Figure 1.
(Lejft)2000-impere Carbon-inode Fluorine ( ell w i t h One inorle i*senibly Kemo~etl. ( R i g h t ) Imohitig Down on 2000- i m p e r e (.ell M ith Otie itlode issembl? Kenioted
CARBON-ANODE FLUORINE CELL R . L. JIurray, S. G. Oshorne, and 31. S. Kircher HOOKER E L t C T R O ( : I I F : \ I I ( , \ L ( ' O \ l l ' 4 \ \ ,
AiTEin 19-12 this company \\-a?reque-ted by the Office of Scientific Research and Development to develop a commerr i d cell for the production of fluorine. Several other groups were working on the prohlem and all findings and discoveries vere ~)oole.d;also, there xi> a f a i r aiiiount of information in theliterature ronceriiing m a l l laboratory cella, particularly as to availaldr electrolyte5 and materials of construction. Holyever, the+% cell5 seldom opewtetl at more than 100 amperes and generally a t considerably less; the;\- tvere essentially laboratory apparntu? and iiot a type adaptalilc t o crpansion t o a large cell operating :It 1500 t o 2000 ampere.>. Tlir prior art n-as reviewed hy Cad?, I l o g e i , ~and , Carl-on i n 1912 ( 2 ) . The pl'(JblC.111roii&ted of tivo parts: ( a ) the choice of elec.ti,oet1 t h e i n a t e r i n l ~of construction and operating teml)ci.:itiii,c., ~ i n d( b ) tlie nicclianicnl design of the cell i t d f . \Tit11 the ercq)tion of one or two commercially impractical c ~ l t c t i ~ o l y ttho-r e ~ , rrcorded and kno~viit o b e available were all different cciiiiliinntions of anhydrous hydrofluoric acid and potwsium fluoride, anhydrous hydrofiuoric acid itself being a noiicoiiclurtor. I3rondl!-, there were four practical ranges of hylrofliioric 249
\I\ing through tion, the vapor pressure omy and ease of operation. ilthough the carbon-anode of hydrogen fluoride over insulators on t h e main cells reached a fairl> sati-factory state of practical deielopcell cover. Individual this electrolyte is high. nient and w e r e operated in large numbers in long-series As other groups were iuspension and gas outcircuits, no adequate 9' i t e m for continuous hydrofluoric working on t h e high lets for each cathode are acid feed w a s deFeloped for use under these conditions. temperature cell and eliminated, and three Intermittent hydrofluoric. arid feed. 1)atchw ice, w a s found insulatorq and one liydroalso on cells uiing to be practical b u t did not gibe the best cell operation. gen outlet R I P U - P ~for KF.2 HF n i t h nickel a number of cathodes. anodes, both on a conThe diaphragm as.jenibly siderable scale, it n a s consists of a grid contnindecided t h a t this com)ntitl slots, 11laced so :hat the solid hai,s hetween the slots pany should ininiediately design and c o n ~ t r u c t n full+izeti lie directly over tlie cathodes. B y Ivelding vertical suq)entled 2000-ampere steel cell using KF.2HF with, if possible, graphite .-creen diap1ir:ignis around t h e edge- of the slots, the hydrogen anodes. T h e large size 11-as chosen partly in tlie Iiopr of saving rising f w m the cathodes is trapped under t l i ~solid liars and time b u t largely because a 2000-ampere cell is pilot plant size, pmsos Irngtli\\-iie t o t h r end of the cell. T h e gi,iti with its and, in terms of actual design and operation, not niuch infornindependiiig screens i.q bolted t o the bottom of :I shallon. box tion can be obtained from anything smaller. d through the main cover of the cell m t l fornie a reLittle 17-as apparent in t h e prior a r t t h a t could tie used in tlir mechanical design of a coniniercial cell. Hon-ever, i i i ninny ivayq phctxtile diaplii,;igiii elnbly. By sucpendiiig t h e anode assciiil)ly Emni t h e rt~inovahlecorer of the deprescd box, the pnra fluorine cell is closely related t o the all-metd oxygen-hydrogcri allel ro\vh of :inotles pass through the slots :inti take up :L poiition cell, because both electrode product. are g:\.w, t h e niaterinl being inside the di:iplirngins, and again a number of separirte suspension decomposed is fed t o an electrolyte which is iiot conwnietl, ancl i d s and gas pipri are eliminated. The fluorine ri.5ing froni the much metal construction can be w e d . On tlie othpr h:ind, IJP+iii$:ic.e of the aIi(JdW pa.ises up through t h e slots and is trapped in