484 KIPPING An’b ROBISON :LL--Organic Derivatives of Silicorl. Part XU.The Condensation Prodztcts of D ~ ~ ~ ~ e r ~ ~ l s i l i c c l n e d i o By FREDERIC STANLEY KIPPING and ROBERT ROBISON.IN the present paper we record the results of a further study ofdiplienylsilicanediol and of the open- and closed-chain condensationproducts derived from this diol (Kipping, T., 1912, 101, 2108,2125). The compounds which have now been obtained are of thefollowing types, of which No. IV has not hitherto been described:I.11.11’1. rv.V.VI.HO SiPh,* OH.HO*SiPh,*O*SiPh,*OH.HO*SiPh,*O*SiPh,*O*SiPh2* OH.It will be seen that diphenylsilicone, SiPh,O, and aIso the closed-chain compound, Ph,Si<O>SiPh,, which theoretically might beformed from anhydrobisdiphenylsilicanediol (II), have not yet* I n a previons paper (T., 1912, 101, foot note to p.210) I mentioned thecircumstances which led tc, the pnblication by Martin of a paper on some siliconcompounds (Be?.., 1912, 45, 403), and I poiiitcd out that hc had given a “veryerroneous ” account of diphenylsilicaiiediol ; hnving speiit a good deal of time iutproving the iuaccuracy of his results, I was able t o suggest exactly liow mid why, i nmy opinion, he had fallen into error. In liis reply (T., 1913, 103, 119), Martintried to show that his results were practically iJentical with rniue a i d that I liadimisrepresented him.As to how far there is any resemblance between Martin’s account of dipheiiy1-silicanediol and my own, readeis of the original papers (not Martin’s tabularcomparisons) may judge.Whether 1 used the word isomeride or modification iii~reference to Martin’s two AlodQikntiowen appears to nie to be iiiiiiiateiial. as in niyopinion his two niodifications have 110 existence. I was not critici5ing his views, but;.his experimental results. His 1)aper certainly suggested that lie regarded his.Jfodijkdionen as isomerides, but if, as he now iniplies, he considered them to-be merely different crystallographic forms, then his iiiethods for the conveisioirof the one into the other seem inexplicable.In a later paper (Ber., 1912, 45, 2097), Martin puLlished without niyIbermission some further results of work carried out a t my suggestion and under mysupervision. I think it is only right to say that, after carefully coilsidering all t h eexperimental data at the time, I told hiin that I could not conseiit to t h cpublication of the results.The whole of the evidcnce indicated that all the‘‘ compounds ” subsequently dcscribed by him were complex mixtures and in myoliiiion the structural f o r i i m l ~ he assigns to these compounds arc devoid o ffouudntion. I?. s. K.ORGANIC DERIVATIVES OF SILICON. PART XXI. 485been isolated. Although we have made several attempts to preparethese substances from diphenylsilicanediol and from anhydrobis-diphenylsilicanediol, we have not obtained any indication of theirexistence; these facts, together with the negative results of ourefforts to produce the corresponding benzyl derivatives (Robisonand Kipping, this vol., p.40), seem to show that (a) the simplesilicones, ths analogues of the ketones, cannot be prepared fromthe diols, SiR2(OH)2, owing to the readiness with which the latterundergo condensation, and ( 6 ) a closed-chain consisting of twosilicon and two oxygen atoms linked together alternately is veryunstable compared with one containing three or four atoms of bothof these elements.The conditions under which diphenylsilicanediol and each of itsopen-chain condensation products undergoes change are, on thewhole, very similar to those which are operative in the case of thecorresponding benzyl derivatives. Small quantities of alkalihydroxide, or of piperidine, or of hydrochloric acid bring aboutcondensation, with formation of one or both of the closed-chaincompounds.The action, however, does not consist merely in theelimination of the elements of water ; in all cases, apparently,hydrolysis and condensation occur simultaneously, and the natureof the final product is largely determined by that of the solventemployed and by the other conditions which prevail; thus, althoughanhydrobisdiphenylsilicanediol in ethereal solution seems to bemainly converted into tetra-anhydrotetrakisdiphenylsilicanediol inthe presence of a small quantity of piperidine, an appreciableproportion of trianhydrotrisdiphenylsilicanediol is also produced.Under similar conditions, dianhydrotrisdiphenylsilicanediol doesnot give pure trianhydrotrisdiphenylsilicanediol, but a mixture ofthe latter witn a very considerable proportion of tetra-anhydro-tetrakisdiphenylsilicanediol.That the formation of any closed-chain compound from thecorresponding