides  requires  great  care,  since 
these  substances  tend  to  retain  both  alkalies  and  acids.  The  acid 
solutions  used  during  the  work  are  liable  to  carry  away  some  of 
the  metallic  acids.  The  errors  act  in  opposite  directions,  the  first 
tending  to  increase,  and  the  latter  to  decrease  the  per  cent.  In 
working  with  these  two  elements  along  these  lines,  some  of  us 
have  passed  through  discouragingly  gloomy  periods.  However, 
many  experiments  along  qualitative  lines  indicate  that,  after  all, 
there  appears  to  be  quite  a  difference  between  these  elements  as 
regards  their  chemical  properties. 

*  Introductory   paper    to   the   session    on    "Rarer   Metals." 

^  Professor  of  inorganic  chemistry,  New  Hampshire   College,   Durham,   N.   H. 

203 


204  C.   JAMES. 

Cupferron  is  a  reagent  that  can  be  used  for  precipitating  tan- 
talum and  columbium  together  from  acid  solution.  The  oxalic 
acid  solution,  strongly  acidified  with  sulfuric  acid,  or  the  hydro- 
fluoric acid  solution  containing  considerable  sulfuric  acid,  is  readily 
precipitated  by  cupferron.  The  precipitation  should  be  carried 
out  in  very  cold  solutions.  The  precipitate  can  be  readily  washed 
and  the  oxides,  obtained  by  igniting  this  precipitate,  appear  to  be 
very  pure.    The  results  seem  to  be  exact. 

When  solutions  of  pure  tantalum  and  pure  columbium,  under 
the  same  conditions,  are  treated  with  cupferron,  a  great  differ- 
ence is  observed  in  the  behavior  of  the  precipitate.  Tantalum 
gives  no  trouble  in  filtering  and  washing,  while  columbium  is 
thrown  down  as  a  sticky  semi-liquid  mass.  It  will  probably  take 
some  time  before  the  conditions  for  an  exact  separation  of  tan- 
talum from  columbium  can  be  achieved. 

Recently  some  interesting  observations  have  been  made  with 
regard  to  the  effect  of  solutions  of  organic  substances,  such  as 
bases,  benzidine,  quinoline,  hexamythelene  tetramine,  piperazine 
hydrate,  quinine,  etc.,  upon  solutions  of  tantalum  and  columbium 
dissolved  either  in  oxalic  acid  solution  or  in  a  solution  of  methyl- 
amine  or  some  similar  substance.  Qualitative  experiments  have 
shown  that  tantalum  is  usually  more  readily  precipitated  than 
columbium.  However,  there  are  cases  where  columbium  solutions 
have  been  precipitated  while  those  of  tantalum  have  remained 
clear.  The  oxide  of  tantalum  obtained  by  some  of  these  tests  is 
extremely  white. 

So  far  as  quantitative  analysis  is  concerned,  the  greatest  prob- 
lem is  found  in  the  case  of  the  cerium  and  yttrium  groups  of 
metals.  The  separation  of  the  two  groups  is  an  extremely  tedious 
matter,  which  is  rarely  carried  out.  The  precipitation  performed 
with  sodium  or  potassium  sulfates  is  far  from  accurate.  The 
precipitated  cerium  group  may  be  as  much  as  fifty  per  cent  too 
high,  while  only  a  small  fraction  of  the  yttrium  group  may  be 
separated  as  such. 

The  most  accurate  method  for  sejiarating  these  elements  is  to 
stir  the  sulfate  solutions  with  potassium  sulfate  until  the  solution 
no  longer  shows  any  neodymium  absorption.  The  precipitated 
double  sulfates  are  filtered  oflf  and  washed  with  a  solution  of 


STATUS   OF   THE   PRODUCTION   OF  RARER   METALS.  205 

potassium  sulfate.  The  filtrate  is  precipitated  with  oxalic  acid. 
The  oxalates  are  filtered  off,  washed,  dried,  ignited,  the  resulting 
oxides  boiled  out  with  water,  filtered  and  washed  with  hot  water. 
These  oxides  are  dissolved  in  the  least  amount  of  hydrochloric 
acid,  the  solution  boiled  and  precipitated  again  with  oxalic  acid. 
These  oxalates  upon  ignition  give  a  portion  of  the  yttrium  oxides. 
The  other  portion  is  separated  from  the  precipitated  double  sul- 
fates by  converting  to  hydroxides  or  oxides.  These  are  then 
dissolved  in  nitric  acid.  A  similar  amount  of  nitric  acid  is  then 
neutralized  by  magnesium  oxide  and  the  solution  added  to  the 
rare  earth  nitrates.  The  liquid  is  then  evaporated  to  crystalliza- 
tion. The  mother  liquor  is  poured  off,  a  quantity  of  bismuth 
magnesium  nitrate  added,  together  with  some  concentrated  nitric 
acid.  The  mass  is  heated  and  allowed  to  crystallize.  The  original 
crystals  are  also  recrystallized.  A  short  fractional  crystallization 
is  carried  out.  All  mother  liquors  that  fail  to  show  neodymium 
or  samarium  absorption  bands  are  placed  aside.  When  no  more 
mother  liquors  of  this  type  can  be  obtained,  the  process  is  stopped. 
The  mother  liquors  are  then  diluted,  the  bismuth  removed  by 
hydrogen  sulfide,  and  the  yttrium  earths  precipitated  by  oxalic 
acid.  This  precipitate  is  filtered  off,  washed  and  ignited.  This 
oxide,  together  with  that  obtained  from  the  potassium  sulfate 
solution  represents  the  total  yttrium  earths  originally  present. 
The  yttrium  earths  at  the  gadolinium  end  give  double  sulfates 
that  are  almost  insoluble  in  potassium  sulfate  solution. 

Some  metals  may  have  been  neglected  either  because  they  were 
considered  to  be  absolutely  useless,  or  because  they  appeared  to 
be  too  rare.  When  an  element  is  condemned  as  being  useless, 
it  is  evident  that  its  characteristic  properties  are  deeply  hidden. 
Many  years  ago,  thorium  oxide  was  a  very  rare  substance,  and, 
one  would  suppose,  considered  useless.  When  it  was  found  to 
be  an  ideal  substance  for  the  Welsbach  mantle,  a  search  was  made 
for  new  mineral  locations.  At  first  the  occurrence  seemed  to  be 
very  limited,  and  the  production  of  a  cheap  mantle  seemed  to  be 
out  of  the  question.  Finally  the  search  for  new  raw  material 
was  rewarded  by  the  discovery  of  monazite  sand.  Today  large 
amounts  of  sand  are  obtained  from  Brazil  and  Travancore.  Many 
other  deposits  are  known,  but  most  of  them  possess   a   lower 


206  C.  JAMES. 

thorium  content  than  those  mentioned.  At  the  present  time  there 
is  enough  thorium  for  mantles  and  for  other  purposes,  if  such 
can  be  found. 

If  we  search  the  Uterature  for  work  on  germanium,  we  shall 
find  little,  apart  from  that  at  the  time  of  its  discovery,  and  that 
done  during  the  last  two  or  three  years.  This  substance,  which 
once  seemed  so  useless,  is  attracting  much  attention  in  the  medical 
world,  because  of  its  action  on  the  blood.  According  to  several 
authorities,  it  should  be  of  great  value  in  certain  cases  of  anaemia. 
This  element  occurs  in  argyrodite  and  canfieldite  (which  appear 
to  be  very  rare),  and  to  a  very  minute  extent  in  some  zinc  ores. 
That  occurring  in  the  zinc  ores  is  concentrated  in  the  regenerated 
zinc  oxide,  which  is  obtained  from  the  retort  residues.  Even 
after  concentration,  the  amount  of  germanium  dioxide  is  still 
very  small. 

Recently  it  has  been  stated  to  occur  in  a  copper  ore  in  Africa. 
Some  of  this  mineral,  which  is  said  to  occur  in  considerable 
quantity,  was  obtained  and  examined.  The  mineral  proved  to  be 
rich  in  germanium,  which  is  easily  extracted  in  an  exceedingly 
pure  condition.  There  is  therefore  a  possibility  that  this  metal 
may  become  sufficiently  plentiful  so  that  its  effect  upon  metals 
and  alloys  may  be  determined.  It  alloys  quite  readily  with  cop- 
per; 5  per  cent  gives  a  pale  gold  colored  alloy. 

G.  Urbain  informed  me,  at  the  New  Haven  meeting  last  April, 
that  he  treated  several  tons  of  zinc  ore  and  obtained  only  a  few 
grams  of  germanium  dioxide.  This  was  finally  loaned  to  a 
doctor,  who  returned  two  decigrams.  Germany,  I  understand, 
has  forbidden  the  export  of  germanium  and  its  compounds. 

Will  thulium  ever  be  of  any  use  ?  It  must  be  admitted  that  it 
is  very  rare  and  extremely  troublesome  to  separate.  The  oxide 
certainly  possesses  characteristic  properties,  for  it  glows  when 
heated.  With  careful  heating  it  gives  a  beautiful  carmine  colored 
light,  which  changes  as  the  temperature  is  raised,  becoming  yellow 
and  then  almost  white.  I  am  optimistic  enough  to  believe  that 
all  these  very  rare  elements  will  prove  to  be  of  great  importance 
in  the  future.  Much  work  may  have  to  be  done,  and  we  must 
not  be  discouraged  by  stone  age  talk  in  the  time  of  super-steel. 

On  the  other  hand  there  are  some  of  the  so-called  rare  elements 


STATUS   OF   THE   PRODUCTION   OF   RARER   METALS.  207 

occurring  fairly  commonly  in  nature  which  have  been  subjected 
to  considerable  research,  and  which,  unfortunately,  are  still  un- 
conquered.    ^Metallic  beryllium  is  a  good  illustration  of  this  group. 

Beryl,  3BeO  .  Al^O.  .  6SiOo,  occurs  in  the  United  States  in 
New  Hampshire,  etc.,  and  in  many  other  parts  of  the  world.  It 
would  seem  that  in  the  case  of  this  element  there  are  perhaps 
three  reasons  why  the  metal  is  not  well  known.  The  mineral, 
beryl,  is  not  readily  decomposed ;  the  separation  of  beryllium  from 
aluminum  is  not  an  easy  matter,  and  the  reduction  of  beryllium 
compounds  presents  great  difficulties. 

Perhaps  the  simplest  way  to  decompose  beryl  is  to  heat  with 
sodium  hydroxide  in  the  following  manner:  The  mineral  ground 
to  200  mesh  is  mixed  with  1.5  parts  of  sodium  hydroxide,  and 
heated  over  a  powerful  oil  burner.  The  mass  first  softens,  then 
fuses  and  boils,  after  which  it  dries  to  a  friable,  bluish  earthy 
mass.  During  the  drying,  the  whole  should  be  well  stirred  so 
as  to  yield  a  line  powdery  product.  Care  should  be  taken  to 
prevent  a  second  fusion  in  which  a  glass  would  result.  Under 
good  conditions  nearly  complete  decomposition  of  the  beryl  is 
obtained  (98  per  cent).  This  friable  product  is  found  to  be 
superior  to  the  glassy  mass  obtained  by  a  second  fusion,  since 
it  is  easily  leached  by  water.  Careful  extraction  by  water  removes 
about  30  per  cent  of  the  total  silica,  and  a  considerable  amount  of 
sodium  hydroxide.  The  amount  of  ber^dlium  found  in  solution 
is  negligible.  This  leaching  is  best  carried  out  by  grinding  with 
water  in  a  ball  mill.  It  is  considered  advisable  to  make  this  ex- 
traction in  order  to  save  mineral  acid  in  the  next  stage  of  the 
process  where  the  alkali  is  removed  by  treatment  with  acid. 

The  powdered  product  of  the  fusion,  or  the  residue  from  the 
leaching  is  stirred  with  water,  sulfuric  acid  being  added  from 
time  to  time  to  neutralize  the  alkali.  It  is  essential  that  the  liquid 
be  neutral  or  slightly  alkaline  towards  the  end  of  the  stirring,  the 
desired  result  being  the  removal  of  the  soda,  leaving  the  beryllium, 
aluminum  and  silica  in  the  residue.  The  mass  is  then  filtered, 
washed  and  treated  with  dilute  sulfuric  acid  to  extract  the  beryl- 
lium and  aluminum.  The  solution  is  filtered  off,  evaporated  to 
dryness  and  gently  heated  to  render  the  silica  insoluble.     The 


2o8  C.  JAMES. 

residue  is  taken  up  with  water,  and  the  resulting  solution  contains 
the  beryllium  and  aluminum  originally  present  in  the  ore. 

A  solution  of  the  sulfates  is  tested  for  the  amount  of  mixed 
AI2O3  +  BeO.  Assuming  that  the  oxides  exist  in  solution  in 
the  same  ratio  as  they  occur  in  beryl,  about  three  times  as  much 
ammonium  sulfate  is  added  as  is  required  by  theory  to  convert 
the  aluminum  sulfate  to  ammonia  alum, 

Al^CSOJs  .  (NHJ3SO,  .  24U,0. 

The  solution  is  concentrated  and  cooled  to  10°  C,  when,  if  the 
concentration  has  been  sufficient,  practically  the  whole  of  the 
aluminum  separates  out  in  the  form  of  alum.  The  liquid  upon 
examination  is  found  to  be  almost  entirely  free  from  aluminum. 

A  small  amount  of  iron  still  remains,  and  this  is  separated  by 
diluting  the  solution,  heating  to  boiling,  and  oxidizing  the  iron 
if  necessary  by  potassium  bromate  or  some  other  suitable  oxidiz- 
ing agent.  The  liquid  is  then  neutralized  by  ammonium  hydrox- 
ide, and  the  iron  precipitated  by  ammonium  acetate  and  a  slight 
amount  of  acetic  acid.  If  too  much  acetic  acid  is  liberated,  more 
ammonium  hydroxide  is  added  to  neutralize  the  great  excess. 
When  a  sample  upon  filtering  and  treating  with  ammonium  sulfide 
gives  a  white  precipitate,  it  is  concluded  that  all  iron  has  been 
removed.  The  whole  is  then  filtered,  the  filtrate  boiled  and  the 
beryllium  precipitated  as  basic  carbonate  by  means  of  ammonium 
carbonate  or  sodium  bicarbonate.  The  basic  carbonate  is  filtered 
off,  washed  with  boiling  water  and  gently  dried. 

This  method,  when  carefully  carried  out,  gives  a  product  which 
is  almost  chemically  pure.  The  regaining  of  the  ammonium 
sulfate  is  an  important  matter,  which  has  not  been  completed  at 
present.  Many  thanks  are  due  to  H.  C.  Fogg,  J.  F.  CuUinan  and 
D.  A.  Newman  for  carrying  out  this  work. 

The  reduction  of  beryllium  compounds  such  as  the  oxide  by 
calcium  and  magnesium,  the  chloride  by  sodium  and  calcium ; 
and  the  electrolysis  of  fused  salts  and  salt  solutions  is  still  under 
investigation.  Before  we  can  make  much  more  progress  with 
regard  to  beryllium,  it  is  essential  that  we  know  more  about  its 
constants,  and  that  we  have  a  simple  method  for  its  quantitative 


STATUS  OF  THE  PRODUCTION   OF  RARER   METALS.  209 

determination.     Such  knowledge  would  allow  us  to  study  solu- 
bility curves  and  alloys  with  rapidity. 

A  few  years  ago  the  difficulties  of  a  zirconium  determination 
were  as  great  as  those  of  beryllium  are  now.  However,  as  zir- 
conium grew  in  commercial  value,  so  the  accuracy  of  its  deter- 
mination improved. 

During  recent  years  new  reagents  have  been  recommended  from 
both  inorganic  and  organic  divisions.  The  organic  section  seems 
to  be  rich  in  reagents  which  may  be  applied  to  the  titanium, 
zirconium,  cerium,  thorium"  family.  Phenylarsonic  acid  is  one 
which  is  being  thoroughly  examined  at  the  present  time.  The 
substituted  phenylarsonic  acids  act  similarly.  This  substance 
precipitates  zirconium  and  titanium  from  solutions  very  acid 
with  hydrochloric  acid,  while  cerium  and  thorivmi  remain  in  solu- 
tion. There  seems  to  be  considerable  difficulty  in  driving  off  all 
the  arsenic  on  ignition.  Igniting  in  a  current  of  hydrogen  rapidly 
removes  all  arsenic. 

Phenylarsonic  acid  precipitates  thorium  from  solutions  con- 
taining ten  per  cent  acetic  acid  and  a  slight  excess  of  ammonium 
acetate.  Under  the  same  conditions  the  metals  of  the  cerium  and 
yttrium  groups  are  not  precipitated.  Cerium  must  be  in  the 
cerous  state.  However,  the  thorium  carries  down  a  small  amount 
of  the  rare  earths,  and  it  is  necessary  to  make  a  second  precipita- 
tion. This  is  easily  performed  by  dissolving  the  thorium  phenyl- 
arsonate  in  a  little  hydrochloric  acid,  diluting  and  adding  acetic 
acid  until  the  solution  contains  about  ten  per  cent.  The  thorium 
is  then  reprecipitated  by  adding  an  excess  of  ammonium  acetate, 
and  a  little  more  phenylarsonic  acid.  This  second  precipitate 
is  once  again  dissolved  in  hydrochloric  acid,  the  solution  diluted, 
and  the  thorium  thrown  down  as  oxalate.  The  oxalate  is  filtered 
off,  washed,  dried  and  ignited  to  oxide.  Thorium  phenylarsonate 
can  be  ignited  directly  to  oxide,  if  a  current  of  hydrogen  be  used 
to  reduce  any  arsenic  remaining  after  the  first  ignition. 

When  hydrogen  peroxide  is  added  to  a  solution  of  cerium 
nitrate  containing  a  little  acetic  acid,  an  excess  of  ammonium 
acetate  and  phenylarsonic  acid,  a  precipitate  of  the  eerie  com- 
pound rapidly  forms.  The  quantitative  nature  of  this  reaction 
has  not  been  ascertained  as  yet. 
15 


2IO  C.   JAMES. 

Phenylarsonic  acid  can  be  easily  prepared  according  to  the 
method  recommended  by  Roger  Adams  (Journal  American 
Chemical  Society,  1922). 

We  must  not  forget  that  long  list  of  elements  known  as  the 
rare  earths,  which  includes  the  members  of  the  cerium  and  yttrium 
groups.  It  seems  unfortunate  that  these  substances,  which  are 
obtained  in  considerable  quantities  as  a  by-product  during  the 
extraction  of  thorium,  have  not  found  many  uses  commercially. 
It  is  true  that  some  are  used  to  a  certain  extent,  however  a  large 
amount  goes  to  waste.  If  we  except  cerium,  the  chemistry  of  the 
remaining  elements  is  a  little  section  all  by  itself.  These  rare 
earth  elements  resemble  a  homologous  series  of  carbon  compounds 
in  many  respects.  Many  properties  when  plotted  against  the 
atomic  weights  give  interesting  curves.  If  the  solubilities  of 
a  set  of  isomorphous  compounds,  containing  the  same  amount  of 
water  of  crystallization  are  examined,  it  is  usual  to  find  that  they 
lie  upon  a  smooth  curve.  On  the  other  hand  a  set  of  compounds 
possessing  two  or  three  states  of  hydration  will  give  a  curve 
resembling  that  of  a  single  substance  at  various  temperatures 
where  two  or  three  states  of  hydration  are  met. 

Unfortunately,  in  this  family  the  separation  of  the  elements 
from  each  other  is  no  simple  matter.  With  few  exceptions,  quan- 
titative analysis  is  unknown.  The  exceptions  include  cerium  and 
those  members  which  lie  at  opposite  ends  of  the  series.  Cerium 
can  be  separated  by  converting  it  into  the  eerie  condition,  when 
its  properties  become  similar  to  those  of  thorium. 

Lanthanum,  which  occurs  at  one  end,  can  be  separated  from 
erbium,  which  occurs  near  the  opposite  end  of  the  series,  by  stir- 
ring the  solution  of  the  nitrates  with  magnesium  nitrate  and  an 
excess  of  bismuth  magnesium  nitrate.  Lanthanum  magnesium 
nitrate  is  precipitated  out,  being  insoluble  in  the  bismuth  mag- 
nesium nitrate  solution.  Erbium  remains  in  solution  in  the  form 
of  the  simple  nitrate. 

There  is  therefore  little  trouble  in  separating  two  elements  that 
lie  far  apart  in  the  series.  It  is  an  easy  matter  also  to  separate 
one  element,  such  as  lanthanum,  from  several  elements  occurring 
at  the  opposite  end  by  the  above  method.  The  greatest  difficulties 
are  encountered  when  an  attempt  is  made  to  separate  two  or  three 


STATUS  OF  THE  PRODUCTION   OF  RARER   METALS.  211 

consecutive  elements  such  as  lanthanum,  praseodymium  and  neo- 
dymium.  In  this  case  the  praseodymium  comes  between  the  lan- 
thanum and  neodymium  when  the  double  ammonium  nitrates  are 
fractionated.  Lanthanum  ammonium  nitrate  is  the  least  soluble, 
while  the  neodymium  salt  is  the  most  soluble.  Praseodymium 
ammonium  nitrate  tends  to  accompany  both. 

The  three  elements  presenting  the  greatest  difficulty  are  dys- 
prosium, holmium  and  yttrium.  In  certain  cases,  the  scarcity  of 
an  element  makes  the  problem  still  more  difficult.  This  is  recog- 
nized in  the  cases  of  europium,  terbium  and  thulium. 

Many  members  of  the  rare  earth  group  can  be  obtained  in  large 
amounts  and  at  a  reasonable  cost  whenever  required.  This  state- 
ment applies  especially  to  cerium,  lanthanum,  praseodymium,  neo- 
dymium and  yttrium.  Of  course  it  is  evident,  when  a  use  is  found 
for  the  elements  mentioned,  that  those  which  are  rarer  will  be 
more  easily  obtained. 

Although  we  often  come  across  the  statement  that  the  rare 
earths  are  no  longer  rare,  we  must  realize  that  this  is  not  general. 
In  fact  europium,  terbium,  thulium  and  celtium  are  exceedingly 
rare.  Some  zinc  ores  contain  more  gallium  than  monazite  sand 
contains  europium. 

Since  the  separation  of  these  elements  is  based  upon  slight 
differences,  the  process  has  to  be  repeated  many  times.  In  some 
cases  thousands  of  operations  have  to  be  carried  out  before  some 
of  the  desired  salt  can  be  obtained  pure.  Fractional  crystallization 
or  fractional  precipitation  can  be  employed.  The  former  is 
usually  selected  because  it  is  cheaper  and  more  efficient  in  the 
long  run  with  large  amovmts.  Perhaps  the  cases  of  lanthanum 
and  yttrium  are  exceptional,  for  in  these  cases  precipitation  plays 
a  big  part. 

With  regard  to  fractional  crystallization  there  are  two  special 
lines  which  are  being  examined  at  the  present  time:  (a)  Solu- 
bility curves  of  various  salts  in  various  solvents;  and  (b)  the 
effect  of  one  rare  earth  salt  upon  another.  In  addition  to  this 
large  amounts  of  dysprosium,  holmium,  erbium,  thulium,  ytter- 
bium and  lutecium  are  being  separated  in  order  to  prove  whether 
there  are  any  other  elements  occurring  in  minute  amounts  in  this 
series.     Welsbach  believes  that  some  of  these  elements,  such  as 


212  C.   JAMES. 

terbium,  thulium,  etc.,  are  complex.  This  is  rather  against  the 
theory  of  Urbain.  and  the  problem  should  be  settled.  It  is  evident 
therefore,  that  the  rare  earths  require  considerable  investigation, 
for  as  yet  we  know  little  about  them.  Only  a  few  of  the  metals 
have  been  obtained  in  a  fused  state.  We  have  learned  much  with 
regard  to  the  structure  of  atoms  from  the  radio-active  elements, 
and  it  is  highly  probable  that  the  rare  earths  will  give  us  a  whole 
lot  more. 

Rare  earth  research  is  slow  and  tedious,  but  it  is  simple  com- 
pared with  what  it  used  to  be.  That  which  required  years  in  the 
time  of  Crookes  can  be  done  now  in  about  as  many  weeks. 

Gallium  and  indium,  two  other  elements  of  group  III,  should 
be  mentioned.  If  only  gallium  could  be  obtained  in  quantity,  it 
would  without  doubt  find  many  uses,  for  it  has  a  low  melting 
point,  and  when  pure  has  many  properties  approaching  those  of 
the  noble  metals.  A  few  years  ago,  after  the  discovery  made  by 
the  Bartlesville  Zinc  Company,  it  looked  as  though  there  would 
be  enough  material  to  supply  all  those  who  desired  to  work  upon 
it.  However,  the  ore  containing  gallium  occurs  only  in  pockets. 
Upon  purifying  this  crude  zinc  by  redistillation,  a  leady  residue 
was  obtained,  which  was  rich  in  gallium  and  indium.  Unfortun^ 
ately,  this  process  has  been  discontinued. 

Indium  occurs  much  more  commonly  in  certain  zinc  by-pro- 
ducts. Some  flue  dusts  have  shown  about  0.5  per  cent  of  this 
element.  All  this  material  passes  through  the  smelters  and  the 
indium  is  lost. 

\\'hen  we  study  a  list  of  rare  metals,  we  note  that  many  ele- 
ments, such  as  titanium,  zirconium,  etc.,  are  commoner  than  many 
of  the  metals  with  which  we  come  in  contact  every  day.  These 
elements  form  stable  compounds  that  are  reduced  with  difficulty. 
Moreover,  the  metals  when  finely  divided  are  very  active,  com- 
bining with  oxygen,  nitrogen,  carbon,  silicon,  etc.  This  great 
activity  and  a  melting  point  beyond  the  range  of  most  furnaces 
easily  account  for  the  stupendous  work  required  for  solving 
such  matters. 

It  is  not  long  since  we  had  to  use  the  greatest  of  care  in  hand- 
ling tungsten  lamps.  It  would  be  a  nightmare  to  a  man  accus- 
tomed to  the  use  of  ductile  tungsten,  to  be  placed  in  a  lamp 


STATUS  OF  THE  PRODUCTION   OF  RARER   METALS.  213 

factor^'  under  the  old  conditions.  Some  of  us  doubtless  remember 
those  old  times  with  the  huge  amount  of  labor  involved.  The 
production  of  ductile  tungsten  at  one  time  seemed  remote,  al- 
though the  number  of  investigators  was  comparatively  great. 
Finally  this  difficult  matter  was  solved,  and  not  until  then  did  the 
tungsten  lamp  really  become  commercial.  Today  tungsten  and 
molybdenum  can  be  worked  as  may  be  desired. 

Tantalum,  which  was  originally  worked  by  Siemens-Halske,  is 
now  being  produced  by  the  Fansteel  Company  in  a  ductile  form. 
They  also  state  that  columbium  can  be  put  on  the  market  in  a 
similar  state. 

It  is  especially  interesting  to  observe  that  with  improved  meth- 
ods, both  zirconium  and  uranium  metals  give  melting  points  very 
different  from  the  figures  obtained  by  earlier  workers.  The  prob- 
lem is  being  attacked  in  the  correct  manner  at  the  present  time. 
The  first  aim  is  to  obtain  pure  metal,  regardless  of  the  cost 
of  the  process,  in  order  to  study  its  properties.  When  these  have 
been  outlined,  it  will  be  much  easier  to  reason  out  a  simpler  plan. 

Zirconium  is  an  element  occurring  generously  in  nature,  so  its 
commercial  possibilities  are  considerable.  One  of  the  great  costs 
is  the  purification  of  the  salts  and  oxide.  The  cost  of  production 
is,  however,  much  less  than  it  used  to  be.  The  indexes  of  the 
various  journals  are  a  good  gauge  of  the  attention  that  the  various 
elements  are  receiving.  If  the  number  of  patents  mean  anything, 
the  future  of  zirconium  should  be  assured. 

Uranium,  a  by-product  obtained  during  the  extraction  of 
radium  salts,  is  easily  purified.  The  new  deposits  of  the  Congo 
indicate  that  there  will  be  no  shortage  of  this  element  in  the  near 
future.  Probably  long  before  this  new  region  is  exhausted,  others 
will  be  discovered.  Since  uranium  is  a  member  of  the  tungsten, 
molybdenum  and  chromium  family,  it  ought  to  have  commercial 
value. 

The  rare  elements  will  also  have  many  uses  when  in  the  form 
of  compounds.  Probably  some  will  be  used  as  catalysts.  Thal- 
lous  chloride  acts  as  an  excellent  chlorine  carrier,  especially  in 
the  chlorination  of  hydrocarbons.  Benzyl  chloride  is  not  pro- 
duced when  toluene  is  used.  Zirconium  has  a  tendency  to  remove 
hydrogen  from  compounds. 


214  ^-  JA^^ES. 

Many  of  the  rare  metals  are  thoroughly  established  in  the  com- 
mercial field,  but  we  must  realize  that  some  are  completely  dis- 
carded. If  the  minerals  are  not  being  used,  of  course  it  does 
not  matter  much,  for  they  can  be  mined  when  wanted.  It  is 
sad,  however,  when  many  rare  metals,  obtained  as  by-products, 
have  to  be  thrown  away. 

As  time  goes  on  the  remaining  territory  of  the  rare  metals 
will  become  more  and  more  difficult  to  explore,  since  the  easier 
ones  fall  before  the  steadily  increasing  power  of  the  investigators. 
It  is  highly  probable  that  in  all  difficult  tasks,  the  worker,  at 
times,  is  liable  to  become  discouraged.  However,  we  have 
many  fine  illustrations  in  the  past  where  the  problems  seemed 
hopeless,  but  where  the  work  was  finally  crowned  with  success. 
It  is  especially  interesting  to  read  the  work  of  Crookes  upon  his 
search  for  certain  elements  giving  phosphorescent  spectra. 

We  who  are  interested  in  the  commercial  application  of  the 
rare  metals,  ought  to  be  thankful  that  we  are  working  at  this 
period,  for  these  substances  are  being  launched  upon  their  journey 
through  the  commercial  world  as  never  before.  It  is  our  duty 
to  assist  in  this  project.  What  greater  reward  can  we  have  than 
to  learn  that  they  have  proved  seaworthy,  and  are  steadily  going 
ahead. 


A  paper  presented  at  the  Forty-third 
General  Meeting  of  the  American  Elec- 
trochemical Society  held  in  New  York 
City,  May  5.  192S,  Dr.  F.  M.  Becket  in 
the  Chair. 


THE  PREPARATION  OF  FUSED  ZIRCONIUM/ 

By  Hugh  S.  Cooper.- 
INTRODUCTION. 

In  the  course  of  certain  alloy  investigations  a  considerable 
quantity  of  zirconium  was  needed  to  pursue  the  work.  This 
metal  is  a  rather  scarce  commodity,  and  therefore  its  preparation 
in  the  laboratory  became  necessary.  Although  nothing  novel 
is  claimed  for  the  process  described  herein,  yet  there  are  a  number 
of  interesting  features  and  essential  precautions  involved  in  the 
production  of  the  metal,  which  are  considered  of  sufficient  im- 
portance to  be  published  in  some  detail.  The  experimental  data 
on  the  establishment  of  the  melting  point  of  zirconium  metal 
are  also  given,  as  well  as  a  brief  description  of  some  new  alloys. 

After  due  consideration  of  the  various  methods  employed  in 
the  past  for  making  zirconium,  it  was  decided  to  adopt  the 
method  in  which  zirconium  tetrachloride  is  reduced  by  sodium, 
because  this  seemed  to  be  the  most  promising.  Previous  ex- 
perience in  producing  anhydrous  chlorides  also  influenced  this 
decision  to  some  extent.  Although  some  zirconium  chloride  was 
made  by  passing  chlorine  over  the  oxide  in  the  presence  of  car- 
bon, the  yields  were  rather  unsatisfactory.  By  far  the  major 
portion  of  the  work  was  carried  out  by  the  action  of  chlorine  upon 
zirconium  carbide,  a  procedure  which  gave  satisfactory  results. 

The  carbide  chlorination  scheme  has  been  used  heretofore  by 
Moissan  and  Lengfeld,^  and  also  by  Wedekind,*  but  a  brief  des- 
cription of  the  apparatus  used  in  the  laboratory,  as  well  as  the 
results  obtained,  will  be  given  because  of  some  rather  important 
considerations  involving  the  purity  and  physical  character  of  the 

^  Manuscript  received  February  14,  1923. 

=  Kemet  Laboratories  Co.,  Inc.,  Cleveland,  O. 

3  Moissan  and  Lengfield.  Compt.  Rend.  122.  651   (1896). 

*  Wedekind,  Preparation  of  Zirconia  and  Tetrachloride  Z.   anorg.   Chem.    33,   81. 

215 


2l6  HUGH    S.    COOPER. 

chloride,   the   latter   greatly   influencing   the   yield   of   zirconium 
metal  during  reduction. 

PREPARATION  OF  ZIRCONIUM  TETRACHLORIDE. 

The  furnace,  as  shown  in  the  accompanying  illustration,  Fig. 
1,  is  of  the  horizontal,  wire-bound  tube  t}"pe  in  which  temperatures 
up  to  1000°  C.  are  obtainable.  The  carbide  is  placed  in  silica 
or  alundum  boats  approximately  7.6  x  15  x  2.5  cm.  (3  x  6  x  1  in.) 
deep.  These  boats  hold  approximately  228  g.  (8  oz.)  of  material 
each  and  are  inserted  in  a  fused  silica  tube  which  fits  snugly  into 


Fig.  1. 

Clilorination   Furnace. 

the  furnace.  The  diameter  of  this  vitreosil  tube  is  about  7 .6  cm. 
(3  in.)  and  the  length  about  92  cm.  (3  ft.)  One  end  of  the 
tube  is  fitted  into  a  terra  cotta  condenser  by  means  of  a  thick 
rubber  stopper.  The  other  end  of  the  condenser  is  also  sealed 
with  a  rubber  stopper  ol  equal  size,  in  the  center  of  which  is  an 
opening  for  the  chlorine  outlet  tube.  This  condenser  is  approxi- 
mately 30  cm.  (12  in.)  in  diameter  by  46  cm.  (18  in.)  long. 

After  placing  the  carbide  in  the  boats  the  tube  is  sealed  and 
the  current  applied  until  a  temperature  of  500°  C.  is  indicated 
by  a  thermo-electric  pyrometer,  the  couple  being  adjacent  to 
the  outside  wall  of  the  silica  tube.  At  this  time  a  stream  of 
chlorine   is  allowed  to  pass  over  the  carbide,  the  temperature 


THE   PREPARATION   OF   FUSED   ZIRCONIUM.  2I7 

being  held  as  close  to  500  or  550°  C.  as  possible,  since  this  seems 
to  be  the  optimum  temperature  from  the  standpoint  of  selective 
separation  of  the  iron.  The  physical  condition  of  the  chlorine 
depends  upon  the  temperature  at  which  it  is  condensed,  and 
this  determines  to  a  great  extent  the  yield  of  zirconium  metal 
which  is  obtained  upon  reduction.  For  example,  that  chloride 
which  has  been  condensed  in  the  terra  cotta  pipe  is  in  a  fine  state 
of  sub-division,  due  probably  to  the  rapid  cooling  at  normal  tem- 
perature, whereas,  that  in  the  end  of  the  silica  tube  near  the 
entrance  to  the  condenser  consists  of  a  heavy  dense  mass  of 
large  crystals.  The  temperature  at  this  point  is  about  200°  C. 
as  a  maximum.  As  soon  as  the  carbide  has  been  converted  to 
chloride,  which  usually  takes  four  or  five  hours,  the  respective 
chlorides  above  mentioned  are  removed  separately  and  placed 
in  glass  stoppered  bottles.  The  voluminous  finely  divided  ma- 
terial is  very  hygroscopic,  rapidly  absorbs  moisture,  and  in  so 
doing  assumes  a  lemon-yellow  color  characteristic  of  the  oxy- 
chloride.  The  heavy  crystalline  material,  on  the  other  hand 
is  much  less  affected  by  moisture  and  can  be  transferred  many 
times  with  slight  absorption  of  water. 

NATURE  OF  CARBIDE. 

An  analysis  of  some  of  the  carbide  used  in  some  of  these 
experiments  is  given  below : 

No.    1  No.    2 

Zr  73.88  83.78 

Fe  0.63  1.00 

Ti  0.41  0.48 

Si  0.10  1.40 

C  23.50  12.96 

98.52  99.62 

Many  experiments  have  conclusively  demonstrated  that  car- 
bide containing  some  free  graphite,  similar  in  analysis  to  No.  1, 
is  more  readily  attacked  by  chlorine  and  at  much  lower  tempera- 
tures than  material  of  the  second  type  which  more  nearly  ap- 
proaches ZrC  in  composition.  As  a  matter  of  fact,  ZrC  shows 
only  a  superficial  attack  at  temperatures  of  800°  C.  and  is  a 
very  dense,  heavy  material.     The  former  is  of  a  light  porous 


2l8  HUGH   S.   COOPER. 

nature,  is  friable  and  gives  practically  the  theoretical  recovery  of 
the  metal  as  chloride. 

A  representative  analysis  of  zirconium  oxide  made  from  chlor- 
ide, by  exposing  the  same  to  the  action  of  steam  with  subsequent 
ignition,  follows: 

Per  Cent 

Zr  O,  99.44 

Ti  O2  0.12 

Fe^O.  0.28 

Si  O5  0.08 

99.92 
REDUCTION  OF  ZIRCONIUM   CHLORIDE. 

The  furnace  depicted  in  Fig.  2  is  55  cm.  (21.5  in.)  in  height 
and  has  an  outside  diameter  of  15  cm.  (6  in.)  It  consists  of  two 
parts,  the  upper  containing  the  periscope  for  observing  the  tem- 
perature of  the  reaction  and  a  tube  connecting  with  the  vacuum 
pump.  The  lower  part  of  the  furnace  consists  essentially  of  a 
cylindrical  tube  having  an  inside  diameter  of  13  cm.  (5  in.)  and 
13  mm.  (0.5  in.)  wall,  to  which  a  base  is  welded,  the  upper  part 
having  a  13  cm.  (5  in.)  opening  with  a  welded  steel  collar.  The 
height  of  the  cylinder  is  33  cm.  (13  in.)  and  the  collar  is  41  mm. 
(1  5/8  in.)  in  diameter.  Just  below  this  collar  is  a  threaded  plug 
in  which  the  terminals  are  placed  which  conduct  the  current  to 
the  inside  of  the  cylinder.  A  lead  gasket  is  used  between  the 
top  and  base  to  make  an  absolutely  air-tight  joint. 

The  heating  unit  consists  of  an  alundum  core  wound  with 
nichrome  and  is  5  cm.  (2  1/8  in.)  inside  diameter  by  23  cm. 
(9  in.)  long.  Between  the  outside  wall  of  this  core  and  the 
steel  shell  the  space  is  insulated  with  sand.  The  steel  cylinders 
in  which  the  zirconium  chloride  and  sodium  are  packed  are  21 
cm.  (8  1/4  in.)  long  by  45  mm.  (1  3/4  in.)  in  diameter,  and  are 
provided  with  screw  tops.  Dense  crystalline  zirconium  chloride 
only  is  employed  to  make  the  metal,  since  it  is  more  permanent 
in  air  and  permits  a  greater  weight  of  material  to  be  used  per 
charge,  due  to  the  smaller  volume.  The  amount  of  chloride  and 
sodium  used  for  each  reaction  is  based  upon  the  equation : 

ZrCl,  -f  2Nao  ->  4NaCl  -f  Zr 

The  theoretical  amounts  would  therefore  be  232  g.  of  chloride 


THE   PREPARATION   OF   FUSED   ZIRCONIUM. 


219 


plus  92  g.  of  sodium,  this  yielding  about  90  g.  of  metal.  In  actual 
practice  230  g.  of  chloride  and  92.  g.  of  sodium  are  used,  as  the 
slight  excess  of  sodium  can  be  easily  washed  out  and  a  complete 
reduction  of  the  chloride  is  thus  assured. 

Zirconium  chloride  and  sodium  in  the  proportions  given  above 
are  rapidly  placed  in  alternate  layers  in  the  iron  cylinder,  the  cap 


riG.    _'. 
Reduction    Furnace. 


is  tightly  screwed  on,  and  the  cylinder  inserted  in  the  furnace. 
The  furnace  top  is  then  bolted  tightly  on  and  the  cylinder  is  then 
exhausted.  At  this  stage  the  current  is  applied  and  within  a 
short  time  a  temperature  of  500°  to  600°  C.  will  have  been  reached. 
Close  observation  at  this  point  will  show  a  sudden  rise  in  tem- 
perature, the  cylinder  reaching  about  900°  to  1000°  C.  It  is 
held  at  this  temperature  for  a  short  time,  current  is  turned  off, 
when  the  cylinder  has  become  cold  the  pump  is  cut  oft  and  the 


220  HUGH    S.   COOPER. 

furnace  opened.  It  is  extremely  important  not  to  remove  the 
cap  from  the  cylinder  until  the  temperature  has  dropped  to 
normal  because  of  the  extreme  activity  of  the  metal,  which  ignites 
upon  the  slightest  friction.  When  the  cylinder  has  been  opened 
a  black  skeleton-like  mass  is  exposed  to  view.  This  is  slowly 
added  in  successive  portions  to  a  large  volume  of  cold  water.  It 
will  be  immediately  found  that  the  heavy  lamellar  material  sinks 
to  the  bottom  while  the  powder  remains  suspended,  and  in  this 
manner  an  easy  separation  is  effected.  The  metal  is  thoroughly 
washed  with  cold,  and  finally  hot,  water  until  entirely  free  from 
sodium  salts ;  it  is  next  dried  for  several  days  at  a  temperature 
not  exceeding  85°  C.  If  this  procedure  is  carefully  followed, 
metal  having  the  following  approximate  analysis  will  be  obtained-. 

Per   Cent 

Zr  99.28 

Fe  0.14 

Ti  0.13 

Si  0.07 

99.62 

The  metal  suffers  little  loss  either  in  concentrated  or  in  dilute 
hydrochloric  or  nitric  acid,  even  when  boiled  therein.  It  is  also 
practically  insoluble  in  dilute  sulfuric  acid,  but  dissolves  com- 
pletely in  boiling  concentrated  sulfuric, 

EXPERIMENTS    TO    DETERMINE    THE    MELTING    POINT    OF 
ZIRCONIUM    METAL. 

The  melting  point  of  zirconium  is  not  known  with  certainly 
even  at  this  late  day.  According  to  Von  Bolton  this  point  lies  in 
the  neighborhood  of  about  2350°  C.  Burgess  states  that  three  ex- 
periments gave  1529°,  1533°,  1523°,  and  he  decided  on  1530°  C. 
as  the  melting  point.  According  to  Guertler  the  melting  point 
is  around  1700°  C.  Having  a  considerable  amount  of  metal  on 
hand,  with  ample  equipment  at  our  disposal,  it  was  decided  to 
make  a  few  experiments  in  an  attempt  to  correct  these  dis- 
crepancies. The  melting  experiments  were  carried  out  in  argon 
and  hydrogen,  as  well  as  in  vacuo.  Two  types  of  furnaces  were 
used.  One  of  these  was  an  Arsem  furnace,  in  which  the  metal 
was  melted  in  especially  prepared  zirconium  oxide  crucibles,  the 


THE   PREPARATION   OF    FUSED   ZIRCONIUM.  221 

Other  of  a  type  used  in  treating  tungsten  and  molybdenum  rods, 
one  end  of  the  metal  being  clamped  to  the  upper  electrode,  the 
other  dipping  into  a  pool  of  mercury  which  acts  as  the  lower 
electrode. 

The  zirconium  metal,  in  large  pieces  which  have  been  carefully 
dried  for  several  days  at  the  prescribed  temperature,  is  weighed 
in  lots  of  35  grams  each.  These  lots,  placed  successively  in  a 
die  and  subjected  to  a  pressure  of  about  35  tons,  yield  rods 
about  6  mm.  (1/4  in.)  square  and  25  cm.  (10  in.)  long.  The 
heating  of  these  rods  in  the  tungsten  treating  furnace  in  an  at- 
mosphere of  argon  or  hydrogen,  has  been  only  partially  success- 
ful up  to  this  time,  and  the  experiments  are  being  continued. 
Traces  of  ox\-gen  still  remaining  in  argon,  as  well  as  moisture  in 
both  gases,  has  prevented  obtaining  full-length  bars  of  completely 
sintered  zirconium  having  a  clean  surface  similar  in  appearance 
to  that  of  tungsten  or  molybdenum.  Nearly  all  such  rods  showed 
superficial  oxidation,  although  in  some  instances  short  lengths 
of  well-fused  metal  have  been  obtained.  These  experiments  indi- 
cate that  if  practically  dry  hydrogen  were  employed  it  should 
be  possible  to  make  solid  bars  of  metal  by  this  method,  and  it 
is  also  believed  that  once  the  metal  reaches  a  fused  state  the 
hydride  will  not  be  formed  at  the  lower  temperatures  upon 
cooling. 

The  melting  experiments  in  the  Arsem  furnace,  utilizing  zir- 
conium oxide  crucibles,  gave  some  gratifying  results.  The  tem- 
perature measurements  were  made  with  a  Leeds  &  Xorthrup 
optical  pyrometer,  especially  calibrated  for  temperatures  up  to 
3000°  C.  The  curve  supplied  with  the  furnace,  in  which  the 
temperature  is  plotted  against  the  power  input,  enabled  us  to 
check  the  temperature  readings  closely.  An  accurate  check  was 
also  obtained  by  use  of  molybdenum  and  tantalum,  which  have 
well-defined  melting  points. 

In  the  first  experiment  a  rod  of  pure  molybdenum  metal  was 
placed  on  one  end  of  a  zirconia  slab,  and  on  the  other  end  a  rod 
of  pressed  zirconium  metal.  Previous  tests  in  alundum  crucibles 
had  shown  that  the  melting  point  of  the  metal  was  above  that 
of  alundum.  The  first  optical  reading  was  taken  at  2420^  with 
8.5  kw.  input,  which  gave  2475°  on  the  chart.     A  final  reading 


222  HUGH    S.   COOPER. 

was  taken  at  9.5  kw.,  corresponding  to  2600°,  and  at  this  tem- 
perature one  end  of  the  molybdenum  rod  had  melted  to  a  small 
button.  The  optical  reading  at  this  time  gave  a  temperature  of 
2630°  C.  Upon  removal  from  the  furnace  the  metal  rod  was 
observed  to  possess  a  well-sintered  appearance,  but  it  was  not 
fused. 

In  the  next  experiment  several  grams  of  metal  were  placed  in 
a  zirconium  oxide  crucible.  The  temperature  reached  2650°, 
with  a  power  input  of  11  to  11.5  kw.  The  zirconium  in  this  in- 
stance was  well-sintered,  but  showed  no  signs  of  fusing  and 
no  loss  in  weight. 

In  the  third  trial  the  metal  was  placed  on  one  end  of  a  zir- 
conium oxide  slab,  with  a  piece  of  pure  tantalum  on  the  opposite 
end.  At  2800°  the  zirconium  had  partially  fused  into  small, 
flat  sections  without  showing  a  complete  melting,  the  tantalum 
had  begun  to  "sweat,"  indicating  a  temperature  close  to  the 
melting  point.  This  was  checked  with  a  further  and  similar 
experiment.  The  temperature  in  this  experiment  reached  2865° 
C.  The  tantalum  showed  distinct  fusion  on  one  corner,  and 
the  zirconium  had  flowed  rather  freely  over  the  sides  of  the 
slab.  This  was  checked  closely  in  a  further  test  in  which  no 
tantalum  was  used.  In  the  sixth  test  several  grams  of  metal 
were  placed  in  the  zirconia  crucible  and  heated  to  2910°.  At 
this  point  the  zirconium  melted  and  took  the  shape  of  the  cru- 
cible. These  experiments  led  to  the  conclusion  that  the  melt- 
ing point  of  zirconium  metal  is  above  that  of  molybdenum  and 
very  close  to  that  of  tantalum,  probably  about  2800°  C. 

ALLOYS  OF   ZIRCONIUM. 

Probably  the  most  interesting  alloys  of  zirconium  yet  dis- 
covered are  those  with  tin  and  with  nickel.  The  former  alloys 
are  exceedingly  pyrophoric  when  the  zirconium  content  exceeds 
60  per  cent,  and  in  this  respect  resemble  the  well  known  cerium- 
iron  alloys. 

Tin  and  zirconium  alloy  readily  with  evolution  of  heat  at 
about  800°  C,  giving  alloys  of  very  high  melting  points.  A 
composition  containing  approximately  25  per  cent  Zr  and  75 
per  cent  Sn  is  very  soft ;  when  heated  to  about  2000°  C.  most 


THE   PREPARATION   OF   FUSED   ZIRCONIUM. 


223 


224  HUGH   S.   COOPER. 

of  the  tin  can  be  removed,  the  zirconium  being  left  behind  as 
an  unfused  mass.  When  40  to  50  per  cent  Zr  is  present,  the 
alloy  begins  to  show  pyrophoric  properties  when  rubbed  across 
a  file,  and  at  60  to  80  per  cent  the  action  is  marked.  At  70 
to  80  per  cent  Zr  the  alloys  probably  equal  the  cerium-iron  alloys 
in  their  scintillating  effect.  Because  of  their  high  melting  point, 
it  is  not  possible  to  produce  rods  and  the  like  by  casting,  but 
these  can  be  readily  made  by  first  pressing  the  zirconium  metal 
into  the  desired  forms  and  then  heating  these  in  the  presence  of 
powdered  or  ingot  tin,  the  latter  being  rapidly  absorbed.  Com- 
positions containing  as  high  as  90  per  cent  Zr  can  be  made  in 
this  manner  and  these  appear  suitable  for  ignitors,  etc. 

When  used  in  percentages  up  to  about  15,  with  small  amounts 
of  aluminum,  silicon  and  tungsten  or  molybdenum,  and  with 
a  base  of  nickel,  excellent  machine  cutting  tools  are  produced 
which  retain  their  cutting  edge  at  a  red  heat. 

Ternary  alloys  have  also  been  made  using  manganese  or  anti- 
mony in  connection  with  zirconium  and  tin,  but  these  do  not  seem 
to  offer  any  advantage  over  the  binary  compositions. 

Fairly  high  percentages  of  zirconium  can  be  added  to  nickel 
before  malleability  is  lost.  The  range  of  toughness  probably  ex- 
tends up  to  about  30  per  cent.  A  20  per  cent  alloy  can  be  drilled 
and  machined ;  but  when  the  zirconium  approaches  50  per  cent 
considerable  hardness  is  manifested,  together  with  some  brittle- 
ness.  These  latter  alloys  can  be  produced  only  at  a  temperature 
around  1700°  C. 

With  gold,  zirconium  forms  straw-colored  brittle  alloys  for 
the  production  of  which  high  temperatures  are  also  required. 
The  zirconium  can  be  almost  entirely  removed  from  the  gold 
by  heating  in  an  oxy-hydrogen  flame.  Attempts  to  make  alloys 
with  antimony  and  zinc  were  unsuccessful,  as  the  metals  vola- 
tilized away  from  the  zirconium  before  alloying  occurred.  At 
about  1500°  C.  zirconium  dissolves  in  copper.  The  effect  is 
to  increase  the  hardness  with  little  change  in  color.  Few  alloys 
with  lead  were  made  and  these  seemed  to  disintegrate  when  ex- 
posed to  the  air  for  some  time.  Alloys  with  aluminum  have  been 
made  in  nearly  all  proportions,  the  action  taking  place  at  about 
1100°  C.    When  the  zirconium  content  is  relatively  low,  consider- 


THE   PREPARATION    OF   FUSED   ZIRCONIUM.  225 

able  toughness  is  manifested,  but  above  35  per  cent  brittleness 
prevails.  Unlike  the  tin-zirconium  series,  the  alloys  exhibit  no 
pyrophoric  properties.  The  effect  of  zirconium  on  aluminum 
appears  to  be  similar  to  that  of  silicon. 

Zirconium  has  been  alloyed  with  magnesium  by  reduction  of 
the  oxide  in  vacuo,  using  a  large  excess  of  magnesium.  Treat- 
ment with  hydrochloric  acid  removes  the  magnesium  without 
affecting  the  zirconium.  If  the  zirconium  is  not  too  high  the 
malleabihty  of  magnesium  is  not  affected  by  addition  of  the 
latter.  Alloys  of  tungsten  can  be  produced  by  pressing  the 
powdered  mixed  metals  into  briquettes.  In  this  manner  as 
much  as  25  per  cent  Zr  has  been  introduced.  Forging  properties 
of  this  series  have  not  yet  been  investigated. 


DISCUSSION. 


J.  W.  Marden'  :  The  paper  given  by  Dr.  Cooper  is  of  interest 
to  the  speaker  because,  in  collaboration  with  Mr.  ]\I.  N.  Rich,  he 
tried  to  make  zirconium  metal  by  the  identical  method  described, 
using  the  Arsem  furnace.  The  results  of  these  investigations 
(which  were  made  in  1919),  were  published  in  Bulletin  186,  U.  S. 
Bureau  of  Mines,  1921.  Although  we  had  some  success,  we 
found  that  it  was  nearly  impossible  to  avoid  some  contamination 
when  the  zirconium  was  heated  in  the  Arsem  furnace.  In  more 
recent  work  in  the  laboratories  of  the  Westinghouse  Lamp  Co., 
much  better  results  have  been  obtained  using  an  especially  con- 
siructed  high-frequency  high-vacuum  induction  furnace,  which 
has  been  described  by  Rentschler  and  ^vlarden  before  the  Ameri- 
can Physical  Society,  April  20,  1923. 

Attempts  were  made  to  fuse  zirconium  in  the  Arsem  furnace 
as  described  by  Dr.  Cooper,  and  the  reasons  for  failure  have 
been  given  on  page  82  of  the  above  bulletin,  which  dealt  briefly 
with  the  impossibility  of  completely  excluding  oxygen  and  carbon 
in  this  kind  of  apparatus.  Even  when  extreme  precautions  were 
observed,  using  a  slow  stream  of  pure  dry  H2  at  the  low  pressure 
of  a  few  mm.,  the  introduction  of  carbon  from  the  heating  helix 

*  Westinghouse  Lamp   Co.,   Bloomfield,   N.   J. 


226  DISCUSSION. 

and  oxidation  from  the  moisture  always  given  off  from  the  large 
amount  of  metal  surface  enclosing  the  furnace  could  not  be 
eliminated.  It  is  well  known  that  the  presence  of  oxide  will 
raise  the  melting  point  considerably. 

Zirconium  oxide  crucibles  were  also  used  in  some  of  the  above 
experiments.  The  preparation  of  these  crucibles  is  described  in 
Bulletin  186. 

Bars  of  sintered  zirconium  were  made  in  a  vacuum  treating 
furnace  in  the  laboratory  of  the  Westinghouse  Lamp  Company 
many  months  ago.  The  vacvmm  used  for  this  work  must  be  of 
the  highest  type,  using  mercury  diffusion  pumps  and  liquid  air 
traps.  These  bars  have  no  superficial  coating  of  oxide.  It  is  of 
interest  to  bring  out  some  of  the  points  about  the  purity  and  the 
methods  of  analysis  of  the  metal  powder.  The  analysis  of  a 
metal  powder  is  attended  with  extreme  difficulty  due  to  the  vola- 
tile gas,  either  in  free  or  adsorbed  state.  Wedekind  has  found 
that  in  a  good  vacuum  these  volatile  gases  can  not  be  all  removed 
from  zirconium  even  at  1,000°  C.  We  have  found  that  zirconium 
powder  often  contains  2  to  9  per  cent  of  moisture,  hydrogen, 
loosely  bound  nitrogen,  etc.  Since  zirconium  increases  only  about 
30  per  cent  in  oxidation  there  is  often  enough  gas  present  in 
weighing  the  sample  to  indicate  many  per  cent,  of  ZrO,.  In  the 
very  painstaking  work  of  Weiss  and  Neumann,  they  found  for 
example  that  0.1006  g.  of  zirconium  yielded  0.1333  g.  ZrOo.  This 
corresponds  to  98  per  cent  total  zirconium,  but  only  91  per  cent 
free  metallic  zirconium. 

Four  years  ago  the  writer  could  not  obtain  over  92.5  per  cent 
free  zirconium  by  the  best  methods  of  preparation.  The  purity 
of  the  powder  was  undoubtedly  greater  than  that  indicated,  but 
the  methods  of  analysis  are  not  yet  satisfactory  for  this  work. 
Analyses  should  be  stated  in  terms  of  free  metal  and  not  total 
metal.  According  to  the  results  we  obtained  the  method  of  des- 
iccation described  by  Dr.  Cooper  would  not  remove  all  of  the 
gases. 

The  melting  point  determination  of  zirconium,  as  with  certain 
other  of  the  rare  metals,  should  be  done  with  extreme  accuracy,  and 
these  determinations  must  be  made  under  conditions  which  preclude 
the  possibility  of  the  presence  of  oxygen  or  carbon.     The  metal 


THE  PREPARATION   OF   FUSED  ZIRCONIUM.  227 

which  is  used  for  this  must  be  analyzed  for  oxygen  and  the  per 
cent  of  oxide  in  the  sample  not  inferred  by  difference,  but  be 
actually  determined  analytically.  The  melting  point  given  by 
Dr.  Cooper  is  near  that  of  the  oxide  or  the  carbide.  The  melting 
point  has  been  determined  in  the  laboratory  of  the  Westinghouse 
Lamp  Company,  and  is  not  nearly  as  high  as  the  value  given  by 
Dr.  Cooper.  Our  metal  melted  sharply  and  did  not  show  any 
gradual  softening.  The  blistering  or  sweating  of  the  high  melt- 
ing point  metals  in  the  Arsem  furnace  may  have  been  indicative 
of  carbide  formations. 

Ruff-  has  suggested  the  formation  of  carbide  under  such  con- 
ditions as  Dr.  Cooper  worked.  This  carbide  was  partially  avoided 
in  the  work  of  Bulletin  186,  U.  S.  Bureau  of  Mines,  by  the  use 
of  purified  dry  hydrogen  to  sweep  away  hydrocarbon  vapors 
from  the  heating  helix  and  moisture  from  the  walls  of  the  con- 
tainer. The  melting  point  of  pure  zirconium  is  discussed  on 
page  97,  Bulletin  186. 

When  the  melting  points  of  the  metals  are  plotted  against  the 
atomic  numbers,  a  regularity  is  observed  which  would  indicate 
the  melting  point  of  zirconium  about  1,700°  C,  or  about  2,000° 
Abs.  This  may  be  somewhat  too  high  or  too  low,  but  roughly 
indicates  where  it  should  be  if  the  atomic  number  and  atomic 
weight  assigned  to  this  element  are  correct. 

Lastly,  the  sintering  of  mixtures  of  tungsten  and  zirconium 
has  been  tried  by  the  writer,  and  it  is  found  that  zirconium  in  a 
high  vacuum  distils  away  from  tungsten  at  temperatures  high 
enough  for  treating  this  metal.  Pure  metallic  zirconium  vola- 
tilizes rapidly  below  2,800°  C,  where  tungsten  is  treated  before 
working. 

H.  S.  Cooper:  It  has  been  stated  that  the  melting  point  deter- 
minations in  the  Arsem  furnace  were  in  effect  comparisons  be- 
tween zirconium  carbide  and  molybdenum  and  tantalum,  because 
it  was  thought  that  the  zirconium  would  be  converted  to  carbide 
in  the  atmosphere  which  prevails  in  a  furnace  of  this  type. 
There  is  no  evidence,  up  to  this  time,  that  this  occurs  when  zir- 
conium is  the  metal  used.  Our  analyses  have  shown  that  there 
is  only  an  absorption  of  carbon  to  an  extent  of  about  0.2  per  cent 

=  Z.   Electrochemie,   24,   157    (1918). 


228  DISCUSSION. 

when  the  metal  is  heated  to  its  melting  point,  and  it  is  unlikely 
that  this  amount  of  carbon  would  materially  affect  the  results 
in  either  direction. 

In  all  of  our  work  we  have  been  careful  not  to  use  amorphous 
zirconium,  as  we  have  found  that  this  grade  of  metal  is  apt  to 
contain  oxygen  to  an  appreciable  extent.  Two  grades  of  metal 
exist  after  reduction,  and  we  have  been  careful  to  pick  clean, 
bright  samples,  which  are  then  pressed  into  the  rods  which  I 
have  described.  When  such  rods  are  used  in  the  Arsem,  I 
seriously  question  whether  there  can  be  any  combination  with 
oxygen,  as  the  furnace  atmosphere  is  decidedly  reducing,  which 
is  evidenced  by  the  discoloration  that  you  have  noted  on  the 
zirconium  oxide  crucibles.  This  change  of  color  on  the  crucibles 
is  not  due  to  carbon  or  carbide,  but  is  an  actual  reduction  of  the 
oxide  to  metallic  zirconium,  which  has  been  proved. 

Mr,  Marden's  criticism  on  the  use  of  the  Arsem  furnace  for 
these  experiments  seems  to  me  to  be  rather  misdirected,  in  view 
of  his  statements  under  the  title  of  "Preparation  of  Coherent 
Metal  in  Arsem,"  on  pp.  94  and  96  of  the  Bureau  of  Mines  Bul- 
letin No.  186 — "Thus  the  experiment  had  accomplished  what  had 
been  considered  impossible,  namely,  the  fusion  of  the  amorphous 
metal."  The  analyses  with  this  statement  is  what  might  be 
expected  by  the  use  of  amorphous  metal. 

It  is  rather  strange  that  having  a  product  of  a  purity  indicated 
by  the  various  analyses  discussed  in  the  bulletin,  that  IMr.  JVIarden 
was  unable  to  produce  an  alloy  of  zirconium  with  tin,  as  these 
alloys  are  simply  prepared.  If  I  have  correctly  interpreted  the 
remarks  made  by  this  gentleman  there  appears  to  be  some  doubt 
in  his  mind  that  the  zirconium-tungsten  alloys  can  be  prepared  in 
the  manner  outlined,  since  he  has  stated  "that  in  his  experience 
the  former  metal  boils  away  from  the  latter  before  alloying 
occurs."  In  this  connection  I  wish  to  state  that  we  have  prepared 
a  great  many  alloys  of  zirconium  and  tungsten.  These  were  made 
by  thoroughly  blending  the  powdered  zirconium  with  powdered 
tungsten,  pressing  the  product  into  rods,  sintering  the  rods  in 
vacuo  and  then  heating  the  same  by  their  own  resistance  up  to 
about  2,200°  C.  There  can  be  no  doubt  that  alloys  of  any  desired 
percentage  of  either  metal  can  be  prepared  in  this  manner,  and 


THE   PREPARATION    OF    FUSED   ZIRCONIUM.  229 

contrary  to  Mr.  Marden's  statement,  if  any  evaporation  of  the 
zirconium  does  occur  the  amount  is  so  slight  as  to  be  invisible 
on  the  surface  of  the  o^lass  enclosure  in  which  the  experiments 
were  conducted. 

W.  C.  Arsem"  {Communicated)  :  It  should  be  remembered 
that  in  a  vacuum  furnace  the  character  of  the  results  de- 
pends on  the  maintenance  of  a  vacuum  as  good  as  can  be  obtained. 
The  best  results  are  not  to  be  expected  unless  the  pressure  is  kept 
low,  probably  around  1  to  10  microns.  It  is  not  sufficient  to 
maintain  a  fairly  good  vacuum  by  an  efificient  pump  acting  against 
a  continuous  leak  in  the  furnace.  Leaks  should  be  absent.  In 
order  to  guard  against  leaks  it  is  necessary  to  make  sure  of  the 
absolute  tightness  of  both  electrode  and  cover  gaskets  by  appro- 
priate tests.  The  technique  for  realizing  this  condition  should 
be  quite  obvious,  although  it  is  often  carried  out  imperfectly 
through  failure  to  recognize  its  importance. 

A  furnace  with  graphite  parts,  allowed  to  stand  open  to  the 
air  when  not  in  use,  absorbs  and  condenses  a  considerable  amount 
of  air  and  moisture,  and  to  avoid  this  condition  it  should  be  kept 
exhausted  when  not  in  use.  When  experiments  are  to  be  tried 
in  high  vacuum  it  is  best  to  run  the  furnace  under  experimental 
conditions  without  a  charge  until  gases  are  well  removed  and  a 
high  vacuum  can  be  maintained  at  a  high  temperature,  then  let 
it  cool  under  exhaust  and  open  it  with  the  temperature  of  the 
cooling  water  above  the  dew-point  to  avoid  condensation  of  mois- 
ture. Then  insert  the  charge,  exhaust  immediately  and  continue 
to  exhaust  at  low  temperature  until  a  high  vacuum  is  obtained 
before  applying  the  current. 

The  presence  of  oxygen  or  water  in  the  interior  means  that 
the  atmosphere  will  eventually  be  chiefly  carbon  monoxide.  This 
is  not  a  "reducing"  atmosphere  except  under  special  conditions. 

With  many  of  the  metals  whose  oxides  are  extremely  stable  we 
have  at  high  temperatures  the  following  reactions : 

CO  -f  M  ±5  ^lO  -f  C 
and 

CO  4-  2^1  ±5  ^lO  -\-  MC 

'  Consulting  Chemical   Engineer,   Schenectady,  N.   Y. 


230  DISCUSSION. 

The  action  which  takes  place  is  really  more  complex  than  the 
equations  indicate,  but  the  net  result  is  that  a  mixture  of  oxide 
and  carbide  can  be  formed  at  least  superficially  by  heating  certain 
metals  in  an  atmosphere  of  CO. 

It  would  be  advisable  in  reporting  results  of  research  of  this 
kind  to  include  in  the  paper  a  complete  log  of  each  furnace  run, 
including  pressure  readings.  Absence  of  these  data  may  lead  to 
much  misunderstanding  and  uncertainty. 


A  paper  presented  at  the  Forty-third 
General  Meeting  of  the  American  Elec- 
trochemical Society  held  in  New  York 
City  May  5,  1923,  Dr.  F.  M.  Becket  in 
the   Chair. 


EXPERIMENTS  WITH  URANIUM,  BORON,  TITANIUM,  CERIUM 
AND  MOLYBDENUM  IN  STEEL 


By  H.  W.  GiLLETT  and  E.  L.  Mack.* 

Abstract. 

Of  U,  B,  Ti,  Zr,  Ce  and  Mo  used  as  alloying  elements  in  heat- 
treated  steels,  only  Mo  has  a  decided  and  consistently  beneficial 
effect.  In  the  types  of  steel  in  which  the  other  elements  were 
used  they  were  either  of  slight  effect  one  way  or  the  other,  or 
decidedly  harmful. 

U  probably  has  a  slight  strengthening  effect,  but  similar  results 
can  be  obtained  by  cheaper  means.  B  and  Ce  are  harmful.  Ti 
and  Zr  have  about  as  much  effect  as  equal  amounts  of  Si.  'Mo  is 
a  real  and  potent  alloying  element. 