cause of t h e extreme corrosiveness and the dangerous nature of the hox, froni wliich it is ivithdrawn through a single pipe in the the liquids and ga.;es handled, a fluorine cell reseiiibles a chlorineside. The large number of insulators, suspension rods, and gas alkali cell in certain respects, and fairly frequent r e n e n d of jo11ie outlet> generally found in multiple electrode cells of the oxygenparts was t o be expect,ed. L o n renelval cost deniaiided rapid and hydrogen type is eliminated, and provision is made for rerien-a1 by cheap replacement of electrodes and diaphragms as unit ahlarge n~senihliesas in chlorine-alkali practice. SIucli o f this genPeniblies, a? in chlorine practice, so th:it t h e actual time-consunieral principle \\-vas incorporated in t h e de.ign of the fir3t full-.-ized ing asenibly of small parts could take place out-ide and :i\v:iy cell and wai virtu:illy complctrly accompliihecl i n tlie final d 4 g n . from the cell. T h e cell itself n-ould then lie +hiit c1on.n only during t h e quick exchange or replacement of such assenililie.: m i d iiot E'oI reasons ~rliich\rill appear later, tlie 2000-:inipcr(~ cell (Figure 1) actually consists of two 1000-:tmpcre cell> comic~ctctliii during their rather s l o ~ construction. T,ow current tleniity and ill a single tank or ccll body; that is, tl1ei.e are tn.o anml economy of floor space called for multiple p r a l l e l electrodes acblies, two diaphragm assemblies, and two cathotle n ~ m n h l i e sin a tive on both sides; this v a s t h e most difficult feature t o incorsingle tank. More t h m tn-o 1000-aniper,eunits CILII t i e incorporated porate n-ithout resorting t o a multiplicity of pipes and elcctiical in a single tank if desirxble; the iize of the cell is thrWhy increased. connections. Figures 2 to ;i are dran-ings of a 1000-ampere cell, built and p u t Some twenty variations were cunsidered before a design \vas iiito operation in June 1914; i t incorporated the reniov:hle diafound which incorporated most of the desiixble featurei. Ha-iphragm and corrected some faults of electrode and diaphragm cally the cell consists of four main parts: :L cell body, a cathode spacing, and may he regarded as representing the final bnhic deassembly, a diaphragm assembly, and an anode
March 1947
INDUSTRIAL AND ENGINEERING CHEMISTRY
sign. I n a large installation of 2000-ampere cells built and operated shortly thereafter, these improvements were not incorporated, since it v-as considered t h a t they had not been sufficiently tested in actual 50-60 operation. 100 CELLBODY. The cell body (Figures 2, 3, and 4.4) 250 is a steel tank provided with a n-ater jacket for heating and cooling. The main cover carries the hydrogen outlet pipe, hydrofluoric acid feed line, nitrogen flush lines, a thernionieter well, and an electrolyte sampling pipe. There are no pipe connections on the sinall cover from \vliich t h e anodes are suspended. Soft copper gaskets, either flat or round, are u5ed t o seal t h e tn-o covers. ASODE.;. T h e anode assembly (Figure 1,B and C) consists of fourteen 18 x 6.25 x 1.25 inch carbon blades arranged in two parallel rows of 5even each. T h e blades are bolted t o a heavy inverted copper channel running hetween the two ron-s, using four bolts and a separate copper cover plate for each blade. T h e separate cover Idntei ninke it possible to replace a single blade without disturbing t h e others. T h e inverted copper channel is suspended by c o p l m rods through two insulators xelded t o the small steel cover. The joints between the carbon and the copper are made \vitli a comniercial carbon-n-ater paste (Sational Carbon Company, Inc.) from ~vliichthe n-ater is driven by heating after t h e bolts are dran-n UI) tight. T h e blade.5 are electrochemically active over the loirer 12 inches so t h a t the fourteen blades correspond t o approxiinntely 14 .;quare feet of operating anode surface, both sides being active. T h e cell runs a t 71.5 amperes per square foot a t 1000 anipereu and 53.5 amperes per square foot a t 750 amperes. Similarly, the 2000-ampere cell, n-hich has tn-o complete sets of electrodes, runs a t 71.5 amperes per square foot at 2000 amperes and 53.5 amperes per square foot at 1500 amperes. CATIIODES. T h e cathode assembly (Figure 5, d and B ) consists of three parallel steel plates welded to cross plates at the ends and suspended through three insulators on t h e main cover. T h e two adjacent suspension rods a t one end of the cell are elect,rical
