INORGANIC CHEMISTRY.THIS Report is divided into two sections, the first of which is a review ofrecent publications of general interest. I n the second section an account isgiven of experimental work on the preparation of pure solid elements, basedon a survey af the literature of the last 15 years or so. Some overlap withearlier AnnuaE Reports has inevitably occurred, but this is not serious.The subject has not hitherto been reviewed systematically, and the classi-fication of the various experimental methods, together with illustrativematter drawn from recent literature, will, it is hoped, be of particular valueto the non-specialist.1. GENERAL.During the past year there has again been great interest in the applicationsof isotopes as indicators of reaction mechanism.Methods of preparation ofthe enriched non-radioactive isotope or of the radioactive isotope whichserves as the indicator are now well established, but only for deuterium andits compounds are the necessary materials widely available, In this case acomprehensive series of experiments has been described on the exchange ofdeuterium in deuterated water with ammonia in the complex ammines ofcobalt,l platinum, and palladium., I n the cobalt complexes it was shownthat the hydrogen of the salts [Co(NH,),]C13 and [Co(en),]Cl, undergoesexchange with deuterium when dissolved in water enriched in deuteriumoxide, This fact had been established by earlier observations on complexcompounds of this class,3 but in the present work kinetic measurementshave been made.I n buffered solutions the interchange reaction proceedsaccording to a pseudo-unimolecular rate law, the rate being also inverselyproportional to the hydrogen-ion concentration. The stability of the com-plex is such that the exchange of deuterium and hydrogen cannot occurby an initial dissociation of ammonia molecules, followed by exchange ofammonia in solution. The observations are readily interpreted, however,by assuming that the metal-ammine group, M-NH,, ionises as an acid,forming a metal amide group, M-NH,, and a hydrogen ion. This behaviourparallels that of aquo-ammines, which form hydroxo-ammines by dis-sociation. There is some evidence from these experiments that the easeof acid dissociation runs parallel to the stability of the ammines.The experiments made with the two salts [Pt(NH3),]C1,,2H20 and[Pd(NH3),]C1,,H20 accord in general with the results obtained with thecobalt complexes.Tetramminopalladous chloride was found to undergointerchange by the acid dissociation mechanism referred to above, and also,in acid solution, by reversible dissociation of ammonia from the complex.J. S. Anderson, H. V. A. Briscoe, and N. L. Spoor, J., 1943, 361.J. S. Anderson, H. V. A. Briscoe, L. H. Cobb, and N. L. Spoor, ibid., p. 367.See, e.g., J. Horiuti and G. Okamoto, Sci. Papers Inst. Phys. Chem. Res. Tokyo,1937, 31, 205EMELBUS : QIWERAT,. 61A study of the homogeneous gas exchange with various halides of thechlorine of HW1 in a product containing 3% of H3'Cl and 9776 of H35Clhas also been r e p ~ r t e d .~ Reaction was found to be rapid with phosphorusand arsenic trichlorides, slower for phosphoryl chloride, very slow for silicontetrachloride and sulphur monochloride, and negligible for carbon tetra-chloride. A mechanism involving additive compounds such as HPCl, andH2SiC16 was postulated to account for the occurrence of the exchange.Exchange between the chlorine of hydrogen chloride and solid potassiumchloride was shown to be limited to the surface layers of the solid.Comparative data on the physical properties of deuterium and hydrogencompounds have been published by A. B. Hart and J. R. Partingt~n,~ whohave measured the dissociation pressures of the ammonia and trideutero-ammonia addition complexes of a variety of metallic halides and of coppersulphate.Similar comparative data on the association of hydrogen anddeuterium fluorides have also been obtained.6 The experiments entailedthe isolation of a quantity of HF or DF in a constant-volume container,and simultaneous measurements of pressure and temperature over a consider-able range. The deviations from perfect-gas behaviour were then used tocalculate the association factor of the gas. Degrees of association of thevapour ranging from 1.2 to 4.5 were observed, and the data as a wholeconformed with a single equilibrium 6HF + (HF),, though the existenceof other polymers is not excluded. The heats of polymerisation were- 40,800 and - 41,100 g.-cals. for (HI?), and (DF),, respectively; further,the polymer in each case being assumed to be a single ring, the correspond-ing strengths of the hydrogen and deuterium bonds between fluorine atomswere 6800 and 6850 g.-cals.The use of a radioactive bromine isotope in studying the exchangebetween methyl bromide and certain inorganic bromides has been describedby G.B. Kistiakowsky and J. R. van Wazer.' Radioactive 80Br with a34-hr. half-life was prepared by irradiating bromoform containing a traceof bromine with slow neutrons. The free bromine, which contained a largepart of the bromine radioactivity, was extracted with cold ammoniumhydroxide solution, time being allowed for the decay of short-lived productspresent. Solid ammonium bromide was then isolated and converted intomethyl bromide by heating with sulphuric acid and methyl alcohol.Theexchwge of bromine between the resulting radioactive alkyl bromide andthe bromides of aluminium, barium, and potassium was followed by measur-ing the loss of radioactivity by the gaseous phase. The experiments, whichwere made in the temperature range 22--227", showed that with aluminiumbromide, which is a good catalyst for the reactions of methyl bromide,exchange of bromine between the solid bromide and the vapour occurred4 K. Clusius and H. Haimerl, 2. physikal. Chem., 1942, B, 51, 347; A , 1943, I, 63.6 R. W. Long, J. H. Hildebrand, and W. E. Morrell, J . Amer. Chem. SOC., 1943,J., 1943, 104.66, 182.Ibid., p. 182962 INORGANIC CHEMISTRY.readily, the activation energy of the reaction being only 4.6 kg.-cals.Forbarium bromide, which is a poor catalyst, the activation energy was 12 kg.-cals., whilst for potassium bromide, which is devoid of catalytic activity,there was no measurable exchange. The use of the radioactive indicatorhas thus served t o indicate a relationship between exchange and catalyticactivity. A somewhat similar use of the radioactive iodine isotope 1281of 25 mins. half-life is described by H. A. C. McKayY8 who has studied theexchange of iodine and alkyl iodides in alcoholic solutions.The structure of diborane, which for many years has been a controversialsubject,g has been re-examined by H. 0. Longuet-Higgins and R. P. Bellylowho have revived a formula (I) in which two hydrogen atoms form a bridgebetween the boron atoms.This formula, which is analogous to that of thedimeric aluminium halides (11), has much chemical and physical evidencein its favour. In particular it accounts for the ease of interconversion of(1.1 (11.) (111.)the boron hydrides, in all of which similar bridges are thought to exist.The existence of products derived from the borine radical, such as BH,,COand BH3,NMe3, is also readily explained, as is the non-existence of methyl-substituted diboranes containing less than two hydrogen atoms. Trimethyl-boron itself is monomeric and pentamethyldiborane is unknown. The boro-hydrides n are formulated similarly, beryllium borohydride, for example,being assigned the formula (111). Physical evidence for these structuresand for the nature of the bonds is fully discussed by the authors in termsof the theory of resonance.There is uncertainty in the interpretation ofelectron-diffraction measurements of bond length, but the vibrational andinfra-red spectra of diborane are more consistent with a hydrogen-bridgestructure than with one resembling ethane.The relative strengths of the N+B bonds in the additive compoundsof trimethylamine with boron fluoride and its methyl derivatives have beenstudied by A. B. Burg and (Miss) A. A. Green.12 This work, in the courseof which the new compounds Me,NBF,Me and Me,NBFMe, were prepared,has shown that the substitution of one methyl group for fluorine in Me,NBF,leads to a large decrease in the N+B bond strength, whereas furthersubstitution causes only a small further decrease.Evidence that gallium forms an unstable borohydride under conditionssimilar to those yielding borohydrides of aluminium, beryllium, and lithiumhas now been obtained.13 Trimethylgalliuni, the improved preparation of* J .Amer. Chem. SOC., 1943, 65, 702.9 For recent reviews, see A. B. Burg, Chem. Reviewe, 1942, 31, I; S. H. Bauer,10 J . , 1943, 250.12 J . Amer. Chem. SOC., 1943, 65, 1838.l3 H. I. Schlesinger, H. C. Brown, and G. W. Schaeffer, ibid., p. 1786.ibid., p. 43.11 Cf. Ann. Reports, 1941, 38, 65EMEL$US : GENERAL. 