When  the  Bureau  of  Alines  was  actively  studying  radium  pro- 
duction it  was  thought  desirable  to  study  the  preparation  of  ferro- 
uranium,  and  this  work  was  assigned  to  the  writers.^  This  gave 
a  stock  of  ferro-uranium.  On  account  of  reported  German  use 
of  U  steel,  the  Watertown  Arsenal  requested  that  an  experi- 
mental series  of  U  steels  be  made  up  and  supplied  to  the  arsenal 
for  study.  Later,  further  series  of  W,  j\Io  and  other  steels  were 
requested.  Since  the  Bureau  was  equipped  for  this  sort  of  small- 
scale  work,  the  navy  then  requested  the  preparation  of  some  high 
Si-Ni  steels,  containing  Zr.  In  that  connection,  Ce  and  B  were 
also  added  to  this  Ni-Si  steel. 

In  the  above  work,  the  Bureau  merely  prepared  and  analyzed 

*  Published  by  permission  of  the  Director  of  the  Bureau  of  Mines.  Manuscript 
received  January  30,  1923. 

*  Department  of  Interior,   Bureau   of  Mines,   Ithaca,   N.   Y. 

'  Gillett,  H.  W.,  and  Mack,  E.  L,.,  Preparation  of  ferro-uranium,  Tech.  Paper  177, 
Bur.  of  Mines   197. 

231 


232  H.   W.   GILLETT   AND  E.   L.    MACK. 

the  steels,  the  testing  being  done  by  other  departments  or 
bureaus.^"*  Co-operative  agreements  were  later  made  with  the 
\\'elsbach  Co.  for  a  further  study  of  Ce  steel,  and  with  the 
Vanadium  Corporation  of  America  for  a  study  of  various  types 
of  Mo  steel.  These  latter  series  have  been  tested  by  the  Bureau 
of  Mines  and  a  comprehensive  series  of  endurance  tests  on  them  is 
still  under  way.  As  regards  data  on  the  physical  properties  of 
the  other  steels,  these  are  wholly  lacking  in  the  case  of  the  steels 
prepared  for  the  arsenal,  and  in  the  case  of  most  of  the  steels  of 
the  Ni-Si-Zr  series  they  are  fragmentary,  in  that  only  normalized 
specimens  and  specimens  subjected  to  a  single  heat  treatment 
and  that  at  a  very  low  draw  temperature  were  tested. 

Detailed  data  on  the  preparation  of  the  steels,  especially  in 
regard  to  recovery  of  the  alloying  elements,  have  been  fully  given 
elsewhere,^  and  will  be  only  briefly  touched  on  here. 

The  steels  were  made  up  in  50  to  lOO-lb.  heats  in  an  indirect  arc 
furnace.  Ferro-alloys  of  readily  oxidizable  alloying  elements 
were  usually  added  at  the  end  of  the  heat,  just  before  pouring. 
Armco  iron  was  used  as  the  base,  by  which  means  sulfur  was 
held  to  0.035  per  cent,  usually  below  0.030  per  cent  and  P  to 
below  0.02  per  cent,  usually  below  0.015  per  cent. 

URANIUM. 

Since  the  arsenal  desired  a  number  of  steels  of  high  U  content, 
attempts  were  made  to  prepare  these.  Steels  analyzing  2  per  cent 
U  and  over  were  prepared,  but  usually  shattered  in  forging. 
Steels  analyzing  over  0.55  per  cent  U  were  slushy  when  poured, 
although  very  hot,  and  all  such  steels  showed  terrific  segregation 
in  different  parts  of  the  ingot.  To  get  a  uniform  U  content  of 
0.35  to  0.50  per  cent  U  it  was  necessary  to  add  over  1  per  cent  of 
U  as  ferro-uranium,  ferro-uranium  alloy  low  in  C  and  high  in  U 
giving  the  best  results. 

Physical  tests  (Bureau  of  Standards,  small  round  test  bars 
cut  from  plates  one-half  in.  thick)   were  made  only  on  some  of 

*  Wheeler,  H.  E.,  Nitrogen  in  Steel  and  the  Erosion  of  Guns,  Trans.  Am.  Insi.  Min. 
and  Met.  Eng.  47,  257    (1922). 

*  Burgess,  G.  K.,  and  Woodward,  R.  W.,  Manufacture  and  Properties  of  Steel 
Plates  containing  Zirconium  and  other  elements.  Tech.  Paper  207,  Bur.  of  Standards, 
1922. 

»  Gillett,  H.  W.,  and  Mack,  E.  L.,  Experimental  Production  of  Alloy  Steels,  Bull. 
199,  Eur.  of  Mines,  1922. 


EXPERIMENTS  WITH  RARE  ELEMENTS  IN  STEEL. 


233 


the  non-segregated  Ni-Si  steels,  to  which  uranium  was  added. 
These  are  given  in  Table  I,  together  with  a  couple  of  comparison 
steels  without  U. 

Table  I. 

Physical  Tests  on  Alloyed  Steels. 

Normalized  from  800  to  840°  C. 


Steel 
No. 

c 

Si 

Mn 

Ni 

u 

Yield 
Point 

Tensile 

El 

1244 

1229 
1228 
1327 
1227 
1237 

0.43 
0.45 
0.63 
0.45 
0.40 
0.49 

1.30 
1.05 
1.20 
2.42 
1.45 
2.20 

0.90 

0.75 
0.84 
0.70 
0.84 
0.94 

3.00 
3.00 
3.00 
2.92 
3.10 
3.05 

0.34 
0.36 
0.37 
0.52 

134,000 
234,000 

169,000 

* 

97,000 
108,000 

184,000 

240,000 
176,000 

134,000 
156,000 

6 
3 

0.5 

18.5 
14.5 

steel 

Red. 

Brinnell 

Heat  Treated.   Quenched   from   800-840°   C.   in   oil; 
175°    C.    draw. 

No. 

Yield 
Point 

Tensile 

El. 

Red. 

Brinnell 

1244 

1229 
1228 

13 

8.5 
2.5 

52 

41 

290 
315 
305 

195,000 
192,000 

310,000 
283,000 
300,000 

10.5 
2.5 
1 

35 

8.5 
3.5 

625 
530 
620 

1227 
1237 

265 
315 

205,000 
258,000 

286 
313 

000 
000 

8.5 
8 

39 
25 

555 
530 

*  Broke  in  rolls. 

The  normalized  U  steels  showed  a  martensitic  pattern,  and 
were  stronger  and  less  ductile  than  the  comparison  normalized 
steels.  The  heat-treated  steels  with  U  show  on  the  average  no 
appreciable  improvement  over  those  without.  The  U  steels 
contain  characteristic  blue  inclusions.  While  great  claims  have 
been  made  for  U  in  high  speed  steel  and  in  ordinary  steels,  the 
first  seem  open  to  grave  question  and  the  second  seem  to  be  cov- 
ered by  the  comment  of  Poluskin*^  to  the  efifect  that,  while  U  may 
somewhat  increase  tensile  strength  and  toughness  without  loss  of 
ductility,  it  does  nothing  that  cannot  be  done  with  cheaper  alloying 
elements.    He  thinks  much  of  the  U  in  steel  is  present  as  oxide. 

The  cost  of  U,  the  difficulty  of  introducing  it  without  excessive 
loss  and  without  the  formation  of  dangerous  inclusions,  together 

8  Poluskin,  E.  Les  aciers  al'  uranium.  Rev.  de  Met.,  17,  421  (1920).  Iron  Trade 
Rev.,  68.  413   (1921).     Iron  Age,  106,  1512  (1920). 

16 


234  ^-   "^^^   GILLRTT  AND  E.   L.   MACK, 

with  the  cessation  of  mining  of  domestic  carnotite,  make  uranium 
steel  arouse  httle  enthusiasm  at  present.  Were  its  alleged 
advantages  more  outstanding  or  the  supply  of  U  larger,  it  would 
deserve  further  study.  It  might  have  use  as  a  scavenger,  but  it 
has  not  impressed  us  as  promising  on  this  score,  as  its  oxidation 
products  do  not  appear  to  be  readily  released  by  the  steel. 

BORON. 

Since  B  is  reputed  to  give  great  hardness  to  steel  some  C-B 
and  Ni-Si-B  steels  were  made  up.  The  only  ferroboron  avail- 
able was  the  thermit  product.  One  lot  contained  two-thirds  as 
much  Al  as  B  and  another  one-third  as  much.  The  ferro-alloy 
was  readily  taken  up.  Adding  it  at  the  end  of  the  heat,  90  per 
cent  of  the  B  or  better  was  recovered.  Even  when  charged  at  the 
start  of  the  heat,  an  80  per  cent  recovery  was  made.  Analyses  of 
different  parts  of  the  ingots  showed  no  segregation  of  boron. 

The  B  steels  with  around  0.10  per  cent  B,  and  with  a  C  content 
of  0.15  to  0.70  per  cent,  had  an  amazing  freezing  range.  They 
started  to  solidify  about  the  usual  temperature,  but  did  not 
become  fully  solid  till  the  temperature  dropped  down,  somewhere 
around  the  melting  point  of  cast  iron. 

During  the  long  freezing  range  the  ingot  was  plastic,  and  when 
poked  it  acted  like  pie  crust  under  the  cook's  thumb.  There  is 
plainly  a  very  low-melting  carbon-iron-boron  eutectic.  This  is 
clearly  shown  metallographically.  Aloreover  the  first  couple  of 
ingots  of  boron  steel  rolled  by  the  Bureau  of  Standards  fell  to 
pieces  of  their  own  weight  when  heated  to  the  usual  rolling 
temperature  and  picked  up  by  tongs,  so  that  the  preheating 
temperature  had  to  be  reduced. 

With  0.30-0.50  per  cent  B  even  low  carbon  steels  lost  a  great 
deal  of  their  ductility,  and  even  0.06  per  cent  B  spoiled  a  0.45  per 
cent  C  steel  for  heat  treating.  The  B  eutectic  in  the  cast  material 
is  a  network,  but  this  can  be  broken  up  and  spheroidized  by  hot 
working  (possibly  also  by  thermal  treatment),  and  in  that  state 
the  steel  is  not  so  brittle.  Heating  to  a  normal  temperature  for 
quenching  causes  a  network  to  reappear  and  gives  a  brittle 
product. 

It  is  within  the  bounds  of  possibility  that  the  steels  might  be 
handled    so    as    to    be    good    for    something,    but    hot-working 


EXPERIMENTS  WITH  RARE  ELEMENTS  IN  STEEL.  235 

processes  as  used  on  other  steels  do  not  produce  anything  worth 
having.  A  purely  scientific  study  of  the  Fe-B-C  system  would  be 
highly  interesting.    One  wonders  what  B  might  do  in  cast  iron. 

If  B  is  to  be  used  as  an  alloying  element,  the  steels  will  have  to 
be  given  special  treatment,  and,  lacking  a  detailed  preliminary 
scientific  study,  it  is  hard  to  see  how  they  can  be  commercially 
useful. 

In  regard  to  boron  as  a  scavenger,  the  fact  that  it  gives  a  high 
recovery  even  when  added  at  the  start  of  the  heat  indicates  that 
it  cannot  be  expected  to  have  strong  deoxidizing  action.  If  it  will 
act  as  a  deoxidizer  it  might,  by  the  formation  of  the  oxide,  tend 
to  flux  out  other  oxides  and  hence  be  beneficial.  One  thinks  at 
once  of  boron  nitride  and  of  the  possibility  that  it  would  remove 
nitrogen.  In  a  British  patent,  Walter,^  a  German,  says  that, 
while  0.2  per  cent  or  more  B  causes  brittleness,  anywhere  from 
0.001  to  0.10  per  cent  B  causes  astonishing  grain  refinement  in 
steel,  and  that  similar  amounts  in  cast  iron  give  stronger  material 
with  graphite  in  spherical  form.  One  would  be  more  impressed 
by  his  claims  if  he  did  not  state  also  that  from  0.007  to  0.01 
per  cent  B  in  a  C  steel  makes  it  self-hardening. 

The  writers  are  inclined  to  feel  that,  while,  on  the  face  of 
returns,  boron  does  not  appear  to  be  of  any  use  in  steel,  a  sys- 
tematic study  of  B  in  steel  might  show  greater  possibilities  than 
can  be  seen  at  present.  This  view  is  based  on  the  fact  that  B 
has  a  real  efifect  and  gives  a  product  with  peculiar  properties, 
which  might  conceivably  be  utilized. 

TITANIUM. 

For  comparison  with  Zr  steels,  which  carry  some  Ti,  various 
plain  Ti  and  Ni-Si-Ti  steels  were  made,  using  a  thermit  ferro- 
titanium  containing  about  one-fourth  as  much  Al  as  Ti.  Adding 
this  at  the  start  of  the  heat,  around  20  per  cent  of  the  Ti  was 
recovered,  while,  by  adding  it  at  the  end  of  the  heat,  around  65 
to  70  per  cent  was  recovered.  Steels  were  made  with  up  to  2 
per  cent  Ti.  Segregation  of  Ti  was  not  troublesome.  The  steels 
containing  Ti  as  alloying  element  were  certainly  no  better,  and 
generally  somewhat  less  ductile  than  comparison  steels  without 

'  Walter,  R.,  British  Pat.  160.  792,  Aug.  23.  1921. 


236  H.   W.  GILIvETT  AND  E.   L.   MACK. 

Ti.    Steels  with  only  a  few  hundredths  per  cent  of  Ti  showed  no 
superiority  over  the  comparison  steels. 

ZIRCONIUM. 

The  work  on  Zr  was  required  because  of  the  high  recom- 
mendation given  a  Ni-Si  steel  carrying  Zr,  by  Mr.  W.  H.  Smith 
of  the  Ford  Motor  Co.  While  "zirconium  steel"  was  loudly 
heralded,  it  is  only  fair  to  say  that  Mr.  H.  T.  Chandler,  formerly 
with  the  Ford  Motor  Co.,  the  metallurgist  in  actual  charge  of  the 
Ford  experimental  work  with  Zr,  considered  the  value  of  this 
steel  to  lie  chiefly  in  the  Ni-Si  combination,  with  the  possibility 
that  Zr  added  something  to  that  combination. 

As  a  result  of  the  agitation  for  Zr  steel,  much  baddeleyite  was 
imported  from  Brazil  at  a  time  when  shipping  was  precious,  and 
ferro-alloy  manufacturers  had  to  displace  the  production  of  ferro- 
alloys of  proven  value  for  that  of  ferrozirconium.  The  navy  was 
not  stampeded  by  the  agitation,  but  decided  to  find  out  what,  if 
any,  virtue  lay  in  the  Zr. 

In  the  work  done  by  the  Bureau  of  Mines  for  the  navy,  some 
75  Zr  steels,  and  an  equal  number  of  comparison  steels  without 
it,  were  made  in  the  preliminary  work  in  which  the  steels  were 
rolled,  heat  treated  (normalized  and  given  a  quench  and  a  single 
low  temperature  draw)  and  tested  for  mechanical  properties  by 
the  Bureau  of  Standards.  In  the  later  work  a  series  of  some  30 
steels,  with  and  without  Zr,  was  made  by  the  Bureau,  rolled,  and 
each  given  three  or  four  different  heat  treatments  by  the  Halcomb 
Steel  Co.  and  tested  for  mechanical  properties  by  the  navy. 

Although  thermit  ferrotitanium  was  found  to  give  fair  recov- 
ery of  Ti,  thermit  ferrozirconium  did  not,  the  recovery  of  Zr 
averaging  not  over  10  per  cent. 

Various  electric  furnace  ferro-alloys  reduced  by  C,  and  carry- 
ing considerable  C,  made  by  the  Bureau  of  Mines  and  by  the 
Southern  Manganese  Co.  also  gave  a  low  recovery,  averaging 
under  5  per  cent. 

Electric  furnace  ferro-alloys  made  by  the  Bureau,  using  Si  as 
reducing  agent,  gave  60  to  80  per  cent  recovery.  An  electric 
furnace  ferro-alloy,  low  in  Si,  made  by  the  Electro  Metallurgical 
Co.,  with  Al  as  reducing  agent,  gave  around  10  per  cent  recovery, 
but  an  alloy  similarly  made  but  in  place  of  Fe  containing  55  per 


EXPERIMENTS  WITH  RARE  ELEMENTS  IN  STEEL.  237 

cent  Ni,  gave  40  per  cent  recovery,  while  the  Electro  Metallurgical 
Go's  Si-Zr  (30  per  cent  Zr-45  per  cent  Si,  reduced  by  C  in  the 
presence  of  Si)  gave  a  55  per  cent  recovery. 

In  the  second  series,  using  Electro  Metallurgical  Go's  ferro- 
alloy, the  Si-Zr  gave  a  50  per  cent  recovery,  Xi-Zr  50  per  cent 
and  a  Si-Ni-Zr  (27  per  cent  Zr,  22  per  cent  Ni,  35  per  cent  Si) 
made  by  melting  together  Si-Zr  and  Ni-Zr,  65  per  cent. 

To  get  these  recoveries,  the  Zr  alloy  had  to  be  added  at  the 
end  of  the  heat.  If  added  at  the  start,  the  steel  contained  only 
traces  of  Zr.  When  remelting  crop  ends  containing  0.20-0.25 
per  cent  Zr  and  0.03  per  cent  Ti,  the  steels  came  out  with  no 
trace  of  Zr  and  under  0.01  per  cent  Ti. 

When  we  consider  the  loss  of  Zr  in  making  the  ferro-alloy 
from  ore,  that  from  ferro-alloy  to  steel,  and  that  in  remelted 
scrap  containing  Zr,  the  recovery  from  ore  to  finished  steel  would 
not  be  over  40  per  cent  and  probably  well  under  that  figure. 

No  matter  what  the  Zr  alloy  used,  steels  finishing  with  from 
0.30  to  0.80  per  cent  Zr  regularly  showed  a  segregation  of  Zr, 
the  top  of  the  ingot  containing  say  30  per  cent  more  Zr  than 
the  butt. 

The  Ni  and  Si  introduced  by  the  Ni-Zr,  Si-Zr  or  Ni-Si-Zr 
alloys  did  not  show  segregation.  With  not  over  0.25  per  cent  in  the 
finished  steel,  segregation  of  Zr  is  negligible.  Full  details  as  to 
recoveries  and  segregation  can  be  found  in  the  report^  on  the 
preparation  of  these  steels. 

The  physical  tests  on  the  Ni-Si-Zr  and  comparison  steels  of 
the  first  series  can  be  found  in  the  rep.ort  of  the  Bureau  of 
Standards.® 

The  sum  total  of  the  tests  by  all  the  co-operating  government 
agencies  led  to  the  conclusion  that  the  Ni-Si  steels  have  good 
mechanical  properties ;  that  these  properties,  measured  by  the 
ordinary  tensile  and  impact  tests,  are  not  materially  injured  by 
the  introduction  of  small  amounts  of  zirconium.  Neither  did 
it  appear  that  the  properties  were  materially  enhanced.  A  steel 
of  0.40  C,  1.45  Si,  0.85  Mn,  3.00  Ni,  rolled  to  one-half  in.  from 
a  3  X  3-in.  ingot,  normalized  from  840°  C.,  gives,  on  0.3-in.  diam- 
eter  by   2-in.   gauge  length   round   specimens,   a   yield   point   of 

5  Gillett,   H.  W.    and   Mack,   E.   L.,   Experimental   Production  of  Alloy   Steels,    P>ull. 
199,   Bur.  of   Mines   1922. 
^Burgess,  G.  K.,  and  Woodward,   R.   W.,  he.   cit. 


238  H.   W.  GILLETT  AND  E.  L.   MACK. 

around  lOO.OCX)  and  a  tensile  strength  of  around  140,000  lb.  per 
sq.  in.,  with  an  elongation  of  say  15  per  cent  and  a  reduction  of 
area  of  say  40  per  cent  with  a  Brinnell  hardness  of  around  270. 
On  quenching  from  840°  C.  and  drawing  3  hours  at  175°  C,  it 
gives  a  yield  point  of  around  240,000  and  a  tensile  strength 
around  280,000  pounds  per  sq.  in.,  with  an  elongation  of  about 
9  per  cent  and  a  reduction  of  area  of  30  per  cent,  a  Brinell  of 
around  550,  and  (on  a  standard  Izod  bar)  around  9  to  12  foot- 
pounds on  the  Izod  test.  The  elongations  would  be  higher  on  a 
standard  0.505-inch  tensile  bar. 

With  0.10-0.40  per  cent  Zr  similar  steels  show  a  tendency 
toward  higher  tensile  strength  and  hardness,  and  lower  ductility 
in  the  normalized  state,  and  approximately  the  same  properties 
with  perhaps  lower  ductility  under  the  heat  treatment  given. 
The  better  Zr  steels  of  this  class  do  not  contain  much  over  0.15 
per  cent  Zr.  The  tests  on  Zr  steels  show  rather  more  variation 
among  steels  of  about  the  same  composition  than  those  on  plain 
Ni-Si  steels.  Since  the  problem  w^as  concerned  with  these  steels, 
and  it  was  necessary  to  use  Si-Zr  alloys  which  introduced  a  good 
deal  of  Si,  nothing  was  done  with  plain  C  low  Si  steels  containing 
Zr.  A  few  high  Si-C  steels  were  made  with  and  without  Zr,  but 
these,  like  the  Ni-Si  steels,  show  no  regular  beneficial  effect  due 
to  Zr. 

A  few  tests  on  the  addition  of  Mo  or  V  to  the  Ni-Si  steels  did 
not  materially  change  the  results  either  on  the  normalized  steel 
or  on  that  given  the  quench  and  low  draw. 

So  far  the  evidence  was  against  any  beneficial  effect  from  Zr 
at  least  in  the  Ni-Si  steels,  but  another  series  was  made  on  which 
each  steel  had  three  or  four  different  treatments,  higher  draw 
temperatures  being  used.  These  steels  were  cast  in  3  x  6-in. 
ingots  and  were  rolled  to  plates  one-quarter  inch  thick,  being 
spread  to  a  little  over  12  in.  wide  by  cross-rolling,  then  straight- 
rolled,  reheated  and  finished  by  straight-rolling. 

For  the  physical  tests,  made  by  the  navy,  tensile  bars  0.5  in. 
wide  by  0.25  in.  thick  by  2  in.  gauge  length,  were  cut  from  the 
plates  with  a  0.06  in.  emery  wheel,  being  finished  by  hand.  The 
bars  were  shouldered  and  held  in  wedge  grips.  Izod  specimens 
were  also  cut,  10  mm.  wide  by  thickness  of  the  plate.  The  notch 
was  cut  by  a  shaper  tool,  being  2  mm.  deep  with  1  mm.  radius 


EXPERIMENTS  WITH  RARE  ELEMENTS  IN  STEEE.  239 

at  the  bottom  (Mesnager  notch).  The  direction  of  impact  was 
parallel  to  the  surfaces  of  the  plate.  The  Izod  values  were  com- 
puted to  standard  square  Izod  bar  size  by  means  of  the  ratio  of 
standard  10  mm.  width  to  the  plate  thickness.  Two  notches  were 
tested  on  each  Izod  bar.  Both  tensile  and  impact  specimens  were 
taken  in  both  longitudinal  and  transverse  directions. 

Steels  of  0.35-0.40  C  and  1.50-2.25  Si  (or  Si  -f  Zr  +  Ti)  gave 
very  uniform  results  between  longitudinal  and  transverse  bars 
on  ductility  and  Izod  tests.  With  higher  carbon  or  silicon  (or 
Si  -|-  Zr  -|-  Ti)  or  both,  the  transverse  bars  generally  fell  below 
the  longitudinal  ones  on  these  tests.  Tensile  strength  and  elastic 
limit  were  of  course  closely  the  same  on  bars  taken  in  either 
direction  on  all  the  steels. 

If  we  assume  that  Zr  or  Ti  are  approximately  equal  to  equiv- 
alent amounts  of  Si,  and  plot  the  properties  of  the  different 
classes,  we  get,  for  the  average  compositions  given  in  Fig.  1,  the 
properties  plotted,  for  bars  taken  longitudinally.  By  comparison 
with  the  data  obtained  by  the  Bureau  of  Standards^"  for  some 
similar  steels  drawn  at  175^  C.  and  from  some  Navy  data  not 
plotted  in  Fig.  1,  it  will  be  found  that  with  draw  temperatures 
below  400°  C,  the  strength  continues  to  increase  while  the 
ductility  remains  about  the  same  as  at  the  400°  draw.  Raising 
the  C  or  Si  too  high  causes  the  ductility  to  increase  with  increas- 
ing draw  temperature  only  slightly,  and  gives  a  dip  in  the  Izod 
curve  with  a  minimum  around  a  525°  C.  draw. 

The  navy  Izod  figures  are  on  bars  with  the  Mesnager  notch 
(1  mm.  radius  at  base)  and  were  taken  on  rectangular  bars  and 
calculated  to  a  10  mm.  square  bar.  The  Bureau  of  Standards' 
Izod  figures  are  on  round  bars,  with  the  one-quarter  mm.  radius 
V  notch,  computed  to  the  standard  round  bar  of  1  sq.  cm.  area. 

The  British  automobile  steel  research  committee,  whose  results 
on  alloy  steels  would  be  interesting  to  compare  with  this  steel, 
used  square  bars  with  one-quarter  mm.  radius  notch.  Conver- 
sion factors,  especially  between  the  two  notches,  are  so  unsatis- 
factory that  no  direct  comparison  can  be  made  of  the  Ni-Si 
steels  and  other  alloy  steels  as  to  impact  results.  Also,  because 
the  Ni-Si  steels  had  to  be  tested  in  flat  bars,  the  ductility  figures 
do  not  compare  exactly  with  data  on  other  steels  from  round  bars. 

•0  Burgess,  G.  K.  and  Woodward,  R.  W.,  he.  cit. 


240 


H.  W.  GILLETT  AND  E.  L.   MACK. 


gSX>,000 

xes;oao 


^  Zoo,ooo 

^    /7S,00O 
J    /SO,OtO 

o 

ft.  7tZS,ooo 


^   Z0O,000 


5> 

^  iTS.Ooo 


/ffTOjOOO 

/25",  000 
40 
30 
zo 
10 


-  o.*7C,z.sos;^Zr*r,,s.oo/^A 

'  0.^1^  C,Z.7S Si i-Zr-t-T,- ,3.00 NiyS^-O",  0/  T 
•■CO'-^O  C,  /.  S-0  S/  ^  3.0*  Ml'  \^«<*-f--tr-eatce<,n 


"0 

ft- 

tf) 

u 

1^ 

(1 

k 

k 

:5 

k 

si  <i 


<i50 


20 


X  3  -^  5  iu 


'Redo 


ction  of  Area 


4<io 


4S0  soo  ^so  600 


Fig.   1. 


EXPERIMENTS  WITH  RARE  ELEMENTS  IN  STEEL.  24 1 

The  Steel  requires  higher  draw  temperatures  to  soften  it  than 
plain  Ni,  or  the  ordinary  Ni-Cr  steels.  It  will  therefore  be  pos- 
sible to  draw  the  steel  at  temperatures  high  enough  to  make 
fairly  certain  that  quenching  stresses  are  released,  and  still  get 
a  high  tensile  strength  combined  with  good  ductility  and  tough- 
ness. The  steels  are  especially  attractive  at  draw  temperatures 
around  400°  C,  for  springs,  and  possibly  for  gears,  or  other 
use,  where  great  strength  or  hardness  combined  with  toughness 
is  desired.  Moreover,  around  this  draw  temperature,  there  is  a 
fair  range  of  compositions  through  which  about  the  same  physical 
properties  are  obtained. 

The  steels  rolled  well  and,  in  the  simple  shape  in  which  they 
were  heat  treated,  did  not  show  quenching  cracks.  We  do  not 
know  what  they  would  do  in  complicated  shapes. 

All  the  Ni-Si  or  Ni-Si-Zr  steels  within  the  limits  of  composi- 
tion tested  gave  good  and  uniform  results  at  the  400°  draw. 
Results  at  draw  temperatures  of  around  500°  C.  were  less 
uniform. 

Up  to  around  0.30  per  cent  Zr  the  effect  of  Zr  seems  to  be 
negligible,  or  at  any  rate  no  more  noticeable  than  the  addition  of 
an  equal  percentage  of  silicon.  Putting  in  0.40  to  0.75  per  cent 
Zr  consistently  decreased  the  toughness  and  such  steels,  as  well 
as  those  with  too  much  C  or  Si  gave  much  more  erratic  results, 
especially  on  transverse  bars.  One  might  expect  that  these  high 
Si  steels  would  tend  to  throw  out  graphite  readily  and  that  the 
impaired  toughness  of  the  higher  C,  high  Si  steels  might  be  due 
to  this  cause.  Microscopic  examination,  however,  has  not  shown 
any  deposition  of  graphite  in  these  steels. 

The  steels  lower  in  C  and  Si  can  just  be  machined  in  the  nor- 
malized condition.  Cooling  in  lime  made  it  possible  to  crop  all 
the  ingots,  of  whatever  composition,  by  sawing,  though  those 
high  in  C  and  Si  would  be  classed  as  steels  difficult  to  machine. 

The  results  of  the  second  series  of  Zr  steels  agreed  with  the 
indications  of  the  first,  to  the  effect  that  the  virtues  of  the  so- 
called  "Zirconium"  (Ni-Si-Zr)  steels  were  due  to  the  combina- 
tion of  Ni^^  and  Si  rather  than  to  Zr.  Zr  probably  has  no  greater 
effect  in  this  type  of  steel  than  so  much  Si.    Zr  leaves  tiny,  sharp- 

'1^  Compare  Hoyt,  S.  L.  Metallography  Part  II,  p.  358,  1921,  for  properties  of  similar 
steels   without  nickel. 

17 


242  H.  W,   GILLETT  AND  E.  L.   MACK. 

cornered  inclusions  in  the  steel,  and  it  is  more  desirable  to  have  a 
steel  free  from  such  inclusions. 

As  to  the  possibilities  of  Zr  as  a  scavenger,  some  very  good 
steels  were  made  by  remelting  crop  ends  containing  Zr,  the 
resulting  steels  coming  out  with  only  a  few  hundredths  of  one 
per  cent  Zr.  On  the  other  hand,  equally  good  plain  Ni-Si  steels, 
into  which  no  Zr  entered,  were  also  made.  Johnson^-  has  studied 
the  Ni-Si  steels  and  has  also  concluded  that  Zr  and  Ti  did  not 
have  any  beneficial  effects.  The  compositions  of  his  steels  are 
given  in  general  terms  only,  so  that  no  real  comparison  with  our 
results  can  be  made. 

CERIUAI   STEELS. 

Cerium  being  in  the  same  group  in  the  periodic  system  as  Zr  and 
Ti,  some  work  on  Ce  was  done  in  connection  with  that  on  Zr.  Since 
the  early  work  showed  that  Ce  had  a  desulfurizing  action,  further 
work  on  this  point  and  on  its  possibilities  as  an  alloying  element 
was  done  in  co-operation  with  the  Welsbach  Co.  Some  "mop- 
ping up"  on  endurance  tests  on  this  problem  is  still  under  way. 

Mix  metal  (Ce,  La,  Nd,  Ph  and  Sa)  was  used  to  introduce  Ce, 
and  the  word  "cerium"  and  all  calculations  involving  percent- 
ages of  "cerium,"  hereinafter  refer  to  the  Ce  group  of  metals 
thus  introduced. 

By  adding  0.50  to  1.0  per  cent  Ce  to  the  steel  just  before  pour- 
ing, we  have  reduced  S  from  0.155  per  cent  to  0.067  per  cent, 
from  0.085  per  cent  to  0.45  per  cent  and  from  0.035  per  cent  to 
0.015  per  cent.  A  strong  SO2  odor  and  the  rising  of  a  reddish 
slag  indicates  that  S  combines  with  Ce  and  rises  to  the  surface, 
where  the  S  burns  out.  When  less  than  0.50  per  cent  Ce  is 
added,  desulfurization  is  slight.  Adding  1  per  cent  Ce  as  soon  as 
the  charge  is  melted  removes  only  a  little  S  and  no  Ce  is  found 
in  the  steel. 

Desulfurization  by  Ce  thus  appears  to  require  the  addition  of 
so  much  Ce  at  the  end  of  the  heat  that  some  will  be  left  in  the 
steel,  and  its  use  would  depend  on  what  the  residual  Ce  does  to 
the  steel.  Somewhere  from  5  to  45  per  cent  of  the  Ce  added  at 
the  end  of  the  heat  is  all  that  is  retained  in  the  steel,  and  if  much 
is  retained  it  segregates  badly.     Such  figures  as  0.60  per  cent  Ce 

^^  Johnson,  C.  M.,  some  alloy  steels  of  high  elastic  limit,  their  heat  treatment  and 
microstructure,  Trans.  Am.  Sec.  for  Steel  Treat.,  2,  501   (1922). 


EXPERIMENTS   WITH   RARE  ELEMENTS   IN    STEEE.  243 

in  the  top  and  0.30  per  cent  in  the  butt  of  a  70-lb.  ingot  are  com- 
mon. If  not  over  0.25  per  cent  Ce  is  retained,  there  is  little 
segregation. 

We  have  not  been  able  to  make  steels  containing  over  0.30  to 
0.40  per  cent  Ce  in  3  x  6-inch  ingots  of  75  to  100  lb.  without 
having  the  ingots  unsound  through  the  formation,  at  least  in  the 
top  of  the  ingot,  which  freezes  last,  and  often  clear  to  the  butt, 
of  very  tiny  hair  cracks  not  visible  without  smoothly  machining 
the  cross  section  of  the  ingot.  Microscopic  examination  shows 
that  there  are  literally  myriads  of  tiny  inclusions  in  a  Ce  steel, 
and  that  if  there  is  enough  Ce  present  and  enough  time  given  for 
it  to  act,  these  inclusions  tend  to  coalesce  and  rise.  If  enough 
large  coalesced  inclusions  are  present  to  be  collected  between  the 
crystals  as  the  steel  freezes,  they  cause  these  inter-crystalline  hair 
cracks.  Possibly  if  enough  time  could  be  given,  all  the  inclusions 
would  coalesce  and  rise,  but,  working  with  not  over  100  lb.  of 
steel,  it  could  not  be  held  long  enough  in  the  ladle. 

The  inclusions,  under  high  magnification,  are  grayish,  some- 
times mottled  with  orange.  They  are  roundish  in  the  ingot.  On 
rolling  or  forging  the  steel  the  inclusions  smash  up  a  trifle  so  as 
to  have  more  irregular  outlines,  but  are  still  more  or  less  roundish. 
They  do  not  draw  into  hair-like  or  knifeblade-like  forms  during 
the  rolling  of  rods  and  plates  as  manganese  sulfide  does.  Inci- 
dentally, the  woody  fracture  of  a  transverse  specimen  of  a  rolled 
plate  that  contains  ordinary  inclusions  may  be  due  to  the  fact  that 
the  inclusions  roll  out  too,  for  one  such  plate  which  is  very  dirty 
from  inclusions  of  Ce  does  not  show  a  woody  fracture,  while  com- 
panion plates,  free  from  Ce  and  immeasurably  freer  from  inclu- 
sions, all  showed  woody  transverse  fractures.  Rolling  or  forg- 
ing, while  it  does  not  flatten  or  draw  out  the  individual  inclusions, 
often  spreads  the  shattered  inclusions  in  well-defined  lines,  so 
that  rods  of  Ce  steel  are  often  seamy  and  plates  laminated. 

It  is  probable  that  the  S  of  the  Ce  steel  is  held  in  these  inclu- 
sions not  as  manganese  sulfide,  as  in  ordinary  steel,  for  the  in- 
clusions are  larger  and  in  greater  mass  though  not  usually  in 
greater  numbers,  in  the  top  of  the  ingot,  and  the  S  is  often  de- 
cidedly higher  in  the  top  of  the  ingot  than  in  the  butt,  the  ratio 
of  segregation  being  usually  higher  than  that  of  the  Ce  present. 


244  H.   W.   GILLETT   AND  E.   L.   MACK. 

Starting  with  material  of  0.03  per  cent  S,  it  was  rare  that  the 
butt  of  an  ingot  of  Ce  steel  would  run  over  0.01  per  cent  S. 

On  the  other  hand,  if  all  the  inclusions  contain  S,  the  per- 
centage of  S  in  them  must  be  very  low.  It  could  not  be  present 
as  any  orthodox  cerium  sulfide  and  account  for  the  great  mass 
of  inclusions.  The  inclusions  are  probably  of  complex  composi- 
tion, and  one  would  naturally  suppose  them  to  be  mostly  oxides. 
That  the  S  is  combined  with  the  Ce  is  probable  from  the  be- 
havior of  a  couple  of  steels  in  which  the  Mn  was  kept  as  low 
as  possible  (and  in  one  case  extra  S  added)  and  the  steel  treated 
with  1  per  cent  Ce.  One  of  these,  a  0.30  per  cent  C  steel,  was 
made  up  for  0.069  per  cent  S.  It  came  out  with  0.032  per  cent 
S,  0.06  per  cent  Mn  and  with  0.05  per  cent  Ce  left  out  of  1.10 
per  cent  Ce  added.  Theoretically,  0.055  per  cent  Mn  is  re- 
quired to  combine  with  the  S  present.  Practically,  on  account 
of  mass  action,  it  is  generally  considered  that  much  more  Mn 
would  be  required  to  prevent  the  presence  of  FeS,  so  that  such 
a  steel  would  be  expected  to  be  red-short.  However,  the  steel 
forged  nicely.  Physical  tests  show  no  difference  from  any  steel 
of  that  general  composition.  No  Al  or  other  special  deoxidizer 
was  used,  but  the  Si  was  raised  to  0.65  per  cent  to  compensate, 
in  killing  the  steel,  for  the  low  Mn.  Some  tests  have  been  made 
using  0.01  to  0.03  per  cent  Ce  as  final  deoxidizer  which  pro- 
duced dead,  de-gasified  steel  on  high  silicon  steels,  but  0.06  per 
cent  Ce  failed  to  kill  a  low  Si  steel.  Ce  is  not  as  strong  a  deoxi- 
dizer and  de-gasifier  as  aluminum. 

Part  of  the  Ce,  or  at  least  of  some  one  or  more  of  the  Ce  group, 
of  metals,  is  probably  present  as  carbide,  for  the  steels  con- 
taining more  than  something  between  0.10  and  0.20  per  cent  Ce 
have  a  decided  acetylene-like  odor.  Steels  high  in  Ce  give  out 
a  strong  odor  on  machining.  All  one  has  to  do  to  pick  out 
such  a  Ce  steel  from  among  others  is  to  rub  it  with  emery  paper 
or  even  with  a  rubber  eraser,  so  as  to  remove  some  invisible 
film,  and  the  characteristic  odor  will  be  easily  detected. 

With  the  low  and  irregular  recovery  of  Ce,  irregular  desulfuri- 
zation,  the  prevalence  of  cracked  ingots  and  the  great  number 
of  inclusions,  it  is  difficult  to  get  steels  by  which  one  can  de- 
termine the   real  alloying  effect,   if  any,  of   Ce.     In   the  Ni-Si 


EXPERIMENTS  WITH  RARE  ELEMENTS  IN  STEEL.  245 

and  in  Ni-Cr  steels,  there  is  some  evidence  that  0.20  per  cent 
Ce  increases  the  propensity  toward  air-hardening,  i.  e.,  that  it 
acts  as  a  true  alloying  element.  But  the  ever  present  inclu- 
sions so  complicate  matters  by  reducing  ductility  that  we  are 
unable  to  state  what  would  be  the  properties  of  a  steel  contain- 
ing Ce  as  alloying  element  and  none  as  non-metallic  inclusions. 

Tests  have  been  made  on  plain  Ce,  Cr-Ce,  Ni-Si-Ce,  and  Ni- 
Cr-Ce  steels.  A  0.45-per  cent  C,  1.30-per  cent  Si,  2.95-per  cent 
Ni  steel  tested  by  the  Bureau  of  Standards  gave  (oil  quenched 
and  drawn  at  175°  C.)  311,000  Ib./sq.  in.  tensile  and  Z7  per  cent 
reduction  of  area  at  a  Brinnell  of  555.  On  this  0.02  per  cent  Ce 
was  used  as  final  deoxidizer.  Other  steels  containing  more  Ce 
showed  similar  strength  but  lower  ductility.  A  couple  of  these 
showed  very  good  impact  tests  and  hence  all  the  Ce  steels  made 
since  have  been  given  the  Izod  test  on  samples  drawn  at  low 
temperatures.  However,  none  of  these  other  Ce  steels  have 
shown  any  exceptional  impact  results. 

Generally  speaking,  the  forged  or  rolled  Ce,  Cr-Ce  and  Ni- 
Cr-Ce  steels,  containing  0.20  to  0.50  per  cent  Ce  quenched  and 
tempered,  are  practically  indistinguishable  on  tensile,  impact 
or  repeated  impact  tests  from  similar  steels  without  Ce,  when 
test  bars  taken  longitudinally  are  considered.  Transverse  test 
bars  from  plates  fall  down  on  ductility,  doubtless  due  to  the 
inclusions. 

When  we  come  to  the  "fatigue"  test,  endurance  against  re- 
peated bending,  the  Ce  steels  regularly  fall  down  in  comparison 
with  similar  steels  free  from  Ce,  and  this  is  the  more  noticeable 
the  harder  the  steel.  This  test  is  probably  more  sensitive  to 
the  presence  of  inclusions  than  any  of  the  other  mechanical  tests, 
and  the  poor  behavior  of  the  Ce  steels  is  obviously  due  to  the 
inclusions.  In  fact,  if  one  looks  at  a  micrograph  of  a  Ce  steel, 
taken  from  a  polished  but  unetched  section  to  show  inclusions 
(see  Fig.  2-5)  he  wonders  why  such  dirty  steel  does  not  give 
poorer  results  on  all  tests  than  it  does.  The  endurance  tests 
show  that,  due  to  the  inclusions,  the  cerium  steels,  especially 
when  treated  to  a  high  hardness,  are  highly  unreliable  against 
repeated  bending. 

It  is  possible  that  some  means  might  be  found  to  control  and 


246  H.  W.   GILLETT  AND  E.  L.   MACK. 


>• 


Img     2.    Unetched    Cross    Section    of  Fig.    3.    Unetched   Longitudinal    Sec- 

Cerium   Steel.     X  100.  tion   Cerium   Steel,     x    100. 


[ 

' 

% 

I 

^  • 

♦ 

•  f» 

.■ 

• 

• 

( 

• 

• 

* 

1 

• 

« 

\ 

• 

• 
• 

* 

L: 

^ 

V 


Fig.      4.    Unetclied      Cross      Section  Fic.      S.    Unetched      Cross      Section 

Cliromium-cerium    Steel,     x  100.  Chromium-cerium    Steel,     x  100. 


e;xpe;riments  with  rare  elements  in  steel.  247 

utilize  the  desulfurizing  action  of  Ce  without  doing  more  harm 
than  good  by  the  retaining  of  inclusions.  We  came  close  to  this 
with  a  couple  of  steels  low  in  manganese,  but  would  be  unwilling 
to  attempt  it  on  a  production  basis. 

Cerium  has  been  claimed  to  increase  the  ductility  of  aluminum 
alloys,  and  we  tried'"  it  out  in  various  light  aluminum  casting 
alloys,  including  the  duraluminum  type,  both  as  cast  and  heat- 
treated,  but  were  unable  to  find  any  improvement  of  any  sort 
due  to  cerium. 

MOLYBDENUM 

After  working  with  U,  Zr  and  Ce,  which  give  low  and  vari- 
able recoveries,  leave  inclusions  in  the  steel,  segregate  and  either 
have  little  effect  on  or  do  actual  harm  to  the  steel,  it  is  a  relief 
to  work  with  an  alloying  element  that  enters  the  steel  without 
loss,  does  not  segregate,  and  has  a  positive  and  very  beneficial 
effect  on  the  steel. 

Most  of  the  elements  above  mentioned  are  not  available  in 
this  country  in  large  quantities.  Mo,  on  the  other  hand,  is  an 
element  of  which  the  United  States  has  an  an  ample  supply,  and 
a  complete  understanding  of  just  what  it  does  in  steel  is  import- 
ant in  the  development  of  our  Mo  resources. 

Present-day  alloy  steels  for  such  uses  as  automobiles  and  air- 
craft require  one  or  more  of  the  following  elements — Mn,  Si, 
Ni,  Cr,  V.  The  first  two  are  not  usually  classed  as  alloying  ele- 
ments, since  they  are  present  in  all  steel,  but  in  amounts  above 
the  normal,  they  do  exert  an  influence  that  justifies  classing  them 
as  alloy  elements. 

The  great  bulk  of  the  Ni  used  is  mined  in  Canada.  The  V 
comes  from  Peru.  Though  we  can,  at  a  pinch,  supply  some  Cr 
and  Mn,  the  Cr  usually  comes  from  New  Caledonia,  Rhodesia 
and  Asia  Minor,  and  the  Mn  from  Brazil  and  Caucasia.  Fe, 
C  and  Si  are  domestic. 

The  development  of  a  home  supply  of  an  alloying  element 
which  can,  in  whole  or  in  part,  replace  or  supplement  the  ele- 
ments of  foreign  origin,  is  of  obvious  importance.  One  de- 
posit in  Colorado  contains  enough  Mo  to  make  some  20  million 

'3  Gillett,  H.  W.,  and  Sclinee,  V.  H.,  Cerium  in  Aluminum  Alloys,  soon  to  be  pub- 
lished in  Ind.  Eng.  Chem. 


248  H.   W.   GILLETT  AND  E.  L.   MACK. 

tons  of  Mo  steel  of  the  usual  Mo  content."  The  Bureau  of 
IMines  is  therefore  interested  in  Mo  steel  both  from  the  point 
of  view  of  preparedness  and  avoidance  of  dependence  on  for- 
eign raw  materials  and  from  that  of  development  of  a  national 
mineral  resource  that  is  as  yet  so  little  used  as  to  justify  includ- 
ing Mo  among  the  rarer  elements  despite  the  size  of  the  deposits. 

Much  work^^  has  recently  been  done  on  Mo  steels,  almost  all, 
save  the  pioneer  work  of  Swinden  that  has  been  done  by  ob- 
servers not  primarily  interested  in  the  sale  of  Mo,  having  been 
published  in  the  last  two  or  three  years.  When  the  Bureau 
started  its  work  on  Mo  steels,  very  little  data  that  was  definitely 
free  from  possible  bias  was  available.  Now,  however,  there  is 
such  a  mass  of  well  agreeing  data  that  there  is  little  chance  for 
argument  as  to  the  facts. 

The  Bureau's  objective  in  the  work  on  Mo  was,  incidentally,  to 
check  up  the  outside  data  on  ordinary  tests,  but  primarily  to  get 
some  idea  of  how  far  Mo  can  replace  other  alloying  elements,  and 
to  secure  data  on  the  resistance  of  Mo  steel  to  shock  and  fatigue, 
points  of  increasing  importance  to  the  engineer  and  on  which 
strong  claims  of  excellence  have  been  made  by  advocates  of  Mo. 

For  two  years  the  work  was  carried  on  under  a  co-operative 
agreement  with  the  Vanadium  Corporation  of  America,  producers 

■*  Compare  Moore,  R.  B.,  Molybdenum.  Political  and  Commercial  Control  of  the 
Mineral  Resources  of  the  World,  Bur.  Mines  Mimeographed  Report  No.  S  of  War 
Minerals  Investigations  Series,  August  25,   1918. 

15  Bullens,  D.   K.,   Steel  and  its  Heat-Treatment  354   (1916). 

Swinden,  T.  Carbon  Molydenum  Steels,  Jour.  Iron  Steel  Inst.,  Carn.  Sch.  Mem., 
3,  66  (1911). 

A  study  of  the  constitution  of  C-Mo  Steels,  Carn.  Sch.   Mem.  5,   100   (1913). 

Wills,  C.  H.,  U.  S.  Pats.  1,278,082;  1,288,345,  Canadian  Pat.  192,341  of  Aug. 
26,  1919,  British  Pat.  150,343  of  Aug.  24,   1920. 

Sargent,  G.  W.,  The  Value  of  Mo  Alloy  Steels,  Trans.  Am.  Soc.  Steel  Treat.,  1, 
589   (1921). 

Cutter,  J.  D.,  Suggested  Methods  for  Determining  Comparative  Efficiency  of  Certain 
Combinations  of  Alloy  Steels,  Trans.  Am.   Soc.  Steel  Treat,  1,  188   (1920). 

McKnight,  C,  Jr.,  A  Discussion  of  Mo  Steels,  Trans.  Am.  Soc.  Steel  Treat,  I,  288 
(1921). 

Schmid,  H.  M.,  Mo  Steel  and  its  Applications,  Trans.  Am.  Soc.  Steel  Treat.,  1,  300 
(1921);  Chem.  and  Met.  Eng.  24,  927   (1921);  Iron  Age,  107,   1,444   (1921). 

Hunter,  A.  H.,  Manufacture  and  Properties  of  Mo  Steel,  Iron  Age,  107,  1,469 
(1921);  Chem.  and  Met.  Eng.  25,  21    (1921).  .         „         ^  „ 

Anon,  Heat  Treated  Castings  of  Cr  Mo  Steel,  Trans.  Am.  Soc.  Steel  Treat.  1, 
588   (1921);   Iron  Age.,    107,   1,052   (1921).  .     ,    „  .  ,         ^„ 

French  H.  J  ,  Effect  of  Heat  Treatment  on  Mechanical  Properties  of  a  C-Mo 
and  a  Cr-Mo  Steel,  Trans.  Am.   Soc.   Steel  Treat.,  2,   769  (1922). 

Dawe,  C.  N.,  Cr-Mo  Steel  Applications  frorn  the  Consumer's  Point  of  view, 
Soc.  Automotive  Eng.,  Annual   Meeting  Jan.,   1922.  .,    ,,,   ,,„-,ox 

Mathews,  J.  A.,  Mo  Steels,  Trans.  Am.  Inst.  Min.  and  Met.  Eng.,  47,  137  (1922), 
Iron   Age,   107,   505    (1921).  ^  .  ,   ^      ,      

Climax  Molybdenum  Co..  Booklet — Molybdenum  Commercial  Steels,   1919. 

Crucible  Steel  Co.,   Booklet— Al-Mo   Steels,  1919.  o.     ,   -,.      . 

Vanick,  J.  S.,  Properties  of  Cr-Mo  and  Cr-V  Steels,  Trans.  Am.  Soc.  Steel  Treat, 
3,  252   (1922). 


EXPERIMENTS  WITH  RARE  ELEMENTS  IN  STEEL.  249 

of  Mo.  For  the  last  six  months  it  has  been  carried  on  by  the 
Bureau  alone,  and  it  will  take  at  least  six  months  more  to  com- 
plete the  time-consuming  endurance  tests  which  are  the  central 
point  of  the  investigation. 

In  the  preparation  of  Mo  steels,  by  adding  ferromolybdenum 
to  the  charge  at  the  beginning  of  the  heat,  the  recovery  was 
found  to  be  quantitative  and  no  segregation  was  found. 

The  most  important  property  of  Mo  in  steel  is  the  control 
it  gives  of  the  development  by  heat-treatment  of  the  properties 
desired.  Comparing  C  steels  with  alloy  steels  as  a  class,  the 
carbon  steels  are  not  so  readily  brought  over  by  quenching,  to 
the  metastable,  hardened  state.  Carbon  steels  require  drastic 
quenching  with  its  attendant  stresses  and  dangers,  they  do  not 
harden  throughout  in  large  pieces,  and  if  heated  too  high  before 
quenching  in  order  to  increase  the  hardening  effect,  they  de- 
teriorate on  account  of  excessive  grain  growth.  The  introduc- 
tion of  alloying  elements,  such  as  Mn  or  Ni,  alone  or  in  com- 
bination with  Cr  or  V,  makes  the  steel  much  more  readily  hard- 
ened, even  in  large  pieces,  and  the  evil  effect  of  over-heating 
diminishes.  The  introduction  of  a  sufficient  quantity  of  the 
proper  alloys  makes  the  steel  so  sluggish  that  even  cooling  in 
the  air  produces  a  self-hardening  or  air-hardening  steel.  By 
proper  adjustment  of  the  alloy  content  and  the  carbon  content, 
any  gradation  between  a  C  steel  that  will  not  harden  at  all  and 
an  air-hardening  steel  can  be  made. 

The  best  classification  of  steels,  which  was  made  by  Aitchison'* 
in  his  excellent  book,  is  on  the  basis  of  the  properties  that  can 
be  developed  in  them,  or,  what  is  almost  the  same  thing,  their 
relative  propensity  toward  hardening,  rather  than  on  chemical 
composition. 

From  this  point  of  view.  Mo  is — C  excepted— the  most  active 
and  potent  element  used  in  steel.  The  propensity  toward  harden- 
ing can  be  shown  by  varying  the  rate  of  cooling  or  by  varying 
the  maximum  temperature  to  which  the  steel  is  heated,  and  cool- 
ing at  a  constant  rate,  since  raising  the  initial  temperature  aids  in 
suppressing  the  stable  change  and  producing  undercooling  or 
hardening,  much  as  increasing  the  cooling  rate  does.     In  either 

"  Aitckison,  L.,   Engineering   Steels,  Van  Nostrand,    1921. 


250 


H.   W.   GILLETT  AND  E.   L.   MACK. 


S33iiD3(r 


EXPERIMENTS  WITH  RARE  ELEMENTS  IN  STEEL.  25 1 

method  the  lowering  and  "spHtting"  of  the  critical  points  on  cool- 
ing will  show  the  propensity  toward  hardening. 

Fig.  6  shows  differential  heating  and  cooling  curves  of  steels 
of  about  0.40  per  cent  C,  1.25  per  cent  Ni,  0.70  per  cent  Cr. 
No.  15  has  higher  C,  but  the  comparison  of  the  steels  is  not 
altered  thereby. 

On  either  the  plain  Ni-Cr  or  the  Ni-Cr-V  steel,  the  critical 
point  on  cooling  at  the  slow  rate  used  (about  75  min.  to  cool  to 
300°  C.)  is  not  appreciably  altered  (slightly  lowered)  by  raising 
the  maximum  temperature  from  775°  to  900°  C.  But  with  the 
addition  of  0.31  per  cent  AIo,  cooling  at  the  same  rate  from 
770°  C,  gives  a  weaker  critical  point  at  the  normal  temperature, 
and  a  new  weak  one  starting  about  525°  C.  and  with  a  maximum 
at  450°  C.  At  the  upper  point  the  austenite  goes  over  to  primary 
troostite,  which  at  the  ordinary  rate  of  cooling  immediately  goes 
over  to  pearlite ;  but  some  austenite  is  retained  unchanged,  which 
at  the  lower  critical  point  goes  over  into  martensite.  This  mar- 
tensite  is  not  stable  at  the  temperature  at  which  it  is  formed, 
and  in  turn  goes  over  to  secondary  troostite  or  sorbite  on  slow 
cooling. 

As  the  maximum  temperature  is  raised,  the  upper  critical 
point  becomes  slightly  lowered  and  progressively  weaker,  and 
the  lower  point  becomes  stronger,  till  at  900°  C.  maximum  tem- 
perature the  upper  point  is  wholly  wiped  out  and  the  steel  shows 
only  the  low  critical  point  corresponding  to  the  formation  of 
martensite,  i.  e.,  is  wholly  air-hardening.  The  propensity  toward 
hardening  is  so  great  that  many  Mo  steels  will  harden  throughout 
on  oil-quenching  in  sizes  which  would  not  harden  at  the  center 
on  water-quenching  without  the  Mo. 

This  same  effect  is  shown  by  Mo  in  all  combinations.  If  we 
leave  out  the  Ni  and  Cr  and  raise  the  Mo  to  say  0.75  per  cent,  we 
get  a  similar  family  of  cooling  curves.  As  we  raise  the  C,  the 
Mn  or  the  Ni,  it  takes  less  Mo  to  shift  the  steel  from  the 
behavior  of  a  plain  C  steel  on  cooling  toward  that  of  an  air- 
hardening  steel.-  With  combinations  of  Cr  and  ]\Io  the  effect 
of  Mo  is  not  quite  so  marked,  but  it  is  still  evident. 

Whatever  the  composition  of  the  steel  in  which  it  is  used,  the 
presence  of  Mo  tends  to  make  the  steel  require  less  drastic 
quenching  and  to  make  it  harden  to  a  greater  depth  on  a  given 


252  H.  W.   GILLETT  AND  E.   L.    MACK. 

quench.  This  slowing  up  of  the  transformation  gives  much 
better  control  over  the  hardening  operation,  and  this  control  is 
given  in  good  degree  by  quite  small  percentages  of  ]\Io,  0.20 
per  cent  Mo  having  a  definite  effect.  If  the  steel  is  to  be  used  in 
the  normalized  condition,  the  air-hardening  properties  may  be  a 
decided  drawback.  Even  if  the  steel  does  not  become  difficult  to 
machine,  the  normalized  steel  is  not  very  good  as  to  elastic 
limit,  ductility  or  single-blow  notched-bar  test.  When  the  Mo 
is  over  say  0.40  per  cent  and  normalized,  plain  V  or  Cr-V  steels 
appear  more  desirable  than  normalized  Mo  or  Cr-Mo  steels, 
unless  tensile  strength  is  the  prime  aim.  Molybdenum  steels 
•should  be  used  in  the  heat-treated  condition  to  secure  the  maxi- 
mum beneficial  eftect. 

Vanadium  does  not,  in  itself,  produce  a  strong  tendency  toward 
air-hardening.  However,  the  greater  hardening  due  to  quench- 
ing from  a  higher  temperature  can  safely  be  made  use  of  in  V 
steels,  because  of  the  marked  ability  of  V  to  inhibit  grain  growth 
of  austenite  at  temperatures  that  would  give  fatally  coarse  grain 
in  C  steel.  j\Iost  alloying  elements  have  this  property  in  some 
degree,  and  Mo  shows  it  strongly,  though  the  efifect  is  probably 
not  quite  so  great  as  with  V. 

Vanadium  steels  generally  show  a  higher  elastic  ratio  than 
other  alloy  steels.  In  this  respect,  ]\Io  has  very  nearly  the  same 
effect  as  V.  The  individual  good  properties  of  both  Mo  and  V 
are  still  in  evidence  when  other  alloying  agents  are  present,  and 
they  may  both  be  used  together.  For  example,  the  addition  of 
V  to  a  Cr-Mo  steel  produces  a  steel  of  remarkable  toughness.^^ 

The  effect  of  Mo  is  shown  not  only  in  the  behavior  on  quench- 
ing, but  in  tempering  also.  A  hardened  Mo  steel  does  not  soften 
on  tempering  with  the  facility  of  a  similar  steel  without  Mo.  To 
bring  the  steel  to  a  given  strength  or  hardness,  it  has  to  be 
tempered  at  a  higher  temperature  or  for  a  longer  time  than  one 
without  Mo.  Steels  high  in  Mo,  especially  in  the  presence  of  a 
good  deal  of  other  alloying  elements  at  first  change  in  properties 
at  a  slow  rate  with  increasing  draw  temperature,  and  more 
rapidly  at  very  high  draw  temperatures  so  that  at  the  highest 
draws  it  may  require  very  accurate  temperature  control  to  get 

"  See  Sargent,  G.  W.  he  cil.  p.  596. 

Crucible  Steel  Co.  Booklet.  Al-Mo  Steels.  39    (1919). 


EXPERIMENTS  WITH  RARE  ELEMENTS  IN   STEEL.  253 

the  same  results  on  consecutive  draw  heats.  But,  if  the  Mo 
content  is  not  too  high,  so  that  the  properties  desired  are  obtained 
in  the  range  of  draw  temperatures  through  which  the  properties 
change  slowly  (and  this  is  the  case  with  most  commercial  Mo 
steels)  less  accurate  control  is  required  and  consecutive  heats 
produce  more  uniform  results  than  with  most  other  alloy  steels. 

The  resistance  to  tempering  shown  by  Mo  steels  holds  promise 
for  these  steels  for  use  at  temperatures  above  normal,  but  no 
extended  study  of  their  properties  at  higher  temperatures  seems 
to  have  been  made.  Another  advantage  in  the  sluggish  nature 
of  a  Mo  steel  is  that  the  higher  draw  temperature  for  a  given 
hardness  or  strength  means  a  better  release  of  quenching 
stresses. 

Table  II  gives  some  of  the  data  secured  on  heat-treated  r^Io  and 
comparison  steels.  These  figures  are  all  on  specimens  heat 
treated  in  1  inch  diameter  or  less.  Were  larger  specimens  used, 
say  3-inch  diameter,  the  depth-hardening  properties  of  Mo  would 
give  the  Mo  steels  an  advantage  that  is  not  shown  by  the  table. 

A  cursory  examination  of  the  table  will  show  that  Mo  has  a 
real  strengthening  effect  and  that  the  heat-treated  Mo  steels 
combine  good  strength  with  good  ductility  and  toughness. 

With  2.5  per  cent  Ni,  0.8  per  cent  Cr  and  0.75  per  cent  jMo, 
one  can  get  results  of  the  same  general  order  as  with  3.5  per  cent 
Ni,  1.5  per  cent  Cr.  Cr-V  and  Cr-Mo  steels  each  have  nearly  the 
same  properties.  On  the  other  hand,  i\Io  finds  its  chief  use  as  an 
addition  together  with  Cr,  Ni,  or  Ni-Cr.  Quite  good  alloy  steels, 
decidedly  ahead  of  plain  C  steels,  can  be  made  with  up  to  1  per 
tent  Mo,  especially  by  increasing  simultaneously  the  Mn  per  cent. 

From  the  results  of  a  couple  of  Ni-Si  plus  Mo  steels,  given 
the  175°  draw,  the  Bureau  of  Standards  concludes^^  that  these 
steels  would  be  superior  with  the  Mo  omitted,  while  Johnson^^ 
secured  good  results  in  such  steels  with  Mo,  but  says  that  it  is 
not  certain  that  the  improved  showing  was  due  to  the  Mo.  In 
the  second  series  of  Ni-Si  steels  made  for  the  navy,  one  was 
included  which  contained  Mo.  This  showed  the  normal  effect  of 
Mo  in  increasing  the  strength  at  high  draw  temperatures  without, 
however,  appreciably  altering  the  ductility.    This  steel  had  rather 

•'  Burgess,  G.  K.,  and  Woodward,  R.  W.,  lec.  cit.  p    133. 
'8  Johnson,   C.   M.,   loc.   cit.   p.    501. 


254 


H.  W.  GILLETT  AND  E.  L.   MACK. 


""      fa 

Stanton 

test 

No. 

blows 

gsss    S S ^ S S S o S S S s ^ ^ S 5 S S So oSSSS 

00  \C  i-O  c        OC  tT  t^  r^  U-,  u".  O  ""-  ^  ""•'-'"•  lt;  O  O  ^C;  ""-  ^  LO  ^  t^  \C  lO  'I-  "^ 

Reduc- 
tion of 

Area 
at 

neck 

Elonga- 
tion 

per  cent 
in 
2  in. 

Ulti- 
mate 
tensile 
in 

lono  lb. 

sq.in. 

•>!i-r>.'oC"^coc— r^ir;(\ioc\coc<r;Lou~-r>.u^ir;oocodcvJo^<Nioco 

O.  0Cr^C>^CNI^CCO\C3Cu--\Ct^C^O"~.  (^-vCtv.OCVCvOOCvOOOCO'-t^ 

i-H.— !!-<.— ■CM'—',— .^.,—,.—  r—   T—   ^-(M— '   —   •—--■  —  -^'— '—'>—'■— '^-'—   Ojr^l»—< 

Yield 

point* 

in 

lOon  lb. 

sq.in. 

r^  t^  O  X  OC  C\  LT,  u-,  r^  ir;  r^  TT  LT,  ^Cm  (^  C^l  ■'I-  vC  vo  -^  \C  t^  LO  f^  "".  OC  CMi^ 

BrinncU 

u-,iricroo'J^c;c:CCCCLaoopc:mmir:iir)u-)Co<J^tJ~-oio>J2 

a 
C 

c 

ffiKS'^'oo  o  5  5  o  5  o  o'o  oj£;i;  5  o'o"o'o"5'o  o  5"o"o'o 

Quen- 
ched 
from 
°C 

CCOCirjCOlOu-linOoOQQ'^Oi'ii^i^i'i^TJTJ^OOOOC^ 

> 

O O TT      .      .      •      -O 

r^ CM CM      .      .      •      -CM 

o o o    •    •    •    'o 

o 

•c5o^cM«^    '    *oo    "ooo    *  coo  o'o      CJO    *    'OOO    ' 

oxcco'CM^t^o.  U-,  CNVCvc^r^occcr^oooo 

■ o  o  d  o  o  C  C:  c  o  o  c;  o  o>  c  o  o  o  c>  o 

2 

f<^  00  rx  O  \0  m  0\  On  CM 

rt^^^CM'^iCMCMCM 

n 

^^^(Mu-)t^lr5-*lOC^•^CM^i^<^CM'*^CC^00lO-HT}•\0■*-Hli-i,-com^ 

rN.rv.r-^.tv.\oooqocMCM^\o\co^.  <^'-o'^u-5t^o\o^t>>o^oovD\o 
<Ddc>dddcJ-^"»-«'.-H'dcio'o'ddo'cddco'dc;ddc)dd 

w 

CM^CMf^OOOCrO\CiOCMOOO-^rM'*C\Cv'OCt^r-iOOrr>CM.— -^l-OO 
CMfO'^'^'^'^'-'^OrOTfco— iCMTl-ro-H^  — —  -^^OCMrtCMf^rOcO'tj-CM 

d  o'  c;  c  c>  CD  d  c  d  d  c>  d  d  d  c"  o'  d  ©  d  o'  d  d  d  d  d  c  ©  c5  d 

U 

TtrO-^Trr^r^-LO-^-^-^r^TTTr'^'^  —  CMCM'^<J^'^<^'3-rO<^'^'^LOrt 

d 

.-Hrr)(M^(M'*>OvOOOr«.aNiO«-iOCMI^\Oa^OOO>Dt>.vOOOrHr<^CMvCrt 

-l-'3--*'0                                  '*l-li-<,-(-^-i--^-^u-;»-lr-t^r-lCMCMCMCM(M 

EXPERIMENTS  WITH  RARE  ELEMENTS  IX  STEEL. 


255 


Table  II. — Continued. 
Physical  Data  for  Alloyed  Steels.     Normalised. 


*  Steel  No.  44  cooled  in  furnace  with  door  open,     t  No  test. 


too  much  Si  to  show  best  results  from  the  addition  of  another 
hardening  element.  .    . 

Some  fairly  good  :^Io  high  speed  steels  have  been  made,  and  it  is 
possible  that  they  will  be  tamed  and  used  some  day.  So  far,  the 
results  are  too  erratic,  and  AIo  does  not  appear  to  be  a  serious 
competitor  of  W  for  this  purpose. 

On  the  single-blow  notched  bar  Izod  tests  (on  the  standard 
round  bar,  V  notch  0.25  mm.  radius  at  base)  and  on  Stanton 
repeated  impact  tests  (5-lb.  hammer,  2-in.  fall,  square  notch,  O.Od 
in.  wide,  0.05  deep,  0.01  in.  radius  at  corners)  the  ]Mo  steels  show 
properties  similar  to  other  alloy  steels  of  their  respective  classes, 
when  due  consideration  is  given  to  the  effect  of  tensile  strength, 
C  content  and  other  variables  outside  of  ^lo  in  these  tests.  (Our 
thanks  are  due  to  Mr.  J.  H.  Nelson,  of  the  Wyman  Gordon  Co., 
for  making  the  Stanton  and  Izod  tests). 

