LABORATORY AND TECHNICAL PRODUCTION OF FLUORWE AND ITS COMPOUNDS By H. R. LEECH M.Sc. F.R.I.C. (RESEARCH DEPT. IMPERIAL CHEMICAL INDUSTRIES LTD. GENERAL CHEMICALS DIVISION WIDNES.) " IN my opinion the hazards of work with fluorine and its compounds have been greatly over-rated. When treated with the respect which is due to it fluorine is just another substance." 1 Fluorine was made on an industrial scale in the U.S.A. in the last war in simple equipment largely of mild steel and handled without difficulty. A large fluorine plant was operated in Germany also. Fluorine can now be made in the laboratory or in industry without difficulty and the chemistry of fluorine can begin to overtake that of other common elements which have been readily available for many years. It is of interest to see the stages of the development of fluorine production the problems which have been solved and those which yet remain.Although it is 62 years since Moissan first isolated fluorine in 1886 there was little further work done for 30 years and developments alillost as revolutionary as Moissan's first preparation may be dated €rom 1913 when Mat>hers and his co-workers first used as electrolyte fused potassium acid fluoride. Between 1886 and 1919 there were only some six papers published by workers in the field of elernenta.ry fluorine between 1919 and 1939 there were a t least 25 papers giving some description of fluorine cells as well as 9 patents and many more papers dealing with the use of elementary fluorine. All fluorine cells employ a potassium fluoride-hydrogen fluoride electro- lyte there are several compounds formed in this system and Cady3 gives the data of Table I (the tension of hydrogen fluoride is the approximate partial pressure at the melting point).Electrolytes in three different working regions have been employed (1) Low-temperature electrolyte a relatively dilute solution of potassium 64.3" 61.8 71.7 68.3 239.0 229.5 fluoride (about KF,13HF -55 -20 -15 -26 -15 - with less than 20y0 by weight of potassium TABLE I Solid phase. HF Eutectic . KF,4HF . Entectic . KF,3HF . Eutectic . nil 17.7 42.1 36.3 49.2 52.1 M.p. - 83.7" - 97 72.0 63.6 65.8 62-4 HF tension Inm. -250 -130 - Solid phase. XIF,2*5HF Eutectic . KF',BHF . Eutectic . KF,HF . Eutoctic . KF "/b 53.8 65.8 59.2 61.5 74.4 75.4 HI? tension nm. 1cz.p. 1 G. H. Cady I d . Eng. Chem. 1947 39 No.3 1OA. 2 W. L. Ago F. C. Mathers B. Humiston and C. 0. Anderson Trans. Electrochem. SOC. 1919 35 335. J . Amer. Chern. Soc. 1934 56 1431. 22 LEECH PRODUCTION OF FLUORINE 23 fluoride) as used by Moissan at subnormal temperatures. (2) High- temperature electrolyte approximately molten KF,HF first introduced by Mathers working temperature about 250". (3) Medium-temperature elec- trolyte approximately molten KPJHF first introduced in 1925 by P. Lebeau and A. Damiens ; 4 working temperature about 100". The partial pressure of hydrogen fluoride over the low-temperature electrolyte is very high and the fluorine produced is heavily contaminated. Over the other two electro- lytes the partial pressure is very much less and the fluorine (and hydrogen) are contaminated with 5-15o/d of hydrogen fluoride a t the working tem- perature.have investigated the possi- bility of finding an electrolyte with lower melting point and draw attention to the advantages of czsium acid fluorides owing to the low melting points the hydrogen fluoride tension of these is negligible. Some attention to this point has also been paid by German workers. In the low-temperature electrolyte the corrosion of the platinum anodes which must be employed is very heavy. indicates a current efficiency of 31% on fluorine and a platinum loss of 5 g. per g. of fluorine (although as recently as 1932 proposals were made for a somewhat modified cell working a t room temperature claiming advantages in a nickel anode).' Moissan found that graphite anodes completely disintegrated almost immedi- ately on starting electrolysis in low-temperature electrolyte and found that platinum in high-temperature electrolyte was too rapidly corroded to be of any value.He did not examine graphite in this electrolyte however and it was left to Argo Mathers Humiston and Anderson to show how effective this was and to open the door to a new development in fluorine production. Medium-temperature electrolyte using a nickel anode was proposed by Lebeau and Damiens for the specific reason that the formation of carbon fluorides reputed to occur in the high-temperature cell was avoided and a purer fluorine obtained. This electrolyte had other advnnhages however and will probably oust all others from genera'l use with in addition an amorphous carbon anode which later work has shown to be possible. Of the work with low-temperature electrolyte there is little more to be said.Moissan showed that copper is inert and anodically passive to fluorine and Soci&t& Poulenc Freres et M. Meslans patented a cell in which the platinum anode was attached to a copper cylinder located centrally in a cylindrical cell and surrounded by a perforated copper diaphragm which did not need to be electrically insulated from the anode. Thus they avoided the difficulty of sealing the anode into the anode compartment in a manner which would retain its insulation value and resistance to fluorine. This cell arrangement proposed as the basis for a manufacturing plant instal- lation has a very modern appearance and was exhibited a t the Exposition Universelle de Paris in 1900. F. C. Mathers and B. T. Stroup 0.Ruff Compt. rend. 1928 181 917 ; P. Lebeau Bull. SOC. d'Encourag. 