INORGANIC CHEMISTRY.1. ATOMIC WEIGHTS.THE period which has elapsed since this subject was last reviewedfor the Annual Reports has seen the redetermination of a numberof atomic weights by chemical methods. Hiinigschmid and hiscolleagues have published papers on radium,l cadmi~in,~germanium,~tantalum,* m~lybdenum,~ tungsten,6 and raniu urn,^ as well as on theratio AgNO, : AgCl,* and American workers have made importantcontributions to this subject dealing with potas~ium,~ rubidium,lOgallium,ll carbon,12 arsenic,13 gadolinium,14 and eur0piurn.1~ I naddition, investigations have appeared on radiogenic lead,16, l7erbium,ls, terbium,lg protoactinium,m and neon.22 Asurvey of some of this new work has appeared in the fifth 23 and t,h&0. Honigschmid and R.Sachtleben, 2. anorg. Chem., 1934, 221, 66.0. Honigschmid and R. Schlee, ibid., 1936, 227, 184.0. Honigschmid, K. Wintersberger, and F. Wittner, ibid., 1935,225,Sl;0. Honigschmid and R. Schlee, ibid., 1935, 225, 64; 1934, 221, 129.0. Honigschmid and G. Wittmann, ibid., 1936, 229, 65.0. Honigschmid and F. Wittner, ibid., 1936, 226, 289.0. Honigschmid and R. Schlee, Angew. Chern., 1936, 49, 464.C. R. Johnson, J . Physical Chem., 1935, 39, 781.0. Honigschmicl and K. Wintersberger, ibid., 1936, 227, 17.cJ 0. Hiinigschmid and W. Menn, ibid., p. 49.lo E. H. Archibaldand J. G. Hooley, J. Amer. Chem. Xoc., 1936,58,70, 618;E. H. Archibald, J. G. Hooley, and N. W, F. Phillips, Tran8. Roy. SOC. Canada,1935, [iii], 29, 111, 155.l1 G. E. F. Lwidell and J.L. Hoffmann, J . Res. Nat. Bur. Stand., 1935, 15,409.l2 G. P. Raxter and A. H. Hale, J. Amer. Chew&. SOC., 1936, 58, 510.l3 G. P. Baxter and L. D. Frizzell, ibid., 1935, 57, 851.l4 C. R. Naeser and B. S . Hopkins, ibid., p. 2183.l5 E. L. Meyer and B. S. Hopkins, ibid., p. 241.l6 G. P. Baxter and C. M. Alter, ibid., p. 467.l7 F. Hecht and E. Kroupa, 2. anorg. C'hem., 1036,226,248.18 0. H6nigschmid, Natumiss., 1936, 24, 619.19 J. K. Marsh, J., 1935, 772.2o A. V. Grosse, J . Amer. Chem. SOC., 1934, 56, 2501; Proc. Roy. Soc.,21 F. G. Ntiiiez, Anal. Fis. Qu,irn., 1935, 33, 633.22 A. von Antropoff, Ber., 1935,68, B, 2389.33 G. I?. Baxter, 0. Hiinigschmid, P. Lebeau, and R. J. Meyer, Bet-., 1935,1935, A, 150, 363.68, -4, 73-84136 INORGANIC CHEMISTRY.sixth24 report of the Committee on Atomic Weights of the Inter-national Union of Chemistry, and in the forthcoming seventh reportthe more recent investigations will be described.In these the readerwill find a complete account of the chemical methods and the data,obtained in each investigation, as well as the table of approvedatomic weight values for the year. It is therefore unnecessary toattempt here a complete summary, and attention will be directedonly to those points which appear to be of special interest andimportance.Some of these new researches are an extension of work mentionedalready25 and lead by different ratios to the same atomic weight.Por example, the ratio TeCl, : 4Ag : 4AgC1 26 gives Te = 127-63, a resultonly slightly higher than that from the corresponding bromide ratioswhich is identical with the figure Te = 127.61 obtained from thesynthesis of silver telluride.2' Similarly, the new work on tantalumchloride 4 confirms the previous value for this element, i.e., Ta =180.88, obtained by the analysis of the bromide, which is more than0.5 unit lower than the old value of 181.4.28 This and similar workon niobium brings the chemical into close agreement with massspectrograph values.The same may now be said of germanium.39 52The work of Honigschmid and Menn on tungsten and Hhigschmidand Wittmann on molybdenum presents many points of interest onaccount of the difficulties which had to be surmounted before anhy-drous specimens of WCI, and MoCI, could be prepared in a highlypurified state and a satisfactory procedure devised for titration withsilver ion.This is the first time that a direct nephelometric com-parison with silver has been used in determining the atomic weightsof these two elements and the values W = 183.92 and Mo = 95.95obtained both from chemical evidence and from the close agreementwith Aston's values are probably very close to the truth.Another interesting revision is that of the atomic weight of radiumcarried out by Honigschmid and Sachtleben,l using as startingmaterial 5 g. of radium bromide containing 1.17% of barium bromidelent for the purpose by the Union Minibre du Haut Katanga ofBrussels. The accepted value for this element was based on the ratiosRaC1, : 2Ag : 2AgC1 and RaBr, : 2Ag : 2AgBr determined by Honig-schmid in 1912 29 with about 1 g.of the chloride supplied by theVienna Radium Institute. The much larger amount made possible a24 G. P. Baxter, 0. Honigschmid, and P. Lobeau, J. Amer. Chem. SOC., 1936,3 5 Ann. Reports, 1934, 31, 95.26 0. Hanigschmid and H. Baudrexler, 2. nnorg. Cirem., 1935, 2%, 91.27 Ann. Reports, 1933, 30, 83.28 Ibid., p. 82; 1934, 31, 96.29 Monntsh., 1912, 33,253; 1913, 34, 283.58, 541-548WHYTLAW-GRAY : ATOMTC WETGHTS. 137more elaborate purification and recrystallisation. The final material,ttfher an extensive series of recrystallisdiions as chloride, was testedspectroscopically by W. Gerlsch and E. Riedl and found to contJniria trace only of barium, estimated at ~.O02-@0O3%.Of this purifiedmaterial 3-5 g. were obtained. Quantities varying from 2 to 3.4 g.were dried and converted into the anhydrous salt by heating ingaseous hydrogen bromide charged with bromine up to a temperatureof 750°, the gas being displaced by nitrogen, and this in turn by air.After weighing, the bromide was converted into the chloride bysimilar treatment with hydrogen chloride and chlorine, followed bynitrogen and air-a method essentially similar to that used first byR. Whytlaw-Gray and (Sir) W. R,amsay 31 when attempting todetermine the atomic weight of radium with very small amounts ofmaterial.Difficulties were experienced in weighing the anhydrous radiumsalts on account of the heat evolution. The final value found for theatomic weight was 226.05, slightly higher than the older value225.97.Success with this element has led Honigschmid andWittner to revise earlier work carried out in Munich on the atomicweight of the parent element uranium, and in a comprehensiveinvestigation recently published they describe the preparation ofanhydrous UC1, and UBr,-every precaution being taken to obtainperfectly homogeneous materials of definite stoicheiometric composi-tion-and their subsequent analysis in terms of silver by standardmethods.The mean values for the nephelometric titrations gave U =238.073 from the chloride and U = 238.076 from the bromide ratios,The halides prepared from minerals from different districts andof varied geological age gave identical values for the atomic weight.Although a number of determinations were also made in which theprecipitated silver halides were collected and weighed, giving valuesof 238.066 from the ratio UC1, : 4AgC1 and 238.088 from UBr, : LiAgBr,yet these were not considered so reliable and were rejected in cal-culating the final mean.Some indication was also obtained thatfusion of the uranium halides before weighing led to a slight dis-sociation and so raised the atomic weight. ,411 the sources of errorindicated by a careful study of the data being taken into account,U = 238.07 is advanced as the most probable value, appreciablylower than the international figure 238.14.At the end of the paper the authors discuss the bearing of theirresults on the uranium-radium series and the origin of actinium andare able to show that they accord satisfactorily with physical data,30 2.anorg. Clwm., 1934, 221, 103.31 Proc. Roy. SOC., 1912, A , 86, 270138 INORGANIC CHEMISTRY.Starting from uranium-lead from Morogoro, the atomic weight ofwhich 0. Honigschmid, R. Xachtleben and H. Baudrexler foundsome years ago to be 206-03,32 and which G. P. Baxter and C. M.Hilton= have since confimed, they point out that this value isidentical with the mass-spectrograph figure for uranium-lead 34 fromthe same source when the latter is corrected to the chemical standard.Aston showed that this lead contained only the isotopes m6Pb(Ra-G)and m7Pb(Ac-D) in the ratio of 93.1 to 6.9. The atomic weight ofradium-@ is hence 206965, from which, by taking into account theenergy changes and the a-particles lost in passing from uranium toradium and from radium to radium-G, they find the calculated atomicweights of uranium and radium to be 238.044 and 226418, insatisfactory agreement with the experimentally found values of238.07 and 226.05.That uranium consists mainly of 238U and contains no higherisotope but only something less than 1% (0.4% estimated fromradioactivity measurements) of 235U, the parent of protoactinium,seems certain from the work of A.J. Demp~ter,~~ who has recentlycompared in his mass spectrograph the doubly-charged ions ofuranium and thorium with the two tin isotopes llGSn and lI9Sn.Reducing the values to the chemical scale and taking into accountthe small amount of 235U, he finds U = 238-028 and Th = 2320024.No indication of a higher isotope was obtained.Thus Aston'sconclusion that 231 is the atomic weight of protoactinium is againconfirmed. This is in accordance with a direct measurement ofthe atomic weight made by A. V. Grosse36 by the conversion ofK,YaF, into Pa20, by evaporation with sulphuric acid, followed byprecipitation with ammonia and subsequent ignition. Though thequantity of material was small (about 50 mg. of oxide), it showed noimpurities when examined in an X-ray spectrograph, but the value230.6 &0.5 might well be checked later on by a standard method.Pa = 231 appears this year for the first time in the internationaltable.Important work on fundamental ratios has appeared during 1936from the Munich laboratories.It has been found possible to carryout the conversion of silver nitrate into silver chloride with gaseoushydrogen chloride with the high accuracy required in this class ofinvestigation.It is obviously an advantage if solution, precipitation, filtration,32 2. anorg. Chem., 1933,214,104; s0e Ann. Reports, 1933,30,34.33 J . Amer. Chem. SOC., 1935, 57, 469.34 3'. W. Aston, Proc. Roy. SOC., 1933, A , 140, 535.35 Nature, 1936, 138, 120, 201.36 Proc. Roy. Soc., 1936,150, 363; Ann. Reports, 1935, 32, 146WYTLAW-GRAY : ATOMIC WEIGHTS. 139transference, and the evaporation of la,rge volumes of liquid can beavoided, and the reaction carried out in the same vessel with gaseousreagents. Dry reactions are, however, subject to errors due to avariety of effects such as adsorption and sublimation, but it is inter-esting to note their increasing use in modern work as exemplified bythe synthesis of silver ~lrlphide,~' ~ e l e n i d e , ~ ~ and telluride,3Q thereduction of silver nitrate to silver by hydrogenY4* the conversion ofbarium perchlorate into barium chloride 41 by hydrogen chloride gas,of silver iodide into silver chloride,42 and of radium and bariumbromides into the chlorides Making use of atechnique similar t o that employed in the reduction of silver nitrate,0.Honigschmid and R. Schlee * obtained for this ratio from eightclosely concordant experiments a mean value of 1.185241 with amaximum divergence of 7 in the last decimal place.Internationalatomic weights being used, the calculated ratio is 10185235. Com-bining the new ratio with the well-established ratios AgNO, : Ag =167479 and AgCl : C1= 4-042592, the following values were obtained:Ag = 107.881, C1 = 35.456, N = 14.009, which, taking into accountthe indirect method of calculation, the authors regard as a confirma-tion of the accepted values rig = 107.880, C1 = 35.457, N = 14.008,which are based on more direct ratios.Another example of stoicheiometric work on a standard ratiois that by C . R. Johnson on the atomic mass of potassium in whichthe experience gained in previous work on sodium 43 has now beenapplied to this element. The investigation on the ratios NaCl : Agand NaCl : AgCl mentioned in the 1934 Report presented a numberof new features, among others the checking of the equal opalescencemethod of nephelometric titration by potentiometric analyses.Inthe latest work, five samples of highly purified potassium chloride,prepared from material from American, German, and Norwegiansources, were referred to three independently purified samples ofsilver, and the fifteen determinations gave values for the KCl : Agratio ranging from 0.691 103 to 0.691 112, giving mean of 0.691 108 &0.0000005, which, with international values for Ag and C1, gaveK = 39.100. This lies midway between the international value37 0. Hilnigsclimid and R. Sachtleben, 2. anorg. Chem., 1931,195,207.88 0. Honigschmid and W. Kapfenberger, ibid., 1933, 212, 198 (see Ann,39 0.Honigschmid and I<. Wintersberger, ibid., p. 242 (see Ann. Reports,40 0. HBnigschmid, E. Zintl, and P. Thilo, ibid., 1927, 163, 66.4 1 0. Honigschmid and R. Sachtleben, ibid., 1929, 178, 1.42 0. Honigschmid and H. Striebel, 2. physikaZ. C7zem., Bodenatein Fest-43 J. Physical Chem., 1933, 37, 923 ; Ann. Reprta, 1934, 31, 96.in a similar way.Reports, 1933, 30, 82, 83).loc. cit.).band, 1936, 283140 INORGANIC CHEMISTRY.K = 39.096, based on the work of G. P. Baxter and W. M. Mac-Nevin *4 and of Honigschmid and Sa~htleben,~~ and the high valueK = 39.104, found previously by Honigschmid and J. G o u b e a ~ , ~ ~and is identical with the value for potassium appearing in the inter-national table prior to 1934. No satisfactory reason for the cause ofthese small differences has so far been given, but a suggestive investi-gation on the ratio of the two main isotopes of this element 39K and41K by A.K. Brewer 47 shows the importance of a knowledge of iso-topic composition in stoicheiometric work of high accuracy. Brewer,using a mass spectrograph of considerable resolving power, hasmeasured the 39K/41K ratio in potassium from a lasge number ofsamples of sea water and found a hardly detectable variation from themean value 14-20 ; minerals exhibited a slightly greater variation(39K/41K = 14-25), whilst in the ashes of plants a change of 15% inthe 41K content was found. Kelp contained the largest amount of41K. The atomic weight of potassium calculated from the sea-waterratio, probable values being assumed for the packing fraction and forthe change to the chemical scale, was 39.094.To account for a difference of 0-01 unit in the atomic weight ofpotassium would require the displacement of the abundance ratio of1.1 units, which is well within the variation found in potassium ofvegetable origin, but the author, after discussing the available data,rejects this explanation.Brewer's results seem well foundedand are in agreement with the values for the same ratio found byA. 0. Nier,48 who finds 39K/41K = 13.96 and an atomic weight of39-096. It may be noted that the rare isotope 40K discoveredrecently by Nier48sa and confirmed by Brewer49 is probably thesource of the p-radiation of potassium. The suggested degradationof 41K to *Wa has been negatived by recent work of F.W. Aston,5*who was unable to detect this isotope even in calcium separated fromvery old potassium minerals and which was reported to have a, highatomic weight : 40K is only present in small amount, and the ratio39K/40K is estimated at about 840011 and would have no detectableeffect on the chemical atomic weight. The atomic weight of the44 J . Amer. Chem. Soc., 1933, 55, 3185.4 5 8. anorg. Chem., 1933, 213, 365 (see Ann. Reports, 1933, 30, 85).46 Ibid., 1927,163, 93 (see Ann. Reports, {bid.).4 7 J. Amer. Chern. SOC., 1936, 58, 370, 365.48 Physical Rev., 1935, 48, 283.48a Ibid., 1936,50, 104; see also G. von Hevesy and M. Logstrup, 2. anorg.49 Physical Rev., 1935, 48, 640; see dso 0. Klemperer, Proc.Roy. SOC., 1935,5O Proc. Roy. Soc., 1935, A , 149, 399; F. H. Newman and H. J. Walke,Chem., 1928, 1'71, 1.A, 148, 638; andH. J. Walke, Nature, 1935,136,755.Phil. Mag., 1935, 19, 767WHYTLAW-GRAY : ATOMIC WEIGHTS. 141closely related element rubidium has been redetermined with modernrefinements by E. H. Archibald and J. G . Hooley and found to be85.481.Until recently it has been assumed that the variation in isotopiccomposition of the elements in chemical and physical processes is,with the exception of hydrogen, too small to influence even the mostaccurate of atomic-weight values. Recent physical considerations,however, show that on account of isotopic exchange occurring duringchemical reactions this view may require modification.H. C . Ureyand L. J. Greiff 51 have calculated from spectroscopic data the equili-brium constants and enrichment factors of several exchange reactionsinvolving isotopes of the lighter elements, and claim that in somecases the theoretical limit to precision in atomic-weight determina-tions has already been reached. They conclude that “ the atomicweights of many common elements as determined by known chemicalmethods are not fundamental constants of nature to more than a,limited precision.” For instance, in a mixture of chlorine andhydrogen chloride, when equilibrium is established the chlorine willbecome richer and the hydrogen chloride poorer in the 37Cl isotopeto an extent which would alter the atomic weight by 0.001 unit.So far, enrichment factors have been calculated only for a fewreactions, but for some of these, experimental confirmation has beenobtained, as, e.g., the concentration of l80 in carbon dioxide inequilibrium with water,52 and 13C from exchange between bicarbon-ate ion and carbon dioxide.53Since oxygen is the standard of afomic weights, the question maywell be asked : What isotopic composition is to be regarded as nor-mal? Oxygen in air has been proved to be heavier than oxygencombined in water.M. Dole54 finds a difference of 6 parts per millionin the densities of the water made by combining these two oxygenswith the sadme sample of hydrogen, whilst N. Morita and T. Titani 55find 7 p.p.m.; that is, if water oxygen is taken as standard, airoxygen has an atomic weight of 16-00012 ; this difference is, however,very small, and unless chemical reactions involve a greater isotopicseparation than has so far been found, it can exert no appreciableinfluence on atomic weights for chemical use.5GNier computes the atomic weight a t 85.45.5 1 J .Amer. Chem. SOC., 1935,57, 321.52 L. H. Webster, M. H. Wahl, and H. C. Urey, J . Chem. Physics, 1935, 3,53 H. C. Urey, A. H. W. Aten, jun.? and A. S. Keston, ibid., 1936, 4, 623.54 Ibid., pp. 268, 778.65 BuU. Chem. Soc. Japan, 1936, 11, 414; see also C. H. Greene and R. J.Voskuyl, J . Amer. Chem. Soc., 1936, 58, 693.66 See also W. Bleakney and J. A. Hipple, jun.? PhySiCd Rev., 1936, @],129.47, 800142 INORGANIC CHEMISTRY.It does, however, seem important to have an invariable standardfor chemical work, if only to be able to detect any considerable iso-topic exchanges which may occur in chemical operations and alsoto check atomic weights from physical data.It is obviouslydesirable, too, that fundamental chemical ratios should be deter-mined with material of known isotopic composition. Since thepreparation and purification of chemical substances are usuallycarried out in aqueous solution, the isotopic ratio of oxygen in freshwater, which appears from numerous measurements from differentlocalities to be remarkably constant, might well be taken as normal.Turning now from chemical to physical methods, progress hasbeen rapid in the determination of the exact masses of isotopes andthe computation of atomic weights from the abundance ratios.Since the survey of the rare earths which revealed some remarkablediscrepancies in the chemical values that have in some cases beenexplained by revision on the chemical side (terbium,lg erbium,l*gadolinium 14), F.W. Aston 57 has applied his methods to titanium,zirconium, calcium, gallium, silver, nickel, iron, hafnium, indium,cadmium, carbon, thorium, and rhodium. Good agreement withchemical values was obtained for the first seven elements, the silvervalue agreeing within the limits of error of the instrument withthe chemical. With hafnium, indium, and cadmium the agreementwas not so good, and for the last element has led to a revision bychemical methods which has confirmed the international value112.41 and exceeds the physical figure by 1 part in 557.Thoriumand rhodium were found to be simple elements. Recently, A. 0.Nier 48a has studied cadmium with a mass spectrograph of highresolving power, and has found somewhat different values for theproportions of the nine isotopes, leading to 112-37 for the atomicweight. Gold, one OI the four elements which has withstood allAston’s attempts at analysis, has recently been resolved by A. 5.D e m p ~ t e r . ~ ~ He finds it to consist of only one.species of atom 197,and its chemical atomic weight 197.2 is very probably too high.One section of last year’6 Report 59 drew the attention of chemiststo the remarkable developments in nuclear physics which enable themasses of atoms to be calculated with a surprising degree of accuracyfrom a knowledge of the energy changes accompanying nucleartransformations. It described how a study of the mass equivalentsof the energy released when light atoms are disintegrated by protonand deuteron bombardment has brought to light an error in the mass-spectrograph value for helium which was subsequently corrected.Since then Aston,60 using his third mass spectrograph, has obtained67 LOC.cit., p. 396.69 Ann. Reporta, 1936,233, 17, 18.6 8 Nature, 1935, 136, 65.6o Nature, 1936, 137, 357, 613WHYTLAW-GRAY : ATOMIC WEIGHTS. 