The  repeated  bending  endurance  tests,  made  on  Upton-Lewis 
machines  (kindly  made  available  for  our  use  by  Sibley  College, 
Cornell  University)  are  still  incomplete,  although  a  large  num- 
ber have  been  made.  It  has  been  claimed  in  technical  and  adver- 
tising literature  that  Mo  steel  is  in  a  class  by  itself  as  to  endurance 
against  repeated  or  vibratory  stresses.     Similar  claims  have  been 


256  H.   \V.   GILLETT  AND  E.  L.   MACK. 

made  for  a  long  time  for  V  steel,  and  no  text-book  discussion 
of  V  is  complete  without  the  statement  that  V  in  steel  enables 
steel  to  withstand  vibratory  stress.  And  in  practice,  it  is  found 
that  both  these  steels  do  give  good  service  under  conditions  of 
severe  repeated  stress. 

The  claims  for  Mo  steel  against  fatigue  were  largely  based 
on  the  evident  asumption  that  since  Mo  steel  has  a  high 
ductility  for  a  given  tensile  strength,  the  toughness  must  give 
good  endurance. 

But  the  work  of  Moore-"  and  McAdam-^  has  shown  that  the 
resistance  to  repeated  bending,  at  least  in  sorbitic  and  pearlitic 
steels,  is  strictly  proportional  to  the  tensile  strength,  while  the 
elastic  limit,  yield-point  and  ductility  have  no  direct  relationship 
to  endurance.  Published  data  of  endurance  tests  on  alloy  steels 
are  incomplete,  but  there  are  indications  that  any  sound,  clean 
steel  is  equivalent  in  endurance  of  any  other  sound,  clean  steel 
of  the  same  tensile  strength. 

Our  work  indicates  that  this  relation  between  tensile  strength 
and  endurance  limit  is  a  good  first  approximation,  at  least  for  a 
pretty  wide  range  of  alloy  steels.  There  are  often  greater  devia- 
tions from  this  relationship  in  sister  bars  of  the  same  steel  than 
among  different  steels  of  widely  different  types.  Slight  dift'er- 
ences  in  surface  finish  or  in  the  cleanliness  of  the  steel  make 
great  differences  in  the  results.  The  only  place  that  we  can  see 
any  effect  of  ductility  is  in  very  hard  steels  as  compared  to  softer 
steels,  the  former  being  more  sensitive  to  surface  scratches  or 
non-metallic  inclusions. 

We  are  led  to  believe  from  our  work  that  the  cleanliness  of  a 
steel,  its  uniformity  of  composition  and  structure,  and  its  freedom 
from  internal  stress  have  far  more  eft'ect  on  its  life  under  repeated 
stress  than  its  composition.  The  quahty  of  the  steel  has  more  to 
do  with  the  endurance  than  the  question  of  whether  it  is  a  C 
steel,  a  Mo  steel  or  a  V  steel. 

Of  course,  if  the  comparison  is  made  between,  say  a  plain 
Cr  and  a  Cr-Mo  steel  having  received  the  same  heat  treatment, 
the  one  containing  Mo  will  be   superior,   simply  because   it  is 

'"  Moore,  H.  F.,  and  Konimers,  J.  B.,  An  Investigation  of  the  Fatigue  of  Metals, 
Univ.   of  111.,  Bull.  No.  8,  19,   (1921),  Eng.   Expt.  Sta.,  Bull.   No.   124. 

"  Mc.Adam,  D.  J.,  Jr.,  Endurance  of  Steel  Under  Repeated  Stress,  Chem.  and 
Met.  Eng.,  25,  1,081    (1921). 


EXPERIMENTS  WITH  RARE  ELEMENTS  IN  STEEL.  257 

Stronger  for  the  same  draw.  There  is  probably  one  real  advan- 
tage in  Mo,  in  that  a  Mo  steel  of  given  strength  requires  a  higher 
draw  temperature  than  one  without  Mo,  and  hence  internal 
stresses  which  quite  certainly  reduce  endurance  are  more  fully 
relieved,^^  As  to  V,  it  is  probable  that,  although  it  is  only  added 
to  steel  for  its  alloying  effect,  its  use  makes  for  cleanliness  in  steel, 
because  of  its  strong  affinity  for  oxygen  and  nitrogen,  and  it  is 
certain  that  since  V  and  Mo  are  both  rather  expensive,  the  steel- 
maker normally  will  not  put  either  into  a  heat  of  steel  without 
taking  particular  pains  with  that  heat.  This  psychological  effect 
on  the  steelmaker  probably  causes  these  steels  to  be  better  made 
than  the  average  steel,  and  hence  they  give  better  endurance  and 
reliability  in  practice. 

We  are  comparing  the  endurance  of  different  classes  of  sim- 
ilar heat-treated  steels  in  which  the  variable  alloying  elements 
are  Ce,  Mo  and  V;  the  latter  as  a  basis  for  the  comparison. 
So  far  the  steels  containing  Ce  and  therefore  full  of  inclusions, 
fall  down  badly  on  the  comparison,  especially  when  in  a  very 
hard  condition.  In  the  grand  average  to  date  Mo  steels  are  a  trace 
ahead  of  V  steels,  but  consideration  of  the  data  shows  that  this 
is  due  to  one  open-hearth  Cr-V  steel  supplied  by  a  commercial 
producer  which  shows  up  poorly  on  most  heat  treatments.  The 
V  and  Mo  steels  made  by  the  Bureau  show  equally  good  endur- 
ance at  equal  tensile  strength. 

Hence,  while  the  exaggerated  claims  for  Mo  steel  in  regard  to 
endurance  cannot  be  corroborated,  it  seems,  nevertheless,  true 
that  a  well-made  Mo  steel  is  at  least  as  good  in  endurance  as  any 
other  well-made  steel. 

Abbott-"  has  summed  up  the  situation  in  a  few  words.  He 
says  that  there  is  no  one  type  of  alloy  steel  that  resists  fatigue 
better  than  any  other,  that  there  is  no  alloy  steel  which  is 
markedly  superior  to  all  others,  each  alloy  steel  requiring  its 
own  particular  heat  treatment,  and  the  choice  of  an  alloy  steel 
depends  largely  on  the  ease  with  which  the  necessary  heat-treat- 
ment can  be  given  it.  He  says  that  on  this  basis  the  outlook  for 
more  extensive  use  of  Mo  steel  is  good. 

--  Compare  Aitchison,  L.,  Engineering  Steels,  204   (1921). 

=3  Abbott,  R.  R.,  The  Heat  Treatment  of  Automobile  Steels,  Iron  Age,  106,  1,110 
(1920). 


258  DISCUSSION. 

Wood,^*  basing  his  conclusions  on  wide  experience  with  results 
in  Liberty  engines,  says  that  Cr-Mo  and  Cr-V  steels  are 
equivalent. 

From  every  point  of  view  it  appears  that  Mo  is  an  alloying 
element  in  steel,  which  in  value  stands  with  Ni,  Cr,  and  V. 
Only  inertia  keeps  it  from  wide  use.  Enough  is  known  of  Mo 
steel  to  make  its  good  qualities  evident.  There  is,  of  course,  valid 
objection  by  steelmaker  and  user  to  adding  another  type  of  steel 
to  the  list  instead  of  following  the  general  trend  of  standardization 
and  simplification.  Since  the  first  cost  of  Mo  steel  today  is  no 
more  than  that  of  any  other  alloy  steel  of  equivalent  properties, 
and  its  use  is  often  attended  with  reduction  in  machining  costs,  it 
will  undoubtedly  be  more  widely  employed.  Tests  to  date  on  the 
use  of  U,  B,  Ti,  Zr  and  Ce  as  alloying  elements  have  not  given  con- 
sistently satisfactory  results.  In  fact,  in  view  of  the  non-metallic 
inclusions  attendant  on  the  use  of  all  these  except  B,  and  of  the 
eutectic  formed  with  B,  we  feel  that  their  use  is  more  likely  to  be 
definitely  harmful  than  definitely  advantageous. 


DISCUSSION. 

H,  W,  GiLLETT :  I  would  like  to  add  a  word  that  is  not  in  my 
paper.  The  market  quotations  for  steel  bars  show  that  it  is  not 
any  more  expensive  to  produce  a  given  set  of  properties  with 
molybdenum  steel  than  with  chrome-vanadium  or  nickel-chrome. 
Personally,  I  am  much  sold  on  molybdenum  steel.  I  do  not  feel 
quite  in  the  frame  of  mind  of  one  advertisement  I  saw  a  couple 
of  days  ago,  however,  where  a  firm  advertised:  "Steel — Super- 
Steel — Molybdenum  Steel !"  Nevertheless,  I  think  it  is  a  valuable 
alloying  element. 

Bradley  Stoughton^:  I  think  this  j^aper  of  Dr.  Gillett's  is 
a  valuable  and  interesting  one.  He  has  given  us  a  lot  of  new 
light  on  two  matters  particularly.  I  refer  to  the  question  of 
segregation  and  sonims.  By  sonims  I  mean  solid  non-metallic 
impurities  in  steel.    They  may  be  anywhere  from  almost  molecu- 

"  Wood,   H.   F.,   Progress  in   Metallurgy  of  Alloy   Steels,   Amer.   Drop   Forger,   Jan. 
1920,  p.  25. 

•  Consulting  Engineer,  New   York  City. 


EXPERIMENTS  WITH   RARE  ELEMENTS   IN   STEELS.  259 

lar  size  up  to  particles  that  are  plainly  visible  under  the  micro- 
scope. But  whatever  they  are,  they  are  common  in  almost  all 
steel,  and  they  make  steel  that  is  not  clean. 

There  are  and  always  have  been  two  grades  of  steel.  There  is 
good  steel  and  super-excellent  steel.  For  years  there  was  only 
one  type  of  super-excellent  steel  and  that  was  crucible  steel.  The 
reason  for  that  was  freedom  from  segregation,  freedom  from 
gases  and  freedom  from  sonims. 

Now  electric  steel  is  attempting  to  invade  the  field  of  super- 
excellent  steel.  Whether  it  succeeds  or  not  depends  upon  the 
amount  of  care  and  the  amount  of  money  that  the  electrical  fur- 
nace people  are  willing  to  spend  on  the  manufacture  of  their 
steel.  They  have  not  made  their  steel  carefully  enough.  They 
have  not  observed  precautions  that  should  be  observed  to  make 
steel  free  from  segregation  and  sonims. 

At  the  present  time,  very  good  steel  is  made  by  the  acid  open 
hearth  process,  and  you  can  get,  by  several  processes,  steel  that 
is  low  in  sulfur,  low  in  phosphorus,  low  in  gases  and  all  other 
impurities,  except  sonims.  The  authors  of  this  paper  have  studied 
really  the  question  of  sonims,  and  they  have  only  scratched  the 
field  where  we  must  have  someone  plow  deep  and  harrow  and 
till  and  cultivate  the  crop.  That  is  no  criticism  of  their  paper. 
It  is  a  good  paper,  but  it  only  begins  to  scratch  the  field  that  needs 
to  be  greatly  worked. 


A  paper  presented  at  the  Forty-third 
General  Meeting  of  the  American  Elec- 
trochemical Society  held  in  New  York 
City,  May  5,  1923,  President  Schlueder- 
bcrg    in    the    Chair. 


SOME  EFFECTS  OF  ZIRCONIUM  IN  STEEL 

By  F.  M.  Becket'. 

Abstract. 
This  paper  refers  briefly  to  the  commercial  development  of 
various  alloys  of  zirconium.  Certain  specific  effects  of  zirconium 
in  steel  are  described  as  determined  in  an  extensive  series  of 
experiments.  Outstanding  eft'ects  include  the  ability  of  zirconium 
to  eliminate  oxygen,  nitrogen  and  sulfur ;  the  remarkable  effect 
of  zirconium  in  overcoming  red-shortness  in  high  sulfur  steels ; 
and  the  striking  improvement  in  physical  properties  of  plain 
carbon  steels  brought  about  by  the  presence  of  zirconium  in  rela- 
tively small  proportions. 


The  title  of  this  paper  purposely  expresses  considerable  limita- 
tion and  implies  brevity.  At  this  time  the  principal  object  is  to 
describe  some  of  the  specific  eft'ects  of  zirconium,  as  determined 
in  the  course  of  an  investigation  which  has  involved  350  experi- 
mental heats  of  steel  made  for  this  particular  purpose.  It  is 
intended  that  a  much  more  detailed  discussion  of  the  results  will 
be  presented  in  the  near  future,  including  descriptions  of  the 
procedures  of  the  steel  making,  the  physical  testing,  the  metallo- 
graphic  studies  and  other  phases  of  the  work. 

Numerous  experiments  on  the  reduction  of  zirconium  ores  and 
the  preparation  of  zirconium  alloys  were  conducted  at  the 
Niagara  Falls  plant  of  the  Electro  ^Metallurgical  Co.  during  the 
period  of  a  few  years  immediately  preceding  the  entry  of  United 
States  into  the  world  war.  These  endeavors  confirmed  in  a  gen- 
eral way  the  published  data  relating  to  the  properties  of  some 
of  the  zirconium  compounds ;  but,  more  particularly,  they  devel- 

1  Chief  Metallurgist,  Union  Carbide  Co.,  New  York  City. 

261 


262  F.    M.  BECKET. 

oped  a  few  important,  unforeseen  results,  which  enabled  the 
author  to  relate  much  more  closely  than  he  had  previously  found 
possible  the  properties  of  zirconium  and  certain  zirconium  com- 
pounds to  the  properties  of  other  more  thoroughly  understood 
refractory  materials. 

Early  in  the  year  1918,  having  been  influenced  by  apparently 
authentic  reports  concerning  the  use  by  Germany  of  remarkable 
ordnance  steels  containing  zirconium — reports  which  were  later 
considered  groundless,  if  the  author  has  been  correctly  informed — 
the  War  Industries  Board  decided  upon  an  intensive  program  of 
experimentation  with  zirconium  in  steel  for  light  armor,  the 
direct  object  being  the  earliest  possible  large  scale  production, 
and  the  Electro  Metallurgical  Co.  was  requested  to  furnish  zir- 
conium alloys  with  this  end  in  view.  A  vast  amount  of  energy 
was  then  expended  in  the  way  of  comparatively  large  scale  experi- 
mentation on  the  production  of  a  variety  of  zirconium  alloys, 
and  the  Ford  i\Iotor  Co.  assiduously  attacked  the  problem  of 
zirconium  steel  with  high  ballistic  qualities.  At  the  date  of  the 
armistice  considerable  tonnages  of  zirconium-silicon  alloy  were 
being  shipped  to  designated  steel  companies  for  the  purpose  of 
large  scale  manufacture,  this  particular  alloy  having  been  selected 
as  the  most  efficacious  after  trial  heats  with  many  other  zirconium 
alloys.  As  a  result  of  the  armistice,  the  major  portion  of  the 
alloy  in  these  shipments  did  not  find  its  way  into  the  nickel-silicon 
steel  for  which  it  was  intended.  However,  this  additional  experi- 
mentation on  the  production  of  zirconium  alloys  brought  still 
more  forcibly  to  the  mind  of  the  author  certain  peculiarities  of 
zirconium. 

The  United  States  Navy  also  became  interested  in  zirconium 
steels,  and  requested  the  co-operation  of  the  Bureau  of  Mines 
and  of  the  Bureau  of  Standards,  According  to  H.  W.  Gillett 
and  E.  L.  Mack,  in  Bulletin  199  of  the  Bureau  of  Mines,  1922, 
entitled  "Experimental  Production  of  Alloy  Steels,"  production 
heats  of  a  series  of  zirconium  and  other  similar  steels  began  in 
September,  1918.  In  this  Bulletin  are  described  fully  the  methods 
involved  in  making  the  experimental  heats  (50  lb.)  of  zirconium 
steel,  and  valuable  information  is  contributed  concerning  the 
recoveries  of  zirconium  obtained  from  several  different  zirconium 
alloys.    Technologic  Paper,  No.  207,  of  the  Bureau  of  Standards, 


some;  effects  of  zirconium  in  steel.  263 

1922,  entitled  "Manufacture  and  Properties  of  Steel  Plates  Con- 
taining Zirconium  and  Other  Alloys,"  by  G.  K.  Burgess  and 
R.  W.  Woodward,  reports  in  detail  the  properties  of  the  zirco- 
nium steels  made  by  the  Bureau  of  Mines.  It  is  the  author's 
understanding  that  as  part  of  the  zirconium  phase  of  the  investiga- 
tions reported  in  the  Governmental  papers  just  mentioned,  it  was 
greatly  desired  to  determine  whether  the  exceptional  properties 
of  some  of  the  steels  made  under  the  direction  of  the  Ford  Motor 
Co.  during  the  summer  of  1918  could  be  properly  attributed  to 
zirconium.  The  conclusions  drawn  by  the  authors  of  Technologic 
Paper,  No.  207,  are  to  the  general  effect  that  no  particular 
enhancement  of  desirable  physical  characteristics  are  to  be 
ascribed  to  zirconium,  at  least  in  the  types  of  steel  tested,  and  that 
the  effects  of  this  addition  agent  may  be  detrimental. 

The  foregoing  statements  have  been  made  to  explain  that  a 
tenacious  enthusiasm  for  zirconium  was  the  result  of  information 
acquired  during  the  smelting  of  zirconium-bearing  materials, 
the  production  of  various  alloys  of  zirconium,  and  the  refining  of 
some  of  these  alloys.  So  impressed  was  the  author  in  respect  to 
certain  properties  of  zirconium,  that  an  extensive  program  of 
experimentation  on  zirconium-treated  steels  was  instituted,  and 
has  since  been  continuously  maintained  with  increasing  encourage- 
ment. This  program  was  launched  with  knowledge  of  the  decid- 
edly skeptical  attitude  the  steel  fraternity  had  acquired  concern- 
ing the  value  of  zirconium  additions  to  steel  in  general,  and  in 
particular  the  role  of  this  element  in  the  excellent  steels  that  had 
occasionally  been  produced  by  the  Ford  jNIotor  Co. 

The  practice  followed  in  the  steel  heats  of  the  present  investi- 
gation has  involved  in  the  great  majority  of  cases  the  melting 
of  a  200  to  350-lb.  charge  of  cold  scrap-steel  in  a  basic-lined 
electric  furnace.  Duplicate  or  triplicate  ladles  have  been  tapped 
from  each  heat  in  order  to  permit  of  a  reliable  comparison 
between  the  effect  of  the  zirconium  alloy  addition  and  that  of  an 
equivalent  addition  of  ordinary  ferro-silicon.  Whether  rolled 
or  forged,  the  ingots  from  any  given  heat  have  been  treated 
identically  so  far  as  was  possible  during  hot  working,  and  all 
annealing,  normalizing,  and  heat  treating  operations  on  the 
finished  product  have  been  likewise  conducted  so  as  to  insure 


264  F.   M.  BECKET. 

Strictly  comparable  results.  The  rolling  and  forging  of  the  ingots 
have  been  performed  under  ordinary  mill  conditions  by  experi- 
enced operators. 

ZIRCONIUM  AS  A  DEOXIDIZER  AND  SCAVENGER. 

Zirconium  has  a  greater  affinity  for  oxygen  than  has  silicon,  and 
due  to  this  fact  increased  recoveries  of  silicon  in  the  finished 
steel  are  obtained  by  the  use  of  zirconium-silicon  alloys.  This 
greater  recover^'  of  silicon  is  quite  marked  when  an  alloy  of  35 
per  cent  zirconium  is  employed.  For  example,  in  a  series  of  40 
heats  of  basic  electric  furnace  steel  an  average  silicon  recovery 
of  98  per  cent  was  realized,  as  compared  with  a  recovery  of  84 
per  cent  for  ordinary  ferro-silicon  added  under  identical  condi- 
tions and  in  equivalent  percentages  of  added  silicon  to  duplicate 
ladles.  This  particular  series  resulted  in  a  56  per  cent  average 
recovery  of  zirconium,  ladle  additions  of  0.15  per  cent  zirconium 
having  been  made  in  all  cases. 

The  rate  of  the  reducing  action  of  zirconium  on  the  impurities 
present  in  molten  steel  is  not  only  more  rapid  than  that  of  silicon, 
but  zirconium  is  the  more  efficacious  in  removing  the  final  traces 
of  oxygen  and  nitrogen.  This  scavenging  power  of  zirconium 
is  demonstrated  in  the  partial  or  complete  elimination  of  the 
banded  structure  in  rolled  or  forged  products,  and  in  an  increased 
rate  of  coagulation  of  emulsified  slag.  Zirconium-treated  steels 
possess  a  cleanness  which  appears  to  be  the  result  of  a  far  more 
deep-seated  action  than  characterizes  the  well-known  deoxidizing 
and  scavenging  agents.  There  seems  to  be  abundant  experimental 
evidence  to  justify  this  assertion,  but  the  relative  brevity  of  this 
paper  precludes  a  discussion  of  this  side  of  the  subject. 

Brief  reference  may  be  made  to  the  analytical  evidence  relating 
to  the  deoxidizing  power  of  zirconium.  By  means  of  new 
methods  of  analysis  developed  by  the  Bureau  of  Standards  for 
the  determination  of  oxygen  and  nitrogen  in  steel,  reliable  data 
have  been  obtained  in  co-operation  with  the  Bureau  on  four  heats 
of  steel  treated  with  zirconium-silicon  (0.15  per  cent  added  Zr) 
and  with  ferro-silicon  in  duplicate  ladles.  The  analyses  show 
that  the  zirconium  treatment  eliminated  from  12  to  84  per  cent 
of  the  total  oxygen  present  in  the  steel    (including  oxygen  as 


SOME  EFFECTS  OF  ZIRCONIUM  IN  STEEL.  265 

FeO,  MnO,  SiOo,  ZrOg  and  silicates),  the  average  being  54  per 
cent.  Or,  expressed  in  another  manner,  the  zirconium-treated 
Steels  showed  a  reduction  in  oxygen  content  of  54  per  cent  as 
compared  with  the  steels  treated  with  ordinary  ferro-silicon. 
Analyses  on  another  similar  series  of  4  heats  gave  0.0035  per  cent 
nitride  nitrogen  for  the  zirconium-treated  steels  as  compared 
with  0.0072  per  cent  for  the  ferro-silicon-treated  steels. 

No  indication  of  the  occurrence  of  inclusions  of  zirconium 
oxide  has  been  observed  in  the  course  of  this  investigation.  All 
the  evidence  obtained  points  to  the  conclusion  that  oxidized  zir- 
conium forms  with  silica  and  oxide  of  manganese  a  fusible  slag, 
which  quickly  rises  to  the  surface  of  the  ladle.  Analyses  of  ladle 
slags  have  confirmed  this  conclusion. 

Minute,  yellow,  cubic  crystals  of  zirconium  nitride  are  gener- 
ally observed  in  steels  treated  with  zirconium  in  excess  of  approx- 
imately 0.10  per  cent.  They  are  strictly  limited  in  number  and 
represent  that  residuum  of  the  nitrogen  content  of  the  steel  which 
was  fixed  by  zirconium,  but  not  slagged  off  prior  to  solidification. 
These  crystals  as  such  do  not  exert  a  harmful  effect  on  the  steel ; 
for  instance,  they  were  present  in  their  usual  amount  in  the  heat- 
treated  steels  whose  properties  are  mentioned  later  in  this  paper. 

Fatigue  tests  to  failure  under  rotary  alternating  stress  have  been 
made  on  23  heats  treated  in  duplicate  ladles  with  zirconium-ferro- 
silicon  (0.04  per  cent  added  Zr)  and  50  per  cent  ferro-silicon. 
The  average  effect  of  0.04  per  cent  added  zirconium  has  been  an 
increase  in  the  endurance  limit  by  1,125  lb.  per  sq.  in.  This  is 
particularly  significant  in  view  of  the  recognized  detrimental 
effect  of  non-metallic  inclusions  upon  endurance  limit. 

ZIRCONIUM   AND   SULFUR. 

When  zirconium  is  added  to  steel  in  excess  of  approximately 
0.15  per  cent,  this  element  assumes  a  new  role  by  chemically  com- 
bining with  sulfur  to  form  an  acid-insoluble  compound  not  de- 
tected by  means  of  the  ordinary  evolution  method  of  analysis, 
and  under  any  given  set  of  operating  conditions  a  linear  relation 
exists  between  the  percentage  of  sulfur  thus  fixed  and  the  amount 
by  which  the  added  zirconium  exceeds  0.15  per  cent.  It  has  been 
reasonably  well  established  that  for  basic  practice  when  the  zir- 
conium-silicon alloy  is  added  in  the  ladle,  every  part  by  weight  of 

18 


266  F.    M.  BECKET. 

zirconium  added  in  excess  of  0.15  per  cent  fixes  0.10  part  by 
weight  of  sulfur  as  an  acid-insoluble,  zirconium-sulfur  compound. 
This  chemical  combination  proceeds  in  as  quantitative  a  degree 
when  the  steel  contains  normal  sulfur  and  manganese  contents, 
as  it  does  in  those  instances  where  the  steel  is  sufficiently  high  in 
sulfur  and  low  in  manganese  to  give  rise  to  an  appreciable  propor- 
tion of  iron  sulfide.  In  other  words,  zirconium  has  a  greater 
affinity  for  sulfur  than  has  manganese.  The  difference  here  in 
affinity  favorable  to  zirconium  is  probably  greater  than  the  cor- 
responding difference  between  manganese  and  iron. 

A  5-ton  acid  open-hearth  heat  and  a  10-ton  basic  electric  fur- 
nace heat  may  be  cited  as  examples  of  the  influence  of  zirconium 
on  sulfur  as  determined  by  the  evolution  method.  In  the  former 
case  an  addition  of  0.27  per  cent  zirconium  as  silicon-zirconium 
lowered  the  percentage  of  sulfur  from  an  initial  value  of  0.040 
per  cent  to  a  final  value  of  0.025  per  cent ;  in  the  latter  a  0.22  per 
cent  addition  of  zirconium  diminished  the  sulfur  from  0.020  to 
0.009  per  cent,  leaving  0.15  per  cent  zirconium  in  the  finished 
product. 

Under  favorable  conditions  the  zirconium-sulfur  compound 
may  be  actually  eliminated  from  the  steel  by  fairly  heavy  additions 
of  zirconium-sihcon  alloy.  Steels  containing  0.08  per  cent  total 
sulfur  have  been  reduced  by  ladle  additions  to  a  total  sulfur  of 
0.048  per  cent,  and  a  corresponding  sulfur  content  of  0.037  per 
cent  as  determined  by  the  evolution  method.  Actual  desulfuriza- 
tion  by  zirconium  is  a  field  more  limited  and  much  less  important 
commercially  than  the  field  covered  by  the  effect  of  zirconium  on 
the  hot-rolling  qualities  of  high  sulfur  steels  now  to  be  described. 

In  order  to  obtain  the  full  beneficial  effect  upon  hot-rolling 
properties,  the  zirconium  alloy  need  be  added  only  in  amount 
sufficient  to  eliminate  the  iron  sulfide  constituent  responsible  for 
red-shortness.  Ingots  containing  0.185-0.200  per  cent  sulfur  and 
only  0.15  per  cent  manganese  have  been  rolled  to  plate  and  sheet 
free  from  cracks  and  seams  when  the  steel  had  been  treated  with 
0.22  per  cent  Zr.  With  steels  containing  sulfur  up  to  0  260- 
0.290  per  cent  similar  results  have  been  obtained  by  the  addition 
of  0.43  per  cent  Zr.  The  untreated  ingots  of  these  steels  have 
broken  to  pieces  in  every  case  on  their  first  pass  through  the  rolls. 


SOME  EFFECTS  OF  ZIRCONIUM  IN  STEEL. 


267 


ZIRCONIUM    IN    HEAT-TREATED  STEELS. 

The  beneficial  effect  of  small  additions  of  zirconium  is  strik- 
ingly demonstrated  in  the  case  of  heat-treated,  ordinary  carbon 
Steels,  To  illustrate,  a  heat  of  0.70  per  cent  carbon  steel  was 
treated  in  one  ladle  with  0.15  per  cent  zirconium  as  a  zirconium- 
silicon  alloy,  and  in  the  other  ladle  with  an  equivalent  amount  of 
ordinary  ferro-silicon.  After  forging  the  ingots  to  one-inch  round 
bars  the  test  data  recorded  in  Table  I  were  obtained  on  the  steels 
quenched  from  825°  C.  in  water  and  drawn  at  the  temperatures 
indicated.  Standard  S.  A.  E.  specification  for  a  much  used  nickel- 
chromium  steel  (2.75  to  3.25  per  cent  Ni ;  0.60  to  0.95  per  cent 
Cr)  are  also  tabulated  for  the  purpose  of  comparison. 


Table  I. 


0.70  per  cent  C 
0.15  per  cent  Zr 


Drawing  Temperature    375°  C. 

Per  cent  Elongation   8.3 

Per  cent  Reduction  of  Area..  23.3 

Yield  Point   185.952 

Ultimate   Strength   227,203  _ 

Izod  Number 7.5 

Brinnell  Hardness  414 

Drawing  Temperature    412°  C. 

Per  cent  Elongation    12.7 

Per  cent  Reduction  of  Area..  45.8 

Yield  Point   172,620 

Ultimate  Strength   198,828 

Izod  Number  14.8 

Brinnell  Hardness  407 


0.70  per  cent  C 
without  Zr. 


375°  C. 
5.2 
6.6 
128.125 
197,800 
7.5 
433 

412°  C. 
7.5 
22.9 
180.180 
207,144 
10.5 
418 


S.  A.  E. 

3450  Ni-Cr 


427°  C. 
12.5 
51.0 
175,000 
200,000 


It  may  be  observed  from  Table  I  that  ordinary  carbon  steels 
in  which  a  small  percentage  of  zirconium  has  been  incorporated 
may  be  made  to  possess  by  suitable  heat-treatment  physical  char- 
acteristics approaching  those  of  the  highest  grade,  heat-treated 
alloy  steels. 

Additional  experimentation  has  demonstrated  that  the  proper- 
ties of  a  number  of  the  well-known  alloy  steels  may  be  improved 
through  the  use  of  zirconium,  and  also  that  by  zirconium  treat- 
ment it  is  sometimes  possible  to  use  advantageously  the  ordinary 
alloying  elements  in  less  than  normal  proportions. 


268  DISCUSSION. 

The  author  does  not  consider  as  relevant  matter  for  this  paper 
a  discussion  of  the  commercial  aspects  of  zirconium  in  the  manu- 
facture of  steel,  nor  does  he  wish  to  engage  in  concrete  prognosti- 
cations. Therefore  it  must  suffice  here  to  state  that  several  steel 
companies  to  whom  zirconium  alloys  were  introduced  have  taken 
advantage  regularly  during  the  past  two  or  three  years  of  the 
excellent  scavenging  properties  of  zirconium.  The  effects  of 
zirconium  on  sulfur  and  in  heat-treated  steels  have  been  drawn 
to  the  attention  of  a  few  steel  manufacturers  only  within  a  com- 
paratively recent  period. 

However,  there  appears  to  be  reasonable  justification  for  the 
optimistic  comment  in  conclusion,  that  in  consideration  of  the 
specific  effects  herein  mentioned  and  the  experimental  intimation 
of  other  effects  now  awaiting  recognition,  zirconium  will  probably 
contribute  its  fair  share  toward  the  progress  of  civilization  through 
assistance  to  the  steel  and  other  metal  industries. 

The  author  acknowledges  the  co-operation  of  his  associates, 
Alexander  L.  Feild,  J.  H.  Critchett,  and  J.  A.  Holladay.  ]\Ir. 
Feild  has  contributed  many  valuable  suggestions,  and  he  has  been 
throughout  in  immediate  charge  of  the  experimental  steel  manu- 
facture and  laboratory  testing,  I\Ir.  Critchett,  by  way  of  sugges- 
tion, has  rendered  much  assistance,  especially  in  connection  with 
the  manufacture  of  zirconium  alloys ;  and  Mr.  Holladay  deserves 
much  credit  for  original  work  on  the  quantitative  determination 
of  zirconium  in  ores  and  steels,  and  for  his  supervision  of  the 
analytical  work  involved  in  this  investigation. 


DISCUSSION. 

E.  F.  CoNE^ :  I  can  not  refrain  from  saying  that  I  think  this 
Society  is  unusually  fortunate  in  hearing  what  seems  to  me  to 
be  an  epoch-making  presentation  of  a  subject  that  is,  particularly 
in  the  future,  going  to  be  extremely  important,  especially  with 
reference  to  the  question  of  sulfur  in  rolling,  and  other  points  of 
equal  importance. 

H.  W.  GiLLETT- :  Have  you  data  on  the  ductile  properties  on 
test   pieces   taken   transversely   instead  of   longitudinally?     The 

1  Assoc.  Editor,  Iron  Age,  New  York. 
»  U.  S.  Bureau  of  Mines,  Ithaca,  N.  Y. 


SOME  EFFECTS  OF  ZIRCONIUM   IN   STEEL.  269 

difference  in  ductility  really  ought  to  show  up  more  strikingly 
in  this. 

F.  M.  Becket:  These  particular  tests  did  not  involve  trans- 
verse sections.  In  other  work,  however,  transverse  testing  has 
brought  out  directly  the  point  you  mention,  and  the  effect  of 
zirconium  on  the  transverse  properties  has  been  rather  marked  in 
improvement. 

H.  W.  GiLLETT :  That  seems  to  indicate  a  cleaner  steel  when 
the  transverse  properties  are  good. 

E.  F.  Cone  :  What  is  the  composition  of  these  silicon-zirconium 
alloys  you  use? 

F.  ^l.  Becket  :  The  composition  of  the  alloys  used  both  com- 
mercially and  in  this  experimental  work  varies  considerably,  ac- 
cording'to  just  what  was  attempted— the  class  of  steel  it  was 
desired  to  produce. 

Naturally,  with  a  silicon-zirconium  alloy,  you  are  limited  by 
the  silicon  content  desired  in  the  finished  product.  In  cases  of 
small  additions  of  zirconium,  it  has  been  used  pretty  largely  as  an 
alloy  containing  approximately  10  per  cent  zirconium  and  40  to 
75  per  cent  silicon.  When  it  has  been  desired  to  introduce  con- 
siderable zirconium  in  relation  to  the  proportion  of  silicon  in- 
troduced, an  alloy  of  35  to  38  per  cent  of  zirconium  and  50  to  55 
per  cent  silicon  has  been  employed. 

H.  W.  GillETT:  At  still  higher  temperatures  is  there  the  same 
improvement  in  ductility? 

F  M.  Becket:  Up  to  the  moment,  the  improvement  at  higher 
drawing  temperatures  has  not  been  so  marked.  It  follows  fairly 
well  the  characteristics  of  your  nickel-chrome  and  other  alloy 
steels,  but  I  do  not  think  the  effect  is  so  forcibly  brought  out  as 
at  temperatures  referred  to  here. 


A  paper  presented  at  the  Forty-third 
General  Meeting  of  the  American  Elec- 
trochemical Society  held  in  New  York 
City,  May  5,  1923,  Dr.  F.  M.  Becket  in 
the   Chair. 


INHERENT  EFFECT  OF  ALLOYING  ELEMENTS  IN  STEEL.' 

By   B.    D.    Saklatwalla.2 

Abstract. 
The  importance  of  the  effect  of  alloying  elements  on  the  purely 
physical  changes  occurring  among  the  constituents  of  steel  is 
brought  out.  Stress  is  laid  on  the  study  of  the  physical  condi- 
tions and  their  alterations  by  alloying  elements,  during  the  period 
of  solidification.  Attention  is  drawn  to  the  importance  of  the 
effect  of  alloying  elements  on  surface  tension  of  the  solidifying 
constituents.  The  idea  is  expressed  of  the  possibility  of  coordina- 
tion and  equivalence  among  alloying  elements  based  on  the  periodic 
system,  especially  referring  to  atomic  volume. 


Steel  at  the  ordinary  temperature  is  a  heterogeneous  con- 
glomerate of  various  crystalline  constituents  cemented  together  by 
the  intervention,  betv^een  the  crystal  faces,  of  a  medium  existing 
in  an  indefinitely  knov^^n  physical  state.  The  composition,  physical 
structure,  and  relative  proportions  of  these  constituents  are  gov- 
erned not  only  by  their  chemistry,  but  also  by  the  thermal  life- 
history  of  the  metal.  The  different  phases  are  in  the  main  made 
up  of  a  metallic  (ferritic)  and  a  carbide  (cementitic)  constituent. 
Oviring  to  this  heterogeneity,  it  is  apparent  that  the  physical  forces, 
not  only  those  at  play  in  the  individual  components,  but  also  those 
existing  between  the  phases,  will  be  of  greater  importance,  from 
an  engineering  standpoint,  than  the  chemical  composition.  Un- 
doubtedly through  change  in  the  chemical  constitution  of  the 
components,  as  a  means  to  an  end,  the  physical  changes  are 
brought  about. 

If,  thus,  to  the  ordinary  constituents,  consisting  of  metallic 

'  Manuscript  received  January  30,  1923. 

»  Gen.   Supt.  Vanadium  Corporation  of  America,   Bridgeville,  Penna. 

271 


272  B.  D.  SAKLATWALLA. 

iron  and  an  iron  carbide,  other  elements  are  added,  changes  in 
the  physical  relations  of  these  constituents  will  take  place.  The 
influences  exerted  by  these  elements  constitute  the  metallurgy  of 
alloy  steels.  The  purpose  of  this  paper  is  to  survey  such  influ- 
ences on  the  physical  relations  of  the  constituents,  and  to  direct 
attention  to  their  study  from  a  physico-chemical  standpoint, 
devoting  special  attention  to  the  period  immediately  preceding 
solidification  of  the  steel,  an  interval  in  its  life-history  hitherto 
rather  neglected. 

The  remarkable  properties  conferred  by  carbon  upon  iron, 
making  it  steel,  are  due  to  the  physico-chemical  interactions 
between  iron  carbide  and  iron.  A  wide  range  of  physical  prop- 
erties suitable  for  particular  engineering  problems  are  obtainable 
from  the  same  chemical  composition  of  the  metal  by  merely  vary- 
ing the  physical  heat  treatment.  A  plain  carbon  steel  can  be  made 
exceedingly  brittle  and  glass  hard  by  quenching  in  cold  water 
from  a  high  temperature,  or  made  ductile  and  malleable  by 
allowing  it  to  cool  gradually  from  the  same  temperature.  The 
discovery  that  additions  of  other  metallic  elements  influence 
these  changes  and  produce  different  results  has  been  more  or 
less  of  an  accidental  nature,  and  the  development  of  alloy  steel 
metallurgy  has  been  more  or  less  empirical. 

The  constantly  increasing  number  of  alloy  steels  brought  out 
in  commerce  makes  it  opportune  to  establish  some  scientific  basis 
for  the  relative  influence  of  the  several  elements  depending  on 
some  equivalence  in  physical  properties  among  them.  Undoubt- 
edly some  such  equivalence  of  the  elements  exists,  as  several 
chemically  different  alloys  can  be  made  to  produce  steels  of  more 
or  less  similar  physical  properties  under  a  divergence  of  heat- 
treatment.  In  order  to  investigate  systematically  the  influence 
of  these  elements  on  the  properties  of  the  steel  components,  it 
seems  logical  to  start  such  study  at  a  period  prior  to  the  solidifi- 
cation of  the  metal.  The  physico-chemical  activity  of  the  con- 
stituents, and  the  change  suffered  by  the  addition  of  alloying 
elements,  will  be  more  pronounced,  and  less  influenced  by 
extraneous  physical  conditions,  in  the  liquid,  or  during  the  solidi- 
fying, rather  than  in  the  final  solid  state.  It  will  not  be  an  exag- 
geration to  assert  that  such  a  study  of  the  inherent  influence  of 
alloying  elements  has  been  greatly  neglected. 


EFFECT  OF   ALLOYING   ELEMENTS   IN    STEEL.  273 

The  splendid  work  of  Bakhuis  Roozeboom,  Willard  Gibbs,  and 
others,  has  given  us  wonderful  insight  into  the  phenomena  of 
solidification  from  the  standpoint  of  thermo-dynamics  and  chem- 
ical constitution.  We  have  applied  these  principles  to  the  study 
of  steels,  and  have  been  able  to  chart  the  solidifying  phenomena 
and  establish  thermal  analysis.  We  are  thus  in  position  to  picture 
the  constitution  of  the  components  in  steel  and  further  verify 
our  picture  by  the  aid  of  the  microscope.  It  does  not  appear 
sufficient,  however,  to  know  the  presence  of  these  constituents 
and  their  chemical  nature,  without  being  able  to  correlate  scien- 
tifically their  chemical  composition  to  their  physical  properties, 
and  the  changes  occurring  during  their  solidification  to  the 
engineering  properties  of  the  solidified  steel. 

The  inherent  physical  effects  of  chemical  elements  undoubt- 
edly start  in  the  liquid  stage,  and,  as  the  physical  properties  of 
the  liquid  from  which  crystallization  takes  place  determine  to  a 
great  extent  the  properties  of  the  crystallized  solid,  the  influence 
of  the  alloying  elements  should  be  studied  in  relation  to  the 
physical  changes  occurring  prior  to  or  during  solidification. 
While  undoubtedly,  by  the  proper  thermal  treatment,  much  can 
be  achieved  in  solid  steel,  yet  it  will  be  right  to  assert  that  the 
inherent  characteristics  of  the  steel  are  defined  up  to  solidification 
in  the  ingot  stage,  and  that  all  later  thermal  manipulations  are  of 
secondary  importance. 

Solidification  in  a  metallic  alloy  such  as  steel  occurs  selectively 
during  an  interval,  the  crystal  growth  starting  from  several 
nuclei  in  the  melt.  According  to  Quincke,  a  separation  of  the 
melt  in  two  liquid  phases  takes  place,  the  one  in  very  much 
smaller  quantity,  the  "oily"  phase,  forming  cell  walls  for  the 
other,  the  whole  forming  a  "foam  structure"  with  several  points 
or  nuclei  for  crystal  growth.  The  application  of  X-ray  analysis 
to  crystal  structure  has  shown  us  that  the  atoms  of  the  crystals 
are  arranged  in  definite  characteristic  space  lattices  in  contra 
distinction  to  an  indefinite  arrangement  in  a  liquid. 

Among  the  immeasurably  large  number  of  atoms  of  the  liquid 
melt  there  will  be  some  which  will  chance  to  have  an  arrange- 
ment corresponding  to  the  space  lattice  arrangement  of  the  solid 
crystals,  or  closely  approaching  it.  We  can  readily  see  that  such 
atoms  will  selectively  assume  the  solid  state  ahead  of  the  others, 

19 


274  ^-  °-  SAKLAT WALLA. 

and  hence  act  as  nuclei  for  crystal  growth.  Solidification  from 
these  nuclei  will  proceed,  at  the  same  time  continuously  dimin- 
ishing the  quantity  of  the  molten  mother  magma,  until  the  amount 
of  liquid,  or  the  spaces  left  between  the  grown  crystals,  will  be  so 
small  as  not  to  allow  further  crystallization.  This  residual 
material  will  therefore  fill  up  the  capillary  interstices  between  the 
crystals  forming  the  so-called  "intercrystalline  cement  medium." 

As  to  the  exact  nature  of  this  medium  there  is  considerable 
uncertainty.  On  account  of  it  not  following  the  crystallization  of 
the  first  successive  solidifying  part  of  the  melt,  it  has  been  com- 
monly called  "amorphous."  Recent  observations  with  the  X-ray 
spectograph  on  the  amorphous  metals  would  lead  us  to  assume 
the  presence  of  extremely  fine  crystal  bodies  combined  with 
colloids  in  this  "intercrystalline  medium."  Its  remarkable  prop- 
erties can  be  explained  more  satisfactorily  on  this  assumption 
than  on  that  of  it  being  "amorphous."  Owing  to  the  importance 
of  this  medium  from  an  engineering  standpoint,  it  deserves 
further  close  study  from  the  standpoint  of  colloid  phenomena 
and  X-ray  analysis. 

The  crystallization  from  nuclei  and  the  growth  of  individual 
crystals  will  depend  on  the  chemical  composition  of  the  melt,  its 
degree  of  under-cooling,  heat  conductivity  and  diffusion  capa- 
bility of  the  resulting  crystals,  etc.  In  these  several  factors,  the 
presence  of  other  alloying  elements  in  the  melt  will  exert  a  great 
influence  on  the  progress  of  crystallization.  For  instance,  slight 
impurities  in  the  melt  have  been  found  to  check  the  velocity  of 
crystallization.  The  impurity  adsorbs  on  the  surface  of  the 
growing  crystal,  thus  checking  the  velocity  of  its  growth.  If  the 
adsorption  on  the  different  faces  of  the  crystal  is  of  a  difterent 
degree,  the  crystallization  velocity  will  be  different  in  different 
directions,  and  consequently  the  soHdified  crystal  can  be  altered 
completely,  for  instance,  from  a  polygonal  to  a  dendritic  form. 
Hence  we  can  readily  see  that  the  presence  of  a  small  amount  of 
another  element  in  liquid  steel  can  materially  influence  the  size 
and  shape  of  the  primary  crystals  and  alter  the  structure  in  the 
solidified  steel.  We  are  all  aware  of  the  importance  of  the 
primary  ingot  structure  in  engineering  practice. 

Another  property  of  growing  crystals  is  that  they  eject  any 
impurities  to  the  surface  of  the  crystal.    The  presence  of  another 


EFFECT  OF   ALLOYING   ELEMENTS    IN    STEEL.  275 

element  may  alter  the  solubility  of  such  impurity  in  the  crystal 
and  consequently  influence  the  degree  of  its  ejection.  This 
ejection  brings  such  impurities  present,  not  only  to  the  surface 
of  the  crystal,  but  into  the  "intercrystalline  cementing  medium." 
Thus,  the  presence  of  another  alloying  element  may  appreciably 
alter  the  amount  of  such  impurity  in  this  medium,  and  hence 
influence  in  a  marked  manner,  in  the  finished  steel,  those  physical 
properties  which  are  a  function  of  the  "intercrystalline  medium," 
such  as  elastic  and  endurance  limits. 

Another  influence  of  the  ejection  of  foreign  elements  by  grow- 
ing crystals  can  be  seen  in  the  case  of  non-corrosive  steel  alloys. 
It  appears  remarkable  that  the  non-corrosiveness  is  brought  about 
when  a  definite  sharply  recognized  percentage  of  the  alloying 
element  is  present.  For  example,  10  to  14  per  cent  chrome  steel 
may  be  cited.  This  phenomenon  is  probably  due  to  the  fact 
that  the  growing  crystal  is  capable  of  keeping  in  solution  a 
certain  percentage  of  the  element,  and  starts  ejecting  it  to  its 
surface  after  this  saturation  is  reached.  The  surface  of  the 
crystal  can  thus  have  a  high  percentage  of  the  element  requisite 
to  give  it  the  necessary  protection  against  corrosion. 

Also  in  connection  with  other  physical  properties  we  are  aware 
that  the  percentage  of  the  alloying  element  has  to  be  beyond  a 
certain  range.  As  examples  may  be  cited  3.5  per  cent  nickel 
steel  and  12  per  cent  manganese  steel.  The  influences  of  these 
percentages  can  probably  be  similarly  explained  on  the  basis  of 
ejection  of  these  elements  to  the  surface  by  the  growing  crystals, 
thus  altering  their  surface  properties  of  adhesion,  etc.,  also  intro- 
ducing a  necessary  amount  into  the  "intercrystalline  cement," 
altering  its  properties.  From  this  standpoint  it  is  also  apparent 
why  the  presence  of  non-metallics  in  the  fluid  steel  exerts  such 
dastardly  pernicious  effect  on  the  physical  properties  of  the 
solidified  metal.  They  not  only  influence  the  process  of  crystal- 
lization, but  through  ejection  get  disseminated  in  the  vital  con- 
stituents of  the  steel. 

Whether  we  agree  with  Quincke  on  the  separation  of  two 
liquid  phases  prior  to  solidification,  or  believe  in  crystallization 
growing  from  the  nuclei  only,  it  is  easy  to  see  that  the  surface 
tension,  with  its  dependent  properties,  of  the  molten  magma  will 
be  of    extremely  great  importance.     Considerable  work  has  been 


276  B.  D.  SAKLATWALLA. 

done  on  measurements  of  surface  tension  of  liquid  metals  by 
several  different  methods.  It  has  also  been  assumed,  since  the 
property  of  surface  tension  of  a  liquid  depends  so  intimately  on 
the  cohesion  of  the  molecules,  and  since  the  properties  of  liquids 
and  solids  show  signs  of  continuity  in  the  two  phases,  that  some 
relation  exists  between  the  surface  tension  of  the  fluid  metal  and 
cohesion  and  tenacity  in  the  solidified  state. 

Also  in  studying  liquids  definite  relations  of  other  physical  prop- 
erties to  surface  tension  have  been  established.  As  such  prop- 
erties of  the  liquids  may  be  mentioned:  molecular  volume,  com- 
pressibility, coefficient  of  thermal  expansion,  vapor  pressure  and 
solubility.  Transferring  these  correlation  of  properties  to  the 
solid  state,  we  find  the  important  relation  of  surface  tension  to 
the  factors  which  we  generally  term  "hardness."  Also  we  are 
aware  from  experimental  data  that  "hardness"  more  than  any 
other  physical  property  forms  a  criterion  for  the  endurance  limit. 
Hence  we  see  the  importance  of  a  study  of  the  surface  tension 
qualifications  of  a  metal  in  the  liquid  state,  and  the  influence  of 
foreign  elements  on  the  surface  tension  in  order  to  arrive  at 
engineering  merits  after  solidification. 

It  is  easy  to  understand  that  a  property,  so  inherently  a  func- 
tion of  the  molecule  itself  as  surface  tension,  should  be  ver}' 
sensitive  to  the  presence  of  a  foreign  element  molecule.  Mole- 
cular forces  of  cohesion  naturally  act  with  greater  energy  between 
two  unlike  than  like  molecules.  Consequently  the  presence  of  a 
foreign  molecule  will  increase  the  cohesive  forces.  This  increase 
can  be  of  such  magnitude  as  to  constitute  chemical  affinity,  and 
bring  about  a  chemical  combination  of  the  metallic  elements, 
forming  inter-metallic  compounds.  It  can  be  of  lesser  intensity, 
constituting  physical  action  only,  bringing  about  an  inter-atomic 
rearrangement  with  a  decrease  of  the  total  volume,  increasing 
hardness. 

It  is  our  practical  experience  that  the  hardness  of  a  metal  is 
generally  increased  by  the  addition  of  another  metal  to  it.  Also 
the  properties  and  nature  of  solid  solutions  find  an  explanation 
in  the  intermolecular  cohesive  forces  dependent  on  surface 
tension.  Further,  the  modern  ideas  of  allotropy  seem  to  be 
finding  explanation  in  the  different  cohesive  forces  in  the  atoms 
causing    the    presence    of    physically    different    but    chemically 


EFFECT  OF   ALLOYING  ELEMENTS   IN    STEEL.  277 

identical  matter,  the  differing  atoms  being  capable  of  interaction 
on  one  another.  If  we  assume  such  explanation  for  the  critical 
points  in  iron,  we  can  readily  understand  how  the  presence  of 
foreign  molecules  will  change  cohesive  forces,  and  exert  an 
influence  on  these  points,  which,  in  turn,  will  alter  the  reactions 
depending  on  these  points,  such  as  thermal  reactions  during  heat 
treatment.  Herein  we  can  find  an  explanation  of  the  great  sus- 
ceptibilit}'  of  alloy  steels  for  thermal  treatment. 

Another  influence,  important  from  a  practical  standpoint, 
which  alloying  elements  can  exert  is  their  influence  on  the  non- 
metallic  impurities  in  steel.  The  viscosity  and  surface  tension  of 
the  melt  can  be  altered  by  the  alloying  elements  to  allow  a  better 
mechanical  separation,  or  the  diffusion  capability  of  the  melt 
can  be  influenced  so  as  to  hinder  or  accelerate  segregation  of  the 
non-metallics.  It  is  also  probable  that  slight  additions  of  elements 
can  greatly  influence  the  colloidal  properties  of  the  non-metallics, 
inasmuch  as  their  presence  can  bring  about  a  flocculation  or  dis- 
persion of  the  impurities,  rendering  their  effect  less  harmful. 

In  the  above  considerations  we  have  enumerated  the  effects 
brought  about  by  the  presence  of  alloying  elements  from  a 
physical  standpoint,  without  entering  into  any  considerations  of 
a  purely  chemical  nature.  In  the  introductory  remarks  we  have 
hinted  at  chemical  equivalence  of  the  alloying  elements.  Un- 
doubtedly the  principle  of  periodicity  among  elements  as  initiated 
by  Wedeleeff,  and  expounded  by  Lothar  Meyer,  Crookes  and 
others,  which  has  given  us  such  wonderful  insight  into  the 
workings  of  pure  chemistry,  can  be  applied,  with  modifications  in 
light  of  our  newer  knowledge  of  atomic  structure,  to  metallurgy. 
If  the  elements  are  arranged  as  a  function  of  atomic  weight  to 
atomic  volume,  or  of  atomic  number  to  atomic  volume,  they  form 
a  series  of  connected  curves,  each  one  representing  a  group  of 
elements  and  consisting  of  an  ascending  and  descending  branch. 
The  properties  of  the  elements  so  arranged  seem  to  bear  marked 
relation  to  their  neighbors  on  the  same  curve.  For  instance,  the 
melting  points,  hardness,  ductility  and  brittleness,  electronic  prop- 
erties, surface  tension,  seem  to  be  coordinated  by  these  curves. 
It  appears  from  this  that  the  atomic  volume,  as  an  inherent  char- 
acteristic of  the  atom,  more  than  any  other  property  is  of  para- 


278  B.  D.  SAKLATWALLA. 

mount  importance.  It  undoubtedly  is  the  criterion  of  the  physical 
qualifications  of  material. 

In  light  of  our  present  knowledge  of  the  structure  of  the  atom 
we  can  see  that  the  atomic  volume  will  be  made  up,  not  only  of  the 
masses  of  the  electronic  constituents,  but  also  the  intra-electronic 
spaces  and  the  intra-atomic  spaces.  The  action  between  atoms  is 
known  to  be  dependent  on  their  relative  arrangements  in  space 
lattices,  and  as  these  are  brought  about  by  forces  acting  over 
intra-atomic  spaces  we  can  readily  see  why  the  atomic  volume 
should  be  a  criterion  of  these  changes.  In  this  arrangement, 
according  to  atomic  volume,  it  is  remarkable  that  the  steel  alloy- 
ing elements  group  themselves  close  together.  Attempts  at  gen- 
eralization among  these  elements  have  been  made,  such  as  the 
theory  put  forward  by  Osmond  that  elements  with  greater  atomic 
volumes  than  that  of  iron  tend  to  raise  and  those  with  atomic 
volumes  less  than  that  of  iron  tend  to  lower  the  transformation 
points,  Arg,  Arg,  and  Ar^.  Also  the  elements  producing  marked 
effects  in  steel, possess  high  melting  points,  a  characteristic  also 
dependent  on  atomic  volume. 

In  practical  application  of  these  considerations  extreme  caution 
should  be  used,  as  the  formation  and  presence  of  definitely  formed 
chemical  molecules  in  place  of  the  individual  atoms  introduce  a 
new  phase  in  the  chemico-physical  equilibrium.  In  such  cases  we 
are  confronted  not  with  atomic  but  with  molecular  volumes,  and 
the  effect  exerted  by  the  addition  of  the  element  is  that  of  the 
compound  formed  and  not  the  element  itself. 

In  the  absence  of  definite  theoretical  knowledge  from  a  physico- 
chemical  standpoint,  we  are  obliged  to  judge  the  merits  of  alloy- 
ing elements  from  the  results  achieved  by  them.  Undoubtedly 
the  use  of  alloying  elements  has  wonderfully  advanced  our 
engineering  practice  in  steel  construction.  The  role  of  these 
elements  has  sometimes  been  minimized  with  the  argument  that 
their  presence  only  retards  or  accelerates  the  thermal  changes 
bringing  about  refinement  of  structure.  It  is  not  beyond  the  pale 
of  possibility  that  similar  refinement  can  be  brought  about  by 
other  and  perhaps  purely  physical  means.  Until  such  time,  how- 
ever, we  can  not  get  away  from  the  fact  that  alloying  elements  in 
steel  have  served  indirectly  as  a  means  to  an  end  to  bring  about 


EFFECT  OF   ALIvOYING  ELEMENTS    IN    STEEL.  279 

these  physical  conditions.    Have  they  not  then  fully  and  justifiably 
played  the  part  credited  to  them  ? 

As  to  the  merits  of  the  different  elements  it  appears  that  each 
one  has  a  definite  role  assigned  to  it  to  bring  out  more  pro- 
nouncedly than  the  rest,  certain  definite  physical  characteristics 
in  the  steel.  The  sole  criterion  of  the  accomplishment  of  these 
characteristics  remains  today,  service.  Let  us  hope  that  more 
scientific  study  of  the  role  of  alloying  elements  in  steel  will  not 
only  give  us  insight  into  the  workings  of  the  alloying  elements, 
but  help  to  bring  out  newer  types  and  compositions  of  steels,  thus 
advancing  not  only  the  art  of  metallurgy,  but  the  hopes  and 
aspirations  of  our  rapidly  striding  civilization. 


A  paper  presented  at  the  Forty-third 
General  Meeting  of  the  American  Elec- 
trochemical Society  held  in  New  York 
City,  May  3,  1923,  G.  B.  Hogaboom  in 
the  Chair. 


NOTES  ON  THE  METALLURGY  OF  LEAD  VANADATES.' 

By  Will  Baughman.^ 

LEAD  VANADATE   ORES   OF  THE   SOUTHWESTERN   UNITED  STATES. 

Vanadium  is  widely  distributed  throughout  the  arid  regions 
of  Cahfornia,  Arizona,  Xew  Mexico  and  Nevada.  The  prin- 
cipal minerals  are  vanadinite,  descloizite  and  cupro-descloizite. 
Minor  amounts  of  psittacinite,  volborthite,  eosite,  endlichite, 
calcio-volborthite  and  vanadiolite  are  also  found.  These  minerals 
are  commonly  associated  with  cerrusite,  wulfenite,  pyromorphite, 
stolzite  and  crocoite. 

The  writer  has  compiled  a  hst  of  over  400  occurrences  of 
vanadium  in  the  four  southwestern  States.  There  are  343  in 
Arizona,  28  in  New  ]\Iexico,  25  in  California  and  19  in  Nevada. 
All  but  8  are  occurrences  of  lead  vanadate  and  similar  minerals. 
Of  the  64  deposits  that  show  commercial  possibilities,  43  occur 
in  shattered  zones  at  or  near  the  contact  of  limestone  and  either 
rhyolite,  basalt,  diorite  or  diabase.  One  of  the  largest  deposits 
consists  of  lenticular  masses  up  to  40  ft.  wide,  5  ft.  thick  and  100 
feet  long  (12.2  x  1.5  x  30.4  m.).  The  strike  of  these  ore  lenses 
is  almost  at  right  angles  to  that  of  intruding  dikes  of  diabase  and 
basalt. 

In  all  the  lead  vanadate  deposits,  the  minerals  are  limited  to  the 
secondary  zone  only,  and  as  a  rule  these  secondary  zones  are 
rather  shallow.  Only  twelve  of  the  deposits  extend  beyond  a 
depth  of  250  ft.  {76  m.).  In  all  cases  no  vanadium  is  found 
below  water  level.  In  the  old  Exchequer  mine  water  was  not 
encountered  till  a  depth  of  900  ft.  (274  m.)  had  been  reached, 
and  then  within  50  ft.  (15.2  m.)  the  cupro-descloizite  and  vanadi- 
nite that  had  persisted,  in  large,  well-defined  lodes,  from  the  sur- 

'  Manuscript  received  January  27,  1923. 

*  Consulting  Electro-Metallurgist,  Los  Angeles,   Calif. 

281 


2  82  WII,L  BAUGHMAN. 

face,  disappeared  altogether  and  were  replaced  by  galena,  chal- 
copyrite,  sphalerite  and  pyrites. 

A.  Ditte^  attributes  the  formation  of  the  various  vanadates  to 
percolating  vanadiferous  waters  acting  on  other  compounds, 
principally  those  of  lead.  Arthur  L.  Flagg  and  the  writer  have 
determined  that  the  igneous  rocks,  associated  with  the  large 
majority  of  the  lead  vanadate  deposits,  contain  from  0.04  to  0.11 
per  cent  of  vanadium  trioxide  and  up  to  five  per  cent  of  sodium 
oxide.  It  is  easy  to  suppose  that  such  rocks  could  readily  become 
a  source  of  sodium  vanadate  solutions.  And  as  a  proof  of  this 
supposition,  great  enrichment  is  generally  found  at  those  places 
where  such  infiltering  waters  would  have  met  ascending  miner- 
alizing solutions  or  previously  formed  bodies  of  cerrusite  and 
allied  minerals. 

Arthur  L.  Flagg,  in  his  examination  of  the  U.  S.  Vanadium 
Company's  deposits,  found  an  unknown  black  mineral,  in  the 
diabase,  that  contains  a  large  percentage  of  vanadium.  Work  is 
still  being  done  on  this  mineral  to  determine  its  characteristics. 
At  present  it  appears  to  be  ilmenite  with  part  or  all  of  the 
titanium  oxide  replaced  by  vanadium  trioxide. 

CONCENTRATION    OF  LEAD   VANADATE    ORES. 

The  concentration  of  lead  vanadate  ores  is  limited  to  gravity 
methods.  Some  attempts  have  been  made,  on  an  experimental 
scale,  to  use  oil  flotation  on  sulfadized  minerals.  Unless  heat, 
pressure  or  a  large  excess  of  sodium  sulfide  is  employed  one, 
more  or  all,  it  is  very  hard  to  sulfadize  any  large  amount  of  the 
lead  vanadates.  In  fact  one  method  of  separation  of  wulfenite 
from  vanadinite,  both  of  which  have  substantially  the  same  spe- 
cific gravity,  is  to  sulfadize  and  float  the  lead  molybdate  and  leave 
the  lead  vanadates  in  the  residue. 

At  S.  G.  Musser's  300  tons  per  day  flotation  plant,  where  wulfe- 
nite was  being  concentrated,  the  ratio  of  molybdenum  oxide  in 
the  heads  was  three  to  one.  But  the  ratio  in  the  concentrates 
was  twenty-five  of  molybdenum  trioxide  to  one  of  vanadium  pen- 
toxide.  The  ore  contained  some  lead  tungstate,  stolzite,  which 
sulfadized  readily.  The  concentrates  often  contained  two  per 
cent  of  tungsten  trioxide. 

•Compt.  Rend.   138,  1303. 


THE   METALLURGY   OF  LEAD  VANADATES.  283 

The  lead  vanadate  minerals  are  all  non-conductors  and  cannot 
be  separated  by  electrostatic  methods. 

The  lead  vanadate  minerals  have  specific  gravities  ranging  from 
6.0  to  7.0  and  the  majority  of  the  gangue  minerals  have  specific 
gravities  of  only  half  that.  It  would  seem  that  such  ores  would 
be  amenable  to  gravity  concentration,  and  several  have  attempted 
to  use  wet  and  dry  gravity  methods  of  concentration.  Little 
success  has  been  had  with  these  methods,  because  of  the  great 
tendency  of  the  lead  vanadate  to  form  slimes  during  the  crushing 
and  grinding.  Often,  too,  the  crystals  are  so  small  as  to  be  almost 
microscopic,  but  even  the  large  specimen  crystals  will  "slime" 
readily.     All  the  lead  vanadate  minerals  are  very  brittle. 

A  common  mistake  has  been  to  employ  ball  mills  for  pulveriz- 
ing of  lead  vanadate  ores.  Six  different  mills  using  this  method 
of  pulverizing  have  proved  failures.  None  has  made  a  recovery 
of  more  than  50  per  cent  of  the  vanadium,  when  making  concen- 
trates containing  6  per  cent  or  over  of  vanadium  pentoxide.  Many 
reports,  made  on  some  laboratory  experiment,  have  stated  differ- 
ently, but  none  has  made  good  in  practice.  R.  L.  Grider,  at  the 
Vanadium  Mines  Corporation's*  deposits  in  New  Mexico,  used 
rolls  and  with  careful  classification  was  able  to  recover  69  per 
cent  of  the  vanadium  in  a  rather  low-grade  concentrate.  The 
Dragon  Mining  Company  was  unable  to  raise  the  grade  of  con- 
centrates beyond  4  per  cent  in  vanadium  pentoxide,  nor  to  make 
a  recovery  in  excess  of  43  per  cent  from  ball  feed.  Changing  to 
rolls  raised  their  extraction  to  nearly  60  per  cent  and  the  grade 
of  the  concentrates  to  5.5  per  cent  of  vanadium  oxide. 

The  Black  Buttes  ore  contains  vanadinite,  wulfenite,  and  cer- 
rusite  as  valuable  minerals,  with  minor  amounts  of  galena, 
stolzite  and  crocoite.  The  gangue  is  primarily  calcite  and  quartz 
with  barytes,  fluorspar  and  iron  and  manganese  minerals.  The 
vanadinite  crystals  are  very  large,  often  being  12  mm.  (0.5  in.) 
long  and  3  mm.  (^  in.)  in  diameter.  The  writer  made  tests  on 
a  one-ton  scale  with  this  ore.^  Ball  mills,  highspeed  rolls,  and  a 
centrifugal  impact  pulverizer  were  used  for  crushing  and  grind- 
ing. The  classified  pulp  was  then  fed  to  an  Isbell  table,  to  a  Senn 
pan  motion  concentrator  and  a  Plumb  pneumatic  jig.  The  results 
are  given  in  Table  I. 

*Min.   Sci.   Press,   113,   389-391. 


284 


WILL   BAUGHMAN. 


The  high  recoveries  made  by  the  centrifugal  impact  mill  are 
due  to  the  fact  that  it  pulverized  the  ore  without  the  formation  of 
a  large  amount  of  slimes.  This  mill  works  on  the  principle  of  the 
Varpart  disintegrator  and  schuledenmuhle,  briefly  described  in 
Richards'  Ore  Dressing.  The  mill  has  no  real  field  for  pulverizing, 
except  where  it  is  desired  to  reduce  slime  losses,  as  in  this  case. 
In  this  field  it  is  supreme. 

Table  I. 
Efficiency  of  Grinding  Mills  on  Vdnadinite  Ores. 


Mesh 

Percentage    Recovered    on 

Mill 

Isbell  Table 

Senn   Table 

Plumb    Jig 

Herman  Ball  

40 
40 
30 

40 

41 
40 

62 
83 

45 
47 
68 

89 

HardingeBall 

Rolls  H  S    

38 
59 

Marks    Mill    Centrif. 