1927 139 15. Trans. Electrochem. SOC. 1934 66 245. " Die Chemie des Fluors " Verlag Springer Berlin 1920. ' W. S. Calcott and A. F. Benning U.S.P. 2,034,458 (1936). G.P. 129,825 (1902). 24 QUARTERLY REVIEWS The early workers in this field laboured under more difficulties than the choice of electrolyte and anode. The raw material for their work was aqueous hydrofluoric acid from which potassium acid fluoride was made and crystallised dehydrated and then decomposed by heating to give the anhydroua acid. It was obviously an advantage of the high-temperature electrolyte that one stage of this preparation could be omitted and then in the 'twenties supplies of the acid fluoride could be purchased and the chemical side of the problem of making fluorine became much simpler.In the 'thirties supplies of anhydrous hydrogen fluoride also became available and the medium-temperature cell with the possibility of electrolyte revivification from this source was a t no disadvantage. Fluorine Cell Development from 1919 to 1939.-The studies on fluorine production by MatJhers and his co-workers were required by the Chemical Warfare Service of the U.S. Government. A cell similar in form to that of Poulenc and Meslans was developed. The heavy copper cont,aining vessel 8 ins. deep 3-5 ins. diameter served as cathode and was externally electrically heated. The sheet-copper diaphragm 2 in. in diameter was slotted a t the lower end and closed a t the bottom. Inside the diaphragm was an Acheson graphite anode suspended from a copper rod which was insulated from the diaphra'gni by rz gland packed with powdered fluorspar rammed hard.It was bettler to insulate the diaphragm for some corrosion occurred when it was anodic. In starting a cell only a low current could be passed. If too high a current density were applied the cell quickly polarised the current fell and a potential of 50-60 volts was required to maintain the original current density. In these circumstances there was the appearance of a gas film round the anode with small sparks through it (this is very similar to the well-known " anode effect " of fusion electro- lysis). This effect was attributed to the presence of small amounts of water in the salt and the formation of an oxygen film owing to its prior discharge. The phenomenon is constantly referred to by later workers.There is no explanation of how the oxygen discharge gives rise to the " gas film " and as will be seen later there is a very interesting field of study,still almost untouched in the phenomena a t fluorine-carbon anodes especially in presence of small quantities of water. Mathers electrolysed a t a current density of 0.1-0-2 amp./sq. in. for several hours. During this period fluorine was not evolved but when i t began to appear the current density could be increased to the normal working figure of 0-5-1.0 amp./sq. in. At 250" t4he cell took 10 amps. a t 15 volts. Some corrosion of the copper parts occurred and copper fluoride thickened the melt a process assisted by the deposition of potassium fluoride as the hydrogen fluoride content decreased.To some extent this process could be avoided by steadily increasing the temperature of elecfro- lysis but this on the other hand increased the rate of corrosion of the copper and the volatiliEation of hydrogen fluoride and above 300° attack on the graphite anode became considerable. As the electrolyte became thicker there was a tendency to frothing and this led to electrolyte being carried into the outlet pipe and then solidifying. The fluorine was then LEECH PRODUCTION OF FLUORINE 25 forced under the diaphragm mixed with the hydrogen and caused explosions. It was not found possible to regenerate the electrolyte by passing ia hydrogen fluoride and therefore when the melting point of the electrolyte was up to about 2S0° it was discarded and a fresh batch introduced.Mathers com- mented that it was remarkable that in the years he was working with fluorine Ruff did not discover this relatively simple method and considered that there were no special difficulties in the manipulation of the apparatus. The problems indicated-sealing in the anode polarisation corrosion of the container outlet-pipe blockage-are the main subjects of attention by subsequent workers. Many cells of a very similar sort of size and capacity were introduced in the next twenty years. in 1921 took up a suggestion of Mnthers and used as container artificial graphite. They had a special ribbed and club-shaped anode to reduce current density and pre-electrolysed at a low current density. They investigated the effect of other fluorides in lowering the melting point of thc electrolyte but found the partial pressurc of hydrogen fluoride was increased by any addition.There T i m some ozone in the anode gas in the early stages but never more than 0.27; G f carbon fluorides. A cell of the Mathers pattern was used by W. R. SmythelO for study of the spectrum of fluorine. The hydrogen fluoride was removed by sodium fluoride and the fluorine then had very little action on glass. J. Simons l1 in 1924 described a cell very similar to that of Mathers he found Portland cement rna'de a n adequate anode seal and the electrolyte was pre-electrolysed a t low current dcnsity in an open cell until fluorine was evolved. Many workers used a cell of the Simons pattern in the next 15 years in some cases monel metal being used in place of copper. S. I?. Whearty junr.,12 used a cast magnesium body and an electrolyte containing 35% of sodium fluoride and working a t 170".M. Bodenstein H. Joliusch and H. Krekeler l3 also used a magnesium alloy cell body but introduced the electrodes from below the sealing in of the electrodes