open-chain compound is a reversible reaction is alsoclearly proved by the preparation of dianhydrotrisdiphenylsilicane-diol from the trianhydro-derivative by thc action of sodium hydr-oxide in acetone-ethereal solution ; sirnilarly, it has been foundthat trisnliydrotetrakisdiphenylsilicanediol may be obtained fromthe tetra-anhydro-compound by hydrolysing the latter with sodiumethoxide in alcoholic chloroform solution, and may be reconvertedinto tetra-anhydrotetrakisdiphenylsilicanediol with the aid ofsodium hydroxide or hydrochloric acid in alcoholic solution.I n view of the facts just stated, it is clear that the relativesolubilities of the two closed-chain compounds in a given sblventwould have a preponderating influence on the relative proportionsK K 486 ICII~PING AKD KOBISON :in which the two substances would be obtained by evaporating asolution of any of the open-chain compounds in the given solventwith a particular reagent.Consequently, the results of experimentssuch as those referred to above give very little information as tothe relative stabilities of the two closed chains, containing, respec-tively, three and four atoms of silicon and of oxygen.It would seem, nevertheless, that a fairly strong inference maybe drawn from a direct comparison of the behaviour of anhydro-bis- and dianhydrotris-diphenylsilicanediol towards hydrochloricacid in methyl-alcoholic solution.Both these compounds arecsonverted into trianhydrotrisdiphenylsilicanediol, and tetra-anhydrot'etrakisdiphenylsilicanediol is not formed in appreciablequantities in either case; but the rapidity with which the condensa-tion occurs in the case of dianhydrotrisdiphenylsilicanediol is verymuch greater than in that of anhydrobisdiphenylsilicanediol. Itseems, in fact, that the closed chain of three silicon and threeoxygen atoms is easily and directly produced from dianhydrotris-diphenylsilicanediol, but that tetra-anhydrotetrakisdiphenylsilicane-diol, in spite of its slight solubility in methyl arcohol, is not formedfrom anhydrobisdiphenylsilicanediol, owing to the relativeinstability of the larger closed chain.A similar conclusion, namely, that the closed chain of threesilicon and three oxygen atoms is the more stable one, may bedrawn from the results of other experiments in which the condi-tions do not favour, although they do not preclude, hydrolysis;thus by the action of heat on diphenylsilicanediol a much largerproportion of trianhydrotris- than of tetra-anhydrotetrakis-diphenylsilicanediol is formed ; again, when dianhydrotrisdiphenyl-silicanediol is heated it is almost quantitatively converted intotrianhydrotrisdiphenylsilicanediol, whereas under similar conditionsanhydrobisdiphenylsilicanediol gives a mixture of approximatelyequal quantities of trianhydrotris- and tetra-anhydrotetrakis-dipheriylsilicanediol (Kipping, Zoc.cit.).From these facts, and from the failure of our attempts to obtaina benzyl derivative corresponding with tetra-anhydrotetrakis-diphenylsilicanediol, we infer that the heterocyclic group of six ismore stable than that of eight atoms.The crystallographic examination of the compounds described inthis paper was carried out by Mr. Vernon Stott under the super-vision of Mr. A. Hutchinson, of the Mineralogical Laboratory,Cambridge; we are greatly indebted to these gentlemen for thereports which they have given usORGANIC DERIVA'I'IVES OF SILICON. PART XXI. 487EXPERIMENTAL.Z~iphenylsilicaizedol.The pure specimens of this compound, prepared more than ayear ago (loc. cit.), have retained their properties; five melting-point tubes containing the crushed powder from one of the speci-mens were heated simultaneously, and in all cases the substanceliquefied with effervescence at 128-1 32O.Diphenylsilicanediol,therefore, is stable a t the ordinary temperature, and the greatdifficulties which attend its isolation (loc. cit.) are due to itssensibility towards alkalis, acids, and other reagents. The crystalsdeposited by the spontaneous evaporation of its solution in ethylacetate were examined.System : anorthic. Subclass : holohedral.a: b : c=0*5657: 1 : 1.700.a = 90°2' ; = 11 1O20' ; y = 87O27'.Forms developed : b(010), nz(llO), p(llO), c(OOl), t(013), ~(116).Angle.110 : 110i i o : oio010 : 001210 : 001Iio : 001010 : 013110 : 013i i o : 013010 : lA0No.161615222626311Table of Angles.Limits.63'2' -65'33'53'10'-55'52'59'34'-6 1'37'89°62'-90030'70°28'-71' 16'7Oo49'-7l057'58"20'-6 1'40'59"26'88'38'Mean.Calculated.64"23' -55'40' -60'51' 60'57'90°11' -70'55' 70"5G'71"19' -60"31' -59'25' 60'1'88'38' 87'51'' The crystals are prismatic in habit. Of the prism faces b , m,and p, b and p are well