INORGANIC CHEMISTRY.1. THE METAPHOSPHATES AND POLYPHOSPHATES OF SODIUM.THE alkali metaphosphates comprise one of the most complicatedand puzzling groups in Inorganic Chemistry. Successive workershave added to rather than lessened the chaotic state. The complexnature of the phosphorus molecule is shown, however, not only inthese compounds but also in the element itself and in its oxides,especially the pentoxide,2 and an understanding of the nature ofthe latter and of the different complex states it can assume mightwell provide a key to the structure of its derivatives. On the otherhand, there seems little doubt that the problem has been mademore difficult by lack of strict analytical control of the initial andfinal products. Owing to the diverse values given to relativelysimple physical constants, one is forced to conclude that impure oreven very impure materials were being used in many cases.Although it is impossible at present to give a completed story ofthe metaphosphates, no apology is necessary for discussing themowing to the enormous importance they have acquired in the lastfew years in industry, especially in connection with the soMeningand conditioning of boiler-feed water, in removing calcareousdeposits from boiler-feed tubes, in the laundry and the textileindustry and in tanning processes.Their use in these diversefields is based on the remarkable fact that the so-calledhexametaphosphates (and one or two polyphosphates) will formsoluble complexes with calcium salts, the calcium becoming anessential part of a very stable un-ionised complex, which permits ofthe calcium-ion content of a water being reduced to a scarcelymeasurable value.The metaphosphates were first prepared in 1833 by T.Graham:who pointed out that when sodium dihydrogen phosphate ordisodium dihydrogen pyrophosphate was heated to 315" asparingly soluble metaphosphate was formed. When this washeated to a higher temperature it fused and, as Graham noted,"on cooling it presents itself as a transparent glass whichdeliquesces in a damp atmosphere and is highly soluble in water.But the fused salt has undergone a most extraordinary and1 Cf. Ann. Reports, 1935, 32, 156. a Cf. i b d . , 1936, 83, 185.8 Phil. Trans., 1833,123, 253116 INORGANIC CHEMISTRY.permanent change of properties.The solution has a very feebleacid reaction when compared with the crystallised biphosphate,”the initial substance. Further, he found that when the solutionwas evaporated at 40°, a gum-like mass was obtained which, ifdried at 204”, consisted of pyrophosphate with some water. Theaqueous solution seemed perfectly stable at ordinary temperaturesand was not much affected by boiling with dilute alkali. Theglassy mass was insoluble in alcohol, and when its solution wasmixed with metallic or alkaline-earth salts voluminous gelatinousprecipitates were produced.The formation of the transparent glass only follows when coolingis fairly rapid. Slow cooling results in crystallisation ; a compound,usually considered the trimetaphosphate, is formed, which has nopower of sequestering calcium ions (but see Kurrol salts, p.117).Graham’s original method of preparation of metaphosphates isthe one used technically to-day. Of the other methods, only oneneed be mentioned-that due to G. von K n ~ r r e . ~ This consists oftreating phosphoric acid on the water-bath with a concentratedsolution of sodium nitrate and then heating the mixture to 330”.Purification can be effected by conversion into the lead salt anddecomposition of this with sodium sulphide. The form so obtainedis usually referred to in the literature as Knorre’s salt.It was early realised that in the change from the insoluble intothe soluble hexametaphosphate or Graham’s salt, a number oftransitions were involved, and attempts were made to explain thesechanges on the assumption that phosphates of increasing molecularcomplexity were formed.The initial insoluble form made at lowtemperatures-commonly referred to as Maddrell’s salt 5-wasconsidered by T. Fleitmann and W. Henneberg6 to be themonomer, as they failed to prepare from it any double salts; theyclaimed, however, to have prepared di-, tri-, and tetra-metaphos-phates, usually by controlled heating and subsequent extractionwith water. The question of the chemical entity of thesecompounds has been the subject of many papers which cannot bediscussed here. The evidence in favour of their existence is basedon (a) formation of double salts, (b) molecular-weight determin-ations (ionisation sometimes considered and sometimes not), and(c) equivalent conductivities of solutions, chiefly the differences inthis property for solutions of dilution v = 32 and v = 1024.The view that Graham’s salt, Le., the product obtained by rapidcooling of the fused mass, is a hexametaphosphate rests partly onthe early work of H.Rose,7 who prepared from it a double silverR. Maddrell, Phil. Mag., 1847,30, 322.Pogg. Ann., 1849, 76, 1.8. anorg. Chem., 1900, 24, 397.Annalen, 1848,65, 328TERREY : METAPHOSPHATES AND POLYPHOSPHATES OF SODIUM. 117sodium salt (AgsNaPeOls), on the Iater work of G. Tammann,*who measured the electrical conductivity of aqueous solutions, andon molecular weights calculated from the lowering of the freezingpoint of aqueous solutions by L.Jawein and A. T h i l l ~ t . ~ Tammannfurther inferred from his conductivity work that sodium hexameta-phosphate normally behaved as a salt of a dibasic acid andsymbolised it as Na,[Na,P,Ol,f. He also considered that Graham'ssalt was a mixture of a t least three different isomeric compoundswith different amounts of sodium in the anionic complex. In additionhe claimed to have made a second soluble form of the monomer(p-form) by neutralising metaphosphoric acid with sodiumcarbonate and evaporating the solution a t 50". This form wasreadily soluble in water and easily changed into orthophosphate.Sodium metaphosphate melts normally a t a temperatureEomewhere between 610" and 640' (610", van Klooster; 619",Jaeger; 640", Pascal).By judicious heating it is said to bepossible to bring about the formation of a crystalline salt whichwill not fuse until a temperature of over 800" is reached. Themodifications existing between the normal melting point and thishigher fusing temperature have been looked upon as octa-salts andare usually classed under the name "Kurrol" salts. Thepreparation in this way seems to be capricious, although as will beseen later, P. Pascal claims to have accurately defined conditionsunder which they can be made. They resemble Maddrel17g salt inbeing insoluble in water, but differ in that with aqueous pyro- andhexa-metaphosphate, in which they are soluble, they givesolutions with a very high viscosity. It should be mentioned thathigher condensed salts, e.g., decametaphosphates, have beendescribed. l1During 1923 and 1924 a series of papers were published byPascal l2 dealing with this subject.His researches are discussedhere only in so far as they relate to the monomer and Kurrol saltsand to the changes which he supposed sodium phosphate underwenton being heated to the fusing point.By allowing water-free ether to act on phosphoric anhydride,Pascal prepared ethyl metaphosphate, presumably in the condensedstate, for he formulates it as the hexa-ester. This ester was runslowly into a solution of sodium ethoxide, the temperature notbeing allowed to exceed' 40". Sodium metaphosphate containingsodium ethoxide was obtained as a tough, transparent, faintlybrown mass. Washing with alcohol, ether, and chloroform8 J .p r . Chem., 1892, 45, 463.10 Cornpt. rend., 1924, 173, 211.12 Compt. rend., 1923, 176, 1398; 1924, 178, 1541, 1906.Rer., 1889, 22, 655.l1 Tammann, Zoc. cit., ref. ( 8 ) 118 INORGANIC CHEMISTRY.removed the ethoxide and unchanged esters, and heating the residuein a vacuum afforded quite pure metaphosphate. It is crystallineand easily soluble in water to give a neutral solution exhibiting allthe characteristic reactions of metaphosphate. Measurements ofthe molecular weight gave a value of 51, which would indicate thatthe salt was completely dissociated. The properties of the saltwere not affected by heating to 600". He concluded thatMaddrell's salt, previously regarded as the monomer, had, owingto the method of formation, already undergone condensation to apolymer .13Pascal stated that the Kurrol salts can be made with certaintyby allowing fused Graham's salt to supercool to 550", seeding themass with crystals obtained from the ignition of monosodiummethyl or ethyl phosphate, and then reheating it cautiously inorder to accelerate crystal growth.Under these conditions thewhole mass is converted into the crystalline, insoluble form. Itcan be heated to 809-811" before fusing, whereupon a liquid isformed which he considered to be essentially different from thatobtained by fusing Graham's salt. On cooling, whether a glass ora crystalline solid was formed, the product was insoluble. Additionof traces of hexametaphosphate to the fused mass resulted in itsconversion on solidification into the soluble form.The capriciousnature of the reverse change has already been mentioned.On heating Kurrol salts to different temperatures rangingbetween 600" and 800" and dissolving the products in salinesolutions, particularly in pyro- and hexameta-phosphates, heobtained viscid liquids. From observations that the viscosityincreased per saEtum according to the temperature to which theKurrol salt had been heated, it was deduced that a t least threedifferent modifications were formed. The elastic solids noted byFleitmann and regarded by Tammann as penta- or deca-metaphos-phates were looked upon by Pascal as Kurrol salts mixed withhexametaphosphates.From a study of the products obtained by heating Maddrell'ssalt to different temperatures and then quenching them in mercury,and taking into account the possible formation of Kurrol salts,Pascal summarised the changes which occurred as follows :Kurrol I1 Kurrol I11,+'-'I 21 j 8100601° 607O 640° .5.Maddrell _I, Tri-salt Tetra-salt -= Hexa-salt - - - Liquid(liquid)13 Cf.P. Nylen, 2. anorg. Chem., 1936, 229, 30TERREY : METAPHOSPHATES AND POLYPHOSPHATES OF SODIUM. 119The tetrametaphosphate has a relatively small range of existence.It can undergo change in two ways, giving the fused hexa-salts at640" or Kurrol salts at 595", fusion in the latter event not occurringuntil 810".In the last two years attempts have been made to follow thechanges by using rather more precise methods. The moreimportant of these include (a) X-ray investigations of the productsformed and ( b ) differential thermal analyses on heating and cooling.In the X-ray work, A.Boull6 l4 carried out two series of experimentson the products obtained by (i) dehydration of Na,H,P,O, up tothe melting point, or by (ii) annealing at different temperaturesthe vitreous mass got after fusion. His results may be summarisedas follows.Above 250°, insoluble Maddrell's salt or m8taphosphate-A' wasformed ; this has a distinct structure. By raising the temperatureabove 400" but keeping it below 550", an insoluble metaphosphate-Bwas formed, again with a characteristic structure. The formationof this modification below 400" was very slow-at 3S0°, even afterseveral days the spectra showed that it was a mixture of A' and B.Above 550" up to the fusing point, a soluble metaphosphate-A wasformed; this gave an X-ray spectrum identical with that of A'.Fusion of A', B, or A followed by rapid cooling gave a vitreousproduct C, which lacked structure.Tempering of C resulted incrystallisation, and whether this was carried out a t 300", 330°, 450",d90°, or 625", the same soluble metaphosphate-A was recovered.The reversible transformations recorded by Pascal seemed non-existent. A metaphosphate prepared by Knorre's method (usuallyconsidered to be the trimer) afforded an X-ray spectrum identicalwith that of A.obtainedresults which supported the X-ray investigations. In thedehydration curve two transformations were distinctly shown :(1) at a temperature of about 430°, corresponding to the passagefrom the A' to the B form; and (2) extending over the range550-590", depending on the rapidity of heating, brought aboutby the change from the insoluble B into the soluble A form.Neither change was reversible, the cooling curve indicating onlythe transformation from the liquid to the solid state A.If theheating were stopped before fusion (625") and the product allowedto cool, a smooth curve was obtained and no changes could bedetected. Heating of Knorre's salt or the product A (obtainedfrom Na2H,P207) up to the melting point failed to reveal anytransitions. Boull6 postulated the following changes, withoutIn the differential thermal analysis the same author54 Compt.rend., 1935, 200, 658. 