by solidified electrolyte was the subject of a patent assigned to I.G. Farbenindustrie.l4 A return to Moissan's cell type was shown by L. M. Dennis J. M. Veeder and E. G. Rochow,15 who describe a V-shaped cell made of copper pipe and several workers used this arrangement in copper or in nickel. In 1925 Lebeau and Damiens introduced a cell using medium-temperature electrolyte with a metal preferably a nickel anode. This was later to become the leading type. Several workers used this electrolyte and its use by Cady l6 is noteworthy because it was the first case in which anhydrous hydrogen fluoride was obtained from a manufacturer and used for regenera- tion of the electrolyte in situ.F. Meyer and W. Sandow Current efficiency was 750/,. This was in 1935. Rer. 1921 54 759. l1 J . Amer. Chem. SOC. 1924 46 2175. l3 Chem. Pabr. 1936 8 283. l4 H. Krekclcr G.P. 522,885 (1931) ; 558,829 (1932). l5 J . Anaer. Chem. SOC. 1931 53 3263 ; L. M. Dennis and E. G. Rochow ibid. lo Astrophys. J. 1921 54 133. 12 J . Physical Chem. 1931 35 3121. 1934 50 879. l6 Ibid. 1935 57 246. 26 QUARTERLY REVIEWS An important study of factors affecting carbon anode behaviour was made by K. Fredenha'gen and 0. Krefft.17 They dried their potassium bifluoride in a stream of dry air a t an elevated temperature and tested for dryness by distilling off some hydrogen fluoride which had to have a specific conductivity a t 0" of less than 0.02 mho.I n an electrolyte so prepared a practically pure fluorine could be generated immediately a t a graphite or amorphous carbon anode and a reproducible current-voltage curve could be obtained. If water were added polarisation occurred and no reproducible curve could be obtained until the water content exceeded 1% but then the anode gas was oxygen. It was found also that for electrolytes with HF KF ratio below 1.8 graphite was unwetted and could then be used satisfactorily as an anode. Above this ratio (38.3% HF) wetting occurred and caused disintegration of the anode. Amorphous carbon anodes behaved similarly but the limit at which wetting began was different-actually a t a much higher hydrogen fluoride content.The use of anodes of graphite or carbon in an electrolyte which did not wet them was the basis of a patent claim by Fredenhagen.l* Eater papers by A. L. Henne19 and Cady D. A. Rogers and C. A. Carlscjn 2O were not so clear as to when a graphite or carbon modo could be employed in medium- temperature electrolyte. American Wartime Developments.-During the second world war fluorine production was developed in U.S.A. in connecfion with the atom bomb in Germany in connection with a wartime requirement for chlorine tri- fluoride. It is surprising that although the principal interest in America before the war had been in high-temperature cells yet the industrial develop- ments were in medium-temperature cells first with nickel anodes later with carbon anodes ; 21 and whilst it was German work which had prin- cipally pointed the way to the possibilities of the medium-temperature cell wartimo expansion there was of the high- temperature cell.22 I n the American developments cells of 1000-2000-amp. capacity with medium- temperature electrolyte were developed by Hooker Electrochemical Co. Harshaw Chemical Co. and E. I. du Pont de Nemours & Co. A large semi-technical installation of 600-amp. cells of high-temperature type is also described by Fowler et al. and smaller installations by Miller and &Bee of medium-temperature cells. A detailed account of the du Pont cell will be given as representative the work on this especially in con- nection with nickel anodes has been described in considerable detail by R. C. Downing.23 The original choice of the medium-temperature electrolyte using nickel anodes might be questioned but comparing the position as now known when carbon anodes have successfully replaced nickel there can be 2.Elektrochenz. 1929 35 670. G.P. 493,873 (1930) ; 511,808 (1930). l9 J . Amer. Chem. SOC. 1938 60 96. 21 For a full account of American wartime developments see various papers ibid. 1947 34 (March). 2 2 German work in the fluorine field is fully described in B.I.O.S. 1595 ; see also H. R. Neurnark Trans. Electrochem. SOC. 1946 91 proprint 3 ; and B.I.O.S. 72 ; 261 ; C.I.O.S. XXII-17; 22/XXIX-l4; F.I.A.T. 838; P.B. 6641. aa P.B. 60796. 2o I n d . Eng. Chem. 1942 34 443. LEECH ; PRODUCTION OF FLUORSNE 27 little doubt that this gives the most economical arrangement. The corrosion is very much less than in high-temperature electrolyte mild steel may be employed for cell construction the control of a temperature of about 100" is much simpler than one about 250" and the anodic polarisation tendency is less pronounced.Fluorine manufacture was started with large numbers of nickel anode cells but the current efficiency was low about 70% the nickel consumption heavy and electrolyte conditioning to remove nickel fluorides was a very big job. Downing states that the sludge contains NiP,,KF. A fluoride of tervalent nickel has not been previously reported. The cells were all finally converted to carbon anodes. The cells had a rectangular steel container about 4 ft. long 2 ft. deep 13 ft. wide provided with a jacket for circulating w-ater to maintain optimum temperature con- ditions.The steel cover carried electrodes and diaphragms in rows along the length. There were two rows of anodes built up each of six carbon plates side by side. Each plate was 18 ins. long 6:; ins. wide 1; ins. thick. To each row of anodes there was a double row of sheet-steel cathodes and between the facing electrodes was a diaphragm solid metal in the gas space perforated monel metal sheet .in the electrolyte. The anodes were bolted to copper carriers and these supported by copper rods which passed through stuffing boxes in the cover. The stuffing box was made gm-tight and electrical insulation provided by rings of '' Teflon " the solid polymer of tetrafluoroethylene which was itself a wartime discovery. 