developed, whilst m is but a small face.The angles of the prism zone, namely, bm=64O23/, m,u=55°40/,pb = 60°57/ all approximate to 60°, and give the crystals a pseudo-hexagonal character. The crystals are terminated by three planesa t each end. These are c(OOl), t(013), and v(i16). The face c(OO1)is quite large, whilst t and v are tiny bevels on the edges cb andcp' respectively.The faces t and v are very bad ones, particularlythe latter, from which no trustworthy measurements could be ob-tained. The faces as a whole were very bad from a goniometricalpoint of view, as may be seen from the range of the angles in thepreceding table.Eflect of Heat 0% Diphenylsilicanediol.Some experiments on the action of heat on diphenylsilicanediolhave already been described (Loc. cit., p. 2137). At 140-180° th488 KIPPING AND ROBISOX :diol loses approximately one molecule of water, and gives as theprincipal product trianhydrotrisdiphenylsilicanediol. Furtherexperiments hhve shown that under these conditions a small pro-portion of the diol is converted into tetra-anhydrotetrakisdiphenyl-silicanediol, but even when these two compounds have been removedas far as possible there remains some oil which is readily soluble inpetroleum containing a little chloroform.As it seemed possible that this oil might contain diphenyl-silicone or diaiihydrobisdiphenylsilicanediol or both these com-pounds, a weighed quantity of the pure substance was heated in axylene bath atl about 125-128O.I n the course of some hours itchanged into a transparent glue, and at the end of about fifteenhours the total loss in weight amounted to 7.5 per cent., but theweight had not become quite constant; the theoretical loss for1 mol. of water is 8.3 per cent. The gelatinous or vitreous massobtained in this way showed no signs of crystallising even whenkept during some months, and it did not crystallise when it wastreated with a little ethyl acetate; this last f a c t showed that,unlike the vitreous product formed at 140-180°, it did notcontain a large proportion of trianhydrotrisdiphenylsilicanediol.The product was fractionally extracted with warm light petroleum,in which it was moderately easily soluble, and finally the wholepassed into solution.The various extracts were then fractionallycrystallised from mixtures of chloroform and light' petroleum, andin this way a small quantity of dianhydrotrisdiphenylsilicanediolwas isolated; the rest of the crystalline material seemed t o consistof a mixture of anhydrobis- and dianhydrotris-diphenylsilicanediol,but a considerable proportion of the original product remained a8an oil.Another sample of the pure diol was heated in a xylene bathat abouf 125-128O; the loss in weight was about 6.8 per cent.at the end of six hours, and 7.8 per cent. at the end of abouttwenty-six hours.From the product, anhydrobis- and dianhydro-tris-diphenylsilicanediol were isolated, but again there remained aconsiderable proportion of an oil, from which crystals could not beobtained, and neither of the closed-chain compounds seemed to havebeen produced.As the loss in weight in these two experiments was not verydifferent from that (8.5 per cent.) required for the elimination ofone molecule of water from one molecule of the diol, and wastherefore much greater than that corresponding with the produc-tion of dianhydrotrisdiphenylsilicanediol, or of a mixture of thiscompound and anhydrobisdiphenylsilicanediol, the results pointedto the formation of diphenylsilicone.The further examination oORGAKIC DERIVATIVES OF SILICON. PART xxi. 489the oily products just referred to did not, however, bear out thisview; when these oils were heated a t about 200° a vigorous effer-vescence occurred, owing t o the liberation of steam, and theproducts, treated with ethyl acetate, immediately crystallised. Thesolids thus obtained gave a small proportion of a very sparinglysoluble powder, and crystals of trianhydrotrisdiphenylsilicanediol,together with what seemed to be tetra-anhydrotetrakisdiphenyl-silicanediol. The behaviour of these oily products was, in fact, justthat of a mixture of anhydrobis- and dianhydrotris-diphenylsilicane-diol. When these oily products were submitted to distillation insteam tho distillate was free from any visible suspended matter;since diphenylsilicone would probably be volatile in steam, thisfact.might be taken as evidence of the absence of this compound.If diphenylsilicone and dianhydrobisdiphenylsilicanediol are notformed when diphenylsilicanediol is heated, i t would seem that apart of the loss which occurs must be due to thO volatilisation ofthe diol or of its condensation products, or to ite oxidation,followed