16 Ibid., p. 832120 INORGSNIC CHEMISTRY.attempting to associate any modification with any molecularmagnitude :Na2H2P20, --+ Meta-A' + Meta-B + Meta-ANo indications were obtained of the transformation associatedby Pascal with the tetra-form. The view that this salt is non-existent has been challenged by P. Bonne?nan.lG Following thetechnique described by F. Warschauer,l7 which consists in heatingto a temperature not exceeding 400" a mixture of copper oxide andorthophosphoric acid and precipitating the copper subsequentlywith sodium sulphide, he obtained a product with an X-raystructure distinct from that of the trimetaphosphate and which,from cryoscopic determinations with fused hydrated sodiumsulphate as the so!vent and from conductivity measurements,agreed with the formula Na4(P03),.On fusion it gave Graham'ssalt. At ordinary temperatures it must be in a state of unstableequilibrium (Pascal). This was demonstrated by annealing : a t375" it was partly, and at 500" completely, converted into thetrirner.Most observersagree on the existence of the trimer. This can apparently exist intwo forms, one of which is soluble, resulting from the crystallisationof fused phosphate or from Knorre's preparatlion, and the otherinsoluble-the so-called Maddrell's salt. From the nature of thecompounds derived from it, Graham's salt is most simply regardedas the hexameta-salt. Beyond these two, there is no real evidencefor the formation of compounds in the dehydration of either themonohydrogen orthophosphate or the dihydrogen pyrophosphate.The changes noted in X-ray structure and in the thermal analysescannot satisfactorily be connected with changes in the extent ofthe polymerisation.Little can be said definitely with regard to the nature of thecomplexes formed by Graham's salt owing to the uncertainty withregard to its constitution and molecular state.If it is assumed tobe the hexametaphosphate, then one molecule of this will reactwith one of a calcium or barium salt to form the soluble complex :LiquidTo attempt to sum up the situation is not easy.N'dNa2(PO3)t.J + Gas04 + Na4[Ca(P03)6] + Na&304or (as more usually written)with sodium and calcium both forming part of the complex anion.Both of the above forms were postulated by Tammann, but the18 Compt.rend., 1937, 204, 865. l7 2. anorg. Chem., 1903, 36, 137TERREY : METAPEOSPHATES AND POLYPHOSPHATES OF SODIUM. 121reasons given in support of these structures are not very con-vincing.Of great importance in the technical use of metaphosphates isthe question of their stability, i.e., the speed of their conversion insolution into the trimeric form or into pyro- and ortho-phosphates,for the trimer and the orthophosphate have no power of removingcalcium ions and the pyrophosphate possesses this property only toa slight extent. This problem has been investigated by L. Germain,l8who carried out measurements on the loss of sequestering power ofdilute solutions towards barium salts, (i) alone and (ii) in contactwith acids and alkalis, for different periods and a t different temper-atures.He found that temperature plays a very important part inthe rate of change. Cold solutions are stable, but at the boilingpoint complete conversion was reached in a few hours. Additionof acids increased the velocity of change at all temperatures.Hydrolysis gave primarily sodium dihydrogen phosphate, so exceptin buffered solutions there is a marked decrease in the value of thepH of the solution, the reaction in water being therefore auto-catalytic. The presence of small amounts of sodium hydroxide orcarbonate, as noted by Graham, reduced the rate of change.PoZyphosphates.-Allied to the metaphosphates are the polyphos-phates of the general formula Nan + 2P,03n + produced by heatingtrisodiurn hydrogen pyrophosphate : l9or by fusing together metaphosphate and pyrophosphate : 20Na4P,07 + NaP03 --+ Na5P3O1,Na4P,07 + 2NsP03 -+ Na6P,Ol3These compounds, like Graham’s salt, possess the property offorming un-ionised calcium derivatives and have a correspondingapplication.There is still considerable controversy as to whetherthey are definite chemical entities or merely mixtures of the twocomponents. There is no evidence for compound formation whenthe two solutions are mixed, and no indication of compounds wasobtained from a thermal study of the Melts cooledquickly give an amorphous glass in which are embedded crystals ofnormal sodium pyrophosphate, and the same compound invariablyseparates on slow cooling.On the other hand, Huber has claimedthat by prolonged annealing of the vitreous mass a transformation18 Chirn. et Ind., 1936, 35, 22.19 J. R. Partington and H. E. Wallsom, Chem. News, 1928, 136, 97.20 F. Schwarte, 2. anorg. Chem., 1895, 9, 249; M. Stange, ibid., 1896, 12,444; H. Huber, ibid., 1936, 230, 123; Angew. Chem., 1937, 50, 323.21 N. Parravano and G. Calcagni, 2. anorg. Chem., 1910, 65, 1122 INORGANIC CHEMISTRY.in the solid state takes place. The temperature and time requisitefor this change are not definite, but depend on the size of thepyrophosphate crystals in the fused mass. At the melting pointthe compound Na5P301, is converted into Na,P,O, and anamorphous polyphosphate. In support of his views he adduces thefact that the dispersive action on calcium soaps increased withcontinued annealing’ and secondly, that it was possible to isolatefrom the aqueous extracts definite salts, e.g., Na6P3010,Na3H,P3Qlo,3H,0, or the corresponding zinc saltZnaNaP3Olo, 96H,O ,formed almost quantitatively when zinc acetate is added to anacetic acid solution of the sodium salt. Compounds of the sametype have also been prepared by Bonneman,16 e.8.’Na3CdP3010, 12H,O and Na2CrP3O1,,6H,O.H.T.2. ANOMALOUS VALENCY IN THE RARE-EARTH ELEMENTS.As stated in last year’s Report, in which this subject was brieflydiscussed, “ it is now & d y established that valencies of two andfour are possible with some of the rare-earth elements, although itis still. correct to say that the characteristic valency of the groupis three.” These elements are therefore properly placed togetheras one in the third group of the Periodic Table.A further reviewof the subject has appeared in which the interpretation of anomalousvalency in terms of electronic configuration, based on magneticand optical data, is discussed. These two reviews give a verycomplete list of references to the literature of the subject coveringthe period up to about 1936.The interpretation of anomalous valency in terms of electronicstructure, the investigation of new or doubtful examples, and theapplication of well-established cases of anomalous valency toprovide more rapid methods of separation and purification of therare-earth elements, continue to excite interest and justify a surveyof the most recent work in this field of investigation.Our present knowledge of the occurrence of bi- and quadri-valencyamongst the rare-earth elements is still well summarised by thediagram of Jantsch and Klemm, reproduced in last year’s Report,or by the following table, due to P.W. Selwood,2 which shows theiso-electronic arrangement of the rare-earth ions for the well-established cases of abnormal valency :1 D. W. Pearce and P. W. Selwood, J . Chem. Educ., 1936, 13, 224; cf.2 J . Amer. Chem. SOC., 1934, 56, 2393.also D. W. Pearce, Chern. Reviews, 1935, 16, 121TERREY AND WALKER : ANOMALOUS VALENCY. 123TABLE I.Iso-electronic Arrangement of the Rare-earth Ions.No. of ?lectrons No.of electrons Spectroscopicin ion. in 4f shell. term. Ions.64 0 1s La3+ Ce4+55 1 2F5/2 ~ e 3 + f ~ r 4 +56 2 3H457 3 41rt~a58 4 b459 5 6 H s l a62 8 '3663 9 6%J a64 10 "*65 11 4 1 i 5 ~ a66 12 3H667 13 2 3 7 / aPr3+ f113+Nd3+Sm3+Sm2+ gE~3+Tb3+fDy3+Ho3+Er3+lcTm3+? Y b 3 +Yb2+ gLu3+:3 En2+ )(Gd3+ Tb4+60 661 768 14 1sAs regards bivalency, the possibility of its occurrence withlanthanum, neodymium, gadolinium, thulium, and lutecium hasbeen suggested, but no direct positive evidence has been obtained.There is, however, scme reason to believe3 that bivalent thuliumhas a transitory existence during the reduction of anhydrousthulium trichloride to the metal.There is more uncertainty with regard to the occurrence ofquadrivnlency in elements other than cerium, praseodymium, andterbium, all of which form oxides MO,, obtained by ignition of thelower oxides or of some of their salts. In the case of the last twometals, however, fusion with an alkali nitrate or in an atmosphereof oxygen is necessary for complete oxidation, and, in spite ofmany attempts, it has not been possible to prepare any Pr*+ andTb4+ salts, analogous to the well-known ceric salts, since the higheroxides are decomposed under the action of acids with formation oftervalent salts.The view that neodymium can form quadrivalentcompounds4 has not been confirmed by G. Jantsch and E.Wiesenberger. They have extended their investigations into thepossible formation of higher-valency compounds of the rare-earthmetals to the case of dysprosium, which, since it bears the samerelationship to terbium as praseodymium does to ceriuni (cf.TableI), might be expected to form a higher oxide. They find that X-rayspectroscopically pure Dy,U, shows no appreciabie gain in weightwhen heated in oxygen or air at 300-1000", whilst Dy,O, con-taining a little Tb,O, shows only a slight gain. The productobtained by fusion of Dy,03 in potassium and sodium nitrates and3 G. Jantsch, N. Skdla, aad H. Grubitsch, 2. anorg. Chem., 1933, 212, 66.4 A. Brukl, Angew. Chem., 1936, 49, 533.5 Monatsh., 1936, 68, 394124 INORGANIC CHEMISTRY.in pota,ssium chloride, after removal of solid matter, contains nohigher oxide which liberates iodine from potassium iodide.W.Klemm and A. Koczy6 have shown that the oxides oflanthanum, cerium, and praseodymium, when treated with hydrogenselenide for about 5 hours a t 600-1000", are converted quanti-tatively into polyselenides of the type M,Se4. Neodymium formsan impure polyselenide, and the other elements give the normalselenides M,Se,. The polyselenides form iSI2Se3 when heated in avacuum. As in the case of the poly~ulphides,~ magneticsusceptibility and other evidence shows that the polyselenidescontain tervalent cations. They are not, therefore, abnormalvalency compounds and are best represented as M,Se,-Se.An investigation of the atomic volumes of the rare-earth elementsin the metallic state by W. Klemm and H. Bommer has disclosedsome important relationships regarding the abnormal valencystates of these elements.By means of X-ray methods, theseauthors have examined the lattice structures of all the rare-earthmetals, except holmium, using a mixture of the metal with3RC1 (R = Na, K, Rb, Cs) obtained by reduction of the rare-earthtrichloride with liquid alkali metals. When the atomic volumescalculated from the lattice-structure data are plotted against theatomic number an interesting periodicity is revealed. The valuesof the atomic volume can be regarded as lying or tending to lie onthree distinct curves, vix., the broken curves 11, 111, and IV ofFig. 