63which by the action of dimethylmercury on gallium has recently beendescribed,l* was treated with an excess of diborane at room temperature.After an induction period a rapid reaction occurred with deposition of afilm of gallium ;.the reaction was represented by the eqllationIt was considered probable from these preliminary experiments that aborohydride of gallium was the first reaction product and that it underwenta rapid autocatalytic decomposition. When the reaction was carried outa t -45" a new compound, dimethylgallium borohydride, (CH,),GaBH4, wasobtained. This substance, which had an extrapolated boiling point of 92",underwent slow decomposition at room temperature. Its reactions arebeing further studied.A non-volatile aluminium hydride, (AlH& has been prepared by0. Stecher and E. Wiberg l5 by passing trimethylaluminium with a largeexcess of hydrogen through a glow discharge. The volatile products werecomplex, but the authors succeeded in isolating and characterising twocompounds, AI,H, (CH,), and Al,H,(CH,)3.Unlike the corresponding galliumderivative, Al,H,(CH,), did not disproportionate to aluminium hydride andtrimethylaluminium.From the non-volatile reaction products a compound having the formulaAIH,,N(CH,), was isolated by treatment with trimethylamine. This losttrimethylamine as it was heated, giving a series of intermediates of the type(AlH,),,N(CH,), and (AlH3)5,N(CH3)3. At 100-135" a product was leftwhich had the empirical formula, AlH,. This was a white, non-volatilesolid, the chemistry of which has not yet been studied.Evidence for the existence of an unstable volatile lower fluoride ofaluminium, (AlF),, has been obtained by W.Klemm and E. Voss.l6 Thiswork arose from the observation 17 that aluminium volatilises at a lowertemperature than usual in the presence of a metallic fluoride. Klemm andVoss showed that when aluminium was heated with aluminium fluoride a ttemperatures between 600" and 1000" and a t pressures of a few mm. asublimate of aluminium and aluminium fluoride was formed. In this tem-perature range the metal is not appreciably volatile, though its fluoridevolatilises to a small extent. Repeated revolatilisation of the sublimatewith excess of aluminium finaily gave a product in which the ratio A1 : Fwas 1 : 1. This result was interpreted by supposing that a volatile fluoride(AlF),, which disproportionates into metal and the trifluoride on condens-ation, is responsible for the transport of metal through the gas phase.The preparation of fluoro-derivatives of non-metallic elements by theSwarts reaction has been further studied by H.S. Booth and his co-workers.l*Fluorination of thiophosphoryl tribromide by antimony trifluoride at 60-70"without a catalyst has yielded the bromofluorides PSF,Br (b. p. 35.6") andl4 E. Wiberg, T. Johannsen, and 0. Stecher, 2. anorg. Chem., 1943, %l, 114.l5 Ber., 1942, 75, 2003.17 C. B. Willmore, U.S. Patent 2,184,706, 1939.16 2. anorg. Chem., 1943, 251, 233.18 Cf. Ann. Reports, 1941,38,16064 INORGANIC CHEMISTRY.PSFBr, (b. p. 125.3").19 These two compounds are noteworthy because oftheir high resistance to hydrolysis : PSF,Br, which was the more resistant,was almost unattacked in 24 hours by aqueous potassium hydroxide a t roomtemperature, though reaction was rapid a t 100".It is of interest that thepartly fluorinated derivatives of phosphorus tribromide are very much lessstable.Attempts to fluorinate sulphur monochloride by the above methodresulted only in decomposition. Boron trichloride has also been examined,and although reaction was carried out at reduced pressure so as to facilitatethe escape from the reaction mixture of partly fluorinated products, onlyboron trifluoride was obtained.20 Reaction temperatures down to - 78"were employed. In addition, the interaction of boron trifluoride and borontrichloride a t 500", fluorination of boron trichloride with calcium fluoride a t160", and the use of the less reactive SbC1,F in place of SbF, were examined,but in each case the result was the same.The preparation of germanium isocyanate has been reported by A.W.Laubengayer and L. ReggeL21 The method of preparation was similar tothat used in preparing isocyanates of silicon, boron, and phosphorus.22Germanium tetrachloride was dissolved in benzene, silver isocyanate added,and the mixture refluxed. The formula of the compound, which had anextrapolated boiling point of 196", was established by analysis, and it wasfound to be rapidly hydrolysed by water and to undergo thermal decom-position a t temperatures above 140".Chlorine azide, which ranks among the most unstable compounds known,has recently been reinvestigated and more fully characterised.,, The pre-ferred method of preparation was by the gradual addition of acetic acid toequimolecular amounts of sodium azide and sodium hypochlorite, followedby distillation in a stream of air or nitrogen a t atmospheric pressure.Thoughthere were many detonations in the course of this work, the compound wasisolated and analysed, and a number of its properties were studied. Liquidchlorine azide did not conduct electricity appreciably, nor did it conductin contact with sodium azide. (Sodium azide dissolved in liquid hydrogenazide conducts readily.24) The compound was soluble in ten organic solventsexamined and thus behaved neither as an ionising solvent nor as a polarcompound. When carried by a stream of Fitrogen into an excess of liquidammonia it reacted quantitatively according to the equation3NR,ClChlorine azide reacted with pentane, forming hydrogen azide and mono-chloropentane, and with metals both a chloride and an azide were produced.I* H.S. Booth and C. A. Seabright, J. Amer. Chem. SOC., 1943, 65, 1834.20 H. S. Booth and C. G. Frary, ibid., p. 1836.21 Ibid., p. 1783.22 G. S. Forbes and H. H. Anderson, ibid., 1940, 62, 761.2s W. J. Frierson, J. Kronrad, and A. W. Browne, ibid., 1943, 66, 1696.24 A. W. Browne and G. E. F. Lundell, ibid., 1909, 31, 435.The term isocyanate is used here for a product which may be acyanate, an isocyanate, or a mixtureEMEL~~US : GENERAL. 65A further series of experimentsZ5 on the interaction of silver azide andchlorine azide in ethyl ether gave a deep blue solid compound of the formulaN,AgCl, which was stable below - 30" but decomposed at higher temper-atures into silver chloride and nitrogen.Cupric azide, which hitherto was believed to be a particularly unstablesubstance, has been re-examined in the course of the past year.2s Thesubstance previously described under this name was found to bk a basicazide, two definite compounds of this class of the formulae Cu(OH)N, andCu(N,),,Cu(OH), being prepared.The pure azide was obtained by theinteraction of cupric nitrate and sodium azide in aqueous solution. Theflocculent precipitate was freed from basic salts by treatment with dilutehydrazoic acid. The pure salt, which when dry was only moderatelysensitive to shock, was also prepared by heating the ammoniate describedby L.M. Dennis.27 Various 4-co-ordinated complex derivatives were alsoprepared, including the tetrammino- complex [ Cu ( NH3)4] (N,) ,, analogousderivatives containing amines and diamines, and also complexes of thetype [Cuxx(NH,),(N,),] which were shown to be, non-electrolytes. Thecomplexes containing ammonia were considerably less sensitive to deton-ation by shock than was the parent azide, whereas the amine complexescould not be detonated, though they burned readily when heated.The study of metallic carbonyls by W. Hieber and his co-workers hasbeen extended to include carbonyls of rhodium and osmium.28 A series ofrhodium carbonyl halides of the type Rh(CO),X, where X = C1, Br, or I,has been prepared by the action of carbon monoxide on the trihalide a tordinary or elevated temperatures, according to the halide ~ s e d .~ g Thesecompounds were volatile and crystalline, and the molecular weight of thechloride corresponded to a dimeric formula, represented in all probabilityby the structure (OC),Rh<Cl>Rh(CO),. c1Rhodium metal, when heated in carbon monoxide at 200"/280 atm.,gave a carbonyl of the formula [Rh(CO),],, which formed orange-yellowcrystals, m. p. 76". This compound resembles cobalt tetracarbonyl, andalso illustrates a point of difference between rhodium and iridium, foriridium carbonyls are not formed directly from the metal.Rhodium halides, when heated in carbon monoxide at 50--80"/200 atm.in the presence of cadmium, zinc, or silver to act as a haIogen acceptor,gave a carbonyl of the formula [Rh(CO),],.If the preparation was carriedout similarly but a t temperatures between 80" and 230°, the product was anew carbonyl of the formula Rh,(CO),,, which was characterised by com-paratively high stability. It was not, for example, attacked by dilute acidsor alkalis, and its solubility in organic solvents was low.25 W. J. Frierson and A. W. Browne, J . Amer. Chem. Soc., 1943, 65, 1698.26 A. Cirulis and M. Straumanis, 2. anorg. Chem., 1943, 251, 332, 335, 341.27 J . Arner. Chem. SOC., 1907, 29, 18.28 Cf. Ann. Reports, 1941, 38, 71 ; 1942, 39, 72.29 W. Hieber and H. Lagally, 2. anorg. Chem., 1943, 251, 96.REP.-VOL. XL. 66 INORGANIC CHEMISTRY.Rhodium carbonyl hydride, Rh( CO),H, was prepared by heating rhodiumin a mixture of hydrogen and carbon monoxide, the former being a t 50 andthe latter at 200 atm.This preparation is analogous to that of cobaltcarbonyl h~dride.