143iiew and more accurate values for a number of isotopes, and thestudy of nuclear energy changes is extending with such rapidity thatexact values for a large number of atomic masses will soon be avail-able. Already a number of computations of values for the lighteratoms have been either from nuclear transformationsalone or from a combination of the two methods. The values soobtained show a close concordance.The great significance of the new work in the interpretation ofnuclear stability and structure is dealt with in another section.These advances promise to furnish chemistry with a table of atomicweights of an accuracy as great as, if not greater than, that attainedso far in a few of the fundamental ratios.At present, however,atomic-weight values of high accuracy can be calculated only (a> forsimple elements, (b) for elements whose isotopic composition has beenmeasured with sufficient pre~ision.~ For example, although theatomic masses of 35Cl and 37Cl 66* 67 are now known with a probableerr'or of & 0.0008 unit, the estimated uncertainty in the latest valuefor the 35C1/37C1 ratio 68 gives an error ten times as great in the atomicweight, i.e., -+ 0.008 unit.In addition, the change from the scale1 6 0 t o that of chemical oxygen involves a slight uncertainty, foreven with this element the limit of accuracy of the 160/180 ratiois still an open question. Recent work in America 69 points to aslightly higher figure for the conversion factor (1-000275) than theusually accepted value of R. Mecke and W. H. J. Childs 70 (1.00022).The difference, however, is only about 1 part in 20,000. At themoment, on account of these uncertainties, and also because the mostaccurate of the chemical values are not those of the lighter elements,a comparison except in a few cases is valueless. I n the appendedtable values for the atomic weights of helium, oxygen, deuterium,carbon, and nitrogen are given and compared with the internationalvalues. They are taken from Aston's latest measurements 60 andfrom H. A.Wilson's values 63 deduced from nuclear reaction energiesalone without the use of mass-spectrograph results. For the mixed61 M. L. Oliphant, A. E. Kempton, and (Lord) Rutherford, Proc. Roy. SOC.,1935, A , 150,241.62 H. Bethe, Physical Bev., 1935, 47, 633.63 H. A. Wilson, Proc. Roy. Soc., 1936, A , 154, 660; see also L. Isakov,Corn@. rend. Acad. Sci. U.R.S.S., 1935, 3, 301.64 T. W. Bonner and W.hl. Brubaker, Physical Rev., 1936, 50, 308.65 0. Hahn, Ber., 1936, 69, 6.66 F . W. Aston, Nature, 1936, 138, 109.1.67 K. T. Bainbridge, Physicul Rev., 1933, 43, 348.6 s A. 0. Nier and E. E. Hanson, ibid., 1936,50,722.69 S. H. Manian, If. C. Urey, and W. lSleakney, J . Amer. C'hem. Soc., 1934,70 2. Physit, 1931, 68, 362.56, 2610144 INORGANIC CHEMISTRY.elements hydrogen, carbon, and nitrogen the isotopic ratios chosenare those which appear to be the most reliable. Mecke and Childs’sconversion factor has been used.Element.Protium, ‘H ............Hydrogen ...............Helium .....................Carbon ..................Nitrogen ..................Fluorine ..................Deuterium, D(2H) ......Atomic Weights.Mass spectrum.1.00790 f 0*000042.01426 f 0.000071.008084.00303 f 0.0001612*011814-008019.0003 & 0.0006Nuclearenergy.1-007692.013721.007874.0025412-01 1814.0081-Internationalvalue.I -1.00784.00212-0014.00819.000H. A.Wilson’s values for lH, D, and He appear to be on the lowside. Probably the most reliable values for the atomic masses arethose recently advanced by M. L. Oliphant 71 and by E. Pollard andC. J. Brasefield 72 by combining both methods. The values found byA. L. Vaughan, J. H. Williams, and J. T. Tate for the isotopicratios with a mass spectrograph were used in computing the values ofnitrogen and carbon.For the proportion of deuterium in normal hydrogen, a mean valuetaken from the recent researches of H.L. Johnston,T4 of N. F. Halland T. 0. and of N. Morita and T. Titani 76 was used, whichgave H/D = 5550/1. The higher value for lH explains to a greatextent the discrepancy between the chemical and the mass-spectro-graph atomic weights mentioned in the 1934 Report.77 The inter-national value is based on results obtained with electrolytic hydrogen,and hence of low deuterium content-probably about 1 part in25,000.78~ 56It has been recognised for some years that the chemical value forcarbon approximates to 12.01. The new physical values confirmthis, and are in close agreement with W. Cawood and H. S. Patter-son’s determinations of the limiting density of ethylene and of carbondioxide, which lead to 12.0108. A full account of this work, whichgives as well as carbon, nitrogen = 14.007 and fluorine = 18.996,has now appeared.79 It may be noted that the value for carbon7 1 Nature, 1936, 137, 396.72 Ibid., p.943.73 Phy8icd Rev., 1934, [GI, 46, 327.74 J . Arner. Chem. Soc., 1935, 57, 404.75 Ibid., 1936, 58, 1915.76 Bull. Chem. SOC. Japan, 1936,11,404.7 7 Ann. Reports, 1934, 31, 98.78 W. Bleakney and A. J. Gould, Phy8icd Reu., 1933, 44, 365; see also7D Phil. !L’ra128., 1936, A, 236, 77.E. Moles, Anal. Fb. Quint., 1935, 33, 721CARTER AND WARDLAW : BLUOMNNE AND ITS COMPOUNDS. 145becomes 12.008 if Aston's value of 14011 is taken for the 12C/13C ratioinstead of the value 91.6/1 of Vaughan, Williams, and Tahe.E. Moles 8o and his collaborators contend that C = 12.009 is acloser approximation, and some very interesting though preliminarywork by G.P. Baxter and A. H. Hale on combustions in oxygenof the aromatic hydrocarbons chrysene , pyrene, triphenylbenzene,and anthracene supports this figure.R. W A .2. FLUORINE AND ITS COMPOUNDS.When Moissan isolated fluorine in 1896 he opened a new chapter ininorganic chemistry, for the new element had many curious propertiesand above all, an amazing reactivity unique amongst the chemicalelements. Moreover, it was soon realised that this new substancecould produce compounds of the highest theoretical value. :Forforty years this element has continued to excite the liveliest interest,and quite recently A. Damieiis 1 and 0. Ruff,2 distinguished workersin the field of fluorine chemistry, have each told the fascinating storyof fluorine and its compounds as it stands to-day.Although inprevious Reports references have been made to isolated discoveriesas they have arisen, the Reporters consider that no apology isneeded for bringing together now, as a connected account, some ofthe important facts about fluorine and its derivatives.The demand for the element itself has led to the elaboration of newmethods of preparation, with the result that the original method ofMoissan is now very little used. Nevertheless, in surveying thesenew methods it will be noticed that they are all based on Moissan'sfundamental idea of making hydrogen fluoride electrolyticallyconducting by the addition of an appropriate metallic fluoride, anddiffer from it only in the proportions and nature of the fluorideadded.It will be recalled that Moissan's process required a platinumor copper vessel with platinum electrodes and a solution containing1 g.-mol. of potassium fluoride to 12 or more g.-mols. of hydrogenfluoride cooled to - 30". Nowadays two main processes are inuse. The first employs the molten acid fluoride, KHF, (m.p. 227"),in an apparatus of copper: g r a ~ h i t e , ~ ~ i l v e r , ~ magnesium,6 or MonelEo Moncttel~,'l936, 69, 342.1 Bull. SOC. chim., 1936, [v], 3, 1.a W. L. Argo, F. C. Mathers, B. Humiston, and C. 0. Anderson, J . PhysicalChew., 1919, 23, 348; K. C. Denbigh and R. Whytlaw-Gray, J. SOC. C'hem.l r d , 1934, 53, 139.4 F. Meyer and W. Sandow, Ber., 1921, 54, 760; 3'.Fichter and K.Humpert, Helv. Chim. Acta, 1926, 9, 467.6 K. Fredenhagen, D.R.-P., 1928, 493,873.(i N. C. Jones, J. Phyaical Cltem., 1929, 33, 801.Ber., 1936, 69, [A], 181146 INORGANIC CHElVfJ.S‘J%Y.A789101112131uetalY7 using a graphite anode, as first proposed by Mathers and hiscollaborators in America.3 The second process, due to P. Lebeauand A. Damiens,* uses the salt KF,3HF (1n.p. 65O) as electrolyte ina copper apparatus wit,h an anode of iron or preferably nickel.Pi. Fredenhageng advocates the use of an electrolyte containing1 g.-mol. of potassium fluoride to about 1.8 g.-mols. of hydrogenfluoride, whilst E’. C. Mathers and P. T. Stroup lo have found asystem approximating to CsF,ZHP (m.p. 19”) as a satisfactoryelectrolyte in a magnesium cell.There is a marked differencebetween the type of cell employed by some workers and that used byMoissan. For example, Moissan’s U -shaped copper or platinumvessel is replaced in Lebeau’s 8 process by a cylindrical vessel ofcopper, magnesium or Monel metal with a copper or magnesiumdiaphragm. Mathers’s process,3 likewise, uses a cylindrical vesselof graphite, copper or magnesium with a diaphragm. In theelectrolysis the fluorine never separates from the anode without somealteration of the anode surface. When platinum was employed asanode in Moissan’s apparatus, a layer of platinous fluoride wasformed, and over it the platinic salt. The latter dissolved in thefluoride bath, with formation of the difficultly soluble K,PtF,, sothat appreciable quantities of platinum were lost in this way (about5-6 g.of platinum for 1 g. of fluorine produced2). I n Lebeau’sapparatus a layer of nickel fluoride forms on the nickel anode, butthis is so thin that a P.D. of 6 volts may be employed. Here again,in fluorine liberation an intermediate formation of a higher fluoride,possibly Nip3, may take part. When graphite is employed as ananode, fluorine is absorbed on the surface and causes an expansionof the crystal lattice of the graphite.11 As a result, the P.D. inspecial circumstances may increase from 8 to 110 volts. Therebythe anode temperature rises until finally an almost explosive decom-position of the surface layers again frees the surface. W. T. Millerand L.A. Bigelow,12 using a heavy nickel U-tube and graphiteelectrodes, with potassium bifluoride as electrolyte, have recentlyshown that fluorine of 94-9974 purity can be obtained.Amongst the remarkable compounds which fluorine forms withother elements those with the halogens are of considerable interest.N. V. Sidgwick has discussed these interhalogen compounds in asrevious Report,13 and it will be found that if the halogen combiningW. C. Schumb and E. L. Gamble, J . Amer. Chem. Soc., 1930, 52, 4302.Compt. rend., 1925, 181, 917.K. Fredenhagen and 0. T. Krefft, 2. Elektrochem., 1929, 35, 670.T’rana. Electrochem. Xoc., 1934, 66, 113.See ref. 30.J . Amer. Chem. Xoc., 1936, 58, 1685.Ann. Reports, 1933, 30, 128CAWTER AND WARDLAW : F L U O J ~ E AND ITS COMPOUNDS.147with fluorine is its neighbour, then this halogen has covalenciea ofone and three (ClF, CIF,). If, however, one member of the seriesseparates the combining halogens, covalencies of 1, 3 and 5 areshown (BrF, BrP,, ErF,), and finally when two members intervenea covalency of seven is attained (IF7). Much of the data relating tothese substances is based on the masterly researches of 0. Ruff andhis collaborators.In recent years three oxygen fluorides OF2,14 02F2,15 and OF15have been prepared and their properties described, but the questionwhether oxy-acids of fluorine exist or not is still under discussion.It is noteworthy that although OF, is very slightly soluble in water,yet it fails to produce hypofluorous acid.It must not be inferredfrom this, however, that the acid cannot exist. It should be remem-bered that hyponitrous acid (H,N20,) is well known, yet it cannotbe synthesised from water and nitrous oxide. The reaction of OFwith water has not been fully investigated, and as 02F2 is onlystable below - 64", a consideration of its action with water does notarise. L. M. Dennis and E. G. Rochow16 have examined theaction of fluorine on a concentrated solution of sodium hydroxide at- 20" and obtained a solution which is relatively stable, liberatesiodine from potassium iodide, and has a high oxidising power whichis not due to potassium ozonate, ozone, hydrogen peroxide, or theoxygen fluoride OF,. They express the opinion that it containseither hypofluorous acid or more probably fluoric acid.Moreover,these workers state that by the electrolysis of a molten mixture ofpotassium hydroxide and fluoride in a silver crucible with a carbonanode, they have isolated a silver salt which has the formula AgFO,.However, G. H. Cady l7 does not accept the interpretation whichDennis and Rochow place upon their experiments, and suggests analternative explanation which does not involve the existence of anyoxy-acids of fluorine. Ruff considers that the silver salt may bethe complex compound AgQF,.The literature contains references to a number of sulphur fluorides :SF,, SF,, SF,, S,F,, S,Fl,. That SF, exists is incontestable, forit has been prepared and its properties examined by a number ofinvestigators since H.Moissan and P. Lebeau l8 obtained it from thereaction between sulphur and fluorine. Moreover, there is no doubtthat S2F10 is another product of this reaction.19 This is proved by itsmolecular weight of 256. In the case of the other fluorides, how-14 P. Lebeau and A. Damiens, Compt. rend., 1927,135, 652.15 0. Ruff and W. Menzel, 2. anorg. Chem., 1933, 211, 204.16 J . Amer. Chem. SOC., 1933, 55, 2431.l7 Ibid., 1934, 56, 1647.18 Compt. rend., 1900, 130, 865.10 K. G. Denbigh and R. Whytlaw-Gray, J . , 1934, 1347148 fNORGAN1C CHEMISTRY.ever, the published data are very confusing. Ruff and his collabor-ators in 1905 attempted to prepare lower fluorides of sulphur bythe reaction between sulphur nitride, N4S4, and hydrogen fluoride,but without success, and they were also unsuccessful when they triedto decompose sulphur chlorides with certain fluorides.Twentyyears later, however, 0. Ruff and E. Ascher observed the formationof a gas froin cobaltic fluoride and sulphur, and a detailed investig-ation led J. Fischer and W. Jaenckner 2o to the conclusion that theliberated gas was SF,. Their analysis gave S : F = 1 : 3.8 or 3-9,and a molecular weight 107 (SF, reqiiires 108). Unfortunately, thisresult was not reproducible, but a very recent re-examination of thereaction by W. Luchsinger 2 l has clarified the matter. He hasshown that pure SF, is not liberated, and that in all reactions ofmetallic fluorides and sulphur, the fluorides SF,, SF,, S,F,, and oftenalso SP, result in varying proportions according to the kind and.amount of the metallic fluoride and the velocity of the reaction.The lower fluorides S2P, and SF, can be removed from the reactionproducts by shaking with mercury, the SF, can then be fractionatedout, and the gas which remains is practically pure SF,.Thisconstitutes the evidence for SF4. What data are available for thelower fluorides ? In 1923 M. Centnerszwer and C . Strenk 22 obtaineda colourless gas from the reaction between silver fluoride andsulphur. It had a varying molecular weight (93-98) depending onthe temperature of the reaction, and it contained S, 64; F, 35(S2F, requires S, 62.8; P, 37.2y0). Although, as the authors say,‘‘ the agreement is not ideal,” yet they conclude that their substanceis disulphur difiuoride.Recently, M. Trautz and K. Ehrmann23have repeated this work and isolated a gas with a molecular weight of98.8 and containing S, 60.98; F, 39.54%. They consider that theirproduct is a mixture of two fluorides : S,F2, 90%, and SF,, 10%.Up to the present, therefore, the experimental evidence shows thatno one has handled either pure S,F, or pure SF2. Ruff states thathe and his collaborators have obtained a gas containing 90-95% ofS,F2, but it is decomposed by light with the separation of sulphurand the formation of SF,. Moreover, the exceptional reactivity ofboth these fluorides, even with the quartz of the walls of the con-taining vessels, rendered isolation of the pure gases impossible.Ruff is of opinion that pure S2F2 and pure SF, will be obtained ifthe experimental work is conducted exclusively in platinum vessels.2o 2.artgew. Chem., 1929, 42, 810.z1 Diss., Breslau, 1936.22 Ber., 1923, 56, 2249; 1925, 58, 914.23 J. pr. Chem., 1935, [ii], 142, 79CARTER AND WARDLAW : ~ U O R ~ I N E AND ITS COMPOUNDS. 