24 This is a hard horny material inert to all reagents including even fluorine under ordinary conditions electrically non-conducting and solid up to a t least 300" The cathode was similarly supported and insulated.The cover was provided with gas-outlet connections and a pipe for hydrogen fluoride addition provided. This cell took about 1600 amps. a t a voltage of 5.5 a t the start gradually deteriorating to 11.0. The working temperature was in the range 95-115" the current density was 0.5 anip./sq. in. the -current efficiency more than 9On/, the anode gas contained 5-15% of hydrogen fluoride and on a hydrogen fluoride-free basis the fluorine content was greater than 95% with a trace of fluorine oxide F,O. If it had not been found possible to use carbon anodes the very hoavy consuniption of nickel and expensive electrolyte regeneration operations would probably have made the high-temperature cell with a graphite anode more economic.The work of Predenhagen and Krefft had indicated clearly the controlling factors although they had suggested a composition between KF,HF and KF,l.GHF when graphite could be employed and a temperature a t the start of 160". With this camposition a t this tem- perature the hydrogen fluoride vapour pressure is high and the generally favourable conditions (low corrosion ease of control) associated with a lower temperature and higher HP proportion (up to KF,2*2HF) are begin- ning to disappear. In this most favourable range it was never found possible to employ graphite but amorphous carbon was satisfactorily used. 24 E. B. Yelton Chem. Eng. 1947 54 No. 3 264; Chem. Met. Eng. 1946 53 No. 4 145. 28 QUARTERLY REVIEWS The cells still showed a tendency to polarisation which has several possible origins as follows.(I) The electrical contact to the anode could deteriorate and indeed it has been suggested that the provision of a permanently satisfactory connection is one of the remaining outstanding problems. The gradual increase of the voltage from 8.5 a t the start to 11.0 or more was largely caused by this deterioration. Each cell W ~ S provided with a subsidiary nickel anode which carried all the current a t the start and the anode gas was mainly oxygen. When the anode gas had reached a certain fluorine content the carbon anodes were switched in. Accumulation of water (from the make-up hydrogen fluoride) might occur and some use of the nickel anode might be necessary again during the course of operation.Downing says that there was some other un- explained effect of this electrolysis at a nickel anode in addition to its effect in removing water. (3) Low hydrogen fluoride concentration could sometimes cause polarisa- tion. It is suggested that 37.5% of hydrogen fluoride equivalent to KF,1.75HF was a limiting figure and that in this case the voltage rise occurred a t the cathode. Normally the bulk of the voltage drop occurs a t the anode. This difficulty was immediately cured by addition of acid the preferred range of acid content being 38-40%. (4) Polarisation could still occur not attributable to any of the above effects it was called " gas film " polarisation and attributed to non- wetting of the carbon anodes by the electrolyte. Pinkston states that the anodes were not wetted by the electrolyte and when removed from the liquid they wore clean and black.This is a repetition of the observation by Cady,20 who further stated that such anodes were not wetted by tap water but carbon cathodes in the same electrolyte did not become non- wetting. It was found that this difficulty was minimised by the addition of 1-2y0 of lithium fluoride to the electrolyte; this addition was made in the first place in an attempt to lower $he melting point of the electrolyte and the Harshaw workers observed this other effect of the addition. Schumb on the other hand states that the lithium fluoride was insoluble and tended to settle out but if the mud a t the bottom of the cell was stirred up there was a temporary improvement of the cell performance as regards polarisa- tion tendency.Aluminium fluoride had a similar effect and equally tended to settle out. The addition of 142% of lithium fluoride became standard practice in the American industrial installations and has been the subject of a ~ a t e n t . ~ 5 Downing states that the two discoveries which eliminated polarisation as an obstacle to the commercial operation of carbon anode cells were the addition of lithium fluoride to the electrolyte and the conditioning of the electrolyte by operating with a nickel anode. The carbon anodes used in these cells were of a special character. The du Pont cell had a carbon especially strengthened by impregnation with carbonaceous material and z 5 W. C. Schumb and A. J. Stevens U.S.P. 2,422,590 (1943). (2) Water in the electrolyte exerted its known effect.LEECH PRODUCTION OF FLUORME 29 rebaking and the Harshaw cell had a carbon rod impregnated with copper the latter was stronger permitted better electrical connection being made and was not so sensitive to acid concentration. It may be said that all this American work solved very satisfactorily the problem of making fluorine on a large scale but left unsolved the interesting theoretical problem of the behaviour of carbon anodes. In low-temperature electrolyte graphite almost