~O The best method of preparation was by tho autoclavereaction of hydrated rhodium trichloride and carbon monoxide at 2OO0/2OOatm. The compound had a melting point of - 10" to - 12", and aboveits melting point lost hydrogen readily, forming the tetracarbonyl.Osmium carbonyl halides and carbonyls have now been fully described.31The former have been obtained from osmium halides by the usual high.pressure reaction with carbon monoxide, and are of several types. Inaddition to the characteristic tetracarbonyl halides, Os(CO),X2, corre-sponding with the iron compounds, compounds of the type OS(CO)~X, andOs(CO),X,, where X = C1, Br, or I , were prepared.In addition, iodidesand bromides of the type Os(CO),X were obtained, the molecular weight ofthe latter corresponding with the double formula (OC),Os<x>Os(CO),. XFor the preparation of pure osmium carbonyls two methods were avail-able. The first, which gave carbonyl halides as well as carbonyls, was bythe action of carbon monoxide on osmium halides in presence of a secondmetal to act as halogen acceptor. The second was by the action of carbonmonoxide on osmium tetroxide, and was similar to the preparation ofrhenium carbonyl from its hept~xide.~, The pentacarbonyl Os(CO), wasisolated from the mixture of products formed in the reaction betweenosmium halides and high-pressure carbon monoxide in presence of a metallicpowder. The oxyiodide, formed from the tetroxide and hydriodic acid,gave the best results.This carbonyl was remarkable for the readiness withwhich it lost carbon monoxide to form OS,(CO)~; indeed, the latter com-pound was the main product of the reaction, and was also formed in quan-tity by the interaction of osmium tetroxide and carbon monoxide a t150°/200 atm. Strong evidence for the existence of osmium carbonylhydride, Os(CO),H,, was also obtained, though this substance has not yetbeen fully characterised.Copper carbonyl, the existence of which has for some time been suspected,has now been prepared by the action of carbon monoxide on heated cuprousoxide.33 It is described as a white, readily sublimable solid, the vapour ofwhich is dissociated at a higher temperature with deposition of copper.The empirical formula Cu(CO), is assigned from preliminary analyses.Inthe same communication the formation of small yields of tellurium oarbonylby the action of carbon monoxide on tellurium is reported, though noneof the properties of this substance has so far been described.Reactions in liquid sulphur dioxide have already been described inthese Reports.34 Further work on the amphoteric behaviour of sulphites80 Cf. W. Hieber et al., 2. anorg. Chem., 1939, 240, 261 ; 243, 145, 156.31 W. Hieber and H. Stahan, Ber., 1942,75, 1472 ; 2. Elektrochem., 1943,49,288.s2 W. Hieber and H. Fuchs, 2. anorg.Chem., 1941, $248, 266.94 Ann. Reports, 1939, 36, 136.P. L. Robinson and K. R. Stainthorpe, Nature, 1944,158, 24EMELLUS : QENERAL. 67in this solvent has now been reported.35 Aluminium chloride has beenfound to react in liquid sulphur dioxide solution with tetramethylammoniumsulphite according to the equations :The second equation represents the redissolution of the sulphite, which isf i s t precipitated, by an excess of the soluble tetramethylammonium sulphite,the latter behaving as a base and being analogous to hydroxides in aqueoussystems or to amides in liquid ammonia. The course of the reaction inliquid sulphur dioxide is readily followed by a conductimetric titration.The extension of this observation on amphoteric behaviour is made moredifficult by the low solubility of many halides in sulphur dioxide.Conducti-metric titration of stannic chloride with tetramethylammonium sulphiteshows, however, that it behaves like aluminium chloride : a sulphite, orwhat is more probably a compound of the type Sn0,,sS02 analogous to anoxyhydrate in aqueous systems, is first precipitated and then redissolves inexcess of the precipitant, forming [ (CH,),N],(Sn( SO,),) (tetramethyl-ammonium orthosulphitostannate), or the corresponding meta-compound[ (CH,)4N],{Sn(S0,),}. Ageing of the precipitated sulphite occurs rapidlyand renders it incompletely soluble. Normally, therefore, in studying thesereactions excess of the aulphite is added and the excess is determined byback titration with thionyl chloride, which behaves as an acid in sulphurdioxide.Silicon tetrachloride and boron trichloride behaved similarly, thoughthere is some doubt as to the composition of the precipitate.With theformer, for example, formation of a true sulphite is very doubtful, and thecomposition is best represented as Si02,xS02. With antimony trichlorideand pentachloride the reaction was again similar. It was found that, whenthe solid precipitated from antimony pentachloride was redissolved in excessof tetramethylammonium sulphite and was treated with excess of thionylchloride, the compound [(CH,),N]SbCI, was formed and could be isolated.The reaction between a solution of tetramethylammonium sulphite andtin illustrates further the analogy between reactions in water, ammonia,and sulphur dioxide.I n the case of the first two solvents the action of an'' alkali " on zinc is represented by the equation :The titrations are made at - 30".Zn + 2KOH + 2H,O = K,[Zn(OH),] + H,Zn + BKNH, + 2NH, = K,[Zn(NH,),] + H,With a solution of sulphite in sulphur dioxide evolution of hydrogen can-not occur, but in its place one would expect sulphur monoxide, since thethionyl radical is the counterpart of the hydrogen ion in water or ammonia.Actually tin was found to dissolve in a solution of tetramethylammonium36 G. Jander and H. Hecht, 2. a w g . Chcm., 1943, ab0, 287, 30468 INORGANIC CHEMISTRY.sulphite in sulphur dioxide, the reactions being represented by theequationsSn -k [(CH,)gN12SO3 + 4so2 = [(CH3).&]2(Sn(SO&) + 2so2so = so2 + sII(CH3)gNI,SO, + s = [(CH,),N12S20,.Analysis of the reaction products showed a ratio of tin dissolved to thio-sulphate formed of 1 : 0.7-0.8, proving that this reaction scheme is sub-stantially correct.H. J. E.2. THE PREPARATION OF THE SOLID ELEMENTS IN A STATE OF PURITY.The preparation of the elements in a state of high purity has engagedconsiderable attention during the past fifteen years, and the Reportersconsider that the subject merits review a t the present stage of its develop-ment. Recent advances have been concerned chiefly with the metals andwith non-metallic elements which are solid under ordinary temperatureand pressure conditions, and it is to these solid elements that discussionis confined in this Report.The subject has been approached from theviewpoint of experimental methods, this approach leading to a better apprecia-tion of modern developments than a discussion of individual elements.The methods treated are designed for use on the laboratory scale, whereeconomic considerations are not paramount, rather than for applicationin industrial practice.An excellent monograph on pure metals,l in which their preparationand properties are authoritatively discussed, has appeared recently, and thisshould be consulted for details and bibliographies relating to earlier methodsof preparing the metallic elements. Two reviews on the availability and useof high-purity metals 2a may also be cited. In one of these an accountis given of efforts by the American Society for Testing Materials to obtainmetal samples of consistently high and accurately known purity; thefollowing percentage purity values for samples obtained or promised illustratethe high purity standards attained in individual cases by the use of specialmethods of preparation or purification : lead, zinc, and platinum, 99.9999 ;bismuth, 99.9984 or better ; cadmium, 99.999 ; gold, 99.998 ; aluminium,99.997-99-991 ; tin, about 99.995 ; copper, 99.994 or better ; silver, 99.983or better ; nickel, 99.97 or better ; magnesium, 99.97.Further improvement8have no doubt been effected since these values were published.The purity values given above and elsewhere in this Report, and in theliterature generally, are in most cases based on determinations (usually byspectrographic methods) of the various impurities present in the samples,since chemical methods for determination of the element concerned ark rarelycapable of establishing its content with the necessary accuracy in an almost1 A.E. van Arkal, " Reine Metalle," Berlin, 1939.T . A. Wright et al., Proc. Amer. SOC. Test. Mat., 1937, 37, I, 531, 538.C. H. Desch, Vortrage Hauptversammlung, 1938, Deut. ffes. Metallk., 1938, 1 ;Met. and Alloys, 1939, 10, No. 4, 204WELCH : PREPARATION OF SOLID ELEMENTS IN A STATE OF PURITY. 69pure sample. This " difference " method of assessing purity must clearlybe used with extreme care, since it presupposes that every impurity has beendetected and determined ; a number of purity estimates made in the literatureare suspect because the absence of conceivable contaminants has not beenfully established, and a need for caution in this respect is evident.The preparation of pure compounds by conventional methods of fractionalcrystallisation, distillation, etc., is outside the scope of this Report, althoughattention may be directed to a recently described technique for manipulatingreadily oxidisable solutions and precipitate^.