80 

The  Marks  mill  resembles  a  centrifugal  separator  in  general 
construction  details.  In  place  of  a  basket  a  table  with  thrower 
blades  or  guides  is  suspended  on  the  shaft.  The  table  or  disc 
revolves  at  2,000  r.  p.  m.  The  ore  is  fed  to  the  center  and  rapidly 
passes  to  the  edge  whence  it  is  thrown  7.6  cm.  (3  in.),  at  a  speed 
of  4.8  km.  (3  mi.)  per  min.,  against  a  heavy,  sectional  cast  iron 
shoe.  The  impact  against  this  shoe  shatters  the  ore  along 
natural  cleavage  lines  and  lines  of  crystallization.  The  shat- 
tered rock  immediately  falls  to  the  screening  apparatus,  the  over- 
size being  returned  for  further  pulverizing.  Primary  feed  may  be 
in  5  cm.  (2  in.)  cubes.  The  shoes  are  kept  clean  by  the  blast  of 
air  caused  by  the  whirling  table,  and,  as  the  ore  is  hit  but  one 
blow  before  being  screened,  a  close  sizing  of  the  product  can  be 
obtained. 

The  disadvantages  of  the  mill  are  that  "rusty"  gold  is  not  given 
the  scouring  or  cleaning  rub  or  twist  that  ball  mills  or  stamps  give. 
It  is  this  same  rub  or  twist  that  produces  slimes.  The  wear  on 
the  shoes  is  also  rather  high.  The  obvious  objection  "that  it  will 
fly  to  pieces"  as  yet  has  had  no  grounds.  Over  40  of  these  mills 
have  been  installed  and  none  has  had  that  trouble  so  far.     Sev- 


the;  metaIvIvURGy  of  lead  vanadates.  285 

eral  have  been  in  steady  use  for  10  years.  A  mill  costs  about 
$2,000,  requires  only  10  horsepower  to  run  it  with  its  accessory 
feeder,  screen  and  elevator,  and  is  capable  of  pulverizing  30  tons 
of  hard  quartz,  per  day  of  24  hours,  to  40  mesh. 

One  promising  field  of  investigation  was  not  followed  out 
according  to  the  writer's  wishes ;  that  is  the  adjustment  of  the 
speed  of  the  table  so  as  to  shatter  the  more  brittle  vanadinite  to  a 
size  permitting  its  passage  through  a  certain  mesh  screen  that 
would  retain  the  less  brittle  gangue  minerals.  The  manufacturers 
of  the  mill  have  conducted  extensive  experiments  in  which  they 
separated  galena  and  sphalerite  in  this  manner.  They  also  have 
separated,  commercially,  strontianite  and  calcite,  recovering  over 
90  per  cent  of  the  strontium  in  a  product  87  per  cent  pure. 

PRODUCTION  OF  VANADIUM   FROM   THE  BLACK  BUTTES  ORES. 

The  metallurgy  of  the  lead  vanadate  ores  of  the  southwest  is 
more  of  an  economic  problem  than  a  technical  one.  There  is  a 
large  amount  of  low-grade  vanadium  ore  available,  most  of  which 
also  carries  valuable  amounts  of  gold,  silver  and  base  elements. 
In  fact,  one  mine  has  been  shipping  vanadium  concentrates  to  the 
smelters  for  the  recovery  of  gold,  silver,  copper,  and  lead,  as 
they  could  find  no  financially  responsible  buyer  who  would  pay  for 
the  vanadium  and  the  precious  and  base  elements  as  well. 

The  Flett-Baughman  Company  became  interested  in  the  Black 
Buttes,  and  first  worked  out  the  concentration  problem.  These 
concentrates  averaged  7.14  per  cent  vanadium  pentoxide,  6.0  per 
cent  molybdic  oxide,  52.4  per  cent  metallic  lead,  2.2  per  cent  com- 
bined chromium  and  tungsten  oxides,  and  $5.00  gold  and  42  oz. 
silver  per  ton.  Only  small  traces  of  phosphorus  and  arsenic 
were  found. 

In  spite  of  the  fact,  that  at  that  time,  there  was  considerable 
activity  in  rare  elements,  and  a  great  many  rare  element  extraction 
plants  were  being  operated,  constructed  or  projected,  the  writer 
was  uable  to  find  any  kind  of  a  reasonable  market  for  the 
concentrates. 

The  following  described  experiments  were  conducted  to  ascer- 
tain which  was  the  best  and  most  economical  method  of  extracting 
the  various  valuable  elements  from  the  concentrates.  The  inves- 
tigation  also   took   into   consideration   the   fact  that   the    ferro- 


286  WILL  BAUGHMAN. 

vanadium  market  is  closely  controlled  through  long  term  contracts, 
by  three  or  four  corporations.  This  rendered  it  imperative  that 
some  new  product  be  developed.  At  first  it  was  planned  to 
produce  C.  P.  vanadic  oxide,  which  was  quoted  at  over  $10.00 
per  lb.  in  the  trade  journals.  However,  the  market  for  this 
product  is  very  limited.  The  writer's  investigations  indicate  an 
annual  consumption  of  not  over  500  lb. 

The  writer  then  started  a  series  of  investigations  in  the  pro- 
duction of  pure  metallic  vanadium,  although,  in  view  of  his 
present  knowledge  of  the  rare  element  business,  he  can  not  at 
present  see  where  he  expected  any  large  market  for  the  pure 
metal.  For  additions  to  steel  the  standard  30  to  40  per  cent 
ferro-vanadium  is  supreme.  Iron  has  a  melting  point  of  1530° 
C.  and  vanadium  1720°  C.  while  the  alloy  (30  to  40  per  cent)  has 
a  melting  point  of  about  1440°  C.  It  will  be  seen  that  the  addi- 
tion of  pure  vanadium  metal  to  steel  would  only  raise  the  melting 
point  and  offer  no  compensating  advantages. 

The  writer  is  not  the  only  one  who  has  made  this  mistake.  To 
his  personal  knowledge  over  twenty  companies  have  been  formed 
to  produce  vanadium  oxide  on  a  large  scale,  all  of  which  expected 
to  sell  all  their  product  at  the  quoted  prices  for  the  C.  P.  oxide. 
One  corporation  spent  $700,000  and  another  $250,000  in  the 
same  vain  attempts.  None  of  these  companies  planned  to  use  a 
process  that  would  produce  an  average  product  better  than  85  per 
cent  pure,  and  some  even  expected  to  sell  the  vanadium  oxide  in 
iron  vanadate  at  the  quoted  figures.  The  preparation  of  99  per 
cent  pure  vanadium  oxide  is  not  the  easy  matter  that  it  would 
appear  to  be ;  in  fact  it  is  a  very  difficult  chemical  operation. 

Hereafter  the  writer  describes  the  results  of  his  investigations 
in 

1.  Chloride  volatilization  applied  to  Buttes  concentrates. 

2.  Baughman  process  for  same  ores. 

3.  Chloridizing  roasting  of  the  U.  S.  Vanadium  Go's  ores. 

4.  Sodium  sulfide  leaching  of  the  "Signal"  ores. 

5.  Production  of  metallic  vanadium. 

CHLORIDE   VOLATILIZATION   OF  THE   BLACK    BUTTES   ORES. 

By  using  a  centrifugal  impact  pulverizer  and  gravity  concentra- 
tion,   the    Black    Buttes    ore    yielded    a    high    grade    concentra- 


THE  METALLURGY  OF  LEAD  VANADATES.  287 

tion,  and  a  high  extraction  was  also  made.  There  was  ample 
water  not  only  for  mill  purposes  but  also  for  power,  so  that  the 
logical  thing  was  to  concentrate  and  then  determine  a  suitable 
method  of  handling  the  concentrates. 

The  Flett-Baughman  Co.  had  been  experimenting  for  some  time 
on  complex  lead-copper-zinc  ores,  and  among  other  plans  had 
done  a  great  deal  of  work  in  applying  chlorine  direct  to  the  ores. 
The  idea  of  chloride  volatilization  was  attractive.  Splendid  results 
could  be  easily  obtained  on  a  laboratory  scale,  so  that  20-lb.  scale 
tests  were  conducted  as  described  below : 

A  rotating  silica  tube  5  ft.  long  and  13  in.  in  diameter  (1.5  m. 
X  0.32  m.)  and  heated  by  an  electrical  resistance  coil  was  used. 
The  volatilized  elements  were  caught  in  a  stoneware  bafifle  tower 
15  ft.  (4.5  m.)  high  and  6  in.  (15  cm.)  in  diameter.  Tempera- 
tures from  200  to  800°  C.  were  employed.  The  majority  were 
run  at  400.°  The  gases  from  the  tower  were  dried  by  passing 
over  sulfuric  acid  and  through  calcium  chloride,  and  returned 
to  the  volatilization  tube. 

Sodium  Chloride.  Temperatures  of  200  to  600°  C.  were  used. 
Percentages  of  salt  between  20  and  40.  Time,  up  to  24  hours. 
The  best  extraction  was  38.5  per  cent  of  the  vanadium  volatilized, 
and  36  per  cent  converted  to  soluble  sodium  vanadate.  This  was 
done  at  600°  with  25  per  cent  salt  and  required  24  hours. 

Magnesium  Chloride.  Hydrated  magnesium  chloride  is  decom- 
posed, by  heating,  to  magnesium  oxide  and  hydrogen  chloride, 

MgCl2  +  H,0  =  MgO  +  2HC1 

.Some  magnesium  oxychloride  is  also  formed,  which  is  in  turn 
decomposed  by  oxygen  in  the  air  into  magnesium  oxide  and 
chlorine, 

2MgO  -f  2HC1  =  MgO  .  MgCl.  +  H.,0 
MgO  .  MgCU  -f  O  =  2MgO  +  C\, 

Thus  a  supply  of  chlorine  and  hydrogen  chloride  are  generated 
in  the  tube. 

The  best  extraction  obtained  was  43  per  cent  of  the  vanadium 
volatilized  and  36  per  cent  of  the  molybdenum,  while  27  per  cent 
of  the  vanadium  and  33  per  cent  of  the  molybdenum   formed 


288  WILI,  BAUGHMAN. 

magnesium  vanadates  and  molybdates.  These  salts  are  soluble 
in  hot  water  and  re-precipitated  on  cooling;  but  this  gives  two 
totally  different  classes  of  product  to  treat  further  and  refine. 

Calcium  chloride.  Two  runs  only  were  made  with  this  reagent. 
Some  calcium  molybdate  and  vanadate  were  formed,  and  some 
oxychlorides  volatilized,  but  the  results  were  so  small  that  analyses 
to  determine  percentages  recovered  were  not  made. 

In  all  three  of  the  above  series  of  experiments  a  gentle  stream 
of  hot  dry  air  was  passed  through  the  tube  and  absorption  tower. 

Ferric  Chloride.  This  is  a  very  expensive  reagent  to  use  for 
the  purpose  of  chloride  volatilization.  Both  ferric  chloride  and 
ferric  sulfate  with  magnesium  and  calcium  chloride  to  form 
ferric  chloride,  by  interaction,  were  tried.  These  are  known  as 
the  Gin  methods.  Ferric  sulfate  and  calcium  or  magnesium 
chloride  interact  thus,  to  produce  calcium  sulfate  and  ferric 
chloride. 

Fe,(SO,)3  +  3CaCL  =:  3CaSO,  +  ZFeCla 

Extractions  in  excess  of  90  per  cent  of  the  molybdenum  and 
vanadium  were  obtained.  Some  magnesium  or  calcium  vanadates 
and  molybdates  were  formed.  The  high  cost  of  the  reagents  pre- 
cludes the  commercial  use  of  this  method  at  present.  Considerable 
lead  chloride  also  volatilized,  and  by  interaction  in  the  tower  re- 
produced lead  vanadates  and  molydates. 

Carhon  Tetrachloride.  One  run  was  made  with  this  reagent. 
An  extraction  of  98.5  per  cent  of  the  vanadium  and  98  per  cent 
of  the  molybdenum  was  obtained.  A  temperature  of  400  to  450" 
C.  was  employed.  Seven  hours  time  was  required.  The  cost  of 
this  chemical  makes  the  process  prohibitive;  also  a  large  excess, 
over  that  theoretically  required,  must  be  used. 

Chlorine  and  Hydrogen  Chloride  Gases.  Chlorine  alone  does 
not  give  a  good  extraction.  It  is  necessary  partially  to  reduce  the 
ore,  preferably  by  reducing  gases,  before  applying  the  chlorine. 
Hydrogen  chloride  alone  gives  a  splendid  extraction  of  the  molyb- 
denum, but  does  not  give  as  satisfactory  results  for  vanadium. 
The  method  used  was  as  follows : 

The  concentrates  were  first  heated  for  2  hr.  at  400°  C.  in  an 
atmosphere  of  natural  gas.    Equal  portions  of  chlorine  and  hydro- 


THE  METALLURGY  Of  LEAD  VANADATES.  289 

gen  chloride  gas  were  then  passed  through  the  tube  for  4  hr. 
longer.  The  reactions  are  rather  complex  but  may  be  expressed 
as: 

•s^VgOs  +  C^'H^  =  -rV^O^  +  pCO  +  qCO^  +  rH^O 
V^O,  +  3CI2  =  2VOCI3  +  O2 

V2O5  4-  6HC1  =  2VOCI3  +  3H2O 
M0O3  +  2HC1  =  M0O2CI2  +  H^O 

WO3  +  2HC1  =  WO2CI2  +  H2O 

CrOa  +  2HC1  =  CrO^Cl^  +  H,0 

The  last  four  are  reversible  and  on  catching  the  volatilized  ele- 
ments in  water  in  the  absorption  tower,  they  are  re-converted  into 
their  respective  oxides  and  the  acid  is  regenerated.  Considerable 
heat  is  evolved  at  the  same  time. 

Some  of  the  molybdenum  and  tungsten  oxides  were  reduced 
during  the  preliminary  reduction.  These  reduced  oxides  interact 
with  chlorine  as  follows : 

M0O2  +  CI.  =  M0O2CI2 
WO2  +  CI2  =  WO2CI2 

The  large  amount  of  regenerated  acid  causes  the  re-solution  of  a 
large  portion  of  the  precipitated  oxides,  and  in  addition  to  this 
some  very  complex  rare  element  compounds  are  formed,  which 
give  all  kinds  of  trouble  in  later  refining  steps.  The  most  serious 
objection  was  the  large  amount  of  gas  required  to  conduct  the 
operation.  Most  of  the  silver  was  also  volatilized  and  formed 
silver  vanadates  that  made  its  recovery  more  expensive.  Lead 
chloride  volatilized  also,  although  most  of  the  lead  was  reduced 
in  the  first  step  to  metal  and  remained  in  the  residue  as  fine 
pellets. 

From  an  economic  standpoint  these  experiments  were  failures. 
Technically  a  large  percentage  of  the  rare  elements  were  recov- 
ered. The  best  run  was  at  400°  C. ;  time  2  hr. ;  reduction,  4  hr. 
chloridizing  volatilization ;  vanadium  recovered  96.5  per  cent ; 
molybdenum  recovered  98  per  cent. 

BAUGHMAN's  process   for  the   TREATMENT  OF  LEAD  VANADATES. 

After  the  failure  of  the  chloride  volatilization  experiments,  the 
Flett-Baughman  Co.  initiated  some  new  experiments  on  the  Black 


290  WILIv  BAUGHMAN. 

Buttes  ore.  The  method  selected  was  first  to  smelt  the  con- 
centrates, producing  a  vanadiferous  slag  and  a  lead  bullion 
containing  the  gold  and  silver.  This  step  was  analogous  to  the 
first  steps  of  Gin's  process  and  to  Grider's  process. 

The  concentrates  were  smelted  in  an  electric  furnace  of  a  tilting 
type.  It  was  constructed  along  the  lines  of  the  Girod  tilting 
resistance  furnace  described  in  Bulletin  77  of  the  Bureau  of  Mines 
page  109.  In  addition  to  the  resistance  heating,  provision  was 
also  made  for  using  it  as  an  arc  furnace  by  placing  three  5  in. 
(12.7  cm.)  graphite  electrode  stubs,  well  rammed  with  magnesite 
and  tar  in  the  bottom,  and  suspending  a  3  in.  (7.6  cm.)  graphite 
electrode,  through  a  hole  in  the  cover,  from  the  ceiling,  by 
wires,  attached  to  the  electrode  holder.  The  electrode  was  raised 
and  lowered  by  passing  these  wires  through  pulleys  to  a  hand 
operated  winch  and  drum.  The  furnace  also  had  a  tap  hole  at  the 
bottom,  the  use  of  which  will  be  seen  later.  The  melting  chamber 
was  15  in.  (39  cm.)  in  diameter  by  22  in.  (54  cm.)  high. 
Alternating  current  was  used  to  heat  the  resistors,  and  direct  cur- 
rent from  a  motor  generator  set  with  two  generators  of  40  to  60 
volts  each  and  capable  of  being  connected  in  either  series  or 
parallel  supplied  the  arc  current.  The  generators  were  each 
capable  of  supplying  a  current  of  900  amp.  The  furnace,  how- 
ever, drew  only  500  amp.  except  during  the  starting  period. 

The  method  of  operation  was  to  charge  150  lb.  (68  kg.)  of 
concentrates,  15  lb.  (6.8  kg.)  of  pulverized  coke  and  30  lb. 
(13.6  kg.)  of  soda  ash,  thoroughly  mixed,  into  the  furnace  and 
melt  the  same  by  the  resistance  heaters.  Two-thirds  of  the  above 
charge  was  put  in  the  furnace  at  the  start,  and  the  balance  as  soon 
as  the  first  portion  was  melted. 

At  the  end  of  1.5  hr.  the  lead  was  tapped  from  the  bottom 
notch,  and  as  soon  as  this  was  completed  a  pipe  was  inserted  in 
the  notch  and  quickly  luted  with  fire  clay.  A  blast  of  air  at  20  lb. 
(9.1  kg.)  pressure  supplied  by  a  large  Crowell  blower  was  then 
forced  through  the  slag.  The  slag  changes  from  green  to  blue 
in  color,  when  cooled  for  tests,  by  plunging  slag  rod  in  water. 
The  furnace  was  then  tilted  and  the  molten  slag  poured  into  hot 
water  to  granulate  it  and  render  it  more  soluble.  Up  to  this 
point  the  process  is  substantially  the  same  as  those  of  Gin  and 
Herrenschmidt. 


THD  METALLURGY  OF  h^AD  VANADATES.  29 1 

The  solution  in  which  the  slag  was  poured  was  filtered  off  and 
the  residue  mixed  with  one  fourth  its  weight  of  caustic  soda.  This 
mass  was  then  roasted  at  a  low  temperature,  about  300°  to  400° 
C,  for  2  hr.  It  was  then  digested  2  hr.  with  the  original  solution 
from  which  it  was  first  filtered.  The  solution  was  filtered  again 
and  the  residue,  which  contained  less  than  3  per  cent  of  its 
original  vanadium  content,  was  discarded.  To  assist  in  the  forma- 
tion of  the  metavanadate  of  soda,  peroxide  of  hydrogen  was 
added  at  first,  later  the  same  results  were  obtained  by  using 
ozone.  This  ozone  was  made  from  oxygen,  from  the  electrolytic 
cells  used  in  a  later  step  in  this  process,  passed  through  an 
ozonater. 

A  calculated  amount  of  sodium  sulfide  was  also  added  to  the 
solution  while  the  roasted  slag  was  being  digested.  This  is  to 
sulfadize,  thus  rendering  it  insoluble,  any  zinc  or  lead  that  might 
have  remained  in  the  slag.  Sodium  aluminate,  also  in  calculated 
amounts,  was  used  to  precipitate  the  phosphorus.  The  digester 
was  made  of  half  inch  sheet  steel ;  it  was  mechanically  agitated 
and  steam  heated. 

Fractional  Crystallization. 

After  the  solutions  were  filtered  from  the  residue,  they  con- 
tained sodium  chloride,  vanadate,  sulfate,  carbonate,  molybdate, 
tungstate,  chromate,  hydroxide  and  aluminate.  This  solution 
was  then  evaporated  in  a  triple  effect  evaporator  to  22°  Be., 
and  the  sodium  sulfate,  aluminate  and  chloride  allowed  to  crystal- 
lize out  and  be  removed.  The  next  fraction  was  removed  at  26° 
Be.  and  contains  most  of  the  vanadium  as  the  various  sodium 
vanadates.  However,  this  fraction  is  not  pure,  as  it  contains 
some  molybdenum  and  chromium  as  complex  molybdo-chromo- 
sodium  vanadates.  The  third  crystallization  is  at  30°  Be.  and  it 
contains  some  sodium  vanadate  with  sodium  carbonate.  These 
salts,  after  calcining,  are  used  for  the  original  flux  in  smelting  the 
ore.  Thus  no  vanadium  is  lost  in  the  cycle  of  the  process.  The 
last  fraction  is  obtained  at  33°  Be.  It  yields  the  sodium  molyb- 
dates,  chromates  and  tungstates  with  some  vanadium  as  complex 
salts.  The  second  and  last  fractions  are  mixed  together  and  used 
in  the  next  step. 

Since  these  experiments,  the  writer  has  developed  a  system 


292 


WILL  BAUGHMAN. 


Flow  Sheet. 
Baughman  Process  for  Treating  Complex  Lead-Rare 


Concentrates 


Element  Ores. 


Coke 

\ 


i     ^ 


Soda  Ash 


Air. 


Caustic  Soda 


Sodium  Aluminate 
Sodium  Sulfide 


r 


No.  1 
No.  2 
No.  3 
No.  4 


OZONATER 
I 

Oxygen       1 


ELECTRIC  FURNACE 

t  ^ >Pb,  Au,  Ag  Bullion 

— >■  Slag  Treatment 

\ 

>►  STIR  TANK 

FILTE  R > \ 

I  1 

>  MIXER 

\  I 

MUFFLE   FURNACE 
\  ->  DIGESTOR    <— 

_J_ I       X 

^  T-TT  'TTJTD    —^     ^  F,  residual  Pb  and  Zn 

i^lLiliR  -T^    -^  Tailings,  Cu,  etc. 

t 
EVAPORATOR 

\ 

CRYSTALLIZING  VATS 

Soda  nitrate,  Sodium  chloride  and  sulfate 
/"^—Sodium  vanadates 

I- 


J 


Sodium  carbonate-^MUFFLE  FURNACE- 
Sodium  tungstates,  molybdates,  chromates 


DISSOLVER< 

-ELECTROLYSIS  CELLS 


^ 


Hydrogen    i  I 

Water^-DECOMPOSITION  VAT 

V }  \ 


Na  OH  Solu. 


I 


MERCURY  PUMP-' 


I 


SUPER  CENTRIFUGE 

I 

FILTER 

PUMP — > 

^  CONDENSING 
(  CHAMBERS 
Molvbdic  Oxide 


^ 


VOLATILIZATION  TUBE- 
— >'First  Step 

''   >       Second  Step  Vanadium  Trioxide 

Residue:  Metallic  Tungsten,  Chromium  and  Titanium  oxides  and  metal 


THE  METALLURGY  OF  LEAD  VANADATES.  293 

of  fractional  crystallization  for  such  complex  mixtures,  that  yields 
each  salt  separately  and  in  pure  condition.  He  has  also  found 
the  method  of  preventing  the  formation  of  the  complex  molyb- 
denum-chromium-vanadium compounds  with  sodium.  This 
system  is  rather  complicated  and  is  too  lengthy  to  describe  here. 

Electrolysis  of  Sodium  Salts. 

The  mixed  vanadates,  chromates,  molybdates  and  tungstates 
of  sodium  were  then  dissolved  to  a  20°  Be.  solution.  This  was 
done  in  a  small  tank  with  a  mechanical  stirrer.  From  there  the 
solution  went  to  the  electrolysis  cells,  which  were  similar  to  the 
mercury  cathode  Solvay  cells  for  producing  chlorine  and  caustic 
soda.  The  salts  are  decomposed,  the  sodium  entering  the  cathode 
and  the  rare  element  oxides  remaining  in  a  semi-colloidal  form 
in  the  electrolyte. 

The  sodium  amalgam  was  kept  in  constant  circulation  between 
the  electrolysis  cells  and  outside  decomposition  vessels  by  a  mer- 
cury pump.  The  amalgam  was  decomposed  by  the  action  of  water 
in  this  outside  vessel,  to  mercury  and  caustic  soda;  considerable 
hydrogen  was  also  evolved. 

The  reactions  in  the  electrolytic  cell  might  be  expressed ; 

2NaV03  +  electricity  =  2Na  (in  Hg.)  -f  V^O,  +  O 

and  the  decomposition  of  the  amalgam  as: 

NayHg^  +  sU.O  =  xUg  +  yNaOH  -j-  sU. 

The  regenerated  caustic  soda  was  used  again  in  the  previous 
steps  of  the  process. 

The  electrolyte  was  also  kept  in  constant  circulation.  It  was 
passed  through  a  Sharpies  centrifuge  and  then  filtered.  This 
eliminated  serious  trouble  that  had  been  encountered  previously, 
due  to  the  semi-colloidal  condition  of  the  suspended  oxides  of  the 
rare  elements.  The  filtrate  is  used  to  dissolve  fresh  amounts  of 
the  crystallized  salts  from  the  previous  step  of  fractional  crystal- 
lization. 

A  current  of  8  volts  was  used.  The  cells  required  1,800  amp. 
Seven  were  connected  in  series.  The  anodes  were  of  platinum 
gauze,  but  fused  iron  oxide  would  have  been  as  satisfactory. 


294  WILL   BAUGHMAN. 

Arrangements  were  made  for  collecting  the  hydrogen  and  oxygen 
evolved,  for  use  in  refining  the  oxides. 

The  electrolysis  of  sodium  vanadate  solutions  in  diaphragm 
cells  has  been  proposed  by  W.  F.  Bleecker^  and  investigated  by 
S.  Fischer.®  The  w^riter  used  a  mercury  cathode  cell,  because  the 
oxides  produced  in  diaphragm  cells  always  contained  impurities 
from  the  disintegration  of  the  diaphragms. 

Separation  of  the  Oxides. 

The  filtered  oxides  were  washed  and  dried  and  then  treated 
by  one  of  the  two  following  methods : 

Electrolytic  Method.  The  mixed  oxides  were  dissolved  in  a 
stoneware  agitator,  by  dilute  hydrochloric  and  sulfuric  acids. 
This  solution,  as  near  neutral  as  possible,  was  again  electrolyzed 
in  a  mercury  cathode  cell.  The  tungsten,  molybdenum  and 
chromium  passed  into  the  amalgam,  the  vanadium  again  separated 
as  the  oxide,  considerable  chlorine  was  given  oflf  and  both  phos- 
phorus and  arsenic,  which  had  been  added  for  purpose  of  testing, 
were  completely  volatilized.  Some  vanadium  remained  in  solu- 
tion. The  mercury  was  pumped  through  a  chamois  skin  amalgam 
filter  instead  of  into  the  outside  decomposition  vessel.  Excess 
mercury  was  strained  and  pressed  out  of  the  amalgam,  which 
was  then  retorted.  The  mixed  chromium,  molybdenum  and 
tungsten  metals  remaining  in  the  retort  were  then  ignited  to  the 
oxide,  (they  were  highly  pyrophoric)  and  the  molybdenum  oxide 
removed  from  the  chromium  and  tungsten,  which  was  discarded, 
by  volatilization.  This  is  identical  with  the  first  step  of  the  other 
method  of  separation  of  the  rare  element  oxides  and  will  be 
described  later. 

Volatilisation  Method.  The  volatilization  tube  previously 
described  was  used.  The  separation  was  conducted  in  two  steps. 
In  the  first  the  molybdenum  trioxide  was  volatilized  in  a  current 
of  oxygen.  The  oxygen  was  a  by-product  of  the  electrolysis  of  the 
rare  element  sodium  compounds.  The  molybdenum  trioxide 
was  completely  volatilized  in  6  hr.  at  a  temperature  of  800"  C. 
Increasing  the  temperature  beyond  this  point  favored  the  forma- 

»Met.  and  Chem.  Eng.  9,  503  (1911). 

« Trans.  Am.   Electrochem.   Soc.  30,   175    (1916). 


THE  METALLURGY  OE  LEAD  VANADATES.  295 

tion  of  molybdo-vanadates,  and  at  850°  only  85  per  cent  of  the 
molybdenum  trioxide  was  volatilized. 

In  the  second  step,  hydrogen  from  the  decomposition  cell  of 
the  electrolysis  cells  was  passed  over  the  mixed  oxides  of  vana- 
dium, tungsten  and  chromium.  At  the  same  time  the  temperature 
was  raised  to  1400°  C.  The  hydrogen  gas  was  preheated  as  was 
the  oxygen  gas  used  in  the  previous  step.  In  this  step  the 
tungsten  and  chromium  were  reduced,  the  former  to  metal  and 
the  latter  to  sub-oxide  and  metal.  Titanium  was  also  reduced 
to  the  lower  oxide.  The  vanadium  pentoxide  was  reduced  to  the 
trioxide  and  then  became  volatile.  At  the  end  of  18  hr.  it  was 
completely  volatilized.  The  tungsten-chromium  residue  was  dis- 
carded. 

Removal  of  Phosphorus  and  Arsenic. 

The  Black  Buttes  ore  contained  no  trace  of  arsenic  and  very 
little  phosphorus.  A  large  majority  of  the  lead  vanadates  con- 
tain these  impurities.  From  acid  solutions  arsenic  may  be  com- 
pletely removed  by  passing  over  copper,  copper  arsenide  being 
formed.  Where  the  solution  is  further  treated  by  passing  over 
iron,  to  reduce  the  vanadium  so  as  to  make  precipitation  easier, 
the  copper  dissolved  by  the  excess  acid  will  be  re-precipitated. 
From  acid  solutions,  preferably  hydrochloric,  the  phosphorus  can 
be  completely  removed  by  precipitation  as  zirconium  phosphate. 
This  precipitation  can  be  obtained  from  quite  strong  acid  solutions. 
Zirconium  hydrate  (crude)  prepared  from  zirkite  is  dissolved  in 
hydrochloric  acid  and  used  as  a  precipitant. 

From  alkaline  solutions  sodium  aluminate,  made  by  dissolving 
aluminum  shot  in  concentrated  caustic  soda  solution,  secures  a 
complete  precipitation  of  the  phosphorus  and  also  a  little  of  the 
arsenic. 

From  neutral  solutions,  strontium  nitrate,  made  by  dissolving 
the  mineral  strontianite  in  commercial  nitric  acid,  will  precipitate 
all  the  phosphorus  and  arsenic,  together  with  most  of  the  tungsten 
and  molybdenum  present,  but  it  precipitates  very  little  vanadium. 

The  electrolysis  of  any  acid  solution  containing  chlorides 
secures  the  complete  volatilization  of  the  arsenic  and  most  of  the 
phosphorus. 


296  WILL  BAUGHMAN. 

CHLORIDIZING  ROASTING  AND  LEACHING  OF  LEAD  VANADATE  ORES. 

At  several  of  the  vanadium  deposits,  of  the  southwestern  part 
of  United  States,  there  are  valuable  amounts  of  gold  and  silver 
in  the  ores,  that  are  not  recovered  by  concentration,  and  not  infre- 
quently there  is  so  much  barytes  present  that  it  is  impossible  to 
obtain  a  high  grade  concentrate.  Where  such  conditions  exist 
and  there  is  a  large  amount  of  low  grade  ore  available,  chloridiz- 
ing  roasting  and  leaching  is  an  ideal  treatment  method. 

These  experiments  were  first  tried  on  the  Buttes  ore,  but 
abandoned  when  suitable  concentration  methods  were  found.  Later 
the  process  was  the  subject  of  an  extensive  investigation  by  the 
U.  S.  Vanadium  Development  Company,  under  the  direction  of 
the  writer.  Later  the  Consolidated  Vanadium  Company  built 
a  25  ton  per  day  plant  to  use  the  same  method,  they  being  cog- 
nizant of  the  U.  S.  Vanadium  Co's  experiments. 

Chloridizing  roasting  and  leaching  has  been  in  successful  use 
in  an  ever  expanding  plant  at  Park  City,  Utah,  at  a  cost  of  about 
$3.00  per  ton.  The  ore  treated  there  contains  6  to  14  oz.  silver 
per  ton,  SLOO  in  gold,  a  couple  of  pounds  of  copper  and  small 
amounts  of  lead  and  zinc.  Recoveries  as  high  as  95  per  cent 
have  been  made,  85  to  90  per  cent  being  common  practice.  This 
process  is  very  simple,  is  economical  on  a  large  scale,  and  is 
capable  of  handling  very  low  grade  ores.^  Briefly  described  it 
consists  of  roasting  with  admixed  fuel  in  a  shaft  furnace,  and 
leaching  with  tower  acid,  precipitating  silver  and  gold  on  copper, 
copper  on  iron,  and  later  electrolytically  recovering  the  lead. 

Roasting. 

A  mixture  of  6  to  9  per  cent  salt,  1  to  3  per  cent  coal  dust,  1  to 
2  per  cent  manganese  dioxide,  and  2  to  8  per  cent  pyrites  is  thor- 
oughly mixed,  moistened  with  tower  acid  till  the  mixture  will 
retain  the  imprint  of  the  fingers  when  tightly  pressed.  This  mois- 
ture varies  from  5  to  10  per  cent  according  to  the  fineness  of  the 
ore. 

A  deep  bed  of  coals  is  started  in  the  bottom  of  the  shaft  roaster 
and  the  roast  mix  charged  to  a  foot  of  depth.  As  soon  as  the 
roast  shows  through  3  ft.  (0.9  m.)  more  of  charge  is  added,  and  at 

'  The    method    is   fully    described    in    Trans.    Am.    Inst.    Min.    and    Met.    Engr.,    49, 
183-197,  and  has  also  been  the  subject  of  several  articles  in  various  mining  journals. 


THE  METALLURGY  OF  LEAD  VANADATES.  297 

the  next  appearance  of  the  roast  4  ft.  ( 1.2  m.)  more,  which  is  about 
the  maximum  depth  to  which  the  blower  can  supply  air.  The 
bottom  of  the  shaft  is  all  a  grate,  with  a  wind  box  underneath. 
Air  is  supplied  at  about  1  lb.  (0.45  kg.)  pressure. 

Temperature  is  controlled  by  rate  of  blast.  It  is  better  to  roast 
too  slowly  than  too  fast.  Temperature  range  may  be  between 
600  and  800°  C.  For  the  U.  S.  Vanadium  ores  650  to  750°  C. 
was  found  best.  Too  high  a  temperature  causes  clinkers  and 
melts  the  salt,  forming  a  thin  glaze  on  the  ore  particles.  However, 
caked  masses  are  a  sign  of  good  roasting.  Slimes  and  fines  cake, 
not  clinker,  readily  at  the  proper  temperatures,  and  form  an  easily 
leached  product. 

Were  a  dry  charge  used,  a  great  deal  of  the  valuable  elements 
would  be  volatilized,  but  the  moisture  is  concentrated  about  a  foot 
ahead  of  the  roasting  zone  and  thus  entraps  any  volatilized  ele- 
ments, so  that  the  only  volatilization  occurs  at  the  end  of  the  roast 
when  the  temperature  is  lowest.  Even  then  the  volatilized  ele- 
ments are  caught  in  the  absorption  tower,  where  the  barren  mill 
brine  is  returned  in  order  to  catch  the  acids  of  the  roasting  fumes. 

The  reactions  during  roasting  are  very  complex.  In  general 
the  following  may  be  said  to  occur : 

Chlorine  is  produced  by  interaction  of  salt  and  sulfur  trioxide, 
obtained  from  the  pyrite,  at  elevated  temperatures, 

2NaCl  -f  2S0,  =  Na^SO,  +  SO^  +  CU 

Chlorine  is  also  obtained  from  salt,  silica  and  oxygen, 

2NaCl  +  SiO^  +  O  =  Na^SiOs  -f  Cl^ 

This  nascent  chlorine  acts  strongly  on  metals  and  sulfides 
present,  and  to  a  lesser  degree  on  oxides, 

Au  +  3Cl  =  AUCI3 
Cu,S  +  4C1  -f  30  =  2CuCl  +  SO3 

Some  metallic  chlorides  are  formed  direct, 

2NaCl  -f  PbSO,  =  PbCL  -{-  Na,SO^ 

Sulfur  dioxide  may  be  converted  in  part  to  trioxide  by  catalysis 
by  silica  or  peroxidized  by  iron  oxide, 

20 


298  WILL   BAUGHMAN. 

2SO,  +  30  +  SiO^  =  2SO3  +  O2  +  SiO^  +  22,600  cals. 
SOo  +  3Fe203  =  SO3  +  2Fe304 

After  reduction  of  the  vanadium  from  the  penta  to  tetra  state 
it  is  readily  attacked  by  the  chlorine  (in  the  roaster)  to  form 
the  volatile  oxychlorides  thus, 

V2O5  +  C  =  V2O4  +  CO 

V2O4  4-  CI,  =  2VO2CI 

V2O,  +  3CI2  =  2VOCI3  +  O2 

These  are  decomposed  by  the  w^ater  in  the  absorption  tower, 
and  redissolved  by  the  excess  acid, 

2VO2CI  +  H2O  =  V2O5  +  2HC1 
2VOCI3  +  3H2O  =  V,0,  +  6HC1 

In  the  tower  chlorine  and  sulfur  dioxide  form  sulfuric  and 
hydrochloric  acids, 

C\,  +  SO2  +  2H2O  =  H2SO,  +  2HC1 

Steam  and  silica  interacting  v^^ith  salt  form  hydrochloric  acid 
in  the  roaster, 

2NaCl  +  SiO,  +  H2O  =  Na^SiOa  +  2HC1 

Sulfur  trioxide,  steam,  and  salt  also  form  hydrochloric  acid  in 
the  roaster, 

2NaCl  +  SO3  +  H2O  =  Na^SO,  +  2HC1 

In  turn  the  sulfuric  acid  in  the  brine  acts  upon  the  salt,  so  that 
the  free  acid  is  hydrochloric, 

H2SO,  +  2NaCl  =  Na^SO,  +  2HC1 

While  unable  to  prove  it  in  every  way  the  writer  has  strong 
evidence  that  salt  and  vanadium  pentoxide  form  sodium  vanadate, 

2NaCl  +  V2O5  +  O  =  2NaV03  +  Cl^ 
6NaCl  +  V.O^  4-  30  r=  2Na3VO,  +  3CI2 

Roasting  in  reverberatories  or  mechanical  furnaces  does  not 
give  the  results  that  the  Holt  shaft  roaster  does.  The  slower 
heating  and  the  much  longer  cooling  is  the  reason  for  this.    At 


the;  me;tai,l,urgy  of  lead  vanadati;s.  299 

Park  City,  Theodore  P.  Holt  used  ore  through  0.25  in.  (0.64  cm.) 
mesh,  but  the  writer  has  determined  that  the  ore  should  be  crushed 
to  0.0625  in.  (1.6  mm.)  mesh  at  least  when  treating  vanadium 
ores. 

The  addition  of  manganese  is  the  result  of  the  writer's  investiga- 
tions. In  his  work  on  complex  lead-zinc-copper  sulfide  ores  he 
found  that  he  could  convert  over  90  per  cent  of  these  metals  to 
sulfates  during  the  roast  by  adding  manganese  dioxide  in  the 
form  of  pyrolusite  to  the  charge.  These  complex  ores  were  after- 
wards leached  with  dilute  sulfuric  acid  to  remove  the  copper, 
which  was  precipitated  as  cuprous  chloride,  and  the  zinc  which 
was  precipitated,  after  purification  of  the  solution,  by  electrolysis. 
The  lead,  gold  and  silver  were  dissolved  by  a  strong  brine.  The 
precious  metals  were  precipitated  on  copper  and  the  lead  by  elec- 
trolysis, at  the  same  time  regenerating  the  chlorine  in  the  brine, 
by  which  the  gold  was  attacked  and  made  soluble.  The  difficult 
step  of  this  process  lay  in  the  roasting  so  as  to  form  a  maximum 
amount  of  sulfate,  without  forming  insoluble  ferrites  or  excessive 
oxide.  The  accidental  addition  of  manganese  dioxide  gave  such 
wonderful  results  that  the  writer  tried  it  in  chloridizing  roasting 
also.  For  vanadium  ores  it  acts  as  an  oxidizer,  and  assists  in 
releasing  a  large  amount  of  acid.  Many  large  and  small  experi- 
ments have  proved  its  value. 

Lixiviaiion  and  Precipitation. 

For  the  U.  S.  Vanadium  Go's  ores  a  pulp  ratio  of  one  to  five 
was  found  best.  A  strong  acid  brine  with  a  gravity  of  20  to 
24°  Be  was  used.  The  temperature  was  to  be  maintained  at  60° 
C.  by  the  use  of  steam.  The  brine  was  applied  in  counter  current 
to  the  ore. 

The  greenish  yellow  solution  was  returned  to  the  tower  and 
leaching  vats  till  it  was  a  strong  green  in  color  and  contained 
10  g./L.  of  vanadium  or  over.  The  acid  solution  was  then  partly 
neutralized  and  passed  over  copper  rififles  to  precipitate  the  gold 
and  silver.  This  step  also  removes  the  arsenic  as  arsenide  of 
copper. 

The  solution  Avas  next  passed  over  scrap  iron  and  the  copper 
precipitated.  At  the  same  time  the  nascent  hydrogen  from  the 
action  of  the  excess  acid  on  the  iron,  and  in  fact  the  iron  itself. 


300  WILL  BAUGHMAN. 

reduced  the  vanadium  from  the  penta  to  tetra  state  and  the 
molybdenum  to  the  molybdous  state.  The  solution  became  a  dark 
blue. 

The  next  step  was  the  electrolytic  recovery  of  the  lead,  as 
sponge  lead,  at  the  same  time  the  vanadium  was  further  reduced 
to  the  tri-valent  state.  Insoluble  anodes  were  used  for  the  sponge 
lead  electrolysis  at  first. 

Flow  Sheet. 
Chloridizing  Plant  for  U.  S.  Vanadium  Development  Co. 

STORAGE    BINS 
Ore                    Salt                Pyrites  Pyrolusite        Coal  Dust 

I  \ \  \  \ 


r 


MIXER        i      y 

I        /-TOWERS-EXHAUST  FAN 

i   t 

HOLT  SHAFT  ROASTER 

i  r- 

LEACH  SYSTEM ^Tailings 

Tronaormagnesite  [^^^eUTRALIZING  TANK        Phosphorus 
Zirconium  Chloride  ]  ■ 

Scrap  copper ^-COPPER  RlFFLES->Gold,  Silver       Arsenic 

Scrap  iron ^IRON  RIFFLES ^Copper 

ELECTROLYSIS  CELLS >^ 


ANODE   COMPARTMENT  Iron  vanadate  and  molybdate 

(to  refinery) 
CATHODE  COMPARTMENT  Sponge  lead 

A  mixture  of  vanadium  and  molybdenum  oxides  together  with 
manganese,  iron,  lime  and  other  elements  as  hydroxides  and 
carbonates  was  obtained  by  using  crude  trona  as  a  precipitating 
agent.  The  writer  found  later  that  calcined  magnesite  was 
cheaper,  gave  a  higher  grade  precipitate,  and  that  the  precipitate 
was  easier  to  filter  than  that  from  the  soda  precipitation. 

Later  on  soluble  iron  anodes  were  used,  in  order  to  lower  the 


THE  METALLURGY  OF  LEAD  VANADATES.  30I 

power  requirements  for  the  precipitation  of  the  lead.  We  were 
surprised  to  find  that  the  vanadium  and  molydenum  were  com- 
pletely precipitated  as  iron  vanadates  and  molybdates  by  purely 
anodic  processes.  Xo  extra  power  was  required,  although  a  dia- 
phragm to  prevent  the  mixing  of  the  anode  products  and  the 
sponge  lead  was  necessary. 

The  mixed  vanadates  and  molybdates  are  of  a  much  higher 
grade  than  any  obtained  by  chemical  precipitation.  They  are  also 
very  granular  and  easily  filtered  and  washed. 

A  certain  amount  of  the  brine  should  be  rvm  to  waste  on  each 
cycle  to  prevent  the  fouling  of  solution  by  sulfates.  The  wash 
water  and  the  salt  added  with  each  roast  will  in  general  keep  the 
brine  up  to  standard. 

The  disadvantages  of  the  process  are  that  it  can  not  be  applied 
to  ores  containing  any  large  amount  of  calcium  or  magnesium, 
and  that  for  its  economic  operation  plants  should  have  a  capacity 
of  at  least  50  tons  per  day. 

The  Consolidated  Vanadium  Co.  built  a  25-ton  plant  using  this 
process,  as  worked  out  by  the  writer,  which  was  closed  for 
internal  and  legal  reasons  shortly  after  its  initial  operation.  The 
best  run  they  made  gave  an  extraction  of  76  per  cent  of  the 
vanadium  in  an  ore  cotaining  only  0.16  per  cent  vanadium  pen- 
toxide.  The  gold  and  silver  recoveries  approximated  those  of 
Holt  at  Park  City. 

The  phosphorus  in  the  ore  was  eliminated  at  the  time  of  neu- 
tralizing the  leach  solution.  This  was  done  by  adding  a  solu- 
tion of  zirconium  chloride,  which  was  prepared  by  dissolving 
crude  zirconium  hydroxide  in  hydrochloric  acid.  The  zirconium 
hydroxide  may  be  prepared  in  any  manner  from  zirkite.  The 
phosphorus  is  precipitated  completely,  even  from  highly  acid  solu- 
tions, as  zirconium  phosphate. 

Chloridizing  roasting  has  been  used  successfully  for  many  years 
in  Colorado  for  the  treatment  of  roscoelite. 

The  precipitation  of  iron  vanadate  and  molybdate  by  anodic 
reaction  is  analogous  to  the  old  Luckow  paint  processes  for  pre- 
paring lead  carbonate  and  chromate.  Warren  F.  Bleecker  has 
patented^  certain  phases  of  the  precipitation  of  vanadium  by  this 
method. 

«U.   S.   Patent   1,105,469. 


302  WILL   BAUGHMAX. 

The  Consolidated  Vanadium  Co.  also  developed  a  soluble  anode, 
which  consisted  of  a  wooden  basket  in  which  machine  shop  turn- 
ings worth  only  $6.00  per  ton  were  used.  The  plates  used  before 
had  cost  $50.00  per  ton. 

The  U.  S.  Vanadium  Co's  ores  contained  an  average  of  $2.00  in 
gold  and  4  oz.  silver  per  ton,  0.57  per  cent  vanadium  oxide  and 
0.52  per  cent  molybdic  oxide.  They  have  a  tremendous  amount  of 
this  grade  of  ore.  It  also  contains  small  amounts  of  copper,  lead, 
arsenic  and  phosphorus. 

The  tests  for  adaptability  of  these  ores  to  chloridizing  roast- 
ing and  leaching  were  concluded  on  200-pound  scale  experiments. 
Extractions  ranging  from  72  per  cent  to  76  per  cent  of  the  rare 
elements,  and  90  to  95  per  cent  of  the  precious  metals,  were 
readily  obtained. 

Sodium  Sulfide  Leaching. 

The  vanadium  minerals  of  the  "Signal"  ores  are  primarily 
cuprodescloizite  and  vanadinite,  with  minor  amounts  of  vanadio- 
lite  and  volborthite.  The  gangue  is  principally  calcite  with  fair 
amounts  of  barytes  and  quartz.  A  typical  analysis  is :  Gold 
$19.00,  silver  6  oz.  per  ton ;  copper  2.5  per  cent ;  lead  3.5  per  cent ; 
vanadium  pentoxide  2.25  per  cent;  lime  25  per  cent;  barytes  18 
percent ;  P,  As,  Ti,  ]\lo,  W,  none. 

The  large  amount  of  lime  prevents  the  use  of  chloridizing 
roasting.  Concentration  is  seriously  interfered  with  on  account 
of  the  bar}'tes  present ;  also  the  gold  and  silver  are  not  amenable 
to  concentration.  Certain  economic  factors  had  to  be  considered 
in  designing  a  process  for  these  ores.  The  plant  had  to  be  simple 
in  construction  and  operation.  The  first  cost  was  to  be  kept  as 
low  as  possible.  It  was  also  desired  that  the  vanadium  be  recov- 
ered as  a  readily  marketed  compound,  so  that  the  expense  of  a 
refinery  and  ferro-alloy  plant  could  be  dispensed  with. 

Alan  Kissock"  has  successfully  employed  sodium  sulfide  for 
the  extraction  of  molydenum  from  wulfenite.  S.  G.  Musser  also 
built  a  plant  using  this  process.  Kissock  used  counter  current 
decantation  for  treating  the  ore.  Musser  used  theoretical  propor- 
tions and  applied  heat  and  pressure.    The  reaction  is  substantially, 

PbMoO,  +  Na,S  =  PbS  -f  Na,MoO, 

»U.    S.   Patent   1,403.035. 


THE   METALLURGY  OE  LEAD  VANADATES.  303 

Both  of  them  pecipitated  the  molydenum  by  calcium  chloride. 
This  was  a  by-product  of  S.  G.  Musser's  plant  for  treating 
residual  brines  from  the  extraction  of  salt  from  sea  water. 

Na^MoO,  +  CaCl^  =  CaMoO,  +  2NaCl 

Alan  Kissock^°  has  patented  the  process  of  using  calcium 
molybdate  so  produced  as  a  direct  addition  to  steel,  the  carbon  in 
the  bath  reducing  the  oxide,  which  readily  alloys  with  the  metal. 
Parenthetically,  it  may  be  remarked  here  that  calcium  vanadate 
can  not  be  employed  in  a  similar  manner.  Both  Mr.  Kissock  and 
his  assistants  and  the  writer  have  repeatedly  tried  to  achieve  this 
end,  but  have  failed  in  all  cases.  The  writer  has  used  scheelite 
concentrates  (CaWO^)  in  a  like  manner  for  adding  tungsten  to 
steel. 

Warren  F.  Bleecker  and  W.  L.  Morrison"  have  described 
experiments  in  which  they  added  calcium  vanadate,  with  suitable 
reducing  agents  as  aluminum  or  silicon,  direct  to  the  bath.  They 
obtained  splendid  results. 

Because  of  the  great  ease  with  which  molybdenum  could  be 
extracted  from  wulfenite  by  sodium  sulfide,  the  writer  initiated 
experiments  to  ascertain  whether  or  not  the  "Signal"  ores  could 
be  treated  in  a  similar  manner. 

Laboratory  tests  soon  showed  that  simple  counter  current  lixi- 
viation  was  not  sufficient.  Heat,  pressure  and  agitation  all  increase 
the  efficiency.  For  some  ores  it  is  necessary  to  add  sodium  poly- 
sulfide,  in  order  to  take  care  of  the  cerrusite  and  similar  minerals 
that  also  consume  sulfur  by  becoming  sulfadized.  Other  ores 
particularly  those  containing  vanadiolite  and  calcium  vanadates 
require  the  addition  of  caustic  soda.  The  sulfadizing  and  forma- 
tion of  sodium  vanadate  reactions  are  very  complex,  but  for  sim- 
plicity's sake  may  be  expressed  thus: 

rzn 

-  Pt 

cuprodescloizite  '  Cl 

3Pb3(VO,),  .  PbCla  -f  10Na,S  =  lOPbS  -f  6Na3VO,  +  2NaCl 

vanadinite 

The  use  of  sodium  sulfide  as  a  solvent  for  vanadium  formed 
part  of  a  patented  treatment  method  of  G.  Fester.^^    Its  similar 

>«U.  S.  Patent  1.385,072. 

"Met.  and  Chfem.  Eng.   13,  492-494   (1915). 

"German  Patent  294,932   (1917). 


(Cu  .  Pb  .  Zn),  V,0,  -f  2  Na.S  =  2  ]  pbj  S  -f  Na,V/J, 


304  WILI.   BAUGHMAN. 

use  has  been  recently  patented  in  the  United  States  by  one  of 
S.  G.  IMusser's  former  laboratory  assistants. 

After  the  usual  laboratory  and  small  scale  experiments,  the 
writer  specified  the  following  procedure,  which  was  carried  out  in 
a  25^-ton  per  day  scale  at  S.  G.  Zinsser's  plant.  The  ore  was 
pulverized  to  0.025  in.  (0.63  mm.)  mesh  in  a  ball  mill.  It  was 
then  charged  into  a  rotating  drum  made  of  0.5  in.  (12.7  mm.) 
boiler  plate,  which  was  8  ft.  (2.4  m.)  long,  and  3.5  ft.  (1.1  m.) 
in  diameter.  It  had  a  tight  fitting  manhole  cover  and  hollow 
axles,  so  that  steam  could  be  supplied  for  heating  and  pressure. 
Several  blades  on  the  inside  of  the  drum  aided  agitation  as  the 
drum  was  revolved.  Three  solutions  were  successively  employed, 
the  one  containing  the  least  sodium  sulfide  first  and  the  strongest 
sodium  sulfide  liquor  last. 

The  second  solution  became  the  first  solution  for  the  next  lot  of 
ore  to  be  treated,  and  the  third  solution  the  second.  A  new  third 
or  strong  sodium  sulfide  solution  was  prepared,  by  dissolving  com- 
mercial sodium  sulfide  in  water  to  saturation  at  average  tempera- 
ture, and  then  diluting  with  an  equal  amount  of  water.  The  ore 
was  digested  for  4  hours,  0.5  ton  to  the  lot.  A  temperature  of 
90°  C.  and  a  pressure  of  120  lb.  was  maintained.  At  the  end  of 
this  period  the  ore  was  discharged  and  filtered.  The  residue  can 
be  easily  treated  by  oil  flotation  for  the  recovery  of  both  the 
precious  and  base  metals.  Experiments  in  this  case  were  made 
with  a  K  &  K  laboratory  flotation  machine. 

The  filtered  solution  was  then  evaporated  in  a  single  effect 
evaporator  to  20°  Be.  and  sent  to  the  crystallizing  tanks,  where 
the  sodium  sulfate  and  chloride  were  crystallized  out.  It  was 
then  evaporated  to  26°  Be.  and  crystallized,  yielding  a  mixture 
of  sodium  ortho,  pyro,  and  meta  vanadates. 

For  a  plant  situated  on  the  desert,  where  the  vanadium  deposits 
are,  the  best  evaporator  would  be  spray  ponds.  These  evaporators 
consist  of  parallel  pipes,  with  many  small  perforations  on  the  top 
side,  which  are  suspended  about  10  ft.  (3m.)  above  a  shallow 
pond.  The  hot  dry  desert  wind  blowing  through  the  spray  causes 
a  very  rapid  evaporation.  A  centrifugal  pump  keeps  the  solution 
in  circulation.  A  3  in.  (7.6  cm.)  centrifugal  pump  will  supply  an 
evaporator  capable  of  evaporating  20,000  lb.  (9,071  kg.)  of  water 
per  day  of  average  desert  weather.     The  writer  has  used  such 


THE  METALLURGY  OF  LEAD  VANADATES.  305 

evaporators  successfully  in  evaporating  borax,  potash,  trona, 
potash  alum  and  nitrate  liquors  at  desert  deposits. 

The  residual  liquor  from  the  sodium  vanadate  crystallization 
contains  some  sodium  vanadate,  sulfide,  sulfite  and  hypo  sulfite. 
It  is  returned  to  the  leach  system  with  the  new  or  strong  sodium 
sulfide  liquor. 

The  mixed  sodium  vanadates  are  a  commercial  product.  They 
can  also  be  converted  into  the  oxide  by  treating  with  either 
sulfuric  or  nitric  acids,  baking  to  dryness,  and  then  washing  or 
rather  leaching  and  filtering  to  remove  the  soluble  sodium  nitrate 
or  sulfate,  and  recovering  the  insoluble  vanadic  oxide.  This 
process  has  splendid  possibilities  in  its  limited  field.  The  cost 
of  plant  and  of  operation  are  both  low.  This  method  can  not  be 
used  if  the  ore  contains  phosphorus,  molybdenum,  tungsten  or 
other  impurity  forming  soluble  compounds  with  the  sodium  sul- 
fide.   These  impurities  would  render  the  product  worthless. 

The  writer  has  developed  a  method  of  using  such  mixed  sodium 
vanadates  with  some  iron  oxide  and  metal  and  aluminum  for 
production  of  ferro  vanadium  by  the  metallo-thermic  method  or 
with  silicon  in  the  electric  furnace.  This  method  will  be  described 
at  some  future  date. 

Metallic  Vanadium. 
The  writer  has  developed  a  method  of  making  pure  metals 
from  difficultly  reduced  oxides.  He  has  prepared  vanadium 
which  was  over  99  per  cent  pure  by  this  method,  and  has 
also  produced  very  pure  lithium,  tantalum,  titanium,  thorium  and 
cerium  by  the  same  method.  The  writer  had  hoped  to  be  able  to 
describe  this  method  in  this  paper  but  business  reasons  have  pre- 
vented, and  he  can  only  hope  to  make  it  the  subject  of  some 
future  paper.  In  the  development  of  a  process  for  making  metal- 
lic vanadium  the  writer  duplicated  the  work  of  previous  investi- 
gators, and  devised  a  method  of  reduction  with  lithium  metal. 
His  experiments  along  these  lines  are  described  hereafter. 

Sefsfrom's  Method.  Sefstrom,  the  discoverer  of  vanadium, 
found  that  on  dissolving  iron  containing  vanadium  with  dilute 
hydrochloric  acid,  that  the  vanadium  remained  in  the  residue  with 
the  graphite  and  other  insoluble  matter.  Ferrovanadium  may 
be  dissolved  in  dilute  hydrochloric  acid,  while  passing  a  stream  of 
21 


306  WILL   BAUGHMAN. 

carbon  dioxide,  and  about  one  half  of  the  vanadium  content  of 
the  ferro-alloy  will  be  recovered  as  vanadium  metal.  The 
vanadium  carbide  and  graphite  present  in  the  ferro  is  also  insolu- 
ble, and  will  be  an  impurity  in  the  vanadium  residue.  Working 
with  ferro  prepared  from  pure  materials  in  magnesite  crucibles 
by  the  alumino-thermic  method,  and  dissolving  the  alloy  with 
C.  P.  acid,  the  writer  was  able  to  prepare  vanadium  metal  over 
90  per  cent  pure.  This  method  has  been  used  for  some  time  in 
Germany,  to  prepare  the  vanadium  metal  sold  to  experimenters 
and  colleges.  It  is  in  the  form  of  fine  glistening  scales,  much 
resembling  graphite  in  appearance.  It  can  be  fused  in  vacuo  only, 
and  even  then  contains  a  large  percentage  of  vanadium  monoxide. 
It  oxidizes  readily  in  the  atmosphere. 

Roscoe's  Method.  This  method  is  fully  described  in  Roscoe  and 
Schorlemmer's  Treatise  on  Chemistry  pp.  279  to  282.  The  writer 
attempted  to  duplicate  this  method  on  a  1  lb.  (0.45  kg.)  scale. 
He  used  a  rotating  silica  tube  4  in.  (10  cm.)  in  diameter  and 
3  ft.  (90  cm.)  long,  which  was  heated  electrically  by  a  resistance 
coil.  The  hydrogen  train  and  other  accessories  were  the  same 
as  specified  by  Roscoe  only  of  suitable  size.  At  the  end  of  6  days 
less  than  half  the  chloride  had  been  reduced  to  metal,  but  on  sub- 
stituting a  smaller  tube  and  using  a  silica  boat  containing  one 
gram,  nearly  90  per  cent  of  the  chloride  was  reduced  to  metal  in 
48  hr.  The  preparation  of  the  chloride  is  a  very  difficult  matter 
in  itself,  and  as  this  method  offered  no  commercial  possibilities, 
no  further  experiments  were  conducted. 

Prandtl  and  Bleyer's  Methods.  They  describe  a  method^^  of 
preparing  metallic  vanadium  up  to  94  per  cent  pure.  They  used 
a  can  10  in.  (25  cm.)  high  and  5  in.  (12  cm.)  in  diameter.  In  the 
bottom  of  this  they  tamped  a  layer  of  fluorspar  1.5  in.  (4  cm.) 
thick.  They  then  placed  a  glass  tube  10  in.  (25  cm.)  long  and 
2  in.  (5  cm.)  in  diameter  in  the  center  and  tamped  fluorspar 
around  this  tube.  The  next  step  was  to  tamp  a  mixture  of 
calcium,  aluminum  and  vanadium  oxide  inside  the  tube  and  with- 
draw the  tube  by  twisting  and  turning.  The  mixture  was  then 
ignited  by  a  "thermit  cherry." 

No  data  were  given  as  to  the  size  of  the  particles  of  aluminum 
and  calcium  nor  the  proportions  of  reducing  agent  and  vanadium 

"Z.  anorg.  Chem.  64,  217-224. 


THE  METALLURGY   OF  LEAD  VANADATES.  307 

oxide,  save  that  there  are  to  be  69  parts  of  calcium  for  31 
parts  of  aluminum.  It  is  assumed  that  they  planned  on  the  fol- 
lowing reaction : 

15Ca  +  lOAl  +  6V2O5  =  SCa^Al^Os  +  12V 

The  degree  of  comminution  of  the  various  ingredients  in  a 
metallo-thermic  reaction  is  of  prime  importance.  Both  the  tem- 
perature and  speed  of  reduction  can  be  controlled  within  certain 
limits,  solely  by  regulating  the  sizes  of  the  different  elements 
and  compounds  used.  Dr.  Saklatwalla^*  has  shown  that  vanadium 
oxide  may  be  reduced  to  metal  in  the  form  of  ferro,  without  the 
excessive  formation  of  carbide  or  the  reduction  of  silica  to  siHcon, 
even  though  carbon  and  silica  be  present  in  large  amounts.  These 
results  were  obtained  solely  by  paying  attention  to  size  of  mate- 
rials. This  explains  why  certain  investigators  have  been  unable 
to  duplicate  the  work  of  others.  Different  sizes  of  materials 
were  used,  hence  a  different  temperature  and  rate  of  reduction. 

Prandtl  and  Bleyer  recommend  the  use  of  old  slag  where  more 
than  100  g.  of  vanadium  oxide  are  reduced,  to  keep  down  the  tem- 
perature ;  but  the  writer  obtained  better  results  when  no  slag  was 
used.  The  writer  also  used  dead  burned  magnesite,  fused  in  the 
electric  furnace  and  then  pulverized,  instead  of  fluorspar  in 
several  runs. 

The  writer  used  a  can  20  in.  (51  cm.)  high  and  12  in.  (30  cm.) 
in  diameter  and  rammed  the  fluorspar  around  a  tube  18  in.  (45 
cm.)  long  and  6  in.  (15  cm.)  in  diameter.  The  fluorspar  or  mag- 
nesite should  be  well  vented.  The  charge  consisted  of  600  g.  of 
small  calcium  shavings,  270  g.  of  minus  40  mesh  aluminum 
powder  and  1100  g.  of  80  mesh  vanadium  pentoxide  that  had 
been  freshly  fused  and  pulverized. 

A  considerable  portion  of  the  vanadium  entered  the  slag  as 
calcium  vanadate,  and  in  8  runs  the  best  metal  was  only  85  per 
cent  pure.  The  impurities  were  calcium  and  aluminum.  Remelt- 
ing  the  regulus  from  the  8  runs  in  an  electric  furnace  and  treating 
the  melt  with  more  vanadium  oxide,  removed  the  remaining  cal- 
cium and  aluminum,  but  the  product  contained  a  high  percentage 
of  vanadium  monoxide. 

"Trans.    Am.    Electrochem.    Soc.,   37,    341    (1920);   Jour.    Ind.    and   Eng.    Chem.    14, 
968-972. 


3o8  WILI,  BAUGHMAN. 

Vogel  and  Tammann^^  produced  vanadium  metal  95  per  cent 
pure  by  using  pure  dry  ammonium-free  vanadium  pentoxide  in 
the  regular  alumino-thermic  method.  The  writer  used  the 
apparatus  of  Prandtl  and  Bleyer,  described  above,  and  40-mesh 
aluminum  dust  with  80-mesh  vanadium  pentoxide.  Out  of  4 
runs  the  best  obtained  was  78  per  cent  vanadium  metal,  the 
balance  was  aluminum  with  a  little  vanadium  monoxide.  The 
point  anent  the  oxide  being  pure,  dry,  and  ammonium  free  is 
important.  Possibly  another  reason  why  the  writer  was  unable 
to  make  the  same  grade  of  vanadium  that  they  did,  is  that  his 
aluminum  dust  contained  some  sodium  and  oxygen.  The  sizes  of 
the  materials  used  may  have  been  different  also. 

The  writer  also  duplicated  the  methods  of  Prandtl  and  Manz" 
who  used  vanadium  trioxide  instead  of  pentoxide  for  the  calcium 
aluminum  reduction.  Vanadium  trioxide  gives  much  better 
results.  There  is  less  slag  loss,  and  as  a  result  of  4  runs  a  metal 
from  89  to  94  per  cent  pure  was  obtained,  which  on  treating  in 
the  electric  furnace  with  more  trioxide  gave  a  metal  96  per  cent 
pure.  Aside  from  the  Baughman  lithium  method  described  here- 
after, the  reduction  of  vanadium  trioxide  by  calcium  and  alumi- 
num gave  the  best  results  of  any  method  tried. 

Prandtl  and  Bleyer^^  also  produced  a  95  per  cent  metal  by 
using  100  parts  of  pure,  fused  and  pulverized  vanadium  pen- 
toxide, 49^  parts  of  aluminum  powder,  and  20  parts  fluorspar 
in  a  magnesium  crucible.  The  writer  attempted  to  duplicate  this 
but  obtained  only  an  81  per  cent  vanadium  metal. 

Ruff  and  Martin's  Methods.    They  describe^^  three  methods: 

1.  Reduction  of  trioxide  by  aluminum  and  a  small  amount  of 
carbon. 

2.  Reduction  by  carbon  in  the  electric  furnace. 

3.  Reduction  of  vanadium  trioxide  by  vanadium  carbide. 
None  of  these  methods  appealed  to  the  author  because  of  the 

use  of  carbon,  as  he  was  searching  for  a  way  of  preparing  a  car- 
bon-free product.  One  run  was  made  in  a  resistance  furnace  at 
a  temperature  of  about  1700°  C,  using  the  third  method.     Ovc" 

"  Z.  anorg.  chem.  64,  223 

^0  Ibid.   79,  209-22. 

'■  Ber.  43,  2602-3. 

*'  Z.  anorg.  chem.  25,  39-56. 


THE  METALI.URGY  OF  LEAD  VANADATES.  309 

three-fourths  of  the  vanadium  was  lost  by  volatiHzation  of  the  tri- 
oxide.  The  regulus  contained  86  per  cent  vanadium  metal,  and 
contained  both  carbon  and  oxide  as  vanadium  monoxide. 

The  writer  has  found  that  some  vanadium  monoxide  is  formed 
before  decarburization  is  complete  in  any  heat  where  decarburiz- 
ing  of  vanadium  carbide  is  attempted. 