immediately on being rendered anodic swells and disintegrates a behaviour which persists until the electro- lyte composition reaches that of medium-temperature electrolyte ; even then polarisation iec so severe as to render graphite unusable. Graphite anodes may be employed in high-temperature electrolyte but even then there is a gradual deterioration.Amorphous carbon completely polarised if not disintegrated in low-temperature electrolyte may be satisfactorily employed (but with some limitations) in medium- and in high-temperature electrolyte. Precisely what is the relation to these phenomena of the behaviour attributed to water in the electrolyte and its own intense polarising influence is un- explained. Would polarisation (and disintegration ?) disappear if rigidly water-free systems were employed ? How does oxygen evolution cause such intense polarisation ? A further unusual phenomenon is the method of evolution of the fluorine itself which A. J. nudge W. N. Howell and H. Hill 26 have related to the high angle of wetting a t the anode but which requires further study and which may cast an interesting light on the puzzling “ anode effect ” observed in fusion electrolyses.An interesting feature of the Pennsylvania Salt Mfg. Co.’s cell described by J. F. Gall and H. C. Miller 27 and by R. W. Porter 28 was that no dia- phragm was interposed between tho electrodes only a gas barrier immersed to a depth of 2 in. to divide anode from cathode gas. The high-temperature cell described by E’owler seems a very practical equipment but requires monel construction instead of steel and corrosion was appreciable in places polarisation skems to have been a recurrent phenomenon and the graphite anodes had a restricted life. Regeneration of the electrolyte with hydrogen fluoride vapour was possible and the cell worked a t an anodic current density of 300 amps.per sq. ft.-four times that of medium-temperature cells-so that a more compact installation was possible. German Wartime Devellspments.-Two German cells developed during the war are worthy of attention. I.G. Farbenindustrie developed for experimental purposes only a cell of medium-temperature type at their research laboratories a t Leverkusen. They had developed a 250-amp. cell by 1940 and a 2000-amp. cell by 1942. The construction was in magnesium alloy with hard-carbon anodes impregnated and re-baked and magnesium sheet cathodes. There was no diaphragm only a gas-separating barrier. The 2000-amp. cell was long and narrow (10 ft. long 1-6 ft. wide and deep) and had external electrolyte circulation and cooling. The electrolyte con- tained about 46% of hydrogen fluoride (KFY2.5HP) the working temperature was 75-85” and in these conditions even when hydrogen fluoride containing 2* Nature 1947 160 604.27 Ind. Eng. Chem. 1947 39 262. Chem. Eng. 1948 55 No. 4 102. 30 QUARTERLY REVIEWS 2-3% of water was used for make-up no trouble was experienced with polarisation. A fluorine plant was erected a t Falkenhagen near Berlin to make 720 tons per annum of fluorine for conversion into chlorine trifluoride. This plant using high-temperature electrolyte was apparently under the direct control of the German High Command. There had obviously been a good deal of study of fluorine production on which the design of the cell was based but it is not known what the antecedents of this plant were. It is indeed noteworthy that the I.G. cell was not developed on a manu- facturing scale for this plant.The O.K.H. cell (Ober Kommando des Heeres) had a capacity of 2000-2500 amps. and was made in magnesium alloy with graphite anodes and silver cathodes ; each electrode was sealed into a diaphragm bell of magnesium alloy slotted in the area where the electrodes were facing each other. Each cell stood on its own weighing machine to control the hydrogen fluoride input and was provided with its own rectifier for D.C. supply. Grezt stress was laid on the purity of the electrolyte and of the make-up hydrogen fluoride ; water and sulphur oxides must be completely removed but some hydrofluorosilicic acid (up to 3% in the electrolyte) was permissible. The electrolyte was purified by pre-electrolysis before being charged into the cell whieh had then to be operated for an hour a t 30-60 volts t o condition the anodes before it was installed in its normal working position.The acid for cell make-up was piirificd by passage with 1-274 of fluorine through a nickel reactor at 300". This acid after condensation was re-vaporised for addition to the cell the addition being controlled by the weight of the cell and a catharometric record of the hydrogen fluoride in the cathodic hydrogen. The cell operated between 2-15" and 253" the starting voltage was 6.0-6-5 which gradnalIy rose over r2 period of 3 months to 11.6 a heavy load of up to 5800 amps. was then app!ied for 5-10 minutes when normal operation wm recommenced the voltage retwning to the initial figure. The anodic current density mas GO-7Q amps. per sq. ft. which n'as very similar to American medium-temperature pracficc and very different from the very high values a t which Fowler's cell ran.A current efficiency greater than 90% was obtained. Some cmrosion of the silver cathode and the magnesium parts occurred so that af+er a year or so the electrolyte was emptied from the cell to remove accumulated sludge. The total li€e of the anode is stated to have been 12 months but the reason for the termination of anode life is not given. A good deal of interesting work appears to have been done in connection with this cell development and some of it has been reported by Neuinark. The absorption of fluorine on graphite and the formation of complex inter- calation compounds is described and i t is suggested that this accounts for the gradual deterioration of the cell voltage. The fluorides of rubidium and caium were also studied as possible alternatives to potassium and the conclusion reached that no advantage was to be observed.Kwnsnik on the other hand states that they would be preferred in the I.G. cell were it not for their high price and scarcity. This seems to require further examination. LEECH PRODUCTION O F FLUORINE 31 Chemical Production of Fluorine.