149Many attempts have been made to bring about direct combinationof fluorine and nitrogen, but so far without success. Nevertheless,in 1928 0. Ruff, J. Fischer, and F. Luft 24 succeeded in isolating NF,as a colourless gas (b.p. - 120") by the indirect method of electro-lysing NH4HF,. The yield of NF, was small and it was accompaniedby small quantities of NH,F, some NHF,, and possibly a secondnitrogen fluoride NF,. These compounds, when associated withNF,, apparently bestow on the trifluoride explosive properties, forif the crude gas is led over manganese dioxide the explosive com-pounds are destroyed and the NF3 is quite stable. In this respectNF, offers a marked contrast to the highly explosive NC1,. When-ever NF, or NHF, was present under a pressure higher than ca,350 mm.it exploded, and Ruff emphasises that these substanceshave not yet been obtained pure in spite of prolonged research.He also states that the claims of 0. T. Krefft 25 to have preparedpure NH2F cannot be admitted. Nitrogen trifluoride is veryreactive and when sparked with hydrogen a shattering explosiontakes place,ZNF, + 3H, = N, + 6HF + 336K,and with water vapour a flame travels through the reaction vessel,filling it with deep brown fumes,2NY3 + 3H20 = GHF + Nz03.In 1890 Moissan described several fluorides of carbon, obtained bythe direct union of the elements, but it is only in the last decade thatpure compounds have been isolated.26.,7 On passing fluorine overcarbon, spontaneous combination takes place with considerabledevelopment of heat. The gaseous products of the reaction arecooled in liquid air and subsequently fractionated, whereby thefollowing have been obtained : CF4, C,P,, C3F8, and otherhomologues up to C,F,,. These carbon fluorides are extremelystable to heat and most chemical reagents. Even moderatelyheated sodium does not decompose them. An ethylene analogue,C,F,, has also been identified in the reaction products; and thisis also formed when carbon tetrafluoride is repeatedly subjected to anelectric arc discharge between carbon electrodes.It forms C,F4Br,with bromine-water ; but, apart from such reactions, which dependupon its unsaturated character, it is very indifferent to most reagents.A solid compound, carbon monofl~oride,~~ CP, is also formed during24 2. anorg. Chem., 1928,172, 417.25 D.R.P., 1932, 448,929.26 P. Lebeau and A. Damiens, Conapt. rend., 1926,182, 1340.27 0. Ruff and R. Keim, 2. anorg. Chem., 1930, 192, 249.28 0. Ruff and 0. Bretsohneider, ibid., 1933, 210, 173.28 F. Ebert, see idem, ibid., 1934, 217, 1150 INORGANIC CHIGMISTRY.the interaction of fluorine and carbon, particularly at low pressuresand high temperatures, e.g., carbon in the form of norit at 420" and25 mm.or of graphite a t 420" and 760 mm. The maximum yield inrelation to the other fluorides is about 3% of the carbon converted.Carbon monofluoride is a grey solid, insoluble in ordinary solventsand indifferent to most reagents. It can, however, be reduced byzinc dust and acetic acid, yielding the original form of carbon fromwhich it was derived.- % 7 2 3 4 5 6 7 8 9 70 77 72FIG. 1.The X-ray analysis 29 (Fig. 1) shows that the introduction of thefluorine atoms takes place without any appreciable alteration in thedistance of the carbon atoms in the basic planes of the graphitelattice. The insertion of the fluorine atoms, however, occurs be-tween the basic planes and expands the carbon lattice in a per-pendicular direction from 3-40 A.to 8.17 A. On regeneration, itshrinks to its original size. This enormous distortion of the latticCARTER AND WARDLAW: FLUOR~NE AND ITS C O ~ O U N D S . 151is sometimes the cause of an explosion in the formation of CP, andthe authors have investigated the conditions for its avoidance. Itis not surprising, therefore, that carbon monofluoride i s formedwhen carbon electrodes are used as anodes in the electrolysis offluorides, and considerable expansion is usually observed.30 Thespecific electrical resistance of CF is over 3000 ohms as comparedwith 0.03 ohm for graphite, and considerable passivity effects arethereby set up. 0. Ruff 31 considers that the structure of carbonmonofluoride closely resembles that of graphite oxide, described byU.HofmannF2 in which oxygen atoms are inserted between thecarbon basic planes of graphite.Amongst interesting new metallic fluorides the isolation of ReF,has already been mentioned in an earlier ReportF3 but attentionshould be directed to an investigation undertaken with the object ofisolating a higher fluoride of copper and silver.34 No fluoride ofcopper higher than CuF, could be prepared, but the existence ofAgF, is now well e~tablished.~~ This compound has been preparedby the action of fluorine at 150-200" on a silver halide, or molecularsilver prepared by reducing silver oxide with formaldehyde, or onfine silver gauze. It is a dark brown powder which with water givesoxygen containing ozone. The fluorine is relatively firmly bound,for only at 440" does the dissociation pressure of argentic fluoridereach 1 atmosphere.This fluoride is strongly paramagnetic, as isto be expected from its electronic structure. It is an excellentfluorinating agent and in many cases can act as a substitute forfluorine.Some extraordinary mistakes by earlier observers are disclosed inrecent publications on the acid fluorides of the metals. I n 1905,E. Bohm 36 recorded the preparation of acid fluorides of cobalt,nickel, and copper of the general formula MF2,5HP,5 or 6H,O.Sixteen years later, F. H. Edminster and H. C . Cooper 37 publishedtheir conclusions from a reinvestigation of these substances. Theystated that the correct formula was MF2,5HP,6H,O, where M couldbe Coy Ni, Cu, or Mn, but added that " It was a surprise to obtainthe acid fluoride by recrystallisation from water." However, rz30 Seeref.11.32 Ber., 1932, 65, 1821 ; U. Hofmann, A. Frenzel, and E. CsalBn, Annnlen,33 Ann. Ileprts, 1933, 30, 91.34 0. Ruff and M. Giese, 2. anorg. Chem., 1934, 219, 143.35 Idem, ibid.; M. S. Ebert, E. L. Rodowskaa, and J. C. W. Frazor, J.Amer. Chern. SOC., 1933, 55, 3056; Naturiuiss., 1934, 22, 561.36 2. anorg. Chem., 1905, 43, 326.57 J . Amer. Chern. Soc., 1920, 42, 2419.Angew. Chern., 1933, 47, 739.1934, 570, 1152 TNORGANIC! CHEMISTRY.recent, investigation by A. Ifurtenacker, W. Finger, and F. Hey 38leads to a very different conclusion. They find that the so-calledacid fluorides of the type MF,,5HF,xH20 do not exist, but that theyare really fluosilicates and must be eliminated from the literature.The story, however, does not end here.According to an earlyinvestigator, 1. E. Willm,39 an acid fluoride HTIF, exists, but in1920 J. Barlot 40 re-examined the compound and concluded that itwas H2T1F,. 0. Hassel and H. Kringstad 41 then took up the matterin 1932 and proved that the formula was really H,T1F3,0.5H,0,and that the substance was isomorplious with a compoundM,(NH,)P,,O-5H2O which they prepared. This year C. Finbak and0. Hassel 42 published an account of a reinvestigation of a numberof so-called acid fluorides. They agree with Kurtenacker and hiscollaborators that the recorded acid fluorides of nickel, cobalt,manganese, and copper are really fluosilicates and, what is par-ticularly interesting, find that the acid fluorides of fhallium also donot exist but are actually fluosilicates.Moreover, a mercurouscompound Hg,F,,4HF,4Hz0, described by E. Bohm36 in 1905, isfound to be Hg,SiF6,2H20. It is obvious that the greatest care isnecessary in investigations where hydrofluoric acid is employed.It is noteworthy that quite recently J. Meyer and W. Taube43state that they have prepared RbHl?, and RbH,F3, but that thesesubstances are commonly contaminated with fluosilicates.Attention should be directed to an ingenious method for deter-mining critical temperatures which has been applied with success toanhydrous hydrogen fluoride by P. A. Bond and D. A. Williams.44A tube of Monel metal is charged with the pure hydrogen fluorideand heated in an air-bath above its critical temperature.Thistube, in an inclined position, is mounted on knife edges and con-nected a t one end by a fine platinum wire to the beam of a balance.The system is then cooled slowly and the critical temperature isclearly indicated by a sudden change in the balance-beam when thecontinuous phase passes to a system of vapour and liquid, withconsequent displacement of the centre of gravity. The criticaltemperature thus determined is 230.2".S. R. C.w. w.38 2. anorg. Chem., 1933, 211, 83.39 Ann. C?&n. Phys., 1865, [iv], 5, 6 .4 0 Cornpt. rend., 1920, 171, 1143.2. anorg. Chem., 1932, 208, 382.*? Ibid., 1936, 226, 175.Ibid., 1936, 227, 337.44 J . Arner.Chem. SOC., 1932, 54, 129CARTER AND WARDLAW : METALLIC CBRBIDES. 1533. METALLIC CARBIDES.Results of great value have been obtained during recent years bythe application of X-ray methods to the structure of the carbides.These substances have long presented a problem in molecularstructure of the very highest interest, for the attention of chemistshas always been arrested by the extraordinary differences inbehaviour exhibited by the metallic carbides in their reactions bothwith water and with acids. In certain cases the reaction productsare of unexpected complexity, and many explanations have beenadvanced to account for these results.There is evidently some connection between the behaviour of acarbide and its position in the periodic table.Those of the first andsecond groups, with the exception of beryllium carbide, give withwater, acetylene only. Beryllium carbide, Be&, and aluminiumcarbide, AI,C,, yield pure methane, but manganese carbide, Mn,C,gives a mixture of methane and hydrogen in equal volumes. Car-bides of yttrium, lanthanum, cerium, and the other rare earths, alsothose of uranium and thorium., are attacked by waher and yield acomplex mixture of gaseous, liquid, and sometimes solid hydro-carbons.In seeking an explanation of these results some early workers onthe carbides did not hesitate to represent them by graphic formule,but others, more cautious, endeavoured to reconcile the formulzeof the carbides with the valency of the constituent elements. Duringrecent years, as a result of the work of M. von Stackelberg and hiscollaborators, the constitution of a number of metal carbides hasbeen disclosed by the use of X-ray methods, and the relationshipbetween chemical reaction and structure made clear for the firsttime.The well-known reaction of calcium carbide with bromine, andfurther, its decomposition with water, have convinced chemists thatin this carbide the two atoms of carbon must be joined :In 1930 von Stackelberg studied the acetylides of the alkali metals,and the carbides of the alkaline-earth metals and the rare-earthmetals. The structure which he finally assigned to CaC,, UC,,Lac,, PrC,, and NdC, is most conveniently described as a face-centred cubic lattice or sodium chloride lattice made up of metalatoms amnd C, groups.1 Z.physsikal. Chern., 1930, B, 9, 437154 INORGANIC CHEMISTRY.As Pig. 2 shows, thc C, groups are all arranged parallel to one edgeof the cube, which is thereby extended in this direction. The disfor-tion may be expressed by an axial ratioc/a of 1*15-1.20. It is not surprising,therefore, that acetylene is liberatedas a primary product from the decom-position with water, for already inthe crystal lattice a C*C bond is present.Prom the fact that acetylene is thesole gaseous product in the case ofcalcium carbide, it must be concludedthat the energy conditions are favour-able for the evolution of this gas. Withthe carbides of uranium and of therare earths, acetylene is evolved butit is accompanied by saturated andunsaturated hydrocarbons of greatcomplexity. The composition (%) ofthe gases evolved from some of these carbides is given by H.Moissan 2 as follows :o=ca @=CFIG.2.La. Ce. Pr. Nd. Sm. Y. Th.C,H, ............... 71 75.5 67.9 66.3 70.6 71.7 47.7C,H, ............... 1.5 4.3 3.0 6.3 7.9 4.6 5.8CH, .................. 27.9 20.3 29.1 27.6 21.5 18.9 29.4H, .................. - - - - 4.7 17.1Later investigations,3 whilst not confirming these analytical resultsin detail, have only served to emphasise the complexity of thegaseous mixtures. In a recent communication N. G. Schmahl*has put forward a theory of carbide hydrolysis based on theseresults of Moissan. Schmahl considers that the following reactionoccurs in the decomposition of the rare-earth metal carbides bywater :4XC2 $- 6H,O = 2X20, (hydrated) + 3C,H2 + CH, + CH,According to Schmahl this CH, radical is changed into ethylene andpropylenc or t o methane and ethane, depending on the heat of thereaction.J. Schmidt has examined this suggestion, and pointedout that Schmahl’s theory demands the formation of acetylene andmethane in a ratio 3 : 1, which would be lowered in those cases wherethe CH, radical is reduced to methane. Actually, if the figuresA. Damiens, Compt. rend., 1913, 157, 214; P. Leboaii and A. Damiens,2 Ann. Chim. Phys., 1896, [vii], 9, 302.ibid., 156, 1987.4 2. h’lektrochem., 1934, 40, 68.Ihid., p. 170CARTER AND WARDLAW : METALLIC CARBIDES. 155given by Moissan are examined, it will be seen that the ratioC,H, : CH, is sometimes greater than 3 and sometimes less.Aboveall, however, Schmahl’s theory fails to recognise that in the crystallattice a C-C bond is already present. In Schmidt’s view thesuggestion of Schmahl is untenable, and there appear sound groundsfor this conclusion. Instead, Schmidt considers that the differencein the reaction products from calcium carbide and the rare-earthmetal carbides is due to the different valencies of the metal atoms.If the metal atom, My is bivalent, as in calcium carbide, then thecarbon is bound as acetylene and the metal as the hydroxideM(OH),. If, however, the metal is in the tervalent state, hydrogenis liberated, and i t is the further reactions between the acetylene andthis hydrogen which produce the complicated mixture of hydro-FIG.3.carbons. Schmidt is of opinion that methane itself is a side productof such reactions. He considers that his views are supported bythe carbides of the tervalent elements aluminium and cerium.Carbides of these elements can be prepared of the general formulaX,(C E C), in which the metal atom has a valency of Onhydrolysis, acetylene alone is liberated, because the metal atonisconcerned do not undergo a valency change but are stable asAl(OH), and Ce(OH),. Reference to the preceding table will showthat the composition of the gaseous products from the hydrolysisof thorium carbide, ThC,, offers a striking contrast to that from theother carbides included. It might be anticipated, therefore, thatthis difference in chemical behaviour would be reflected in thestructures of these carbides, and it is interesting to find that this isso.The structure of ThC, (Fig. 3), although similar to that of the6 J. F. Durand, Bull. SOC. cl~im., 1924, 38, 1141; L. Damiens, tibid., 1914,15. 370156 INORGANIC CIIEMISTRY.CaC, type, has the C, groups all parallel to a cube face but with theiraxes in two mutually perpendicular directions. The distortionproduced gives rise to a tetragonal lattice with an axial ratioc : a = 0.903.Particularly interesting are the detailed structures which havebeen evolved for the carbides of aluminium and beryllium, Al,C,and Be,C, for they explain in a most convincing way why methaneis the sole hydrocarbon produced by hydrolysis. Beryllium car-bide, Be2C,7 has been shown t o have an antifluorite structure, inwhich each beryllium atom is surrounded by four carbon atoms andeach carbon atom has eight beryllium atoms as neighbours.TheC-C distance is 3.06 A., and Be-C is 1.9 A. This separation of thecarbon atoms in the crystal lattice, combined with the bivalency ofthe beryllium atoms, explains most effectively why methane resultsfrom the hydrolysis.is complicated, and is best understood by reference to the diagramsin the original publication. However, it can be described as a layerlattice in which three layers of carbon atoms are interleaved withfour layers of aluminium atoms so that each unit has the composition(Al4C3lW and is, in itself, saturated with regard to valency.Eachaluminium atom is siirrounded by four carbon atoms at a distanceof 1.9-2-0 A., while the carbon atoms have either 5 or 6 aluminiumatoms as neighbours. With such a structure the formation ofmethane, by hydrolysis of the carbide, would be expected. It iswell established that hydrogen and a number of hydrocarbons areproduced when either Ni,C or Fe,C is treated with hydrochloric acid.To explain this result, J. Schmidt assumes that the CH, group isformed initially, and it then undergoes hydrogenation to methaneor polymerisation to ethylene. This ethylene can then be furtherchanged by hydrogenation or renewed polymerisation or reactionwith a new CK, group. I n the decomposition of Fe3C, polymeris-ation must occur to a marked degree, for in addition to gaseoushydrocarbons, liquid and solid hydrocarbons may be f ~ r m e d .~H. A. Bahr and T. Bczhr lo state that they observed the formationof liquid hydrocarbons in the decomposition of Ni,C with hydro-chloric acid, but J. Schmidt could not confirm this. Incidentally,the decomposition of Fe,C is complicated by the separation ofelementary carbon, which is accelerated by ferrous ions.11 A carbonseparation can also take place when Ni,C is decomposed, but it hasThe detailed structure of aluminium carbide7 M. von Stackelberg and F. Quatram, 2. physikal. Chem., 1934, B, 27,50.8 M. von Stackelbsrg and E. Schnorrenberg, ibid., p. 37.9 E. D. Campbell, Amer. Chem. J., 1896, 18, 836.lo Ber., 1928, 61, 2177.11 R.Schenck and R. Stenkhoff, Z. anorg. Chem., 1927,161, 287WARDLAW : CO-ORDINATION COMPOUNDS. 