^ Special methods of separ-ation applied to groups of closely similar elements are of special importance,however, in the preparation of the pure elements themselves, and the mostrecent work in this field is briefly discussed below.Methods of isolating orpurifying the elements are then considered under appropriate sub- headings.Separation of Closely Similar Elements by Chemical Means.-The difficultyof separating scandium from rare earths, thorium, etc., is well known, andan exhaustive experimental study and comparison of the available methodsis of considerable interest.Almost all the previously known methods areconsidered to be unsatisfactory, except fractional condensation of thechloride or sublimation of the acetylacetone complex ; the former givessatisfactory separation from all the common impurities except thorium,and possibly manganese, and the latter permits sharp separation from thorium,zirconium, hafnium, and rare earths, although iron, aluminium, and probablyberyllium, accompany the scandium. Both methods suffer from the dis-advantage that moderate or large quantities of material cannot be ex-peditiously handled, and the yields of purified material are poor.A newand useful method investigated in some detail consists in extracting withether an acid solution of the scandium preparation containing ammoniumthiocyanate ; the scandium is strongly concentrated in the ether phase.A single extraction permits recovery of 94% of the scandium present,separation from magnesium, calcium, rare earths, thorium, and manganesebeing almost complete ; ferrous iron, titanium, zirconium, hafnium, anduranium are largely removed, but beryllium , aluminium, indium, molyb-denum, rhenium, ferric iron, and cobalt may accompany the extractedscandium. Application of the method to a large quantity of 75--80~0pure scandium oxide resulted in extraction of 90% of the contained scandiumin a spectroscopically pure form.A recent method for separation of the rare-earth elements, depending ondifferences in the stability of their amalgams, shows considerable promise.When rare-earth acetate solutions are shaken with sodium amalgam,europium, samarium, and ytterbium are rapidly transferred to the amalgamphase ; the other rare-earth metals give amalgams much less readily underthe conditions proposed, their amalgam-forming power diminishing withincreasing atomic number.The method has been successfully applied t oseparation of neodymium-samarium and samarium-gadolinium mixtures,S. Rihl and R. Fricke, 2. anorg. Chem., 1943, 251, 405.W. Fischer and R.Bock, ibid., 1942, 249, 116. J. K. Marsh, J . , 1942, 39870 INORGANIC CHEMISTRY.from which samarium amalgam is rapidly obtained with little contaminationby neodymium or gadolinium.' Further purification is carried out byfractional decomposition of the amalgam with water or dilute acid, thesamarium dissolving preferentially and leaving the other rare-earth metalsin the amalgam. By one application of this two-fold reaction the neodymiumcontent of a mixture with samarium is reduced from 70 to O.O1~o-a separ-ation which is remarkable in comparison with those achieved by the classicalfractionation methods. Separation from gadolinium is considered to beequally rapid. By a very similar process, ytterbium may be isolated frommixtures with lutecium and thulium,* ytterbium preparations containingless than O .O l ~ o of the accompanying rare-earth metals being readilyobtained. By addition of samarium and its subsequent removal as amalgam,small quantities of ytterbium remaining iu lutecium preparations may beremoved, the ytterbium being co-extracted with the samarium; by thismeans lutecium salts containing only 0.001 % of ytterbium are obtainable.Application of the same amalgam procedure to a mixture of gadolinium,samarium, and europium acetates affords a pure gadolinium acetate solutionand an amalgam containing the europium and samarium; these two metalscan be separated subsequently .Q An electrolytic method of amalgam form-ation gives results similar to those just described in that europium, ytterbium,and samarium are concentrated in the mercury phase.lo These new methods,judiciously combined with the older processes of fractionation, have greatlyfacilitated the separation of some of the individual rare-earth metals.Incertain caaes, however, the exclusive use of the fractionation methodsappears essential, and such methods have recently been used to isolateabout 12 g. of holmium oxide containing not more than O.l')'o of erbiumand less than 0.08% of dysprosium and yttrium.llA preliminary investigation of the possibility of separating the rareearths by fractional base-exchange with zeolites has recently been de-scribed.12 Fractionation is obtained if a concentrated solution of rare-earth salts is treated with a quantity of zeolite insufficient to exchangewith all the earths; the metals most strongly held by the zeolite are thosewhioh show the smallest ionic radii in their crystalline compounds.Fractional removal from the zeolite is also possible, the ions of larger radiusbeing removed preferentially. A chromatographic separation prooess hasalso been applied to the rare-earth elements.13The separation of zirconium and hafnium still remains a process in whichmethods of fiactionation are indispensable.A careful study has now beenmade of the optimum conditions for effective separation of these two metalsby fractional precipitation of their ferrocyanides.14 Four successive7 J. I(. Marsh, J., 1942, 523.10 H. N. McCoy and R. P. Hammond, J . Amer. Chem. Soc., 1942,64,1009.11 W.Feit, 2. anorg. Chm., 1940, 243, 276.1) R. G. Ruseell and D. W. Pearce, J . Amer. Chem. SOC., 1943, 65, 1924.13 0. Eriimetrii, T. G. Sahama, and V. Ksnula, Ann. Acad. Sci. Fennicoe, Ser. A,16 W. C. Schumb and F. K. Pittman, Ind. Eng. Chem. (Anal.), 1942,14,512.Idem, ibid., 1943, 8 . 9 Idem, ibid., p. 531.1943, 67, No. 3, 6 WELCH: PREPARATION OF SOLID ELEMENTS IN A STATE OF PURITY. 71fraotionations by the procedure recommended have increased the hafniumoxide content of a mixture of the two oxides from 12 to 20,36, 62, and SO%,respectively. The precipitation of zirconium and hafnium phosphates hasalso been investigated in a similar manner ; 15 phosphate fractionationshave hitherto been rendered difficult in this case by the gelatinous natureof the precipitates, but in the new procedure precipitation of the zirconium-hafnium phosphate in a granular form which is easily filtered off is securedby spraying the sulphate solution [Z-5 yo of (Zr,Hf)OSO,] and phosphoricricid (245%) (both in 10% sulphuric acid) at equivalent rates into a largevolume of lo?; sulphuric acid held at 75".This technique, combined with anew method of reconverting the phosphates into the soluble sulphates for thenext fractionation, has given new scope to a classical separation process.In a typical series of seven precipitations, in each of which about 56% of thedissolved material was converted into phosphate, hafnium oxide was en-riched from 13% in the original oxide mixture to 93%, the hafnium in theconcentrates representing 10% of the total quantity of hafnium in thestarting material.Two successive treatments of material low in hafnium,followed by recrystallisation of the oxychloride, have yielded zirconiumsalts in which hafnium could not be detected by spectrographic examination.Preparation of Elements by Thermal Decomposition of their Compounds.-Thermal decomposition of 'a suitable compound appears t o be the simplestconceivable method of isolating &n element, and although this method hasrelatively few applications to solid elements it has certain interestingpossibilities and deserves brief mention. Pure metals of the platinum groupare customarily prepared by ignition of the purified salts obtained in thecourse of their separation.ls Sodium, potassium, rubidium, and cesiumhave been prepared by pyrolysis of their azides in a high vacuum; 1' thereactions occur at moderate temperatures (275-395"), and after distillation(in the same apparatus) the metals are spectroscopically pure and gas-free.Decomposition methods are particularly applicable to the less volatileelements, which cannot distil out of the heated zone and recombine withother decomposition products.This is well illustrated by C. W. von Bolton'soriginal method of preparing tantalum,l8 in which rods of tantalum dioxidewere heated to a high temperature in a vacuum by passage of an electriocurrent. An interesting recent example of a reaction of the same generaltype occurs in the preparation of pure germanium by decomposition of itsnitride ; l9 treatment of germanium tetrachloride with ammonia affords theimide, Ge(NH),, together with ammonium chloride ; the latter is washedout of the product with liquid ammonia, and the imide heated in nitrogen.A t about 150" germanam, Ge,N3H, is formed, and at 350" this is convertedl6 E.M. Larsen, W. C. Fernelius, and L. L. Quill, Id. Enp. Chem. (Awl.), 1943,15,l6 E. Wichers, R. Gilchriat, and W. H. Swanger, Trans. Amer. Inst. Min. Met. Eng.,512.1928,76, 602; E. Wichere, J. Res. Nat. Bur. Stand., 1933,10, 819.R. Suhrmann and K. Clusius, 2. anorg. Chem., 1926, 152, 52.l 8 2. Elektrochem., 1906,11, 45.le R. Schwarz, Die Chemie, 1042, 66, 4672 INORGANIC CHEMISTRY.into the nitride, Ge,N,; dissociation of the nitride into germanium andnitrogen takes place at 1000".Interesting information on the properties of carbonaceous materialobtained by heating sucrose a t temperatures between 300" and 1100" isgiven in a recent paper.20 Material prepared by heating sucrose in hydrogenat 1000-1100" for 10 hours consists of substantially pure carbon in the formof graphite crystallites about 10 x 30 x 30 A.; specimens prepared at lowertemperatures contain several yo of hydrogen and oxygen." Hot-wire " Metha&.