-Perhaps a word should be said with regard to the attempted chemical production of fluorine. This was of course studied from Davy onwards and he was well aware of the difficulty he would have in finding vessels which would withstand fluorine a t the high temperatures presumably necessary. Platinum and gold are not particularly good in this connection. Brnuner was reported to have been successful in making fluorine in 1882 by thermal decomposition of the tetrafluorides of lead and cerium but in 1916 0.Ruff 29 cast doubt on this and B. Brauner 30 agreed with him. The studies which produced the Mathers cell also embraced chemical preparations and G. L. Clark 31 describes methods of preparing complex salts such as 3KE',HF,PbP and Na,PbF the former loses hydrogen fluoride a t 250" and evolves fluorine at 300° and the latter may decompose a8t 250". The studies of M. Jellinek and A. Rudat 32 on the reducibility of fluorides by hydrogen are interesting and show very high heats of formation e.g. for FeF,. A patent for the production of fluorine by action of oxygen on the oxyfluorides of titanium zirconium or hafnium was taken out in 1926.33 The difluoride of silver AgP, was first reported by M.8. Ebert E. 1,. Rodowskas and J. C. W. Frazor in 1933,34 and it is suggested,35 without confirmation that thc equilibrium pressure of fluorine over this salt reaches 1 atmosphere a t about 450". It is probable that with increasing knowledge of materials resistant to fluorine it would be possible to find a fluoride which would evolve fluorine on heating and this may be a mettiis of generat- ing very pure fluorine. During the war very considcrable use was made in America of fluorine carriers espccially cobalt trifluoride for fluorinating hydrocarbons. In addition AgF, MnF, CeF, PbF, HgF, and others were studied. These higher fluorides were made by treatment of lower- valency compounds with electrolytically produced fluorinc aiid i t is probable that this method of applying fluorine ma'y undcrgo further development.Handling of Flnor~e.-Consiciernble confusion anrJ mi.;apprehcnsion have been prevalent with regard t o the dificulCies of wxking wit,h fluorine. It reacts fairly rcsdiiy with many organic m3,terials such as rubber aiid textiles and with moisture the hydrogen fluoride present in gaa direct from the cell reacts of course with glass ; but after romsval of this hydrogen fluoride fluorine can be handled readily in glass harder glasses being more resistan5 and silica excellent.36 Tlrus pure fluorine free from hydrogen fluoride out of contact with rubber is quite inert in glass or metal equipment. For connections especially in metal pressure or force joints as commonly used in domestic plumbing and used by Henne l9 for making connection to his cell are very useful.For stopcock lubricants Dry metals also are resistant. 29 2. anorg. Chem. 1916 98 27. a1 J . Amer. Chem. SOC. 1919 41 1477. s a 2. canorg. Chem. 1925 175 281. 33 N. V. Phillips Gloeilampenf~brikien B.P. 362,918 (1926). 3 p J . Amer. Chem. SOC. 1933 55 3056. 35 0. Ruff and M. Giese 2. anorg. Chem. 1934 219 143. 36 S . Aoyama and A. Kanda Bull. Chem. SOC. Japan 1937 12 409. Ibid. p. 38. 32 QUARTERLY RE VIEWS material of the fluorocarbon type is desirable and this a t present is not available commer cia11 y . For removing hydrogen fluoride the cooling possible with solid carbon dioxide is inadequate for the vapour pressure is still appreciable at -60". If cooling is to be resorted to liquid air must be used. A better method is the use of sodium or potassium fluoride as absorbents as first used by Moissan.The former is preferred as it does not give any liquid products which may cause blockages in the apparatus. These have been generally used by fluorine workers and the sodium fluoride is preferably used in a granular or pelleted form. Data on the effect of cooling and of sodium fluoride absorption in removing hydrogen fluoride have been given by the du Pont workers.21 Porter describes how sodium fluoride absorbers are regenerated in situ by heating. At the O.K.H. plant in Germany potassium fluoride was used and liquid products drained off. Even a t temperatures up to 100" at least mild steel copper nickel monel aluminium magnesium and brass are practically completely inert to fluorine and there is no need to use platinum. Atl higher temperatures platinum is not very good nickel or monel is best being useful even a t 600°.21 37 Soft metals lead and tin are not resistant and soldering cannot be used.Amongst non-metallic materials sintered alumina now becoming commercially available is resistant up to a t least 700°,38 and there is no need to go to the specially made fluorspar or fluoride porcelains described by Damiens 39 and by 0. Ruff and A. Riebeth.40 Fluorine compression is not very easy mechanically although it has been done up to 40 Ibs./sq. in.21 For higher pressures the gas is liquefied by cooling with liquid air or nitrogen and then allowed to evaporate under pressure into a suitable container. H. von Wartenberg 41 has referred to the use of a cylinder