157not yet been demonstrated that it can be influenced by concentrationof nickel ions.5An X-ray investigation of these carbides does appear to throwsome light on their behaviour with acids, for it indicates that theindividual carbon atoms in the lattice of Fe,C and Ni3C l2 are separ-ated from one another, and this implies the possibility of theformation of a primary CH, group. This is especially so in Fe,C,for its method of preparation and its existence in equilibrium withiron (ferrite) indicate that the metal in the carbide is in the lowerstate of oxidation, i.e., bivalent. By analogy one might assume asimilar method of decomposition for Mi@ but, as is well known,methane and hydrogen are the only products.The crystal structureof Mn,C is not known accurately, but evidently it must be funda-mentally different from that of Fe3C. F. Fischer and I?. Bangert 13have prepared a manganese carbide of a different type by thereaction of manganese oxide and methane. When it was decomposedwith water, the evolved gas contained 2.5% of unsaturated hydro-carbons, 45% of saturated hydrocarbons, and 52.5 % of hydrogen.The unsaturated hydrocarbons contained on an average at leastfour carbon atoms. This carbide agrees with the formula (Mn,C,),.A. Westgren l4 has recently examined the crystal structure of amanganese carbide to which he assigns the composition Mn,C3.A magnesium carbide, Mg,C,, is known to give pure allyleneCH, - C = CH on hydrolysis but its crystal structure is unknown.15The fact that only one hydrocarbon is formed suggests, by analogywith the carbides of the CaC, type, that the C-C-C bond is alreadypresent in the crystal lattice.S.R. C. w. w.4. CO-ORDINATION COMPOUNDS.The old idea that the term co-ordination compound is restrictedto the well-known cobalt, chromium, and platinum ammines isundoubtedly disappearing, and it is being realised that these complexsalts represent only a special section of an exceedingly wide anddiverse class of substances. In an introduction to the symposiumon complex inorganic compounds arranged by the AmericanChemical Society, L. F. Audrieth l pointed out how recent develop-la B. Jacobsen and A.Westgren, 2. physikal. Chern., 1933, B, 20, 361:J. Schmidt, 2. anorg. Chem., 1933, 216, 85.l3 Breninsto$-Chem., 1929,10, 261.11 Jernk. Arm., 1935, 118, 231.1 Chem. Reviewe, 1936, 19, 65.l5 See N. GI-. Schmahl, ref. (4)158 INORGANIC CHEMISTRY.nients in theoretical chemistry and in research technique had givennew prominence to this field of investigation. The Bronsted conceptof acid-base equilibria has directed attention to complex ions in thedevelopment of acidic and basic properties in solution, whilst thecver-changing theories of atomic structure have demanded investiga-tion of the physical and chemical properties of compounds char-acterised by the covalent link. Interest in the modern theories ofoptical activity has stimulated investigations of the optical propertiesof complex inorganic compounds, whilst improved apparatus andthe development of new research technique, such as are available inX-ray and electron-diffraction methods and in measurements ofdipole moments, have thrown a flood of light on the detailedstructure of co-ordination compounds.A few years ago it was quite true to say that the available evidenceindicated that the arrangement of the valencies of 4-covalent atomswas almost always tetrahedral.This was the general conclusionarrived a t by the classical methods of stereochemistry and it wassupported by a substantial mass of physical data. I n the discussionon modern stereochemistry, held by the Chemical Society, S. Sugden2reviewed recent work on 4-covalent complexes of bivalent nickel,palladium, platinum, and copper, and showed that in the case ofthese elements there is definite evidence for the frequent occurrenceof a planar configuration.Sugdcn pointed out that there are nowthree main lines of evidence which lead to this conclusion in the casesof n i ~ k e l , ~ palladium: and p l a t i n ~ m . ~ Erst, there is the occurrenceof cis-trans-isomerides when two unsymmetrical chelate groups area.ttachcd to the metallic atoms. Examples of this are found in theisomcric nickel derivatives of benzylmethylglyoxime andisomeric glycine derivatives of palladium and platinum (I) andScconclly, there is the evidence provided by the resolution4-covalent compound of bivalent platinum. This work of W.H.Mills and T. H. €1. Quibell was discussed in the Report last year.Finally, there are the X-ray studies by E. G. Cox, W. Wardlaw, andcollaborators who, in the last two years, have demonstrated aa Nature, 1936, 137, 543.4 F. W. Pinkard, E. Sharratt, W. Wardlaw, and E. G. Cox, J . , 1934, 1012.5 A. A. Grunberg and B. W. Ptizyn, J . pr. Ohm., 1933, [ii], 136, 143.6 cJ., 1935, 830.7 See E. G. Cox, F. W. Pinkard, W. Wardlaw, and K. C. Webster, J . ,S . Sugden, J., 1932, 246.1‘338, 459; E. G . Cox, W. Wardlaw, and K. C. Webster, J., 1936, 1476WARDLAW : CO-ORDINATION COMPOUNDS. 159planar configuration in no less than 14 derivatives of nickel, palla-dium, and platinum. Thus there is, in Sugden's view, abundantevidence for planar structures in co-ordination compounds of4-covalent nickel, palladium, and platinum.Obviously there must be a theoretical explanation why in certaincases the structure is tetrahedral, and in others planar, and thereader will probably recall that in 1931 L.Pauling * put forward atheory to explain this. In connection with this theory it should benoted that the electrons in an octet are divided into two sub-groupsof 2 and 6, and those of an 18 group into three sub-groups of 2, 6,and 10. The values of the second quantum number for these sub-groups are II: = 1, 2, 3, and the spectroscopic designations of thethree are s, p , and d respectively. Pauling accepts the idea that abond involves two electrons, and he lays down that when the linksare tetrahedrally distributed these bonds are compounded ofclectrons in s and p levels.GeneraJly, it might be said that thelinks are of the form sfi3. When however onc d level is used and thelinks are of the type spad, Pauling predicts an arrangement of fourbonds at 90" in one plane. Such vacant d levels, with energiescomparable with s and $3 levels, are found in the atoms of thetransition elements. Moreover, since the d electrons are chieflyresponsible for the magnetic moment of the atom, sharing of themshould reduce this property so that bivalent nickel, which is para-magnetic in its simple salts, should becomc diamagnetic in its4-covalent planar compounds. It is interesting to find that therecorded magnetic evidence for nickel agrees with Pauling's theory,for all the planar compounds investigated have proved to bediamagnetic.With palladium and platinum, the magnetic evidencedoes not appear to be so significant, for all simple and complex saltsof these elements are diamagnetic, but the diamagnetism of theplanar compounds is in accord with Pau1ing:'s views. Sugdenindicates, however, that SL real discrepancy arises with certain cupriccompounds. E. G. Cox and K. C. Webster ti have found from X-raystudies that the cupric derivatives of p-diketones have a planarconfiguration. These compounds are paramagnetic, and there isnot a vacant 3 d level in the cupric ion. This is readily seen from theelectron distributions for nickel and copper atoms, which show thatthe difference between nickel and copper is not merely the additionof an electron in the nickel atom but a change in the number ofelectrons in the d level :Ni (At.no. 28) ...... Is2 282 2p6 388 3p6 3d8 48'(At. no. 29) ...... 28* 2p6 h2 3p6 3d1° 48l8 J . Amer. Chm. Soc., 1931, 58, 1367.J . , 1936, 731160 INORGANIC CHEMISTRY.The planar configuration for 4-covalent cupric compounds is nowfirmly established. J. M. Robertson 10 found that in the phthalo-cyanines of nickel, platinum, and copper the metal atom and thefour surrounding nitrogens lie in one plane. Again, an X-rayinvestigation of the crystal structure of cupric chloride dihydrate,CuC12,2H20, by D. Harker,ll shows that each copper atom isattached to two chlorine and two oxygen atoms by covalent linkingsdirected to the corners of a square.To illustrate how X-ray resultscan give evidence of this planar structure in simple derivatives of4-covalent copper, details of a recent investigation l2 of CuC12py2may be given. This substance, which crystallises well frommethyl alcohol, was prepared under a variety of experimentalconditions with the object of discovering the theoretically possiblenWFIG. 1.cis- and trans-isomerides. Under all conditions it was found tocrystallise in the same trans-planar form. It is evident that 4-covalent copper compounds of cis- configuration, like those ofpalladium, are generally unstable when chelate groups are absent.That CuC12py2 has a trans-planar configuration follows from aconsideration of its cell dimensions as determined by X-ray methods.The short length of the c axis (3.84 A.) shows that the pyridine ringsmust be coplanar (or very nearly so), since this value is of the sameorder as the distance of approach of =CH- groups in differentorganic molecules (m.3-7 A.). Reference to a model shows thatthis can only be so in a molecule of trans-planar configuration (Fig. 2)for with it cis-planar configuration the parallel arrangement of thepyridine rings shown in Fig. 1 is impossiblo on account of thelo J., 1935, 613.18 E. G. Cox, E. Sharratt, W. Wardlaw, and K. C. Webster, J., 1936, 129.l1 2. K h t . , 1936, 93, 136WARDLAW : CO-ORDINATION COMPOUNDS. 161proximity of the two rings (minimum distance 1.9 A. instead of3.7 A.). In order to obtain the requisite clearance between therings in a cis-planar structure they must be rotated about theN-Cu bonds through approximately 40°, thus increasing the thick-ness of the molecule and necessitating a c-axis of at least 4-5 A.FIG.2.This is impossible in view of the experimental datum that c is 3-84 A.On the same grounds, a molecule in which the distribution of coppervalencies is tetrahedral is excluded. Since the j ~ yMe:NO\] distribution of valencies in nickel and copper>M 4-covalent compounds may be planar, and themetal atoms are not greatly different in radius, it ’\ (111.) is to be expected that corresponding nickel andcopper compounds will sometimes be iso-morphous. This has proved to be the case with the methylethyl-glyoxime derivatives (III).12 In view of Sugden’s findings 13 thatthe nickel derivatives of unsymmetrical glyoximes are planar, thisresult idds additional weight to the conclusion that 4-covalentcopper may be planar in its cupric compounds.In a paper on the stereochemistry of the metallic phthalocyanines,R. P.Linstead and J. M. Robertson l4 have shown that bivalent4-co-ordinate beryllium, manganese, iron, and cobalt, like nickel,copper, and platinum, all exhibit planar symmetry in the crystalsof their phthalocyanine derivatives (see Pig. 3). The fact thatcobalt is tetrahedral l5 in the group CoC1,” makes the result with thcphthalocyanine derivatives particularly interesting, and providesthe first example of this metal exhibiting planar symmetry. Themost remarkable of these results, however, is provided by beryllium,for which a tetrahedral symmetry is well established by bothchemical and physical considerations.Beryllium is tetrahedral1s H. J. CavellmdS. Sugden, J., 1935, 621.14 J . , 1936, 1736.16 H. M. Powell and A. F. Wells, J., 1935, 359.1 CEt*N,OH j2REP.-VOL. XxxIU. B162 INORGANIC CHEMXSTRY.in its bcnzoylpyruvic acid derivative l6 and in basic berylliumacetate,17 and BaBeF, is isomorphous with BaS04.1S That theFIG. 3.very simple atom beryllium, which normally contains no d electrons,should adopt a planar distribution of valencies, appears inexplicableon Pauling’s theory. Linstead and Robertson point out that thetheoretical difficulty can be avoided by the assumption that inberyllium phthalocyanine the metal is combined with only twonitrogen atoms, but there seems no justification for arbitrarilydifferentiating between this cornpoiind and the other covalentmetallic phthalocyanines which resemble it so closely in crystallineform.The general conclusions from these facts are : (1) that in themetallic phthalocyanines, and probably also the correspondingporphyrins, the rigid planar organic portion of the molecule imposesits steric requirements upon the metal, and (2) that there is moretolerance in the distribution of valencies about 4 co-ordinate-metalatoms than has hitherto been realised.The work already surveyed has shown that many 4-hovelentcomplexes of copper, nickel, palladium, and platinum actuallypossess a planar configuration, and that this configuration ismaintained even when considerable changes are macle in the natureof the co-ordinate groups.On the other hand there is definiteevidence that a change in the principal valency of the metal maylead to a change in the spatial distribution of the bonds. Forexample, trimethylplatinic chloride, Pt (CH,),Cl, in which platinumis 4-covalent but also quadrivalent, has been shown l9 to be non-planar and is most probably tetrahedral. Again, nickel carbonyl,Ni(CO),,20 in which the nickel has no principal valency, possesses a16 W. H. Mills and R. A. Gotts, J., 1926, 3121.17 (Sir) W. Bragg and G . T. Morgan, Proc. Roy. SOC., 1923, A , 104, 437.18 N. N. Ray, 2. anorg. Chem., 1931, 201, 259.19 E. G. Cox and K. C. Webster, Z.Krist., 1936, A, 90, 561.20 L. 0. Brockway and P. C. Cross, J . Chem. P~Q&cL?, 1935, 3, 825WARDLAW : CO-ORDINATION COMFOUNDS. 163tetrahedral structure. Eecenf investigations have provided addi-tional examples of this change of structure with change in theprincipal valency of the central element. For instance, in strikingcontrast to the planar 4-covalent dcrivahives of bivalent copper, ithas been proved that some 4-covalent cuprous compounds aretetrahedral. Similarly, silver has been found to be planar whenbivalent and tetrahedral when univalent. Prior to this year, theonly evidence on the configuration of either 4-covalent cuprous orargentous derivatives mas that of F. Hein and H. Regler, who, in apreliminary note,21 claimed to have effected a partial resolutionof an argentous derivative of S-hydroxyquinoline. Since then,these authors 22 have described their work in detail, and state thatby the use of bromocamphorsulphonate they have obtainedevidence that their silver compound is optically active and musttherefore have a tetrahedral disposition for the valency bonds of thesilver atom.They were unable however to retain the activitywhen the optically active acid was removed.+ I-INO, eE. G. Cox, W. Wardlaw, and I<. C. Webstter 23 imve examined byX-ray methods potassium cuprocyanide, K&u( CN),], and theisomorphous substances t etrakis - t hioacetamide cuprous andaxgentoua chlorides [as (IV)]. In eaeh of these substances a tetra-hedral valency distribution bas been found, and it will be noted thatin every case the effective atomic number of the metallic atom isthat of an inert gas.It is also noteworthy that the complex ionCu(C"),"' has exactly the same number of electrons, and pre-sumably the same electronic distribution, as tho neutral complexNi(CO),, which liz~s been shown to be tetrahedral. This tetrahedraldistribution of valencies in a co-ordinated cuprous complex has alsobeen found by F. G. Mann, D. Purdie, and A. P. Wells 24 in themolecule [Et3As->CuI),. Silver, with SL principal valency of two21 Nntzcrwiss., 1935, 23, 320.82 Ber., 1936, 69, 1692.a3 J., 1936, 775.14 Ibid., p. 1503164 INORGANIC CHEMISTRY.and a covalency of four, has proved 23 to be planar as a result of anX-ray examination of the argentic derivative of picolinic acid.Thiso=c M c=oPV.)result is based on the high birefringence of the compoundisomorphism with the corresponding copper derivative, which hasbeen shown to be trans-planar as in (V). Further, E. G. Cox andK. C. Webster 25 have demonstrated that potassium auribromide,K[AuBr4],2H20, in which gold is tervalent, possesses the ion AuBr4‘with a planar configuration. The X-ray evidence shows that thewater in the auribromide is held mainly as water of crystallisation,and that the substance does not contain the sexacovalent complex[AuBr,,2H20]’. This appears to be the first example of a tervalentmetal with a planar distribution for its four valency bonds. Itwill be realised that a definite advance in the stereochemistry ofthese so-called currency metals has been made.Some years agoR. Dickinson 26 investigated the complex cyanides of potassiumwith zinc, cadmium, and mercury of the type K,[X(CN),], and foundthat the co-ordinated cyanogen groups had a tetrahedral arrange-ment around the central metal atom. W. H. Mills and R. E. D.Clark 27 have prepared compounds of the type (VI), where M = Hg,Cd, and Zn, in order to investigate the stereochemistry of thesemetals in the 4-covalent state. Various alkaloid salts were in-vestigated, but in no case could direct evidence of their opticalresolution be obtained. Nevertheless, other results of great interestwere recorded, but as these cannot be satisfactorily summarised,the original papers must be consulted.Outside the range of the transition elements, the stereochemistryof 4-covalent tin and lead has been considered.A tetrahedraldistribution for the four valencies of stannic tin is fimly establishedby chemical and physical methods. An optically active compoundof tin was prepared by W. J. Pope and S. J. Peachey 28 in 1900, and2s J., 1936, 1636.26 J. Amer. Chem. SOC., 1922, 44, 774.a7 J., 1936, 175.28 P., 1900, 16, 42, 116WARDLAW : CO-ORDINATION COMPOUNDS. 