Muthmann  and  Weiss  Method}^  This  method  consists  of 
reduction  with  "misch  metal,"  a  mixture  of  cerium  and  other 
rare  earth  metals.  The  writer  was  unable  to  obtain,  at  the  time 
of  these  experiments,  any  misch  metal  but  did  obtain  some 
cerium  metal.  He  made  three  100  g.  runs  with  the  best  product 
containing  84  per  cent  vanadium. 

Baughman's  Lithium  Method.  Theoretically,  lithium  should 
be  a  better  reducing  agent  than  calcium,  aluminum  or  cerium, 
as  shown  by  the  following: 

3V2O5  +  lOAl  =  5AI2O3  +  6V  +  638,500  cal. 

3V2O5  +  15Ca  =  15CaO  +  6V  +  648,000  cal. 

3V2O5  -f  7>4Ce  =r  7i/^Ce02  +  6V  +  360,000  cal. 

3V2O5  +  30Li  =  ISLi^O  +  6V  +  825,000  cal. 

This  proved  to  be  the  case.  The  writer  used  lithium  pellets 
about  the  size  of  BB  shot  in  the  same  apparatus  as  was  used  for 
the  previous  experiments  in  calcium  and  aluminum  reduction. 
A  metal  containing  95  to  97  per  cent  vanadium  is  readily  obtained. 
An  excess  of  vanadium  oxide  must  be  used,  as  a  considerable 
amount  is  lost  in  the  slag  as  lithium  vanadate.  The  reduction  is 
so  rapid,  however,  that  very  little  vanadium  is  lost  by  volatiliza- 
tion even  when  the  trioxide  is  used. 

Lithium  is  such  an  expensive  reducing  agent  that  the  writer 
then  turned  his  attention  to  making  metallic  lithium  from 
lepidolite,  lithia  mica,  of  which  there  are  large  deposits  in  Cali- 
fornia. He  finally  worked  out  a  method  of  producing  lithium 
metal  within  reasonable  cost  and  was  planning  to  use  the  lithium 
reduction  method  to  produce  large  amounts  of  vanadium  metal, 
when  the  thought  occurred  that  vanadium  oxide  might  be  reduced 
to  the  metal  by  the  same  method.  With  minor  changes  the 
method  was  successful.     It  was  also  found  applicable  to  reducing 

«Liebig  Ann.   337,  370;  355,   58. 


3IO  WILL   BAUGHMAN. 

titanium,  thorium,  uranium,  cerium,  and  tantalum  from  their 
oxides.  On  account  of  business  reasons  the  writer  cannot 
describe  this  method,  nor  the  technic  that  he  has  worked  out  of 
using  sodium  vanadate,  iron  oxide,  and  metal  and  aluminum  shot 
for  producing  ferrovanadium. 

Werner  von  Bolton's  Method.-'^  This  is  a  method  of  reducing 
columbium  or  tantalum  oxides  to  metal.  It  consists  in  preparing 
the  oxide  in  the  form  of  filaments  with  paraffine,  calcining,  and 
then  heating  by  an  electric  current  in  a  high  vacuum.  The  writer 
used  vanadium  trioxide,  which  is  a  conductor,  but  was  unable  to 
obtain  the  metal.  The  trioxide  was  reduced  to  monoxide  and 
dioxide  but  not  to  the  metal. 

The  writer  also  attempted  to  produce  metallic  vanadium  by- 
electrolysis  of  vanadyl  salts  with  a  mercury  cathode  cell,  in  the 
same  manner  that  metallic  chromium,  tungsten  and  molybdenum 
can  be  prepared.  The  experiments  all  gave  negative  results. 
Vanadium  does  not  form  amalgams  and  in  aqueous  solutions 
it  is  always  anodic  in  properties. 

Other  methods  for  making  metallic  vanadium  are  those  of 
Gin,  Beckman  and  Cowper  Coles.  Dr.  S.  Fischer^^  investigated 
Cowper  Coles'  electrolytic  method,  and  found  it  to  be  the  forma- 
tion of  a  coating  of  platinum  hydride  instead  of  metallic 
vanadium. 

Beckman' s  Method.  Dr.  Beckman's  method^^  consisted  of 
using  an  igneous  electrolyte  of  fused  calcium  oxide,  and  adding 
excess  vanadium  oxide  while  passing  direct  current.  The  writer 
used  the  furnace  described  before  for  smelting  the  Black  Buttes 
ore.  Instead  of  trying  to  produce  the  metal  the  writer  attempted 
to  produce  a  ferrovanadium.  Scrap  steel  weighing  50  lb.  (22.7 
kg.)  was  first  melted,  then  50  lb.  (22.7  kg.)  of  crude  calcium  vana- 
date charged  on  top  and  melted.  At  20  min.  periods  for  4  hr., 
20  lb.  (9  kg.)  of  vanadium  pentoxide  was  added.  At  the  end 
of  thfs  period  the  metal  was  tapped  and  cast  in  pigs.  It  con- 
tained 2.67  lb.  (1.21  kg.)  of  vanadium  metal  and  carbide.  A 
direct  current  of  about  500  amp.  at  80  volts  was  used. 

Dr.  Beckman   gave  no   operating   data   in   his  paper  on   this 

20  Zeit.  elecktrochem.   11,  4S   and   722. 

=•  Inst.  Min.  and  Met.  Eng.  1898-99  pp.   198-200. 

"Trans.  Am.  Electrochem.   Soc.   19,  171    (1911). 


THE  METALLURGY  OF  LEAD  VANADATES.  3II 

method,  and  as  apparently  insignificant  details  determine  the 
success  or  failure  in  this  class  of  work,  the  writer  decided  to 
drop  this  line  of  investigation. 

Gin's  Methods.  Gustave  Gin  describes  his  two  methods,  in 
detail,  in  his  "Memoir  on  Vanadium,"-^  to  which  the  reader  is 
referred.  The  first  method  consists  of  electrolyzing  molten  cal- 
cium and  vanadium  fluorides,  adding  vanadium  tetroxide  from 
time  to  time.  The  second  uses  a  calcium  and  ferrous  fluoride 
electrolyte,  and  vanadium  is  supplied  to  the  bath  by  special  anodes 
composed  of  vanadium  trioxide  and  carbon.  The  cathodes  in 
both  methods  are  iron,  copper  or  other  metal  with  which  it  is 
desired  to  alloy  the  vanadium,  or  lead,  which  is  later  volatilized  if 
vanadium  metal  is  desired.  This  latter  is  an  object  that  is  diffi- 
cult to  achieve.  The  methods  are  better  suited  for  producing 
ferrovanadiuni. 

It  was  the  writer's  privilege  to  be  Dr.  Gin's  assistant  when  he 
was  developing  these  two  processes.  In  modified  form  the  second 
method  was  later  used  at  the  works  of  Paul  Girod  at  Ungine, 
France.  Technically  both  methods  are  operative,  but  are  not  in 
wide  use  at  present  because  the  electrically  fused  alumina  linings 
often  failed  before  a  run  was  half  completed.  The  amount 
of  carbon  tetrafluoride  formed  at  the  anode,  while  not  large  in 
proportion  to  the  amount  of  fluorine  in  use  in  the  bath,  was  still 
enough  to  require  the  use  of  tight  fitting  goggles  and  aspirators 
by  the  furnace  operators.  Instead  of  using  calcium  fluoride  in 
the  second  method  calcium  vanadium  fluoride,  as  made  for  the 
first  process,  was  found  necessary  and  the  addition  of  tetroxide 
of  vanadium  was  found  desirable,  so  that  the  final  process  became 
a  combination  of  the  original  two.  The  cost  of  manufacture  by 
these  methods  is  high.  In  fact  it  can  not  compete  with  electric 
furnace  reduction,  using  silicon  as  reducing  agent,  or  with  the 
alumino-thermic  method. 


DISCUSSION. 

B.  D.   Saklatwalla'  :    The  first  thing  that  is  remarkable  is 
the  large  number  of  various  occurrences  which  the  author  de- 

-» Trans.  Am.  Electrochem.   Soc.  16,  439   (1909). 
*  Vanadium   Corp.   of  America,   Bridgeville.   Pa. 


312  DISCUSSION. 

scribes.  Vanadium  is  one  of  the  most  widely  disseminated  ele- 
ments that  we  know.  It  occurs  on  every  continent  of  the  globe. 
To  find  scattered  occurrences  of  vanadium,  therefore,  should  not 
appear  strange,  but  the  difficulty  has  been  that  we  do  not  find 
them  as  commercial  deposits.  They  are  of  an  erratic  nature  and 
do  not  persist. 

Now  the  metallurgy  of  lead  vanadates  has  not  been  commer- 
cially developed,  not  because  it  is  a  difficult  problem  from  a 
metallurgical  standpoint,  but  because  it  had  no  particular  com- 
mercial application. 

As  to  the  leaching  methods  and  treatment,  Mr.  Baughman  is 
right  when  he  considers  all  these  roundabout  leaching  processes 
as  not  commercial,  because  the  losses  are  high. 

He  then  describes  his  method  of  smelting  out  lead,  and  then 
taking  the  slag  and  fusing  it  with  sodium  hydrate  and  making 
a  sodium  vanadate.  I  am  inclined  to  believe  that  is  superfluous. 
The  slag  that  you  can  get  by  reducing  the  lead  out  of  the  lead 
vanadates  would  be  perfectly  amenable  to  reduction  directly, 
either  by  means  of  aluminum  or  silicon,  or  by  carbon  in  the 
electric  furnace.  So  the  problem  of  getting  vanadium  out  of 
lead  vanadates  is  not  a  difficult  metallurgical  problem.  It  has 
not  been  commercially  exploited  for  the  reason  that  there  are 
no  lead  vanadates  to  exploit  commercially.  But  at  the  present 
time,  since  the  radium  industry  has  been  practically  shut  down 
in  this  country,  and  which  was  a  source  of  vanadium  obtained 
as  a  by-product,  there  has  been  activity  in  development  of  other 
vanadium  minerals,  and  probably  this  impetus  to  search  might 
reveal  larger  deposits  of  lead  vanadates  or  other  vanadates. 

Colin  G.  Fink^:  Formerly  all  ferro-vanadium  was  made  by 
the  Goldsmith  process.  Dr.  Saklatwalla  has  recently  published 
a  paper  in  the  "Electrical  World,"^  on  the  production  of  ferro- 
vanadium  in  the  electric  furnace.  It  is  another  ferro-alloy  which 
has  submitted  to  electric  furnace  methods,  an  alloy  which  for 
years  has  been  thought  impossible  to  produce  by  any  but  the 
alumino-thermic  method. 

'  Consulting    Metallurgist,    New    York    City. 

'  Electric   FBrnace   makes   Ferro-Vanadium   by   B.   D.    Saklatwalla    and    A.    Anderson. 
Electrical  World,  February,   1923. 


THE  METALLURGY  OF  LEAD  VANADATES.  313 

W.  C.  Arsem':  About  sixteen  years  ago  I  made  some  vana- 
dium on  a  laboratory  scale  and  determined  the  melting  point  to 
be  1,650°  C.  This  was  made  by  reducing  the  tri-chloride  with 
magnesium,  in  a  vacuum,  similar  to  the  classic  research  followed 
by  Sir  Henry  Roscoe,  who  reduced  the  di-chloride  and  tri-chloride 
with  sodium  in  hydrogen. 

Will  Baughman  (Communicated)  :  The  statement  that  64  lead 
vanadate  deposits  show  commercial  possibilities  is  not  only  the 
writer's  opinion  but  is  based  upon  reports  made  by  competent 
mining  engineers,  familiar  with  the  characteristics  of  the  lead 
vanadates,  who  have  examined  a  majority  of  these  deposits  at 
various  times. 

Lead  vanadates  generally  occur  in  well  defined  veins  and 
should  not  be  confused  with  carnotite  or  roscoelite  deposits  that 
occur  in  small,  irregular,  scattered  pockets.  To  those  who  have 
made  a  study  of  the  genesis  of  the  lead  vanadates,  the  probable 
existence  of  lead  vanadate  ore  chutes  can  be  determined  with  as 
much  assurance  as  the  probable  existence  of  the  commoner  metals. 

No  vanadium  deposit  persists  in  depth.  All  stop  at  the  zone 
of  ground  waters.  In  the  arid  regions  of  United  States  this  may 
mean  a  considerable  depth.  At  least  one  lead  vanadate  mine 
extends  to  900  ft.  vertical  depth,  or  1300  ft.  on  the  ore  body. 
The  wonderfully  rich  and  unique  deposit  in  the  Peruvian  Andes 
is  no  exception.  In  fact  it  is  a  rather  shallow  and  superficial 
deposit,  the  zone  of  ground  waters  being  at  100  to  150  ft.  depth.^ 

The  lead  vanadate  deposits  of  United  States  have  not  been 
developed,  because  of  economic  conditions,  not  a  lack  of  potential 
ore.  Of  the  several  attempts  made  in  the  past  to  develop  these 
ores  all  failed  for  reasons  other  than  lack  of  ore,  save  one  project. 

The  lead  vanadate  miner  can  not  ship  his  concentrates  to  some 
treatment  plant.  He  must  refine,  manufacture  and  then  market 
his  product,  which  is  a  serious  undertaking. 

One  company  owns  a  deposit  of  ore  that  is  practically  free  of 
impurities,  that  is  readily  concentrated  by  mere  roasting  and 
which  contained  at  first  seven  times  as  much  vanadium  as  the  best 
run  of  mine  lead  vanadates.     Through  being  the  first  large  pro- 

*  Consulting    Chemical    Engr.,    Schenectady,    N.    Y. 

» Miller    and     Singewald.       Mineral     Deposits    of    South     America.       D      F      Hewett 
Vanadium  in  Peru.     Trans.  A.   I.  M.  E.  Vol.  40. 


SH  DISCUSSION. 

ducers,  and  selling  under  contract  systems,  this  company  offers 
a  problem  in  financing  and  competition,  for  which  the  writer 
knows  no  parallel. 

Before  the  discovery  of  the  unique  deposit  owned  by  this  com- 
pany, the  lead  vanadates  were  the  principal  source  of  supply. 
As  soon  as  this  deposit  is  reduced  to  low  grade  ore,  so  that  the 
production  costs  will  be  higher  than  they  were  a  few  years  ago, 
then  the  lead  vanadates  may  again  become  the  principal  source  of 
vanadium. 

Hewett's  description  of  this  property  shows  that  the  very  rich 
ore  occurred  as  shallow  gash  veins  in  a  lense  shaped  mass  300 
ft.  long,  28  ft.  wide,  and  200  ft.  on  slope  to  ground  waters.  On 
an  optimistic  basis  this  would  indicate  less  than  100,000  tons  of 
1  to  20  per  cent  ore,  while  consular  reports  show  over  12,000  tons 
of  40  per  cent  concentrates  have  been  shipped.  This  would 
indicate  that  this  deposit  is  approaching  exhaustion.  Also  the 
first  material  mined  ran  as  high  as  20  per  cent  vanadium  oxide, 
which  was  raised  to  as  high  as  80  per  cent  by  roasting.  This 
roasted  material  has  steadily  fallen  off  in  grade.  Consular  reports 
show  that  concentrated  material  recently  shipped  contained  only 
16  to  20  per  cent. 

During  the  period  that  the  highest  grade  ore  was  being  mined, 
this  company  sold  ferro  for  less  than  $2.50  per  lb.  of  vanadium 
content,  or  about  $1.00  less  than  the  writer  estimated  that  the 
more  favorably  situated  lead  vanadate  deposits  could  produce  it. 

On  the  other  hand,  the  use  of  vanadium  may  fall  off.  The 
same  development  of  electric  furnace  practice  that  allows  Dr. 
Saklatwalla  to  produce  ferro-vanadium  in  the  electric  furnace, 
has  also  made  it  possible  to  use  titanium,  or  other  cheap  nitro- 
gen and  oxygen  removers,  and  with  better  furnace  control,  pro- 
duce a  steel  for  many  purposes  superior  to  the  old  vanadium 
steel.  The  Ford  Motor  Co.,  formerly  one  of  the  largest  vanadium 
users,  has  used  little  for  some  time  past. 

The  writer  did  not  intend  to  infer  that  he  considered  all  the 
methods  discussed  in  the  paper  as  non-commercial.  He  considers 
the  chloride  volatilization  method,  ball  mill  grinding,  and  the 
ideas  of  producing  99  per  cent  vanadium  oxide  or  vanadium 
metal,  in  order  to  avoid  competition,  as  impractical.  He  considers 
chloridizing  roasting  for  ores  difificult  to  concentrate  and  low  in 


THE  METALLURGY  OF   LEAD  VANADATES.  315 

lime  or  magnesia,  sodimii  sulfide  leaching  for  similar  ores  high  in 
lime  or  magnesia,  and  the  smelting,  refining  method  for  concen- 
trates, as  methods  having  excellent  commercial  possibilities. 

Dr.  Saklatwalla  suggests  the  direct  reduction  of  slag,  from 
smelting  lead  vanadates,  to  ferro  vanadium.  The  lead  vanadates 
all  contain  one  or  more  of  the  elements  phosphorus,  molybdenum, 
arsenic,  tungsten,  copper,  and  chromium.  These  elements  would 
enter  the  final  product,  making  it  worthless.  Some  kind  of  refin- 
ing system  is  absolutely  necessary. 


A  paper  presented  at  tl\e  Forty-third 
General  Meeting  of  the  American  Elec- 
trochemical Society  held  in  New  York 
City,  May  5,  1923,  Dr.  F.  M.  Becket 
in   the  Chair. 


PREPARATION  OF  METALLIC  URANIUM/ 

By  R.  W.  MooRE.2 

Abstract. 
A  method  for  the  preparation  of  metallic  uranium  in  a  very 
pure  state  is  described,  also  a  method  of  fusing  the  metal  to  form 
buttons  or  small  pigs,  which  may  be  rolled  down  to  give  thin  sheets. 


For  the  preparation  of  this  metal  in  a  state  of  high  purity,  the 
old  method  of  the  reduction  of  the  anhydrous  chloride  with 
metalHc  sodium  seems  to  be  the  one  giving  the  best  results.  This 
method  has  been  used  by  a  number  of  investigators,  including 
Peligot,^  Zimmerman,*  Moissan,'  Mixter,®  Roderburg,^  Fischer,^ 
and  Lely  and  Hamburger.^ 

The  method  which  we  have  used  is  in  general  that  outlined  by 
the  last  named  investigators,  with  several  modifications  which 
make  for  simplification.  As  Lely  and  Hamburger  point  out, 
there  are  several  conditions  which  must  be  fulfilled  if  high  purity 
of  the  metal  is  to  be  attained.  These  are,  in  brief,  the  production 
of  the  chloride  in  a  pure,  dense  form,  which  does  not  take  up 
moisture  rapidly.  This  eliminates  the  action  of  water  during  the 
reduction,  and  the  attack  of  moist  chlorides  i.  e.,  hydrochloric 
acid)  on  the  reduction  bomb.  The  purer  the  chloride,  the  higher 
is  the  temperature  produced  during  reduction  and  the  coarser  the 
particles  of  metal  produced.  This  condition  is  desirable  since  it 
results  in  less  oxidation  taking  place  during  the  removal  of  the 

•  Manuscript  received  February   1,   1923. 

^  Research   Laboratory,   General    Electric   Co.,   Schenectady,   N.   Y. 

»Ann.   Chim.    Phys.    (4,)    17,   368. 

<  Ber.    deutsch.    Chem.    Ges.    13,    348    (1882). 

"Compt.   rend.   122,   1088. 

»Z.  anorg.   Chem.   78,   231    ri912). 

'Z.   anorg.  Chem.  81.   I,    122. 

«Z.    anorg.    Chem.   81,    II,    189. 

9  Z.   anorg.   Chem.   87,  209. 

317 


3l8  R.    W.    MOORE. 

other  products  of  the  reduction.  If  the  chloride  is  pure  the  heat 
of  the  reaction  is  sufficient  to  fuse  part  of  the  metal  product  into 
the  form  of  small  pellets, 

PREPARATIOX   OF  THE   CHLORIDE. 

The  preparation  of  UCI4  is  most  easily  carried  out  by  the  reac- 
tion of  SoCl,  on  uranium  oxide,  (UsOg),  a  method  similar  to 
that  used  by  Arsem^°  for  making  ThCl^  and  also  used  by  Matignon 
and  Bourion,^^  Colani/-  and  Lely  and  Hamburger.^^  An  easy 
method  of  carrying  out  this  reaction  was  found  to  be  as  follows : 
The  oxide  of  uranium  was  placed  in  quartz  or  porcelain  boats, 


r/G.  1 

/  ^^^/fe>  at/iss   rcB£. 
Z  C^fff£    t/ei « 
3  //vi.£-r  /^^^  ci- 

and  these  were  inserted  into  a  5  cm.  (2  in.)  porcelain  tube, 
resistance  furnace.  In  one  end  of  this  tube,  an  empty  boat  was 
placed  and  redistilled  SgClo  was  allowed  to  flow  into  this  drop  by 
drop  through  a  tube  connected  to  a  separatory  funnel.  The  other 
end  of  the  furnace  tube  was  closed  with  a  rubber  stopper  with  a 
large  outlet  tube  opening  under  S^Cl,  contained  in  a  bottle. 

The  furnace  was  inclined  towards  the  outlet  end  to  allow  any 
excess  SjCL  condensing  in  the  cool  end  of  the  tube  to  flow  out. 
The  tube  was  brought  up  to  200°-300''  C,  the  S^Clj  started  drop- 
ping in,  and  the  temperature  of  the  furnace  gradually  raised  to 
500°  C,  at  which  temperature  it  was  held  for  three  or  four  hours. 
Under  these  conditions  uranium  oxide  is  converted  to  a  greenish, 
coarse  crystalline  mass,  which  absorbs  moisture  only  slowly.    At 

"U.  S.  Patent  1,085.098. 
"Ann.   Chim.   Phys.    (8)   5,    127    (1905). 
"Ann.  Chim.   Phys.   (8)    12,   59   (1908). 
»»  hoc.  cit. 


PREPARATION   OF    METALLIC    URANIUM.  3I9 

500°  C,  the  UCI4  did  not  melt  nor  sublime,  but  remained  in  the 
boats  in  the  form  of  a  compact  mass  of  coarse  crystals.  It  still 
contained  some  oxide,  and  for  this  reason  required  sublimation. 

SUBLIMATION  OF  THE  UCL4. 

An  easy  and  convenient  method  of  carrying  out  this  sublimation 
was  found  to  be  as  follows :  A  large  hard  glass  tube,  about  4  cm. 
(1.6  in.)  in  diam.  was  bent  as  shown  in  Fig.  1.  The  tube  was 
filled  with  CI  which  was  bubbled  through  SgClg  to  make  sure  it 
was  dry.  The  UCI4  was  emptied  from  the  bottles,  in  which  it 
had  been  sealed,  directly  into  the  tube,  so  as  to  avoid  exposure  to 
the  air,  and  shaken  down  into  the  lower  end  of  the  bend. 

The  outlet  end  was  closed  by  a  stopper  carrj'ing  a  small  tube. 
The  part  of  the  tube  containing  the  UCl^  was  heated  to  a  bright 
red  heat,  with  a  moderate  current  of  CI  passing  through  it.  The 
UCI4  sublimed  in  the  form  of  dark  red  vapors,  which  deposited 
close  to  the  hot  zone  in  the  form  of  a  mass  of  coarse,  greenish 
crystals.  Besides  this  product,  there  was  formed  a  considerable 
amount  of  a  fluffy,  golden-yellow,  crystalline  substance  that  depos- 
ited in  the  cooler  part  of  the  tube.  This  was  apparently  an  addi- 
tion product  of  UCI4  with  SjCl,,  for  on  replacing  the  CI 
with  dry  N,  and  heating  the  tube  containing  these  crystals,  they 
were  decomposed  into  UCl^  and  SoClg.  The  UCI4  thus  sublimed 
was  poured  directly  into  a  bottle  containing  dry  N,  and  sealed 
until  it  was  used  for  reduction. 

REDUCTION. 

The  sodium  used  for  reduction  was  all  sublimed  in  vacuum  in 
an  apparatus  similar  to  that  suggested  by  Lely  and  Hamburger.^* 
This  was  arranged  so  that  the  cylinder  containing  the  sodium  was 
heated  in  vacuum  Avith  the  same  heating  arrangement  as  used 
later  for  heating  the  reduction  bomb.    See  Fig.  2» 

The  resublimed  sodium  was  cut  up  into  small  pieces  under 
redistilled  benzol,  which  had  stood  over  sodium  for  weeks.  It 
was  dried  in  a  vacuum,  and  opened  up  under  an  atmosphere  of  dry 
nitrogen.  About  25  per  cent  excess  Na  was  used  for  the  reduc- 
tion. The  UCI4  was  broken  up  into  small  lumps  in  an  atmosphere 
of  dry  nitrogen,  and  this  was  mixed  with  the  sodium  by  shaking 
in  a  bottle  filled  with  dry  N. 

1*  Loc.  cii. 


320 


R.   W.    MOORE. 


The  reduction  was  carried  out  in  a  steel  bomb  in  vacuum.  The 
simplest  manner  of  accomplishing  this  was  to  use  a  steel  cylinder 
closed  at  one  end,  with  a  steel  cap  screwed  in  the  other  end,  using 
a  fine  thread.  A  copper  gasket  was  used  under  the  cap,  and  the 
cap  was  screwed  down  by  hand.  This  arrangement  allowed  the 
gas  in  the  bomb  to  escape,  while  little  or  no  sodium  was  lost  dur- 
ing the  reduction.  The  bomb  was  exhausted  at  the  same  time  as 
the  chamber  in  which  it  was  heated.  The  arrangement  for  carry- 
ing out  the  reduction  is  shown  in  Fig.  2. 


3  c/fP  aP  Bo/^3 

4  CoPPsiP  G^s^rr 

5  BO/^B 

7   i.ow£-J?  P^^^  ar  y/rct/uPf    t^rsss'L. 

6  SfteSr     /^'CJrft.    S*^'i-i.O 

9  /JLOrvot/^  rva£ 

10  nt>--r BOS/vt/M    Wf£ 

11  ^Li/^o<//^  ru3£ 

IZ   /MSl/L/>r£0    TSV/^/K/liS 
/3  f/ff£  BPIC^ 

II   rMfPfe  ceuPi-e 
IS  ^iABS^Jf  sraPP£-jf 


Fig.  2 

The  bomb  was  first  filled  with  dry  nitrogen,  and  the  UCI4 
sodium  mixture  poured  in  and  pressed  down.  The  space  above 
the  charge  was  filled  with  dry  nitrogen,  the  bomb  closed  up  and 
placed  inside  the  heater  in  the  reduction  chamber.  This  was 
exhausted  to  about  25  microns,  and  the  bomb  gradually  heated 
until  a  sudden  rise  of  temperature  was  shown  by  the  pyrometer. 
The  bomb  was  then  cooled  under  vacuum  and  the  vacuum  broken 
with  dry  nitrogen. 


EXTRACTION  OF  THE  METAL. 

The  product  of  the  reduction  was  a  sintered  grayish  mass.    It 
contained  U,  NaCl,  Na,  with  possibly  some  small  amounts  of 


PREPARATION   OF    METALLIC    URANIUM.  32 1 

UCI4  and  uranium  oxide.  The  excess  sodium  was  removed  with 
absolute  alcohol,  the  NaCl  completely  washed  out  with  water,  and 
then  the  heavy  brownish  residue  washed  with  dilute  (2  per  cent) 
acetic  acid.  The  acid  was  washed  out  completely  with  water  and 
the  residue  washed  with  acetone  and  dried  in  vacuum.  This 
washing  was  carried  out  as  rapidly  as  possible  to  avoid  oxidation 
of  the  wet  metal.  It  was  found  necessary  to  break  the  vacuum 
after  the  metal  was  dry,  with  dry  nitrogen,  for  the  finer  portions 
of  the  metal  were  very  pyrophoric.  This  was  true  to  such  an 
extent  that  the  metal  could  not  be  transferred  from  one  container 
to  another  without  handling  it  entirely  in  an  inert  atmosphere, 
such  as  nitrogen. 

The  resulting  metal  was  a  very  heavy,  brownish  powder,  con- 
taining a  considerable  proportion  of  small  round  sintered  balls. 
The  yield  was  usually  above  90  per  cent.  The  coarser  portions 
of  the  metal  (remaining  on  80-mesh  screen)  were  quite  pure, 
analyzing  as  high  as  99.8  per  cent  uranium.  The  finer  portions 
were,  of  course,  more  affected  by  oxidation  during  the  washing, 
and  for  this  reason  were  not  so  pure. 

FUSION  OF  THE  METAL. 

Since  uranium  reacts  with  almost  all  gases  at  high  tempera- 
tures, and  alloys  readily  with  most  metals,  such  as  W,  Mo,  Fe,  Ni, 
etc.,  the  problem  of  melting  or  working  the  powdered  metal  into 
solid  form  ofifers  considerable  difficulty.  This  was  finally  accom- 
plished by  fusing  the  metal  on  a  water-cooled  table  with  an  arc  in 
an  atmosphere  of  argon  at  a  pressure  of  50  to  100  microns.  The 
apparatus  used  for  this  purpose  is  shown  in  Fig.  3. 

Pellets  or  discs  of  two  sizes  (about  2.5  cm.  and  3.75  cm.  diam. 
and  0.5  cm.  thick)  were  made  by  pressing  the  powdered  metal  in 
a  mold  using  a  hydraulic  press.  In  order  to  prevent  spontaneous 
ignition  of  the  metal,  the  mold  was  filled  with  N  and  the  metal 
powder  poured  into  the  mold  through  a  stream  of  N.  After 
pressing,  the  metal  no  longer  took  fire,  but  the  discs  were  pre- 
served in  an  atmosphere  of  dry  N  to  prevent  oxidation. 

The  large  discs  were  clamped  in  the  upper  electrode,  in  the 
apparatus   shown,  and  the   smaller  placed  on  the  water-cooled 


322 


R.   W.    MOORE. 


table ;  usually  two  discs  were  used,  placed  one  on  top  of  the  other. 
The  large  discs  were  first  sintered  by  placing  on  the  table  and 
passing  an  arc  over  them. 

After  the  air  had  been  exhausted  from  the  globe  to  0.5  micron, 
the  globe  was  washed  out  by  passing  in  argon  to  a  pressure  of  a 
few  mm.  and  again  exhausted.  Then  the  globe  was  filled  with 
argon  at  a  pressure  of  about  75  to  100  microns,  an  arc  started  by 
in  contact  with  the  discs  on  the  table,  and  then  the  arc  was  moved 


6  &t.^ss  at/4,  a 


Fig.  3. 


pushing  down  on  the  upper  electrode  to  bring  the  disc  of  uranium 
around  over  the  surface  of  the  discs  by  manipulating  the  upper 
electrode.  In  the  course  of  one  or  two  minutes  the  surface  of  the 
discs  could  be  all  brought  to  fusion.  The  discs  were  then  cooled, 
turned  over  on  the  table,  and  the  other  side  fused  as  before.  By 
using  care  not  to  keep  the  arc  on  too  long,  the  whole  pellet  could 
be  melted,  provided  too  much  oxide  was  not  present.  In  most 
cases,  there  was  sufficient  oxide  present  to  prevent  complete 
fusion ;  the  oxide  appeared  to  be  very  difficult  to  fuse. 


PREPARATION    OF    METALLIC    URANIUM.  323 

In  order  to  obtain  metal  nearly  free  from  oxide,  a  depression 
was  cut  in  the  water-cooled  table,  the  metal  discs  were  placed  on 
the  edge  of  this,  and  the  melted  uranium  caused  to  run  out  of 
the  unfused  portion  of  the  discs  into  the  depression  by  tilting  the 
table  slightly.  If  the  table  were  made  thin  and  kept  well  cooled, 
the  melted  uranium  solidified  immediately  in  the  depression 
without  attacking  the  metal  of  the  table,  which  was  made  of  iron 
or  monel  metal.  The  small  pigs  of  metal  formed  in  this  way 
were  at  times  remelted  on  a  smooth  table  to  form  fiat  smooth 
buttons. 

PROPERTIES  OF  METALLIC  URANIUM, 

The  metal  thus  formed  had  about  the  appearance  of  polished 
iron.  It  oxidized  quite  readily ;  a  brightly  polished  piece  became 
tarnished  quite  brown  when  exposed  to  air  for  two  or  three  days. 
The  metal  was  very  ductile ;  some  buttons  formed  as  above  were 
rolled  cold  from  a  thickness  of  about  5  mm.  to  small  sheets  about 
0.375  mm.  (0.015  in.)  thick. 


DISCUSSION. 

Chas.  a.  Doremus^:  I  desire  to  call  the  attention  of  the 
Society  to  the  fact  that  Fig.  3  is  substantially  Robert  Hare's  elec- 
tric furnace,  invented  in  1842,  and  which  was  described  to  this 
Society  some  years  ago,  and  later  by  Edgar  F.  Smith  in  his  book 
on  "Chemistry  in  America." 

In  Hare's  furnace,  the  top  electrode  was  movable.  He  made  a 
great  many  interesting  experiments.  It  was  unquestionably  the 
first  electric  furnace  in  this  country,  if  not  in  the  world. 

J.  W.  Harden  and  H.  C.  Rentschler^  :  This  paper  is  of  great 
interest  to  us,  since  we  are  interested  in  the  preparation  and  prop- 
erties of  certain  rare  metals.  This  discussion  is  intended  to  bring 
out  some  additional  questions  and  difficulties  which  are  found 
during  the  preparation  of  the  extremely  active  element  uranium. 

1  New  York  City. 

'Research    Lab.,   Westinghouse   Lamp    Co..   Piloomfield,    N.    J. 


324  DISCUSSION. 

We  have  many  samples  of  uranium  powder  by  practically  all  of 
the  methods  now  given  in  the  literature,  and  we  are  thoroughly 
familiar  with  the  process  used  by  Mr.  Moore. 

We  do  not  agree  that  the  reduction  of  the  chloride  with  metallic 
sodium  is  the  method  which  gives  the  best  results  for  the  prepara- 
tion of  uranium  or  indeed  of  some  other  metals,  such  as  thorium, 
zirconium,  etc.  Burger,^  for  example,  in  1908  describes  the 
reduction  of  uranium  oxide  with  calcium,  in  which  he  claims  a 
high  degree  of  purity.  This  is  the  same  method  which  was  used 
in  1904  by  Huppertz^  for  other  rare  metals.  The  advantage  of 
using  uranium  oxide  in  place  of  uranium  chloride  if  it  can  be 
satisfactorily  reduced  is  obvious,  since  the  difficulty  of  the  prepara- 
tion of  the  pure  dry  chloride  is  costly,  tedious  and  difficult.  The 
chloride  must  be  distilled  at  high  temperatures,  and  even  under 
the  most  exacting  of  conditions  it  seems  almost  impossible  to 
avoid  traces  of  oxy-chlorides  and  also  contamination  from  silica. 

In  the  reduction  of  the  chloride  with  sodium,  if  there  is  any 
trace  whatever  of  moisture  of  oxy-chloride  where  oxide  can  be 
formed,  the  sodium  will  not  reduce  the  oxide.  It  is  therefore 
next  to  impossible  to  get  100  per  cent  uranium,  that  is,  as  metallic 
uranium  free  from  uranium  oxide  in  this  way. 

Mr.  Moore  states  in  his  paper  that  his  uranium  powder  was 
brown.  We  all  know  that  molybdenum  powder  or  tungsten  pow- 
der or  other  metal  powders  when  in  reasonably  coarse  condition 
as  Mr.  Moore's  probably  was,  are  gray  and  not  brown.  The 
brown  color  indicates  the  presence  of  considerable  amounts  of 
oxide.  It  is  possible  to  have  a  sample  of  uranium  powder  which 
will  analyze  a  high  percentage  of  total  uranium,  but  which  will 
also,  if  figured  in  terms  of  oxide,  show  a  considerable  amount  of 
uranium  oxide  present,  or  a  much  lower  percentage  of  free  metal- 
lic uranium  than  of  total  uranium.  This  is  due,  of  course,  to  the 
high  atomic  weight  of  uranium  and  the  low  atomic  weight  of 
oxygen.  We  should  like  to  ask,  therefore,  if  Mr.  Moore  actually 
determined  the  percentage  of  oxygen  or  uranium  oxide  in  this 
sample.  Our  experience  indicates  to  us  that  a  sample  of  uranium, 
of  the  coarseness  indicated  by  Mr.  Moore,  would  have  a  consid- 

*  Burger;   Dissertation,   Basel    (1908). 
Mluppertz,  Chem.  Cent.  1,  1383   (1904). 


PREPARATION   OF    METALLIC   URANIUM.  325 

erable  percentage  of  oxide  if  it  had  much  of  a  brown  color,  per- 
haps 10  or  20  per  cent  of  oxide  figured  as  UgOs,  or  a  much  larger 
percentage  figured  as  a  lower  oxide.  We  have  made  many  sam- 
ples of  this  kind. 

We  should  also  like  to  inquire  whether  the  percentage  of  iron 
has  been  determined.  Mr.  Moore  has  given  no  data  in  his  paper 
with  regard  to  the  iron  content  of  his  uranium  sample,  which  we 
understand  was  made  in  an  iron  bomb.  Fisher  and  RideaP  and 
others  have  found  that  with  the  chloride  method  of  reduction, 
the  uranium  thus  produced  contains  a  considerable  amount,  say 
from  0.5  to  2  per  cent  of  iron.  We  should  like  to  know  if  the 
iron  has  been  actually  determined,  since  this  has  important  bear- 
ing on  the  methods  of  determining  the  purity  of  uranium  and 
its  apparent  melting  point.  If  there  is  much  iron  present,  and  the 
percentage  or  uranium  metal  is  determined  by  simply  burning  to 
oxide  and  getting  the  increased  weight,  the  larger  increase  due 
to  the  presence  of  a  small  amount  of  iron  would  make  up  for  the 
presence  of  a  considerable  amount  of  uranium  oxide  in  the  sample. 

Furthermore,  we  should  like  to  know  the  melting  point  of  the 
buttons  which  Mr.  ]\Ioore  has  prepared  by  arc-melting  on  pieces 
of  monel  metal.  We  infer  from  his  paper  that  the  melting  point 
of  the  beads  is  fairly  low,  and  should  like  to  point  out  that  when 
the  reduction  is  carried  out  in  an  iron  bomb,  there  is  no  difficulty 
in  getting  a  powder  which  consists,  partly  at  least,  of  little  beads 
of  apparently  fused  metal.  This  is  not  the  case  if  the  presence  of 
iron  is  excluded. 

It  is  hoped  that  a  paper  can  soon  be  published  describing  meth- 
ods by  which  uranium  powder  is  now  being  produced  in  the  Re- 
search Laboratory  of  the  Westinghouse  Lamp  Company,  which, 
as  has  been  stated,  is  not  brown  but  has  the  appearance  and  the 
fine  pressing  quality  of  a  good  sample  of  molybdenum  powder. 
This  kind  of  uranium  powder  can  be  pressed  into  any  desired 
shape,  and  with  proper  precautions  against  oxidation  and  spon- 
taneous combustion  in  the  air  can  be  sintered,  treated  into  bars  or 
into  solid  buttons  of  any  desired  shape  suitable  for  example  for 
X-ray  targets  or  other  commercial  purposes. 

In  conclusion,  the  present  writers  have  worked  with  the  method 

3Z.  anorg.  Chem.  81,   170,    (1913). 


326  DISCUSSION. 

described  by  Mr.  Moore,  but  were  not  able  to  obtain  samples  of 
uranium  powder  by  any  means  free  of  oxide.  It  is  character- 
istic of  uranium,  if  it  contains  even  very  small  quantities  of  almost 
any  kind  of  metallic  impurities,  that  when  heated  in  a  vacuum  or 
in  an  inert  environment  beads  of  low  melting  point  metal  separate 
and  run  away  from  the  remainder  of  the  mass.  We  experienced 
considerable  trouble  in  finding  a  suitable  substance  upon  which 
to  support  uranium  during  its  heat  treatment.  Uranium  alloys 
with  most  metals,  (even  tungsten)  under  proper  conditions,  and 
interacts  with  such  refractories  as  lime,  magnesia,  etc.  We  have 
devised  a  method  for  making  pure  thorium  oxide  crucibles,  which, 
when  properly  heat  treated,  have  served  excellently  for  this 
purpose. 

R.  W.  Moore  (Communicated)  :  In  reply  to  the  point  raised  by 
Dr.  Doremus,  I  would  say  that  there  was  no  intention  of  even 
suggesting  that  this  type  of  furnace  is  new.  Similar  furnaces 
have  been  used  for  various  purposes  in  the  past.  The  present 
paper  simply  shows  the  application  of  this  type  to  this  particular 
problem,  with  some  details  modified  to  fit  this  case.  The  method 
of  separating  the  metal  from  the  oxide  and  obtaining  a  small 
pig  of  pure  cast  metal  is  believed  to  be  new. 

Messrs.  Marden  and  Rentschler  bring  out  several  points  on 
which  there  may  be  differences  of  opinion.  As  regards  the  rela- 
tive values  of  the  methods  of  reduction  by  calcium  and  sodium, 
much  may  be  said.  It  would  seem  that  the  reduction  products 
from  the  sodium  reduction,  namely  NaCl,  excess  Na,  and  possibly 
undecomposed  UCI4,  should  be  much  more  easily  extracted  from 
the  uranium  metal  than  the  CaO  and  possibly  undecomposed  ura- 
nium oxide  produced  during  the  reduction  by  calcium.  In  the 
case  of  thorium,  with  which  we  tried  both  methods,  the  sodium 
reduction  gave  the  best  results. 

In  regard  to  the  difficulty  of  preparing  the  UCI4,  I  do  not  agree 
with  Messrs.  Marden  and  Rentschler.  It  is  not  particularly  costly, 
nor  tedious,  and  certainly  not  difficult.  It  does  require  care.  The 
sublimation  does  not  require  high  temperature ;  it  is  readily  carried 
out  in  a  hard  glass  tube. 


PREPARATION    OF    METALLIC    URANIUM.  327 

Messrs.  Marden  and  Rentschler  state  that  the  brown  color  of 
the  powdered  uranium  metal  indicates  the  presence  of  considerable 
amounts  of  oxide.  Unquestionably,  the  color  indicates  oxidation 
of  the  surface  of  the  particles.  Uranium  is  quite  easily  oxidized ; 
as  mentioned  in  the  paper,  even  rolled  and  worked  metal  pieces, 
when  polished  bright  will  turn  quite  brown  in  a  few  hours  in  the 
air.  Naturally,  during  the  process  of  washing  out  the  NaCl,  Na, 
etc.,  from  the  reduction  mass,  the  metal  is  constantly  subject  to 
oxidation,  and  the  particles  tarnish,  but  probably  only  on  the  sur- 
face, since  the  particles  are  dense,  and  the  oxygen  probably  does 
not  penetrate  to  any  appreciable  depth.  Since  our  analyses  show 
a  percentage  of  uranium  as  high  as  99.8  per  cent  this  would  seem 
to  be  the  case.  Of  course,  the  0.2  per  cent  oxygen  (assuming  the 
difference  to  be  all  oxygen)  may  mean  several  per  cent  of  oxide. 
The  fact  that  there  was  some  oxidation  of  the  particles  was  the 
reason  for  the  process  described  in  the  paper  of  melting  the 
metal  away  from  the  oxide  coating.  The  fact  that  the  metal  so 
obtained  was  very  ductile  would  seem  to  indicate  that  it  was  quite 
well  freed  from  oxide. 

It  would  seem  that  any  process  of  reduction  would  be  subject 
to  this  difficulty  of  oxidation  of  the  metal  during  extraction  of 
the  reduction  products.  If  Messrs.  Marden  and  Rentschler  have 
found  some  method  of  avoiding  this  I  shall  be  interested  to  learn 
what  it  is. 

In  regard  to  the  possible  content  of  iron  in  the  uranium,  I  would 
refer  to  the  paper  by  Messrs.  Lely  and  Hamburger,®  in  which  they 
state  that  if  the  uranium  chloride  (and  also  thorium  chloride) 
is  kept  dry,  there  is  no  trace  of  iron  in  the  metal  produced  in  the 
steel  bomb.  We  did  not  analyze  our  metal  for  iron,  there  being 
no  reason  to  expect  contamination  from  this  source.  After  many 
reductions,  the  bomb  used  showed  no  sign  of  any  attack  by  the 
reduction  materials.  Furthermore,  in  cases  where  a  small  amount 
of  iron  had  alloyed  with  the  melted  metal  during  fusion  of  the 
pellets  in  the  furnace  described,  the  resulting  alloy  was  very 
brittle.  The  metal  we  obtained  was  very  ductile.  This  would 
indicate  the  absence  of  any  considerable  amount  of  iron.  Although 

«  Z.  anorg.  Chem..  87,  209. 


328  DISCUSSION. 

we  have  no  positive  evidence  that  traces  of  iron  may  not  have 
been  present,  the  indications  are  that  it  was  probably  not  present 
Amounts  such  as  suggested  by  Messrs.  Harden  and  Rentschler 
are  entirely  out  of  the  question. 

The   melting   point   of    the   metal    produced    by   the    method 
described  in  the  foregoing  paper  has  not  as  yet  been  definitelv 


A  paper  presented  at  the  Forty-third 
General  Meeting  of  the  American  Elec- 
trochemical Society  held  in  New  York 
City.  May  5,  1923,  Dr.  F.  M.  Becket  in 
the  Chair. 


EXPERIMENTS  RELATIVE  TO  THE  DETERMINATION  OF  URANIUM 
BY  MEANS  OF  CUPFERRON.' 

By  Jas.  a.  Holi.aday  and  Thos.  R.  Cunningham.^ 

Abstract. 

A  description  is  given  of  experimental  work  concerning  the 
determination  of  uranium  by  precipitation  with  cupferron.  Data 
are  cited  to  prove  that  quadrivalent  uranium  can  be  quantitatively 
precipitated  by  cupferron  from  solutions  containing  from  4  to  8 
per  cent  of  H.SO^  (sp.  gr.  1.84),  that  aluminum,  calcium,  mag- 
nesium and  phosphorus  remain  in  solution  and  can  be  completely 
separated  from  the  uranium  by  filtration,  and  that  the  precipitate 
of  UCCeHsN.O-,)^  can  be  quantitatively  converted  to  U.O,  by 
ignition. 


PRELIMINARY  REMARKS. 

Recent  years  have  witnessed  a  marked  increase  in  interest  con- 
cerning uranium,  and  in  experimental  work  looking  to  the  dis- 
covery of  new  uses  for  its  compounds  and  alloys.  Coincident  with 
and  resulting  directly  from  this  activity  there  has  arisen  a  need 
for  more  satisfactory  analytical  methods  for  the  determination 
of  the  element.  Without  going  into  an  exhaustive  discussion  of 
the  present  state  of  the  art,  it  may  be  stated  that  although  several 
of  the  commonly  used  methods  are  capable  of  yielding  accurate 
results,  the  necessary  separations  are  accomplished  by  reactions 
requiring  numerous  time-consuming  and  laborious  re-precipita- 
tions, particularly  the  separation  of  uranium  from  vanadium  and 
of  uranium  from  aluminum.  The  experimental  work  described 
in  this  paper  had  for  its  object  the  development  of  a  procedure 

'  Manuscript   received   January    11,    1923. 

a  Union  Carbide  and  Carbon   Research  Labs.,  Inc.,  Long  Island  City,  N.   Y. 

22  329 


330  JAS.   A.    HOLLADAY   AND  THOS.   R.    CUNNINGHAM. 

free  from  these  objections,  /.  e.,  one  based  on  sharp,  clean-cut 
reactions. 

It  has  been  shown  by  W.  A.  Turner*  that  vanadium  can  be 
quantitatively  separated  from  uranium,  phosphorus,  and  arsenic 
by  precipitation  with  cupferron  in  a  10  per  cent  sulfuric  acid 
solution.  Under  these  conditions  aluminum,  calcium,  magnesium 
and  phosphorus,  impurities  usually  found  in  carnotite,  pass 
quantitatively  into  the  filtrate,  while  iron  (titanium  and  zirconium) 
is  completely  precipitated  wuth  the  vanadium.  The  reliability  of 
these  separations  has  been  confirmed  in  this  laboratory.  Recently 
V.  Auger*  has  gone  on  record  to  the  effect  that  quadrivalent 
uranium  can  be  quantitatively  precipitated  from  an  acid  solution 
by  cupferron  as  a  brown,  flocculent  precipitate  having  the  formula 
U(C6H5N202)4.  However,  this  article  makes  no  mention  of 
the  necessary  acidity  nor  of  the  behavior  of  aluminum,  calcium, 
magnesium,  zinc  and  phosphorus.  Believing  that  these  reactions 
might  prove  to  be  better  suited  to  the  separation  and  determina- 
tion of  uranium  and  vanadium  than  any  previously  proposed, 
experiments  were  carried  out  to  obtain  information  on  the  fol- 
lowing points. 

1.  To  confirm  Auger's  statement  that  quadrivalent  uranium  is 
precipitated  by  cupferron  in  acid  solutions,  and  to  find  out 
whether  uranium  is  also  precipitated  when  present  in  a  still  lower 
state  of  oxidation  than  U^^. 

2.  To  determine  within  what  limits  of  acidity  the  precipitation 
is  complete. 

3.  To  learn  whether  the  uranium  precipitate,  UCCgHsNoO,)*, 
can  be  quantitatively  converted  to  UgOg  by  ignition. 

4.  To  ascertain  whether  aluminum,  calcium,  magnesium,  phos- 
phorus, and  zinc  can  be  quantitatively  separated  from  uranium 
by  proper  regulation  of  the  acidity. 

PREPARATION   OF  STANDARD  SOLUTIONS. 

In  order  to  carry  out  the  proposed  study  of  the  reactions,  the 
following  standard  solutions  were  prepared: 

1.  Uranyl  Sulfate,  UOz{SO^)o,  Solution.  Prepared  by  dis- 
solving 1.2  grams  of  the  C.  P.  salt  in  water  and  making  the  solu- 

'Am.  J.  Sci.  42,  109-10   (1916). 
«Compt.  Rend..   170.  995-6   (1920). 


THE  DETERMINATION  OF  URANIUM.  331 

tion  up  to  500  cc.  in  an  accurately  calibrated  500  cc.  volumetric 
flask.  If  the  uranyl  sulfate,  UOaCSOJ^.  had  been  pure,  100  cc. 
of  the  solution  should  have  contained  0.1237  g.  of  uranium.  The 
actual  uranium  content  was  determined  by  the  following  three 
methods : 

(a)  Precipitation  as  (NHJ2U2O7  and  Weighing  as  UgOs- 
The  uranium  in  a  50  cc.  aliquot  part  of  the  solution  was  precipi- 
tated with  ammonium  hydroxide^  and  ignited  to  UsOg.  The 
weight  of  the  precipitate  of  UgOs  was  0.0750  g.,  corresponding  to 
0.1272  g.  U  in  100  cc.  of  the  solution. 

(b)  Precipitation  as  (NHJ2UO3V2O5 .  HgO  and  Weighing  as 
2UO3 .  V2O5.  The  uranium  in  a  25  cc.  ahquot  part  of  the  solution 
was  precipitated  as  ammonium  uranyl  vanadate,  and  ignited  and 
weighed  as  2UO3 .  V2O5  according  to  Blair.«  The  result  obtained 
by  this  method  was  0.1276  g.  of  U  in  100  cc.  of  the  solution. 

(c)  Reduction  with  Zinc  and  Titration  with  0.1  N  KMn04. 
A  50  cc.  aliquot  part  of  the  solution  was  acidified  with  6  cc,  of 
H2SO4  (sp.  gr.  1.84),  diluted  to  100  cc.  cooled  to  room  tempera- 
ture, and  passed  through  a  Jones  reductor  having  a  zinc  column 
about  25  cm.  (10  in.)  long.  The  uranium  was  completely  removed 
from  the  reductor  by  the  use  of  125  cc.  of  water.  Approximately 
six  minutes  were  consumed  in  passing  the  solution  and  washings 
through  the  reductor.  When  the  amount  of  uranium  to  be 
reduced  exceeds  about  0.3  g.,  a  preliminary  reduction  in  the 
beaker  with  5  g.  of  zinc  is  necessary.  The  solution  was  vigor- 
ously stirred  for  1.5  minutes  to  re-oxidize  the  small  amount  of 
uranium  reduced  below  the  uranous  (11(804)2)  state,  and 
titrated  with  0.1  A/"  KMn04  that  had  been  standardized  against 
Bureau  of  Standard's  sodium  oxalate.  By  this  procedure  the 
result  0.0631  U  in  50  cc,  or  0.1262  in  100  cc,  was  obtained. 

A  resume  of  the  results  obtained  by  the  three  methods  follows : 

Grams  U  in  100  cc. 

Amount  theoretically  present    0.1237 

Amount  found  by  weighing  UsOs  0.1272 

Amount  found  by  weighing  2110.3. V20:i    0.1276 

Amount  found  by  zinc  reduction 0.1262 

Average    0.1270 

»  C.  A.   Pierle,  J.   Ind.   and   Eng.   Chem.,    12,    1,   60. 
•  "Cliemical   Analysis  of  Iron,"  ji.   210. 


332  JAS.   A.    HOLLADAY   AND  THOS.   R.    CUNNINGHAM. 

2.  Sodium  Vanadate  Solution.  Prepared  by  covering  2  g.  of 
pure  VoOj  with  hot  water,  and  adding  Na202  in  small  amounts 
until  the  VjOj  had  dissolved.  The  resulting  solution  was  boiled, 
filtered,  and  made  up  to  500  cc.  in  a  volumetric  flask.  If  the 
V2O3  had  been  pure  the  vanadium  value  of  100  cc.  of  the  solution 
should  have  been  0.2240  g.  A  determination  made  of  a  25  cc. 
aliquot  part  of  the  solution  by  reduction  with  H2O2  in  concentrated 
sulfuric  acid  solution,  followed  by  titration  with  0.1  A/"  KMn04 
(Cain  and  Hostetter's  method)  yielded  the  result  0.2020  g.,  while 
another  determination  made  by  passing  a  25  cc.  aliquot  portion, 
acidified  with  6  cc.  of  H2SO4  (sp.  gr,  1.84)  and  diluted  to  100  cc, 
through  a  Jones  reductor  into  ferric  phosphate  solution  and 
titrating  with  0.05  A^  KMn04,  gave  an  identical  result. 

3.  Phosphorus  Solution.  Prepared  by  dissolving  0.1065  g.  of 
ammonium  phosphate,  (NH4)2HP04,  in  250  cc.  of  water  in  a 
volumetric  flask.  The  phosphorus  content  of  a  25  cc.  aliquot  part 
of  this  solution  was  determined  by  precipitating  with  "molybdate 
solution"  and  filtering  and  washing  the  ammonium  phospho- 
molybdate.  The  "yellow  precipitate"  was  subsequently  dissolved 
in  NH4OH,  acidified  with  H2SO4,  and  the  resulting  solution 
passed  through  a  Jones  reductor  into  ferric  phosphate  solution 
and  titrated  with  a  solution  of  KMn04  (1  cc.  =r  0.0000431  g.  P) 
that  had  been  standardized  against  Bureau  of  Standards  sodium 
oxalate.  The  actual  phosphorus  value  of  the  solution  was  found 
to  be  0.000114  g.  per  cc.  as  against  the  theoretical  of  0.0001  g. 

4.  Aluminum  Solution.  Prepared  by  dissolving  8.9  g.  of 
ALClg .  I2H2O  in  water,  adding  H2SO4,  evaporating  until  all  free 
H2SO4  had  been  expelled,  dissolving  in  water,  and  filtering  and 
making  up  to  500  cc.  in  a  volumetric  flask.  One  hundred  cc.  of 
the  solution  was  found  to  contain  0.2014  g.  of  aluminum. 

5.  Calcium  and  Magnesium  Sulfates.  In  the  experiments 
where  known  amounts  of  calcium  and  magnesium  were  added 
weighed  amounts  were  employed  of  the  c.  p.  salts,  CaS04  .  2H2O 
and  UgSO, .  7U,0. 

GENERAL  DESCRIPTION  OF  EXPERIMENTS. 

"Synthetic"  solutions  containing  known  amounts  of  one  or 
more  of  the  elements  under  consideration — uranium,  vanadium, 


THE  DETERMINATION  OF  URANIUM.  333 

aluminum,  calcium,  magnesium,  phosphorus  and  zinc — were  pre- 
pared by  measuring  with  accurately  calibrated  pipettes  aliquot 
portions  of  the  standard  solutions  or  in  a  few  instances  (calcium 
and  magnesium)  by  weighing  the  salts.  When  vanadium  was 
present,  the  "synthetic"  solution  (volume  100  cc.)  was  acidified 
with  12  cc.  of  H2SO4  (sp.  gr.  1.84),  treated  with  enough 
KMn04  (approximately  0.1  A'')  to  give  a  permanent  pink  color, 
and  cooled  to  10°  C.  The  vanadium  was  precipitated  by  addition 
of  an  excess  of  a  cold  6  per  cent  solution  of  cupferron  and  the 
precipitate  (mixed  with  paper  pulp)  was  filtered  and  washed  with 
cold  10  per  cent  HoSO^  containing  1.5  g.  of  cupferron  per  L. 

If  the  determination  of  vanadium  was  a  part  of  the  program, 
the  paper  holding  the  cupferron  precipitate  was  dropped  into  an 
Erlenmeyer  flask,  and  treated  with  30  cc.  of  HjSO^  (sp.  gr.  1.84) 
and  10  cc.  of  HNO3  (sp.  gr.  1.42)  and  evaporated  to  fumes. 
After  several  successive  evaporations  with  10  cc.  portions  of 
HNO3  to  destroy  carbonaceous  matter  and  one  evaporation  with 
10  cc.  of  water  to  expel  every  trace  of  HNO.,,  the  vanadium  was 
reduced  with  H.O,  and  titrated  with  0.05  N  KMnO^  (Cain  and 
Hostetter's  method).  The  filtrate  from  the  cupferron  precipitate 
was  evaporated  to  a  volume  of  about  50  cc,  20  cc.,  of  HNO... 
(sp.  gr.  1.42)  were  added,  and  the  evaporation  was  continued 
until  clouds  of  sulfur  trioxide  were  evolved.  A  second  evapora- 
tion with  HNO3  was  made  to  destroy  all  organic  matter,  and  the 
solution  was  finally  evaporated  with  10  cc.  of  water  to  remove 
all  HNO3.  The  solution  was  then  diluted  with  the  volume  of 
water  necessary  to  give  the  desired  acidity — for  example,  137  cc. 
of  water  if  8  per  cent  acidity  was  desired — cooled  to  room  tem- 
perature, and  passed  through  a  Jones  reductor  in  the  manner 
previously  described,  the  reductor  then  being  washed  with  100  cc. 
of  the  same  strength  (8  per  cent  in  the  example  cited)  sulfuric 
acid.  The  quadrivalent  uranium  solution  was  finally  cooled  to 
5°-10°  C.  and  treated  with  an  excess  of  a  freshly  prepared 
6  per  cent  solution  of  cupferron. 

The  precipitate  does  not  begin  to  form  until  from  5  to  10  cc. 
of  cupferron  have  been  added.  Some  ashless  paper  pulp  was  intro- 
duced and  the  brown  precipitate  was  filtered  on  an  11  cm.  paper. 
The  precipitate  of  U(CH..N202)4  was  washed  with  cold  5  per 
cent  H2SO4,  containing  1.5  g.  of  cupferron  per  L.,  and  ignited  in 


334  JAS.   A.    HOLLADAY   AND  THOS.   R.    CUNNINGHAM. 

a  weighed  platinum  crucible,  first  at  a  low  temperature  and  then 
at  1000-1050°  C.  in  an  electric  muffle  furnace  into  which  a  cur- 
rent of  oxygen  was  passed.  The  crucible  and  precipitate  were 
then  cooled  and  weighed  and  the  amount  of  uranium  was  cal- 
culated from  the  weight  of  UgOg.  As  a  check  on  the  gravimetric 
method  the  precipitate  was  fused  with  K2S2O7,  dissolved  in  100 
cc.  of  6  per  cent  H2SO4,  and  the  uranium  determined  by  passing 
the  cold  solution  through  a  Jones  reductor  and  titrating  with  0.1 
N  KMnOi  as  previously  outlined. 

Experience  indicated  that  high  uranium  results  are  always 
obtained  when  uranium  oxide  (or  ammonium  di-uranate)  is  dis- 
solved in  HNO3,  and  evaporated  to  fumes  with  H2SO4,  prelimi- 
nary to  reduction  with  zinc  and  titration  with  KMnO^.  Addition 
of  water  and  evaporation  to  fumes  a  second  time  does  not 
eliminate  the  error,  which  is  apparently  due  to  obstinate  retention 
of  nitric  acid  by  the  uranium  compound  and  subsequent  reduc- 
tion of  the  nitrate  to  hydroxylamine,  NH2OH,  which  is  oxidized 
by  KMn04.  Uranium  compounds  should  therefore  be  dissolved 
in  H2SO4,  or  fused  with  K2S2O7  and  dissolved  in  H2SO4,  rather 
than  dissolved  in  HNO3  and  evaporated  with  H2SO4,  as  a  pre- 
liminary to  passage  through  the  Jones  reductor. 

When  elements  such  as  aluminum,  phosphorus,  etc.,  were 
present  in  the  "synthetic"  solutions,  the  filtrate  from  the 
U(C6H5N202)4  was  evaporated  with  HNO3  as  already  described, 
and  the  elements  in  question  were  determined  by  the  usual 
methods. 

1.    Experiments  to  Determine  Whether  Uranium  in  the  Quadri- 
valent Form  or  in  a  Still  Lower  State  of  Oxidation  is 
Quantitatively  Precipitated  by  Cupferron. 

The  factors  that  were  kept  constant  in  these  experiments  were : 

1.  The  amount  of  uranium  present,  0.0842  g.  in  each  case. 

2.  The  volume  and  acidity.  In  each  experiment  100  cc.  of  the 
solution  containing  6  cc.  of  H2SO4  (sp.  gr.  1.84)  were  passed 
through  the  reductor,  which  was  then  washed  with  100  cc.  of  6 
per  cent  H2SO4.  The  uranium  was  therefore  precipitated  from  a 
solution  having  a  volume  of  200  cc.  and  containing  12  cc.  of 
H2SO4  (sp.  gr.  1.84). 


THE  DETERMINATION  OF  URANIUM, 


335 


3.  Time  of  passage  of  solution  and  washings  through  redac- 
tor, about  six  minutes. 

The  conditions  that  were  varied  were : 

1.  The  temperature  of  the  solution  passed  through  the  reduc- 
tor.  In  Experiments  Nos.  1  and  2  the  solutions  were  at  room 
temperature,  while  in  No.  3  the  solution  was  heated  to  boiling 
previous  to  reduction  to  increase  the  amount  of  uranium  reduced 
below  the  quadrivalent  form. 

2.  In  Experiment  No.  2  the  solution  was  given  a  preliminary 
reduction  with  2  g.  of  zinc  before  being  put  through  the  reductor, 
the  object  being  to  reduce  as  much  uranium  as  possible  to  a  lower 
state  of  oxidation  than  the  quadrivalent  form.  Similarly,  in 
Experiment  No.  1  the  reduced  solution  was  vigorously  stirred  for 
three  minutes  before  precipitating  the  uranium  with  cupferron, 
while  in  Experiments  2  and  3  the  uranium  was  precipitated 
immediately  after  the  reduction. 


Expt. 
No. 


Acid- 
ity 
Per 

Cent. 


Weight  of  U 
Used 


0.0842 
0.0842 
0.0842 


Weight   of  U   Found 

by   Weighing 

the    UaOs 


0.0843 
0.0840 
0.0844 


Weight  of  U  Found 

by  Zn  Reduction 

and  KMnOi 

Titration 

g. 


0.0842 
0.0835 
0.0837 


These  results  constitute  reasonably  conclusive  evidence  that : 

1.  Auger's  statement  that  quadrivalent  uranium  can  be  quan- 
titatively precipitated  with  cupferron  is  correct. 

2.  That  uranium  present  in  a  lower  state  of  oxidation  than  the 
quadrivalent  form  is  also  precipitated. 

3.  That  the  precipitation  can  be  made  in  a  solution  contain- 
ing 6  cc.  of  H2SO4  (sp.  gr.  1.84)  per  100  cc. 

4.  That  when  the  precipitation  is  made  from  a  solution  con- 
taining 6  cc.  of  H,SO,  (sp.  gr.  1.84)  per  100  cc.  the  presence  of 
zinc    sulfate    does    not    lead    to    any    contamination     of    the 

U(CeH,NA)4. 

5.  That  the  precipitate  of  UCCeHsNoO.)^  can  be  quantitatively 

converted  to  U.O,. 


336 


JAS.   A.    HOLIvADAY   AND  THOS.   R.    CUNNINGHAM. 


The  following  two  additional  experiments  furnish  further  con- 
firmation of  the  above  statements  and  also  illustrate  the  separa- 
tion of  uranium  from  vanadium : 


Expt. 

No. 

1 
Acid-    Weight  of  U 
Pe^             Used 
Cent.              ^• 

Weight  of  U 

Found  by 

Weighing 

the    UjOg 

g- 

Weight  of  U 

Found   by 

Zn  Reduction 

and   KMnO* 

Titration 

g. 

Weight    of 

Used 
g> 

Weight    of 

V 

Found 

g- 

4 

S 

6 
6 

0.0127 
0.1270 

0.0124 
0.1264 

0.0123 
0.1262 

0.1300 
0.0260 

0.1295 
0.0260 

2.    Experiments  to  Determine  Within  What  Limits  of  Acidity 
the  Precipitation  is  Complete. 

The  experiments  already  cited  under  (1)  show  that  complete 
precipitation  is  obtained  from  a  solution  containing  6  cc.  of 
H2SO4  (sp.  gr.  1.84)  per  100  cc.  and  the  results  of  experiments 
shown  under  (4)  prove  that  an  acidity  as  high  as  8  per  cent  can 
be  successfully  employed.  Qualitative  tests  showed  that  when 
the  concentration  of  H2SO4  was  increased  much  above  8  per  cent 
the  uranium  is  not  completely  precipitated.  Inasmuch  as  the 
results  of  the  experiments  shown  under  (4)  prove  that  sharp 
separations  from  the  accompanying  impurities  can  be  obtained  in 
a  6  per  cent  H2SO4  solution,  there  is  no  reason  for  using  higher 
concentrations  than  6  per  cent  or  7  per  cent. 

3.  Experiments  to  Determine  Whether  the  Uranium  Precipitate, 
U{CQHr^No0.j)i,  Can  be  Quantitatively  Converted  to  U^O^ 

by  Ignition. 

The  results  of  numerous  experiments  tabulated  under  1,  2  and 
4  prove  that  the  precipitate  can  be  quantitatively  ignited  to  UaOs- 

4.  Experiments  to  Determine  Whether  Aluminum,  Phosphorus, 
Calcium  and  Magnesium  can  be  Quantitatively  Separated 

from,  Uranium  by  Proper  Regulation  of  the  Acidity. 

(See   Table  on  next  page.) 

Experiments  6  and  7  show  that  if  the  acidity  is  reduced  as  low 
as  2  per  cent  or  3  per  cent  the  uranium  precipitate  drags  down 
aluminum  and  probably  phosphorus,  while  Experiment  8  proves 


THE  DETERMINATION  OF  URANIUM. 