-In " hot-wire " methods the vapour of a volatilecompound of the desired element is thermally decomposed or reduced a t thesurface of a wire heated to a suitable high temperature by passage of anelectric current, the element being deposited on the wire as a more or lesscoherent coating.In the preparation of a metal by this means a thin" starting wire " may be drawn from a previously prepared specimen of themetal itself, the thermal reaction then being employed to build up a homo-geneous rod which may reach a diameter of several millimetres; since thedeposited metal does not come into contact with any extraneous material,contamination is minimised.The apparatus generally used in the hot-wire technique consists of asuitably designed glass or quartz bulb with heavy sealed-in leads whichconduct the heating current for the wire, the latter being supported by theleads in the centre of the bulb. Suitable provision is made for introdu9ingthe reactants into the bulb, and for pumping off any volatile decompositionproducts.The principal experimental difficulty is the maintenance of thewire at a constant temperature; if, as in most cases, the material depositedconducts electricity, the current passing must be continuously increasedas deposition proceeds, sometimes from a fraction of an ampere at thebeginning of the experiment to several hundred amperes at the end.A review of applications of the hot-wire technique, by one of its principalexponents,21 illustfates its versatility. In the majority of the applicationsthe reaction taking place a t the wire is thermal decomposition of a halide ofthe element (frequently the iodide); if a supply of the element in a finely-divided form, prepared by some other method, is placed in the bulb andheated to a suitable temperature, the liberated halogen may react con-tinuously with it, and thus replenish the halide required for decompositiona t the wire.Copper, titanium, zirconium, hafnium, thorium, vanadium,chromium, molybdenum, tungsten, rhenium, iron, and nickel have allbeen prepared by this form of the hot-wire technique, the deposition temper-atures for these elements varying from 600" t o 2000°, and the temperatureof the "reserve" of crude metal ranging from 20" to 800". In otherapplications a mixture of hydrogen with the vapour of a halide is passedover the heated wire, and the halogen hydride liberated in the consequentreduction is pumped off; beryllium, silicon, and vanadium have beenprepared by this method. Niobium, tantalum, and platinum are best20 U.Hofmann and F. Sinkel, 2. anorg. Chern., 1940, 245, 85.21 A. E. van Arkel, Metallw., 1934,13,405,511WELCH: PREPARATION OF SOLID ELEMENTS IN A STATE OF PURITY. 73obtained by decomposing a halide (or carbonyl halide in the case of platinum)at the hot wire, and removing the gaseous products by direct pumping. Inall these methods, essential conditions for deposition of the element are thatthe compound employed should be appreciably dissociated at temperaturesbelow the melting point of the element, and that the vapour pressure of theelement should be considerably less than that of the compound at thetemperature chosen for the heated wire.The most fruitful applications of hot-wire methods have been to elementsof high melting point which are not readily obtained in massive form byfusion, notably to titanium, zirconium,22 hafnium,23 thorium,24 niobiumand tantalum,Z5 rhenium,26 and boron.In several of these examples hot-wire methods afford the only means of preparing the element in a highlyductile form, other methods giving products which are rendered brittle bytraces of oxygen, nitrogen, or other impurities. I n one of the more recentaccounts of hot-wire methods 27 the preparation of ductile titanium isdescribed ; crude titanium, prepared by reduction of the tetrachloride (or,less satisfactorily, potassium or sodium titanifluoride) with metallic sodium,is confined with a little iodine in a bulb heated to 500" or more, the titaniumbeing deposited on a wire heated to 1300".As indicated above, the halogenis continuously re-used to form titanium tetraiodide, in which form the metalis conveyed from the supply of crude material and deposited by decom-position on the wire. The titanium is o.btained in rods up to 7 mm. indiameter, which contain about 0.14% of iron (introduced from the vesselused in preparing the crude metal) and a little silicon, but are otherwise pure.In the original paper the effects of wire temperature and other conditionson the reaction and its products are discussed in some detail.Commercial " pure " boron from various sources has recently been shownto contain less than 80% of the element ; the balance consists of oxygen andaluminium or magnesium, the latter evidently being introduced during thereduction process.28 Special intere.st therefore attaches to a new preparationof genuinely pure boron by the hot-wire method.29 The reaction employedwas the reduction of boron tribromide by hydrogen, rendered particularlysuitable by the ease with which the tribromide can be purified by con-ventional vacuum fractionation methods in all-glass apparatus.Since boroncould not be used in the " starting wire," deposition was carried out onLz A. E. van Arkel and J. H. de Boer, 2. anorg. Chern., 1925, 148, 345; J. H. deBoer and J. D. Fast, ibid., 1926, 153, 1 ; 1930, 187, 177; C. J. Smithells, Metal Ind.(Lond.), 1931, 38, 336.aa J. H. de Boer and J. D. Fast, 2. ar2.org. Chern., 1930,187, 193.24 Ref. (I), p. 215.25 W. G. Burgers and J. C. M.Basart, 2. anorg. Chern., 1934, 216, 223; K. Moers,a6 C. Agte, H. Alterthum, K. Becker, G. Heyne, and K. Moers, 2. anorg. Chem.,Metallw., 1934,13, 640.1931,196, 129.J. D. Fast, ibid., 1939, 241, 42.A. W. Laubengeyer, D. T. Hurd, A. E. Newkirk, and J. L. Hoard, ibicE., 1943,a6 E. H. WhslowandH. A. Liebhafsky, J. Amer. Chem. SOC., 1942,84, 2725.66, 1924.0 74 INORGANIC CHEMISTRY.0.01-in. tungsten or (preferably) " hydrided " tantalum wires, the latterconsisting merely of tantalum filaments pre-treated with hydrogen at a hightemperature before use. With the wire a t about 1300" and a partial pressureof 18 mm. of boron tribromide in the reacting mixture (total pressureatmospheric), the boron was produced as crystals up to 1 mm. in lengthwhich were readily detachable from the wire ; as much as 0.5 g.of crystallineproduct could be obtairled in a single run. The boron wm shown by spectro-graphic examination to be free from non-volatile impurities such as silicon,carbon, or tantalum from the wire. In this case an application of the hot-wire method has permitted a complete re-examination of the properties ofboron, carried out on material of proved purity. It is noteworthy that thecrystalline boron had a hardness of 9.3 on the modified Moh scale, approach-ing that of boron carbide. An amorphous form of boron could be pre-pctred by using a lower wire temperature, or increasing the boron tribromidepressure in the reaction zone.Reduction Metho&s.-Conventional methods of reducing oxides, halides,and other compounds by purely chemical means have in recent years beenrefined by the introduction of new reducing agents and a variety of newtechniques. The classical method of reducing a metallic oxide with hydrogenis still applicable, however, in some cases ; it is stated, for example, that purecobalt and nickel are obtained by this simple method: and the percentagepurity of cobalt prepared from cobaltous oxide and hydrogen at 550-1200"has been given as 99.81-99*89~&~ Iron of purity superior to that of elec-trolytic iron has recently been produced on a fairly large scale (about 500 g.of product per day) by a process involving hydrogen reduction of ferricoxide.31 The starting material for this process was commercial electrolyticsheet iron containing O - O l l % of copper and 0.018% of phosphorus.Theseimpurities were largely removed by dissolving the metal in high-purityhydrochloric acid and allowing the solution to stand in contact with excessof the iron. The ferrous chloride was crystallised, dried, and convertedinto ferric oxide by treatment with steam and air at 250" ; after being washedwith dilute hydrochloric acid and water, the oxide contained only one-quarter to one-third of the nickel (04014%) present as impurity in the originaliron. The ferric oxide, contained in alumina boats, was reduced by hydrogena t 760", electrically heated tube furnaces being employed for this operation ;in order to minimise spontaneous oxidation of the metal powder on removalfrom the furnace, the reduced material was sintered in nitrogen at 900"before the furnace tubes were opened.A preliminary melting of the ironpowder was carried out in a nitrogen atmosphere; slight oxidation duringthis process was unavoidable, but it served to remove some of the morereadily oxidisable impurities from the metal. Finally, the oxygen wasremoved from the main bulk of material by a process of melting in anatmosphere of hydrogen at successively reduced pressures (15,6, and 3 cm.).The product from the last melting operation was estimated to contain atG. F. Huttig and R. Kassler, 2. anorg. Chem., 1930, 187,25.mal F. Adcock, J . SOC. Chem. Ind., 1940,69,28WELCH : PREPARATION OF SOLID ELEMENTS IN A STATE OF PURITY. 