1, which is reproduced from Klemm and Bommer's paper.The meaning of these curves is interpreted by them on theassumption that the rare-earth metals are built up from positiveions and an electron gas of valency electrons.Curve I11 containsthe atomic volumes of those elements which show only the normalvalency of three and contain only triply charged ions and threevalency electrons per atom in the metal. One would expect tofind a gradual decrease in atomic volume with increasing atomicnumber (lanthanide contraction), and actually the decrease fromlanthanum to lutecium is 4.5 C.C. The curve is made up of twoparts, vix., from lanthanum to gadolinium and from gadolinium tolutecium, corresponding with the usual sub-division of the rare-earth elements into two groups. Curve I1 joins the atomicvolumes of europium and ytterbium to that of barium, and it isconcluded that these two rare-earth metals contain mostly doublycharged positive ions, corresponding with their ready tendency toform bivalent compounds.For samarium, the third element of6 2, anorg. chn., 1937, 233, 84.7 W. Klemm, K. Meisel, and H. U. von Vogel, ibid., 1930, 190, 123.8 Ibid., 1937, 231, 138TERREY AND WALDR : ANOMALOUS VBLENCY. 125this type, the atomic-volume data are uncertain, but it seemsprobable from the greater instability of its bivalent compoundsthat the value of its atomic volume must lie somewhere betweenthat of europium and Curve 111. The lowest curve, IVY purportsto represent the atomic volumes of metals consisting entirely ofM4+ ions, and its position is determined approximately by theexperimentally determined point for the atomic volume of hafnium.FIG.1.The three metals which can form quadrivalent compounds, wiz.,cerium, praseodymium, and terbium, all have atomic volumes lyingbelow the normal curve 111, and this is especially marked withcerium. The atomic-volume curve reproduces, therefore, the mainwell-established features of abnormal valency in the rare-earthelement^,^ vix., that (1) the tendency towards bivalency is greaterthan that towards quadrivalency, and (2) the occurrence ofabnormal valency (both bi- and quadri-) is more noticeable in thefirst group (La --+ Gd) than in the second (Gd + Lu).Klemm and Bommer have deduced similar conclusions from9 Cf. Ann. Reporb, 1936, 33, 179 (diagram)126 INORGANIC CEEMISTRY.measurements of the magnetic susceptibilities of the rare-earthelements in the metallic state.The susceptibility of the metal willbe determined practically only by the paramagnetism of thepositive ions, since their diamagnetism and the weak temperature-independent paramagnetism of the electron gas can be neglected.Thus the paramagnetic behaviour of the metal will depend onwhether it contains only M3+ ions, or also M2+ or M4+ ions, and itshould be possible to determine what ions are present in any metalfrom magnetic measurements. The problem is complicated bythe occurrence of ferromagnetic phenomena in the yttrium-earthmetals, and the results are best discussed by considering thefollowing two groups of elements separately.These metals have no ferromagneticproperties. The effective magnetic moment, p&., is derived fromthe atomic susceptibility, Xat., by the expression pee.= 2-841/xat. . T,where T is the absolute temperature. The values so obtained canbe compared with the moments pcalc. of the various ions, whichcan be calculated from spectroscopic theory 10 and agree withthe values obtained experimentally from measurements with solidsalts such as M2(SO4),,8H,O. The values of pdf. at roomtemperature found by Klemm and Bommer, and those of pcalc.which they quote, are given in Bohr magneton units in Table 11.(1) Lanthanum-samarium.TABLE 11.pca1c..r--- Ipelf.. MA+. M3+. Ma+.- La, ..................... very small - 02-56 - Ce ..................... 2.34 0- 3-61 - Pr ..................... 3.22Nd ..................... 3.56 3.62 3.66 2.8 .....................1-55 3.40 Sm 2.07 -Lanthanum is very weakly paramagnetic, as is to be expected ifit contains only La3+ ions. Cerium has a value of peff. intermediatebetween those of pcdc. for Ce3t and Ce4+, but much nearer theformer, indicating a preponderance of Ce3+ ions. With neodymiumthe result is not so satisfactory, since pdf. is in reasonableagreement with either Nd3+ or Nd4+, both of which have practicallythe same value of pcalc.$ but at any rate Nd2+ can be excluded.For samarium the result suggests that, although it consists mostlyof Sm3+, yet the proportion of Sm2+ is considerable, wix., more than20%. The value of pa. found for praseodymium is the only reallyanomalous one, since it indicates almost equal proportions of Pr3+and PI?+ in the metal, which is contrary to other evidence.Thelo Cf. J. H. van Vleck, “The Theory of Electric and MagneticSusceptibilities,” 1932TERREY AND WAL-R : ANOMALOUS VALENCY. I27magnetic moments of cerium and praseodymium metals have alsobeen measured by a group of Russian workers l1 and found to be11.1 and 16.1 Weiss magnetons, respectively. I n these units thetheoretical values for Ce3+ and Pr3+ are 12-6 and 17.9, respectively,so the above-mentioned data are confirmed. The results forlanthanum, cerium, and neodymium are also confirmed by the workof F. Trombe,12 who obtained a value of 1’7.8 Weiss magnetons forpaf. of neodymium metal, compared with the theoretical value forNd3+ of 18.0.In this group, owing to the occurrenceof ferromagnetism to a greater or less extent, it is better to consider,not pd., but the pmamagnetic moment calculated according toPLpara.= 2.841/xat. (T - O), where 0 is the Curie temperature.Details concerning the ferromagnetic behaviour of this group willbe found in Klemm and Bommer’s paper. It is sufficient to statehere that gadolinium is the only metal of the series which manifeststhe complete beliaviour of a ferromagnetic substance, and thisobservation is confirmed by Trombe. In Table 111, Klemm andBornmer’s values of ppara, and palc. (Bob magnetons) are shown.(2) Europium-lutecium.TABLE 111.P m .EU ..................... 8.3Gd ..................... 7.8Tb ..................... (9.0)Dy .....................10.9’Ho - .....................Er ..................... 9.5Tm ..................... 7-6Yb ..................... 0Lu ..................... 0r-- M4-+.1.53.47.99.710.610.69.67.64.5pI2aJc.eM 3 - k . M2+.3.4 7.97.9 9.79.7 10.610.6 10.610.6 9.69.6 7.67-6 4.54.5 0J\0 SOFor the elements gadolinium, erbium, thulium, and lutecium theva,lues of bma agree well with those of pcalc. for M3+, and showthat only normal ions are present in those metals. With dysprosiumthe result indicates that M2+ ions might also be present, but as inthe above-mentioned case of neodymium, this possibility is excludedfrom consideration of the atomic-volume curve and of chemicalbehaviour. The magnetic data for europium and ytterbium showthat M2+ ions preponderate in these metals, and the proportion ofM3+ probably does not exceed 2 or 3%.For gadolinium metalTrombe has obtained a value for p of 39-28 Weiss magnetonscompared with a theoretical value of 39.26.11 L. F. Vereschtschagin, L. V. Schubnikov, and B. C. Lawrev, PhysikaE.2. Sovietunion, 1936, 10, 618.18 Ann. Phyaique, 1937, [xi], 7, 386128 INORGANIC CHEMISTRY.The results of the atomic-volume and magnetio-susceptibilitymeasurements do not provide any indication of a tendency to formanomalous valency compounds amongst the rare-earth elementsother than cerium, praseodymium, and terbium on the one handand samarium, europium, and ytterbium on the other. W. Noddackand A. Brukl13 have attempted to get evidence for the lowervalency state in the case of other elements from measurements ofthe current-cathode potential curves during the electrolysis ofsolutions of the normal ions, using a dropping-mercury cathodeand a large stationary mercury anode.In the case of Sm3+, Eu3+and Yb3+, from which bivalent compounds can be obtained byelectrolytic reduction, the curves show two distinct reductionstages corresponding to H3f -> M2+ and M2+ -T M (amalgam).The differences in the characteristic potentials of the two stages forthe three elements are 14300, 0.575, and 0.290 volt, respectively.The curves of all the other rare-earth ions also show breaks whichindicate two reduction stages, but the differences in the characteristicpotentials are much less than for the three elements just mentioned,in most cases about 0.1 volt or less.With Sc3+ and Gd3+, however,the differences are 0.160 and 0-145 volt, respectively, and Noddackand Brukl suggest, therefore, that salts of these two elements might,under suitable conditions, be prepared in the bivalent form. Theydo not consider that the results can be attributed to impurities,since the materials used were in most cases X-ray spectroscopicallypure, i.e., contained not more than 0.05% of any other rare-earthor disturbing impurity. It should be noted, however, that thulium,which is the most likely element after samarium, europium, andytterbium to form bivalent compounds, gives a difference betweenthe two reduction potentials of only 0.08 volt. Moreover, thereactions of scandium at the dropping-mercury cathode have beenstudied in detail by R.H. Leach and H. Terrey,l4 who found noevidence for the lower-valency Sc2+ ion.H. T.0. J. W.3. PREPARATION OF PURE COMPOUNDS AND METHODS OFSEPARATING RARE EARTHS.Several papers on the preparation of pure bivalent compounds ofsamarium, europium, and ytterbium have been published duringthe last two years. Work of this kind has an important practicalapplication, since the utilisation of anomalous valencies renderspossible rare-earth separation methods, which have many advantages13 Angew. Chem., 1937, 50, 362. l4 Trans. Faraday SOC., 1937, 33, 480TERREY AND WALKER : PREPARATION O F PURE COMPOUNDS. 1%over the more classical methods of fractional crystallisation andprecipitation.It may be recalled that, in general, four methodshave been used in the preparation of bivalent from the normaltervalent compounds, via, (1) the high-temperature reduction ofthe anhydrous trihalides with hydrogen or ammonia, (2) thethermal decomposition of the tri-iodides, (3) the electrolyticreduction of the tervalent ion in presence of sulphate ion, withconsequent precipitation of the sparingly soluble bivalent sulphate,and (4) the reduction of the tervalent ions by metals such as zincin a reductor.The instability of Sm2 + compounds makes their preparation moredifficult than that of Eu2+ and Yb2+ compounds. F. D. S. Butementand H. Terrey l5 have described a modification of the method ofC. Matignon and E.Cazes 16 for preparing anhydrous samarouschloride by means of the high-temperature reduction of the tri-chloride with hydrogen.W. Kapfenberger l7 has made a further study of L. F. Yntema’smethod l8 of separating europium from other rare-earth elementsby electrolytic reduction of a solution of the trichloride at a mercurycathode in presence of sulphuric acid. Btarting with a 3%europium material, containing mostly gadolinium and samariumand traces of terbium, a 98% europium product was obtained afterthree electrolyses. It is claimed that after one or two furtherelectrolyses an X-ray spectroscopically pure product (< 0.1 yoimpurities) can be obtained.An improved method of purifying europium has been describedby H. N. McCoy.lg The starting material is partly purified byelectrolytic reduction, followed by precipitation of europoussulphate, which is then converted into the trichloride and reducedby the zinc reductor method20 to europous chloride.The finaland complete separation depends on the fact that a concentratedsolution of the latter, containing up to 30% of other rare-earthchlorides, gives with concentrated hydrochloric acid a crystallineprecipitate of EuC1,,2H20, which is practically free from otherrare-earth elements. A few further precipitations render thematerial spectroscopically pure, as shown by both its absorptionand its emission spectra. The crystals of EuC1,,2H20, which arerapidly oxidised on exposure to air, could not be distinguished underthe microscope from those of BaC1,,2H20, which are similarlyprecipitated by concentrated hydrochloric acid, and the twohydrates are probably isomorphous.An X-ray examination byl5 J., 1937, 1112.1 7 2. anal. Chem., 1936, 105, 199.l9 Ibid., 1937, 59, 1131.l6 Comnpt. rend., 1906, 142, 83.l8 J. Arner. Chem. SOC., 1930, 52, 2782.2O H. N. McCoy, ibid., 1936, 58, 1577, 2278.REP.