337 


that  a  sharp  separation  of  uranium  from  aluminum  and  phos- 
phorus is  obtained  with  an  acidity  of  4  per  cent.  Experiments 
9  and  10  show  that  uranium  can  be- quantitatively  precipitated 
from  a  6  per  cent  H2SO4  solution,  and  that  aluminum  is  not 
carried  down,  while  Experiments  11  to  14  inclusive  illustrate  the 
separation  from  aluminum,  calcium,  magnesium,  and  phosphorus 
under  similar  conditions.  Experiments  15  and  16  prove  that  the 
precipitation  of  uranium  is  complete  with  an  acidity  of  8  per  cent. 


Acid- 

Kxpt. 

ity 

No. 

Per 

Cent 

6 

2 

7 

3     ' 

8 

4 

9 

6 

10 

6 

11* 

6 

12* 

6 

13* 

6    ; 

14* 

6 

15* 

8 

16 

'  i 

Added 


Uranium 


Found 

as 
UtOg 

8- 


\'anadiuin 


Phosphorus 


0.210s 

0.2105 
0.1684 
0.0318 
0.0635 
0.0635 
0.0635 
0.4210 
0.2105 
0.4210 
0.1684 


Found 
with  Zn 

and 
KMnO* 

8' 


Added 
8- 


Found 
g* 


Added 
g. 


Found 
g- 


Wt.  of 

AI2O3 
Added 
g- 


0.2188 
0.2137 
0.1679 
0.0316 
0.0628 
0.0625 
0.0636 
0.4210 
0.2103 
0.4208 
0.1688 


0.2108 
0.2097 
0.1675 
0.0314 
0.0622 
0.0624 
0.0628 
0.4194 
0.2100 
0.4200 
0.1681 


0.00057 

0.00057 
0.00114 


0.0260 
0.1300 
0.1010 
0.0202 


0.0260 

0.1622 ; 

0.0203 


0.00114 

0.00342 
0.00285 
0.00285 
0.00285 
0.00114 


0.1140 

0.1140 

0.00114  0.1980 
0.0760 
0.1885 
0.1885 
0.1885 
0.1885 
0.1885 
0.0580 
j0.1980 


0.00112 
0.00340 
0.00284 
0.00288 
0.00285 
0.00116 


•  In  experiments  11  to  IS  inclusive  there  were  present  in  addition  to  the  elements 
shown  in  the  tabulation  0.05  g.  CaO  and  0.03  g.  MgO.  Zinc  sulphate  was  of  course 
present   in   all   of   the   experiments. 


Experiments  11  and  12  also  illustrate  the  accuracy  of  the  separa- 
tion of  vanadium  from  uranium,  aluminum,  calcium,  magnesium, 
and  phosphorus. 

The  following  tabulations  show  the  averages  of  the  uranium 
and  vanadium  results  obtained  in  the  entire  series  of  experiments 
excepting  Nos.  6  and  7. 

Uranium 

Average  weight  used    0.1431  g. 

Average  weight  found  by  weighing  the  UaOs 0.1429  g. 

Average  weight  found  by  Zn  reduction  and  KMn04  titration. 0.1424  g. 

Vanadium 

Average  weight  used    0.0606  g. 

Average  weight  found  0.0608  g. 

23 


338  JAS.   A.    HOLLADAY   AND  THOS.   R.    CUNNINGHAM. 

CONCLUSIONS. 

Study  of  the  results  of  these  experiments  leads  to  the  following 
conclusions : 

1.  Quadrivalent  uranium,  or  uranium  in  a  lower  state  of  oxi- 
dation than  the  quadrivalent  form,  can  be  completely  precipitated 
with  a  freshly  prepared  solution  of  cupferron  from  solutions  con- 
taining from  2  to  8  cc.  of  H2SO4  (sp.  gr.  1.84)  in  100  cc. 

2.  The  precipitate  can  be  quantitatively  converted  to  U3O,  by 
ignition. 

3.  If  the  amount  of  sulfuric  acid  in  the  solution  is  less  than 
4  cc.  per  100  cc,  aluminum,  and  probably  phosphorus,  will  be 
carried  down  with  the  uranium,  while  if  the  acidity  exceeds  8  cc. 
the  uranium  will  not  be  completely  precipitated.  If  the  acidity  be 
maintained  between  4  and  8  per  cent  (preferably  at  about  6  per 
cent)  a  sharp  separation  of  uranium  from  aluminum,  zinc,  calcium, 
magnesium,  and  phosphorus  can  be  obtained  by  a  single  precipita- 
tion. 

A  large  amount  of  convincing  data  has  been  presented  to  prove 
that  uranium  and  vanadium  can  be  separated  and  determined  with 
a  satisfactory  degree  of  accuracy  in  the  presence  of  widely  vary- 
ing amounts  of  iron,  aluminum,  calcium,  magnesium  and  phos- 
phorus, by  a  process  involving  the  following  steps : 

1.  Precipitation  of  the  vanadium  and  iron  from  a  12  per  cent 
H.SO4  solution  in  which  the  uranium,  vanadium,  and  iron  are 
present  in  the  higher  states  of  oxidation,  vis.,  in  the  sexivalent, 
pentavalent  and  trivalent  forms.  Uranium,  aluminum,  calcium 
(unless  present  in  amount  sufficient  to  precipitate  out  as  the 
difficultly  soluble  CaSO^),  magnesium,  and  phosphorus  pass  quan- 
titatively into  the  filtrate  when  the  solution  is  filtered.  Vanadium 
can  be  determined  in  this  precipitate  by  any  of  the  usual  methods. 

2.  Destruction  of  the  cupferron  by  evaporation  of  the  filtrate 
with  nitric  acid. 

3.  Reduction  of  the  uranium  by  passage  of  the  solution 
through  a  Jones  reductor.  The  zinc  sulphate  introduced  into 
the  solution  does  not  interfere  with  the  subsequent  reactions. 


THE  DETERMINATION  OF  URANIUM.  339 

4.  Precipitation  of  the  uranium  from  a  6  per  cent  H0SO4  solu- 
tion with  cupferron,  followed  by  filtration  and  washing  to  remove 
aluminum,  zinc,  calcium,  magnesium  and  phosphorus. 

5.  Ignition  of  the  uranium  precipitate  to  UsOg.  After  having 
weighed  the  UgOg,  its  uranium  content  may  be  checked  by  fusing 
it  with  K2S2O7,  dissolving  the  fusion  in  H2SO4,  and  passing  the 
solution  through  a  Jones  reductor  and  titrating  it  with  0.1  N 
KMnO,. 

The  reactions  upon  which  this  method  is  based  are  more  sharp 
cut  and  dependable  than  any  other  with  which  the  writers  are 
familiar.  The  procedure  has  been  applied  to  the  analysis  of 
uranium  and  vanadium  ores  and  alloys  with  excellent  results. 


A  paper  presented  at  the  Forty-third 
General  Meeting  of  the  American  Elec- 
trochemical Society  held  in  New  York 
City,  May  4,  1923,  President  Schlueder- 
berg   in    the   Chair. 


COBALT— ITS  PRODUCTION  AND  USES.* 

By   C.    W.   Drurv.2 

Abstract. 

In  the  preparation  of  the  following  paper  dealing  with  the 
production  and  uses  of  the  metal  cobalt,  an  attempt  has  been  made 
to  review  only  the  essential  points.  An  extensive  study  of  the 
occurrences,  metallurgy,  uses,  and  alloys  of  the  metal  cobalt  was 
published  recently,^  and  the  reader  is  referred  to  that  report 
for  any  detailed  information. 


In  Table  I  is  given  the  production  of  cobalt  ores  in  the  various 
countries.  The  figures  in  the  table  are  interesting  since  they 
show  the  history  of  the  mining  of  cobalt.  The  deposits  of  cobalt 
were  small  and  often  the  metal  w-as  obtained  as  a  by-product. 
Austria  was  the  first  large  producer,  followed  by  Germany,  Nor- 
way and  Sweden,  Spain,  Germany,  New  Caledonia  and  Canada. 
The  consumption  of  cobalt  compounds  has  gradually  increased 
from  that  sufficient  to  supply  the  pottery  industries  of  Europe, 
which  would  perhaps  be  approximately  10  tons  in  1860,  to  a 
world's  supply  of  400  tons  in  1920.  The  use  of  cobalt  compounds 
for  coloring  glass  has  been  known  by  the  Chinese  for  perhaps  60 
or  75  years.  No  record  is  available  of  the  production,  but  it 
must  have  been  small.  At  present  the  Chinese  import  cobalt 
supplies  from  America.  A  few  remarks  summarizing  the  metal- 
lurgy of  cobalt  have  been  incorporated  in  the  paper. 

In  Table  II  a  list  of  the  commercial  compounds  of  cobalt,  show- 
ing the  range  in  cobalt  content,  has  been  prepared. 

'  Manuscriot    received   February   24,    1923. 

'  Professor  of  Metallurgical   Research,  Queen's  University,  Kingston,  Ont. 

'  Report  of  Ontario  Bureau  of  Mines,   XXVII,   Part  3,   1918. 

341 


342 


C.  W.  DRURY. 


ORES. 

The  mining  of  cobalt  ores  has  been  carried  on  in  Europe  for  a 
considerable  time  but  no  records  of  production  are  available  pre- 
vious to  1856.  With  the  discovery  of  the  cobalt  deposits  in 
Canada  in  1903,  the  mining  of  cobalt  ores  in  Europe  and  New 
Caledonia  practically  ceased.  The  mining  of  cobalt  ore  flourished 
best  previous  to  1860,  in  Austria,  in  Germany  between  1872-1876 
and  1889-1893,  in  Norway  from  1877  to  1893,  in  Sweden  1876- 
1893,  and  in  New  Caledonia  between  1893  and  1908. 

Cobalt  ores  are  usually  associated  with  nickel,  iron,  copper, 
and  silver  minerals.  Bismuth,  antimony,  arsenic,  sulfur,  man- 
ganese and  lead  are  also  often  present. 

The  ore  of  New  Caledonia  is  a  hydrated  oxide  of  manganese, 
cobalt  and  nickel,  high  in  iron.  The  ores  of  West  Australia  are 
practically  free  from  nickel.  In  Africa  at  the  Union  Miniere  du 
Haut  Katanga,  cobalt  is  associated  with  copper.  In  the  United 
States  at  Fredericktown,  Missouri,  nickel,  copper  and  lead  are 
the  chief  metallic  constituents  of  the  cobalt  ore.  The  ores  of 
Cobalt,  Canada,  contain  chiefly  arsenic,  antimony,  iron,  and  copper, 
and  in  addition,  lead  and  bismuth. 

The  world's  consumption  of  cobalt  amounts  to  approximately 
400  tons.  The  chief  source  of  supply  is  Cobalt,  Canada,  but  a 
small  quantity  (80  to  150  tons)*  has  been  obtained  within  the  last 
two  years  from  Queensland,  West  Australia.  The  shipments 
from  Australia  have  been  made  in  the  form  of  a  concentrate  con- 
taining from  20  to  33  per  cent  cobalt.  The  consumption  is  dis- 
tributed between  cobalt  metal,  oxide,  and  salts  in  about  the  fol- 
lowing proportion :  metal  175,  oxide  200,  salts  25  tons. 

The  tons  of  cobalt  ore  mined  in  the  United  States  are  not 
given  separately,  but  the  production  of  cobalt  oxide  has  been 
recorded.  Between  1870  and  1902,  the  quantity  of  cobalt  oxide 
produced  in  the  United  States  varied  from  5,000  to  10,000  lb.  per 
annum.  In  1903  and  1908,  the  production  increased  to  120,000 
and  100,000  lb.  respectively.  No  further  production  has  been 
reported  until  1920  when  102,000  lb.  was  produced.  These  large 
recoveries  were  due  to  the  operation  of  the  ^lissouri  mines  and 
smelter. 

The  ores  of  Ontario  continue  to  supply  practically  the  world's 

♦Mining  Magazine,  26,  97   (1922). 


COBALT — ITS  PRODUCTION    AND   USES. 


343 


requirements  of  cobalt.  A  few  years  ago,  it  was  thought  that  the 
deposits  of  Belgian  Congo  would  produce  sufficient  cobalt  to 
satisfy  the  demands  of  Europe  and  Asia,  but  so  far  little,  if  any, 
has  been  recovered,  and  it  is  doubtful  whether  the  recovery  of 
cobalt  from  the  copper  ores  would  prove  a  commercial  operation. 
Several  reports  concerning  the  deposits  of  West  Australia  have 
appeared  in  the  technical  journals.  In  1922  a  crushing  and  con- 
centrating plant  at  a  cost  of  $200,000  was  constructed.  To  con- 
centrate at  a  profit  an  ore  for  the  cobalt  content  alone  presents  a 

Table  I. 
Tons  of  Cobalt  Ore  Mined. 


Year        Ge 

rmany 

Austria 

1 
Norway 

Sweden 

CaSnia         ^P*'" 

Canada 

1856 

6 

136 

1 
... 

....          I 

1861 

1 

.  .  . 

1 

1871    : 

18 

25 

4 

1881     1 

33 

■46 

80 

556 

02 

1891 

176 

187 

244 

60 

1901 

36 

... 

3J23 

1904 

41 

8.964 

"I6* 

1909 

979 

1,533 

1911 

852 

1916 

400 

1917 

337 

1918 

380 

1919 

298 

1920 

283 

1921 

127 

1922 

... 

....    j 

221 

*  Figures  fc 

)r  Cana 

dian  produc 

tion 

a 

re  gn 

ren  in   tons 

of  cobalt  meta 

. 

For  complete  Table  see  report  of  Ontario  Bureau  of  Mines  XXVII,  Part  3,   1918. 

difficult  problem.  It  is  true  the  ores  of  Cobalt,  Canada,  are  con- 
centrated, but  the  silver  recovered  is  charged  with  the  costs  of 
operation.  Unless  the  costs  of  mining,  milling  and  refining  the 
ores  of  West  Australia  are  low,  there  will  no  doubt,  be  many 
difficulties  to  overcome  before  the  deposits  are  fully  developed. 


METALLURGY. 

The  common  methods  employed  to  treat  cobalt  ores  are  either 
chemical  or  smelting.  The  chemical  method  is  employed  more 
for  the  fairly  pure  ores,  and  consists  generally  in  dissolving  the 


344  C.   W.  DRURY. 

metallic  constituents  of  the  ore  in  acid,  followed  by  precipitations 
to  remove  the  various  metals  and  impurities.  The  smelting  process 
which  is  used  almost  entirely  on  the  ores  of  Cobalt,  Canada,  pro- 
duces first  a  speiss.  The  speiss  contains  approximately  35  to 
40  per  cent  cobalt  and  nickel,  15  per  cent  iron,  35  per  cent  arsenic, 
and  1100  oz.  of  silver  per  ton,  in  addition  to  lead  and  copper. 

The  separation  of  the  several  metals  in  speiss  from  the  cobalt 
presents  numerous  difficulties.  These  are  due  mainly  to  the 
similar  properties  of  the  three  metals,  iron,  nickel,  and  cobalt.  It 
is  impossible  in  a  commercial  operation  to  separate  the  previously 
mentioned  metals  by  fire  methods,  by  depending  on  the  different 
degrees  of  oxidation  or  reduction.  Therefore,  in  the  standard 
process  for  treating  speiss,  all  the  constituents  are  dissolved  in 
acid,  sulfuric  being  commonly  employed.  The  necessity'  of  add- 
ing acid  to  dissolve  all  other  metals  in  addition  to  the  cobalt  adds 
greatly  to  the  cost.  The  similarity  of  the  properties  of  the  three 
metals  still  exists  after  being  rendered  soluble.  Alkaline  hydrates 
or  carbonates,  the  cheapest  precipitants,  precipitate  under  ordi- 
nary conditions  hydrates  or  carbonates  of  the  three  metals.  The 
large  quantities  of  iron,  arsenic,  and  acid  which  must  be  removed 
retain  considerable  quantities  of  cobalt  solutions,  and  are  a 
source  of  heavy  losses.  For  the  foregoing  reasons,  it  is  necessary 
to  operate  on  dilute  solutions  to  effect  anything  approaching  effi- 
ciency in  the  various  precipitations. 

The  following  figures  show  the  extent  of  the  removal  of  the 
impurities  in  the  metallurgy'  of  cobalt. 


analysis 

Co 

Xi 

Fe 

As 

s 

Cu 

SiOz 

Ore 

5 

4 

10 

14 

7 

1 

20 

Oxide 

70.5 

1.0 

0.25 

trace 

0.1 

0.03 

0.2 

The  corrosive  effect  of  solutions  obtained  at  the  different  stages 
in  the  metallurg}-  of  cobalt  is  serious.  The  handling  of  large 
tonnages  containing  sulfuric  acid,  copper  and  ferric  sulfates 
presents  a  problem  most  difficult  to  solve.  Practically  every  metal 
or  alloy  on  the  market  has  been  tested  in  the  solutions,  but  so 
far  nothing  has  been  found  which  will  withstand  the  corrosion 
and  abrasion.  It  may  be  of  interest  to  the  members  of  this 
Society  to  know  in  the  production  of  1  lb.  of  cobalt  it  is  neces- 
sarv  to  handle  3000  lb.  of  solution. 


COBALT — ITS   PRODUCTION    AND   USES.  345 

After  the  sulfur,  iron,  arsenic  and  copper  have  been  removed, 
the  separation  of  the  cobalt  from  nickel  is  the  next  operation. 
The  cobalt  is  precipitated  as  cobaltic  hydrate  Co  (OH).;  by  hypo- 
chlorite solutions.  To  get  a  pure  product  the  oxide  is  dissolved 
and  reprecipitated.  In  the  precipitation  of  cobalt,  chlorine  is 
evolved,  which  is  hard  on  the  workmen,  and  the  moist  chlorine 
gas  is  very  corrosive  on  exhaust  fans  and  pipe  lines. 

The  metallurgy  of  cobalt  presents  some  interesting  problems, 
to  those  engaged  in  the  study  of  finding  a  suitable  material  to 
resist  the  combined  corrosive  effect  of  acid  solutions  of  copper 
and  ferric  sulfates. 

Recently  a  patent^  was  granted  covering  the  treatment  of  cobalt 
ores  with  chlorine  gas.  It  is  planned  to  operate  the  process  at 
the  plant  of  the  Coniagas  Reduction  Co.,  Thorold,  Ont.  Little  is 
known  at  present  of  the  details  of  the  process,  but  it  is  under- 
stood that  the  arsenic  and  iron  are  volatilized  as  chlorides  at 
certain  temperatures. 

USES. 

Cobalt  metal  is  used  chiefly  in  the  manufacture  of  stellite,  and 
as  one  of  the  main  constituents  of  permanent  magnets.  The 
superiority  of  stellite  as  a  cutting  tool  has  been  definitely  estab- 
lished. The  addition  of  cobalt  permits  magnets®  to  be  made  of 
less  than  half  the  weight  of  those  made  of  ordinary  tungsten 
magnet  steel. 

The  oxide  is  used  mainly  for  coloring  in  the  ceramic  and  enamel 
industries  and  in  the  preparation  of  cobalt  salts.  Cobalt  salts  are 
used  as  driers  in  paints  and  varnishes,  as  catalytic  agents  in  the 
hydrogenation  of  oils,  in  the  preparation  of  certain  printing  inks, 
and  in  stains  in  the  ceramic  industries.  Cobalt  silicate  possesses 
a  rich  blue  color  and  is  used  extensively  in  the  china  trade.  In 
enamels,  cobalt  is  used  to  neutralize  the  yellow  tinge  due  to  iron 
oxide. 

The  salts  of  cobalt  which  are  at  our  disposal  in  commercial 
quantities  are  all  of  the  cobaltous  or  divalent  type.  It  has  been 
found  that  although  they  can  be  readily  used  in  the  manufacture 
of  driers,  and  worked  like  the  various  compounds  of  manganese, 

'  E.  W.  Westcott,  U.  S.  Patent  No.   1,406,595. 

"Honda  and  Saite,  K.  S.  Magnet  Steel.  Electrician,  85.  705.  (1920);  Steels  for 
Permanent  Magnets,  Electrician,  86,  327,   (1921);  Kayser,  Electrician,  88,  421    (1922). 


346 


C.   W.  DRURY. 


lead,  zinc,  calcium,  aluminum,  etc.,  the  organic  compounds 
formed,  which  are  the  basis  of  the  so-called  driers,  are  not  effi- 
cient while  in  the  cobaltous  state.  The  formation  of  trivalent 
cobalt  compounds  is  sought  in  the  making  of  driers.  The  value 
of  cobalt  compounds  depends  not  on  their  power  to  dry  linseed 
oil,  but  on  their  ability  to  make  the  lower  priced  semi-drying 
oils  act  like  linseed  oil. 

Table  II. 
Composition  of  Commercial  Cobalt  Compounds. 


Cobalt 

Cobaltous   Oxide    

Cobaltous  Cobaltic  Oxide.. 

Cobaltic    Oxide    

Cobalt  Acetate    

Cobalt  Borate   

Cobalt   Carbonate    

Cobalt   Chloride    

Cobalt  Hydrate: 

Black    

Pink    

Cobalt  Nitrate  

Cobalt  Linoleate : 

Solid     

Liquid    

Cobalt    Phosphate    

Cobalt   Resinate    

Cobalt  Sulfate  

Cobalt  Ammonium  Sulfate. 

Cobalt    Tungate    stereoiso- 
mer of  linoleate  


Formula 


Co    

CoO     

C0.O4    

C02O3 

Co(C.H302),.4H.O 
2CoO  .  2B2O3 .  4H.0 

C0CO3     

C0CU.6H.O    


Theoretical 
Per  Cent 

Cobalt 


Per  Cent 
Cobalt    in 
Com- 
mercial 
Product 


Co(OH)3  

Co(OH)=     

Co(N03)2.6H.O 


78.65 
73.43 
71.00 
23.70 
32.60 
49.58 
24.80 

53.64 
63.44 
20.27 


Co(Ci8H3i02)2    1       9.56 


Co3(PO.)s.2H.O    

Co(C«H,=O02    

CoSO*.7H.O    

CoSO..(NH02SO..6H2O 


43.92 

4.31 

20.90 

14.93 

9.56 


I  97.5 
75.0 
70.5 

23.6 

30.0 
43.5 
24.0 

S0.+ 
57-62 
20 

7.-7.5 

5.0 

41.0 

1.5  (fused) 

20-21 
14.5 


In  conclusion  it  may  be  added  that  unless  some  extensive  cobalt 
deposits  are  found,  the  source  of  cobalt  is  limited.  The  cobalt 
content  of  the  present  ores  is  gradually  decreasing  and  the 
impurities  are  increasing,  which  has  a  tendency  to  raise  the  cost 
of  production.  The  superiority  of  cobalt  and  its  compounds  in 
the  stellite,  magnet  steel,  ceramic,  paint  and  varnish  industries 
has  been  established.  The  properties  of  cobalt  and  its  com- 
pounds are  remarkable,  varying  from  imparting  great  hardness 


COBALT — ITS   PRODUCTION    AND   USES.  347 

and  strength  in  stellite,  high  magnetic  retentivity  in  permanent 
magnets,  beautiful  blue  color  in  china  and  enamels,  to  its  action 
as  a  catalytic  agent  in  the  oxidation  or  hydrogenation  of  oils.  The 
demand  for  cobalt  or  its  compounds  is  becoming  greater,  but  its 
use  will  no  doubt,  be  confined  to  those  industries  in  which  the 
price  of  the  raw  material  is  small  in  comparison  with  the  results 
achieved  or  savings  effected. 


DISCUSSION. 


Kenneth  S.  Guiterman^  (Communicated)  :  I  have  read  this 
paper  with  exceeding  interest.  Having  in  mind  the  fact  that  the 
cobalt  industry  as  such  would  undoubtedly  be  materially  benefited 
through  a  better  understanding  of  the  metallurgy,  I  venture  to 
emphasize  some  of  the  salient  features  which  have  apparently 
escaped  the  attention  of  Prof.  Drury. 

As  is  undoubtedly  well  appreciated,  the  primary  cause  of  the 
high  cost  of  producing  cobalt  has,  in  the  past,  been  a  consequence 
of  the  cumbersome  and  highly  unsatisfactory  method  of  separat- 
ing it  from  its  constituent,  nickel.  This  has  almost  universally 
been  accomplished  through  the  medium  of  sodium  or  calcium 
hypochlorite.  Through  the  addition  of  this  oxidizing  agent  to  an 
essentially  neutral  solution  containing  both  cobalt  and  nickel,  it 
was  possible  to  precipitate  a  hydrated  oxide  of  cobalt.  Unfor- 
tunately, the  precipitation  was  extremely  imperfect,  in  that  the 
precipitant  likewise  reacted  with  nickel.  The  net  result  of  this 
was  that  the  separation  had  to  be  carried  out  through  numerous 
fractionizations,  each  product  thereof  containing  varying  propor- 
tions of  cobalt  and  nickel.  Hence,  it  is  obvious  that  an  operating 
plant  became  burdened  with  large  quantities  of  intermediary  by- 
products, none  of  which  were  suitable  for  the  market  and  all  of 
which  necessitated  re-treatment. 

In  1914  and  1915  the  Research  Department  of  the  American 
Smelting  and  Refining  Company,  under  my  direction,  undertook 
the  development  of  a  new  and  more  efficient  process  for  the  sep- 
aration of  cobalt  and  nickel.     This  work  was  largely  a  conse- 

iNew  York  City. 


348  DISCUSSION. 

quence  of  the  Smelting  Company  having  in  its  possession  large 
tonnages  of  cobalt-nickel  speiss.  This  speiss,  of  course,  locked  up 
an  appreciable  amount  of  both  gold  and  silver.  Hence,  it  was 
primarily  with  a  view  to  recovering  these  precious  metals  that 
the  process  was  devised. 

After  some  months  of  laborious  and  pains-taking  work,  a 
method  of  electrochemical  separation  was  devised,  whereby  it 
became  possible  to  separate  cobalt  from  nickel  electrolytically  and 
without  the  formation  of  by-products,  under  conditions  of  operat- 
ing efficiency  in  excess  of  98  per  cent.  The  process,  as  patented 
by  myself,  consisted  briefly  of  sulfating  the  speiss,  followed  by 
a  solution  thereof  in  water.  After  the  preliminary  removal  of 
iron,  arsenic,  copper,  etc.,  the  resultant  cobalt-nickel  sulfate  solu- 
tion was  evaporated  to  a  high  degree  of  concentration.  Salt  was 
added  slightly  in  excess  of  the  theoretical  quantity  necessary  to 
produce  nascent  sodium  hypochlorite.  The  solution  was  then 
electrolyzed  in  soapstone  hopper-bottom  tanks,  under  conditions 
of  great  velocity  of  circulation  and  high  current  density.  Copper 
cathodes  and  graphite  anodes  were  employed.  The  solution  was 
maintained  faintly  acid  at  all  times  through  the  addition  of  a 
solution  of  sodium  carbonate  of  approximately  N /\  normal.  In 
order  to  preclude  the  formation  of  insoluble  carbonates,  the  addi- 
tion of  the  sodium  carbonate  was  made  in  the  form  of  a  cloud 
over  the  reservoir  containing  the  circulating  sulfate  solution. 
Regulation  of  the  acidity  was  maintained  throughout  the  entire 
process  by  frequent  electrotitrometric  determinations,  litmus  and 
other  indicators  being  worthless,  because  of  the  intense  green 
color  of  the  solution. 

The  current  efficiency  of  the  process  was  excellent,  and  as  I 
have  stated  above,  the  separation  of  the  cobalt  from  nickel  was 
all  that  could  be  desired.  The  end-point  of  the  reaction  was  mani- 
fest through  the  practical  absence  of  cobalt  in  solution.  As  soon 
as  this  moment  was  reached,  the  entire  solution  was  filter-pressed, 
thereby  removing  the  hydrated  oxide  of  cobalt  from  suspension. 
This  was  washed  in  the  usual  method  with  water  and  possibly 
dilute  sulfuric  acid.  The  filter-pressed  cakes  after  discharge 
were  then  calcined,  so  as  to  produce  the  gray  or  black  oxide  of 
cobalt  as  might  be  desired. 


COBALT — ITS    PRODUCTION    AND    USES.  349 

Throughout  the  entire  operation  of  the  plant,  the  process  func- 
tioned without  difficulty,  and  the  most  high-grade  product  was 
put  on  the  market.  Insofar  as  costs  were  concerned,  I  may  say 
that  they  were  low,  and  would  have  permitted  of  the  active  com- 
petition of  cobalt  versus  nickel,  without  appreciable  danger  to 
the  former. 

The  above  briefly  described  method  is,  in  my  judgment,  emi- 
nently superior  to  the  others  which  have  so  far  been  devised, 
including  that  of  treatment  with  chlorine  gas.  The  objection  to 
all  of  these  is  that  by-products  either  form  or  are  so  liable  to 
formation  as  to  render  the  process  hazardous.  No  such  condition 
presents  itself   in  the  electrolytic  method. 

E.  O.  Benjamin-  :  A  use  was  devised  for  cobalt  by  I.  H. 
Levin,  as  an  oxygen  electrode  in  electrolytic  cells,  claiming  a 
higher  efficiency  or  lowering  of  the  oxygen  over-voltage.  But 
since  the  time  of  the  appearance  of  that  patent",  as  well  as  the 
description,  I  have  made  some  experiments  which  do  not  seem 
to  confirm  that  claim.  I  have  found  the  efficiency  of  a  cobalt 
electrode  is  somewhat  lower  than  that  of  a  nickel  electrode. 

Colin  G.  Fink*  :  May  I  ask  Mr.  Benjamin  if  a  cobalt-plated 
iron  electrode  was  used? 

E.  O.  Benjamin  :   Yes. 

Colin  G.  Fink  :  Our  tests  have  shown  that  a  cobalt-plated  iron 
electrode  is  decidedly  Better  as  an  oxygen  electrode  than  a  nickel- 
plated  iron  electrode.  Perhaps  you  did  not  get  enough  cobalt  on 
your  electrode. 

O.  C.  Ralston-"'  :  I  am  a  little  surprised  to  hear  the  present 
commercial  methods  of  separating  nickel  and  cobalt  accused  of 
being  so  inefficient.  It  recalls  a  little  piece  of  work  done  by  M.  J. 
Udy  and  myself  some  years  ago  on  separating  these  two  metals 
from  each  other.  Chlorine  was  used  to  oxidize  the  cobalt  to  the 
higher  stage  of  oxidation  in  the  presence  of  finely  divided  calcium 
carbonate  to  cause  its  hydrolysis  to  the  black  oxide.  As  long  as 
the  solution  was  kept  cold  only  the  cobalt  precipitated  and  the 

-  Consulting    Engr.    and    Chemist,    Newark,    N.    J. 

3U.   S.   Pat.    1,214,934. 

<  Consulting  Metallurgist,  New  York  City. 

'  U.    S.    Bureau    of    Mines.    Berkeley,    Calif. 


350  DISCUSSION. 

separation  was  practically  quantitative.  In  fact,  Mr.  Udy,  who 
did  most  of  the  experimental  work,  told  me  that  he  found  he 
could  use  it  also  as  an  analytical  method  and  that  it  seemed  to 
be  more  sensitive  than  the  dimethyl  glyoxime  separation. 

Colin  G.  Fink  :  I  may  add  to  Mr.  Ralston's  remark  that  Prof. 
Edgar  F.  Smith  and  Prof.  H.  S.  Lukens,  of  the  University  of 
Pennsylvania,  have  worked  out  an  analytical  method  for  the 
separation  of  cobalt  from  nickel^  Cobalt  is  deposited  as  an  oxide 
at  the  anode,  and  nickel  as  metal  at  the  cathode. 

R.  B.  Moore"  :  The  price  of  cobalt  oxide  for  a  good  many 
years  was  from  $1.00  to  about  $1.50  a  pound,  and  then  it  went 
up  to  $4.50.  We  tried  to  investigate  why  that  was,  but  without 
success,  unless  it  was  that  at  that  time  the  principal  cobalt  prop- 
erty in  this  country  was  absorbed  by  foreign  interests.  However, 
the  question  that  is  obvious  is,  how  long  would  a  $3.00  price  last 
if  other  companies  got  into  the  game?  There  are  a  number  of 
small  cobalt  deposits  in  this  country,  and  naturally  if  there  were 
a  chance  of  their  succeeding  we  would  like  to  see  them  do  some- 
thing. But  under  such  conditions,  would  the  price  of  $3.00  a 
pound  suddenly  drop? 

C.  W.  Drury  (Communicated)  :  Mr.  K.  S.  Guiterman  empha- 
sizes the  electrochemical  method  of  precipitating  cobalt  compared 
with  that  employing  hypochlorite  solutions.  In  the  electrochemi- 
cal method,  it  appears  that  the  salt  is  electrolyzed,  giving  chlorine 
and  caustic.  These  two  products  unite  in  the  cell,  giving  what 
Mr.  Guiterman  calls  "nascent  hypochlorite." 

To  produce  cobalt  as  cobaltic  hydrate,  a  certain  quantity  of 
hypochlorite  is  necessary.  The  standard,  as  well  as  the  Guiter- 
man method,  requires  hypochlorite,  and  the  whole  question  under 
discussion  appears  to  be  whether  hypochlorite  can  be  prepared  more 
cheaply  from  calcium  bleach,  liquid  chlorine  and  soda,  or  by  elec- 
trolyzing  the  salt  solution  as  in  the  Guiterman  method.  Special 
attention  has  been  given  to  the  development  of  cells  to  produce 
chlorine  efficiently,  e.  g.,  Nelson,  Allen  :Moore,  and  Townsend, 
and  even  in  the  best  an  energy  efficiency  of  60  per  cent  is  high. 

•Trans.  Ain.  Electrochem.  Soc,  27,  31    (1915). 
7  do  The  Dorr  Co.,  New  York  City. 


A  paper  presented  at  the  Forty-third 
General  Meeting  of  the  American  Elec- 
trochemical Society  held  in  New  York 
City.  May  4,  1923,  President  Schlueder- 
berg    in    the    Chair. 


CHROMIZINC' 

By  F.   C.  Kelley.' 

Abstract. 

It  is  the  purpose  of  this  paper  to  give  a  brief  summary  of 
the  work  which  has  been  done  to  date  upon  the  diffusion  of 
metals  in  the  soHd  state,  and  to  describe  in  detail  the  process 
of  chromizing,  and  its  effects  upon  the  physical  and  chemical 
properties  of  iron.  The  practical  application  of  this  process 
is  also  considered. 

There  are  many  other  metals  which  diffuse  in  the  same  manner 
when  brought  into  contact  with  each  other  at  temperatures  below 
their  melting  points.  This  field  may  be  the  subject  of  a  future 
paper. 


SUMMARY   OF   LITERATURE. 

It  has  long  been  known  that  solid  bodies  are  capable  of  diffus- 
ing into  one  another.  The  old  cementation  processes  are  based 
upon  this  fact,  but  it  is  only  within  comparatively  recent  years 
that  any  practical  use  has  been  made  of  this  knowledge. 

Faraday  and  Stodard  in  1820  while  experimenting  on  the 
alloys  of  iron  observed  that  steel  and  platinum  wires  when  tied 
together  in  bundles  could  be  welded  at  a  temperature  considerably 
below  that  at  which  either  of  the  metals  melted.  Upon  etching 
the  welded  mass  with  acid,  the  iron  appeared  to  be  alloyed  with 
the  platinum. 

Chemoff  in  1877  discovered  that  if  two  surfaces  of  iron  are 

'  Manuscript  received  February  1,   1923. 

»  Research  Laboratory.     General  Electric  Co.,  Schenectady,  N.  Y. 

3M 


352  F.  C.  KELLEY. 

heated  to  650°  C.  in  intimate  contact  with  each  other  they  will 
unite. 

Spring  in  1882  discovered  that  alloys  might  be  produced  by 
compression  of  their  constituent  metals  in  a  fine  state  of  division. 

Hallock  in  1888  showed  that  similar  results  to  those  of  Spring 
could  be  obtained  at  higher  temperatures  without  pressure. 

Roberts-Austen  in  1896  published  results  of  experiments  on 
diffusion  of  gold  in  solid  lead  at  various  temperatures,  and  in 
1900  published  additional  data  on  the  same  work  in  the  Pro- 
ceedings of  The  Royal  Society. 

C.  E.  Van  Ostrand  and  F.  P.  Dewey  of  the  U.  S.  Geological 
Survey  in  1915  checked  up  Roberts-Austen's  work. 

Tycho  Van  Aller  of  the  General  Electric  Company  patented  in 
1911  a  process  (calorizing),  which  depends  upon  the  diffusion 
of  aluminum  into  metals  below  its  melting  point. 

E.  G.  Gilson,  of  the  General  Electric  Company,  patented  in 
1914  another  process  of  calorizing  metals,  in  which  greater  pene- 
tration of  aluminum  is  obtained  by  operating  at  higher  tempera- 
tures, and  in  an  atmosphere  which  protects  the  aluminum  from 
oxidation.     Hydrogen  is  the  usual  atmosphere  used. 

Cowper-Coles,  in  June,  1902,  and  in  August,  1906,  received 
patents  on  coating  iron  with  zinc  to  protect  it  from  corrosion. 
The  zinc  forms  an  alloy  with  the  iron  at  a  temperature  below 
the  melting  point  of  zinc,  and  the  surface  of  the  coated  metal 
is  nearly  pure  zinc,  which  resists  corrosion. 

Collins  and  Capp,  of  the  General  Electric  Company,  patented 
a  process  of  sherardizing  in  January,  1916,  in  which  they  de- 
scribe an  improved  sherardizing  process,  in  which  the  zinc  con- 
tent of  the  sherardizing  mixture  and  the  temperature  are 
correlated  in  a  new  way. 

Calorizing  and  sherardizing  are  two  commercial  processes 
which  depend  upon  the  alloying  of  metals  at  temperatures  below 
their  melting  points.  In  the  case  of  calorizing  I  refer  to  the 
Van  Aller  process. 

Chromizing,  the  subject  with  which  this  paper  is  chiefly  con- 
cerned, is  another  patented  process  which  depends  upon  this 
same  property  of  diffusion  of  metals  at  temperatures  below  their 
melting  points. 


CHROMIZING.  353 

THE    METHOD. 

The  process  consists  of  packing  the  material  to  be  treated  into 
a  powdered  mixture  of  ahimina  and  chromium.  The  amount  of 
each  material  used  in  the  mixture  is  45  per  cent  of  alumina  and 
55  per  cent  of  chromium  by  weight.  The  material  is  usually 
packed  into  a  tube  of  iron,  and  then  heated  at  1300  to  1400°  C. 
in  hydrogen,  in  vacuum  or  in  some  neutral  atmosphere,  for 
lengths  of  time  depending  upon  the  penetration  and  concentration 
of  chromium  desired. 

Where  a  protective  atmosphere  like  hydrogen  is  used  it  is 
absolutely  necessary  that  it  should  be  free  from  all  oxygen  and 
water  vapor,  for  at  the  high  temperatures  at  which  this  work 
is  carried  on  the  chromium  powder  would  be  rapidly  consumed 
by  oxidation.  In  fact  as  soon  as  a  film  of  oxide  is  formed  on 
the  surface  of  the  fine  particles  they  refuse  to  react  with  the 
metal  to  be  chromized.  This  purification  is  accomplished  by 
first  passing  the  hydrogen  through  a  sulfuric  acid  tower,  to 
remove  most  of  the  water.  The  gas  is  then  passed  over  a  copper 
gauze,  heated  to  about  600°  C,  to  get  rid  of  any  oxygen  present 
by  combining  the  oxygen  with  the  hydrogen,  the  copper  acting 
as  a  catalyzer.  The  water  formed  in  the  copper  furnace  is  then 
taken  out  by  passing  the  gas  through  additional  sulfuric  acid 
towers,  after  which  it  is  passed  over  potassium  hydroxide  to 
take  out  any  sulfuric  acid  vapors.  Finally  it  is  passed  over 
phosphorus  pentoxide  to  remove  the  last  traces  of  moisture. 
This  gas  then  goes  directly  to  the  chromizing  furnace. 

The  furnaces  used  for  this  work  consist  of  alundum  tubes 
wound  with  molybdenum  wire  as  a  heating  unit.  These  tubes  are 
placed  in  a  suitable  furnace  casing  and  surrounded  with  alumina 
powder,  which  acts  as  a  heat  insulating  material. 

The  hydrogen  atmosphere  of  this  furnace  serves  a  double 
purpose.  It  not  only  prevents  the  burning  up  of  the  chromium, 
but  it  also  protects  the  molybdenum  from  oxidation,  and  thus 
enables  us  to  attain  with  ease  the  high  temperatures  at  which 
we  operate. 

In  order  to  indicate  the  size  of  the  furnace  of  this  type  which 
may  be  used  to  advantage,  I  may  say  that  we  have  operated  two 
furnaces,  which  were  each  made  of  four  alundum  tubes,  60  cm. 


354  F-  C.  KELLEY. 

long,  20  cm.  inside  diameter  and  13  mm.  walls,  placed  end  to 
end  in  a  metal  casing,  and  held  in  line  by  means  of  a  strip  of 
sheet  molybdenum  25  mm.  wide  and  0.63  mm.  thick  bound 
around  the  tubes  at  the  joints.  The  tubes  at  the  joints  were 
supported  in  the  furnace  casing  by  fire  brick  cut  to  fit  the  tubes. 
The  molybdenum  band  also  serves  to  keep  the  alumina  from 
falling  into  the  furnace  through  the  joints  formed  at  the  ends 
of  the  tubes. 

The  winding  for  these  tubes  was  molybdenum  wire  1.91  mm. 
diameter,  and  there  were  two  windings  on  each  tube.  Each 
winding  consisted  of  22  turns  12.7  mm.  apart,  with  the  exception 
of  the  two  end  windings,  which  were  wound  8.5  mm.  apart  to 
compensate  for  radiation  at  the  ends  of  the  furnace.  The  furnace 
casing  was  2.74  meters  long  and  53.3  cm.  square.  The  inside  was 
lined  with  a  single  row  of  fire  brick,  and  the  space  between  the 
brick  and  furnace  tube  was  filled  with  alumina. 

The  furnace  was  operated  directly  from  a  1000-volt  a.  c. 
generator,  by  means  of  a  resistance  in  the  field  of  the  machine. 
Two  large  transformers,  connected  in  parallel,  were  used  to 
step  the  voltage  down  from  1000  to  as  low  as  12  volts.  The 
windings  of  the   furnace  were  all  connected  in  parallel. 

These  furnaces  will  carry  without  trouble  a  charge  weighing 
136  kg.  (300  lb.)  distributed  over  1.83  m.  (6  ft.)  of  its  length 
at  1350°   C. 

The  chief  use  for  these  furnaces  was  in  chromizing  turbine 
buckets,  which  have  been  installed  in  various  turbines  through- 
out the  country,  in  order  to  test  them  for  corrosion  under  actual 
operating  conditions. 

MATERIALS   REQUIRED. 

In  chromizing  it  is  necessary  to  have  powdered  chromium  of 
at  least  95  per  cent  purity,  for  chromizing  iron  which  is  intended 
to  withstand  corrosion.  Powdered  AI2O3  is  necessary  as  a  di- 
luting agent,  and  to  prevent  excess  sintering  of  the  powdered 
material  at  high  temperatures.  It  is  also  necessary  to  have  pure 
hydrogen,  free  from  moisture  and  oxygen.  And  last  of  all  it 
is  necessary  to  have  a  furnace  which  will  operate  at  a  temperature 
of  1300°  C.  or  higher,  and  in  an  atmospliere  of  hydrogen. 


CHROMIZING. 


355 


STRUCTURE. 
Chromized  iron,  with  which  I  am  going  to  deal  chiefly,  when 
examined  under  the  microscope,  has  a  structure  which  seems 
to  be  characteristic  of  all  metallic  coatings  obtained  by  diffusion. 
This  chromium-iron  alloy,  which  is  a  solid  solution  of  chromium 
in  iron,  is  made  up  of  an  area  of  elongated  grains,  with  their 
longer  axes  perpendicular  to  the  surface  chromized,  and  the  line 


Fig.  1. 
Clironiized   iron   heated  at   1,350°    C.    for   4   hr.     x   70. 


of  penetration  is  generally  very  sharp.  Sometimes  there  seems 
to  be  a  cylinder  of  grains  arranged  in  this  manner  on  the  surface 
of  a  chromized  iron  rod.  Then  again  there  seems  to  be  two  or 
more  such  bands  or  cylinders,  which  contain  varying  amounts 
of  chromium,  for  each  band  etches  up  differently.  The  one  with 
the  highest  chromium  content  etches  up  the  slowest. 

Fig.   1  illustrates  the  chromized  iron  surface,  consisting  of  a 
layer  of  elongated   grains.     The  longer  axis   of  each  grain  is 


356 


F.  C.  KELLEY. 


at  right  angles  to  the  surface  chromized.  Fig.  2  illustrates  a  piece 
of  chromized  iron,  showing  two  distinct  bands  on  the  surface. 
If  we  were  to  analyze  samples  from  each  band,  we  would  find 
that  the  inner  band  had  a  lower  chromium  content  than  the 
outer  band. 


Fig.   2 
Chromized  iron   showing  banded  structure   of  the  chromized  section. 

Fig.  3  also  shows  three  distinct  bands  outside  of  the  iron  core. 
The  outer  band,  which  is  very  narrow,  seems  to  be  made  up  of 
almost  pure  chromium,  sintered  together  and  alloyed  to  the 
chromium-iron  band  just  underneath.  This  is  a  cross  section  of 
a  sample  fired  in  pure  chromium  at  1350°  C.  for  3  hr. 

If  we  chromized  a  sample  at  the  same  temperature  as  the 
sample  shown  in  Fig.  1,  but  for  twice  the  length  of  time  (8  hr.) 
we  would  get  no  sharp  line  of  penetration,  and  the  large  elon- 
gated grains  we  would  find  had  broken  into  somewhat  smaller 
grains,  as  is  shown  in  Fig.  4.     If  we  analyzed  the  surface  coat- 


CHROMIZING. 


357 


ings  of  samples  shown  in  Fig.  1  and  4,  we  would  find  that  the 
sample  of  Fig.  1  would  show  the  higher  per  cent  of  chromium. 
The  chromium  content  of  the  other  sample  would  be  reduced, 
due  to  greater  diffusion  for  it  has  been  fired  twice  as  long. 


Fig.  3. 
Chromized  iron  heated  in  pure  chromium  powder  at   1,350°   C.  for  3  hr.     x  57. 


CONTACT  PROCESS. 
In  chromizing,  even  at  these  high  temperatures,  the  vapor 
pressure  of  chromium  is  very  low,  and  all  of  the  chromium  which 
is  taken  up  by  the  iron  must  be  in  contact  with  it.  If  it  is  de- 
sired to  increase  the  chromium  content  of  a  surface  coating,  it 
is  necessary  to  give  the  sample  a  second  treatment.  This  fact  is 
illustrated  by  examination  of  the  somewhat  sintered  chromizing 
mixture  where  it  has  been  in  contact  with  the  iron  treated.  The 
surface  of  the  sintered  mixture  is  white,  showing  that  all  of  the 
chromium  has  been  taken  up  by  the  iron  surface  in  contact  with 


358 


F.  C.  KELLEY. 


these  finely  divided  particles,  leaving  only  the  white  AI2O3  behind. 
If  this  sintered  piece  of  mixture  is  broken  at  right  angles  to  the 
surface  examined,  chromium  particles  will  be  found  just  under 
the  AI2O3  surface. 


Fig.  4 
Chromized  iron  heated  at  1,350°  C.  for  8  hr.     x  42. 


THE  EFFECT  OF  CARBON. 

In  order  to  get  the  best  chromizing  results  it  is  necessary  to 
use  an  iron  or  steel  of  low  carbon  content,  for  iron  of  high 
carbon  content  does  not  chromize  well.  The  carbon  seems  to 
retard  the  penetration  of  the  chromium.  It  is  possible  to  chro- 
mize it  if  it  is  first  decarbonized  by  firing  in  hydrogen.  Another 
essential  point  is  to  have  the  samples  to  be  chromized  well  cleaned 
and  free  from  oxide  or  rust. 


CHROMIZING.  359 

THE  DIFFICULTIES. 

The  AUOo  if  new  should  first  be  fired  before  using,  in  order 
to  drive  out  any  moisture  which  it  may  have  taken  up.  Then 
it  may  be  mixed  with  chromium  powder  and  kept  in  closed 
cans.  This  mixture  may  be  used  over  and  over  again,  and 
chromium  added  at  intervals  to  maintain  the  chromium  content. 

The  determination  of  the  amount  of  free  chromium  metal 
present  is  one  of  the  worst  troubles  with  which  we  had  to  con- 
tend. We  were  not  able  to  determine  the  amount  of  chromium 
present  exclusive  of  the  oxides  of  chromium.  The  only  way 
that  we  could  check  up  our  mixture  was  to  put  test  samples  in 
each  run  and  compare  these  samples  with  others  which  we 
considered  good. 

It  is  almost  impossible  to  run  a  furnace  of  the  type  which 
I  have  described  without  getting  some  oxidation,  because  the 
AI2O3  used  as  the  furnace  insulation  and  also  as  a  part  of 
the  chromizing  mixture  is  active  towards  moisture.  But  where 
the  furnace  and  mixture  are  being  constantly  used,  there  is  little 
chance  to  take  up  moisture,  and  under  these  conditions  we  get 
the  best  results.  The  oxidation  of  the  chromium  mixture  always 
took  place  to  some  extent  at  the  open  end  of  the  containing 
vessel,  and  this  powder  at  the  end  was  always  discarded,  so  as 
not  to  contaminate  the  rest  of  the  mixture  when  used  again. 

CHROMIZING  DATA. 

In  order  to  give  an  idea  of  the  amount  of  chromium  taken  up 
by  a  sample  and  the  penetration  I  shall  give  the  data  shown  in 
Table  I.  These  samples  were  about  1.27  cm.  x  1.27  cm.  x  1.27 
to  1.59  cm.  (H  X  ^  x  ^  to  ^  in.)  Three  samples  were  used 
and  one  sample  was  taken  out  after  each  chromizing  run  at 
1300°  C.  for  3  hr. 

Table  II  contains  the  data  on  samples  heated  at  1200°  C, 
1350°  C.  and  1400°  C.  for  3  hr.  periods. 

Fig.  5  shows  the  effect  of  time  upon  the  penetration  of  chro- 
mium at  1300°  C.  The  samples  were  rechromized  into  a  new 
mixture  after  each  run  of  3  hr. 

Fig.  6  shows  the  effect  of  temperature  upon  the  penetration, 
where  the  time  of  heating  at  each  temperature  is  maintained 
constant  for  3  hr. 


36o 


F.  C.  KELLEY. 


We  must  remember  that  in  firing  a  sample  of  iron  in  a  chro- 
mizing  mixture  and  in  a  hydrogen  furnace,  that  a  sample  which 
is  fired  at  1400°  C.  must  be  brought  up  through  the  range  of  tem- 
peratures between  1200-1400°  C,  and  that  chromizing  and  dififu- 

Table  I. 
Chromizing  Data. 


Sample     '      Weight    before 
jtq                    chromizing 

Weight   after 
first    chromizing 

Difference  in 

weight    in 

grams   and 

per   cent 

Weight   after 

second 
chromizing  g. 

2.1  19.7518 

2.2  '          21.9863 

2.3  23.1170 

1 

19.8861 
22.1382 
23.2802 

0.67   per  cent 
0.1343  g. 
0.69   per  cent 
0.1519  g. 
0.706  per  cent 
0.1632  g. 

22.2325 
23.3780 

Difference 

in    weight    in 

per  cent  and  grams. 

Weight  after  3rd 
chromizing   g. 

Difference  in          ^--f^^^  P|- 
^^af^p^r   f.T           <=^-r" 

2  1 

0.178 

0.343 
0.558 

22 

1.12  per  cent 
0.2462  g. 

1.13  per  cent 
0.2610  g. 

2.3 

23.3275              1.77  per  cent 
0.4105  g. 

Table  II. 


Sample 
No. 


Weight 

before 

chromizing 

g- 


H.1.12 

16.5594 

H.1.135 

20.7165 

H.1.14 

21.9600 

Weight                Difference  in        '  Tempera- 
after  weight    in    per    cent  ture    in 
chromizing               a^d   grams  °C. 
g- 


16.5776 
20.8590 
22.220 


0.11  per  cent 
0.0182  g. 
0.69  per  cent 
0.1425  g. 
1.18  per  cent 
0.2600  g. 


Average 
Penetra- 
tion 
mm. 


sion  are  taking  place  during  the  time  that  the  sample  is  being 
heated  through  this  range.  It  is  almost  impossible  to  put  a  chro- 
mizing charge  into  a  hydrogen  furnace  at  1200-1400°  C.  without 
having  the  entire  charge  blown  out  of  the  containing  tube  due  to 
the  sudden  expansion  of  gases.     Even  if  this  were  possible  it 


CHROMIZING. 


361 


would  be  hard  to  judge  just  when  the  samples  within  this  heat 
insulating  mixture  came  up  to  any  given  temperature.  These 
figures  then  must  be  taken  as  the  average  obtained  in  standard 
practice. 

It  is  almost  impossible  to  chromize  a  steel  of  high  carbon  con- 
tent, such  as  drill  rod,  until  after  the  carbon  content  has  been 
greatly  reduced  by  decarbonization,  as  by  firing  in  hydrogen.  If 


1^5 

1 

LOO 

(0 

ys 

5 

^ 

^ 

.50 

1 

( 

.25 

^ 
^ 

^ 

( 
n 

TFMP 

FfffJU 

/rfT. 

Curve   showing  the  change  in   penetra- 
tion   of   chromium    with    the   time. 


/200  1500  f^OO 

Fig.  6. 

Curve  showing  the  effect  of  temperature 

u|)on  the  penetration  of  chromium. 


a  sample  of  drill  rod  is  chromized  at  1300^  C.  in  the  regular  way, 
we  notice  that  it  has  lost  in  weight,  but,  upon  examination,  we 
find  that  some  chromium  has  been  taken  up  by  the  iron.  When 
we  polish  a  cross  section  of  the  sample,  and  examine  it  under  the 
microscope,  we  find  that  the  penetration  is  very  irregular  and 
varies  so  much  that  it  is  not  possible  to  state  even  the  average 
penetration  for  such  a  sample.     There  seem  to  be  shiny  needle- 

24 


362 


F.  C.  KELLEY. 


like  projections  from  the  well-defined  chromized  ring  at  the  edge, 
which  is  about  0.05  to  0.076  mm.  (0.002-0.003  in.)  in  width. 
Upon  refining  a  sample  a  second  and  still  a  third  time,  we  notice 
that  the  sample  begins  to  increase  in  weight,  because  the  carbon 
is  nearly  all  taken  out  of  the  sample  by  the  hydrogen.  The  chro- 
mized ring  takes  on  a  much  more  regular  shape  or  has  a  more 
uniform  penetration,  and  with  each  firing  takes  up  an  increasing 
amount  of  chromium. 

The  data  given  in  Table  III  are  obtained  by  firing  three  drill 
rod  samples  of  about  0.8  per  cent  carbon  in  chromizing  mixture 
at  1300°  C.  for  3-hr.  periods,  removing  one  sample  after  each 
firing. 

Table  III. 
Chromizing  Drill  Rods 


Sample 
No. 

Weight   before 

chromizing 

g- 

Weight   after 

first     chromizing 
g. 

Difference  in 

weight  in   g.   and 

per   cent 

Weight   after 

second 

chromizing 

g. 

3.1 

3.2 
3.3 

19.6248 

20.6339 
19.7910 

19.6152 
20.6256 
19.7890 

0.05  per  cent 
—0.01  g. 

0.04  per  cent 
—0.008  g. 

0.01  per  cent 
—0.002  g. 

Removed 
20.6958 
19.8630 

Difference  in  weight 
in  g.  and 
per   cent 

Weight    after 

third  chromizing 

g- 

Difference  in 

weight  in  grams 

and  per  cent 

Average  pene- 
tration  of   Cr 
mm. 

,, 

0  076 

3.2 

0.34  per  cent 
+0.07  g. 

0.37  per  cent 
+0.074  g. 

Removed 
19.9783 

0  127 

3.3 

0.957  per  cent 
+0.1893  g. 

0.420 

In  order  to  show  the  effect  of  diffusion  at  1350°  C,  I  took 
eight  especially  turned  sample  rods  and  chromized  them  all  at 
1350°  C.  for  3  hr.  I  took  out  sample  No.  1  and  had  the  chro- 
mized surface  turned  off  to  the  depth  of  penetration  of  the 
chromium.  The  remaining  seven  samples  were  then  fired  in 
hydrogen  for  an  equal  length  of  time  at  1350°  C.  and  sample 
No.  2  was  taken  out  and  the  chromized  surface  turned  oflF.  This 
was  repeated  until  we  had  chromized  four  times  and  reheated  in 


CHROMIZING. 


)63 


hydrogen  four  times.  The  samples  were  taken  out  as  above,  one 
by  one  in  their  order.  After  each  treatment  the  turnings  were 
analyzed  for  chromium  content.  The  results  are  given  in 
Table  IV. 

Table  IV. 


Time 

Time  heated 

Temperature     of 

Per   cent 

Sample 

chromized 

in    Ha 

chromizing   and 

of 

No. 

hr. 

hr. 

heating  °C. 

chromium 

1 

3 

1350 

10.42 

2 

3 

3 

1350 

6.97 

3 

6 

3 

1350 

12.15 

4 

6 

6 

1350 

8.70 

5 

9 

6 

•     1350 

15.50 

6 

9 

9 

1350 

9.62 

7 

12 

9 

1350 

14.39 

8 

12 

12 

1350 

9.77 

We  must  remember  that  this  analysis  represents  the  average 
chromium  content  of  the  layer  turned  ofif.  The  chromium  content 
of  this  layer  near  the  surface  of  the  sample  is  much  higher  than 
the  average  shown  by  analysis.  The  fact  that  the  percentage  of 
chromium  content  decreases  each  time  after  firing  in  hydrogen 
is  due  to  increased  penetration  of  the  chromium  into  the  chro- 
mized layer.  Since  the  surface  is  not  in  contact  with  chromium 
when  reheated  in  hydrogen,  it  is  not  able  to  take  up  additional 
chromium.  The  diffusion  of  the  chromium,  gained  by  chro- 
mizing, through  a  greater  volume  of  the  sample  by  means  of 
firing  in  hydrogen  decreases  the  percentage  of  chromium  content. 

We  notice  that  with  each  additional  chromizing  treatment  the 
tendency  of  the  chromized  layer  is  to  increase  in  percentage  of 
chromium  content  above  that  of  the  previous  chromizing,  in  spite 
of  the  fact  that  between  each  chromizing  the  chromium  content 
of  this  layer  has  been  reduced  by  diffusion.  That  is,  the  chro- 
mium content  of  the  layer  is  reduced  on  the  average  by  hydrogen 
firing  4.24  per  cent,  and  for  every  time  it  is  rechromized  it 
increases  on  the  average  of  5.4  per  cent,  so  the  net  result  is  a 
continual  percentage  increase  of  the  chromium  with  each 
chromizing. 

Fig.  7  is  a  photograph  about  actual  size  taken  of  samples  which 
have  been  alternately  chromized  and  heated  at  1350°  C.    The  first 


364 


F.  C.  KELLEY. 


sample  shown  at  the  left  is  chromizecl,  and  from  left  to  right  they 
are  alternately  chromized  and  heated  so  that  the  last  sample  has 
been  chromized  four  times  and  heated  four  times.  The  pene- 
tration is  quite  sharp  in  each  sample  with  the  exception  of  sample 
No.  2,  which  has  been  heated  in  hydrogen  for  3  hr.  after  the  first 
chromizing.  The  diffusion  of  the  chromium  has  decreased  the 
distinctness  of  the  chromized  laver. 


Fig.  7. 

Samples  of  chromized  iron  showing  the  effect  of  alternate  chromizing  and 
heating,     x   1. 

The  effect  of  concentration  of  chromium  powder  is  shown  when 
we  pack  the  samples  to  be  chromized  into  pure  chromium  powder, 
for  the  penetration  at  any  given  temperature  for  a  given  length 
of  time  is  greatly  increased.    This  is  shown  in  Table  V. 


Table  V 


j 

Tem- 

Sample 

Weight    before 

Weight 

after 

gain    in 

Per 

cent 

Time 
of 

pera- 
ture of 

Pene- 

No. 

chromizing 

chromizing 

weight 

gam 

chro- 

chro- 

tration 

g- 

g- 

g. 

in 

mizing 

mm. 

weight 

hr. 

•c 

1 

43.8459 

45.8186 

1.9727 

4.5 

3 

1350 

0.852 

2 

43.7816 

45.5701 

1.7885 

4.10 

3 

1350 

3 

43.7585 

44.9604 

1.2019 

2.74 

2 

1350 

0.533 

4 

43.8480 

44.8967 

1.0487 

2.40 

2 

1350 

•  •  •  • 

5 

43.7020 

44.7787 

1.0767 

2.46 

1 

1350 

0.406 

6 

43.7895 

44.7177 

0.9282 

2.12 

1 

1350 



The  percentage  gain  in  weight  is  much  greater  where  pure 
chromium  is  used,  for  the  pure  chromium  particles  are  fused  to 
the  surface  in  a  much  closer  arrangement,  and  the  rate  of  diffu- 
sion being  so  nuich  greater  also  helps  to  account  for  the  increase 
in  weight. 


CHROMIZING. 


365 


In  order  to  get  some  of  the  physical  characteristics  of  chro- 
mized  iron,  I  fired  some  iron  wire  in  vacuum  at  1300°  to  1400"  C. 
for  1.5  hr.  and  obtained  the  results  recorded  in  Table  VI. 

TabIvE  VI. 
Physical  Characteristics  of  Chromized  Iron. 


Tem- 
perature 
of   chro- 
mizing 
"C. 

Time            Weight 
chro-              before 
mized          chromizing 
hr.                      g. 

1 

Weight 
after 
chromizing 
g- 

1 

Per  cent    [     Diameter 
increase            before 

in               chromizing 
weight                mm. 

Diame- 
ter   after 
chro- 
mizing 
mm. 

\m\~       1.5      '      2.2673 

1400            2               2.2666 

i                 J 

2.568 
2.7705 

11.7              0.89 
22.6 

1.0 

Per    cent 

increase 

in    diameter 

Resistance 
before  chro- 
mizing   in    mi- 
crohms  per 
cc. 

Resistance 
after   chro- 
mizing in 
michrohms 
per  cc. 

Specific 

gravity  before 

chromizing 

Specific 

gravity    after 

chromizing 

1300-  \ 
1400   ( 
1400 

12.9 

11.53 

86.2 

8.11 

7.62 

This  wire  had  a  large  grain  structure,  but  was  not  brittle.  It 
was  surprisingly  soft  for  the  amount  of  chromium  which  it  had 
taken  up.  The  chromium  had  diffused  entirely  through  the  wire. 
It  was  heated  in  the  open  air  by  passing  a  current  through  it  at 
1050°  C.  for  200  hr.  without  burning  out,  thus  showing  the  pro- 
tective value  of  chromium  as  far  as  oxidation  is  concerned. 


RESISTANCE   TO    CORROSION. 

In  testing  samples  of  chromized  iron,  we  ran  them  in  salt  spray 
along  Avith  blanks  and  found  that  after  a  month  the  blank  sample, 
3  mm.  (0.125  in.)  thick,  was  about  half  corroded  away,  while  a 
chromized  sample  had  here  and  there  slight  signs  of  attack. 

Chromized  samples  tested  along  with  sherardized  iron  samples 
in  salt  spray  after  six  weeks  showed  only  slight  attack  and  held 
up  under  test  just  as  well  as  the  sherardized  samples. 

The  samples  under  test  showed  up  so  well  that  we  made  addi- 
tional tests  upon  turbine  buckets.  These  chromized  nickel  steel 
buckets  showed  up  so  well  that  the  Turbine  Department  decided 


366 


F.  C.  KELLEY. 


Fig.  8. 

Chromized    and    uncliroinized    nickel    steel    turbine   buckets 
after   one   year    of   actual  service. 


CHROMIZING.  367 

to  put  them  into  various  turbines  throughout  the  country,  and 
into  some  of  the  turbines  of  ocean-going  vessels.  The  best  com- 
parison of  resistance  to  corrosion  of  chromized  and  unchromized 
turbine  buckets  under  service  conditions  is  illustrated  in  Fig.  8 
of  this  paper.  These  buckets  were  run  side  by  side  in  the  same 
wheel  of  a  turbine  for  one  year.  The  unchromized  nickel  steel 
bucket  at  the  left  has  its  edge  entirely  corroded  and  eroded  away, 
and  in  addition  the  face  of  the  bucket  is  badly  corroded.  The 
chromized  bucket  on  the  right  is  in  perfect  condition,  showing  no 
signs  of  corrosion. 

EFFECT  OF  HEAT  TREATMENT. 
In  cases  where  the  material  chromized  must  stand  high  tension 
and  fatigue  stresses,  the  high  temperature  of  chromizing  lowers 
the  resistance  of  the  material  to  these  stresses,  but  by  proper  heat 
treatment  the  original  properties  may  be  almost  completely 
restored. 

EFFECT  OF  CARBONIZING. 

Carbonizing  of  chromized  iron  lowers  its  resistance  to  corro- 
sion, and  polished  samples  of  chromized  iron  which  have  been 
case  hardened  will  show  numerous  globules  of  water  if  allowed 
to  stand  in  the  open  air  for  only  a  short  time. 

Chromized  iron  itself  is  quite  soft  and  ductile,  but  by  case 
hardening  and  heat  treatment  it  may  be  made  very  hard. 

RESISTANCE  TO  ACIDS, 

Chromized  iron  samples  were  tested  in  10  per  cent  HCl,  HNO3 
and  H2SO4.  They  stood  up  for  five  months  in  the  10  per  cent 
HNO3  without  discoloring  the  solution  or  showing  any  signs  of 
attack,  but  they  broke  down  almost  immediately  in  the  other  two 
acids. 

ADDITIONAI,  PROPERTIES. 

In  addition  to  these  characteristics,  chromized  iron  has  a  silver 
color,  it  takes  a  high  polish,  and  the  most  remarkable  thing  about 
it  is  its  softness  even  where  large  percentages  of  chromium  are 
present, 

OTHER  CHROMIZED  METALS. 

There  are  other  metals  which  may  be  chromized  besides  iron, 
but  under  somewhat  different  temperature   conditions.      Nickel 


368  DISCUSSION. 

may  be  chromized  if  the  temperature  used  does  not  exceed 
1300°  C.  If  a  higher  temperature  is  used  the  eutectic  alloy  of 
(Cr-Ni)  is  formed  and  the  whole  mass  melts.  The  composition 
of  this  alloy  is  (42  per  cent  Ni,  58  per  cent  Cr)  and  it  melts  just 
under  1300°  C.  There  does  not  seem  to  be  much  trouble  with 
fusion  when  chromizing  at  1300°  C,  because  the  rate  of  diffusion 
of  chromium  into  nickel  is  slow  at  this  temperature. 