75most 0-006~0 of total impurities.Melting of the iron was in each case carriedout in a high-frequency induction furnace; heating in such furnaces isproduced within the mass of metal itself by alternating currents of largemagnitude, induced in the metal by a suitably placed primary coil suppliedfrom a high-frequency oscillator. The simplicity, flexibility, and all-roundefficiency of induction furnaces renders them eminently suitable for meltingor heat-treatment of moderate masses of metal; high temperatures arereadily attained, and the complete absence of fuel combustion productsand of external heating devices simplifies the problem of preserving the moltenmetal from contamination. The choice of a suitable inert refractory materialstill remains, and is a source of difficulty in many studies on pure metals;in the work on iron just described, impervious sintered alumina crucibleswere selected for the melting operations.Brief reference may be made to several other elements which are preparedin the pure state by hydrogen reduction processes.Molybdenum is obtainedby reducing the dioxide in hydrogen at 900-1200"; the resulting powderis pressed into bars, which are heated electrically to just below the meltingpoint in an atmosphere of hydrogen. This sintering process expels volatileimpurities, and completes the reduction of any residual oxide. The ingotsof metal can afterwards be '' swaged " and drawn into wire, or rolled intosheet.32 Pure rhenium metal is prepared by hydrogen reduction of am-monium per-rhenate, the reaction being completed a t 1000°?3 Metallicvanadium, stated to be 99435% pure, is obtainable by reduction of speciallyprepared vanadium trichloride with hydrogen, but the reaction is slow.3aElemental arsenic containing less than 0.002 yo of antimony, 0.0002~0of iron, 0.005% of sulphur, and 0.01% of phosphorus is stated to be obtainedby reducing recrystallised ammonium dihydrogen arsenate with ammoniaa t 1000°.35The use of novel methods of reduction is well illustrated by two processesfor the preparation of pure metallic chromium.36 An essential conditionfor effective reduction of chromic oxide by hydrogen is the maintenance ofthe pressure of water vapour produced in the reaction a t a low value.37In the first method this condition is secured by placing thin layers of chromicoxide (prepared by heating redistilled chromium trioxide) between platesof metallic tantalum, and " hydriding " the tantalum by heating in com-mercial hydrogen, from which the impurities are not absorbed; when thetantalum ia saturated with hydrogen the pressure in the reaction chamber,heated at lOOO", is reduced, and the chromic oxide then undergoes reductionby the pure hydrogen evolved by dissociation of the tantalum hydride;the low pressure prevailing ensures rapid removal of water vapour fromthe reaction zone.In the second method calcium hydride, CaH',, is em-31 C . J. Smithells, Metal. Ind. (Lond.), 1931, 38, 336.33 L. C . Hurd and E. Brimm, Inorganic Syntheses, 1939, 1, 175.34 T. Doring and J.Geiler, 2. anory. Chem., 1934,221, 56.35 -4. dePassill6, Compt. rend., 1934, 198, 1781.3' P. P. Alexander, Met. and Alloys, 1934, 6, 37.37 H. von Wartenberg and S. Aoyama, 2. Elektrochem., 1927,33, 14476 INORGANIC CHEMISTRY.ployed as the reducing agent; this liberates hydrogen on heating, and themetallic calcium set free is available to react with water vapour formedwhen the chromic oxide is reduced. The water is thus removed rapidlyby chemical means, and the reaction responsible for its removal yields a freshquantity of hydrogen gas for reduction purposes. The effectiveness of thisingenious process may be judged from the claim that chromic oxide can bereduced completely by calcium hydride in 30 minutes at a temperature aslow as 470".The chromium produced is 99.95% pure, the chief impuritybeing calcium. It is stated that a similar method of reduction has beensuccessfully used with oxides of thorium, beryllium, vanadium, and boron.Metallic calcium proves to be a valuable reducing agent for the preparationof certain pure metals. Granules of metallic chromium which are moderatelyductile can be obtained by reducing chromic chloride or chromic oxide withcalcium in a steel bomb heated (by induction) in an atmosphere of argon.38The successful preparation of even moderately pure chromium from theoxide and calcium is remarkable, for under low pressures calcium oxide isitself reduced by chromium; evidently the equilibrium is to a large extentdependent on pressure, the high pressure developed in a bomb favouring theformation of metallic chromium. This pressure effect may well repay furtherstudy in other similar cases.In its simple form the reduction of chromicoxide by calcium is of little practical value, for chromium is not readilymelted and cast, and the granular product cannot be pressed or sintered to acoherent mass of metal. A chromium powder that can be sintered is pre-pared by carrying out the reduction at 1000" in a flux of molten calciumand barium chlorides, in an argon atmosphere; the use of a bomb is un-necessary in this case. The initial product, obtained by extracting solublematerial from the cooled melt with water and dilute nitric acid, is given asecond similar treatment with a small quantity of calcium to ensure reductionof small inclusions of oxide.Sintering of pressed bars of the pure chromiumis carried out first at 1300" in a vacuum, and then at 1600-1700" in argon,the bars being placed on beryllium oxide refractory supports; the sinteredmetal is brittle at room temperature, but can be rolled under barium chlorideat about 1250". It is stated that thorium, uranium, and vanadium can beprepared by methods similar to that just described.The preparation of pure titanium and zirconium by reduction of theiroxides and halides has been discussed in some and particulars havebeen given of the reduction of the dioxides with calcium in a calcium chlorideand barium chloride flux. Although they are ductile at moderate temper-atures, the metals obtained still contain a little oxygen, the presence of whichis admittedly due to the equilibria referred to above.This oxygen cannotbe removed by any known deoxidiser.A simple and elegant method for preparation of pure rubidium or casiumfrom a halide by reduction with calcium has been described.*O The halide-t 8 W. Kroll, 2. anorg. Chem., 1936, 226, 23.40 F. C. Schmidt, F. J. Studor, and J,. Sottysiak, J . Amer. Chem. SOC., 1938, 60,as Idem, ibid., 1937, 234, 42.2780WELCH: PREPARATION OF SOLID ELEMENTS IN A STATE OF PURITY. 77calcium mixture is placed in a nickel tube, which is practically sealed by bend-ing over the ends and suspended in an evacuated glass enclosure. Thenickel tube is heated by induction; the alkali metal then distils through theseams, and is collected in a suitably placed bulb.The very strong affinity of elemental zirconium for oxygen, combinedwith the refractory nature of the oxide, renders the metal a particularlyuseful, if expensive, reducing agent.The alkali metals have been obtainedin a very pure state by reduction of their sulphates, chromates, dichromates,molybdates, and tungstates with zirconium.41 Excess of zirconium must beused, otherwise the reactions are explosive. For the preparation ofpotassium, rubidium, and czsium, heating of mixtures of the chromates withzirconium powder (1 part to 4 parts, by weight) at 700-800" is recom-mended; a similar mixture with the molybdate, heated at 550", is preferredfor sodium. Lithium is obtained from the chromate (1 part to 8 of zirconium,at 450-600"), but the yield is small.These reactions are useful because themixed starting materials are stable for indefinite periods in air, the reactiontemperatures are moderate, no readily volatile products other than thealkali metals are obtained, and under optimum conditions the yields aregood.Preparation of Metals by Electrolytic Methods.