-VOL. XXXIV. 130 INORGAXIC CHEMISTRY.L. Pauling21 has established the isomorphism of europuus andbarium sulphates, and therefore also of strontium sulphate.The electrolytic preparation of ytterbous sulphate has beenfurther investigated by J. K. Marsh 22 and by A. BruldZ3 Hithertothis salt has been prepared by reduction at a mercury cathode, andonly in small quantities. Marsh states that electrodes ofamalgamated lead are more conveaient and may be employedsuccessfully in cells of low resistance.The purity of the lead isvery important, and the freedom of the electrolyte from heavymetals is a factor on which the stability of the reduced salt islargely dependent. Further reduction is very slow when theYb,O, content of the solution falls to 10-15 g./l. This representsa 95 o/o efficiency, with a concentrated sulphate solution initially,and marks the usual limit to which it is worth working. In fourprecipitations the purity of the sulphate can be raised to -iOOyo.Ytterbium acetate undergoes ready reduction in presence of aceticacid when treated with sodium amalgam, and addition of sulphateion then gives a precipitate.Owing to the difficulties in recoveringthe rare earth in presence of much sodium, it is doubtful whetherthis method of reduction can compete in utility with the electrolyticone, which Marsh has used to obtain 95% lutecia from ytterbiurn-lutecium mixtures.Brukl has considerably improved the electrolytic preparation ofytterbous sulphate at a mercury cathode by using an electrolytecontaining sulphuric acid. Starting with a material containing a tleast 96% of ytterbia, he obtained a very pure product by a singleelectrolysis, and from a mixture with lutecium containing about66% of ytterbium a single electrolysis gives 98 % ytterbous sulphate.The mother-liquor can be electrolysed further after the addition offreshly precipitated strontium sulphate, on which the isomorphousytterbous salt is deposited.This method is stated to be veryeffective for the removal of ytterbium from its mixtures withthulium and lutecium, such as are obtained in the fractionation ofthe yttrium earths. Brukl has also applied the electrolyticreduction of ytterbic salts as a method for the quantitativedetermination of the metal. The reduced ytterbous solution isoxidised with ferric alum, and the ferrous ion produced titratedwith permanganate.The earliest rare-earth separation method utilising the anomalousvalency of an element was that in which cerium is separatedeffectively from its neighbours in one step by oxidation to the ceric21 J . Amer. Chern. Xoc., 1937, 59, 1132.22 J., 1937, 1367.23 Angew. Chenz., 1937, 50, 26TERREY AND WALKER : PREPARATION OF PURE COMPOUNDS.131state. J. A. C. Bowles and H. M. Partridge24 describe a methodwhereby a satisfactory separation of ceric cerium and lanthanumcan be obtained by fractional precipitation of the sulphates. Themethod is based on measurements with the glajss electrode of thep , at which the various rare-earth sulphates are precipitated fromsolution on the addition of sodium hydroxide. I n the case of cerica.nd lanthanum sulphates the p , necessary for precipitation (5.78and 7.62, respectively) differs sufficiently to make possible analmost quantitative separation of Ce(SO,), from La,(SO,),. Ii'orthe precipitation of the normal tervalent sulphates there is a totalpE diEerence of only 1-46 between lanthanum and ytterbium, sothat, in general, separation of the rare-earths by this method isimpracticable.I n conclusion, reference may be made to the absorption spectraof non-tervalent rare-earth corn pound^.^^ Those of solutions ofthe tervalent ions have been much investigated, but the origin ofcolour in the rare-earth compounds is a complicated problem whichis as yet incompletely solved.It is known that the 4f electrons,which are responsible for the paramagnetism, are also responsiblefor the colour, and it might be expected that ions with the samenumber of 4f electrons would show similar absorption spectra andhence similar colours. Very few data are available, however, onthe absorption spectra of non-tervalent rare-earth ions. F.D. S.Butement and H. Terrey l5 have investigated the absorptionspectra of Eu3+ and 5m2+ from the point of view of their iso-electronic structure and find that the spectra of pink solutions ofEuCI, and of red solutions of SmCl, show a general similarity withregard to the wave-lengths of the chief absorption maxima. Thesimilarity is not complete, for whereas the bands of Eu3+ show theextremely narrow appearance characteristic of many rare-earthbands in the visible region, those of Sm2+ are among the broadestshown by any rare-earth compound. Comparison of the absorptionspectrum of samarous chloride solutions with that of samaricchloride solutions 26 shows that the shift of the valency electron inSm3f Sm2f is responsible for a profound change in theabsorption.Similarly, H. N. McCoy2' has shown that theabsorption spectrum of a solution of europous chloride in the visiblerange is quite different from that of the europic salt. The solutionof the former (20-30%) has a greenish-yellow colour like that of aconcentrated solution of chlorine. It does not show any of the24 I n d . Eng. Chem. (Anal), 1937, 9, 124.25 Cf. Pearce and Selwood, Zoc. cit., ref. ( l ) , p. 227.z 6 W. Prandtl and K. Scheiner, Z. anorg. Chem., 1934, 220, 107.27 J . Amer. Chem. SOC., 1936, 58, 1580132 INORGANIC CHEMISTRY.absorption bands of the trichloride in the visible region, but absorbscompletely below about 4480 A. There is a rough similarity withthe absorption spectrum of GdC1,,26 which also shows no bands inthe visible region, but in this case it appears that appreciableabsorption commences only in the ultra-violet region a t about3 0 0 0 ~ .A more detailed study of the absorption spectra of therare-earth ions which have an iso-electronic arrangement would beof considerable interest.H. T.0. 5. W.4. THE LOWER OXY-ACIDS OF BORON.A. Stock and E. Kuss observed that when either of the hydridesof boron, B2H6 or B4H,,, was allowed to act on a concentratedpotassium hydroxide solution a salt of the empirical formulaKOBH, was formed :B,H,, $- 4KOH -+ 4KOBHz + H2This salt possessed marked reducing properties and was only stablein the presence of strong alkali. In more nearly neutral solution itunderwent decomposition, hydrogen being evolved and metaborateremaining :2HOBH, + 2H20 --+ 2HB0, + 5H2Shortly afterwards, M.W. Travers, R. C. Ray, and N. M. Guptafound that when magnesium boride (prepared by heating to redness1 part of boric anhydride and 2$ parts of magnesium powder) wastreated with water, hydrogen containing only traces of boronhydrides was. given off, and that the resulting solution actedtowards organic and inorganic compounds as a powerful reducingagent. It had the property of evolving hydrogen on treatmentwith acids, and after acid treatment it would react with iodine.The original solution was not very stable in the air, absorbingoxygen and forming metaborate. They concluded from a study ofthe solution-the volume of hydrogen evolved and amount ofiodine taken up-that it contained the compound H,B,O,.Whensolutions were kept in sealed tubes, particularly after ammonia hadbeen added, a change took place involving an increase in thequantity of hydrogen evolved on addition of acids, and theysuggested that two isomeric compounds, which they denoted asH4 H2B202 and H2 H4B202, were present. They were at that timeunable to isolate from their solutions any definite solid product.1 Ber., 1914, 47, 3115.“Some Compounds of Boron, Oxygen and Hydrogen,” H. K. Lewis& co., 1916TERREY : THE LOWER OXY-ACIDS OF BORON. 133Later, R. C. Ray3 described the isolation of the potassium saltof one of the above acids. The magnesium boride was treatedwith dilute (N/lQQ) potassium hydroxide until the magnesium wasprecipitated ; this was filtered off, and the solution fractionallycrystallised.The first fractions consisted of metaborate andmagnesium hydroxide ; on further concentration, a crystalline solidseparated which could be recrystallised from water freed fromcarbon dioxide. The crystals, probably belonging to the cubicsystem, were slightly deliquescent and readily soluble in water.In the dry state they were quite stable, but evolved hydrogenslowly on exposure to air. Aqueous solutions behaved in theirreducing properties like the original magnesium solutions. Treat -ment with acids brought about decomposition with hydrogenevolution, and the resulting solution absorbed iodine. The atomicratio of the hydrogen given off to the iodine taken up was foundto be 2.From theequivalent conductivity of solutions it was concluded that twoatoms of boron were present in the molecule, and that the aboveformula should be doubled, i e ., K,02B,H4. The reactions withacid and iodine were expressed as follows :Analysis showed that the compound was KOBH,.H,B,(OK), + 2H2SO4 --+ BKHSO, + B,(OH), + 2HZB,(OH), + I, + B,O, + 2HI.It should be noted that the product obtained after oxidation withiodine is diboron dioxide and not the typical oxide of boron.As mentioned above, the possibility of an isomeric form of thissalt was indicated by Travers. This second form has now beenisolated by Ray.4 Magnesium boride was prepared as before, butprior to the treatment with alkali, 2% of magnesium powder andISYO of boric acid were added. The middle fraction of thecrystallised solution was fractionally recrystallised, leading to theseparation of two compounds-a dipotassium salt p-K,H,B,O,isomeric with the one earlier prepared, and a dipotassium saltK,H,0,B2, derived from another acid H4B20,.Treatment of thefirst salt with dilute potassium hydroxide resulted in the formationof a tetrapotassium salt, K4H,B,0,. This could also be obtainedby treatment of the cooled magnesium boride mass with ratherstronger alkali (N/ZQ) than was used in the earlier case. All thesecompounds are strong reducing agents and from molecular-weightand conductivity measurements seem to contain two atoms ofboron. The derivative from the acid H,B,O, differs from theothers in that no hydrogen is evolved on treatment with acid.J ., 1922,121, 1088. * Trans. Paraday SOC., 1937, 33, 1260134 INORGANIC CHEMISTRY.Ray has suggested the following structures for the isomericcompounds :[ H ~ ~ B - O H - H+ H*-B**OH - 13+1 H-B-OH J - H+ HO**B*-H -H+and [ :: 3a - Compound. b-Compound.These are cis- and transforms, the double bond implying inabilityof the boron atoms to rot,ate. They can lose two or four atoms ofhydrogen. In the case of the a-compound it is considered that theadditional two atoms of hydrogen can be more easily lost giving (I),B**OH H * B OH H B OH .. . ..R * * O H H B OH OH i: H(1.) (11.) (I11 .)but with the p-compound the loss of two atoms of hydrogen wouldlead to either (11) or (111), and since the double bond has beenremoved, these are identical.By the loss of two atoms of hydrogenthe a-compound should give a similar derivative, but this has notbeen experimentally realised.Another lower boric acid of the formula M,B,O, has been recentlyisolated by E. Wiberg and W. Ru~chrnann.