Molybdenum  and  tungsten  may  also  be  chromized,  but  in  order 
to  get  any  penetration  a  temperature  of  1600°  C.  is  necessary, 
which  is  above  the  critical  point  of  cr}'stallization,  and  the  wire 
obtained  is  of  large  grain  structure  and  very  brittle. 

ANOTHER  APPLICATION. 

Chromizing  may  be  used  for  another  purpose  than  protection 
from  corrosion.  It  may  be  used  to  prevent  the  flow  of  a  metal 
like  copper  on  iron  at  a  temperature  above  the  melting  point  of 
copper.  In  a  case  like  this  it  is  better  to  oxidize  the  chromized 
metal  first  before  attempting  to  use  it.  If  the  chromized  metal, 
say  iron,  is  used  to  prevent  copper  from  wetting  it  in  hydrogen  at 
above  1200°  C,  it  will  if  fired  for  long  enough  time  eventually 
alloy  with  the  copper,  due  to  the  lowering  of  the  concentration  of 
chromium  at  the  surface,  due  to  dififusion.  But  it  will  resist  alloy- 
ing for  a  limited  length  of  time  even  at  this  high  temperature. 

This  is  an  interesting  field  of  research,  and  there  are  indica- 
tions that  there  may  be  future  developments  and  applications  for 
metals  treated  by  dififusion  processes. 


DISCUSSION. 

CoLTN  G.  FiNK^ :  Mr.  Kelley  states  that  hydrogen  performs 
two  functions.  One  is  to  keep  the  chromium  in  reduced  condi- 
tion, and  the  other  is  to  prevent  the  molybdenum  resistor  from 
oxidizing.  Without  the  hydrogen,  the  alloying  between  chromium 
and  iron  would  probably  not  take  place.  Hydrogen  is  the  only 
efficient  "flux"  that  I  know  of  commercially  for  this  case. 

L.  O.  Hart-  :  I\lr.  Kelley's  paper  opens  up  a  new  field.  There 
seems  to  be,  from  a  manufacturing  standpoint,  some  objections  to 

'  Consulting   Metallurgist,   New   York   City. 
*  Driver-Harris  Co.,  Harrison,  N.  J. 


CHROMIZING.  369 

this  process.  The  requirements  of  temperature  and  purity  of 
hydrogen  mean  that  the  process  of  chromizing  necessarily  must 
be  an  expensive  one.  It  requires  expert  supervision  and  ex- 
pensive apparatus,  and  it  seems  to  me  that  a  number  of  the  results 
might  be  obtained  more  cheaply  by  making  the  articles  of  a 
chrome-iron  alloy  rather  than  chromizing  a  steel  or  a  nickel  steel. 
If  the  process  were  capable  of  operation  at  a  cost  comparable 
with  sherardizing,  I  think  that  chromizing  would  have  a  much 
wider  application  than  it  has  now. 

H.  K.  Richardson^:  We  have  three  things  to  say  about  this 
process.  At  present  we  are  using  a  process  in  an  experimental 
way,  of  chromium  plating  nickel-steel  wire.  This  wire  is  heated 
in  its  final  stage  of  preparation  to  about  1,100°  C,  and  for  less 
than  half  a  minute.  The  process  is  continuous,  whereby  a  strictly 
adherent  coating  of  ductile  chromium  is  made  upon  an  under- 
coating  of  nickel  steel. 

I  would  like  to  speak  about  one  or  two  of  our  observations. 
Mr.  Kelley's  curve  on  page  361,  regarding  the  penetration  in  time, 
does  not  seem  to  be  borne  out  in  its  lower  regions  by  our  experi- 
ence. We  have  a  penetration  of  about  0.01  mm.  in  a  half  minute 
or  less. 

Regarding  the  amount  of  chromium,  we  put  8  per  cent  or 
thereabouts  upon  a  wire,  and  that  coating  after  passing  through 
the  process  at  1,100°  C.  can  be  drawn,  under  the  right  conditions, 
from  25  mils  to  10  mils,  with  little  cracking  on  the  surface. 

Now  some  friends  have  taken  this  coating  and  have  submitted 
it  to  X-ray  analysis.  The  resulting  spectrogram  shows  that  the 
chromium  has  inter-penetrated  the  nickel-steel  lattice  and  as  such 
has  made  a  much  more  dense  alloy  than  Mr.  Kelley  shows  here. 
There  is  no  indication  at  250  magnifications  of  any  crystals  what- 
soever on  the  coating.  Sometimes  the  coating  can  not  be  seen 
at  250  magnifications.  The  only  way  that  we  can  find  out  that 
we  have  a  coating  is  by  special  etches.  That  is,  when  things  have 
been  done  rightly.  We  do  not  always  get  the  result,  for  some- 
times, due  to  faulty  cleaning,  we  have  a  line  of  demarcation 
between  the  chromium  coating  and  the  nickel-steel  under-body. 

In  our  own  work  we  can  not  use  any  nickel-iron-chromium 
alloy,  because  it  would  have  too  high  a  resistance,  and  it  would 

'  Westingliouse  Lamp   Co.,   Bloomfield,   New  Jersey. 

25 


370  DISCUSSION. 

have  a  wrong  coefficient  of  expansion.  So  we  are  limited  to  an 
under-body  which  has  the  right  coefficient  of  expansion.  The 
chromium  serves  only  the  purpose  of  making  the  contact  between 
the  glass  and  the  right  coefficient  under-body. 

F.  C.  KelIvEy  :  In  answer  to  Dr.  Fink's  statement,  that  hydrogen 
is  acting  as  a  flux  in  this  process,  and  is  the  essential  thing  which 
makes  it  work,  you  may  call  hydrogen  a  flux,  or  whatever  you 
will.  It  is  not  essential  to  the  operation  of  the  process,  but  it  is 
the  most  convenient  way  of  preventing  oxidation  and  of  carrying 
on  the  process.  It  can  be  done  in  a  lamp  exhausted  down  to  very 
low  pressures.  I  will  go  so  far  as  to  say  that  from  my  experience 
and  knowledge  of  the  facts,  an  iron  wire  can  be  chromized  inside 
of  a  lamp  where  the  pressure  is  extremely  low,  as  low  as  the  best 
vacuum  we  know, 

I  have  treated  cold-rolled  iron  in  this  same  way  and  at  these 
same  temperatures  in  vacuum,  and  produced  the  same  results. 
I  should  say  that  good,  clean  contact  surfaces  between  the  pow- 
dered chromium  and  iron,  and  an  atmosphere  where  oxidation 
can  not  take  place,  or  a  vacvmm,  are  the  essential  conditions  under 
which  this  temperature  treatment  should  take  place. 

In  regard  to  the  wire  to  which  IMr.  Richardson  has  referred, 
he  is  dealing  with  a  nickel-iron  alloy  wire,  I  assume,  with  a  high 
nickel  content.  Nickel  and  chromium  form  a  eutectic  alloy,  which 
melts  below  1,300°  C,  and  in  his  case  a  ternary  alloy  is  probably 
formed.  It  is  an  entirely  different  material  from  cold-rolled 
iron  which  my  data  cover.  As  to  the  nickel-steel  buckets,  to  which 
I  referred  in  my  paper,  I  wish  to  make  clear  that  the  data  given 
in  this  paper  do  not  deal  with  the  rate  of  penetration  of  chromium 
in  nickel  steel,  but  only  in  cold-rolled  iron.  I  have  no  accurate 
data  on  the  diffusion  of  chromium  into  nickel  steel  or  nickel-iron 
alloys  of  high  nickel  content  at  1,100°  C,  but  I  do  know  that  it 
is  possible  to  chromize  pure  nickel  in  this  same  way  and  at  lower 
temperatures,  and  that  chromized  nickel  steel  resists  corrosion  to 
a  marked  degree. 

You  can  chromize  many  other  metals.  In  fact,  this  diffusion 
of  metals  at  high  temperatures  and  below  their  melting  points 
occurs  generally.  This  is  a  wide  field  for  investigation  and  little 
is  known  about  what  is  really  going  on  outside  of  the  fact  that 
diffusion  takes  place. 


A  pafer  presented  at  the  Forty-thira 
General  Meeting  of  the  American  Elec 
trochemical  Society  held  in  New  York 
City,  May  5,  1923,  Dr.  F.  M.  Becket  in 
the  Chair. 


THE  PREPARATION  OF  PLATINUM  AND  OF  PLATINUM-RHODIUM 
ALLOY  FOR  THERMOCOUPLES.' 

By  Robert  P.   Neville.* 

Abstract. 
The  Bureau  of  Standards  has  prepared  in  its  laboratories 
thermo-element  platinum  and  platinum-rhodium  alloy  for  standard 
thermo-couples,  to  determine  what  performance  might  justly  be 
required  of  such  instruments.  Melting  of  the  pure  metal  and 
of  the  alloy  was  carried  out  in  an  Ajax-Northrup  high  frequency 
induction  furnace,  in  crucibles  of  lime  or  thoria.  Platinum  and 
platinum-rhodium  alloy,  superior  in  quality  to  the  best  material 
of  this  kind  formerly  in  the  possession  of  the  Bureau,  was 
prepared. 


I.     INTRODUCTION. 

One  of  the  essential  properties  of  thermocouples  is  constancy 
of  calibration.  Deficiencies  in  this  property  may  be  due  to  several 
causes,  chief  among  which  are  inhomogeneity  in  the  alloy  wire 
and  contamination  of  either  the  pure  metal  or  the  alloy.  Deterior- 
ation may  be  due  either  to  introduction  of  impurities  during  use, 
which  is  especially  true  of  rare-metal  couples,  or  to  impurities 
in  the  metal  and  alloy  from  which  the  thermocouple  was  made. 
Lack  of  constancy  in  calibration  caused  by  contamination  during 
use  may  be  prevented  by  proper  precautions,  but  a  solution  of 
the  problem  when  the  deterioration  is  due  to  a  lack  of  sufficient 
purity  in  the  original  metals  is  less  easily  attained. 

As  a  part  of  its  general  investigation  of  the  metals  of  the 
platinum  group  the  Bureau  of  Standards  was  desirous  of  making 

'  Published  by  the  permission  of  the  Acting  Director  of  the  Bureau  of  Standards  of 
the  U.    S.   Department  of   Commerce.      Manuscript   received    February   2,    1923. 
*  Associate  chemist,  Bureau  of  Standards,   Washington,  D.   C. 

371 


372  ROBERT    P.    NEVILLE. 

up  in  its  own  laboratories  standard  rare-metal  thermocouples  to 
determine  what  performance  might  justly  be  required  in  such 
instruments. 

II.      MELTING  TECHNIQUE. 

r  1.  Requisites:  Platinum  melting  is  usually  done  with  an  oxy- 
hydrogen  or  oxy-gas  flame  on  a  fire-clay  or  lime  refractory. 
When  extreme  purity  of  the  fused  metal  is  of  utmost  importance, 
the  method  of  heating  and  the  composition  of  the  crucible  must 
be  considered  with  respect  to  other  conditions  than  simply  re- 
fractoriness and  sufficiently  high  temperature.  In  addition  to 
possessing  the  usual  necesssary  refractory  qualities,  the  container 
in  which  pure  platinum  is  melted  should  be  a  material  free  from 
all  substances  which  might  alloy  with  the  metal,  either  directly 
or  after  reduction  by  the  molten  platinum.  It  also  must  be  a 
material  which  does  not  appreciably  dissociate  when  heated  to 
very  high  temperatures  under  vacuum. 

The  first  essential  of  a  method  of  heating  is  the  attainment 
of  the  temperature  at  which  platinum  melts,  but  in  addition  to 
this  the  method  also  must  be  such  as  not  to  promote  decomposi- 
tion of  the  refractory.  Calcium  oxide  is  sufficiently  reduced 
by  an  oxy-hydrogen  flame,  especially  if  the  flame  is  deficient  in 
oxygen,  for  calcium  to  be  detected  in  the  platinum  thus  melted. 
Other  oxides  behave  in  a  similar  manner.  The  method  of  heating, 
therefore,  must  be  such  as  not  to  favor  the  reduction  of  the 
refractory. 

^  2.  Furnace :  The  Ajax-Northrup  high  frequency  induction 
furnace  is  particularly  well  adapted  to  the  melting  of  pure  plat- 
inum, and  the  preparation  of  platinum  metal  alloys.  A  descrip- 
tion of  the  furnace  and  a  mathematical  explanation  of  the  theory 
of  its  method  of  heating  may  be  found  in  a  paper  by  Dr. 
Northrup.^  A  small  inductor  coil  especially  adapted  to  melting 
small  amounts  of  platinum  was  made  for  this  work.  It  differed 
from  the  usual  coils  only  in  size,  the  inside  diameter  being  3  cm., 
and  the  length  10  cm.  A  25  kva.  converter  supplied  the  high 
frequency  current. 

3.  Refractories :  Molded  crucibles  of  the  necessary  size,  shape 
and   composition  were  not  available.     Hand  tamped   crucibles, 

»  E.   F.  Nortliiiip,  Tians.   Am.   Electrochem.   Soc.   35,  09-158    (1919). 


PREPARATION  OF  PLATINUM  FOR  THERMOCOUPLES.  373 

usually  either  of  lime  or  thoria,  were  made  from  materials  pre- 
pared in  the  Bureau's  laboratories  for  this  purpose. 

Calcium  oxide  has  the  advantage  of  being  the  least  expensive 
and  the  most  easily  purified  of  the  refractory  materials  tried. 
Its  use  is  advantageous  for  small  melts  where  solidification  of 
the  metal  is  allowed  to  take  place  in  the  container.  The  crucibles 
are  less  troublesome  to  make,  the  resulting  ingots  may  be  easily 
cleaned  with  hydrochloric  acid,  and  the  quality  of  the  metal 
produced  compares  favorably  with  the  best.  The  special  purifi- 
cation of  the  material  employed  at  the  beginning  of  the  work  was 
later  found  unnecessary.  The  oxide  obtained  by  igniting  "c.  p." 
calcium  carbonate  at  about  1000°  C.  in  an  electric  muffle  furnace 
was  found  just  as  satisfactory. 

The  thoria  used  was  prepared  from  the  "c.  p."  nitrate  of  com- 
merce by  successive  precipitations  as  Th(OH)4  and  finally  as 
the  oxalate,  and  ignition  in  an  electric  muffle  furnace  at  about 
1000°  C.  Calcination  at  a  higher  temperature  would  have  been 
desirable,  but  means  of  obtaining  a  higher  temperature  without 
danger  of  contamination  were  not  available.  It  was  necessary 
to  use  thoria  whenever  large  quantities  of  the  purest  metal  attain- 
able were  desired,  or  whenever  the  melt  was  to  be  poured.  Cruci- 
bles of  thoria  have  a  high  density,  and  unusual  mechanical 
strength  for  an  unsintered  material.  Their  very  low  thermal 
conductivity  makes  them  more  suitable  than  lime  for  large  melts, 
where  the  heat  capacity  is  much  greater.  Ingots  melted  in  thoria, 
however,  are  very  troublesome  to  clean.  Several  fusions  in 
potassium  pyrosulfate  are  necessary  to  dissolve  the  refractory 
still  remaining  after  all  possible  has  been  removed  by  mechanical 
means. 

The  best  lot  of  platinum  that  has  been  prepared  up  to  the 
present  time  was  melted  in  thoria.  It  was  found  to  be  10  micro- 
volts thermoelectrically  negative  at  1200°  C.  to  the  Bureau's 
standard,  known  as  "K."  This  standard,  K,  was  a  melt  made  in 
lime  and  was  45  microvolts  negative  at  1200°  C.  to  the  best 
Heraeus  thermo-element  wire  formerly  used  as  a  standard.  This 
later  melt  in  thoria  then  superseded  K  as  a  standard. 

Zirconium  oxide,  if  pure,  would  probably  serve  as  well  as 
thoria.    A  few  melts  were  made  in  zirconium  oxide  prepared  by 


374  ROBERT   P.    NEVILLE. 

igniting  Kahlbaum's  zirconium  nitrate  in  an  electric  muffle  fur- 
nace. Platinum  melted  in  crucibles  of  this  oxide  was  about  30 
microvolts  positive  to  the  standard  then  used,  or  40  microvolts 
positive  to  the  present  standard.  The  number  of  these  melts  was 
insufficient  to  justify  definite  conclusion  as  to  its  suitability. 

The  same  general  method  of  making  crucibles  was  used  for 
the  different  refractory  materials.  The  procedure  was  that  of 
tamping  the  refractory  in  a  cylindrical  crucible  of  alundum  of 
very  thin  walls.  Thoria,  in  the  dry  powdered  form,  is  sufficiently 
coherent  when  slight  pressure  is  applied  for  crucibles  to  be  made 
without  moistening.  Calcium  oxide  packs  less  easily,  so  this 
material  was  moistened  with  petroleum  ether.  The  hydration  of 
calcium  oxide  necessitated  the  use  of  petroleum  ether,  but  it 
was  also  used  with  thoria  or  zirconia  if  moistening  was  necessary, 
because  of  its  rapid  evaporation  and  the  consequently  quick 
drying  of  the  crucible.  The  mandrel  was  removed  after  tamping, 
and  if  the  material  was  thoria,  the  crucible  was  ready  to  use. 
Lime  crucibles  were  dried  and  lightly  calcined. 

4.  Melting:  A  number  of  experiments  were  made  to  deter- 
mine the  best  method  of  heating,  and  to  work  out  details  of 
melting  and  crucible  making.  The  melting  of  previously  fused 
platinum  in  the  induction  furnace  was  a  simple  matter  so  far 
as  ability  to  obtain  the  necessary  temperature  was  concerned. 
The  melting  of  sponge,  however,  was  found  to  be  more  difficult. 
Sponge  could  be  quickly  heated  to  about  1500°  or  1600°  C,  but 
it  was  very  difficult  to  continue  heating  from  this  point  up  to  the 
melting  point  of  platinum.  If  the  sponge  was  compressed  it 
would  then  heat  as  readily  as  the  solid  platinum.  The  method 
tried  first  was  to  melt  a  small  piece  of  the  compressed  sponge 
and  then  to  add  to  this  the  remainder  of  the  platinum  as  un- 
compressed sponge.  The  time  required  for  the  addition  of  this 
uncompressed  sponge  allowed  excessive  shrinkage  of  the  refrac- 
tory, and  thereby  frequently  caused  failure  and  loss  of  the  melt. 
For  this  reason  it  was  found  more  satisfactory  to  compress  all 
the  sponge  into  pellets. 

The  compression  block  used  for  this  purpose  was  a  steel  cylin- 
der of  uniform  diameter,  highly  polished  and  "glass"  hard  on 
the  inner  surface,  closed  with  a  tightly  fitting  plug  at  one  end 


PREPARATION  OF  PLATINUM  FOR  THERMOCOUPLES.  375 

and  a  removable  plunger  at  the  other.  With  the  plug  in  place 
the  block  was  filled  with  sponge  and  the  plunger  inserted.  Pres- 
sure was  applied  to  the  plunger  until  the  sponge  was  compressed 
into  a  compact  mass,  after  which  the  plug  was  released  and  the 
pressure  reapplied  which  forced  out  the  plug  followed  by  the 
pellet  of  platinum.  The  cylindrical  pellets  made  in  this  manner 
were  1  cm.  in  diameter  and  about  1.5  or  2  cm.  long.  They  had 
the  bright  metallic  appearance  of  fused  metal  and  possessed 
sufficient  mechanical  strength  so  that  no  particular  care  was 
necessary  in  handling  them. 

Experiments  were  carried  out  varying  the  rate  of  heating,  the 
temperature  of  the  melt,  length  of  time  the  melt  was  kept  molten, 
and  the  number  of  repeated  meltings.  In  these  experiments  the 
metal  was  allowed  in  every  instance  to  solidify  in  the  crucible  in 
which  it  was  made.  Superheating  to  any  extent  was  found  not 
only  to  be  of  no  advantage  but  undesirable,  and  to  subject  the 
melt  to  danger  of  loss  through  shrinking  and  cracking  of  the 
refractory.  Excessive  heating  increased  liability  of  contamina- 
tion. The  method  found  to  produce  the  best  results  consisted  in 
a  rapid  heating  to  a  temperature  just  below  the  melting  point, 
followed  by  a  much  slower  heating  to  fusion.  After  dropping 
the  temperature  so  as  to  permit  partial  solidification,  and  re- 
melting  two  or  three  times  with  alternate  scant  solidification,  the 
metal  was  allowed  to  cool  slowly  to  below  the  freezing  point. 
The  slower  rate  of  heating  just  before  fusion  kept  the  tempera- 
ture from  suddenly  running  up  too  high  when  the  metal  melted. 
Likewise  there  was  more  certainty  that  the  metal  was  not  being 
heated  to  an  excessive  temperature  when,  instead  of  being  kept 
continuously  molten,  it  was  remelted  several  times  with  alternate 
scant  solidification.  If  the  temperature  was  allowed  to  run  too 
high,  or  the  metal  kept  molten  an  excessively  long  time,  shrinkage 
took  place  in  the  refractory  (especially  if  it  was  thoria),  which 
allowed  cracks  to  develop  into  which  the  metal  would  run,  and 
cause  the  ingot  to  have  an  irregular  shape  and  uneven  surface. 
A  calcination  of  the  refractory  at  a  higher  temperature  would 
have  been  a  means  of  preventing  this,  if  calcination  could  have 
been  accomplished  without  contamination.  A  method  of  calcining 
thoria  crucibles  under  vacuum  in  a  tungsten  shell  will  be  discussed 
below. 


376  ROBERT   P.   NEVILLE. 

Several  experimental  melts  were  carried  out  in  which  attempts 
were  made  to  control  shrinkage  cavities  by  regulating  the  method 
of  solidification,  so  as  to  obtain  progressive  freezing  of  the  ingot 
from  the  bottom  toward  the  top.  In  this  manner  the  diminution 
in  the  volume  of  the  metal  upon  transition  from  the  liquid  to 
the  solid  phase  could  be  localized  at  the  top  of  the  ingot.  The 
method  of  controlling  the  order  of  freezing  consisted  in  cooling 
through  the  solidification  temperature,  by  gradually  lowering  the 
crucible  down  through  the  furnace  inductor  coil  without  any 
change  in  the  power  input  of  the  furnace.  This  progressive 
freezing  of  the  melt  was  obtained  by  lowering  the  crucible  through 
the  inductor  coil  by  means  of  a  screw  in  the  crucible  support,  the 
power  input  remaining  unchanged  during  the  process.  The  bottom 
thus  began  to  cool  first,  and  solidification  was  progressive  from 
the  bottom  to  the  top  of  the  ingot.  Thus  since  the  direction  of 
freezing  was  entirely  lengthwise  in  the  crucible,  the  shrinkage  was 
localized  at  the  top,  and  any  cavity  was  at  the  top  rather  than  in 
the  interior  or  on  the  side  of  the  ingot.  The  method  was  not 
entirely  successful,  because  the  longer  time  required  for  solidifi- 
cation in  this  manner  often  caused  the  failure  of  the  refractory 
and  consequent  loss  of  the  melt. 

5.  Casting :  The  preparation  of  platinum  alloys  introduces  a 
difficulty  which  does  not  accompany  the  melting  of  pure  platinum, 
namely,  inhomogeneity  in  composition,  resulting  from  selective 
freezing  upon  solidification.  Selective  freezing  may  be  prevented 
by  extremely  slow  cooling  with  stirring,  or  by  so  sudden  a  tran- 
sition from  the  liquid  to  the  solid  phase  that  segregation  can 
not  take  place.  Casting  the  melt  in  a  chill  mold  is  the  obvious 
solution,  hence  the  first  requirement  is  a  crucible  of  sufficient 
mechanical  strength  to  permit  pouring.  This  mechanical  strength 
was  not  present  in  the  crucibles  used  for  melting  pure  platinum, 
where  solidification  took  place  in  the  crucible.  This  again  brought 
up  the  question  of  a  feasible  method  of  calcining  crucibles. 

The  preparation  of  hard-burned  crucibles  from  compressed 
refractory  powders,  without  the  calcination  of  the  crucibles  before 
their  removal  from  the  shells  in  which  they  were  molded,  was 
practically  impossible.  Tungsten  seemed  to  be  the  only  practical 
material  in  which  this  calcination  could  be  carried  out  without 


PREPARATION  OF  PLATINUM  FOR  THERMOCOUPLES.  377 

detriment  to  the  quality  of  the  crucible.  Graphite  so  used  caused 
the  formation  of  carbide  in  the  refractory.  Recent  work*  had 
shown  that  thoria  is  slightly  reduced  by  tungsten  at  temperatures 
below  2300°  C,  but  apparently  not  enough  to  interfere  with  its 
utilization   for  the  present  purpose. 

Crucibles  were  made  by  tamping  thoria,  previously  calcined  at 
1800°  C,  in  cylinders  of  sheet  tungsten.  The  thoria  lined 
tungsten  shells  were  calcined  in  an  electric  vacuum  furnace  to 
about  1800°  C.  The  resulting  sintered  crucibles  were  very  hard 
and  possessed  good  mechanical  strength,  but  as  a  precaution 
they  were  not  used  without  the  reinforcement  of  an  outer  crucible 
of  alundum.  Any  space  intervening  between  the  thoria  crucible 
and  the  alundum  shell  was  filled  in  carefully  with  finely  ground 
thoria.  Castings  were  in  a  few  instances  made  from  uncalcined 
crucibles,  but  the  fragility  of  the  thoria  was  a  source  of  annoyance. 

The  melting  procedure  for  casting  was  the  same  as  usual  except 
that,  instead  of  allowing  the  melt  to  solidify  in  the  furnace,  the 
crucible  was  removed  from  the  coil  and  the  melt  poured  into 
a  graphite  mold,  made  by  drilling  out  the  desired  ingot  shape  in 
a  large  block  of  Acheson  graphite. 

A  melt  to  be  poured  must  be  superheated  somewhat  more  than 
one  permitted  to  solidify  in  the  crucible,  or  solidification  will 
take  place  before  pouring  is  possible,  particularly  when  the 
melting  has  been  done  in  a  sintered  crucible  whose  thermal 
conductivity  is  greater.  The  pouring  temperature  was  kept  as 
near  the  freezing  point  as  possible,  as  high  casting  temperatures 
were  found  to  cause  unsoundness  in  the  ingot.  The  amount  of 
gas  dissolved  by  the  molten  metal,  especially  the  platinum-rhodium 
alloy,  apparently  increased  as  the  temperature  of  heating  was 
raised,  which  seemed  to  cause  more  blow  holes  on  freezing. 
However,  if  the  melt  was  held  for  a  short  time  at  the  lowest 
temperature  permitting  pouring  without  premature  freezing,  little 
gas  was  evolved  on  solidification  in  the  mold,  and  chances  for  a 
sound  ingot  were  greater. 

III.      MECHANICAL  WORKING. 

1.  Rolling:  The  ingots  were  rolled  through  5  cm.  (2  in.) 
diamond  grooved  hard  steel  rolls.     The  grooves  were  graduated 

*  C.  J.  Smithells,  Reduction  of  Thorium  Oxide  by  Metallic  Tungsten  Tour 
Chem.  Soc.   (Lon.),   122,  2236   (1922),  ^        '    •" 


37^  ROBERT  P.    NEVILLE. 

in  size  from  19  mm.  (3/4  in.)  square  for  the  first  to  2  mm.  (5/64 
in.)  for  the  last,  which  had  slightly  rounded  corners.  In  order 
to  prevent  contamination  during  mechanical  working  particular 
care  was  taken  to  keep  the  roll  surfaces  in  the  best  possible 
condition.  Spectrographic  analysis  revealed  no  trace  of  iron  in 
pure  platinum  after  rolling. 

2.  Draw'mg:  Sapphire  dies  were  used  for  drawing  the  wire 
from  the  5/64-inch  rod.  Before  drawing,  the  wire  was  cleaned 
by  rubbing  between  filter  paper  saturated  with  alcohol  to  remove 
grease  and  any  adhering  flakes  of  metal.  The  reductions  in  the 
dies  were  0.0076  mm.  (0.003  in.)  at  each  draft  at  the  start. 
The  last  few  drafts  were  slightly  less.  The  final  diameter  of  the 
wire  was  0.63  mm.   (0.0246  in.) 

Platinum  is  so  malleable  that  unevenness  of  the  ingot  and  many 
other  defects  may  be  rolled  out  and  obscured.  Such  flaws  pos- 
sibly may  be  cold-welded  so  that  they  are  as  sound  as  any  portion 
of  the  metal,  but  since  there  was  some  uncertainty,  discards  were 
always  made  from  both  ends  of  the  drawn  wire.  The  wire  was 
cleaned  after  drawing  in  the  same  manner  as  after  rolling. 

IV.      THERMO-ELEMENT  PLATINUM. 

1.  Sponge:  The  separation  of  the  metals  and  the  purification 
of  the  sponge  will  not  constitute  a  part  of  this  paper.  It  is 
assumed  here  that  the  materials  melted  were  in  every  case  of 
the  highest  degree  of  purity  attainable.  Preparatory  to  melting 
the  sponge  was  pressed  into  small  cylinders  of  sufficient  density 
to  permit  heating  by  direct  induction. 

2.  Melting:  Several  different  lots  of  platinum  were  melted 
with  slight  variations  in  method.  This  description  will  follow 
in  detail  the  method  of  melting  the  best  of  the  large  ingots  made. 

The  melting  was  done  in  a  thoria  crucible  made  in  an  outer 
crucible  of  alundum  as  described  above.  Its  inside  diameter  was 
slightly  greater  than  that  of  the  cylinders  of  compressed  sponge. 
The  cylinders  of  sponge  were  placed  in  the  crucible,  one  on  top 
of  the  other,  until  the  crucible  was  full,  and  the  whole  set  in  the 
coil  of  the  furnace  ready  for  melting.  The  graphite  mold  was 
set  a  few  inches  away  from  the  furnace,  and  the  remainder  of 
the  compressed  cylinders  for  the  charge  placed  conveniently  for 


PREPARATION  OP  PLATINUM  FOR  THERMOCOUPLES.  379 

quick  addition  after  fusion  had  begun.  The  furnace  was  started 
at  a  power  input  of  5  kw.  and  reduced  to  about  4  kw.  before 
fusion  had  begun,  which  usually  occurred  in  about  a  minute. 
Quick  adjustment  of  power  input  permitted  ready  control  of  the 
temperature  of  the  melt.  As  the  first  part  of  the  charge  melted 
and  sank  the  remaining  cylinders  were  added.  As  soon  as  the 
mass  was  thoroughly  liquid  after  all  the  sponge  had  been  added, 
it  was  allowed  to  cool  to  superficial  solidification  and  again  melted. 

Two  or  three  successive  remeltings  with  alternate  superficial 
solidification  were  a  means  of  preventing  unintentional  super- 
heating, and  at  the  same  time  assured  that  the  time  the  melt  was 
molten  was  long  enough  for  the  volatihzation  of  any  remaining 
salt  or  other  foreign  matter  in  the  sponge.  Upon  melting  the 
last  time  the  temperature  was  carried  on  up  until  judged  high 
enough  for  the  metal  to  remain  liquid  until  it  could  be  poured. 
The  power  was  cut  ofif,  the  crucible  removed  with  tongs  and  the 
melt  poured  as  quickly  as  possible.  Solidification  was  almost 
instantaneous. 

The  main  body  of  the  ingot  was  about  1  cm.  in  diameter  and 
6  cm.  long.  The  top  part  of  the  mold  was  of  a  larger  diameter 
so  as  to  provide  for  a  head  which  would  solidify  last  and  confine 
the  shrinkage  cavity  to  the  top.  The  ingot  (124  g.  in  weight) 
was  sound,  of  smooth  surface,  and  with  no  sign  of  a  defect. 
The  shrinkage  cavity  on  solidification  was  localized  at  the  top 
of  the  head,  which  was  cut  oflf  before  the  mgot  was  rolled. 

3.  Wire :  Since  this  ingot  was  cast,  no  cleaning  was  necessary 
before  rolling  as  was  the  case  when  the  melt  was  allowed  to 
solidify  in  the  furnace.  Pure  platinum  is  so  ductile  that  rolling 
and  drawing  are  simple  matters.  A  slight  contamination  of 
very  pure  platinum  noticeably  increases  the  hardening  resulting 
from  the  cold  working  during  drawing.  After  roUing  and  draw- 
ing to  0.63  mm.  (0.0246  in.)  wire,  without  any  annealing  during 
the  process,  the  pure  platinum  wire  was  still  soft.  After  cleaning 
and  making  discards  from  the  ends,  the  wire  was  ready  for 
testing  and  cutting  into  convenient  lengths  for  thermocouples. 

Spectrographic  analysis  failed  to  reveal  the  presence  of  any 
impurit}\  Results  of  thermo-electric  comparisons  and  service 
tests  are  discussed  below. 


380  ROBERT  P.   NEVILLE. 

V.     THERMO-ELEMENT    ALLOY. 

1.  Materials  and  Melting:  The  platinum  sponge  used  in  mak- 
ing the  90  per  cent  platinum- 10  per  cent  rhodium  alloy  was  the 
same  as  that  used  for  the  platinum  element,  and  was  handled  in 
the  same  way.  The  rhodium,  however,  was  not  added  as  sponge 
but  in  the  form  of  a  fused  ingot.  The  rhodium  sponge,  approxi- 
mating in  weight  a  tenth  of  the  total  weight  of  alloy  to  be 
made,  was  compressed  into  a  cylinder  and  fused  under  vacuum. 
This  preliminary  vacuum  fusion  eliminated  any  gaseous  and 
volatile  matter  in  the  sponge  and  facilitated  the  addition  of  a 
definite  amount  of  rhodium  to  the  melt. 

The  melting  procedure  for  the  rhodium  differed  little  from  the 
melting  of  pure  platinum.  The  crucible  was  of  powdered  cal- 
cium oxide  pressed  into  a  shell  of  alundum.  This  crucible,  con- 
taining the  compressed  rhodium  sponge,  was  placed  in  the  bottom 
of  a  closed-end  hard  glass  tube  and  set  in  the  coil  of  the  induc- 
tion furnace.  The  rhodium  was  fused  under  vacuum,  held  molten 
a  few  minutes,  and  allowed  to  cool  in  the  crucible  while  the 
vacuum  was  maintained. 

The  charge  was  then  accurately  calculated  on  the  basis  of  the 
weight  of  the  cleaned  rhodium  ingot,  the  platinum  weight  being 
nine  times  the  weight  of  rhodium.  In  melting  the  alloy  the  pro- 
cedure Avas  the  same  as  described  above  for  melting  pure  platinum. 
The  rhodium  ingot  was  dropped  in  while  the  platinum  was  molten 
so  it  would  dissolve  quickly  and  not  be  exposed  to  the  air  long 
while  at  a  high  temperature.  The  method  of  pouring  the  alloy 
differed  from  that  of  pure  platinum  only  in  that  more  care  had 
to  be  taken  that  the  alloy  was  not  too  hot  when  poured ;  otherwise 
an  unsound  ingot  resulted. 

2.  Wire:  The  alloy  was  rolled  and  drawn  to  0.63  mm. 
(0.0246  in.)  wire  in  the  same  manner  as  the  pure  platinum, 
except  for  annealing,  which  was  unnecessary  with  the  latter. 
The  alloy  hardens  more  with  deformation  than  pure  platinum, 
so  it  was  annealed  at  frequent  intervals  during  the  mechanical 
working.  The  finished  wire  was  as  smooth  and  uniform  as  the 
pure  platinum  wire. 

The  alloy  was  tested  spectrographically  for  contamination,  with 
negative   results.      Xo    difficulty    was    experienced    in    preparing 


PREPARATION  OF  PLATINUM  FOR  THERMOCOUPLES.  38 1 

alloys  of  the  desired  composition  by  direct  synthesis.  Complete 
homogeneity,  however,  was  the  doubtful  point,  so  thermo-electric 
tests  were  made  to  detect  any  inhomogeneity  in  composition. 
Careful  thermo-electric  comparisons  made  at  frequent  intervals 
along  the  entire  length  of  the  wire  indicated  a  maximum  dif- 
ference in  composition  corresponding  to  less  than  1°  at  1200°  C. 
In  lengths  suitable  for  thermocouples  there  was  no  significant 
difference  in  e.m.f.  between  the  opposite  ends. 

VI.  SERVICE  TEST. 
One  of  the  thermocouples  made  as  described  and  designated 
as  CI  was  subjected  to  continuous  heating.  After  a  flash  an- 
nealing, which  consisted  in  heating  the  cold-drawn  wire  to  about 
1500°  C.  by  the  momentary  passage  of  an  electric  current,  the 
couple  was  compared  with  the  standard  couple.  It  was  then 
heated  for  25  hours  at  about  1600°  C.  by  passing  an  electric 
current  through  the  wire  suspended  in  air,  and  again  compared 
with  the  standard.  The  platinum  and  alloy  were  found  to  have 
dropped  5.5  and  14.5  microvolts,  respectively,  at  1200°  C,  the 
equivalent  of  less  than  1°  C,  which  is  about  the  usual  drop  noticed 
upon  annealing  preparatory  to  calibration. 

Life  tests  made  at  the  Bureau  on  thermocouples  of  commercial 
manufacture  have  shown  that  the  usual  change  in  calibration 
resulting  from  18  to  24  hours  heating  at  1500°  to  1600°  C.  sub- 
sequent to  preliminary  annealing  is  from  3°  to  10°  C. 

Upon  calibrating  CI,  subsequently  to  the  treatment  mentioned 
above,  its  calibration  curve  was  found  to  be  almost  identical  with 
the  standard  temperature-e.m.f.  curve  for  platinum-platinum- 
rhodium  thermocouples  as  given  by  the  Geophysical  Laboratory 
of  the  Carnegie  Institution.'* 

VII,      SUMMARY. 

1.  As  part  of  its  general  investigation  of  the  platinum  metals 
now  in  progress,  the  Bureau  of  Standards  desired  to  prepare 
in  its  own  laboratories  standard  rare-metal  thermocouples  in 
order  to  determine  what  performance  might  justly  be  required 
of  such  instruments. 

6  Adams,  Bull,  A.  L  M.  M.  E.,  159,  2111   (1919). 


382  DISCUSSION. 

2.  A  method  of  melting  was  developed  which  consisted  in 
fusing  the  sponge  in  crucibles  of  pure  thoria  or  lime  by  means 
of  an  Ajax-Northrup  high  frequency  induction  furnace.  The 
crucibles  were  made  by  tamping  the  powdered  material  around 
a  mandrel  in  an  outer  crucible  of  alundum  or  tungsten.  The 
melts  of  alloy  were  cast  in  a  chill  mold  and  those  of  pure  platinum 
were  usually  allowed  to  sohdify  in  the  furnace. 

3.  Platinum  and  platinum-rhodium  alloy  superior  in  quality 
to  the  best  material  of  this  kind  formerly  in  the  possession  of 
the  Bureau  were  prepared.  Thermocouples  were  made  from 
this  material  which  drop])ed  off  on  25  hours  heating  at  about 
1600°  C.  the  equivalent  of  about  1°  C. 

The  author  wishes  to  make  several  acknowledgments.  The 
preparation  of  the  platinum  and  rhodium  sponges  was  done  by 
E.  Wichers,  chemist;  the  spectrographic  analyses  were  made  by 
W.  F.  Meggers,  physicist ;  the  thermocouple  service  test  was  made 
by  W.  F.  Roeser,  laboratory  assistant ;  the  entire  work  was  con- 
ducted under  the  supervision  of  E.  Wichers  and  Louis  Jordan, 
chemist. 

Other  papers  dealing  with  the  investigation  of  platinum  metals 
by  the  Bureau  of  Standards  are  as  follows: 

G.  K.  Burgess  and  P.  D.  Sale.  A  Study  of  the  Quality  of  Platinum 
Ware,   Bureau  of  Standards   Sci.  Papers,  No.  254. 

G.  K.  Burgess  and  R.  G.  Waltenberg,  Further  Experiments  on  the 
Volatilization  of  Platinum,  Bureau  of  Standards  Sci.  Papers,  No.  280. 

L.  J.  Gurevich  and  E.  Wichers,  Comparative  Tests  of  Palau  and  Rho- 
tanium  Ware  as  Substitutes  for  Platinum  Laboratory  Utensils,  Ind.  Eng. 
Chem.  11,  570  (1919). 

E.  Wichers,  The  Preparation  of  Pure  Platinum,  J.  Am.  Chem.  Soc 
43,   1268    (1921). 

E.  Wichers  and  L.  Jordan,  Investigations  on  Platinum  Metals  at  the 
Bureau  of  Standards,  Trans.  Am.  Electrochem.  Soc.     This  volume. 


DISCUSSION. 

H.  K.  Richardson^  :  The  author  says  that  the  preparation  of 
hard-burned  crucibles  from  compressed  refractory  powders,  with- 
out calcination  of  the  crucibles  before  their  removal  from  the 
shells  in  which  molded,  is  practically  impossible.  We  have  been 
using  thoria  crucibles  the  last  year  or  so,  which  were  made  by 

*  Westinghouse  Lamp   Works,   Bloomfield,  New  Jersey. 


PREPARATION   OF   PLATIXUM   FOR   THERMOCOUPLES.  383 

practically  standard  methods  of  the  ceramic  art,  i.  e.,  by  casting 
and  also  by  pressing  methods.  In  both  cases  we  have,  by  suit- 
able calcination  and  burning,  obtained  a  crucible  which  has  a 
perfectly  smooth  surface,  and  to  which  we  have  found,  in  one  or 
two  cases,  that  platinum  scrap  when  melted  does  not  adhere. 

So  far  we  have  not  been  able  to  make  a  crucible  larger  than  2 
inches  in  diameter  by  6  inches  long.  These  crucibles,  when 
made  in  a  furnace  of  the  carbon-plate  resistance  type,  are  not 
satisfactory  for  use  in  induction  furnaces,  because  they  take  up 
carbon  from  the  atmosphere  of  the  furnace  and  do  not  make  a 
satisfactory  container  at  2,200  to  2,500°  C,  due  to  conducting  the 
current  and  breaking  down. 

These  crucibles  were  made  up  for  uranium  research.  One 
crucible  has  been  carried  to  approximately  2,500°  C.  at  least  seven 
times.  Molten  material  at  this  temperature  has  dropped  into 
the  crucible  without  cracking  same,  showing  in  a  practical  way 
their  low  coefficient  of  expansion.  Under  this  condition  of  use 
they  do  not  soften  or  lose  shape. 

In  the  manufacture  of  the  crucible,  the  thoria  is  very  sensitive 
to  taking  up  various  materials ;  but  after  once  being  made,  except 
for  metallic  iron  in  an  oxidizing  atmosphere,  the  metals  do  not 
seem  to  react  much  with  the  crucibles. 

H.  T.  Reeve-:  Has  Dr.  Jordan  ever  tried  working  down  a  bar 
of  sintered  platinum  to  wire.  Melting  seems  unnecessary  when 
it  requires  such  troublesome  methods  to  prevent  contamination. 

F.  E.  Carter^:  What  does  Dr.  Jordan  mean  by  the  inhomo- 
geneity  of  the  wire?  Does  he  mean  that  one  end  of  a  thermo- 
element wire  has  a  different  composition  from  the  other,  or  that 
there  is  coring  of  the  crystals? 

Also,  on  page  372,  it  is  stated  that  "the  container  in  which  pure 
platinum  is  melted  should  be  a  material  free  from  all  substances 
which  might  alloy  with  the  metal,  either  directly  or  after  re- 
duction by  the  molten  platinum."  Actually,  if  you  use  the  oxy- 
hydrogen  flame,  and  under  oxidizing  conditions,  the  purity  of 
the  lime  crucible  does  not  seem  to  be  of  much  importance.  The 
impurities  are  taken  out  of  the  platinum  by  the  lime  rather  than 
the  impurities  of  the  lime  by  the  platinum.     I  have  found,  for 

-  Western    Electric   Co.,   New  Yorlc  City. 

=  .Afetallurgist,   Baker   &   Co.   Inc.,    Newark,   N.    T. 


384  DISCUSSION, 

instance,  I  could  get  a  purer  platinum  by  melting  in  a  lime  cruci- 
ble with  oxy-hydrogen  flame,  making  sure  it  is  thoroughly  oxi- 
dizing, than  by  using  the  high-frequency  induction  furnace. 

When  the  metal  is  cast  into  a  graphite  mould,  has  the  author 
ever  had  any  indication  of  the  platinum  being  attacked  by  the 
carbon?  Is  a  carbide  formed  under  these  conditions?  I  notice 
the  author  used  a  closed  end,  hard-glass  tube  when  working  with 
a  vacuum  in  the  high-frequency  furnace.  I  have  found  it  more 
convenient  to  work  the  other  way  around,  /.  e.,  to  use  the  ordinary 
vitreosil  insulator  closed  by  a  disc  of  transparent  quartz  at  the 
top,  and  evacuate  at  the  bottom. 

Louis  Jordan*:  In  regard  to  making  the  crucibles,  Mr.  Rich- 
ardson calls  attention  to  the  statement  that  it  is  practically  im- 
possible to  make  them  by  ordinary  methods.  Since  the  preparation 
of  Mr.  Neville's  paper,  work  with  methods  of  casting  the  cruci- 
bles has  been  in  progress. 

With  regard  to  reaction  with  tungsten,  there  is  a  reference  at 
the  bottom  of  page  Z77  to  the  reduction  of  thorium  oxide  by 
metallic  tungsten.  The  work  cited  was  at  a  somewhat  higher 
temperature  than  that  employed  for  the  platinum  melting,  and 
at  1,800°  C.  we  did  not  notice  any  reaction  between  the  tungsten 
and  the  thoria.  This  was  a  thoria-Hned  tungsten  shell,  rather 
than  a  tungsten-lined  thoria.  It  was  calcined  in  an  Arsem  fur- 
nace, but  we  found  no  decrease  in  the  purity  of  the  platinum 
melted  in  such  crucibles,  and  no  apparent  change  in  the  crucible 
indicating  contamination  by  carbon. 

It  is,  of  course,  not  necessary  actually  to  melt  the  compressed 
platinum  sponge.  That  is  an  European  practice,  I  believe.  It  was 
not  difficult  to  melt  the  compressed  platinum  sponge.  A  consid- 
erable charge  of  platinum  sponge  was  melted  in  a  high-frequency 
induction  furnace  in  a  few  minutes,  and  it  was  as  easy  to  melt 
completely  as  to  sinter  and  hammer  the  metal  sponge. 

We  did  not  find  any  trouble  in  casting  the  pure  metal  or  its 
alloys  in  Acheson  graphite  moulds.  The  mould  was  a  chilled 
mould  for  the  amount  of  metal  we  used,  and  freezing  took  place 
instantly  and  without,  as  far  as  we  could  see,  any  reaction  with 
the  carbon, 

*  Bureau  of  Standards,  Washington,  D.  C. 


A  paper  presented  at  the  Forty-third 
General  Meeting  of  the  American  Elec- 
trochemical Society  held  in  Nciv  York 
City,  May  5,  1923,  Dr.  F.  M.  Becket  in 
the  Chair. 


INVESTIGATIONS  ON  PLATINUM  METALS  AT  THE  BUREAU 
OF  STANDARDS.^ 

By  Edward  Wickers^  and  Louis  Jordan.' 

Abstract. 
The  Bureau  of  Standards  has  vmdertaken  a  comprehensive 
investigation  of  the  platinum  metals,  involving  the  purification 
of  all  metals  of  the  platinum  group,  critical  studies  of  analytical 
separation  of  the  platinum  metals,  the  melting  and  mechanical 
working  of  the  pure  metals  and  their  alloys,  the  study  of  selected 
alloys  with  respect  to  their  suitability  for  platinum  ware,  and  the 
determination  of  a  variety  of  physical  properties  of  such  metals 
and  alloys.  The  first  three  phases  of  this  investigation  are 
actively  in  progress ;  the  last  two  phases  are  to  be  undertaken  in 
the  immediate  future. 


I.      INTRODUCTION, 

A  considerable  amount  of  work  on  platinum  and  platinum 
group  metals  has  been  carried  out  or  is  in  progress  in  various 
divisions  of  the  Bureau  of  Standards.  It  is  believed  that  it  will 
be  of  interest  to  give  a  brief  account  of  this  research,  and  call 
to  the  attention  of  those  interested  the  activities  of  the  Bureau  of 
Standards  in  this  field.  It  is  the  Bureau's  desire  to  assist  both 
users  and  the  manufacturers  in  improving  the  standards  of 
quality  and  performance  of  platinum  and  platinum  alloy  products. 

The  importance  of  platinum  and  platinum  metals  for  chemical 
laboratory  ware,  catalysts,  resistance  thermometers,  thermo- 
electric pyrometers,  electrical  contacts,  dental  alloys,  standards  of 
mass  and  length,  as  well  as  their  wide  use  in  jewelry  and  in 

*  Published  by  permission  of  the  Acting  Director,  Bureau  of  Standards,  Department 
of  Commerce.     Manuscript  received  February  2,   1923. 

*  Chemists,  Bureau  of  Standards. 

385 


386  EDWARD   WICHERS   AND   LOUIS   JORDAN. 

numerous  other  miscellaneous  but  important  applications,  are  all 
too  well  known  to  require  more  than  mention. 

1.    Beginning  of  Platinum  Work  at  the  Bureau  of  Standards. 

In  1910  the  American  Chemical  Society  formed  a  committee 
on  quality  of  platinum  laboratory  utensils,  with  Dr.  W.  F. 
Hillebrand,  chief  of  the  division  of  chemistry  of  the  Bureau  of 
Standards,  as  chairman.  This  committee  made  two  reports,  the 
first  in  191 P  and  a  supplementary  one  in  1914.*  In  the  first 
report  were  summarized  the  principal  difficulties  experienced 
with  laboratory  ware  at  that  time,  namely,  (1)  undue  loss  of 
weight  on  ignition;  (2)  undue  loss  of  weight  on  acid  treatment, 
especially  after  ignition;  (3)  discoloration,  crystallization,  or 
frosted  appearance  of  the  surface  after  ignition;  (4)  adherence 
of  crucibles  to  platinum  triangles  after  ignition;  (5)  alkalinity  of 
surface  after  ignition;  (6)  blistering;  (7)  development  of  cracks 
after  continued  heating. 

Following  the  suggestions  of  this  committee  as  to  points 
requiring  investigation.  Burgess  and  Sale,"*  of  the  Bureau  of 
Standards,  determined  losses  on  ignition  and  on  treatment  with 
acid  for  a  number  of  platinum  utensils.  In  order  to  show  the 
relation  of  these  losses  to  the  composition  of  the  platinum  ware, 
they  developed  a  thermoelectric  test  for  purity,  a  test  which  did 
not  injure  the  article  tested,  and  which  furnished  data  which 
made  it  possible  to  classify  the  metal  in  terms  of  its  content  of 
iridium,  at  that  time  the  most  common  impurity  or  alloying 
element. 

This  method  consists  in  clamping  or  arc  soldering  two  pure 
platinum  wires  to  opposite  sides  of  a  platinum  dish  or  crucible 
and  connecting  these  wires  with  a  millivoltmeter.  With  one 
junction  at  room  temperature  or  cooled  in  an  air  jet,  the  other 
junction  was  heated  in  a  small  blast  flame  to  a  definite  tempera- 
ture, say  1,100°  C,  and  the  thermoelectromotive  force  of  the 
impure  platinum  of  the  crucible  against  the  pure  platinum  wire 
was  read.  From  a  chart  of  isothermal  curves  of  electromotive 
force  in  millivolts  against  the  percentage  of  iridium  alloyed  with 
platinum,  the  amount  of  impurity  in  the  crucible  was  found  in 

'J-  Ind.  Eng.  Chem.,  3,  686-91  (1911). 
*J.  Ind.  Eng.  Chem.,  6,  512-13  (1914). 
»  B.  S.  Scientific  Papers  No  2.54;  1915. 


INVESTIGATIONS  ON   PLATINUM   METALS.  387 

terms  of  the  equivalent  iridium  content.  The  crucibles  tested  for 
losses  on  ignition  and  acid  treatment  were  then  classified  accord- 
ing to  their  iridium  content  or  in  one  or  two  instances  accord- 
ing to  their  rhodium  content,  when  this  was  known  to  be  the 
alloying  and  hardening  element. 

Burgess  and  Waltenberg"  carried  out  further  tests  on  the  vola- 
tilization of  platinum,  working  over  a  range  of  temperatures 
from  700  to  1,200°  C.  In  testing  for  volatilization  losses  the 
ware  was  heated  in  an  electric  resistance  furnace,  but  always  with 
a  stream  of  air  passing  through  the  heated  chamber,  since  losses 
in  weight  of  platinum  on  ignition  seem  to  be  influenced  by  the 
presence  of  oxygen.  The  data  obtained  from  these  experiments 
indicated  that  above  900°  C.  the  volatilization  of  platinum  con- 
taining iridium  was  greater  than  that  of  pure  platinum,  and 
increased  with  the  iridium  content  and  with  temperature ;  the 
loss  of  platinum  containing  rhodium  was  less  than  for  pure 
platinum  at  all  temperatures. 

2.    Present  Need  for  Research  on  Platinum  Metals. 

It  was  not  possible  for  the  platinum  committee  to  obtain 
reliable  information  as  to  the  composition  of  platinum  ware 
further  than  that  given  by  the  thermoelectric  test.  That  is  to 
say,  qualitative  information  as  to  the  nature  of  the  impurity 
could  not  be  obtained,  and  little  could  be  done  in  correlating 
composition  with  quality  of  service  in  the  absence  of  definite 
knowledge  of  the  nature  of  the  impurities. 

The  committee  in  their  supplementary  report  outlined  in  the 
following  sentences  the  procedure  which  seemed  to  them  desira- 
ble in  continuing  work  on  the  quality  of  platinum  ware.  ''This  in- 
formation .  .  .  can  be  gained  only  by  carrying  out  an  elabo- 
rate investigation  involving  the  preparation  of  the  pure  metals 
and  some  of  their  alloys,  and  also  by  the  careful  analysis  of 
commercial  ware.  It  is  hoped  that  in  time  the  Bureau  of  Stand- 
ards may  be  able  to  take  up  such  an  investigation.  .  .  .  The 
investigation  should  not  be  restricted  to  a  study  of  the  subject 
from  the  point  of  view  of  the  chemist  alone,  but  should  be  made 
comprehensive  as  to  the  physical  behavior  of  the  metals  and  their 
alloys  so  that  all  users  of  platinum  might  benefit." 

•B.  S.  Scientific  Papers  No.  280;  1916. 


388  EDWARD   WICKERS   AND   LOUIS   JORDAN. 

II.       CURRENT    INVESTIGATION    OF    PLATINUM    METALS. 

The  Bureau  of  Standards  has  recently  been  able  to  commence 
this  comprehensive  investigation  of  platinum  metals.  The  major 
phases  of  this  work  are  the  preparation  of  all  the  platinum  metals 
in  a  state  of  very  high  purity ;  the  development  and  critical 
examination  of  methods  of  analysis,  not  only  chemical,  but  also 
spectrographic  and  thermoelectric  methods ;  the  development  of 
the  technique  of  melting  and  mechanical  working  of  metals  and 
alloys  of  extreme  purity ;  the  preparation  and  testing  for  quality 
of  platinum  laboratory  ware  of  accurately  controlled  composition, 
the  determination  of  selected  physical  properties  of  metals  and 
alloys  of  composition  identical  with  the  ware,  and  the  correlation 
of  composition,  physical  properties,  and  quality  of  service ;  the 
determination  of  the  most  important  physical  constants  and  the 
physical  and  chemical  behavior  of  all  available  platinum  metals 
and  alloys  in  so  far  as  the  facilities  of  the  Bureau  permit. 

1.    Purification  of  Metals. 

It  is  obvious  that  the  first  essential  in  an  investigation  such  as 
was  just  outlined  is  the  preparation  of  each  of  the  platinum 
metals  in  the  highest  possible  degree  of  purity.  This  important 
feature  has  been  neglected  too  often  in  the  past,  and  to  this 
neglect  are  undoubtedly  due  many  of  the  questionable  data  on 
physical  properties  found  in  the  literature.  This  work  of  purifica- 
tion has  already  progressed  to  a  stage  where  quantities  of  each 
metal,  except  ruthenium,  sufficient  for  our  immediate  needs  have 
been  prepared.  Ruthenium  has  been  omitted  thus  far  because 
its  scarcity  probably  will  prevent  any  extensive  application. 

The  purification  of  platinum  and  palladium  has  been  reduced 
to  a  routine  procedure.  A  preliminary  paper  on  pure  platinum 
has  been  published,  this  paper  dealing  particularly  with  the  con- 
tamination of  platinum  with  calcium  when  the  metal  is  melted  in 
lime  crucibles  under  unfavorable  conditions.''  The  preparation 
of  pure  osmium,  involving  a  new  method  for  converting  osmium 
tetroxide  to  quadrivalent  osmium  chloride,  will  be  published 
shortly  in  connection  with  the  re-determination  of  the  atomic 
weight  of  osmium,  the  latter  work  having  been  done  at  The  Johns 
Hopkins  University. 

'E.  Wichers,  J.  Am.  Chem.   Soc,  43,   1268    (1921). 


INVESTIGATIONS  ON   PLATINUM    METALS.  389 

The  Study  of  methods  of  purification  has  developed  consider- 
able material  suitable  for  publication,  but  which  will  first  be 
supplemented  with  additional  work.  It  may  be  stated  that  the 
simple  process  of  repeated  precipitation  with  ammonium  chloride 
is  an  entirely  feasible  method  of  purifying  platinum.  The  exist- 
ing literature  contains  but  little  useful  information  on  the  puri- 
fication of  most  of  the  platinum  metals,  especially  iridium  and 
rhodium. 

It  was  realized  from  the  first  that  the  presence  of  very  small 
quantities  of  impurities  in  the  metals  prepared  would  have  to  be 
detected  by  other  than  chemical  methods.  For  this  purpose  spectro- 
graphic  analysis,  and  the  comparison  of  thermoelectric  force,  and 
the  coefficient  of  electrical  resistance,  have  been  used  with  much 
success.  Spectographic  examination  has  been  applied  to  all  of 
the  metals,  but  the  other  two  methods  have  been  applicable  only 
in  the  case  of  palladium  and  platinum,  the  two  metals  which  can 
be  readily  drawn  into  wire. 

Thermoelectric  comparison  has  been  found  to  be  particularly 
useful  in  controlling  the  purification  of  platinum.  Observations 
are  made  at  an  approximately  fixed  temperature  (1,200°  C.) 
against  an  arbitrary  standard.  Readings  can  be  completed  in  a 
few  minutes,  and  the  sensitiveness  is  far  in  excess  of  the  require- 
ments. The  e.  m.  f.  can  be  measured  to  tenths  of  microvolts 
without  difficulty,  and  dififerences  of  10  or  even  15  microvolts 
can  hardly  be  interpreted  in  terms  of  a  definite  impurity  even  by 
means  of  the  spectroscope.  All  evidence  indicates,  however,  that 
the  purest  samples  (as  prepared  from  the  usual  sources)  are 
the  most  negative.  The  best  samples  of  platinum  thus  far  pre- 
pared gave  an  e.  m.  f .  of  about  30  microvolts  negative  to  the  best 
material  (consisting  of  a  single  sample)  to  which  the  Bureau 
had  had  access  previously.  One  of  these  best  samples  is  now 
used  as  the  standard  for  thermoelectric  comparison,  and  all 
platinum  prepared  is  required  to  give  an  e.  m.  f .  of  not  more  than 
15  microvolts  positive  to  this  sample  at  1,200°  C. 

A  series  of  alloys  of  rhodium  in  platinum  and  one  of  iridium 
in  platinum,  both  from  1  per  cent  down  to  0.001  per  cent,  were 
prepared  to  determine  the  thermoelectric  behavior  of  dilute 
alloys  toward  pure  platinum.  It  was  found  that  in  these  two 
series  the  variation  in  e.  m.  f.  was  directly  proportional  to  com- 


390  EDWARD   WICHE;RS   and   LOUIS    JORDAN. 

position ;  that  is,  the  isothermal  curve  of  e.  m.  f .  plotted  against 
composition  was  practically  a  straight  line,  for  alloys  up  to  1 
per  cent.*  Thermoelectric  comparison  has  been  applied  in  a 
similar  way  to  palladium,  particularly  to  the  metal  which  is  being 
used  to  determine  the  palladium  melting  point  on  the  optical 
scale  of  temperature. 

Less  use  has  been  made  of  the  determination  of  the  cqefificient 
of  electrical  resistance.  This  method  is  far  less  sensitive  than  the 
thermoelectric  comparison,  takes  more  time,  and  so  far  as  it  has 
been  used  has  not  given  much  additional  information.  It  is  gen- 
erally accepted  that  the  coefificient  of  resistance  increases  with 
increasing  purity.  The  thermo  element  platinum  purchased  by 
the  Bureau  in  the  past  few  years  has  usually  had  a  coefificient 
(between  0°  and  100°)  of  about  0.003910,  while  one  of  a  group 
of  three  samples,  which  were  of  a  foreign  manufacture,  gave  a 
value  of  0.003917.  A  sample  of  American  manufacture,  recently 
received,  gave  a  value  of  0.003906  as  received,  and  0.003917  after 
heating  at  1,500°  for  several  hours.  Samples  of  platinum  pre- 
pared at  the  Bureau  have  given  a  coefificient  up  to  0.003917  as 
drawn  and  up  to  0.003922  after  a  period  of  heating  such  as  that 
just  described. 

This  method  as  well  as  the  thermoelectric  comparison  fails  to 
give  any  information  as  to  the  nature  of  the  impurity  or  impuri- 
ties which  are  present.  For  this  purpose  the  method  of  spectro- 
graphic  analysis  is  used.  The  application  of  this  to  platinum  has 
been  described  by  Meggers,  Kiess  and  Stimson.^  The  sensitive- 
ness of  the  method  to  the  most  persistent  impurities  was  deter- 
mined by  examining  series  of  progressively  diluted  alloys,  down 
to  0.001  per  cent.  It  is  believed  that  the  presence  of  0.001  per 
cent  of  any  of  the  usual  impurities  in  the  platinum  metals  can 
be  detected  in  the  spectrogram.  The  aim  in  routine  purification 
is  to  prepare  material  which  shows  no  lines  of  any  impurities, 
except  in  some  cases  the  faintest  lines  of  the  elements  in  the 
refractories  used  for  melting.  Sometimes  even  these  can  be 
eliminated. 

Table  I  will  serve  as  an  illustration  of  the  way  in  which 
spectrographic  analysis  and  the  thermoelectric  comparison  were 

'  C.  O.  Fajrchild.  Communication  to  the  Philosophical  Society  of  Washington, 
Feb.   11.  1922. 

•  B.   S.    Scientific    Papers   No.   444,   "Practical    Spectrographic   Analysis;"    1922. 


INVESTIGATIONS  ON   PLATINUM   METALS. 


391 


used  in  the  control  of  the  process  of  purification.  No.  94 
is  a  sample  taken  from  a  500-g.  lot  of  sponge  of  commer- 
cial purity  purchased  from  an  American  refiner.  No.  95 
is  metal  from  the  first  precipitation  of  ammonium  chloroplatinate. 
No.  96  is  from  the  second  and  No.  105  is  from  the  third  and  last 
precipitation.  All  were  melted  in  pure  lime  in  the  induction 
furnace,  as  will  be  described  in  a  subsequent  part  of  the  paper. 

The  values  for  e.  m,  f.  are  those  found  against  our  standard  at 
1,200°  C.  The  values  are  given  in  microvolts  and  all  are  positive. 
The  figures  given  in  the  spectrographic  analysis  are  intensities 
estimated  relative  to  those  of  platinum  lines,  the  faintest  of  which 
are  designated  as  1  and  the  strongest  as  10.  These  values  are  not 
interpreted  in  absolute  proportions  present,  except  that  the  "trace" 
of  rhodium  in  No.  96  is  estimated  as  less  than  0.001  per  cent. 


Table  I. 

Results  Obtained  in  Purifying  Platinum. 

Thermo clcciric   Comparison. 


No.  94 

No.  95 

No.  96 

No.    105 

e.  m.  f 

442 

57 

24 

8 

Spectrographic  Analysis.     Figures  indicate  estimated  relative  intensities. 

Palladium 
Rhodium    . 
Copper    . . . 
Iridium  . . . 
Ruthenium 

Iron     

Tin    

Lead   

Calcium    . . 


3 

2 

1  — 

0 

1 

1  — 

trace 

0 

1 

trace 

0 

0 

0 

0 

0 

0 

0 

0 

0 

0 

0 

0 

0 

0 

0 

0 

0 

0 

0 

0 

0 

0 

trace 

trace 

trace 

trace 

It  is  interesting  to  note  that  the  original  source  of  this  platinum 
was  probably  platiniferous  copper  or  nickel  ore,  rather  than 
alluvial  platinum  deposits.  This  is  indicated  by  the  predominance 
of  palladium  among  the  impurities  and  the  absence  of  iridium. 

2.     Analytical  Methods. 
With  a  quantity  of  each  of  the  pure  metals  (except  ruthenium 
as  noted)  at  hand,  it  became  possible  to  undertake  some  of  the 


392  EDWARD    WICKERS    AND   LOUIS   JORDAN. 

Studies  contemplated  when  the  work  was  begun.  A  critical 
investigation  of  analytical  separations  and  methods  of  determina- 
tion seemed  to  be  of  prime  importance,  both  as  a  means  of  pro- 
viding adequate  control  of  the  composition  of  alloys  used  in  other 
phases  of  the  general  research,  and  because  of  the  lack  of 
accepted  standard  methods  for  the  analysis  of  various  articles  of 
commerce  containing  one  or  more  of  the  platinum  metals.  The 
evaluation  of  crude  platinum  metals,  ore  concentrates,  catalytic 
masses,  manufacturers'  scrap  and  sweeps,  and  the  control  of  com- 
position of  alloys  for  electrical  work,  jewelry  and  dental  work, 
are  matters  of  every-day  necessity  in  the  platinum  industry. 

The  Bureau's  study  of  analytical  methods  is  at  present  directed 
mainly  toward  two  problems.  The  first  is  the  accurate  determina- 
tion of  iridium  in  platinum  alloys,  and  the  second  an  accurate 
and  reasonably  rapid  method  for  the  partial  or  complete  analysis 
of  crude  platinum  concentrates  or  native  grain  platinum.  The 
work  on  the  latter  has  been  begun  only  recently,  and  may  be  said 
to  show  promise.  The  principal  novel  feature  of  the  method  is 
the  avoidance  of  the  separation  of  the  concentrates  into  two 
fractions,  respectively  soluble  and  insoluble  in  aq^ta  regia.  If 
this  is  successful  it  will  permit  of  the  determination  of  total 
iridiuiTu  rather  than  the  portion  which  is  insoluble  in  aqua  regia. 
The  high  relative  cost  of  iridium  makes  the  proper  evaluation  of 
the  native  platinum  important. 