containing fluorine a t 2000 lbs./sq. in. supplied to him by I.G.Farbenindustrie before the war and I.G. practice has been described.22 Gas compressed to 400 lbs./sq. in. was apparently used in America during the war but only under stringent precautions. A trace of grease or other organic matter could react violently and cause such a local high temperature as to set the metal of the valve burning leakage through the packing of the gland could have the same effect. Containers of compressed fluorine were therefore stored behind heavy steel or thick brick walls and valves manipulated by remote control. In the light of this it seems improbable that compressed fluorine is ever likely to become an article of commerce. In the liquefaction and re-evaporation of fluorine Ruff and 0. Bret- schneider 42 pointed out that explosive compounds of an unknown char- acter possibly C1OF or 0, accumulate in the less volatile residues.Both I.G. Farbenindustrie and Penn. Salt follow Ruff's advice and heat the fluorine to 300° to decompose these impurities before liquefying it. W. R. Myers and W. B. delong Chem. E n g . Progress 1948 44 359. 3* 0. Hannebohn and W. Klemm 2. anorg. Chem. 1936 229 336. Compt. rend. 1931 192 1235 ; A. Damiens and L. Domange ibid. p. 1711. 4o 2. anorg. Chem. 1928 173 373. 4l Ibid. 1939 243 406 ; 1941 247 135. 4 a Ibid. 1934 217 1. LEECH PRODUCTION OF E'LUORTNE 33 Anhydrous Hydrogen Fluoride.-Moissan's procedure for making his anhydrous liquid has already been described. The thermal decomposition of acid potassium fluoride and the conditions required for production of acid of low moisture content have been studied by K.Fredenhagen and G. Cadenbach ; 43 and although on a small scale fluorine cell workers were content to purchase the acid fluoride and discard it when partly exhausted yet for large-scale manufacture a supply of anhydrous acid was essential for both medium-temperature and high-temperature cells. A demand for anhydrous hydrogen fluoride (" AHF ") in industry first developed about 1930 in the U.S.A for the manufacture of the chlorofluoro- methanes the "Freons ".44 About 10 years later a further demand developed also in the U.S.A. for the use of this acid as alkylation catalyst for producing aviation petrol in the petroleum industry. The American wartime fluorine industry thus found the plant in existence to supply its raw materials. Production of some 25,000 tons of AHF was estimated in 1944.I n Germany a special plant was built in Bavaria to supply the O.K.H. plant although I.G. also had an AHF plant at Leverkusen. The method of manufacture is by the action of concentrated sulphuric acid on fluorspar at a temperature of about 250". The fluorspar should be high grade may have been subjected to a flotation process to remove impurities should be low in silica galena and moisture and have a minimum of 98-99Yo CaF content. It is finely ground dried mixed with acid and fed into the reaction vessel. Penn. Salt in their plant a t Cornwell Heights use a rotary kiln externally gas fired and the acid is absorbed in weaker aqueous solution to give an 80% solution which is fractionally distilled to give an anhydrous acid and a weaker solution returned to the absorption system.The plant a t Stulln Bavaria with a capacity of 3000 tons of AHF per annum uses a stationary retort the acid-spar mixture being forced through by slow paddle agitators. The acid is con- densed by cooling t.0 give a product with 10% of water which is batch- distilled through fractionating columns to give a pure anhydroun acid. The I.G. plant with a capacity of 10,000 tons per annum used rotary kilns of steel plate brick-lined the acid is condensed by cooling and the crude product contains 5% of water but a purer product is obtainable by separately collecting the condensate from the final condenser in the series. Some Recent Developments in Fluorine Compounds.-Finally i t is of interest to draw attention to developments in the field of fluorine com- pounds more particularly those involving elementary fluorine in their production.Hence we will not refer further to developments in recent years of boron trifluoride and fluoroborates of the chlorofluoro-derivatives of methane and ethane such as CF,Cl, which have become known generally under their trade name of " Freons " or to the polymers of tetrafluoro- ethylene and chlorotrifluoroethylene. The most outstanding developments of recent years have been in the field of the fluorocarbons. Carbon tetrafluoride was very incorrectly 4 4 J. R. Callahan Chem. Met. Eng. 1945 52 94. Ibid. 1929 178 289. U 34 QUARTERLY REVIEWS described by Moissan 45 and only properly identified by Lebeau and Damiens in 1926.46 It is a gas liquefied a t atmospheric pressure a t - 128' stable and chemically inert unattacked by most reagents.Ruff and R. Keim 47 in 1930 and Ruff and Bretschneider in 1934 42 described a number of fluorocarbons formed by the action of fluorine on carbon up to C5FE2 and J. H. Simons and L. P. Block 48 went up to C7F14. These compounds analogues of the hydrocarbons are all stable and inert and above the butane derivatives actually have boiling points lower than the corresponding hydrocarbon. In addition a t temperatures below about 400° Ruff and Bretschneider showed that different forms of carbon give rise to a solid monofluoride (CF), apparently formed by intercalation of fluorine molecules betn-con the la'ycr planes of the graphite crystallites. It is a chemically inert compound which howcver decomposes a8bove 400" into carbon tetra- fluoride and higher fluorocarbons ; in presencc of fluorine particularly this decomposition