165this tetrahedral configuration was also revealed in the four atoms ofiodine which surround the metal in stannic iodide.29 Until recently,however, no attempt had been made to determine whether thistetrahedral distribution also holds for 4-covalent compounds ofstannous tin.The preliminary results of an investigation ofK2[SnC1,],2H,0 have been published,30 and they show that incontrast to SnI, the complex ion SnC1,” is planar. The X-rayresults plainly prove that the tin is 4-covalent, an\d not sexacovalentas the presence of the two molecules of water might suggest. Otherstannous compounds lead to a similar conclusion. Further, lead,which is tetrahedral in the 4-covalent and quadrivalent leadcompound, Pl~Ph,,~l proves to be planar in its bivalent and 4-coval-ent derivatives such as lead benzoylacetonate, lead salicylate, andPbC12,2CS(NH,)2.30 This short summary indicates that investig-ations on 4-covalent compounds are yielding results of considerableinterest and valuc.Although the structure of compounds of the type AB, has notyet been determined, suggestions have been made as to possibleconfigurations.Iron pentacarbonyl, Fe( CO),, is a substance ofthis type, and two possible structures have been advocated. Thefirst is a tetragonal pyramid in which the apical carbon monoxidemolecule is further removed from the central atom than the otherfour ; such is the structure advanced by W. Graff under and G. Hey-mann 32 to explain the small dipole moment. Thc second, whichJ. S. Anderson 33 considers more probable, is a trigonal bi-pyramid.It provides the closest packing possible in a 5-covalent arrangement,and affords a satisfactory explanation of the ready formation ofFe,(CO), as formulated by N. V. Sidgwick and R. W. Bailey 34 anddescribed in a previous Report.35Anomalies in the parachor of co-ordination compounds wereobserved by S.S ~ g d o n , ~ ~ who found that beryllium in the basicpropionate Be,O( C,H,*CO,), and t’he acetylacetonate appearedto have a variable negative parachor. Co-ordination compounds ofthallium and aluminium displayed a similar anomaly, and in lastyear’s Report attention was directed to some results of F. G. Mannarid D. P~rdie,~’ who found that in certain series of organic metallic29 It. G. Dickinson, J . Amer. Chem. Soc., 1923, 45, 958.30 E. G. Cox, A. J. Shorter, and W. Wardlaw, Nature, 1937, 139, 72.31 W. H. George, Proc. Roy. SOC., 1927, A, 113, 585.32 2. physikal. Chem., 1932, B, 15, 377.33 J , , 1936, 1283.34 Proc. R o g . SOC., 1934, A, 144, 521.36 “ Parachor and Valoncy,” London, 1930, p.145.37 ?7., 1935, 1549.Awn. Reports, 1934, 31, 1041.66 INORGANIC CIIEMISTRY.compounds, both simple and complex, the metal atom showed anapparent parachor which fell steadily as the homologous series wasascended. For example, in the homologous series PdC1,,2R2S theparachor of palladium fell from 36 for the methyl to - 7 for then-amyl compound. Sugden sought to obviate the anomaly in thecompounds of beryllium, thallium, and aluminium by a singlet-linktheory of co-ordination ; in the case of palladium derivatives,however, the deficit cannot be explained by substituting a singletlinkage for the co-ordinate linkage, but must be regarded as a realeffect. Mann and Purdie suggested that the effect might beexplained, in part a t least, as due to the molecular shape, since thetrans-planar arrangement of groups about +,he palladium conEers amolecular configuration which might well be associated withanomalous packing effects.This explanation may account in somemeasure for the results of the parachor in the homologous seriesconsidered by Mann and Purdie, but it cannot be regarded ascomplete in view of some new results by J. S. Anderson 33 on theparachors of metal carbonyl compounds. These compact, non-planar molecules show a large anomaly in their parachors. Nickelcarbonyl, which is tetrahedral,20 and has a close-packed structure,gives a parachor which is a few units greater than four times theobserved parachor of carbon monoxide. For iron pentacarbonyl theobserved total parachor is less than four times that of carbon mon-oxide, I n the ca'rbonyls the available evidence indicates thatthe carbon-oxygen linkage differs very little from that in carbonmonoxide, BO that the assumption that the parachor of co-ordinatedGO is the same as that of free carbon monoxide should give anapproximate value for the parachor of the metal atom.Theparachors of the tricarbonylnitrosylcobalt, Co( CO),(NO), anddicarbonyldinitrosyliron, Fe(CO),(NO),, have also been determined.The assumption is made that as NO must be closely related to COin the nitrosocarbonyls, it is therefore reasonable to calculate theparachor on the relation PNO - 2'00 = PN - Pa. The parachoraof the metals calculated in this way are given in the followingtable :Ni(CO),.Co(CO),NO. Fe(CO),(NO),. Pe(CO),. CO.249.8 252.5 300.6 61.6 P ... ... ... ... ... ... ... ... 255.3Apparent P of metal 8.9 - 8.9 - 18.5 - 7.4From Sugden's atomic number-parachor curve, iron, cobalt, andnickel should have nearly equal atomic parachors of about 50(Cr = 54-3, Cu = 46). It is plain that all four substances show alarge deficit, just as do the co-ordination compounds of beryllium orpalladium. No adequate explanation of this anomaly has yet beenadvancedWARDLAW : CO-ORDINATION COMPOUNDS. 167I n $1) c most stable derivatives of cuprous copper the co-ordinationnumber appears to be four, als, e . ~ . , in K,[Cu(CN),]. It is interestingto find, therefore, that cuprous and silver iodides form compounds 24with tertiary phosphines and arsines analogous to the well-knownnon-ionic aurous chloride derivatives, R,P(As)+ AuCI, and thatmany members of this series possess considerable stability in spiteof the fact that the metal atom has apparently a co-ordinationnumber of 2.Actually, one would have expected the stable com-pounds to be of the type (PR,),CuI, where copper would have anatomic number of 36 and thereby attain the electronic structure ofkrypton. X-Ray analysis 24 of the arsine compound shows that thetrue molecule is not R,As+CuI, but that it really consists of foursimple units (Et,As+CuI),, and this is in accordance with molecular-weight determinations by the cryoscopic method. The phosphinederivative Et3P+ Cul is strictly isomorphous with the triethyl-arsine compound and has therefore the same structure. The detailedstructure shows that the four cuprous atoms are arranged at theapices of a regular tetrahedron and the four iodine atoms lie eacha t the centre, but above the plane, of one face of the tetrahedron.Beyond each cuprous atom is an arsenic atom lying on the elongationof the axis joining the centre of the tetrahedron to the copper.The three ethyl groups are then joined to each arsenic atom, so thatthe tetrahedral angle is subtended both a t the arsenic and at thefirst carbon atom of the ethyl groups.The stability is conferred byeach iodine atom, in addition to being covalently linked to itsoriginal copper atom, also being joined by two co-ordinate links tothe other two copper atoms of the same tetrahedral face.Eachcopper atom acquires seven electrons and is identical, therefore, inboth co-ordination number and electronic structure, with that inBenzoinoxime, as is well known, i s one of &number of organicreagents employed in the detection and estimation of metals. Thecopper derivative, discovered by F. Feig1,38 is a deep green amorphouscompound, insoluble in water and organic solvents, to which formula(VII) was assigned. Until recently, however, this structure couldnot be regarded as fully established. It is unusual to find thehydrogen atom of the secondary alcoholic group replaced bycopper; moreover, benzoinoxime is a reducing agent, and onewould expect it to reduce a proportion of the cupric salt to thecuprous state.From the analytical data, structure (VIII), whichis that of a cuprous compound, is a possible alternative to (VII).Convincing evidence that the structure is' (VII) has been obtainedby treating Feigl's compound with alcoholic hydrogen chloride. A3a Ber., 1823, 56, 2083,K3CCu(CN) 4168 INORGANIC CHEMISTRY.green crystalline salt (IX) then separates with a molecule of alcoholof crystallisation. With hot water, two molecules of hydrogenchloride are eliminated from (IX) and the original compound (VII)is produced. A compound of formula (VITI) on treatment withH HI /cu CPh=N<o>Cl CPh=N\(VII.) (VIII. ) PX.)CHPh-Q CBPh-O\ CHPh-0 c1 I I %CU’ \cuC P h = N dHCHPh-O> CPh=N\QH I ,cuc1 [g::$J(X.1 2 (XI.)alcoholic hydrogen chloride would give the derivative (X) .WhilstFeigl’s compound must therefore have structure (VII), there is, atpresent, no direct evidence whether the ring is five- or six-membered.It) has also been shown 39 that nickel, palladium, and platinum givecovalent compounds of the type (XI), where iU = Ni, Pd, or Pt,and it has also been demonstrated that in the case of nickel theoxime may function as a chelate group attached by two co-ordinatelinks. It will therefore be seen that benzoinoxime forms co-ordination compounds with a number of metals and can act as achelate group in three ways. In Feigl’s compound it may beattached to a copper atom by two principal valencies; with otherbivalent metals it may be associated either by one principal valencyand one co-ordinate link as in (XI), or by two co-ordinate links.When anhydrous cuprie chloride reacts with a glyoxime, such asdimethyl- or benzylmethyl-glyoxime, in ethyl-alcoholic solution,green crystalline co-ordination compounds (XII) are produced of atype different from the more familiar nickel derivatives.I n these(XII) + EtOH -1 &Oz = HZO -j-copper compounds the oxime is functioning as a chelate groupattached by two co-ordinate links. When methylglyoxime isemployed, not only is a co-ordination compound of type (XII)obtained, but a second product of entirely different constitution39 J. S. Jennings, E. Sharratt, and W. Wardlaw, J., 1936, 818WARDLAW : CO-ORDINATION COMPOUNDS. 169(XTII) may be produced.A new chelate group is formed40 in thereaction. This methylalkoxyglyoxirne may readily be distinguishedfrom the familiar dialkylglyoximes by the colour of its nickel deriva-tive. For example, the dimethylglyoxime of nickel is crimson, hutthe methylethoxyglyoxime (XIV) is orange.Although gold may be tervalent, there is no evidence that atervalent gold ion can exist. However, when suitably co-ordinrtted,this metal can form part of a tervalent cation as in (Auen2)Br3,where gold has a covalency of i o ~ r . ~ l This covalency of four isalso exhibited in the dialkyl compounds of tervalent gold, whichhave been shown to have the symmetrical constitution (XV),and tribromogold is similarly forrn~lated.~~ When these dialkylcompounds are treated with ethylenediamine there is an interestingdifference between the behaviour of the diethyl and the di-n-propylderivative. Diethylbromogold 43 yields directly the co-ordinationcompound (XVI), but the di-n-propyl compound 44 gives the inter-mediate derivative (XVII) , which in chloroform slowly yields amixture of the original di-n-propylbromogold and the di-n-propylanalogue of (XVI).When diethyl- or di-n-propyl-bromogold isBr BrjEt,*&NH2*yH2] Br Pr-**tNH,.C2H4*NH2-t~~~-~r II PrEt' 'NH,*CH2 I (XVI.) Pr (XVII.)heated with silver cyanide,45 the corresponding cyano-compoundis obtained as a colourless crystalline non-electrolyte which provesto be tetrameric in freezing bromoform. Reactions with ethylene-diamine lead to crystalline salts R,Au*CN*en*CNAuR,.In theReport for 1933 it was stated that C. S. Gibson 46 had announced thepreparation of two auric derivatives of unusual type (XVIII)40 E. Sharratt and W. Wardlaw, J., 1936, 563.4 1 C. S. Gibson and VC'. M. Colles, J., 1931, 2410.42 A. Burawoy and C. S. Gibson, J., 1935, 217.43 C. S. Gibson and J. L. Simonsen, J . , 1930, 2531.44 A. Burawoy and C. S. Gibson, J., 1935, 219.45 A. Burawoy, C. S. Gibson, and S. Holt, ibid., p. 1024.4G Nature, 1933, 131, 130170 INORQANIC CHEMISTRY.and (XIX). He now states *' that in the light of further experi-mental work the constitutions of these compounds must bc correctedto (XX) and (XXT) respcctively.1 (XIX.)BrIt will be recalled that a few years ago H. Carl~ohn,~* by a simplebut ingenious reaction, prepared a nitrate of iodine, co-ordinated withone or two molecules of pyridine.A solution of iodine in chloroformwas added to silver nitrate dissolved in a chlorof orm-pyridinemixture ; silver iodide was precipitated and dipyridinc iodinenitrate INO,,Zpy crystallised from the filtrate. A molecule ofpyridine was lost when the compound was left in a desiccator oversulphuric acid. More recently,49 the bromine analogue has beenprepared by a similar method. H. Carlsohn obtained it as a hygro-scopic substance by adding bromine in chloroform to silver nitratein a chloroform-pyridine mixture, removing the silver bromide, andadding the filtrate to ether containing a little pyridine. A dipyridinebromine perchlorate has also been isolated.Apparently co-ordina-tisn with pyridine is necessary for the isolation of these interestingsubstances.The fundamental principles of Werner's theory are now so firmlyestablished that it can be asserted with confidence that when asubstance is formulated at variance with these principles then theformulation is incorrect. In the course of his researches onruthenium, A. Joly Ii0 prepared its trichloride, RuC13, and noted thatit would absorb ammonia with the formation of an ammine, whichwas stated to be Ru,C1,,7NH3. By the action of water it yielded anintense violet-red solution from which was obtained a crystallinehydroxy-compound, to which the formula Ru,Cl4(OH),,7NH3,3H,Owas assigned. This hydroxy-compound, also obtained by the action47 J., 1936, 324.48 Habilitat.Schrift., 1932 (S. Hirzel, Leipzig).49 M. I. Uschakov and T7. 0. Tchiatov, Ber., 1935, 68, 824; H. Carlsohn,6o Compt. rend., 1892, 114, 391; 115, 2399.ibid., p. 2209WARDLAW : CO-ORDINATION COMPOUNDS. 171of ruthenium trichloride on aqueous ammonia, is known as“ruthenium red” and dyes animal fibres in red shades. Bothformuke seem unlikely in the light of modern ideas on molecularstructure, and an investigat’ion by G. T. Morgan and F. H. Burstall 51has shown that the Ru,C1,,7NH3 is not a simple substance but amixture of arnrninated chlorides, probably including the hexammine[Ru(NM,),]Cl,. Moreover, ‘( ruthenium red ” proves to be theco-ordination compound [Ru( OH)C1(NH3),]CI,€I,O, in whichruthenium has the usual co-ordination number six.With hydro-chloric acid it gives the yellow salt [RuCI~(NH,)~]C~,~H,O. It isinteresting to find that when the ammonia molecules are replacedby ethylenediamine, ethylamine, or pyridine, the tinctorial poweris greatly reduced. Morgan? and Eurstall suggest that a possibleexplanation lies in the fact that ammonia may induce a differentdistribution of the hydroxy- and chloro-groups from that of theseother ammines. It is obvious in the octahedral arrangement shownby ‘( ruthenium red ” and its analogues that the hydroxy- and chloro-groups may occupy cis- or trans-positions relative to one another.F. H. Burstall52 has recently described a series of very stable redco-ordination compounds [Ru 3dipy]X,,yH,O which have beenobtained by the action of 2 : 2‘-dipyridyl on ruthenium salts.Bythe inbraction of this diamine and ruthenium trichloride at 250°,an almost quantitative yield of the complex chloride[nu 3dipy]C12,6H,0may be isolated from the reaction products :2RuC1, + 8CIoHgN2 = 2[Ru Sdipy]Cl, + C20H,4N4 + 2HUThe stability of these tris-2 : 2’-dipyridylruthenous salts is remark-able; they can be boded with concentrated (50%) potassiumhydroxide solution without destruction of the complex cation. Asthe base, dipyridyl, forms a chelate ring in the same way as ethylene-(1.1 (11.) (111.)diamine [cf. (I)] these trisdipyridylruthenous salts should be capableof resolution. With the aid of d- and Z-ammonium tartrates, thecomplex [Ru 3dipy-j has been resolved into optically active forms(11) and (111), and the optically active bromides isolated. These51 J., 1036, 41..53 Ibid., p. 173172 INORGANIC CHEMISTRY.gave [OL];& + 860" and - 815". I n 1931 R. Charonnat 53 obtainedd- and I-forms of the complex salt R[Ru(NO)(C5H5N)(C,0,),],where R is NI-1, or C,H,N. Here the ruthenium is present as acomplex anion. H. Gleu and K. Rehm 54 record that they haveobtained luteo- and purpureo-salts of ruthenium. The former arethe stable hexammines, [Ru(NH3),],(S0,),,5H,0 and[WNH, 1 6 I H (8 0 4 1 2,[Ru(NH,)5Cl]Cl,.and the latter are the pentammines [Ru(NH,),Br]Br2 andThese compounds are analogous wit'h the corresponding well-knowncobalt and chromium ammines, but the hexammines of rutheniumare colourless like those of rhodium.All the compounds areparamagnetic, with a moment of 2 Bohr magnetons at 20".The heats of formation and solution of some isomeric coba1ta.m-mines have been determined by T. C. J. Ovenston and H. Terre~.~5The method adopted was to decompose the ammine with an excessof M-sodium sulphide. The reaction can be represented for thetetrammine salts by the equation2[Co(NH3),C1,] C1+ 3Na,S = CO&, + 8NH, + GNaC1-I- 2Q cals.If the heats of formation, in solution, of sodium sulphide andchloride and ammonia (or other amine) and the heat of formnt' ionof solid cobaltic sulphide are substituted in the equation, then therequired heat of formation can be calculated from &. The resultsare summarised below :Heat of formation, Weat of solution,cals .cals.trans-[Co(NH,),C1,161 .................. 214,880 - 8290trans- [ Co en,CI,]Cl ..................... 171,920 - 5340css- 7 7 Y Y .................. 214,170 -9510......................... C'LS- 172,830 -8010Although much work has been done on the vapour tensions ofammines, mention should be made of a very interesting seriesof results obtained by G. Spacu and P. Voiche~cu.~~ I n a study ofcertain ammoniates of the thiocyanate, formate, acetate, chromate,glycdate and other salts of the ion Cu2+, the strength of the bondbetween the metal ion and ammonia was found to be inverselyproportional to the base strength of the anion of the salt, underotherwise comparable conditions.In his papers reporting the discovery of unusually stable com-53 Ann.Chim., 1931, 16, 126.54 2. csnorg. Chem., 1936, 22'9, 237.5 5 J., 1936, 1660.56 Z. anorg. Chern., 1936, 226, 273WARDLAW : CO-ORDINATION COMPOUNDS. 173plexes with o-phenanthroline (IV), I?. Blau 57 draws attention to theexistence of two different ferric complexes. The blue one isobtained only by oxidation of the ferrous complex whose formulationis well established as [Fe(CI2HSN2),l2 + ; analysis of its chloroplatinateshowed it to be [Fe(C,2HsN2),]3+. By contrast, direct reaction ofo-phenanthroline and ferric salts led to the formation of brownsolutions from which Blau isolated no solid compounds. A. Gaines,L. P. Hammett, and G. H. Walden 58 have now obtained from thesebrown solutions an interesting crystalline salt of definite compositionwhose properties correspond to the formula of a tetraphenanthroline-dioldiferric chloride (V) .The iron atoms have acovalency of 6H 1and the unusually low magnetic susceptibility of the new complexsuggests the partial neutralisation of the magnetic moments of thetwo atoms, and is additional evidence of a polynuclear structure.There is nothing repugnant in this formulation. The ease offormation and stability of the double “ 01 ” bridge are well estab-lished for cobaltic and chromic ammines and are basic features ofthe theories of colloidal oxides which E. Steasni 59 and A. W.Thomas 60 have developed.Many facts have come to light from a study of the rotatory powerof complex inorganic compounds, with a covalency of six.Studiesof such compounds have been made by F. M. Jaeger,61 J. Lifschitz,62C. H. Johnson,G3 R. Samuel,G4 and many others. A. Werner,65 itwill be remembered, carried out many transformations involvingoptically active compounds in his efforts to find a relationshipbetween configuration and direction of rotation. Typical of theseinvestigations are the following :Z-[Co en,Cl,]Cl+ K,CO, -+ d-[Co en2C0,]C1 + 2KC1Z-[Co en,Cl(SCN)]Cl + NaNO, -+ d-[Co en,(NO,)(SCN)]Cl + NaClZ-[Co en2C1(NO,)]C1 + KCNS -+ Z-[Co en,(N02)SCN]C1 + KCl57 Monatsh., 1898, 19, 647.5 8 J. Amer. Chem. SOC., 1930, 58, 1668.59 Collegium, 1932, No. 751, 902.60 J . Amer. Chem. Soc., 1934, 56, 794.61 “ Optical Activity and High Temperature Measurements,”,NewYork, 1930.62 2.physihl. Chem., 1923, 105, 27; 1925, 114, 485; Rec. trav. chim.,63 Trans. Paraday Soc., 1932, 28, 845; 1933, 29, 626.64 ibid., 1935, 81, 423.1922, [iv], 41, 13.6 5 Ber., 1912, 45, 1228174 INORGANIC CHEMISTRY.He assumed that no change in configuration took place during thesereactions and formulated the structures in accordance with thisassumption. The complex salts [Co en2C1( SCN)]Cl, [Co en2C1(N0,)]C1,and [Co en2(N02)( SCN)]Cl were then resolved as bromocamphor-sulphonates. Werner now assumed that the complex ions of thesame configuration form, in each caBe, the less soluble salt with theactive acid. The experimental results obtained agreed with thosebased on his first assumption, for the Z-[Co en,Cl(SCN)]+,d- [Co en2C1(N02)]+ ,and d-[Co en,(NO,)(SCN)I+ ions crystallised as bromocamphor-sulphonates in the &st fractions.Werner's views have beencriticised by F. M. Jaeger,c6 who states that these rules are quiteillusory and that a much better criterion of analogous spatialconfigurations is based on a comparison of the crystal form. J. P.Mathieu,e7 on the other hand, from studies of the optical absorptionand rotation of many optically active complex salts, finds somesupport for Werner's rule relating configuration and solubility of thediastereoisomerides. W. Kuhn and K. Bein 68 have deducedabsolute configurations of inorganic compounds from the theory ofthe origin of optical rotation. J. C. Bailar and R. W. Auten G9 haverecently shown that Werner's assumption that atoms or groups inthe complex ion are always displaced by others without change ofconfiguration cannot be maintained.They proved that whensilver carbonate reacts with Z-[Co en,C12]@l it gives E-[Co en,CB,]Cl,whereas potassium carbonate gives the dextrorotatory product.A more recent study 70 has demonstrated that an excess of silvercarbonate gives the l ~ v o - and a deficiency the dextro-rotatory salt.Potassium carbonate always gives a dextrorotatory product. Ofthe mechanisms suggested for the Walden inversion, on the basis ofthe reactions of organic molecules, is one 71 which states thatinversion accompanies every reaction. This means that everyreaction which involves a single step in the substitution of one groupby another on a tetrahedral atom should lead to inversion.Accord-ingly if the over-all reaction takes place in an odd number of stepsthe product should be the enantiomorph of the original material.The object of 8ome recent work by J. C. Bailar, J. H. Haslam, andE. M. Jones 72 was to see if this theory might be applied to &covalent6 6 Ref. 61, pp. 93, 139.Compt. rend., 1934, 199, 278; 1935, 201, 1183.68 2. anorg. Chem., 1934, 218, 321.69 J . Arner. Chem. SOG., 1934, 56, 774.' 