~ When a boric esteracts on boron trichloride, a chloroboric ester is formed :BCI, + .ZB(OR)3 + 3B(OR),ClThe latter esters on treatment with pure sodium amalgam in thecomplete absence of water undergo reduction, with formation ofesters of the type B,(OR),. Hydrolysis of these in a vacuum(complete absence of air) with water results in the liberation of thecorresponding acid, which is obtained in the form of a white solid :B,(OR), + 4HOH -+ B,(OH), + 4ROHLike the other acids, it is a powerful reducing agent.presence of water, hydrogen is evolved and boric acid formed :In theB,(OH)4 + 2HOH + 2B(OH), + H,The solutions are stabilised by acids and alkalis ; e.g., in the presenceof B~-acid or -alkali a solution of the acid of the same concentrationwas scarcely affected by exposure to the air for 4 hours.Solutionsof silver nitrate and potassium permanganate are readily reduced,but the acid is not oxidised by iodine.This acid is apparently derived from the oxide B,O,, which wasti Ber., 1937, 70, 1303TERREY : THE LOWER OXY-ACIDS OF BORON. 135recognised by Travers and his co-workers,2 who obtained it byoxidising H2B202 with iodine. It has also been identified in anumber of other reactions. For instance, it is formed in thefollowing hydrolyses :(1) B2C14 + 2H,O 4 B202 + 4HC1In this reaction no hydrogen is evolved.6(2) BzHsK, + 4H2O + B202 + 6H2 + 2KOHHere on hydrolysis with alkalis only six molecules of hydrogen areevolved,This reaction is realised when the ammonium salt of B4H10, i.e.,(NH4),[B4H,] is treated with hydrochloric acid and the resultingproduct hydrolysed; only 9 molecules of hydrogen are evolved,whereas for completz oxidation to boric acid 11 molecules shouldbeFor an unsaturated oxide it is apparently extremely stable,especially in the presence of acids or alkalis.In acid solution it isunaffected by iodine, and in the presence of lime it can be heatedin the air without undergoing complete oxidation. On the otherhand, mere exposure of the oxide itself to the air results in theformation of boric acid.These lower boric acids can all be considered to be formed by thehydrolysis of B4H10 :(3) B4K10 + 4H20 -+ 25,02 + 9H2U) 2B4HI3 + 8H20 -+ 4B2MdOH), + 2H2(2) B,HlO + 4H,O -+ 2H4B2(OH), (two forms) + 3H,(3) B&l0 + 4H20 + 2B2(W2 + 7H2(4) B4H1, + 8 H 2 0 2B2(0H)4 9H2although this has only been experimentally carried out in the casesof (1) and (4)-in (4) B20, is actually formed.Interconversion isonly possible in the cases of (2) and (3), although in these, oxidationwith iodine leads to B202, the oxide from which (4) is derived.Stability seems to decrease in passing from (1) to (4) : B,H,(OH),is certainly the most unstable, and B2(OH), is the only one so farisolated in the solid state.H.T.A. Stock, A. Brandt, and H. Fischer, Ber., 1925, 58, 643.A. Stock, W. Sutterlin, and F. Kurzen, %. unorg. Chem., 1935, 225, 225.8 A. Stock, E. Wiberg, and H. Martini, Ber., 1930, 68, 2927136 INORGANIC CHEMISTRY.5. RECENT WORK ON THE OXIDES OF THE HALOGENS.By passing a mixture of oxygen and fluorine at a pressure of10-15 mm. through an ozoniser cooled in liquid air, 0. Ruff andW. Menzel1 obtained an oxide of fluorine F20,. This oxide can beobtained in the form of orange-coloured crystals which melt a t- 160" to a cherry-red liquid, and this can be distilled a t lowtemperatures, Le., below - loo", without decomposition, but a thigher temperatures the characteristic brown colour of the vapoursdisappears and there is formed first an oxide F O which subsequentlya t still higher temperatures breaks down into oxygen and fluorine.The change F,O, --+ 2F0 took place in an irreversible manner,but the amounts of the two gases present in a mixture startingfroni P20, were dependent on the temperature ; e.g., at - 94" thegaseous mixture consisted of 98.6% F20, + 1.4% PO, but a t - 52"complete conversion into FO had taken place.The boiling andthe freezing point of the second oxide were - 185" and - 223"respectively. Its presence as an intermediate phase in thedecomposition was deduced from the fact that the mixturecontaining F,O, and its decomposition products was completelyabsorbed by hydriodic acid, which was considered not to be feasibleif it were a mixture of oxygen and fluorine.Although vapour-density measurements of the gas were carried out and the oxidisingvalue determined, it is obvious that neither would distinguishbetween 2F0 and a mixture F2 and 0,.Two papers dealing with the thermal decomposition of F,O, haverecently been published by P. Frisch and H. J. Schumacher,2 whofollowed two lines of attack. The rate of decomposition between- 25" and - 60" was found to be homogeneous, and could berepresented by an equation of the typeCF2021 d[Bj 0 2 2 = 1012.4 x 10-17.00/457T -Such an equation could not represent the course of the reaction if itoccurred in stages, as suggested by Ruff, but indicated that directdecomposition to fluorine and oxygen took place. This wasconfirmed by examination of the absorption spectrum, vapourpressures, boiling and freezing points, and action on hydriodic acidsolutions, all of which agreed with the properties of an equimolecularmixture of oxygen and fluorine, and these authors conclude thatF O has no real existence.From the time of N.A. E. Millon in 1843, numerous observers haveclaimed to have made chlorous anhydride, CI2O,, by the action of2;. anorg. Chem., 1933, 211, 204.Ibid., 1936, 229, 423; 2. physikal. Chem., 1936, B, 84, 322TERREY : RECENT WORK ON OXIDES OF THE HALOGENS. 137some reducing agent on chloric acid. Millon3 himself used amixture of arsenious acid and dilute nitric acid, J. Schie14 usedsucrose, L. Carius benzene, and M. Hermann naphthalene.Early examination of the absorption spectrum of this oxide showedit to be identical with that of chlorine d i ~ x i d e .~ . ~ The existenceof two different compounds with the same absorption spectrumwas questioned by A. Sch~ster,~ and closer investigation of theso-called Cl,O, has shown that in general a mixture of chlorinedioxide, chlorine, and often a little carbon dioxide comprised thegas under examination, thus vindicating the use of absorptionspectra for identification purposes,The same fate has befallen the last recommended reducing agent-undecenoic acid. M. Kantzer 10 considered he had proved theexistence of this oxide by allowing sulphuric acid to react withpotassium chlorate in the presence of the above acid. From a studyof the absorption spectrum of the gas so prepared, C.F. Goodeveand F. D. Richardson l1 have shown that again the product is thedioxide.By the action of sunlight on chlorine dioxide cooled below 20°,Millon obtained a red liquid which decomposed in the dark athigher temperatures. A similar liquid was prepared by E. J.Bowen,12 who later showed that on continued illumination it wasconverted into colourless dichlorine heptoxide, Cl,07. From aconsideration of its properties he concluded that it differed fromany of the known oxides of chlorine. By analysis of the red liquidM. Bodenstein, P. Harteck, and E. Padelt,13 showed that theoxygen : chlorine ratio was 3 : 1, and from determinations ofthe molecular weight from measurements of the depression of thefreezing point of carbon tetrachloride, concluded that in thedissolved state the oxide had the formula Cl,06. Further work l4on this oxide has shown that in the vapour phase it exists in themonomeric form, but from magnetic measurzments it was concludedthat an equilibrium between the two forms existed in the liquidand the solid state.15 In the latest paper l6 the purification of thecompound is further described. It is shown that, not only is it3 Ann.Chim. Phys., 1843, 7, 298.5 Ibid., 1867, 140, 317. 6 Ibid., 1869, 151, 63.7 W, A. Miller, Phil. Mag., 1845, 27, 81.8 I ) . Gernez, Compt. rend., 1873, 74, 660.9 Rep. Brit. ASSOC., 1880, 258.11 Ibid., 1937, 205, 416.1s 2. anorg. Chem., 1925, 147, 233.14 C. F. Goodeve and F. A. Todd, Nature, 1933,132, 514.15 J.Farquharson, C. F. Goodeve, and F. D. Richardson, Trans. Baruduy18 C. F. Goodeve and F. I). Richardson, J., 1937, 294.Annalen, 1858, 108, 128.lo Compt. rend., 1936, 202, 209.l2 J . , 1923, 123, 2328; 1925,127, ,510.SOC., 1936, 32, 790138 INORGANIC CHEMISTBY.necessary to distil away the more volatile fractions, but owing tothe presence of a non-volatile residue, possibly produced by theaction of the reagents on the glass, it is essential to distil thehexoxide before determining its physical properties. The meltingpoint measured in quartz sampling tubes mas found to be3.50"&0-05", a value much higher than that previously recorded.The vapour pressure was measured over the range - 40" to + ZOO,and it was found that the values for the liquid and the solid lay ontwo straight lines which could by expressed by the followingequationsLiquid : loglopm.-- - 2070/17 + 7.1Solid : log,,p,,,, = - 2690/T + 9.3Calculation of the latent heats of evaporation and sublimation gavevalues of 9-5& 1 and 12-3&0-5 kg.-cals. per g.-mol., respectively.These values are much higher than those of any of the other oxidesof chlorine. This is coupled with the highest melting point andhighest density. Goodeve has suggested that these values,especially when compared with the heptoxide, can be accounted foron the assumption of a symmetrical non-planar structure (I) for/*\ O\ - /O 0 c1 c1-0 7 clG O ? 0 1'0(I.) Non-planar symmetrical, Cl,O,. (11.) Non-planar unsymmetrical, ClzO,.this oxide and an unsymmetrical one (11) for the heptoxide.Thesymmetry of Dhe hexoxide permits the more ready formation ofcrystals in which the molecules can pack more closely together.In 1930 E. Zintl and G. Reinicker l 7 showed that when brominewas passed over specially active mercuric oxide the issuing gascontained a small amount (about 4%) of an oxide which proved onanalysis to be Br,O. W. Brenschede and H. J. Schumacher18failed to isolate the gas in this way, but obtained it in concen-trations greater than 50% of the total bromine taken by theaddition of mercuric oxide to bromine dissolved in carbon tetra-chloride. The reaction was considered to occur through theintermediate formation of mercuric oxybromide :HgO + Br,+ RgOBr,HgOBr, + Br,+ HgBr, + Br,OEvidence for the entity of the oxide and for its molecular state wasobtained from its absorption spectrum, and cryoscopically.Itssolution in carbon tetrachloride was stable below - 20" in the darkl5 Ber., 1930, 83, 1098. l8 2. anorg. Chem., 1936, 226, 370TERREP : RECENT WORK ON OXIDES OF THE HALOGENS. 