occurs explosively.The American wartime fluorine cell developments were largely for the production of fluorocarbons. Two types of compound were required a volatile liquid and a lubricating material ; for the former perfluorodimethyl- cycbhexaiie C6Flo(CF,) was made for the latter various types of hydro- carbon or chlorohydrocarbon were fluorinated. A very wide range of fluoro-compounds was made both of aliphatic and polycyclic character mostly saturated. McBee 21 has described the production of perfluoro- benzene c6F6 but says very little regarding its properties. It is a liquid b.p. 81-82". The properties of these compounds have been described by Fowler and by Cady and Grosse.21 A variety of methods of preparation was employed.The direct fluorination of hydrocarbons with fluorine over a silver-copper catalyst is described by Cady but in general the process used a fluorine carrier of the type of cobalt trifluoride. In many cases hydrocarbon materials were chlorinated and the chlorine replaced with fluorine by treatment with hydrogen fluoride or other agent before com- pletion of the substitution and saturation in the above way. Sulphur hexafluoride was made in 1900 by Moissan and Lebeau by burning sulphur in fluorine and can still only be made by this or a related method. It has a very high electrical breakdown capacity and this com- bined with its low boiling point has made it particularly suitable for filling the Van de Graaff high-voltage generat0r.4~ The production of sulphur hexafluoride on a laboratory scale has been described by Schumb,21 and on a technical scale by Porter.28 The initial product of combustion is contaminated with lower fluorides with the highly toxic S2F1,, and with hydrogen fluoride.After an alkaline wash the gases are passed through a heated metal tube to decompose other impurities again alkali-washed before drying and compressing into cylinders. The chlorine fluorides were only discovered some twenty years ago and 45 Compt. rend. 1800 110 951. ** l b i d . 1926 182 1340 ; 1930 191 930. 4 7 2. anorg. Chem. 1930 192 249 49 W. W. Buechner R. J. Van de Graaff A. Sporduto L. R. Macintosh and E. A. 4 8 J . Awaer. Chem. SOC. 1939 61 2962. Burrill Rev. Sci. Instr. 1947 18 754. LEECH PRODUCTION OF FLUORINE 35 during the war the trifluoride especially underwent some development.Ruff et aZ.50 in 1928 reported that the action of fluorine on hydrogen chloride a t liquid-air temperature led to the formation of chlorine monofluoride boiling a t - 103". K. Freclenhagen and 0. Krefft 51 showed that an equimolecular mixture of fluorine and chlorine exploded on sparking and Ruff 52 stated that fluorine readily burns in chlorine. Then in 1930 Ruff and H. Krug53 showed that fluorine and chlorine mixed in the proportion of 1 3 combined gently when heated to about 300" to give the trifluoride ClF, b.p. 11.3". This is a very reactive material in many reactions simi- lar to fluorine itself. The fluorine made in the O.K.H. cells at Falkenhagcn was entirely converted into chlorine trifluoride which was to be empioyed apparently as an incendiary agent in a manner not clearly indicated.The very violent reaction which chlorine trifluoride shows with water and organic matter would appear to make i t fairly effective for incendiary purposes whilst like fluorine itself it can readily be handled in steel or other metal equip- ment ; and unlike bottled fluorine chlorine trifluoride does not develop a high pressure in containers and there is therefore not the same risk of uiicontrollable escape. In the manufacturing process described fluorine from the cells is mixed with three times its volume of chlorine in stages in the first stage on heating to ZOO" the monofluoride is formed and finally the trifluoride is formed a t 250-280". The reaction vessels are nickel throughout with shell-and-tube heat exchangers cooled with air when the reaction is proceeding normally The chlorine trifluoride is condensed out by cooling and stored in steel vessels of 5 m.capacity. The efficiency of conversion of fluorine into the trifluoride is 95%. Chlorine trifluoride is also stated to be one of the products made from the fluorine from the Pennsylvania Salt Mfg. Co.'s cell already described,28 by a process similar to the German. It is stated that under controlled conditions reaction of chlorine trifluoride with organic compounds leads to the introduction of both fluorine and chlorine into the organic molecule. The possibility of using higher fluorides such as cobalt trifluoride as fluorine carriers had been proposed before the war by Ruff. The American fluorocarbon production led to a very considerable development of this technique and e.g.Fowler has considered the various fluorides which may be used. Cobalt trifluoride was principally used manganese trifluoride and silver difluoride to a less extent the latter largely by McBee and his school. These higher fluorides are relatively stable compounds in themselves made by the action of fluorine at somewhat elevated temperatures (200-400") on lower fluorides or chlorides. They yield up their fluorine in a controlled manner to organic materials in the liquid or the vapour state a t tem- peratures from 150" to 400". These compounds may find many uses for laboratories whose fluorine requirements hardly justify the installation of a fluorine cell and their post-war production has been referred to by Porter.2* 6o 0. Ruff E. Ascher J. Fischer and F. Lanss 2. nnorg. Chem. 1928 1'96,258. 61 2. physikal. Chem. 1929 141 221. 6aZ. angew. Chem. 1929 42 807. 63 2. anorg. Chem. 1930 190 270. Little work has been reported on these compunds.