0 J. C. Bailar, F. G. Jonelis, and E. H. Huffman, ibid., 1936, 58, 2224.71 A. R. Olson, J . Chem. Physics, 1933, 1, 418; E. Borgmann, M. Polanyi,and A. Szabo, 2. physikat. Chem., 1933, By 20, 161.J . Arner. C h m . SOC., 1936, 58, 2226WARDLAW : CO-ORDINATION COMPOUNDS.175inorganic compounds. The case selected was the reaction ofammonia with E-[Co &n2Cl2]C1. At - 77" or - 33" with liquidammonia, the product was Z-[Co en,(NH,),]Cl,, but a t 25" andhigher it was the d-form. These investigators argue that the twochlorine atoms in the complex ion must be attached to the cobaltatom in exactly the same way, and occupy like positions in themolecule. They assume, therefore, that the same mechanismfunctions in their displacement from the complex. If this is correct,then the conversion of the dichloro-salt into the diammino-derivativemust take place in an even number of steps, and the theory men-tioned would allow no inversion. Actuadly, the reaction does leadto an inversion, for the product a t low temperatures is laevo- andthat a t higher temperatures is dextro-rotatory.The authors 72point out, however, that it is possible for the displacement in thecomplex of a chlorine atom by a molecule of ammonia to producesuch a profound change in the complex ion that the second step ofthe reaction does not follow the samc mechanism. E. Bergmann 7Zahas recently stated that the theory mentioned above 7 1 is concernedexpressly with substitations of ions for polar bonds and the reactionstudied by Bailar is not of this type.R. Tsuchida, M. Kobayashi, and A. Nakamura 73 have reportedthat when solutions of certain racemic complex compounds areshaken with powdered quartz, it preferentially adsorbs one antipode,compounds of the same configuration being preferentially adsorbed byquartz of a given sign of rotation.As an example, when a solutionof chloroamminobisdimethylglyoxime cobalt [Co(dm),Cl(NP-I,)],Wdm being dimethylglyoxime, is shaken with the quartz powder thesupernatant liquid is optically active. This method, it is claimed,may be used to determine whether a given compound is cis- or trans-.I n the case just cited, the complex compound must be of cis-configuration, for the trans- woulci be incapable of resolution.Y, Shibata 74 and his colleagues have studied the catalytic oxidationof certain racemic amino-acids in the presence of optically activecomplex compounds, and state that one isomeride of the amino-acid is oxidised faster than the other. They explain this as beingdue to an " enzyme-like action " of the inorganic complex.J. C.Bailar 75 suggests as an alternative explanation that one form of theamino-acid becomes part of the complex while the other does not,120 J . Afmer. C k e m Xoc., 2937, 59, 423.J . Ch,ern. SOC. Japan, 1935, 56, 1339; Bull. C'kem. SOC. Japan, 1936,74 Y. Shibats and R. Tsuchida, i b i d . , 1929,4, 142; Y . Shibata, Y. Tanah,75 Chem. Reviews, 1936, 19, 82.11, 38.and S. Gocla, ibid., 1931, 6, 210176 INORGANIC CHElklISTRY.and subsequent oxidation might destroy one or the other. C. E. M.Pugh's results 76 are not in entire accord with those of Shibata.An interesting account by N. F. Hall 77 of the acid-base propertiesof complex ions has appeared. He summarises the results on theacid strength of various ammino-cations from the work of Lamb,Werner, and Bronsted.As would be expected, those cations withthe greatest tendency to liberate ammonia should be the weakestacids, and this is, in general, the case. It is also interesting tonotice from Werner's work 78 on 6-covalent-metal cations that thecentral atom confers acid strength in the diminishing orderIn connection with complex anions, A. Hantzsch remarked thatamong the oxygen acids those with the most oxygen were in generalthe strongest, and that, as a rule, complex anions tend to be weakbases. In two recent papers, 1. M. Kolthoff and W. J. Tomsicek 79have brought forward the striking fact that, although [Fe(CN),]3-and [Mo(CN),I4- are weak bases, yet [Fe(CN)J4- is about as stronga base as the benzoate ion.An important extension to the chemistry of the metal carbonylshas been made by W.Manchot and W. J. Manchot.80 They haveisolated Ru(CO), as a very volatile crystalline compound (m. p.- 22") by the action o f carbon monoxide on finely divided rutheniumat 180" and a pressure of 200 atmospheres. Moreover, by the actionof light, or better, by heating its benzene solution, they have pre-pared from the pentacarbonyl another derivative, Ru2(CO),, asorange-yellow crystals which in the absence of air only decompose a t200". By the action of the halogens on this enneacarbonyl, theyhave formed compounds of the type RuX,(CO), and have alsoobtained a nitrosyl compound, Ru(NO), or RU(NO)~, from theinteraction of nitric oxide and Ru,(CO),.These results emphasisethe periodic relationship which is found between iron and ruthenium.The three carbonyls c"r(CO),, Mo(CO),, and W(CO), have been thesubject of an extensive investigation by W. Hieber, E. Romberg,and F. Miihlbauer.81 These carbonyls prove to be isomorphous, allforming colourless, strongly refracting, volatile, orthorhombiccrystals, readily soluble in inert organic solvents. Compared withother metal carbonyls they are very stable, the vapour decomposingslowly only above 120". The boilingpints (abs./l atm.) are Cr(CO),Pt4+, Ru4+, Cr3+, Coat.76 Biochem. J., 1933, 2'4, 480.7 7 Chem. Reviews, 1936, 19, 89.7 8 " New Ideas on Inorganic Chemistry," London, 1911, p. 201.7O J . PhysicaE Chem., 1935, 39, 945.8O 2.anorg. Chem., 1936, 226, 385.81 lbid., 1935, 221, 321, 332, 337, 349WARDLAW : CO-ORDINATION COMPOUNDS. 177420*5", Mo(CO), 429-4", W(CO), 448.0". As in the cases of thenickel, cobalt, and iron carbonyls, it is possible to replace the co-ordinated CO group by suitable organic units. For example, bythe use of pyridine, the derivatives Cr(CO), 3py, Mo(CO), 3py, andW(CO), 3py have been prepared. Derivatives of similar type havebeen obtained by the use of such chelate groups as ethylenediamine,o- phenanthroline, or cccc ' -dipyrid yl .There is no doubt that at present most chemists consider that theelectron-pair theory of the co-ordinate link offers the best explan-ation of the many properties of co-ordination compounds. Theformation of complex compounds of the olefins with metallic saltssupplies an interesting test of the application of the lone-pair bondtheory.The first compound of this series was obtained in 1831 byZeise,82 who isolated a substance with the empirical formulaK[PtC13,C2H,],H20 from a reaction mixture of chloroplatinic acidand alcohol. Later, K. Birnbaum 83 prepared similar compoundswith propylene and amylene, and C. Chojnacdki 84 obtainedK[PtBr,,C,H,]. This series of substances is obviously derivedfrom a platinous derivative of the general type PtX,Y, where X isa halogen atom and Y a molecule of a hydrocarbon. Substances ofthis type are bimolecular, but it seems most unlikely that Zeise'ssalt is other than unimolecular, for it has the covalency of 4 which isnormally associated with bivalent platinum.Now if the ethylenemolecule is associated in the usual way with the central metal atom,il "lone pair " of electrons must be available. These are notapparent in the usual formulation for ethylene and allied molecules.One possible mechanism 85 supposes that the lone pair is producedby an elect'romeric change, R.CH=CKR R-6H-EHR. Thisinvolves (a) an opening of the double bond, and ( b ) leaving one ofthe carbon atoms-that which bears the residual positive charge-with a sextet of electrons. Discussing this possibility in the forma-tion of K[PtCl,,C,H,], J. S. Anderson 86 states that it is not alto-gether free from objection on physical grounds. The hypotheticalpolarised state of the bond concerned, supposing as it does thecomplete transfer of one electron, represents an excited state of theinolecule and could be represented at room temperature by only a\very lowsupposedplatinumprobability of occurrence.Alternatively, it might bethat rearrangement takes place in the field of the adjacentatom which possesses a high electron affinity. In either82 Pogg. Ann., 1831, 21, 497.S3 Annalen, 1868, 145, 67.84 Jahresber., 1870, 23, 510.8 5 G. M. Bennett and G. H. Willis, J., 1929, 256.86 J., 1936, 1042178 INORGANIC CHBMISTRY~.case the reactions concerned should involve a high energy ofactivation. There are no data available as to the energy of activ-ation either for the reactions of the type involving formation ofcomplex compounds of olefins with metallic salts or for addition-compound formation by aromatic hydrocarbons, but reactions ofboth kinds proceed either very rapidly or fairly rapidly at roomtemperature.Anderson concludes that, on the evidence so faravailable, any attempt to formulate compoundssuch as Zeise's salt in terms of the electron-pairbond theory possesses a certain artificiality. T. KR A. Ashford and M. S. Kharasch 87 have proposed I R for compounds of the type PtC1,,C,H4 a structure c < ~ which embodies the idea of the electron-pair bond h . 7 but appears t o imply a change in the valency ofcl/p\cl the platinum from two to four. This is difficultto reconcile with their stat,ement that it is wellestablished that these compounds are derivatives of platinousplatinum.It should be mentioned that R. F. Hunter and R.Samuel 88 believe that the conception of the lone pair of electrons asan agent for true chemical linkage is in direct contradiction to theresults of band spectroscopy.Much work of value has been omitted from this report owing tolimitations of space, but the Reporter has attempted to deal with ava'riety of topics in the hope that this course will make the reportof wider interest.\pt/C1c11% ..( '\ R R>y g>cw. w.5. THE RARE EARTHS.As a general rule, chemists now recognise that the elementsof the rare earths are not a confused collection of metals but sub-stances of the highest scientific interest. It is, of course, quitetrue that there are members of the group so scarce and so difficultto separate that they are little more than names in the list of knownchemical elements.Nevertheless, it is not always realised that thegroup as a whole is as plentiful in nature as lead, zinc, or cobalt,and that cerium, the masb abundant member, is more plentiful thansilver, gold, or platinum. The following table gives the estimatedoccurrence of the elements in the earth's crust :Element La Ce Pr Nd Sm Eu Gd Tb D y Ho Er Tin Yb LuAt.No. 67 58 50 60 62 63 64 65 66 67 68 69 70 71"/o x 10'' 7 31 6 18 7 0.2 7 1 7 1.2 6 1 7 1.5$ 7 J. Amer. Gh,em. Xoc., 1936, 58, 1733.1 V. M. Goldschmidt and L. Thomassen, VVidensEa(ps.-Slc~~fte~, I, Matemat.-88 Nature, 1936, 133, 411.Naturu. K h s e (Krbthnia), 1924, No. 5, p. 49CARTER AND WARDLAW : THE RARE EARTHS.179It shows that the elements of even atomic number are always moreabundant than their neighbours of odd atomic number, and thissuggests that their stability is closely related to their detailed atomicstructures.At one time the metals of the rare earths were supposed to havean invariable valency of three, so that cerium, being quadrivalent,was considered an intruder. Although it is still correct to saythat the characteristic valency of the group is three, yet it is nowfirmly established that both higher and lower valencies are possible.Por inshnce, cerium, praseodymium, and terbium can be quad-rivalent, and samarium, europium, and ytterbium may be inducedto show a valency of two. G. Jantsch and W. Klemmy2 prominentworkers on these anomalous valencies, have published a diagramwhich shows the results of recent investigations. In this diagram(see figure) lines above the central horizontal line represent quadri-valency, and lines below show bivalency.The length of the line[Cp = Lu (at. no. 71) and Tu = Tm] Igives an approximate measure of the stability, and the size of thepoint denotes the relative stability of the electronic configurationof the Me3+ ion. There is evidence in the cases of lanthanum,gadolinium, thulium, and lutecium that the lower-valency compoundsexist, but these have not yet been obtained in a form capable ofdetailed study. Lanthanum was stated3 to form an oxide of thequadrivaleiit element, but this could not be confirmed by G.Jantsch and E.Wie~enberger.~These variable valencies are proving most useful in makingpossible new methods of separation. Cerium, for example, haslong been separated from its neighbours by boiling an oxidisedsolution containing thein, whereby ceric salts are readily hydrolysedand precipitated as basic compounds. The new processes, however,utilise a valuable observation first made by Jantsch and his col-laborator~,~ that when the rare-earth metals become bivalent,2 Z . anorg. Chin., 1933, 216, 80.3 I. M. Kolthoff and 1%. Elmquist, J . Amer. Chem. SOC., 1831, 53, 1230.4 Monatsh., 1932, 60, 1.5 G. Jantsch, H. Alber, and H. Grubitsch, iKom~%h., 1929, 53-64, 305180 INOELGBNIC CHEMISTRY.the solubilities of their sulphates resemble those of barium andstrontium.This fact led L.F. Yntema 6 to explore the possibility of separ-ating europium from other rare earths by electrolytic reductionin the presence of the sulphate ion. He dissolved a mixture ofthe oxides of samarium, europium, and gadolinium in hydrochloricacid, added a small amount of dilute sulphuric acid, and electrolysedthis solution in a two-compartment cell with a mercury cathodeand platinum anode. As the electrolysis proceeded, colourlesseuropous sulphate, EuSO,, separated. A spectrographic examin-ation of this material showed only a trace of samarium. Thismethod successfully accomplishes one of the most difficult separ-ations in the field of analytical chemistry. states thatif the material contains less than 2% Eu,O, there is a greaterdifficulty in effecting a separation, but co-precipitation with theisomorphous strontium sulphate is helpful.He confirms that theelectrolytic method gives good yields of europous sulphate of highpurity. have shown that bivalentytterbium sulphate may similarly be precipitated from an acidsolution by electrolytic reduction in the presence of the sulphateion, In this way ytterbium may be separated from yttrium,erbium, and thulium. The precipitate has the variable compositionYbSO,,xH,O and is a very light green crystalline compound. W.Prandt19 has used this method to prepare pure ytterbium, butstates that it is not always successful. D. W. Pearce,lo however,has found it satisfactory for the separation of ytterbium fromthulium and from lutecium fractions.33. N. McCoy l1 has shownthat reduction by zinc, in a modified Jones reductor, changesEuCl, into the dichloride. Investigators 9* 11 on. the sulphates ofytterbium have all commented on the reaction of the yellowish-green YbSO, with the dilute acid as soon as the current is stopped :ZYbSO, + 2H' = 2Yb"' + 250,'' + H,It appears that the order of decreasing stability amongst the sul-phates of the bivalent rare-earth metals is europium, ytterbium,samarium,During recent years several attempts have been made to evolvemethods of separation depending on partial thermal decompositionof rare-earth metal sulphates. For example, H. H. Willard andA. BruhlR. W. Bell and L. F. YntemaJ . Amer. C'hem. SOC., 1930, 52, 2782.Aiagew.Ckena., 1936, 49, 159.J . Arner. Chem. Sac., 1930, 52, 4264.@ 2. anorg. Chem., 1932, 209, 13.lo Thesis, Illinois, 1934.l1 J . Amer. Chem. SOC., 1936, 58, 1677CARTER AND WARDLAW : THE RARE EARTHS. 