-A number of metals areprepared in the pure state by electrolysis of solutions or fused salt meltscontaining their compounds ; methods of electrolytic refining in which ananode of previously prepared crude metal is employed are discussed separatelybelow.Metallic gallium is successfully prepared by electrolysis of a solutionof gallium hydroxide in sodium hydroxide solution, platinum electrodesbeing used.42 The temperature of the solution is kept above'30°, the meltingpoint of the metal, and the liquid gallium is collected in a shallow glass cupbelow the cathode, with which it remains in electrical connection.Theinitial product is freed from traces of lead, tin, and platinum by washingsuccessively with hydrochloric acid (1 : l), concentrated nitric acid, and dilutehydrochloric acid ; the purified gallium is spectroscopically pure exceptfor a faint trace of iron. Direct electrolysis of an alkaline extract of germaniteore has recently been employed to give a deposit of gallium and ger-manium; 43 the latter is removed b'y treating the deposit with chlorine anddistilling off the resulting germanium tetrachloride, and the residual galliumtrichloride, after removal of lead, antimony, and molybdenum by precipit-ation with hydrogen sulphide, is used for the electrolytic preparationof galliummetal. This new method affords a simple and rapid means of obtainingpure gallium from its principal natural source.Tin is an interesting example of a metal the properties of which areconsiderably affected by traces of impurity; it has been shown that as littleas 0.0035% of bismuth imparts an unusual " cored " structure to the cast41 J.H. de Boer, J. Broos, and H. Emmens, 2. anorg. Chem., 1930,191, 113.4a F. Sebba and W. Pugh, J., 1937, 1371.D. J. Lloyd and W. Pugh, ibid., 1943, 878 TNORQANfC CHEMISTRY.metal, and inhibits the transition to grey tin a t low temperatures.44 Com-mercial samples of supposedly “ pure ” tin all showed the cored structureafter casting, and a “ structurally pure ” product, shown to be free frombismuth, was obtained only by electrolysis of a solution of stannous chloridecontaining some suspended metastannic acid to adsorb impurities.Needle-like crystals of pure thorium are stated to be obtained by electro-lysis of an aqueous solution of thorium sulphamate.*5The technique of preparing metals by electrolysis of salt melts is illustratedby the production of pure tantalum,4* and niobium 40from melts containing complex fluorides of the metals.In the case of uraniumthe electrolyte consists of equal parts by weight of calcium and potassiumchlorides containing the green complex fluoride KUF,; this is fused at775” in an electrically heated graphite crucible, which serves as the anode.The cathode is a molybdenum strip suspended in the centre of the crucible.As electrolysis (requiring 30 amp.at about 5 volts) proceeds, uraniumseparates as a tree-like deposit on the cathode, which is removed and re-placed by a new molybdenum strip a t intervals; additions of the electrolyteconstituents are made when necessary. The cooled cathode material,containing solidified salts from the bath, is washed with water, dilute aceticacid, alcohol, and ether, and dried ; insoluble calcium fIuoride in the productis readily washed away from the very much denser uranium metal. Theuranium is obtained as a coarse, grey powder, which is pressed into pelletsand fused in a vacuum in an induction furnace. The metal then contains0.06% of carbon, 0.05% of iron, and 0.01 yo of silicon. I n this process thecomplex fluoride was adopted as electrolyte after tests had shown theunsuitability of a uranyl salt or uranium trioxide; these gave a deposit ofuranium dioxide at the cathode during electrolysis.A closely similar technique is used in the case of the thorium,47 exceptthat potassium and sodium chlorides, with KThF,, are employed in themelt to obviate difficulties due to formation of calcium fluoride, which inthis case is not easily washed out of the product.The pressed, sintered,and degassed metal is stated to be very soft, and to contain only 0.02%of carbon, 0.05% of silicon, and 0.005% of iron.Tantalum and niobiumproduced by electrolysis, under similar conditions, of the complex fluoridesK,TaF, 48 and K,NbF, 49 are of comparable purity; in these instances,however, the “ anode effect,” well known in the electrolysis of melts of thistype, is troublesome unless tantalum or niobium pentoxide is also added tothe molten electrolyte.Further useful details of the technique of electrolysis of fused salts areprovided by a long paper on the isolation of pure rare-earth metals,50 and afull description of the preparation of pure metallic scandium.51 Pure4 4 C. W. Mason and W. D. Forgeng, Met. and Alloys, 1935, 6, 87.15 R. Piontelli and A. Giulotti, Chim. e Z’InCE., 1939,21,478.46 F. H. Driggs and W. C . Lillienduhl, I d . ErLg. Chem., 1930, 22, 518.4 7 Idem, ibid., p.1302. 4 8 Idem, &id., 1931,23, 634.49 C. W. Bake, ibid., 1935, 27, 1166.O1 W. Fischer, K. Briinger, and H. Grieneisen, 2. a w g . Chcm., 1937, 231, 54.F. Trombe, Ann. Chim., 1936,6,349WELCH : PREPARATION OF SOLID ELEMENTS IN A STATE OF PURITY. 79cerium, for example, is prepared by electrolysis, a t 800-850", of a fusedmixture of anhydrous cerous chloride (60%) and potassium chloride (40%),with an addition of about 5% of calcium fluoride. The apparatus foundmost suitable is a graphite crucible (anode), with a central rotating cathodeof molybdenum or tungsten, shielded along most of its length by a concentricquartz tube ; the metal collects in a, sintered fluorite or quartz crucible fittedinto the bottom of the graphite container, into which the lower end of thecathode projects.The cerium obtained contains potassium, which is re-moved by fusion in a vacuum in a tube furnace, or in a cathode-ray furnaceof interesting design which permits efficient attainment of very hightemperatures. Lanthanum, neodymium, and praseodymium are obtainedby very similar methods, but in the preparation of samarium and gadoliniumthe method is modified by the u8e of a pool of molten cadmium as the cathode ;the cadmium is afterwards distilled from the resulting alloy by heatingin a vacuum a t 1300". An analogous method is applied in the preparationof s~andium,~l the cathode consisting of molten zinc of high purity; evenwith special precautions during electrolysis, some oxidation of the scandium-zinc alloy is difficult to prevent, and the oxide is subsequently " filtered off ''from the molten metal, with inevitable loss of scandium, by passing it througha tungsten crucible with a perforation in the bottom. The zinc is finallydistilled off by slow heating to 1250" in a vacuum ; the scandium then remainsin a highly sintered condition.Although the percentage purity of themetal obtained is given as only 94-98%, this figure must be consideredin relation to the high reactivity and affinity for oxygen associated withscandium metal.Puri$cation of Solid Elements by Distillation.-A number of solid elementsare prepared in the pure state by applying some method of purification to acrude material, rather than by direct production from a purified compound.I n a number of cases, distillation at high temperatures has been investigatedas a means of effecting the necessary purification. A review is available 52covering the technique and experimental difficulties involved in some detail,and describing experiments (on the laboratory scale) on the distillationof chromium, aluminium, silicon, beryllium, iron, copper, nickel, tin, andlead.The purification of magnesium by distillation methods has also beendiscussed. 53A particularly difficult case of purification of a metal by distillationarose in work carried out at the National Physical Laboratory on the pro-duction of pure beryllium,54 and this merits brief description. Berylliumprepared by electrolysis of fluoride melts, although 99.6-99.7 yo pure,obstinately retains a little oxygen, which cannot be excluded by any simplemodification of the electrolysis technique ; the oxygen present causes51 W.Kroll, iWetaZZw., 1934, 13, 726, 789; Metal Ind. (Lond.), 1935, 47, 3, 39, 103,53 W. Kaufmann and P. Siedler, Z. Eleklrochem., 1931, 37, 402; J. HQenguel and54 H. A. Sloman, J . Inst. Metals, 1938, 49, 366.155.G. Chaudron, Cornpt. rend., 1931,193, 77180 INORGANIC CHEMISTRY.deposition of a beryllium-beryllium oxide eutectic in the solidified metal,and this is said to be responsible for the brittleness previously associatedwith metallic beryllium. A distillation apparatus was eventually constructedin which the beryllium, contained in a crucible of sintered beryllium oxide,was heated to about 1900" by induction; the vapour issued through a" baffle," designed to prevent collection of splashes of the molten metal,and was condensed on a water-cooled silica surface.The whole apparatuswas kept under high vacuum. Although distillation was slow and thequantity of product small, beryllium of purity estimated a t 99.95-99-97was obtained; this met'al was ductile, and contained none of the eutecticpreviously mentioned. Attempts to deoxidise beryllium by chemicalmeans-including fusion in the flame of an atomic hydrogen blowpipe-were unsuccessful, and apparently distillation is the only known means ofpreparing an oxygen-free metal.A simple and ingenious apparatus designed for the distillation of zincon the laboratory scale 55 is noteworthy as a prototype ; the zinc is condensedon a graphite sleeve so designed that the crystals grow downwards awayfrom the surface, and are readily removed without metal which has beenin contact with graphite.I n the commercial distillation of zinc a productof 99.994% purity is stated to be obtainable; 56 the use of carborundumas a refractory material in the apparatus may be noted.Tellurium is obtained in the pure state by vacuum distillation of a com-mercial electrolytic product containing selenium, copper, iron, and someoxide. 