IiF.3HF 1iF.2Hl' TiF.RT
25 1
rapper. nickel Irun
Copper, nickel Iron
Copper, nickel Iron
('opper. niagnesiuin
C'ripper, riiagnesiuiri
Copper, magnesium
iridium Xickel
Xickel, gra h i t e Grapiite
conductors; the .ingle rod a t the opposite end of the cell is for supnort only. DI.~I~FIR.A(+w. T h e diaphragm assembly (Figure 5 , C arid U ) consists of :I iwtaiigulnr steel angle frame: an inverted steel channel, forming tlic grid, is welded through the ends of t h e flnme. T h e hoslilce structures, made of sis-mesh steel screen, :ire welded to the nnglec and the tn-o Fides of the inverted chnnnel, Forniing four .separate pardlel diaphragms, lxtn-een and evenlJspaced from, the anodes and cathodes a t all points. The angle frame iq bolted to the bottom of the 'nos set into the in:iiii cover and makes the diaphragm a-sembly renioJ-able. .'ji.i~Pr.E\Ii:s.TBR'i EQL-IPIIEST. Sitrogen flush lines are n-eltitd t o the anode and cathode chambers tor flushing before disniaiitling and lxfore talciiig samples of electrolytr,. A thermomet,er 1 ~ 1 is 1 provided on the rathode side. The hydlofluoric x i d i:. f i d through a pipe submerged about S inches in the electrolyte. This is a simple open-ended pipe: elaborate sparger pipes lvei'cl found t o be unnecesiary, because niising is very rapid. The electrolyte level is read by bubbling nitrogen through a submerged rluurtei,-inch pipe and reading the preas~ireon a gage or manometer. Electrolyte sample> are taken through a 1-inch pipe on the cathode side fitted n-ith a gate valve. T h e gate d v e is screived down onto a copper sleeve liner Trhich project& beloTv the electrolyte level and prevents air from entering the cathode chamber, where it would combine esplosively Ivith the hydrogen. Any of t,lme pipes ivhich projPct into the clect,rolyte are renewable. A
a
a
I
'1
Figure 3.
Sectional Elevation of 1000- i n i p e r e Fluorine Cell
252
Vol. 39, No, 3
INDUSTRIAL AND ENGINEERING CHEMISTRY
A C
Figure 1.
itiocle
ntid
(:ell (:ross Section of 1000-lmpere Fluorine Cell
thrradcd nipple one size largt:~ i- T\.eltletl to tliv pi1)c : ~ i i c l s ~ ' e \ i . into :I coupling welded to the cell cover. As f:ir LIS possible, :ill pipes :ind bus bars, etc., ai'(: placed a t on(: e n d of the cell so t h t the space over and above the cell may 1)~; clear for :inode :ind diaphragm renen.al. For en,y removal, tlir small c o v c ~form ' ivliicli the anodes are suspended i.s kept free from pipes, and the nitrogen flush line enters the :Ul[Jde c1iamhc.r through the d e of the main coyer. r..i'ivm. T h e electrod?.> are suspended by q ~ p f or q stet.1 bars ixishing through insulators of the stuffing-hoi tylie. Their are Ixiclmi x i t h ring:: of ~~olytctrafluoroetliylei~c, in :rccordn~icc with tlic, nietliod devcloped bj- the 111.1Font ('omliaiiy: top , bottom i,iiigs containing calcium fluoride are used t o ri1.r -tiffili ~ * ~ i m x i t o L m EThe . electrolyte gives h t t k trouhli3. .\t t I plant the same electrolyte n-a> used for over :I ~ P : I I , ; i i i oiic c:i.c* it has been n s d for t w o years. Sludge must be iwnovrd oec;izionally, nnd finally the electrolyte must be diacardrd or -om(. n ~ t l ~ o t l of filtration or settling must he utilized. POI..UUZ.ITIOS. A11 fused salt cells are suhirct t o tlie 1)lienonie n a of po1:irization and anode effect, in which tlie the crll for a given amperage will suddenly inci value, often severnl times its normal oper:iting fi creme in resistance is generally at the anode but occ:ision:~llyrrt the cntliode. Often it appears t o he due to the foi,niiitiori of :I gas filni between the electrode and the electrolyte, ivhich act. a n insulator, I n fluorine cells, n-ater in the rlectrolyte, high current density, nnd impropcr conditions of temperature and elrctrolyte concentration tend to produce polarization. Carbonanode cells are considerably more susceptible than those n-ith nickel modes.