181R. D. Fowler l2 determined the products formed in such decom-positions and also the dissociation pressures of the pure anhydroussulphates at definite temperatures. They then attempted tomaintain the partial pressure of sulphur trioxide above the heatedisomorphous sulphates a t a value intermediate between thoseof the constituents, and thus bring about complete decompositionof one compound to the insoluble oxide without affecting theothers, By adopting these principles, a separation of cerium fromother earths was achieved. Cerous sulphate ignited readily toceric oxide, whilst the associated rare-earth metals remainedas the sulphates.It was found that the separation of praseodymiumfrom lanthanum was not completely successful, because the sulphateof the former ignited only partially to the higher-valency oxide,and the success of the separation requires the formation of thishigher oxide of praseodymium. Similar work has been carriedout recently by L. Wohler and K. F1i~k.l~ Attempts to separaterare earths which did not oxidise on ignition were unsuccessful,because the decomposition products formed solid solutions with noappreciable soiubility differences between the constituents. D. W.Pearce l4 mentions that he and Knlischer have determined thetemperatures at which certain nitrates of tho rare-earth metalslose some or all of their water of crystallisation. When certainmixtures of carefully heated, partly dehydrated nitrates are ex-tracted with anhydrous ether a t low temperatures separationsare obtained.The magnetic susceptibilities of the rare-earth metals, withanomalous valencies, afford an interesting approach to the studyof the electronic structures of the rare-earth metal ions.The largeparamagnetism of these metals and their salts is well known.Susceptibility determinations of praseodymium and cerium inthe quadrivalent state were described in 1925,15 and later investiga-tors l6 have examined compounds of bivalent samarium, tervalentgadolinium, and both bi- and ter-valent europium and ytterbium.The results from the various researches bear out the law thatthe susceptibility of a quadrivalent rare-earth metal ion approachesthat of the ion of a metal with atomic number one less, but havinga valency of three.Also the susceptibility of a bivalent rare-earth metal ion approaches that of the ion of a metal with atomic12 J . Amer. Chem. SOC., 1932,54,496.l3 Ber., 1934, 67, 1679.l4 Chem. Reviews, 1935,16, 121.It S. Meyer, Phy8ikd. Z., 1925, 26, 51, 479.l6 W. Klemm and J. Rockstroh, 2. anorg, Chem., 1928,176,181 ; W. Klemmand W. Schuth, ibid., 1929,184,352 ; P. W. Selwood, J . Amer. Cham. ~SOC., 1933,55,4869; 1934,56, 2392; G. Hughos and D. W. Percrce, ibid., 1933,55,3277182 INORGANIC CHEMISTRY.number greater by one, but with a valency of three.These factsmust mean that, on reduction, the third electron-the one not inuse for valency purposes--is suppressed and becomes associat'edwith the 4, orbital groups. On oxidation this inner group thengives up the electron as a valency unit. For example, the diEerencebetween ytterbium (at. no. 78) in the bivalent and in the tervalentstate is expressed by the electronic distribution :I<. 1;. M . N . 0.Yb3+ .................. 2 8 18 18 + 13 8Yb2+ .................. 2 8 18 18 + 14 8I n view of the fact that the rare-earth metals differ, as far as theorbital electrons are concerned, only in the distribution in the fourthquantum group, it might be expected that the colour of their saltswould be due mainly to the degree of incompleteness of this shell.Thereby colour relations might be thought to be periodic, but thisidea is not substantiated by the facts, as the following table shows.Colours of tervalent ions.57 La 71 Lu Colourless58 Ce 70 Yb ? 958 Pr 69 Tm Green60 Nd 68 Er Red62 Sa 66 Dy Yellow63 Eu 65 Tb Faint rose64 Gd colourlossThis intleresting colour sequence has been discussed by variouswriters, e.g., L.F. Yntema and J. D. Main Smith,18 but it cannotyet be considered as fully explained. There is, incidentally, aninteresting resemblance in colour in the cases of certain bi- andter-valent ions :At. No. ..................... 63 63 69 70Colour ........................ Palo yellow Colourless Green GreenIt should be mentioned that magnetic susceptibilities, when care-fully measured, afford a most accurate means of analysis, becausethe magnetic susceptibilities of mixtures of rare-earth compoundsare additive.The strengths of the rare earths as bases have very frequentlybeen determined, and the results show that there are considerabledifferences in this property among the various members of thegroup.B. S. Mopkins l9 has produced a table which gives theIon ........................... Sm3t EU2+ Tm3 + 133%-l7 J. Amer. Chem. SOC., 1926, 48, 1598.la Nature, 1927, 120, 583.lS J . Chem. Educ., 1936, 13, 363CARTER AND WARJ3LAW : THE RARE EARTIIS. 183relative atrengkh of any rare earth, where the strength of P(OH),is hken its unity.At. No. .................. 39 57 t 9 60 62 G4 66Element ...............Y La I'r Nd Sm Gd DyRel. basicity ............ 1 1300 80 47 8 3.4 0.5It is safe to say that lanthanum hydroxide is the strongest tervalentbase known. It is also important to notice that lanthanum is eleventimes as basic as praseodymium, while neodymium is nearly sixtimes as basic as samarium. The order of decreasing basicitybecomes a matter of importance, because some of the most usefulmethods of separation are based upon the differences in this property.The latest investigations of G. EndresZ0 and of B. 5. 130pkins~~and his collaborators lead with great certainty to the results thatthe order of decreasing basicity throughout the rare-earth groupis exactly the order of increasing atomic number. If yttrium (at. no.39) were included in the rare-earth group it would form an exception,for Hopkins l9 states that in basicity it falls between illinium(at. no.61) and samarium (at. no. 62), and G. Endres 2o places itbetween gadolinium (64) and dysprosium (66).J. Newton Friend 22 has carried out a series of solubility investiga-tions on the selenates, nitrates , and double magnesium nitratesof lanthanum , praseodymium , and neodymium , largely withthe object of facilitating the separation of these rare-earth metals.Finally, mention should be made of the interesting results obtainedby G. Jantsch and his c~llaborators,~~~ 24 who have prepared thealmost complete series of the chlorides , bromides , and iodides ,and determined their melting points. It will be seen from thefigure given by Jantschz4 that the melting points are arrangedvery regularly, and it is noteworthy that in the first half of theseries the chlorides possess the highest melting points, and in thesecond half the iodides.W. Klemm 25 suggests that this is due tothe fact that in the second half of the group another lattice typeoccurs.Prom this short summary of modern work on the rare earthsit will be clear that the study of this group'is far from exhausted.i h c h still remains to be done on the chemistry of anomalous valenciesin spite of the work of investigators in different parts of the world,20 2. anorg. Ci~em., 1932, 205, 321.21 J . Arner. Chem. SOC., 1933, 55, 3117, 3121.22 J . , 1928, 1820; 1930, 1633, 1903; 1931, 1802; 1932, 707, 1083; 1035,83 2.anorg. Chem., 1929,185, 49; 1931,2ol, 207.24 Ibid., 1932, 207, 357.Angew. Chem., 1934,47, 21.356, 824, 1430184 INORGANIC CHEMISTRY.and the group of elements as a whole presents many fresh problemsto the (=ourageous and resourceful investigator,S. R. C. w. w.6. SOME ELEMENTS AND COMPOUNDS.There has been a considerable output of interesting and importantwork, although in the main the year's work has followed closelyalong established lines. Again, full use has been made of thefacilities provided by modern chemical equipment and methods.P. W. Schenli and H. Platz 1 have announced the preparation of ahitherto unknown peroxide of phosphorus. On passing a mixtureof the vapour of phosphoric oxide (P205) and oxygen a t a pressure ofca.1 mm. of mercury through a hot discharge tube, a bluish-violetproduct separated behind the discharge zone and was stable for aday at room temperature, if moisture was excluded. Its aqueoussolution was colourless, and slowly liberated iodine from potassiumiodide. The product is considered to contain about 2% of a newperoxide of phosphorus of the empirical formula PO,.The hydrides of phosphorus have recently been subjected torenewed investigation and some important facts have been disclosed,To the liquid hydrogen phosphide which accompanies gaseous phos-phine when phosphorus is acted on by aqueous potassium hydroxide,the formula P2H, (H.W. 66) has been generally ascribed, althoughthe molecular weights of 74.4-77.0, found by L. Gattermann andW.Hausknecht,2 are by no means in good agreement with it. Thediscrepancy has usually been attributed to the presence of higherhomologues. Liquid hydrogen phosphide has now been carefullyprepared by P. Royen and K. Hill3 in a pure state, and densitydeterminations are in satisfactory agreement with the formulaY,H,. Moreover, a careful search for other homologues has demon-strated their absence. A solid yellow hydride of phosphorus is alsoproduced during the action of aqueous potassium hydroxide onphosphorus. The empirical formula, P,H, has 'long been ascribed tothis substance, as the result of the work of P. Th4nard,4 and ofGattermann and Mausknecht,z who regarded its formation as due toa breaking up of the P,H, molecule :5P2H4 = 6PH, + 2P2H* ,Xafurwiss., 1936, 94, 651.Ber., 1890, 23, 1179.Z .anorg. Chem., 1936, 229, 97.Ann. Chim. Phys., 1845, [iii], 14, 6 ; Annalen, 1845, 55, 27CARTER AND WARDLAW SOME ELEMENTS AND COMPOUNDS. 185A cryoscopic determination of molecular weight by R. Schenckand E. Buch raised the formula Y,H to I?,,€€,. Direct estimationsof phosphorus and hydrogen in the solid hydride, made by R.Schenck and by A , Stock,' indicated a composition varying betweenP,,H,.,, and P12H6.4, the difference being due, presumably, toexperimental errors. P. Royen and K. Hilly3 however, have nowreinvestigated this substance and consider that the yellow hycirideis not a definite compound, but that it arises from adsorption of PH,on amorphous yellow phosphorus, these products having been formedby the decomposition of liquid hydrogen phosphide :3P2H4 = 2P + 4PH3It is pointed out that the analytical methods employed for thedeterminations of phosphorus and hydrogen are very exact (accuracyfor hydrogen, -+0.03%) and the variations found are due simply tothe different amounts of phosphine adsorbed according to theexperimental conditions.Royen and Hill * consider that they have substantiated theirsorption theory by the artificial production of a similar substance bybringing phosphine into contact with amorphous phosphor us, althougha product of composition higher than Pi&l4.12 was not obtained.Very little is known about the modifications of phosphoric oxide,or phosphorus pentoxide, as it is frequently called.Reference toany standard text-book reveals the conflicting nature of the availabledata. It seems agreed, however, that there is a form without arecognisable crystalline structure, and a vitreous form. Otherforms, of more or less doubtful existence, are a crystalline form (a),produced by distillation at comparatively low temperatures, say350", and a second crystalline form ( b ) , stated to be produced fromthe vitreous modification by prolonged heating and having a meltingpoint of 569". A third crystalline form ( c ) , stable above 570", isthought to exist in the absence of the lower-melting form. A. N.Campbell and A. J. R. Campbell have investigated the amorphous,the vitreous, and the low-temperature crystalline form (a), producedfrom the amorphous form a t any temperature between 350" and 600°,provided the heating be not prolonged.These workers have deter-mined densities, solution tensions, and solubility in chloroform, andconclude that, of the three modifications, the only homogeneous formis the vitreous. This is the most stable, since it has the lowestBer., 1904, 37, 915.Ibid., 1903, 36, 991, 4202.Ibid., 1909, 42, 2849.2. anorg. Chem., 1936, 229, 369.it l ' r m s . Faraday Xoc., 1935, 31, 1567186 INORUANIC CHEMISTRY.solubility aiicl the highest density. It is suggested that, as thevitreous form is produced from the amorphous modification, a truesolution of amorphous in vitreous is formed as an intermediateproduct. It is also pointed out that, if the two allotropes aresufficiently stable to form a true solution, their structural units mustbe very different.The nitrides of the non-metals are very diverse in character andinclude a group distinguished by the most extraordinary stability.Prominent in this group are the nitrides of phosphorus, and althoughin the past much research has been done on these and allied sub-stances, they still continue to excite the marked interest of manyinvestigators.Published work dealing with the nitrides of phos-phorus has appeared in recent years from H. Moureu l o and hiscollaborators, P. Renard,ll V. F. Postnikov and L. L. Kuzmin.12Phosphorus pentanitride, P3N5, can be obtained in good yield byA. Stock's method l3 from ammonia gas and P,S5, but H. Moureuand P.Rocquet l4 have described another method which uses thechloronitride of phosphorus, discovered by Liebig and prepared fromphosphorus pentachloride and ammonium chloride. The simplereaction may be expressed thus :CI,PC13 + H3N = (C1,PN) + 3HC1but the chloronitride is really the polymer (PNCl,),. By the actionof liquid ammonia, the chloronitride is changed into phospham,probably by the reactionWhen phospham is heated in a vacuum a t 380" it yields PN,H by thereaction PN(NH,), = PN2H + NH,, and if the temperature nowrises above 400°, pure P N is obtained from the decomposition3PN,H = P,N5 + NH,. 'Tie pentanitride is a light amorphouspowder, insoluble in cold water and the usual solvents. It isattacked only very slowly by concentrated sulphuric acid, but it isquantitatively converted into orthophosphoric acid and ammoniaby sulphuric acid a t the boiling point :It is therefore conveniently analysed by the Kjeldahl method.12When heated to 730" in a vacuum, P3N5 yields PN, which sublimes.PNC1, + 2NH3 = PN(NH,), + 2HCl2P$?, f- 5H2SO4 + 24M,O= 6H3PO4 + S(NH,),SO,10 H.Moureu and A. M. de Ficquelmont, Compt. rend., 1934, 198, 1417;11 Ibid., p. 1159; B2cE.Z. SOC. chim., 1933, [iv], 53, 692; Ann. C?~im., 1835,11 J . AppL. Chem. Russia, 1935, 8, 429.l3 A. Stock and B. Hoffmann, Ber., 1903, 36, 314.l4 Cwrnpt. rend., 1934, 198, 1691.also refs. (14), (15), and (16).xi], 3, 443CARTER AND WARDLAW : SOME ELEMENTS AND COMPOUNDS. 187This, according 10 W. Jfoiireii a,nd l?. Rocquet,*j can exist in twoforms.The more stable red form reduces warm concentratedsulphuric acid, but this reachion is very slow in the cold. The otherform obtained from P3N5 at temperatures in the neighbourhood of720" is yellow, and is readily soluble in sulphuric acid, reducing iteven a t room temperature. Neither form shows crystal structurewhen examined by X-rays. H. Moureu and G. Wetroff 1 6 haveadded a new nitride to the list. When the products of the reactionof phosphorus trichloride on liquid ammonia are heated in a vacuuma t 550", a white, insoluble, non-volatile substance is obtained, spon-taneously inflammable in air. This nitride, P4NG, heated above750" in a vacuum, gives PN, which condenses in a pure state.Until recently, the heats of formation of the metallic nitrideshave been derived indirectly.Now B. Neumann, C. Kroger,and their collaborators l7 have evolved a method whereby thesemeasurements can be made from the direct union of the metal andthe gas, and their work has disclosed some very interesting factsabout this chemical reaction. Certain of the metals studied wouldunite with nitrogen under a pressure of 5-25 atmospheres and atemperature of 500-1000" with sufficient velocity to give a measur-able rise of temperature within one or two minutes. Manganese,chromium, and lithium would do this, but aluminium, beryllium,and magnesium required a catalyst, of which sodium fluorideproved to be the best. The reactions of a number of metals such asnickel, cobalt , aluminium, and beryllium were catalysed by lithiumnitride.It was found that thorium must be very pure to react withnitrogen and, curiously, its reactivity was not improved by sodiumfluoride. A relation-ship has been established between the heats of formation of thenitrides and the atomic numbers of the metallic elements, and onthe basis of this the authors have deduced values for the nitrides ofother elements such as scandium, vanadium, and tungsten. As theease with which a metal takes up nitrogen varies very much with itscondition, this has been studied for molybdenum-iron alloys andfor molyLdenum by A. Sieverts and his collaborators.18In 1909, A. Hantzsch l9 determined the molecular weight ofchamber crystals (HO*SO,*ONO) cryoscopically in sulphuric acidand obtained values varying from 70.8 to 72.5, compared with theIn all cases the nitrogen must be oxygen-free.15 Bull.SOC. chim., 1936, [v], 3, 1801.16 Compt. rend., 1935, 201, 1381.1 7 %. nnorg. Chela., 1931,196, 65; 1932, 204,81; 207, 133, 145; 1934, 218,18 A. Sievarts and K. Briining, Arch. Eisenhiittenw., 1933-34, '7, 641;10 2. physikal. Chem., 1909, 65, 57.379.A. Sieverts and G. Zapf, 2. anorg. Chem., 1936, 229, 161188 INORGANIC CHEMISTRY.calculated value of 127 for HNS05. He concluded, therefore, thatthis substance behaves as an electrolyte, and suggested that dissocia-tion into NO+ and HS0,- ions took place. Twenty-one yearslater, from an examination of the conductivity of nitrosyl perchloratein nitromethane, Hantzsch and K.Berger 2o deduced that nitrosylperchlorate, like nitrosyl sulphate, exists as a salt-like compound[NO]-'[X]-, where X = C10, or SO,. A recent investigation byW. R. Angus and A. H. Leckie of the Ramnn spectrum of nitrosylsulphate (chamber crystals) ,21 and nitrosyl perchlorate 22 has givenresults which can be interpreted only on the assumption that thesesubstances have an ionic structure.I n order to substantiate the deductions from Ramanmeasurements,electrolytic experiments 23 were undertaken. A qualitative demon-stration that nitrosyl sulphate has an ionic structure was made byelectrolysing a solution of nitrosyl sulphate in sulphuric acid betweena platinum anode and an iron cathode. The lower portion of aglass IJ -tube was filled with a concentrated solution of nitrosylsulphate in sulphuric acid, and the upper portion of each limb filledup with more sulphuric acid; the electrodes were clipping into theacid.The iron cathode provided a source of ferrous sulphate in thecathode limb where NO+ would be discharged. After the currenthad passed for some time, an intense brown colour developed in thecathode limb, which suggests that NO+ ions are discharged a t thecathode to give the well-known brown PeSO,,NO. The authorsconclude that, although, unfortknately , quantitative conductivitymeasurements have not yielded results of high accuracy, they haveindicated that nitrosyl sulphate and nitrosyl perchlorate are electro-lytes. No salts of nitrosylsulphuric acid could be isolated in spiteof several attempts.This fact supports the view that the substanccis a salt, and actually the salt-like configuration is the ionised formof the structure hitherto accepted by most chemists, SO,(OH)*ONO.The authors also discuss the theoretical possibility of the existenceof such a radical as NO+. The most important criterion is theease with which a neutral molecule can lose an electron and becomepositively charged. To bring this about, a certain ionisation poten-tial is required, and it follows that the lower the ionisation potentialthe greater the probability of the existence of that particular ion.For nitric oxide the ionisation potential is 9.5 v0lts.