139but decomposed a t room temperature, producing oxygen, carbonylchloride, chlorine, and bromine.hadpreviously found that (Br,O& was formed as a white crystallinesolid by the action of ozone on bromine. It is only stable a t lowtemperatures, and possibly exists in two modifications with atransition temperature a t - 35".R. Schwarz and M. Schmeisser20 have now succeeded inisolating another oxide of bromine, BrO,. This can be preparedwith a yield of 80% calculated on the bromine used by the passageof a 1 : 5 mixture of bromine and oxygen through a U-shapedozoniser furnished with aluminium electrodes, the greater part ofwhich was immersed in liquid air.A favourable yield dependedon the design of the ozoniser and on the temperature. No oxidecould be detected when the gases were allowed to flow into theuncooled ozoniser and then subsequently passed through a coolingbath. If the oxygen content of the mixture was not too high theformation of ozone was practically nil, although any formed couldbe removed by fractionation, and excess of bromine could bepumped off at - 30". The oxide was obtained in the form of anegg-yellow-coloured solid. It has no definite melting point, butdecomposes spontaneously a t about 0" into oxygen and bromine.This decomposition was utilised in its analysis, the oxygen evolvedbeing measured, and the bromine determined iodometrically.Nothing is yet known about its molecular state.It was noted that if the oxide was cautiously warmed, thedecomposition seemed to take place through an intermediatestage. Together with elementary bromine, a dark brown and awhite substance were formed.The nature of these is the subjectof further investigation.J. I. 0. Masson21 has made some very interesting observationson the mode of formation of the lower oxides of iodine, which arelikely to lead to a complete elucidation of the structure of thesecompounds. The original method of preparation, due again toMillon, was t o heat iodic acid or iodine pentoxide with concentratedsulphuric acid until iodine simultaneously boils off with oxygen.According to conditions, basic iodous iodate OKIO, (iodine dioxide)or iodous sulphate (IO),S04,9H,0 (P.Chrktien's sulphate 2 z ) canseparate out.,, These products arise from the thermal decom-position of the pentoxide, which takes place in stnges :I n addition to this oxide, B. Lewis and H. J. Schumacher1205 _3 1 2 0 3 + 0, L3 I, + 240,19 2. physikal. Chern., 1928, A, 138, 462.21 Natwe, 1937, 139, 150.23 Cf., however, Ann. Reports, 1935, 32, 159.2o Ber., 1937, 70, 1163.22 See Contpt. rend., 1896, 123, 814140 INORGANIC CHEMISTEY.The iodine sesquioxide formed possesses basic properties 24 andcan unite with the sulphuric acid present, to form the sulphate, orwith iodic acid, giving initially iodate which undergoes hydrolysiswith the formation of the more insoluble basic salt.Masson hasshown that the decomposition of the pentoxide can be arrestedexactly a t the middle stage by using fuming sulphuric instead ofthe concentrated acid, such acid being strong enough to stabilisecationic tervalent iodine even a t 220".Of greater importance is the realisation by him that the reactionis a reversible equilibrium. It is displaced to the left by water togive the stable iodate ion, and to the right by acids sufficientlystrong to convert 1,03 into a salt. Thus the sulphate can beobtained in a quantitative yield by the action of concentratedsulphuric on a mixture of iodine and the pentoxide, the compoundseparating in a pure yellow form.The reaction goes through anintermediate stage, shown by the formation of a very deep brownsolute which, it is suggested, is the sulphate of the tervalent radicalI,+.H. T.6. POLYMORPHISM OF ELEMENTS AND INORGANIC COMPOUNDSAT HIGH PRESSURES.I n 1935, P. W. Bridgman 1 described his redesigned apparatusby means of which substances can be subjected to pressures up to50,000 kg./cm.,2 and reported results obtained with a number ofelements. He has now2 described the effects of high pressure onabout 90 inorganic compounds over a temperature range of - 79"(solid carbon dioxide) up to + 200". In his initial work Bridgmanwas limited to pressures of the order of 12,000 kg./cm.2 owing tothe bursting of the containing cylinders.This difficulty wasovercome by making the external walls of the cylinder cone-shapedand supporting this in an external sleeve. Application of pressureon the piston forces the cylinder with an equal pressure into thesleeve so that the external pressure on the cylinder keeps pace withthe internal. The cylinder was constructed of " solar," i.e., awater-hardening silicon-manganese steel, and the piston (whichwas about 6 mm. in diameter) of " carboloy " t h e cementedcarbide of tungsten and cobalt. Temperature control was attainedby means of a bath fixed to the sleeve. For low temperatures,24 Cf. F. Fichter and H. Kappeler, 2. anorg. Chent., 1915, 91, 134.Physical Rev., 1935, 48, 893.Proc.Nut. Acad. Sci., 1937, 23, 202; Proc. Arner. Acad., July 1937TERREY : POLYMORPHISM OF ELEMENTS, ETC. 141solid carbon dioxide, and for higher temperatures, oil electricallyheated, were used, the temperature being measured by a thermo-couple; 200" was about the maximum temperature attainable a thigh pressures owing to the softening of the steel. As in the earlierwork, the pressure was determined from that in the press and thearea of the piston, and the volume change from the movement ofthe piston.For the elements discussed in the 1935 paper, transitions arerecorded in the cases of bismuth (four solid forms 3), thallium (3forms), tellurium (3 forms), gallium (3 forms), and mercury. I nFig. 2 the phase transitions of bismuth are shown, the abscismFIG.2.denoting pressure in kg./cm.,2 and in Table I V the transitionparameters for mercury. I n this case extrapolation of thePress.,kg. /cm.2.10,00015,00020,00025,00030,00035,000TABLE IV.Av, Latent heat,Temp. 104dt/dp. ~rn.~/g x lo6. cal./g.-109" 77 104 0.518- 73 67 71 0.495- 43 57 51 0.483- 17 48 41 0.511+ 5 40 36 0.586 + 23 32 32 0.694temperature-pressure results suggests that the transition should bein the neighbourhood of liquid-air temperatures a t ordinarypressure. This has not been observed, although X-ray spectra,Physical Rev., 1935, 4'7, 427142 INORGANIC CHEMISTRY.have been taken at these temperatures. This may be due to theviscid resistance to transition a t low temperatures.With the following elements no changes were observed : Li, K,Ca, Cd (solid carbon dioxide temperatures), Mg, Ba, In, C (graphite),Ge, Sn, Pb, P (dense black form), As, Sb, S, Se (amorphous andmetallic) (at solid carbon dioxide ar,d higher temperature).A few experiments on alkali halides are also given in this paper.Earlier work had shown that rubidium chloride, bromide, or iodidechanges from the simple cubic sodium chloride structure to thebody-centred cesium chloride structure.This occurred with thepotassium salts at higher pressures, but pressures still higher than50,000 kg.lcm.2 seem requisite for lithium and sodium. Attemperatures below 200" the transition pressures for rubidium andpotassium increase in the order I, Br, C1.This is the reverse of theorder for the corresponding ammonium salts. Above ZOO",however, the transition curves for potassium iodide and bromidecross. The transition parameters for the potassium halides aregiven in Table V.TABLE V.Change ofSalt. kg./cm.2. of vol. dt/dp. cals./g. kg.-cm./g.Fractional Latent internalPress., decrease heat, energy,KC1 ............ 19,900 0.108 +0.200 -1.97 + 1006KBr ......... 19,300 0.102 -0.188 t l . 2 2 680KI ............ 17,850 0.087 +0.430 -0.42 472The fluorides seem different from the other halides. Measure-ments on czsium fluoride, the only czsium halide to crystallise inthe simple sodium chloride system and containing the alkali elementwhich would be expected to show transition at the lowest pressure,showed that it did not undergo any change.Of the 90 inorganic compounds more recently examined, 35 wereshown to exhibit polymorphism.The relevant data for an averagetemperature of about 100" in the case of 10 of these are given inTable VI.Some general observations have been made by the author basedon a statistical study of all his results. Since transitions wereusually observed irrespective of the temperature-the transitionlines passing through the entire temperature range for a pressurechange of 10,000 kg./cm.2 or less-and since a t the absolute zero atransition line must be vertical, it follows that most of the transitionlines will strike the pressure axis at 0" K., and that polymorphismmust be a common phenomenon at the absolute zero,In the earlier work up to 12,000 kg./cm.2 it was found that 14out of the 59 transition linen were of the ice type, i.e., the phasTERREY : POLYMORPHISM OF ELEMENTS, XTC.143Substance .HgCLAgNO,KCNNaClQ,NaBrQ,NaClQ,KIO,CSC10,TlClO,Mean press.,16,6002,20024,00038,50022,5004,40021,20019,00032,20015,00032,00028,00036,00018,90025,30032,50012,8001,70011,900kg.TABLE VI.Mean dtldp.-0.0180-1670.046-0.0940.056-0.0880.0900.0420.005 - 0.046-0.016-0.0100.008 - 0.0270.017 + 0.01 8-0.0130.0740.114-0.125Mean vol.change, yo.0.71.10.61.32.72.98.00.40.40.11.11.53.02.50.43.70.44.92-2Mean latent heat,kg. - em. 18.29.04.58.48.437.085.0220-0360.0197.02.0100.0350.063-0116.030-00%4.648-013-0stable at higher temperatures had the smaller volume.This typeof change is more common a t higher pressures, for 19 cases or 43%were recorded.Another difference found between transitions a t low and a t highpressures was that the internal energy of the higher-pressure formwas greater than that of the lower-pressure form in 78% of thetransitions--the reverse is the case for normal transitions a t lowpressures. Neither the latent heat nor the volume change has anystatistical trend with pressure, so that AE = L - p . A v tends toincrease proportionally to the pressure. This means that thermaleffects become less, and mechanical effects, expressed by p.Av, moreimportant at high pressures.From Schottky's theorem, the total internal kinetic energy underthe experimental conditions is expressed byAEEnetic = 4 p .A ~ - Land L can be neglected in comparison with 4 p . A ~ ; since Av isnegative, the total intlernal kinetic energy decreases on passing fromthe low- to the high-pressure phase. Part of AEkinetic arises froma change in the zero-point energy. If this increases on passing tothe high-pressure form, then the part of AE arising from the internalmotion of the electrons of the atom or molecule must decrease bymore than 4p.Av. On the view that the atoms in the solid have asimilar electronic distribution to that in the isolated state, this mustmean an increase in the size of the atom on passing to the high-pressure form.As Bridgman states : (' This result is so highlyparadoxical that one seems almost driven to the conclusion that a144 lNORGANIC CHEMISTRY.important fraction of the electrons are in an essentially differentstate from that in the free atoms and that polymorphic transitioninvolves an important change in the state of these electrons. Itsuggests itself therefore that the clue to the explanation of thesepolymorphic changes which up to now has so obstinately resistedtheoretical attack is to be found in something similar to theco-operative phenomena between the electrons of the entirestructure which are already known to be determinative forelectrical resistance. ”The same author 4 has investigated the effects of applying ashearing stress simultaneously with the pressure.It might beexpected that transitions which would not occur under anycombination of temperature and hydrostatic pressure might beexhibited under conditions where the shearing stress was verygreatly increased. With the possible exception of lithium, thiswas not found to be the case. Certain observations, however, onthe properties of crystals and glasses are worthy of mention.(a) The shearing stress a t plastic flow of a material has usuallybeen considered to be independent of the pressure. It was found,however, that the shearing strength of a crystal increased withincrease of pressure. With many metals the increase might be asmuch as ten-fold.( b ) Under conditions of great stress and high pressure, it mightbe thought that the crystalline state would break down completely,giving an amorphous mass.This is not so, for the crystalline stateis retained and may even be built up under the severest conditions,e.g., quartz glass tends to crystallise, and transitions were observedjust as they were under pure hydrostatic pressure.