57Puri$cation of Metals by Xintering or Fusion in a Vacuum.-Passingreference has already been made to sintering and vacuum fusion processes,which are of special value in dealing with metals of high melting point andhigh oxygen affinity, respectively.Both methods are used in the preparationof pure, ductile tantalum and A crude tantalum powder isobtained by reduction of the double potassium fluoride, K,TaF,, withsodium ; this is pressed into bars and sintered in a vacuum a t a temperaturejust below the melting point to remove volatile impurities. Alternatively,the crude metal is converted into tantalum hydride by heating in hydrogenat IOOO", and the hydride is decomposed by sintering at 1500" or above.The sintered bars of tantalum produced,in either process are brittle, andvacuum fusion is necessary to render the metal ductile; since the meltingpoint is very high (2800-2850"), the fusion process is carried out by strikingan electric arc between a block of the sintered material and an electrode oftantalum or tungsten.The final product is sufficiently ductile to be rolledwithout difficulty into sheets 0.1 mm. thick. Ductile niobium is producedby an exactly similar method.5 5 E. C. Truesdale and G. Edmunds, h e r . Inst. Min. Met. Eng., Inst. Metalss6 H. Matthies, Metall u. Erz, 1936, 33, 280.5 7 F. C. Kracek, J. Amr. Chem. SOC., 1941, 63, 1989.5 a C. J. Smithells, Metal I d . (Lond.), 1931, 38, 336.Divn., Tech. Publ. 1033 (1939)WELCH : PREPARATION OF SOLID ELEMENTS IN A STATE OF PURITY. 81Final purification of cathodic nickel has been effected by annealingthe metal in hydrogen a t 1050" to remove carbon and sulphur, and thenfusing in a vacuum in specially bonded magnesia crucibles.59 The productis 99.94% pure, the chief impurities being iron (0.03%) and cobalt (0.016%).Attention may be directed to two papers dealing with the theoreticalaspects of degassing of metals,m in which it is shown that maximum ratesof degassing are reached over certain temperature ranges, which are simplyrelated to the melting points of the metals.The results clearly have apractical bearing on the purification of metals by sintering in a vacuum.Electrolytic Puri$cation of &etch.-This method of refining, so wellknown in industrial practice, finds useful applications in the preparationof high-purity metals on the laboratory scale. Recent work in this directionis well exemplified by a method for producing 99.999% pure copper.61The first stage in the refining of commercial electrolytic copper to this purityis an electrolysis at low current density through a bath of sulphuric acidand copper sulphate, anodes of the crude metal and cathodes of pure coppersheet being employed in the usual manner.This electrolysis is stated tofree the copper from all the important impurities except sulphur ; the latteris removed by air-blowing the surface of the molten metal, a process whichnecessarily introduces oxygen, with clay, iron, and graphite from the meltingcrucible. Final purification is effected by a second electrolysis through acopper nitrate electrolyte. In each of the electrolyses " starting sheets ')are first obtained by deposition on stainless steel plates, from which they canafterwards be stripped.The purified metal is cast into oxygen-free rods bymelting in hydrogen in a specially designed casting apparatus.Metallic indium estimated to contain not more than O - O O l ~ o of totalimpurities has been prepared by electrolytic refining of a, " pure )' com-mercial product through an indium chloride solution, two successive electro-lyses being required.62 Other metals to which electrolytic refining is par-ticularly applicable include zinc,63 lead,64 mangane~e,~5 and silver.66A particularly novel method of electrolytic purification of aluminium,in which both electrodes are composed of molten metal, has been de~cribed.~'The lowest and densest layer in the cell is an aluminium-copper alloy con-taining 33% of copper ; above this is the electrolyte, consisting of moltenalkali-metal and aluminium fluorides and barium chloride ; the upper layeris the cathode of pure, molten aluminium, which is less dense than the fused6' L.Jordan, W. H. Swanger, et al., J. Reg. Nut. Bur. Stand., 1930, 5, 1291.60 G. F. Huttig, H. Thiemer, and W. Breuer, 2. anorg. Chem., 1942, 249, 134; G. F.61 J. S. Smart, jun., A. A. Smith, jun., and A. J. Phillips, Amer. Inst. Min. Met.Huttig and H. H. BIudau, ibid., 1942, 250,36.Eng., Inst. Metals Divn., Tech. Publ. 1289 (1941).G. P. Baxter and C. M. Alter, J. Amer. Chem. SOC., 1933, 55, 1943.R. S. Russell, Proc. Awl. Inst. Min. Met., 1932, 87, 145.63 W. Hiinig, Metallu. Erz, 1936, 33, 274.6 5 H. H. Oaks and W. E. Bradt, Trans. Amer. Electrochem.SOC., 1936,69,567.66 G. P. Baxter and 0. W. Lundstedt, J. Amer. Chem. Soc., 1940, 62, 1829.H. Diirr, Gieeeereipraxb, 1938, 69, 11482 INOROBNIC CHEMISTRY.salt bath. During electrolysis the less electropositive impurities (iron,silicon, etc.) remain in the anode layer, and the more electropositive ones(magnesium and lithium) dissolve in the electrolyte 86 their chlorides.Special Purification Methods applicable to Individual Elements,-Inaddition to the more or less general methods of preparation and purificationdescribed above, there are numerous special methods designed to removespecific impurities from particular elements. These cannot be enumeratedhere, but brief reference may be made to the purification of iodine by re-moval of other halogens and organic matter,66 and the extraction of iron,silica, and other impurities from commercial silicon by means of acids.68It has recently been pointed out 69 that the most troublesome impurityin commercial sulphur is organic material originating from the hydro-carbons always associated with non-volcanic sulphur deposits.Decom-position of organic material leads also to the presence of hydrogen per-sulphides. Four successive distillations of commercial sulphur do notsuffice to remove the impurities, the presence of which is indicated by thedevelopment of black specks when the sulphur is boiled in a clean glass tube.The method of purification recommended i s as follows : the sulphur (1 kg.)is raised slowly to the boiling point in a Pyrex flask, and boiling continuedfor 3 4 hours after addition of 6 g. of magnesium oxide; this aerves toremove acid impurities and decompose hydrogen persulphides. Thesulphur is allowed to stand for some hours a t 125", and the clear moltenmaterial is decanted from a black sludge of impurities, through a Pyrexwool filter. The sulphur is then heated a t the boiling point for four suc-cessive periods of about 30 hours with 10-g. portions of magnesium oxide,the liquid being filtered after each period of boiling. The purified sulphurcontains no detectable impurity. The method described is intended foruse with Amerioan sulphur, in which there is no arsenic, selenium, ortellurium.Preparation of Elements in Special Allobropic -Forms.-Attention may bedirected here to recent work on the synthesis of diarn~nd,~o in which Moissan'sexperiments were repeated under a variety of conditions with moderntechnique ; molten iron containing carbon was quenched in water- or liquid-air-cooled vessels after heating a t temperatures as high as 3000". Graphitewaa also subjected to a momentary pressure of about 120,000 kg. per sq. cm.at 3000-3200". Although a few very small fragments with the propertiesof diamond were produced in some experiments, consistent yields of diamondswere never obtained. ' These experiments lend added interest tothe discovery,made by X-ray analysis, that eleven out of twelve (' artificial diamonds "allegedly prepared by J. B. Hannay in 1879-1880 are indeed diamonds, a tleast one of them having the rare " type I1 " structure.'l Hannay's attempts6 a N. P. Tucker, J. Iron Steel Inst., 1927,115,412 ; -4. B. Kinzel and T. R. Cunning-ham, Amer. Inst. Min. Met. Eng., Inst. Metals Divn., Tech. Publ. 1138 (1939).6D R. F. Bacon and R. Fanelli, lnd. Eny. C'hem., 1912, 34, 10.13.'13 P. L. Giinther, P. Geselle, and W. Rebentisoh, 2. anorg. C'hem., 1943, 250,357.71 F. A. Bannister and (Mrs.) K. Lonsdale, Nature, 1943,151, 334; (Lord) Rayleigli,ibid., p. 394; F. A. Bannister and K. Lonsdale, Min. Mag., 1943, 26, 315WELCH : PREPARATION OF SOLID ELEMENTS IN A STATE OF P ~ I T Y . 83a t diamond synthesis were carried out by heating paraffin, bone oil, andmetallic lithium to a red heat in a very strong iron tube.72Black phosphorus has recently been prepared from the white form bymomentary application of a pressure of about 100,000 kg. per sq. cm., a troom temperaf~re.~~ The black form is stated to be unstable under ordinarytemperature and pressure conditions, and to revert to white phosphoruson keeping; the red form is the most stable allotrope. A. J. E. W.H. J. EMEL~US.A. J. E. WELCH.7 2 Nature, 1880, 22, 355; Proc. Roy. SOC., 1880, A , 30, 188, 450; 1882, A , 32,407.P. L. Giinther, P. Geselle, and W. Rebentisch, 2. anorg. Chem., 1943, 250, 373