The  work  on  the  determination  of  iridium  in  platinum  alloys 
will  be  published  shortly.  No  new  method  is  proposed,  but  the 
old  method  of  Deville  and  Stas  is  brought  up  to  date,  and  made 
to  conform  to  modern  laboratory  methods.  The  procedure  con- 
sists of  fusing  the  alloy  with  ten  or  more  parts  of  lead,  and 
parting  the  resulting  lead  ingot  first  with  nitric  acid  and  then 
with  dilute  aqua  regia.  The  factors  of  temperature  of  the  lead 
fusion,  time  of  fusion,  proportion  of  lead,  and  concentration  of 
aqua  regia  used  in  the  second  parting,  as  well  as  the  influence 
of  the  presence  of  iron,  ruthenium,  and  rhodium  have  been  care- 
fully studied.  The  method  was  found  to  be  capable  of  giving 
results  of  great  accuracy,  except  that  slightly  low  results  for  the 
iridium  content  are  obtained  in  alloys  containing  about  15  to  20 
per  cent,  corresponding  to  contact  point  metal.  This  error  can 
be  corrected  by  a  second  separation  of  iridium.     It  is  proposed 


INVKSTIGATIONS  ON   PLATINUM   METALS.  393 

to  submit  this  method  to  commercial  laboratories  for  comment 
before  it  is  published. 

3.     Technique  of  Melting  and  Working. 

(a)  Refractories.  The  first  method  employed  for  melting  the 
pure  platinum  sponge  was  the  usual  one  of  fusion  on  lime  in  an 
oxyhydrogen  blast  flame.  Calcium  was  always  detected  in  metal 
melted  in  this  manner.  Contamination  by  calcium  was  serious 
■  whenever  the  blast  was  allowed  to  become  deficient  in  oxygen 
while  the  metal  was  molten.^"  Platinum  melted  in  lime  in  the 
Ajax-Northrup  high-frequency  induction  furnace,  with  free 
access  of  air  and  without  excessive  superheating,  was  of  satis- 
factory purity  as  determined  by  the  thermoelectric  tests. 
although  spectrographic  evidence  of  calcium  usually  was  found. 

Small  quantities  of  platinum  melted  on  pure  magnesia  in  the 
oxy-hydrogen  flame  with  an  excess  of  oxygen  were  of  high  purity. 
Melts  of  platinum  in  magnesia  in  the  induction  furnace  were 
seriously  contaminated  with  magnesium  when  a  graphite  or 
tungsten  shell  was  used  outside  the  refractory  liner.  The  reduc- 
tion of  the  refractory  may  have  been  caused  by  the  carbon  or 
carbon  monoxide  or  by  the  metallic  tungsten.  However,  a  con- 
sideration of  the  qualities  required  in  a  refractory  for  general  use 
in  melting  the  metals  of  the  platinum  group  and  their  alloys,  led 
to  the  belief  that  thorium  or  zirconium  oxide  should  be  more 
satisfactory  than  lime  or  magnesia,  and  tests  with  magnesia  in 
the  induction  furnace  were  discontinued. 

Thorium  oxide  has  an  exceedingly  high  heat  of  formation,  a 
high  fusion  point,  low  thermal  conductivity,  and  can  readily  be 
made  into  refractory  shapes  of  good  mechanical  strength  without 
the  use  of  any  additional  substance  as  a  binder.  It  therefore  gave 
promise  of  being  suitable  for  crucibles  for  melting  all  of  the 
platinum  metals,  and  of  being  little  liable  to  dissociation  at  high 
temperatures  and  reduced  pressures.  These  latter  character- 
istics are  required,  because  it  is  desirable  to  carry  out  the  fusi.on 
of  certain  platinum  metals  in  a  vacuum,  as  in  the  case  of  palladium 
and  rhodium.  Thoria  has  been  used  as  the  refractory  in  all  melts, 
both  in  air  and  in  vacuum,  when  the  highest  purity  was  required. 

Xtvy  pure  zirconium   oxide   is   difficult  to  obtain.     The  best 

'•  See  footnote   7. 

26 


394  EDWARD   WICKERS   AND   LOUIS   JORDAN. 

oxide  available,  prepared  by  ignition  of  Kahlbaum's  zirconium 
nitrate,  was  used  as  the  refractory  in  a  few  melts  of  platinum. 
The  quality  of  the  resulting  metal  as  judged  by  the  thermoelectric 
test  was  not  as  good  as  that  of  metal  fused  on  lime  or  thoria. 

(b)  Furnace.  The  high-frequency  induction  furnace  proved 
to  be  a  convenient  and  satisfactory  means  of  melting  platinum 
metals  with  the  minimum  contamination.  The  range  of  tempera- 
tures available  is  sufficient  for  melting  the  most  refractory  of  the 
metals.  Temperature  control  and  the  control  of  the  atmosphere 
over  the  molten  metal  or  melting  under  vacuum  are  easily  accom- 
plished. '  Homogeneous  alloys  are  readily  made ;  small  melts  can 
be  cast  in  chill  molds ;  the  location  of  shrinkage  cavities  can  be 
controlled ;  and  a  very  accurate  synthesis  of  alloys  is  possible. 

(c)  Working  of  Platinum  Metals.  The  mechanical  working 
of  the  pure  metals  and  alloys  presents  no  particular  difficulty 
in  so  far  as  contamination  during  rolling  and  drawing  are  con- 
cerned. Careful  attention  to  the  condition  of  the  surface  of  steel 
rolls,  and  the  use  of  jewel  (sapphire  and  diamond)  dies  for  wire 
drawing,  allow  satisfactory  working.  ^^lore  complete  details 
of  the  technique  of  melting  and  working  as  involved  in  the 
preparation  of  the  Bureau's  standard  thermocouples  are  given 
in  a  paper  by  R.  P.  Neville  to  be  presented  at  this  meeting. 

4.    Quality  of  Platinum  Laboratory  Ware. 

The  results  of  the  first  three  major  phases  of  the  general  in- 
vestigation of  the  platinum  metals,  namely  the  purification  of 
metals,  methods  of  analysis,  and  melting  and  working  technique, 
are  rapidly  becoming  available.  It  is  thus  possible  to  undertake 
the  study  of  platinum  alloys  with  reference  to  their  suitability 
for  platinum  laboratory  ware. 

In  the  Bureau's  experience,  based  on  the  small  portion  of  its 
platinum  laboratory  ware  purchased  from  commercial  sources 
and  on  the  few  samples  of  ware  submitted  from  other  labora- 
tories for  tests  of  quality,  some  of  the  difficulties  mentioned  in 
the  platinum  committee's  first  report  are  no  longer  so  frequently 
encountered.  The  most  serious  point  of  failure  in  platinum  ware 
at  present  seems  to  be  the  tendency  to  develop  cracks  after  con- 


INXTiSTlGATIONS  ON   PLATINUM   METALS.  395 

tinned  heating.     The  causes  of  this  failure  and  methods  for  its 
prevention  are  apparently  unknown. 

The  present  plan  for  this  portion  of  the  general  investigation 
is  to  determine  mechanical  and  certain  other  physical  properties 
of  test  specimens  and  crucibles  made  from  selected  platinum 
metal  alloys,  to  make  accelerated  service  tests,  and  to  attempt 
to  correlate  the  two  series  of  data. 

5,     Physical  Properties  of  Platinum  Metals. 

The  preparation  of  the  pure  platinum  metals  and  their  alloys 
for  the  purposes  of  the  several  phases  of  the  general  investigation 
thus  far  outlined,  provides  an  opportunity  for  the  measurement 
of  a  variety  of  physical  properties  of  such  materials.  The  plat- 
inum metals  have  properties  of  unusual  interest,  and  are  in  con- 
stant use  in  the  prosecution  of  scientific  investigations.  Some 
of  the  data  already  reported  in  the  literature  are  contradictory, 
and  many  more  are  doubtless  incorrect  or  very  inaccurate,  both 
because  of  faulty  measurements  and  because  the  degree  of  purity 
of  the  materials  studied  has  frequently  been  ignored. 

In  so  far  as  the  nature  of  the  samples  available  from  the 
preceding  phases  of  this  work  and  the  facilities  of  the  Bureau 
of  Standards  will  permit,  the  more  important  electrical,  thermal, 
optical,  mechanical  and  various  other  miscellaneous  physical 
properties  of  the  pure  metals  and  selected  alloys  of  the  platinum 
group  will  be  determined. 

It  is  believed  that  the  results  obtained  from  the  investigations 
outlined  will  be  of  interest  and  value  to  all  users  and  manu- 
facturers of  platinum.  Manufacturers  of  platinum  have  already 
expressed  their  interest  in  this  investigation.  Detailed  reports  on 
the  various  phases  of  the  research  will  be  made  as  rapidly  as 
the  progress  .of  the  work  warrants.  The  Bureau  will  welcome 
correspondence  or  conference  with  both  manufacturers  and  users 
of  platinum,  with  a  view  to  establishing  closer  contact  with  the 
outstanding  problems  of  the  industry. 


DISCUSSION. 
F.  E.  Carter^  :    I  have  just  two  remarks  to  make.     I  do  not 
altogether  agree  with  the  authors  that  the  thermoelectric  test  is 

'  Metallurgist,  Baker  &  Co.,  Inc.,  Newark.  N.  J. 


396  DISCUSSION. 

better  than  the  coetiicient  of  electrical  resistance.  I  think  that 
the  figures  given  in  the  same  paragraph  show  that  you  readily 
distinguish  between  the  purity  of  platinum  by  the  latter  test.  I 
would  be  interested  to  learn  whether  the  Bureau  is  working  out 
electrolytic  methods  of  separation  of  the  platinum  metals,  for 
I  think  the  future  lies  along  that  line.  Getting  only  partial 
separation  and  having  to  repeat  the  precipitation  is  a  great 
nuisance  in  chemical  separations.  As  an  example  of  what  I 
mean,  Dr.  Jordan  states  that  the  method  for  determining  iridium 
in  platinum  was  found  capable  of  giving  results  of  great  accuracy 
except  that  slightly  low  results  for  the  iridium  content  are  ob- 
tained in  alloys  containing  about  15  to  20  per  cent,  where  actually 
greatest  accuracy  is  required. 

The  manufacturer  does  not  take  into  account  whether  there  is 
half  per  cent  or  one-third  per  cent  of  iridium  in  platinum ;  in 
such  cases  he  pays  for  platinum  only.  But  if  appreciable  quan- 
tities are  present  it  is  different;  he  pays  for  the  iridium,  the 
amount  of  which  he  must  know  accurately,  because  the  difference 
in  price  between  iridium  and  platinum  is  great. 

I  quite  agree  with  the  statement  that  in  platinum  ware  the 
tendency  to  develop  cracks  upon  continued  heating  is  serious. 
This  cracking  appears  to  take  place  most  frequently  when  deter- 
mining, say,  volatile  matter  in  coal  or  in  other  materials.  I  think 
it  has  something  to  do  with  the  formation  of  carbides. 


A  paper  presented  at  the  Forty-third 
General  Meeting  of  the  American  Elec- 
trochemical Society  held  in  New  York 
City,  May  5,  1923,  Dr.  F.  M.  Becket  in 
the  Chair. 


SOME  NOTES  ON  THE  METALS  OF  THE  PLATINUM  GROUP.' 

By  Fred  E.  Carter.^ 

Abstract. 
Some  general  remarks  on  metals  of  the  group  are  given,  par- 
ticular mention  being  made  of  the  liability  to  gas  absorption  and 
of  the  consequent  difficulties  of  melting.  Results  are  given  to 
show^  that  the  addition  of  iridium  to  platinum  raises  considerably 
the  temperature  required  for  annealing.  Some  alloys  of  the 
platinum  metals  among  themselves  are  discussed,  particularly 
those  of  platinum  and  iridium. 


When  the  comparative  rarity  of  the  platinum  metals  is  taken 
into  consideration,  it  is  quite  remarkable  that  so  much  attention 
has  been  given  to  them ;  this  statement  applies,  however,  more 
accurately  to  the  chemical  rather  than  to  the  physical  side,  because, 
although  large  numbers  of  complex  salts,  etc.,  of  the  platinum 
metals  have  been  prepared  and  investigated,  the  physical  prop- 
erties of  the  metals  of  the  group  and  their  alloys  have  not  been 
nearly  so  exhaustively  examined.  The  average  analyst  has  felt 
that  there  is  something  uncanny  about  the  platinum  metals ;  he  has 
found  that  it  is  practically  impossible  to  get  complete  quantitative 
separation  by  a  simple  laboratory  operation  and  that  to  make 
such  separations  complete  precipitations,  etc.,  must  be  carried  out 
many  times  in  succession.  He  ordinarily  considers,  for  example, 
platinum  ammonium  chloride  as  an  insoluble  salt,  particularly  in 
the  presence  of  alcohol,  yet  actually  there  are  many  salts  which 
may  be  present  in  the  solution,  and  in  which  the  precipitated  salt 
may  be  appreciably   soluble.     The  author  prefers   therefore   to 

^  Manuscript  received  February  9,    1923. 
=  Baker  and  Co.,  Inc.,  Newark,  X.  J. 

397 


398  FRED  E.  CARTER. 

avoid  in  this  paper  the  chemistry  of  the  platinum  metals,  and  to 
write  down  some  facts  on  the  subject  from  the  physical  and 
metallurgical  standpoint. 

It  is  necessary,  first  of  all,  to  emphasize  the  fact  that  much 
that  has  been  published  on  the  physical  properties  of  the  platinum 
metals  is  quite  erroneous,  because  the  metals  used  in  the  tests 
have  been  by  no  means  pure,  although  this  difficulty  is  not  at  all 
peculiar  to  the  group  in  question,  but  applies  generally  to  data 
on  the  metallic  elements  and  their  alloys.  The  platinum  metals 
further  add  to  our  troubles  by  their  susceptibility  to  gases.  It 
is  not  intended  in  this  paper  to  present  precise  measurements  of 
physical  standards,  but  rather  to  give,  as  is  indicated  by  the 
title,  some  random  notes  on  the  platinum  metals  and  to  point  out 
certain  facts  which  may  be  of  general  interest. 

The  platinum  group  of  metals  is  composed  of  platinum,  iridium, 
osmium,  palladium,  rhodium,  and  ruthenium.  They  occur  prac- 
tically always  in  the  elemental  state,  so  that  their  metallurgy  is 
comparatively  simple  and  need  not  be  discussed  here.  AH  the 
platinum  metals  are  white  in  color;  most  writers  differentiate 
between  their  appearance,  platinum  being  called  tin-white, 
rhodium  aluminum-white,  osmium  bluish,  etc.,  but  actually  it  is 
difficult  to  distinguish  the  group  members  by  appearance  only. 
The  author  would  hesitate  to  say  that  there  is  any  difference  in 
color  between  platinum,  palladium,  and  rhodium ;  iridium  might 
be  conceded  a  more  brilliant  white  appearance  and  ruthenium  is 
whiter  still ;  osmium  certainly  has  a  bluish  tinge.  Such  differences 
as  are  noted  here  might  easily  be  due  to  surface  oxide  films  or  to 
variable  crystal  grain  sizes,  rather  than  to  inherent  different 
shades  of  color. 

The  metals  do  not  oxidize  at  ordinary  temperatures,  but  on 
heating,  certain  of  the  group  oxidize  and  volatilize.  Table  I, 
based  on  the  contradictory  literature  on  the  subject,  shows  what 
probably  happens  when  a  hypothetical  mixture  of  all  the  metals 
is  gradually  heated  up. 

One  of  the  most  interesting  general  properties  of  the  platinum 
group  of  metals  is  their  capacity  for  dissolving  gases,  for  therein 
probably  lies  the  reason  for  their  great  activity  as  catalytic  bodies. 
This  property  is  extremely  disagreeable  to  the  manufacturer ;  if 
the  composition  of  the  gas  used  for  melting,  say,  platinum,  is 


THE    METALS  OF  THE   PLATINUM   GROUP.  399 

incorrect,  gas  may  be  dissolved  by  the  molten  metal  and,  since 
the  solid  metal  is  a  much  poorer  solvent  than  the  molten,  set  free 
on  solidification,  the  metal  then  "spits"  in  the  same  manner  as 
does  solidifying  silver  which  has  been  saturated  with  oxygen  in 
the  molten  condition.  Generally,  however,  the  gas  has  not  the 
opportunity  to  escape  in  this  way,  and  is  partly  entrapped  in  the 

Table  I. 

Results  of  Heating  the  Platinum  Group  Metals. 

100°  C.     OsOi  begins  to  evolve   (if  the  metal  is  very  finely  divided  the 

vapor    is   observed    at   considerably    lower    temperature,    but 

compact  metal  does  not  oxidize  appreciably  below  a  dull  red 

heat). 

450°  C.     Pt  oxidizes  to  black  PtO  (if  metal  is  finely  divided  and  oxygen 

is  passed). 
500°  C.     PtO  decomposes  to  give  Pt  and  Pt02. 
550°  C.     Pt02  decomposes  to  Pt  and  0=. 

600°  C.     Rh    and    Ru,    if    finely    divided,    oxidize    to    black    Rh^O.    and 
bluish  RuOj. 
Pd  begins  to  oxidize  to  PdO,  giving  blue  and  red  colors. 
700°  C.     Pd  oxidizes  to  PdO. 
800°  C.     Ir  begins  to  oxidize  to  IrOj. 
900°  C.     PdO  is  decomposed  to  Pd  and  O. 
1,000°  C.     Ir  begins  to  volatilize  freely  as  oxide. 

RuOj  partially  decomposes  to  Ru  and  O2.    If  oxygen  is  passed, 

some  RuO*  is  formed  and  evolved. 
Pd,    Pt   and   Rh   begin    to   volatilize    appreciably   in    the    order 
named,  as  metals. 
1  150°  C.     Rh.O.  decomposes  to  Rh  and  O. 
1,550°  C.     Pd  melts. 
1,755°  C.    Pt  melts. 
1,950°  C.     Rh  melts. 
2,350°  C.     Ir  melts. 
2.450°  C.     Ru  melts. 
2.500°  C.     Rh  boils. 
2,520°  C.     Ru  boils. 
2.540°  C.     Pd  boils. 
2.550°  C.     Ir  boils. 
2,700°  C.     Os  meks. 
3.910°  C.    Pt  boils. 

The  temperatures  given  from  2350°  C.  upwards  are  questionable. 

bar  as  small  gas  inclusions.  These  often  do  not  appear  until  the 
bar  is  rolled  down  to  thin  sheet  and  annealed,  when  the  surface 
is  found  covered  with  numerous  gas  blisters.  A  bar  that  is  much 
gassed  swells  badly  on  solidifying  and  is  remelted  forthwith,  but 
the  manufacturer's  trouble  chiefly  comes  when  not  so  much  gas  is 
trapped  that  swelling  occurs  and  only  shows  up  in  the  finished  bar 


400  FRED  E.  CARTER. 

as  a  few  blisters  scattered  throughout  the  bar.  It  is  hoped  that 
x-ray  examination  may  eventually  be  useful  in  showing  gas 
bubbles  in  the  interior  of  ingots,  but  the  usefulness  of  this  method 
of  examination  has  not  yet  been  proved  in  the  case  of  platinum. 

Coal  gas  and  oxygen  or  hydrogen  and  oxygen  are  generally 
used  for  melting  the  metals,  and  it  is  obvious  that  great  care  must 
be  taken  to  have  the  correct  proportions  of  gas  or  hydrogen  and 
oxygen.  Platinum  must  be  melted  in  a  distinctly  oxidizing 
atmosphere,  otherwise  the  blister  trouble  will  appear;  palladium, 
if  melted  by  a  reducing  flame,  is  absolutely  friable,  the  well- 
formed  crystal  grains  being  apparently  without  cohesion  and  easily 
separable  by  the  fingers ;  rhodium  blisters  if  melted  under  oxidiz- 
ing conditions ;  iridium  behaves  like  platinum ;  osmium  and 
ruthenium  rapidly  volatilize  in  an  oxidizing  flame.  It  will  be 
evident  that  difficulties  arise  when,  say,  alloys  of  platinum, 
iridium,  and  rhodium,  or  of  platinum,  palladium,  and  osmium 
have  to  be  melted;  in  such  cases  experience  has  pointed  out  the 
special  precautions  that  must  be  taken. 

The  obvious  method  of  overcoming  such  difficulties  is  to  melt 
in  vacuo  and  by  electricity,  but  even  then  the  troubles  are  not  yet 
avoided,  because  refractories  which  are  suitable  for  oxy-hydrogen 
gas  melting  may  react  with  the  molten  metals  under  such  condi- 
tions. Wichers^  has  shown  that  platinum-calcium  alloys  are 
formed  if  platinum  is  melted  in  a  lime  crucible  with  a  reducing 
atmosphere  existing  in  the  crucible.  If  such  alloy  formation  is 
to  be  avoided  excess  oxygen  in  the  oxy-hydrogen  flame  must  be 
used.  Also  it  was  shown  that  if  platinum  is  melted  electrically  in 
a  magnesia  crucible  an  alloy  of  platinum  containing  about  three 
per  cent  magnesium  may  be  produced.  In  parenthesis,  it  may  be 
observed  that  such  results  force  us  to  the  conclusion  that  our 
ideas  of  stability  of  compounds  must  be  modified  when  we  get 
into  the  higher  ranges  of  temperature. 

Of  course,  graphite  would  be  the  ideal  crucible  material  to  use 
for  melting  in  vacuo,  but  unfortunately  the  platinum  metals  are 
readily  attacked  by  carbon.  It  is  not  even  necessary  actually  to 
melt  the  metal  in  carbon  vapor  for  the  platinum  to  be  rendered 
quite  dark  in  appearance,  to  be  strongly  modified  in  its  micro- 
scopic structure,  and  to  be  made  absolutely  brittle.    One  way  of 

»  j.  Am.  Chem.   Soc.  43,  1268  (1921), 


THU    METALS  OF  THE  PLATINUM   GROUP.  401 

avoiding  these  troubles  is  to  use  the  old  "French  method,"  which 
consists  in  pressing  Pt  sponge  into  a  briquet  and  heating  to  about 
1,000°  C.  In  this  way  the  grey  platinum  mass  is  gradually 
"metallized,"  and  can  then  be  worked  down  to  thin  sheets  in  the 
usual  way.  The  finer  the  state  of  division  of  the  original  plat- 
inum, the  more  readily  does  this  metallizing  take  place.  The 
process,  however,  lacks  one  advantage  of  the  ordinary  melting 
process,  namely,  the  refining  effect  {i.  e.,  removing  the  base  metal) 
of  the  lime  on  the  molten  metal. 

Some  indication  has  been  given  above  of  how  the  loss  of  the 
platinum  metals  by  volatilization  takes  place,  which  loss  is  of 
course  important  in  crucible  ware ;  it  is  necessary  to  decide  what 
is  the  alloy  which  will  lose  least  weight  when  heated  to  1,000°  or 
1,200°  C.  for  several  hours.  Platinum-iridium  alloys,  high  in 
iridium,  lose  in  weight  considerably  at  these  temperatures,  owing 
to  volatilization  of  the  iridium  and  must  be  avoided;  platinum- 
rhodium  is  practically  eliminated  by  the  high  cost  of  the  rhodium, 
chemically  pure  platinum  is  good  so  far  as  constancy  in  weight 
is  concerned,  but  is  rather  soft.  The  Bureau  of  Standards*  has 
therefore  recommended  as  a  compromise  that  a  small  amount  of 
platinum  metals  (chiefly  iridium)  other  than  platinum  may  be 
present,  suggesting  that  the  alloy  used  should  not  show  more 
than  1  m.  v.  against  chemically  pure  platinum  at  1,100°  C. ;  this 
corresponds  to  about  0.3  per  cent  iridium.  Crucibles,  etc.,  made 
from  such  material  are  constant  in  weight  and  reasonably  stiff. 

The  metals  of  the  group  alloy  with  one  another  in  all  propor- 
tions, the  alloys  being  solid  'solutions,  as  is  usual  in  the  case  of 
combinations  amongst  the  closely  related  elements  of  a  group. 
There  does  not  appear  to  be  any  case  where  the  meUing  point  of 
the  alloy  is  lower  than  either  of  the  constituents  (as,  for  example, 
occurs  with  gold  and  copper),  but  always  the  melting  points  of 
the  alloys  are  intermediate  between  those  of  the  constituent 
metals.  Micro-photographs  show  that  the  addition  of  a  second 
platinum  metal  to  platinum  itself  causes  a  distinct  refining  of  the 
crystal  structure.  For  example,  the  crystal  grains  of  a  series  of 
annealed  iridio-platinum  alloys  show  with  increasing  iridium  a 
decreasing  size  of  grain.     Metallurgical  examination  also  shows 

*  Bureau  of  Standards,  Sci.  Paper  254. 


402 


FRED  E.  CARTER. 


that  the  alloys  are  homogeneous  and,  after  adequate  annealing, 
are  practically  free  from  any  "coring"  in  the  crystal  grains. 

The  temperature  required  to  render  platinum  dead  soft  is 
comparatively  low,  but  this  temperature  is  considerably  raised 
by  the  addition  of  even  a  small  percentage  of  iridium.  Fig.  1 
shows  the  large  effect  of  traces  of  iridium  on  the  annealing  point 


120 

( 

■* 

►—a 

" 

-^ 

^ 

lOO 
90 

\N" 

^ 

\ 

\ 

\ 

\ 

•\ 

80 

1 

\ 

1 

\  , 

70 

\ 

\ 

^ 

CRUC. 

50 

S) 

^ 

^*~~^ 

k— Q— ^ 

;^ — A — 1 

40 
30 
20 

C.P. 

lO 

. 

zoo         300        400  30O        •OO  700  800         900        lOOO 

TEMPERATURE.  PESREES    CEMTIGR.ADE 
Fig.   1. 


of  platinum.  The  curve  marked  "C.  P."  is  for  platinum  of  a  high 
degree  of  purity  (temperature  coefficient  of  resistance,  0.00391), 
and  that  marked  "Cruc."  is  for  platinum  containing  0.1  per  cent 
iridium  (0.48  m.  v.  against  C.  P.  Pt),  such  as  is  used  for  crucible 
ware,  etc.  The  furnace,  a  platinum-wound  electric  tube  furnace 
equipped  with  platinum-rhodium  thermocouple,  was  slowly 
brought  up  to  temperature,  and  the  samples  of  metal,  1.9  x  1.3  x 
0.3  cm.  (^  X  ^  X  ^8  i»-)  ^vere  introduced  and  kept  in  the  fur- 


THE   METALS  OF  THE  PLATINUM   GROUP.  403 

nace  for  5  min.  after  they  had  reached  the  furnace  temperature ; 
the  hardness  was  tested  in  a  Brinnell  machine. 

It  is  necessary  to  make  some  further  reference  to  Fig.  1.  Rose' 
showed  that  traces  of  impurity  raise  the  annealing  temperature 
of  gold  appreciably ;  hydrogen  was  found  to  be  especially  effec- 
tive in  this  respect,  0.002  per  cent  raising  the  temperature  of 
annealing  from  150°  C.  to  over  300°  C.  Phelps*'  confirmed  Rose's 
results.  It  was  believed  by  the  present  author  that  a  similar  effect 
had  been  shown  for  platinum,  although  the  impurity  is  not  neces- 
sarily hydrogen.  However,  another  factor  may  have  been  influ- 
ential in  the  results  here  obtained.  It  is  well  known  that  an 
increase  in  the  amount  of  cold  work  done  on  a  metal  previous  to 
annealing  causes  a  decrease  in  the  temperature  required  to  anneal, 
and  it  was  thought  possible  that  the  two  samples  of  platinum  in 
Fig.  1  were  not  in  exactly  the  same  strained  condition,  in  spite 
of  the  fact  that  both  had  been  cold  rolled  from  }i  in.  to  ^  in. 
Another  sample  of  platinum,  not  quite  so  pure  (it  gave  0.05  m.  v. 
positive  to  the  platinum  previously  used),  was  rolled  in  the  same 
way,  and  tested,  and  it  was  found  that  the  temperature  required 
for  annealing  was  practically  as  high  as  that  necessary  for  the 
"crucible"  platinum  of  the  figure.  The  same  sample  was  then  re- 
melted  and  cold  rolled  from  ^i  in.  to  j4.  in.  instead  of  from  ^/^  in. 
to  %  in.  The  annealing  temperature  now  was  even  lower  than 
that  shown  in  the  figure.  There  seems  to  be  little  doubt,  there- 
fore, that  in  the  case  of  almost  pure  platinum  the  previous  history 
of  the  sample  has  more  eflfect  on  the  annealing  temperature  than 
has  the  purity. 

It  will  be  seen  from  the  figure  that  for  pure  platinum  the 
required  temperature  is  about  650°  C,  while  for  platinum  with 
only  0.1  per  cent  iridium  the  temperature  is  about  1,000°  C. 
Further  additions  of  iridium  do  not  raise  the  annealing  point 
much.  For  example,  platinum  with  ten  per  cent  iridium  requires 
1,150°  C.  to  become  fully  annealed  in  five  minutes;  complete 
crystallization  of  the  alloys  containing  20  and  25  per  cent  iridium 
may  be  brought  about  at  this  same  temperature,  but  require  a 
considerably  longer  time. 

It  is  unnecessary  here  to  do  more  than  to  draw  attention  to  the 

"J.  Inst,  of  Metals,  10,  ISO  (1913). 
"J.  Inst,  of  Metals,   12.  125    (1914). 


404  FRED  E.  CARTER. 

slight  increase  in  hardness  of  platinum  on  annealing  at  about 
300°  C.  This  phenomenon  of  a  slight  hardening  at  tempera- 
tures just  below  that  at  which  softening  begins  seems  to  be  a 
general  one  in  commercially  pure  metals  and  in  alloys.  It  cer- 
tainly is  quite  pronounced  in  many  gold  alloys  for  which  the 
author  has  drawn  curves  similar  to  Fig.  1.  It  is  interesting,  in 
view  of  some  theories  which  have  been  advanced  in  explanation, 
to  find  this  phenomenon  occurring  in  the  case  of  a  metal  of  such 
extreme  purity  as  the  platinum  used  here. 

PHYSICAL    CHARACTERISTICS    OF    EACH     METAL. 

Platinum.  Electrical  resistance  at  0°  C.  is  60.5  ohms  per  mil 
foot  (10.06  microhms  per  cm.  cube)  for  hard  drawn  platinum, 
and  59.8  ohms  per  mil  foot  (9.96  microhms  per  cm.  cube)  for 
the  annealed  material.  The  temperature  coefftcient  of  resistance 
is  0.00392  or  even  slightly  higher  for  the  extremely  pure  metal.^ 

The  melting  point  is  1,755°  C.,^  apparently  being  the  same  for 
metal  melted  in  air  or  in  vacuo.  The  melting  point  is  depressed 
by  the  presence  of  traces  of  carbon  in  the  metal. 

The  Brinnell  hardness  is  about  110  in  the  hard  worked  and  47 
in  the  annealed  condition.  The  Erichsen  number  for  ductility 
of  the  annealed  sheet  0.040  in.  thick  is  12,2  mm. 

Pt  wire  is  drawn  down  commercially  directly  to  0.0007  in., 
while  if  drawn  by  the  Wollaston  method  (that  is,  a  platinum  core 
and  a  covering  tube  of  a  metal,  e.  g.,  silver,  which  can  be  dissolved 
off  later  without  attacking  the  platinum,  are  drawn  down  together) 
the  diameter  may  be  made  one-tenth  or  even  one-hundredth  of 
this  size. 

Iridium.  The  melting  point  is  about  2,350°  C.®  and  possibly 
higher.  This  metal  is  little  used  except  in  alloy  form.  It  is  gen- 
erally stated  to  be  quite  a  hard  metal,  but  actually  such  statements 
are  made  from  tests  with  a  very  impure  material.  The  chemically 
pure  metal  is  fairly  soft — about  the  same  as  90  Pt  10  Ir.  Brinnell 
hardness,  172  (cast).     Iridium  is  insoluble  in  aqua  regia. 

Osmium.  The  melting  point  is  about  2,700°  C,  but  this  figure 
must   be   considered   as   only    an   approximation.      It   volatilizes 

"  Bureau  of   Standards. 

'  Bureau   oi   Standards,  Circular  35. 

•  Loc.  fir. 


THE    METALS  OF   THE  PLATINUM    GROUP.  405 

rapidly  as  osmium  tetroxide  if  heated  in  air,  and  the  melting 
should  be  done  in  vacuo;  even  in  vacuo  osmium  on  heating  close 
to  its  melting  point  volatilizes  in  the  form  of  a  brown  vapor. 
Osmium  is  insoluble  in  aqtia  regia. 

Palladium.  The  melting  point  is  1,550°C.  ;^  as  ordinarily 
melted  the  metal  retains  considerable  quantities  of  gas,  as  is 
shown  by  the  fact  that  if  it  is  remelted  in  vacuo  there  is  a  violent 
evolution  of  gas  just  at  the  melting  point.  The  metal  forms 
different  oxides  which  are  stable  only  within  certain  narrow  limits 
of  temperature.  If  an  ingot  of  palladium  is  allowed  to  cool  slowly 
it  becomes  coated  with  thin  oxide  films  of  red,  green  and  blue. 
If  it  is  desired  to  have  a  bright  finish  to  a  bar,  it  is  only  necessary 
to  quench  it,  red-hot,  in  water.  Brinnell  hardness,  49  (Cast). 
Palladium  is  soluble  in  concentrated  nitric  acid  and  in  aqua  regia. 

Rhodium.  This  metal  has  been  obtained  in  a  high  state  of 
purity,  since  it  is  used  for  making  the  10  per  cent  alloy  with 
platinum,  as  the  positive  element  in  precious  metal  thermo- 
couples; if  the  metal  is  even  slightly  impure  the  curve  for  the 
electromotive  force  against  platinum  at  once  shows  discrepancies. 

The  melting  point  is  1,950°  C,"  if  melted  in  air  the  metal  is 
coated  with  a  blue  oxide  film,  but  in  vacuo  the  metal  is  perfectly 
white.  Brinnell  hardness,  139  (Cast).  Rhodium  is  insoluble  in 
aqua  regia. 

Rtithetiium.  The  melting  point  is  about  2,450°  C.,*  but  it  is 
not  at  all  certain  that  the  metal  has  ever  been  obtained  in  the  chem- 
ically pure  state.  Melted  in  air  it  is  coated  with  a  blue-black 
oxide;  melted  in  vacuo  it  remains  quite  bright,  although  a  black 
deposit  settles  in  cooler  parts  of  the  apparatus.  Brinnell  hardness, 
220  (Cast),  but  the  pure  metal  would  certainly  be  considerably 
softer  than  this.     Ruthenium  is  insoluble  in  aqua  regia. 

ALLOYS. 

The  metals  of  the  platinum  group  form  many  useful  alloys  with 
other  metals  outside  the  group,  of  which  may  be  cited  palladium- 
gold  alloys  for  laboratory  ware,  etc.,  palladium-silver  for  contacts, 
platinum-copper  alloys  of  remarkably  high  electrical  resistance, 
etc.  Discussion  of  these  would  lead  too  far  afield,  and  in  this 
paper  mention  will  be  made  only  of  the  alloys  formed  among  the 
platinum  metals  themselves. 


406  FRED  E.  CARTER. 

Platinum-iridium  alloys  undoubtedly  constitute  the  most  im- 
portant series.  "Crucible  platinum"  is  platinum  with  a  small 
quantity  of  iridium  in  it  (less  than  0.3  per  cent.)  This  iridium  is 
sufficient  to  stiffen  the  pure  metal  slightly  and  probably  helps  to 
reduce  the  tendency  to  form  large  crystals.  Ordinary  commercial 
platinum  is  by  no  means  pure  platinum ;  it  contains  from  1  to  3 
per  cent,  iridium,  which,  although  double  the  value  of  platinum,  is 
not  worth  while  extracting,  owing  to  the  chemical  difficulties 
involved ;  also  traces  of  all  the  other  platinum  metals  are  present, 
together  with  appreciable  quantities  of  iron.  Alloys  useful  to 
the  jewelry  world  are  platinum  with  5  to  10  per  cent  iridium, 
known  to  the  trade  as  "hard"  platinum ;  here  again  the  iridium 
includes  all  the  other  platinum  metals  in  small  quantity.  C.  P. 
platinum  with  10  per  cent  C.  P.  iridium  would  be  much  softer 
than  the  ordinary  commercial  "10  per  cent."  The  15  and  20  per 
cent  iridium  alloys  are  used  for  electrical  contacts  and  for  hypo- 
dermic needles,  and  indeed  in  many  places  where  a  hard  precious 
metal  alloy  with  reasonably  good  working  properties  is  required. 
The  25  and  30  per  cent  iridium  alloys  are  considerably  harder  and 
are  rather  difficult  to  work  without  special  precautions.  They  are 
chiefly  used  for  hypodermic  needles. 

The  approximate  figures  for  Brinnell  hardness  of  some  typical 
commercial  iridio-platinum  alloys  are  shown  in  Table  II.  These 
alloys  are  widely  used  for  resistance  wires  where  a  precious  metal 
alloy  is  required.  The  approximate  resistances  of  some  of  the 
commercial  alloys  are  given  in  Table  III.  Alloys  made  from  pure 
materials  have  resistances  as  shown  in  Table  IV. 

The  addition  of  iridium  to  platinum  decreases  the  rate  at  which 
the  latter  dissolves  in  aqua  regia;  platinum  with  20  per  cent 
iridium  is  very  slowly  dissolved,  while  the  25  and  30  per  cent 
alloys  are  practically  unattacked. 

Platinum-rhodium.  The  only  important  alloy  of  these  metals  is 
that  containing  10  per  cent  rhodium,  used  at  the  present  time  for 
the  positive  element  of  the  well-known  Pt-PtRh  thermocouple ; 
although  the  electromotive  force  developed  by  this  couple  is  only 
about  60  per  cent  of  that  given  by  the  corresponding  Pt-Ptir 
couple,  it  is  preferred  on  account  of  the  low  volatility  of  the 
rhodium  compared  with  the  iridium,  and  the  consequent  greater 
constancy  of  e.  m.  f.     A  great  many  industries  require  accurate 


THE   METALS  OF  THE  PLATINUM  GROUP. 

Table  II. 
Hardness  of  Commercial  Iridio-Platinnm  Alloys. 


407 


Composition 

Brinnell   Hardness 

Pt 

Ir 

Hard 

per  cent 

per  cent 

Worked 

Annealed 

95 

5 

170 

110 

90 

10 

220 

150 

85 

15 

280 

190 

80 

20 

330 

230 

75 

25 

370 

270 

70 

30 

400 

310 

Table  III. 
Resistances  of  Commercial  Iridio-Platinnm  Alloys. 


Composition 

Resistance 

Pt                                      Ir 
per   cent                        per    cent 

Microhms 
per    cm.    cube 

Ohms 
per  mil  ft. 

95 
90 
85 
80 
75 

5 
10 

15      • 
20 
25 

20.0 
26.6 
30.8 
3U 
34.9 

120 
160 
185 
200 
210 

Table  IV. 
Resistances  of  Pure  Iridio-Platinnm  Alloys. 


Composition 

Resistance 

Pt 

Ir 

Microhms 

Ohms 

per  cent 

per  cent 

per    cm.    cube 

per  mil   ft. 

99.9 

0.1 

11.0 

660 

99.8 

0.2 

n.3 

67.9 

99.0 

1.0 

12.4 

74.7 

98.0 

2.0 

15.0 

89.9 

96.0 

4.0 

17.3 

104 

94.0 

6.0 

19.5 

117 

4o8  FRED  E.  CARTER. 

temperature  control  at  some  stage  of  manufacture,  and  it  is 
essential  to  have  reliable  thermocouples.  Pt-PtRh  certainly 
remains  the  most  constant  in  e.  m.  f.  and,  with  care  in  manu- 
facture, can  be  made  to  agree  to  the  standard  curve  of  Day  and 
Sosman  to  v^^ithin  a  degree  or  two.  Certain  other  alloys,  with 
the  rhodium  somewhat  above  or  below  10  per  cent  are  used  in 
thermocouples,  but  that  containing  exactly  10  per  cent  seems  the 
most  satisfactory. 

The  10  per  cent  rhodium  alloy  is  much  softer  (Brinnell  number, 
90  when  annealed)  than  the  corresponding  iridium  alloy  and  also 
has  lower  electrical  resistance  (110  ohms  per  mil  ft.;  18.3 
microhms  per  cm.  cube). 

PlafiniDJi-palladiinii  alloys  are  used  to  some  extent  in  jewelry; 
the  addition  of  the  palladium  does  not  harden  the  platinum  much, 
and  the  resulting  alloys  are  readily  workable. 

Platiniim-osmhim  alloys  have  been  made  containing  up  to  30  per 
cent  osmium ;  they  are  extremely  hard,  the  osmium  having  about 
two  and  one-half  times  the  hardening  effect  of  iridium.^"  The 
osmium  also  increases  the  electrical  resistance  of  platinum  about 
two  and  one-half  times  as  much  as  does  the  same  amount  of 
iridium.  They  are  not  used  commercially,  because  annealing  at 
even  a  dull  red  heat  is  sufficient  to  expel  some  of  the  osmium 
and  thus  alter  the  composition  of  the  alloys. 

Indium-osmium  alloys  occur  in  the  natural  state  as  osmiridium ; 
the  grains  are  extremely  hard  and  are  used  as  tips  for  fountain 
pens.  The  alloys  are  now  being  made  artificially  in  any  desired 
proportions  and  by  suitable  treatment  crystal  grains  of  the  proper 
size  for  pens  are  obtained. 

Palladium-osmium  alloys  are  easily  workable,  but  cannot  be 
heated  without  losing  osmium. 

There  are  also  several  ternary  and  quaternary  alloys  finding 
commercial  application  which  may  be  mentioned.  Platinum- 
iridium-os)yiium  alloys  are  used  for  sparking  points ;  platimtm- 
iridium-rhodium  alloys  are  used  for  radio  tubes ;  plafinum-pal- 
ladium-osmium  alloys  were  formerly  used  in  jewelry,  but  the 
partial  volatilization  of  the  osmium  as  tetroxide  was  disagreeable 
and  platinum-palladium-rhodium  alloys  are  now  preferred. 

"Johnson,  V,.  P.  29/23  0910);  Heraeus.  C.  P.  239,704  (1913):  Zinimermann,  U.  S. 
P.   1,055,199  (1913). 


INDEX 


PAGE 

Acheson,  Dr.  Edward  G.,  and  His  Work — F.  A.  J.  FitzGerald S 

Air  Electrode,  Electrotitration  with  the  Aid  of  the — N.  Howell  Furman  79 

Alkaline  Solutions,  The  Hydrogen  Electrode  in— A.  H.  W.  Aten 89 

Alloy  for  Thermocouples,  The  Preparation  of  Platinum  and  of  Plati- 
num-Rhodium— Robert    P.    Neville 371 

Alloying  Elements  in  Steel,  Inherent  Effect  of — B.  D.  Saklatwalla 271 

American  Electrochemist  Abroad,  Opportunities  for  the — C.  G.  Schlue- 

derberg    21 

Annual  Report  of  the  Board  of  Directors 12 

Annual  Report,   Secretary's 13 

Annual  Report,  Treasurer's 17 

Arcs,  Carbon,  The  Relation  Between  Current,  Voltage  and  the  Length 

of— A.  E.  R.  Westman 171 

Arsem,  W.  C. — Discussion 166  ct  seq.,  229,  313 

Artificial  Magnetite,   Oxygen  Overvoltage  of,  in   Chlorate  Solutions — 

H.  C.  Howard   51 

Aten,  A.  H.  W. — Discussion 77  et  seq. 

Aten,  A.  H.  W. — The  Hydrogen  Electrode  in  Alkaline  Solutions 89 

Base   Metal,  The  Influence  of,   on   the  Structure  of  Electrodeposits — 

W.  Blum  and  H.  S.  Rawdon See  Vol.  44 

Baughman,  Will — Discussion 313  et  seq. 

Baughman,  Will — Notes  on  the  Metallurgy  of  Lead  Vanadates 281 

Becket,  F.  M..— Discussion 268  et  seq. 

Becket,  F.  M. — Some  Effects  of  Zirconium  in  Steel 261 

Benjamin,  E.  O. — Discussion 75  ct  seq.,  349 

Benzene,  Electrolytic  and  Chemical  Chlorination  of — Alexander  Lowy 

and  Henry   S.  Frank 107 

Blum,    W.    and    H.    E.    Haring — Current    Distribution    and    Throwing 

Power  in  Electrodeposition See  Vol.  44 

Blum,  W.  and  H.  S.  Rawdon— The  Influence  of  the  Base  Metal  on  the 

Structure  of   Electrodeposits See   Vol.  44 

Board  of  Directors,  Annual  Report  of  the 12 

Boron,  Uranium,  Titanium,  Cerium  and  Molybdenum  in  Steel,  Experi- 
ments with— H.  W.  Gillett  and  E.  L.  Mack 231 

Brooke,   Frank   W.— Methods    of    Handling  Materials    in   the   Electric 

Furnace  and  the  Best  Type  of  Furnace  to  Use 149 

Caplan,    P. — Discussion 75 

Caplan,  P.,  M.  Knobel  and  M.  Eiseman— The  Effect  of  Current  Density 

on    Overvoltage    55 

I  409 

27 


410  INDEX. 

PAGE 

Carbon  Arcs,  The  Relation  Between  Current,  Voltage  and  the  Length 

of— A.  E.  R.  Westman 171 

Carter,  F.  K.— Discussion 383,  395 

Carter,  Fred  E. — Some  Notes  on  the  Metals  of  the  Platinum  Group.. 397 
Cerium,  Uranium,  Boron.  Titanium  and  Molybdenum  in  Steel,  Experi- 
ments with— H.  W.  Gillett  and  E.  L.  Mack 231 

Chemical,  and  Electrolytic,  Chlorination  of  Benzene — Alexander  Lowy 

and   Henry  S.  Frank 107 

Chlorate  Solutions,   Oxygen   Overvoltage   of   Artificial  Magnetite   in — 

H.   C.   Howard 51 

Chlorides,  The   Reduction  of   Some  Rarer   Metal,  by   Sodium — M.   A. 

Hunter  and  A.  Jones  See  Vol.  44 

Chlorination  of  Benzene,  Electrolytic  and  Chemical — Alexander  Lowy 

and  Henry   S.  Frank 107 

Chromizing— F.    C.   Kelley 351 

Cobalt — Its  Production  and  Uses — C.  W.  Drury 341 

Cone,  E.   F. — Discussio>i 268 

Conversion    of   Diamonds   to    Graphite   at   High   Temperatures,   The — 

M.  deKay  Thompson  and  Per  K.  Frolich 161 

Cooper,  H.  S. — Discussion 227  et  sea. 

Cooper,  Hugh  S. — The  Preparation  of  Fused  Zirconium 215 

Crosby,   E.   L. — Discussion 200 

Cunningham,   Thos.   R.,   and   Jas.   A.   Holladay — Experiments   Relative 

to  the  Determination  of  Uranium  bj'  Means  of  Cupferron 329 

Cupferron,  Experiments  Relative  to  the  Determination  of  Uranium  by 

Means  of — Jas.  A.  Holladay  and  Thos.  R.  Cunningham 329 

Current  Densitj',  The  Effect  of,  on  Overvoltage — M.  Knobel.  P.  Caplan 

and    M.    Eiseman 55 

Current  Distribution  and  Throwing  Power  in  Electrodeposition — H.  E. 

Haring  and  W.  Blum See  Vol.  44 

Current,  Voltage  and  the  Length  of  Carbon  Arcs,  The  Relation  Be- 
tween— A.  E.  R.  Westman 171 

Dawson,  F.    G. — Discussion — 186 

Detinning,   Electric  Furnace,   and   Production  of   Synthetic  Gray  Iron 

from   Tin-Plate   Scrap — C.   E.   Williams,   C.   E.   Sims  and   C.  A. 

Newhall     191 

Diamonds,    the    Conversion    of,    to    Graphite    at    High    Temperatures 

M.  deKay  Thompson  and  Per  K.  Frolich 161 

Doremus,   Chas.   A. — Discussiion 323 

Drury,  C.  W. — Cobalt — Its  Production  and  Uses 341 

Drury,  C.   \Y.— Discussion 350 

Edward  G.  Acheson  and  His  Work— F.  A.  J.  FitzGerald 5 

Effect  of  Current  Density  on  Overvoltage.  The —  M.  Knobel,  P.  Caplan 

and    M.    Eisiman 55 


INDEX.  4H 

PAGE 
Effect  of  Iron  on  the  Electrodeposition  of  Nickel,  The — M.  R.  Thomp- 
son   See    Vol.   44 

Eiseman,  M.,  M.  Knobel,  and  P.  Caplan— The  Effect  of  Current  Density 

on    Overvoltage 55 

Electrically   Heated   Apparatus,    Heat    Insulating  Materials    for — J.   C. 

Woodson    127 

Electric  Furnace    Detinning  and    Production   of   Synthetic   Gray   Iron 
from  Tin-Plate  Scrap — C.  E.  Williams,   C.  E.   Sims  and  C.  A. 

Newhall 191 

Electric  Furnace,  Methods  of  Handling  Materials  in  the,  and  the  Best 

Type  of  Furnace  to  Use — Frank  W.  Brooke  149 

Electrochemist  Abroad,  Opportunities  for  the  American — C.  G.  Schlue- 

derberg    21 

Electrode,  Air,  Electrotitration  with  the  Aid  of  the — N.  Howell  Furman  79 

Electrode,  Hydrogen,  in  Alkaline  Solutions — A.  H.  W.  Aten 89 

Electrodeposition",  Current  Distribution  and  Throwing  Power  in — H.  E. 

Haring  and  W.  Blum  See  Vol.  44 

Electrodeposition  of  Iron,  Notes  on  the — Harris  D.  Hineline 119 

Electrodeposition  of  Nickel  on  Zinc,  The — A.  Kenneth  Graham, 

See  Vol.  44 
Electrodeposition  of  Nickel,  The  Effect  of  Iron  on  the — M.  R.  Thomp- 
son    See  Vol.  44 

Electrodeposits,  The  Influence  of  the  Base  Metal  on  the  Structure  of — 

W.  Blum  and  H.  S.  Rawdon  See  Vol.  44 

Electrolytic  and  Chemical   Chlorination   of  Benzene — Alexander  Lowy 

and  Henry  S.  Frank   107 

Electrotitration  with  the  Aid  of  the  Air  Electrode — N.  Howell  Furman  79 
Experiments  Relative  to  the  Determination  of  Uranium  by  Means  of 

Cupferron — Jas.  A.  Holladay  and  Thos.  R.  Cunningham 329 

Experiments   with   Uranium,   Boron,  Titanium,  Cerium   and  Molybde- 
num in  Steel— H.  W.  Gillett  and  E.  L.  Mack 231 

Fink,  Colin  G.— Discussion 53,  167  ct  seq..  312,  349,  368 

FitzGerald,  F.  A.  J. —Discussion 147,   168 

FitzGerald,  F.  A.  J.— Dr.  Edward  G.  Acheson  and  His  Work 5 

Forty-third  General  Meeting,  Proceedings  of 1 

Frank,    Henry    S.,    and    Alexander    Lowy — Electrolytic    and    Chemical 

Chlorination   of   Benzene 107 

Frolich,  Per  K.,  and   M.  DeKay  Thompson — The  Conversion  of  Dia- 
monds to  Graphite  at  High  Temperatures 161 

Furman,    N.   H. — Discussion 87 

Furman,  N.  Howell — Electrotitration  with  the  Aid  of  the  Air  Electrode  79 

Fused  Zirconium,  The  Preparation  of — Hugh  S.  Cooper 215 

General    Meeting,   Forty -third,    Proceedings   of 1 

Gillett,  H.  W.,  and  E.  L.  Mack — Experiments  with  Uranium,  Boron,  Ti- 
tanium, Cerium  and  Molybdenum  in  Steel 231 


412  INDEX. 

PAGE 

Gillett,  H.  W.— Discussion 201  et  seq.,  258,  268  et  sea. 

Graham,  A.  Kenneth — The  Electrodeposition  of  Nickel  on  Zinc. 

See  Vol.  44 
Graphite.  The  Conversion  of  Diamonds  to,  at  High  Temperatures — M. 

DeKay  Thompson  and  Per  K.  Frolich 161 

Gray  Iron,   Synthetic,   from  Tin-PIate  Scrap,   Electric  Furnace   Detin- 
ning  and  Production  of — C.  E.  Williams,  C.  E.  Sims  and  C.  A. 

Newhall     191 

Guests  and  Members  Registered  at  the  Forty-third  General  Meeting...   18 

Guiterman,  Kenneth  S. — Discussion 347  et  seq. 

Handling  Materials  in  the  Electric  Furnace.  Methods  of.  and  the  Best 

Type  of  Furnace  to  Use — Frank  W.  Brooke 149 

Haring,    H.    E.   and    W.    Blum — Current    Distribution    and    Throwing 

Power  in  Electrodeposition See  Vol.  44 

Hart,  L.  O. — Discussion 368    et  seq. 

Heat   Insulating  Materials    for    Electrically    Heated    Apparatus — J.    C. 

Woodson    127 

Hering,    Carl — Discussion 77,    146 

Hineline,  Harris  D. — Notes  on  the  Electrodeposition  of  Iron 119 

Holladay,  Jas.  A.  and  Thos.  R.  Cunningham — Experiments  Relative  to 

the  Determination  of  Uranium  by  Means  of  Cupferron 329 

Horsch,  W.  G.— Discussion 54,  75,  88 

Howard,  H.   C. — Discussion 50,   54 

Howard,  H.  C. — Oxygen  Overvoltage  of  Artificial  Magnetite  in  Chlo- 
rate   Solutions 51 

Hunter,   M.   A.  and  A.  Jones — The  Reduction  of  Some   Rarer  Metal 

Chlorides  by  Sodium See  Vol.  44 

Hydrogen  Electrode  in  Alkaline  Solutions,  The— A.  H.  W.  Aten 89 

Influence  of  the  Base  Metal  on  the  Structure  of  Electrodeposits,  The — 

W.  Blum  and  H.  S.  Rawdon See  Vol.  44 

Inherent  Effect  of  Alloying  Elements  in  Steel — B.  D.  Saklatwalla 271 

Insulating  Materials,  Heat,   for   Electrically   Heated   Apparatus — J.   C. 

Woodson    127 

Investigations  on   Platinum   Metals  at  the   Bureau   of   Standards — Ed- 
ward Wichers  and  Louis  Jordan   385 

Ionization  Problems,  Newer  Aspects  of — Hugh  S.  Taylor 31 

Iron,  Notes  on  the  Electrodeposition  of — Harris  D.  Hineline 119 

Iron,  Synthetic  Gray,   from  Tin-Plate  Scrap,   Electric  Furnace  Detin- 
ning  and  Production  of — C.  E.  Williams,  C.  E.  Sims,  and  C.  A. 

Newhall     191 

Iron,  The  Effect  of,  on  the  Electrodeposition  of  Nickel — M.  R.  Thomp- 
son   See  Vol.  44 

James,  C. — Present  Status  of  the  Production  of  Rarer  Metals 203 

Johnston,    John — Discus^non 48 


INDEX.  413 

PAGE 

Jones,  A.  and  M.  A.  Hunter — The  Reduction  of  Some  Rarer  Metal 

Chlorides  by  Sodium See  Vol.  44 

Jordan,    Louis     and     Edward    Wichers — Investigations     on     Platinum 

Metals  at  the  Bureau  of  Standards 385 

Jordan,    Louis — Discussion 384 

Kelleher,    J. — Discussion 188 

Kelley,    F.    C— Chromizing 351 

Kelley,  F.  C. — Discussion 370 

Knobel,  M. — Discussion 54,  75,  77  et  seq.,  104  et  seq. 

Knobel,  M.,  P.  Caplan,  and  M.  Eiseman — The  Effect  of  Current  Den- 
sity   on    Over  voltage 55 

Knobel,  M. — The  Reactions  of  the  Lead  Storage  Battery 99 

Lead  Storage  Battery,  The  Reactions  of  the — M.  Knobel 99 

Lead  Vanadates,  Notes  on  the  Metallurgy  of — Will  Baughman 281 

Lind,  S.  C. — Discussion AS  et  seq..  168  et  seq. 

Lowy,    Alexander    and    Henry    S.    Frank — Electrolytic    and    Chemical 

Chlorination   of   Benzine 107 

Mack,  E.  L.  and  H.  W.  Gillett — Experiments  v/iih  Uranium,  Boron,  Ti- 
tanium, Cerium  and  Molybdenum  in   Steel 231 

Magnetite,  Artificial,  Oxygen  Overvoltage  of,  in  Chlorate  Solutions — 

H.  C.  Howard   . . .' 51 

Marden,  J.  W.,  and  H.  C.  Rentschler — Discussion 323  ef  seq. 

Marden,  J.  W. — Discussion 225    et  seq. 

Members  and  Guests  Registered  at  the  Forty-third  General  Meeting...   18 
Metal  Chlorides,  The  Reduction   of   Some  Rarer,   by   Sodium — M.  A. 

Hunter  and  A.  Jones See  Vol.  44 

Metallic  Uranium,  Preparation  of — R.  W.  Moore 317 

Metallurgy  of  Lead  Vanadates,  Notes  on  the — Will  Baughman 281 

Metals  of  the  Platinum  Group,  Some  Notes  on  the — Fred  E.  Carter... 397 

Metals,  Rarer,  Present  Status  of  the  Production  of — C.  James 203 

Methods  of  Handling  Materials  in  the  Electric  Furnace  and  the  Best 

Type  of  Furnace  to  Use — Frank  W.  Brooke 149 

Molybdenum,  Uranium,  Boron,  Titanium,  and  Cerium  in  Steel,  Experi- 
ments with— H.  W.  Gillett  and  E.  L.  Mack 231 

Moore,   R.    B. — Discussion 350 

Moore,  R.  W. — Discussion 326    et  seq. 

Moore,  R.  W. — Preparation  of  Metallic  Uranium 317 

Moore,    W.    C. — Discussion 49 

Neville,  Robert  P. — The  Preparation  of  Platinum  and  of  Platinum- 
Rhodium  Alloy   for  Thermocouples 371 

Newer  Aspects  of  Ionization  Problem.s — Hugh  S.  Taylor 31 

Newhall,  C.  A.,  C.  E.  Sims,  and  C.  E.  Williams — Electric  Furnace  De- 
tinning  and  Production  of  Synthetic  Gray  Iron  from  Tin-Phte 
Scrap    191 


414  INDKX. 

PAGE 
Nickel,  The  Effect  of  Iron  on  the  Electrodeposition  of — M.  R.  Thomp- 
son    See  Vol.  44 

Nickel,  The  Electrodeposition  of,  on  Zinc — A.  Kenneth  Graham, 

See  Vol.  44 

Notes  on  the  Electrodeposition  of  Iron — Harris  D.  Hineline   119 

Notes  on  the  Metallurgy  of  Lead  Vanadates — Will  Baughman 281 

Opportunities  for  the  American  Electrochemist  Abroad — C.  G.  Schlue- 

derberg    21 

Overvoltage,  Oxygen,  of  Artificial  Magnetite  in  Chlorate  Solutions — 

H.  C.  Howard    51 

Overvoltage,  The  Effect  of  Current  Density  on — M.  Knobel,  P.  Caplan, 

and  M.  Eiseman    55 

Oxygen    Overvoltage   of    Artificial    Magnetite   in    Chlorate    Solutions — - 

H.  C.  Howard    51 

Platinum  Group,  Some  Notes  on  the  Metals  of  the — Fred  E.  Carter. ..  .397 
Platinum   Metals  at  the  Bureau  of   Standards,   Investigations  on — Ed- 
ward Wichers  and  Louis  Jordan 385 

Platinum,   The   Preparation   of,   and   of   Platinum-Rhodium   Alloy    for 

Thermocouples — Robert  P.  Neville   371 

Preparation  of  Fused  Zirconium,  The — Hugh   S.  Cooper 215 

Preparation  of  Metallic  Uranium — R.  W.  Moore 317 

Preparation  of  Platinum  and  of  Platinum-Rhodium  Alloy  for  Thermo- 
couples. The— Robert  P.  Neville 371 

Present  Status  of  the  Production  of  Rarer  Metals — C.  James 203 

Proceedings  of  the  Forty-third  General  Meeting 1 

Ralston,  O.  C. — Discussion 87,  349  et  seq. 

Rarer   Metal   Chlorides,   The  Reduction  of    Some,   by   Sodium — M.   A. 

Hunter  and  A.  Jones  See  Vol.  44 

Rarer  Metals,  Present  Status  of  the   Production  of — C.  James 203 

Rawdon,  H.  S.  and  W.  Blum — The  Influence  of  the  Base  Metal  on  the 

Structure   of   Electrodeposits See   Vol.  44 

Reactions  of  the  Lead  Storage  Battery,  The — M.  Knobel 99 

Reduction  of   Some  Rarer   Metal   Chlorides  by  Sodium,  The — M.  A. 

Hunter  and  A.  Jones  See  Vol.  44 

Reeve,    H.    T. — Discussion 383 

Relation   Between  Current,  Voltage  and   the   Length  of   Carbon   Arcs, 

The— A.   E.   R.  Westman 171 

Rentschler,  H.  C.  and  J.  W.  Marden — Discussion 323    et  scq. 

Report,  Annual,  of  the  Board  of  Directors 12 

Report,    Annual,    Secretary's ■ 13 

Report,   Annual,   Treasurer's 17 

Report  of  Tellers  of  Election 4 

Rhodium-Platinum    Alloy    for    Thermocouples.    The    Preparation    of 

Platinum  and  of — Roliert  P.    Neville 371 

Richardson,  H.  K. — Discussion 369,  382 


INDEX. 


415 


_    ,  ,           „  PAGE 

baklatwalla,    B.    D.— Discussion 312 

Saklatwalla,  B.  D.— Inherent  Effect  of  Alloying  Elements  in  Steel 271 

Schluederberg,    C.   G.— Discussion 200 

Schluederberg,  C.   G.— Opportunities   for  the  American  Electrochcmist 

Abroad    21 

Scrap,  Tin-Plate,  Electric  Furnace  Detinning  and  Production  of  Syn- 
thetic Gray  Iron   from— C.  E.  Williams,   C.  E.   Sims  and   C.A. 

Newhall     jni 

Secretary's  Annual   Report    J3 

Sims,  C.  E.,  C.  E.  Williams  and  C.  A.  Xewhall— Electric  Furnace  De- 
tinning  and  Production  of  Synthetic  Gray  Iron  from  Tin-Platc 

Scrap    291 

Sodium,   The  Reduction   of   Some    Rarer   Metal    Chlorides  by— M.   A. 

Hunter  and  A.  Jones  '  See  Vol.  44 

Some  Eflfects  of  Zirconium  in  Steel — F.  M.  Becket 261 

Some  Notes  on  the  Metals  of  the  Platinum  Group— Fred  E.  Carter. ..  .397 
Steel,  Experiments  with  Uranium,  Boron,  Titanium,  Cerium  and  Mo- 
lybdenum in— H.  W.  Gillett  and  E.  L.  Mack 231 

Steel,  Inherent  Effect  of  Alloying  Elements  in— B.  D.  Saklatwalla 271 

Steel,  Some  Effects  of  Zirconium  in — F.  M.  Becket 261 

St.   John,   Ancel — Discussion jgp 

Storage  Battery,  Lead,  The  Reactions  of  the— M.  Knobel 99 

Stoughton,,    Bradley — Discussion 258 

Synthetic   Gray   Iron    from   Tin-Plate   Scrap,    Electric  Furnace   Detin- 
ning and  Production  of— C.  E.  Williams,  C.  E.  Sims  and  C.  A. 

Newhall     joj 

Taylor,  H.  S. — Discuss-wn   A9  et  sea 

Taylor,  Hugh  S.— Newer  Aspects  of  Ionization  Problems '. 31 

Tellers  of  Election,  Report  of 4 

Tin-Plate  Scrap,  Electric  Furnace  Detinning  and  Production  of   Syn- 
thetic Gray  Iron   from— C.  E.  Williams,  C.  E.  Sims  and  C.  A. 

Newhall    jgj 

Titanium,  Uranium,  Boron.  Cerium  and  Molybdenum  in  Steel,  Experi- 
ments with— H.  W.  Gillett  and  E.  L.  Mack .' 231 

Thermocouples,  The   Preparation   of   Platinum   and  of  Platinum-Rho- 
dium Alloy  for— Robert  P.  Neville 37I 

Thompson.   M.  DeKay  and   Per  K.  Frolich— The  Conversion  of  Dia- 
monds to  Graphite  at  High  Temperatures 161 

Thompson,  M.  R.— Discussion 86  ^-^  seq. 

Thompson,   M.    R.— The   Effect   of    Iron   on    the   Electrodeposition    of 

Nickel   See  Vol.  44 

Throwing    Power,    and    Current    Distribution,    in    Electrodeposition— 

H.  E.  Haring  and  W.  Blum See  Vol.  44 

Treasurer's  Annual  Report    \y 


4i6  INDEX. 

PAGE 

Uranium,  Boron,  Titanium,  Cerium  and  Molybdenum  in  Steel,  Experi- 
ments with— H.  W.  Gillett  and  E.  L.    Mack 231 

Uranium,  Experiments  Relative  to  the  Determination  of,  by  Means  of 

Cupferron — Jas.  A.  Holladay  and  Thos.  R.  Cunningham 329 

Uranium,  Metallic,  Preparation  of — R.  W.  Moore 317 

Vanadates,  Lead,  Notes  on  the  Metallurgy  of — Will  Baughman 281 

Voltage,  Current,  and  the  Length  of  Carbon  Arcs,  The  Relation  Be- 
tween— A.   E.  R.   Westman 171 

Weir,  Helen — Discussion 104  et  scq. 

Westman,  A.  E.  R. — Discussion 187  et  seq. 

Westman,  A.  E.  R. — The  Relation  Between  Current,  Voltage  and  the 

Length  of  Carbon  Arcs 171 

Wichers,    Edward     and     Louis    Jordan — Investigations     on     Platinum 

Metals  at  the  Bureau  of  Standards 385 

Williams,  C.  E.,  C.  E.  Sims  and  C.  A.  Newhall — Electric  Furnace  De- 
tinning  and  Production  of  Synthetic  Gray  Iron  from  Tin-Plate 
Scrap    191 

Williams,  C.  E. — Discussion .201  et  scq. 

Woodson,   J.    C. — Discussion 147 

Woodson,  J.  C. — Heat  Insulating  Materials  for  Electrically  Heated  Ap- 
paratus     127 

Zinc,  The  Electrodeposition  of  Nickel  on — A.  Kenneth  Graham, 

See  Vol.  44 

Zirconium,  Fused,  The  Preparation  of — Hugh  S.  Cooper 215 

Zirconium  in  Steel,  Some  Effects  of — F.  M.  Becket 261 


"^w 


TP 
250 

V.-43 


Electrochemical  Society 
Transactions 


>^nnliaH  sS 


3eFIAU 


Engineering 


PLEASE  DO  NOT  REMOVE 
CARDS  OR  SLIPS  FROM  THIS  POCKET 


UNIVERSITY  OF  TORONTO  LIBRARY 


£:noi^