( c ) Resistance to plastic flow became greater the more finely thestructure was broken up by continued deformation. Thus acleavage plate of graphite, in which slip easily occurred initially,hardened rapidly on breaking down, and finally became anabrasive, rubbing metal off the hardened steel surfaces of thetesting apparatus. The hoped-for transition from graphite todiamond did not take place.As might be expected, welding of the material was almostgeneral.Under the enormous stresses the molecules are broughtinto the range of each other’s forces, where they attract just as theydo in a solid piece of matter.With certain substances, permanent non-reversible changes indensity were observed, especially where two forms are capable ofexistence. For instance, chalcocite was changed throughout itsmass into cubic cuprous sulphide and calcite became denser, owing* Phpical Rev., 1935, 48, 825; J . Geol., 1936, 44, 663TERREY SOME ELEMENTS AND COMPOUNDS. 145presumably to partial conversion into aragonite. A few changes ofa purely chemical character occurred, these being verified (owing tothe small amounts of material available) by X-ray analysis; e.g.,bismuth oxide was reduced to elementary bismuth, stannic oxidegave stannous oxide, and similar changes almost certainly occurredwith mercury and lead salts.A mixture of copper and sulphur wasconverted into cuprous sulphide, the product being a mixture ofchalcocite and the cubic variety. From the above-mentionedtransformation, this would indicate that chalcocite is the form firstproduced under these conditions.H. T.7. SOME ELEMENTS AND COMPOUNDS.In this section no attempt has been made to give a comprehensivesurvey, and the topics dealt with represent only a fraction of thework carried out during the past year in Inorganic Chemistry.The selection of these topics has been somewhat arbitrary.Although much interesting work has been carried out on molecularstructures,l these have been deliberately omitted, as have alsophase-rule studies and alloy systems.Rare-earth EZeme?zts.-By the action of alkali metal on thecorresponding halide, W.Klemm and H. Bommer have succeededin isolating all the rare-earth metals except holmium, and haveinvestigated their lattice structures and magnetic properties.Prior to this work the only rare-earth elements isolated in therelatively pure state were lanthanum, cerium, praseodymium, andneodyrni~rn,~ and gadolinium.4~ 5 The purity of samarium obtainedby Muthmann, Hofer, and Weiss is open to question, and the publisheddata on erbium seem to indicate thaL a very impure specimen wasbeing used, for the lattice constants 6 found are much greater, andthe magnetic susceptibility very much smaller, than thoserecorded in the present paper.Wohler's method of preparation was selected as the most suitablewhen dealing with small quantities, and as giving a product whichCf.E. G. Cox, A. J. Shorter, W. Wardlaw, and W. J. R. Way, J., 1937,1556; H. D. K. Drew and N. H. Pratt, ibicl., p. 506; H. D. K. Drew andF. W. Chattaway, ibid., p. 947; M. Burawoy, C. S. Gibson, G. C. Hampson,and H. M. Powell, ibid., p. 1690.2. anorg. Chem., 1937, 231, 138.W. Muthmann, H. Hofer, and L. Weiss, Annalen, 1901, 320, 231.G. Urbain, P. Weiss, and F. Trombe, Compt. rend., 1935, 200, 2132.Trombe, Bull. Xoc. chim., 1935, 2, 660.6 J . C. McLennan and R. J. Monkman, Trans. Roy. SOC. Canada, 1929, 23,255.M.Owen, Ann. PhysiE, 1912, 37, 657146 INORGANIC CHEMISTRY.could be used directly for the determination of the more importantphysical properties, the alkali chloride present acting as a calibratingmaterial for the X-ray structure and as a diluent in the magneticmeasurements. The reduction was carried out with liquid alkalimetal which avoided the difficulty experienced by E. Zintl andS. Neumayr,8 who from observafions on the reduction of CeCl, withsodium or potassium vapour and from the behaviour of ceriummetal with potassium vapour concluded Ohat the -rare-earth metaldissolved the alkali element with alteration in the lattice spacing.I n addition to potassium, Klemm and Bommer used rubidium andcmium, since it was felt that these elements with their much widerspacings would not so readily dissolve in the rare-earth element.That this procedure was justified follows from the observations that(a) the spacings obtained were identical irrespective of the alkalimetal used, ( b ) no change was experienced by allowing liquid alkalielement to act on the rare-earth metals, and ( c ) the spacings andmagnetic susceptibilities of the products obtained agreed well withthe better values already recorded in the literature.Difficulty was encountered with samarium, ytterbium, andeuropium, the three elements giving lower chlorides. Here it wasfound that a t high temperatures a back reaction tended to occur :2KC1+ Sa SaC1, + 2K.The best results were obtained withpotassium a t as low a temperature as 250".The purity of the starting product was investigated in every caseeither magnetically or spectroscopically, the estimated impuritiesbeing less than 1% except with Eu (4% Tm), Tm,03(4-7~0 Yb,03and 5.7% Lu203) and Tb, which was an 85% product.Thechlorides were prepared from the oxides by 0. Honigschmid andH. Holch's method,g and the rubidium and cmium by heating thechloride with calcium turnings in an evacuated vessel, followed bydistillation. 10Scandium-Although many attempts have been made to isolatescandium, it is only during the past year that the element has beenobtained in a relatively pure condition. It is somewhat lesspositive l1 than aluminium, but owing to the small amounts of theoxide normally available, electrolytic methods as used for aluminiumare not feasible, and Wohler's method is unsatisfactory owing tothe extraordinary reactivity of scandium and its salts.By usingmolten zinc as cathode and scandium chloride dissolved in theeutectic mixture of potassium and lithium chlorides as electrolyte,2. Elektrochem., 1933, 39, 85.Z . an'org. Chem., 1927, 165, 294; 1928,177, 94.10 Cf. W. Biltz, F. Weibke, and H. Eggers, ibid., 1934, 219, 119.l1 R. H. Leach and M. Terrey, Trans. Faraday SOC., 1937, 33, 480TERREY : SOME ELEMENTS AND COMPOUNDS. 147W. Fischer, K. Brunger, and H. Grieneisen l2 succeeded in isolatinga scandium-zinc alloy from which the zinc and any alkali metalcould be removed by distillation. As a container for the moltenelectrolyte, a magnesite crucible was used.This fitted tightly intoa cylindrical graphite container which acted as anode. Contactwith the molten zinc was obtained by means of a tungsten wirewhich at the temperature of the bath (700-800") was found not toreact with the molten zinc or subsequently formed alloy.The oxide used for the starting material contained small amountsof Zr, Hf, Th, Yb, and Lu. Conversion into the chloride, whichwas attained by heating in a stream of chlorine charged withsulphur monochloride at 1200" and fractionation by volatilisation,gave a still purer product.The alloy resulting from the electrolysis contained 2% ofscandium and about 0.1% of oxide. Unfortunately, in the removalof the zinc this oxide concentrated in the residual scandium, a veryimpure material resulting.This trouble was overcome by filteringthe molten zinc-scandium alloy through a tungsten filter, althoughit reduced the scandium content to about 1%.The element was obtained in the form of a sintered, light greymass, which darkened on exposure; fusion seemed to indicate thatit possessed a silver-white, metallic glance. Analysis showed amet'al content of 94--98y0, the impurity being chiefly oxide whichmay account for its hard and brittle properties. The meltingpoint was found to be approximately 1400" and the density of asample (9S-4y0 Sc) sintered at 1250" was 3-05, which increased to3.08 on fusion.GakZium.-F. Sebba and W. Pugh 13 have described their processfor the preparation of relatively large quantities of pure gallium.The possibility of separating this element by electrolysis has beenrealised since the early work of P.E. Lecoq de Boisbaudran,14 whoobtained it in this way from an alkaline gallate solution. Laterworkers have preferred dilute acid or ammoniacal solutions. Thelatter baths are obviously unsuitable from concentration con-siderations. The use of alkali gallate had been criticised by L. M.Dennis and J. A. Bridgman l5 on the grounds that deposition wasslow a8nd that the metal contained alkali, The authors, however,showed that these criticisms are not valid, current efficiencies of25-30~0 being possible if last traces were not removed, and thatthe metal after being washed was spectroscopically free fromalkali.The greatest difficulty was in the design of a suitable form of12 2.anorg. CJzem., 1937, 231, 54. l3 J., 1937, 1373.14 Bull. SOC. chim., 1879, 31, 50. l5 J . Amer. Chem. Soc., 1921, 43, 274.This value gives an atomic volume of 14.5 C.C148 INORGANIC CHEMISTRY.cathode. From cold solutions the metal separates in the form oftrees with a large surface and a maximum amount of impurities,from a warm solution (above 30") in the form of globules whichdrop off the cathode and, being no longer cathodic, dissolve in theelectrolyte. By using a cathode in the shape of an inverted cup,the metal could be collected in the molten condition and keptcathodic. Platinum, although slightly attacked, was preferred forelectrodes.The gallium so obtained was contaminated by traces of lead, tin,and platinum.These were removed by washing with dilutehydrochloric acid followed by dilute nitric acid, the resulting lossof gallium being only of the order of 5%. The removal of platinumin this way is rather surprising, but only traces of iron could bedetected spectroscopically in a sample of oxide prepared from themetal.Oxides.-(i) I n the AnnuaZ Reports for 1935 l6 am account wasgiven of P. W. Schenk's early investigations on sulphur monoxide.This work has been extended.l7 The formula has been confirmedby a quantitative synthesis from sulphur dioxide and sulphur.The oxide is also formed in the thermal dissociations of thionylchloride at 900°~100 or of thionyl bromide a t 52O"~lO". Atlower temperatures the reverse change takes place, SO unitingreadily with chlorine or bromine.The halogen can be removed a tlower temperatures from the thionyl compounds by means ofmetals such as sodium, tin, magnesium, or aluminium. I n thedecomposition 250 +- SO, + S the sulphur left behind containsappreciable amounts of oxygen. A product of approximatecomposition S20 is formed, which in a vacuum a t 100" gives rise toa mixture of sulphur dioxide and monoxide with a high proportionof the latter and can be used as a suitable source of the gas.Attempts to prepare the corresponding selenium compound SeOhave so far been unsuccessful.l*(ii) Further work has been carried out l9 on the bluish-violetperoxide of phosphorus, PzOs, formed when a mixture of thepentoxide and oxygen is passed through a hot discharge tube, andthe best conditions for its preparation defined.It is insoluble inchloroform, and this fact can be utilised to separate it from thepentoxide. It has been shown that the pentoxide alone whensublimed through the tube gives this oxide, and phosphorusseparates. With water P,06 gives a per-acid, probably H,P20,.It is noteworthy that permonophosphoric acid, H,PO,, originallyprepared by J. Schmidlin and P. &Iassini2* by the action ofl8 Ibid., 1937, 233, 401.2o Ber., 1910, 43, 1162.16 P. 151.lg P. W. Schenk and H. Rehaag, &id., p. 403.l7 2. anorg. Chem., 1936, 229, 305TERREY : SOME ELEMENTS AND COMPOUNDS. 