~4 The ionisa-tion potential of N, --+ N,+ is very considerably higher, and that20 2.anorg. Chem., 1930, 190, 321.21 Proc. Roy. Soc., 1935, A , 149, 327.22 Ibid., 1935, A , 150, 615.23 W. R. Angus and A. H. Leckie, T'mns. Paraday Xoc., 1935, 31, 958.24 J. T. Tate and P. T. Smith, Physical Rev., 1932, 39, 270CARTER AND WARDLAW : SOME ELEMENTS AND COMPOUNDS. 189for 0, -+ O,+ also very much higher than that for NO -+ NO+.The value for the ionisation potential of 0, is given by R. S. Mullikenand D. S. Stevens 25 as 12.2 volts. It is clear that for the nitric oxidemolecule the ionisation potential is decidedly low, as in general fordiatomic molecules the value is above 10 volts. This may arise fromthe fact that nitric oxide is an odd-electron molecule, having 15 extra-nuclear electrons. When it becomes ionised it is isoelectronic withcarbon monoxide and nitrogen.Since isoelectronic structuresexhibit many similar properties, it is possible that the NO+ radicalis a relatively stable one.Although nitrosoamine, NH,*NO, is probably not stable a t roomtemperature, there is good evidence to show that R. Schwarz andH. Giese 26 have obtained this compound by the interaction of solidanhydrous ammonia and solid dinitrogen trioxide, N,O,, a t lowtemperatures. The reaction was conducted at the melting point ofanhydrous ammonia, and the experiments exhibit some novelfeatures. Liquid anhydrous ammonia (m. 60 c.c.) was poured intoa quantity of liquid air contained in a large porcelain dish, which wascovered with a loosely fitting wooden lid. The ammonia soon solidi-fied, and after the workers’ hands had been suitably protected withgloves, the ammonia and liquid air were ground with a pestle into athin paste.Dinitrogen trioxide in a smaller amount (1-2 g.) wassimilarly crushed in liquid air, and the two reagents were transferredto a wide-necked flask which was rapidly connected, through aground joint, t o a pump, and the excess of liquid air removed. Thereaction mixture was shaken and allowed to warm somewhat, where-upon the light blue colour of the mass changed into the orange-redcolour characteristic of the nitrosoamine. After the whole of theammonia had melted, a clear red solution remained. (Some nitrogenwas evolved, and this was swept out by ammonia gas and collectedin an azotorneter containing dilute acid.) The ammonia was care-fully distilled off , and the solution gradually assumed a deep purple-red colour. This is probably the real colour of the nitrosoamine, andthe orange-red appearing in its preparation with dinitrogen trioxidemay be due to admixture with nitric oxide, which has a yellowcolour when dissolved in liquid ammonia. The authors representthe formation and decomposition of nitrosoamine thus :N,O, + ZNH, = (NH,)NO, + H,N*NO2H2N*N0 = (NH4)N02 + N2Nitrosoamine cannot be obtained in any solvent other than liquidanhydrous ammonia, for it at once breaks up into ammonium nittrite25 Physical Rev., 1934, 44, 720.46 Ber., 1934, 67, 1108190 INORGANIC CI-IBMISTRY.and nitrogen.The nitrosoamine was also formed when the dinitrogentrioxide was replaced by nitrosylsulphuric acid, mO]HSQ,,nitrosyl perchlorate, [NO]ClO,, or nitrosyl chloride, NOC1. Theproposed formulation of the nitrososmine is further supportedby the fact that methylaniline and dinitrogen trioxide at- 5" yield N-nitrosomethylaniline, NPhMe*NQ ; also that nitricoxide under pressure acts on potassium amide in solid ammonia togive the nitrosoamine :KNH, + 2NO = 9CNO -+ NH,*NOThe authors suggest two possible structures for the nitrosoamine,H,N--N--O and H-N=-N-OH, of which the latter is in accordancewith its deep colour, whilst the former is more consistent with itsmode of formation.Probably the two forms are in tautomericrelation to one another.In their study of the decomposition products of carbon suboxide(C,O,), ,4.Klcmenc, R. 'Mreclrsberg, and C. LVagner 27 hare made thefascinating observation that carbon may probably exist in a gaseousform, a s dicarbon C,. The reaction is consideredto be C , O , ~ C O , +C, and these authors state that at 200" the equilibrium constant E =P,, .Pc,/Pc,,2 is ca. 10-7. The equilibriuni is constantly disturbedby the polymerisation process, C,(gas) + graphite. Dicarbon is acarmine-red gas, soluble in water. It rapidly polymerises to apurplish-red, finely divided carbon, which gives an X-ray diagramidentical with that of graphite. In the early stages of the decom-position of the suboxide C,Q,, the head of the Swan band at 4737 A.is clearly visible, and this is known to be characteristic of C,.Theformation of dicarbon may be an intermediate stage in the oxidationof carbon. H. G. Grimm 28 has calculated that the change fromdicarbon gas to solid carbon, as diamond, is strongly exothermic,and is 100,000 cals. The absorption spectrum of gaseous carbonsuboxide bas been recently studied in detail by IH. W. Thompsonand N. K e a l e ~ . ~ ~In a paper on the cbemicd nature of graphite, A. E. Balfour, H. L.Riley, and R. XI. Robinson bring forward scversl coizsiderationswhich, in their opinion, show the aromatic character of the carbonhexagon planes in pure graphite. N. M. Adam 31 pointed out sometime ago that, if one of the carbon atoms in a hexagon plane of tfhegraphite lattice is selected, then the three valency bonds lead to27 %.&:lektrochcm., 1934, 40, 488; 2. physikal. Qhem,, 1934, 170, A, 97." 8 Z. Elektrochem., 1934, 40, 461.2s Proc. Roy. Soc., 1936, A , 167, 331.30 J., 1936, 456.31 !,!'rams. Paraday Xoc,, 1931, 30, 57CAETER AND WARDLAW : SOME ELEMENTS AND. COMPOUNDS. 191three aromatic hexagons, suggesting a similarity, “ though thismay be only superficial,” between the structure of graphite andtriphenylmethyl. Riley and his co-workers believe that thisformal similarity is reflected in a very fundamental way in thechemical properties of these two substances. A. Frenzel and U.Hofmann 32 prepared graphite bisulphate in which the carbonhexagon planes remained intact, the hydrogen sulphate iona havingpenetrated between them, increasing their distance apart. Graphitemonofluoride (see p.149) is a similar type of compound, thoughmuch more stable. In these compounds, Riley states, the hexagonplanes are playing the part of macro-positive radicals. The analogybetween graphite and triphenylmethyl is emphasised from the factthat in liquid sulphur dioxide triphenylmethyl chloride is an electro-lyte, and contains the ions Ph,C+ and Cl-. On the other hand,K. Fredenhagen and G. Cadenbach 33 and K. Fredenhagen and H.Suck 34 have prepared the compounds C,K and C,,K, and A. Schleedeand M. Wellmann 35 have shown that in these compounds the alkali-metal atoms have penetrated, and formed layers between, the hexa-gon carbon planes. These compounds, according to Riley, areobviously analogous to the alkali-metal triphenylmethyls, and thehexagon planes are acting as macro-negative radicals.Otherarguments are brought forward to support this interesting idea.In the literature many methods are given for the preparation ofcuprous oxide, but the colour of the product varies widely : it maybe yellow, orange-yellow, orange, red, and even dark reddish-brown.There are marked differences of opinion as to whether the pre-parations with yellow and red colours should be considered asidentical. In many older and also in some newer text-books, thered product is regarded as the oxide Cu,O and the yellow and orangecompounds as cuprous hydroxide. F. Gebhardt, R. Kohler, andE. Korner 36 have shown that the yellow compound obtained byreduction of Felrlirmg’s solution with glucose, gelatin, or sugar, atboiling heat, is identical with the red crystalline oxide. In order tosettle the question, M.Straumanis and A. Cirulis 37 have preparedcuprous oxide by a, wide range of methods and a t various temper-atures, and submitted the products to an X-ray investigation. Thered product was obtained in a high state of purity by reduction of asolution of cupric hydroxide in concentrated ammonia with hydrazine32 Z. Elektrocimn., 1933, 40, 511.33 Z . anorg. Chem., 1927, 158, 249.34 Ibid., 1929, 178, 353.35 2. physikal. Chem., 1932, B, 18, 1.37 2. amrg. Chem., 193s. W, 107.36 Kolloid-Z., 1933,639 267192 INORGANIC CHEMISTRY.hydratein a hydrogen atmosphere. The yellow product was isolatedfrom the reaction between cupric nitrate in ammoniacal solutionand hydrazine and 2N-potassium hydroxidein a nitrogen atmosphere.The yellow product was also obtained by other methods.Theidentity of the red and the yellow oxide was established from thefact that they had identical lattice dimensions. The yellowpasses into the red modification on growth of crystallites, e.g., onignition, as is shown by the increase in sharpness and number ofdiffracted Rontgen lines with the redness of the material.Lead monoxide, as is well known, occurs naturally in two crystal-line forms, yellow and red. These can also be prepared in thelaboratory by artificial means. The difference between the formshas been attributed to polymorphism, the red being regarded as themore stable form at the ordinary temperature and at all temperaturesup to the transition point.Some years ago the existence of thesetwo forms was in dispute, but M. P. Applebey and R. ID. Reidasisolated the varieties in well-crystallised forms and brought forwardevidence derived from solubility measurements and examinationof crystalline structure which showed clearly that the two modi-fications were polymorphic forms. E. Rencher and M. Bassibre 3*now report the results of an X-ray investigation of two forms of themonoxide E- and p-PbO. The cc-form was obtained as an orange-yellow compound by the dehydration of lead hydroxide (producedfrom a lead salt and alkali) at 130°, and also by heatling lead carbon-ate to 260". An investigation of this modification in a dilatometergave a sharp contraction at 530°, indicating a transition into a newform, a lemon-yellow power, which they designate p.These twoforms gave distinctive X-ray diagrams. When the p-form wasmelted and allowed to solidify, the X-ray diagram showed that it wasstill p, so presumably a p-+ a transition did not occur. The thermaldecomposition of lead dioxide or red lead always led to a-lead oxide,provided that the temperature was below 530". If a sodium hydr-oxide solution of concentration above 30% acted on lead hydroxideat 20" a grecnish-yellow p-lead oxide was formed. After longstanding, this p-form slowly changed into a carmine-red a-form.S. S. Bhatnagar and G. S. Ba140consider that pure nickeloxide,NiO, is green, and that black samples owe their colour to adsorbedoxygen. They also state that their magnetic-susceptibility deter-minations, made at 25-366", show that the high values of % forNiO recorded in the literature are due to traces of nickel formed byreduction during its preparation.On the other hand, W. Klemm38 J., 1922, 121, 2129.3D C m p t . rend., 1936, 202, 765.J. Indian Ghem. SOC., 1934, 11, 603CARTER AND WARDLAW: SOME ELEMENTS AND COMPOUNDS. 193and K. Haas 41 express the view that the variable values for themagnetic susceptibility of nickel oxide, NiO, are due to partialsplitting up into nickel and a higher oxide at temperatures above400". In an earlier Report, attention was directed to some work by(Miss) W. R. A. Hollens and J. F. Spencer 42 on the supposed sub-halide of cadmium, Cd,Cl,, and the so-called cadmous hydroxide andoxide, Cd,O.The cadmium atom (at. no. 48) has the electronicstructure 2, 2.6, 2.6.10, 2.6.10.0, 2, from which it is seen that theatom has two s-electrons in the Ol shell and the bivalent cadmiumion a complete N,,, shell. Both cadmium and the Cd++ ion must bediamagnetic, but the Cd+ ion with one odd electron will be para-magnetic. Thus if solutions of cadmium in molten cadmium chloridecontain cadmous chloride, CdC1, in appreciable amount the systemwill be paramagnetic. Measurements of the solid systems, made byHollens and Spencer, show these to be diamagnetic ; hence the existenceof a sub-chloride, CdC1, has to be excluded. This result has been con-firmed by J. Farquharson and E.He~mann.4~ These authors pointout, however, that the measurements do not exclude the existence ofa birnolecular sub- chloride, Cd,Cl,, because such a substance wouldbe diamagnetic. A definite verdict on this matter is given by thework of R. E. Hedges and H. Terrey,*4 who have examined by X-raymethods the so-called sub-halide, Cd,Cl,, prepared by solution ofcadmium in molten cadmium chloride, and find that the structure isidentical with that givenby the normal chloride, CdCl,, and themetal.Powder photographs of the so-called cadmous oxide, Cd,O, preparedfrom Cd,Cl, by decomposition with water, were taken, and com-pared with those from the normal oxide. It was then quite evidentthat thq so-called sub-oxide is merely a mixture of the normaloxide with very finely divided metal.These findings are in har-mony with the results obtained from other physical measurementson the solid product, e.g., density, heats of solution, etc., and it mustbe concluded that the sub-halides and sub-oxides of cadmium areincapable of existence as solid phases.It has always been questionable whether the formula for the higheroxides of the alkali metals M,O, is not better halved. F. Ephraim 4 ~ iexpresses the opinion that there is little to be said against theformulation MO,. Nevertheless, the double formula is possiblyeasier to construct from the usual valency considerations. L.Pauling has recently raised the question whether the potassium41 2. anorg. Chern., 1934, 219, 82.42 J., 1934, 1062.43 Trans. Paraday Soc., 1935, 31, 1004.44 Ibid., 1936, 32, 1614.45 " Inorganic Chemistry," Gurney and Jackson, London, 1926, p. 341.REP.-VOL. XXXIII. 194 INORGANIC CHIElVlISTRY.oxide should not be KO, instead of K20,. E. W. Neurna~m,~~ atthe suggestion of L. Pauling, studied the magnetism of this peroxide,found the paramagnetism of the expected magnitude for the simplermolecule, and concluded therefore that KQ, was really present.W. Klemm47 has discussed the exact measurements, and statesthat here is a case where such measurements cannot decide betweenthe two €ormula and that we must await further investigations,especially those of the lattice structure, before a final opinion can begiven. This year the structure of potassium tetroxide has beendetermined by X-ray methods, by V. Kassatochkin and V. K o ~ o v , ~ ~and they state that the formula KOz is supported.J. F. Spencer and (Miss) G. T. Bddie 49 have successfully preparedlithium alum, despite the fact that the probability of its existencehas been denied. To prepare the alum, molecular proportions oflithium sulphate monohydrate and the octadecahydrate of alumin-ium sulphate were dissolved in the minimum quantity of cold water.The solution was concentrated considerably by evaporation on asand-bath, and cooled in a freezing mixture of ice and salt withvigorous stirring, whereupon it crystallised suddenly and depositeda mass of small crystals. The mother-liquor after a further slightconcentration deposited small transparent crystals on keepingin the freezing mixture. Both crops of crystals contain H20, 49.0[Li2S04,N2(S04)3,24H20 requires H20, 48.93%]. The crystals areisotropic, a combination of cube and octahedron. M. Mousseronand P. Gravier 5O conclude from solubility, density, viscosity, anddilatometric measurements that sodium alum is stable only between1.1" and 39". The heat of formation of A12(S04),,18H,0 andNa,SO4,10H,O is -3980 g.-cals., and the heat of dissolutionin water-8500 g.-cals. A stable hydrate containing 4H20 has also beenobtained at 15" in a vacuum. In the literature it is generally statedthat two alums are known containing tervalent titanium,Rb,S04,Ti2(S0,),,24H,0 and the cesium analogue. J. Meyer andH. Meissner 51 have recently attempted to extend the series, buttheir attempts to prepare titanium-potassium, -ammonium, and-thallous alums failed. They state that pure RbTi( S04),,12H,0could not be obtained.In a continuation of the studies of the phosphates, H. Bassett,W. L. Bedwell, and J. B. Hutchinson s2 have examined the p p o -46 J . Chem. Phyasics, 1934, 2, 31.47 Angew. Chem., 1935, 48, 617.48 J . Chem. Physics, 1936, 4, 458.2iy Nature, 1936, 138, 169.j0 Bull. SOC. chim., 1932, [iv], 51, 1382,61 J . pr. Chern., 1936, [ii], 143, 70.J., 1936, 1412UARTER AND WARDLAW : SOME ELEMENTS AND COMPOUNDS. 195phosphates of some bivalent metals, and have noted that there is amarked tendency for the formation of solid solutions containingsodium, although definite double salts also occur. They make theimportant suggestions ( a ) that the water molecules are distributedso as to give cations with even co-ordination numbers, and (b) thatreplacement of [M(H,O),]" by ~a2(H20),]** or [M(H20),]"* by 2Na'occurs owing to approximate equality of molecular volumes. Insupport of this view it is mentioned that the Na,P2O7,10R2O whichseparates in large transparent crystals from solutions containingmagnesium, cobalt, nickel, or zinc pyrophosphate contains a smallamount of these in solid solution. This is explicable on the basis ofthe above theory and indicates that Na,P,O,,ZQH,O is probablyAlthough indium is present in minute amounts in a number ofminerals, it is one of the rarest of metals, and its scarcity hasrestrictedthe investigation of its chemistry. In an arc-spectrographic deter-mination of indium in minerals, F. 1cT. Brewer and (Miss) E. Baker 53have made the valuable observation that indium is present in un-usually large amounts in the mineral cylindrite. This mineral,obtained from the Santa Cruz mine, Poopo, Bolivia, is a sulphide oflead, antimony, and tin, and has been shown to have an indiumcontent estimated a t @1-1%. Brewer and Baker 54 have alsofound that indium is present in large traces in some chalcopyritesand as a general impurity in metallic tin, and they have described itsextraction and concentration from these sources.The subject of the nomenclature and classification of inorganiccompounds is one of great and ever-increasing difficulty, and fornearly twenty years chemists of many countries have been tryingto devise a systematic international nomenclature. In his lecture,delivered before the Chemical Society this year, C. Smith 5 5 gavcan account of the agreement that has been reached, and as modernchemical nomenclature is a subject of the deepest concern to allchemists, this address deserves the closest study.Finally, attention should be directed to the recent publication by(Sir) G. T. Morgan and F. H. B ~ r s t a l l , ~ ~ which gives a survey ofmodern developments in inorganic chemistry."a(H20)sl;. "a~(H~O>,1'"P~0,1"".s. It. c.w. w.S. It. CARTER.W. WARD LAW.R . WHYTLAW- GRAY.$3 J . , 1936, 1286.56 " Inorganic Chemistry," Heff er, Cambridge, 1936.64 Ibid., p. 1290. 5 5 Ibid., p. 1067