149concentrated hydrogen peroxide on phosphoric oxide, and later byF. Fichter21 by the anodic oxidation of phosphates in potassiumfluoride solution, can easily be prepared by the interaction ofhydrogen peroxide and phosphoric oxide in acetonitriIe solution,yields of 65-68y0 calculated on the oxide content being obtained.22As is well known, this acid is a powerful oxidising agent, convertingmanganous salts into permanganate even in the cold.(iii) I n addition to the two simultaneous reactions in thehomogeneous gas phase already recorded in the case of carbons ~ b o x i d e , ~ ~ vix., polymerisation and the decomposition C,O, -fiCO, + C, a third reaction may also occur, C,O, + 2C+ CsOz,giving rise to a new oxide of carbon, O=C-C=C-C=C=O.This oxide 24 was obtained in a 3% yield during the polymerisationa t 200" of the suboxide prepared from malonic acid.It is describedas a very stable oxide of b. p. 105", as calculated from observationson its vapour pressure a t lower temperatures. It shows no tendencytowards polymerisation ; a t room temperatures it is slowlyconverted into carbon dioxide and an uninvestigated tricarboxylicacid, C,,H1,012. It is interesting t o note that the formation ofthis oxide was never detected in the suboxide arising from diacetyl-tartaric acid. Does this point to a difference in the oxide preparedin the two ways?Early work by electron-diffraction methods indicated that thesuboxide was a linear and symmetrical molecule. This view hasreceived further support from an examination of its ultra-violet,Raman, and infra-red 25 spectra.(iv) A.Guntz and F. Benoit,26 by heating barium with its oxidea t 1150", claim to have prepared Ba20. Their claims were basedon the production of a homogeneous reddish-brown mass and onthe fact that, if any free barium were present, it would, owing toits higher density, sink to the bottom of the melt.* As indirectevidence, they adduced the fact that aluminium reacts energeticallywith baryta, which would not be expected from the known heats offormation of the two oxides, and that reduction of the barium oxideonly proceeded as far as the suboxide stage.The existence of the suboxide is of interest in that oxide cathodes21 Helv. Chim. Acta, 1918, 1, 297.22 G. Toennies, J . Amer. Chem. SOC., 1937, 59, 555.23 A. Klemenc, R. Wechsberg, and G. Wagner, 2.physikal. Chern., 1934,24 Klemenc and Wagner, Ber., 1937, 70, 1880.2 5 R. C. Lord, junr., and N. Wright, J . Chem. Physics, 1937, 5, 642.2 6 Bull. Soc. chim., 1924, 35, 700.* This statement appears to be erroneous : the densities of baryt'a and ofA , 170, 97.the metal at room temperature are 5-39 and 3.63, respectively150 INORGANIC CHEMISTRY.only show a marked electron emission when the ratio Ba : 0 isgreater than unity although this barium excess need not be greaterthan 2-3%.27Reinvestigation of this problem by M. Schrie12* leads fairlydefinitely to the conclusion that this suboxide does not exist.Indirect calculations by Guntz seemed to show that the reactionBaO + Ba + Ba,0 was exothermic to the extent of 15.5 cals.A direct attempt to measure this heat evolution when the twocomponents were heated together failed to indicate any heatchange.The X-ray structure of the product was the same as thatof the original barium oxide, and only a slight change of densitycould be found, which might be expected from the packing ofbarium atoms into the barium oxide lattice owing to its solubilityat high temperatures.The composition of the mass only approximated to that of acompound when the components were taken in the required ratio.Barium oxide is readily soluble in molten barium, and only whenthe ratio Ba : BaO is greater than 2.5 : 1 is the barium at all readilyevolved, even on heating in a vacuum a t 900".(v) Further evidencez9 against the existence of a stable leadsuboxide has been obtained by measurements of the magneticsusceptibility of the product of vacuum decomposition of leadoxalate at 200-300".To compensate for this, a new oxide of lead,Pb,O,,, is said 30 to be formed by the dissociation of the dioxide orthe interaction of lead carbonate and oxygen a t 365-460" underan oxygen pressure of 200 atmospheres.(vi) The formation of titanium monoxide when the dioxide andmetal are heated together in a vacuum has been proved by anexamination of its crystal str~cture.~l It has a melting point of1750", and dissolves in dilute acids with evolution of hydrogen.In connexion with the oxides of titanium, it may be mentionedthat the equilibrium 2Ti0, + H,CTi,O, + H,O has beenmeasured by N. Nasu.3,Boron Hydrides.-By the action of potassium amalgam on thehydrides B,H,, BLfHl0, and B,H9, Stock and his co-workers33prepared a number of borane salts, e.g., B,H, + K2+ K,B,H6.The method used was to shake the hydride for some hours with theamalgam, run off the greater part of the mercury, and distil off the27 T.P. Berdennikowa, Physikal. 2. Sovietunion, 1932, 11, 77.28 2. anorg. Chem., 1937, 231, 313.2B L. We10 and M. Petersen, Physical Rev., 1936, [ii], 49, 864.30 C. Holtermann and P. Laffitte, Compt. rend., 1937, 204, 1813.31 W. Dawihl and K. Schroter, 2. anorg. Chem., 1937, 233, 188.s2 Kinz. no Kenk., 1935, 12, 371 ; Sci. Rep. TGhoku, 1936, 25, 510.a3 2. anorg. Chem., 1935, 225, 225TERREY : SOME ELEMENTS AND COMPOUNDS. 151remainder a t as low a temperature as possible (170-190"), thesalts being left behind in the form of iion-volatile white solids.The salt K2B2H6 could be heated to 300" without change, but a thigher temperatures about one-third sublimed unchanged, theremainder undergoing decomposition :The substance KB2H2, which may be K,B4H4, dissolved in waterwith the evolution of hydrogen.In continuation34 of this work,the corresponding sodium and calcium salts have been prepared,and the properties of the higher members more fully investigated.In each case two equivalents of metal per molecule of hydriderepresented the composition of the salt. At 170" the derivativesfrom B4H,, and B5H, lose hydrogen according to the equationsK,B$&j __t KB&& + K + 2H2K2B4H10 K2B4H8 + HZ2K2B6Hg = K4B10H16 + HZHeated to high temperatures (450"), K,B,HS gave K2B,H,.Allreact in the same way with hydrochloric acid, part forming theoriginal hydride and part being chlorinated to the chloro-hydride.Rontgen photographs of the sublimed potassium salts showed amarked similarity, indicating that they belonged to a series.Unsuccessful attempts were made to prepare KBH, by the actionof atomic hydrogen on &@&6). From Grimm's hydridedisplacement law, [BH,] should act on the one hand as an anionlike [OH]- or [NH2]-, or on the other hand like [NH,]+ or [CH,], sosalts such as k[BH,] should be capable of existence. I n connexionwith active hydrogen, reference should be made to the paper byH. Kroepelin and E. V0ge1,~~ who have investigated the action ofthis gas on nearly 90 inorganic salts.By the action of carbon monoxide under pressure on B2H6,A.B. Burg and H. I. Schlesinger 36 have obtained borine carbonyl :B2H, + 2CO 3 2BH,CO. The molecular s-tate was found fromvapour-density measurements, and it is suggested that thecompound is held together by a co-ordinate link, the electronsarising from the carbon monoxide. An addition compound isformed with ammonia, BH,CO + 2NH, + BH,CO(NH,),.With triethylamine the following reaction takes placeThe last compound is very stable and resists heating at 125" forsome hours. From measurements of the rate of decomposition ofthe carbonyl, it is concluded that BH, has a, transitory existence.In the early days, diborane B,H6 was considered to be akin toBM,CO + (CH3)3N + (CH,)zN,BH, + CQ34 A.Stock and H. Laudenklos, 2. anorg. Chern., 1936, 228, 178.3 5 Ibid., 1936, 229, 1. 36 J. Arner. Chem. SOC., 1937, 59, 780152 INORGANIC CHEMISTRY.ethane. Subsequent research modified this view and tended tolink it up with ethylene. Its unsaturated character was indicatedby its behaviour in liquid ammonia, in which an addition compoundB&,(NH,), is formed, and by the action of alkali metals. Onelectrolysis in ammonia, B,H,(NH,)2 behaves as a salt and isregarded by Stock as [B,H4](NH4), :[B,H41(NH4)2 2 62H4 + 2hH4This leads to the conception of B,H, as an H,[B,H4]. Thebivalent negative radical B2H4 has an ethane-like structure, theboron and carbon in each case being four-covalent. This is morestable than the ethylenic B,H,, which accounts for the stabilityof the metallic boranes.Yurther support for the ethylenic nature of B2H6 is claimed fromparachor measurements 38 [the value 121 -9 harrnonises best with[H,-B = B-H,]--H,' '-, i.e., with two single parachors for 2 boronatoms (2 x 16.4), 4 hydrogen atoms (4 x 17.1), a double link(23.2), and 2 electrovalencies (- 2 x 1.6) = 121.21, from ultra-violetabsorption spectra, dipole, and magnetic measurements.On the conception of the boron hydrides as acids, it is possible tobuild up structures for the two series : B,(H,+4), e.g., B,H&B,H,, etc.; and Bn(HnA6), e.g., B4HI0, B5Hll, etc. The extraelectron required to complete the boron octet and for it to assumea neon-like structure is derived electrovalently from hydrionsoutside the co-ordinate sphere.Recent determinations of the structure of diborane by electrondiffraction 39 are a t variance with any such formuh, and it appearsthat any structure with a double bond between the boron atoms isuntenable. The values obtained for the B-B and the B-Hdistance were 1-86 and 1.27, respectively, and shortening of theB-B bond below the single-bond value was not observed, whichrules out the double bond. The B-H distances are greater thanthe single-bond separations for other diatomic hydrides, so thebonds are weaker than single ones. The distances are compatiblewith structures representing resonance among the seven possiblearrangements,- -H H .. .. H H .. ..H:B:BH H:B B:H etc.H H H Hgiving each 6/7 single and 1/7 zero bond character.37 E. Wiberg, Rer., 1936, 69, 2816.38 A. Stock, E. Wiberg, and W. Mathing, ibid., p. 2811.39 5. H. Bauer, J. Amer. Chem. SOC., 1937, 59, 1096; cf. also Bauer andPauling, ibid., 1936, 58, 2403TERREY : SOME ELEMENTS AND COMPOUNDS. 153Phosphorus Hydrides.-In last year’s Reports 4O doubt wasexpressed with regard to the existence of the higher hydrides ofphosphorus. Further work by P. Royen 41 has confirmed the viewthat these are merely sorption complexes of phosphine and yellowamorphous phosphorus of indefinite composition. The solid“ hydride ” PI2H6, obtained by the decomposition of P,H4, iswithout structure and evolves only phosphine on heating; nohydrogen is given off until the decomposition temperature ofphosphine is reached. The rate of decomposition falls withdecreasing hydrogen content, but no discontinuity is shown a t thesecond solid hydride (P,H2) stage. The desorption is irreversible,although by irradiating white phosphorus and phosphine under30-40 atmospheres’ pressure an amorphous product containing upto 14% of phosphine is formed.With regard to the addition compounds which these hydridesform with piperidine, whence it was deduced that they possessedacidic properties, it is now shown that these also are adsorptioncomplexes. In the addition of piperidine, part of the phosphorusis replaced but there is no stoicheiometric relationship between thephosphorus, hydrogen, and piperidine.I n conclusion, the Reporter feels that the congratulations of allinorganic chemists must be showered on Dr. J. W. Mellor, F.R.S.,on the completion of his stupendous work.42 H. T.H. TERREY.0. J. WALKER.40 Ann. Reports, 1936, 83, 185.4 1 2. anorg. Chem., 1936, 229, 369; cf. I. Mathieson and W. Wrigge, ibid.,42 “ A Comprehensive Treatise on Inorganic and Theoretical Chemistry.”1937,232,284