by the volatilisation of some of the products of oxidation.A weighed quantity (0.478 gram) of the diol was therefore heatedin a bromobenzene bath a t a temperature which varied from130-140° in order to ascertain whether a loss greater than thetheoretical would take place.A t the end of six hours the loss was7.1 per cent., a t the end of nine hours 7.8 per cent., at the endof fourteen hours 8.6 per cent'., a t the end of sixteen and a-halfhours 9 per cent., and at the end of twenty-two hours i t was 9.5 percent. During the first period of heating the sides of the weighingbottle containing the substance became coated with oily drops,which extended right up to the neck of the bottle, but no increasein this deposit seemed to occur during the remaining periods.The vitreous product when rapidly heated to and kept a t 200°during about five minutes evolved vapours which could be clearlyseen; the total loss in weight had then reached 10 per cent.It isevident, therefore, that since the loss in weight is not entirely dueto the escape of water vapour, the quantitative results cannot betaken as pointing to the formation of diphenylsilicone. The oilymather, in which the presence of this compound was suspected,is, no doubt, merely a mixture of anhydrobis- and dianhydrotris-diphenylsilicanediol, the separation of which, as already shown, isa task of very considerable difficulty, particularly when only smallquantities of material are available.Action of Heat on Dipherylsilicanediol in Xylene Solution.diol was heated in boiling toluene and boiling xylene solution,As these experiments failed to give the desired products, the pureI490 KIPPING AND ROBISON :is moderately soluble in both these liquids at their boiling points,and the hot saturated solutions give when cooled a considerabledeposit of the unchanged substance; if, however, the solutions areheated for a short time, decomposition sets in, with liberation ofsteam.I n experiments with each of the solvents, in which thesolutions were boiled for a short time only, dianhydrotrisdiphenyl-silicanediol was isolated from the product. A xylene solution whichhad been boiled during one hour gave on evaporation a productwhich was readily and completely soluble in cold chloroform, anddid not, therefore, contain any considerable proportion of theunchanged diol ; from this product a large proportion of trianhydro-trisdiphenylsilicanediol was isolated, and smaller proportions ofanhydrobis- and dianhydrotris-diphenylsilicanediol. There thenremained a relatively very small quantity of a viscid oil, fromwhich crystals could not be obtained.All these results seem toshow that diphenylsilicone and dianhydrobisdiphenylsilicanediolare not formed in appreciable quantities by heating the diol aloneor in solution; the diol first undergoes condensation, givinganhydrobis- and dianhydrotris-diphenylsilicanediol, and if thetemperature is sufficiently high these compounds are almost com-pletely transformed into trianhydrotris- and tetra-anhydrotetrakis-diphenylsilicanediol, the former of which is always produced inmuch the larger proportion.Action of Piperidine on Diphenylsilicanediol.Diphenylsilicanediol undergoes condensation in acetone solutionin the presence of a trace of piyeridine.When such a solutionwas allowed to evaporate spontaneously, it gave a solid, somewhatsticky residue, which was readily and completely soluble in coldchloroform, and therefore did not contain any appreciable propor-tion of the original diol. This residue was gently warmed on awater-bath in order t o expel the piperidine, and then dissolvedin ethyl acetate; the solution ultimately deposited crystals oftetra-anhydrotetrakisdiphenylsilicanediol (m. p. 200°), togetherwith a small proportion of a colourless powder, which was onlysparingly soluble in ethyl acetate, and probably represented acondensation product more complex than tetra-anhydrotetrakis-diphenylsilicanediol. The mother liquors from the tetra-anhydro-compound gave crystals melting indefinitely from about 160° to170°, which were probably a mixture of trianhydrotris- and tetra-anhydrotetrakis-diphenylsilicanediol, but the presence of the formerwas not conclusively establishedORGANIC DERIVATIVES OF SILICON.PART XXI. 491A ction of Ammoluium Hydroxide o n Diphenylsilicanediot.I n the presence of a little a.mmonium hydroxide, diphenyl-silicanediol undergoes condensation in acetone solution a t theordinary temperature, and the solution deposits an oil. From thisproduct, with the aid of chloroform and light petroleum, crystalsmelting a t 113O, which dissolved in a 5 per cent. solution ofpotassium hydroxide, were obtained ; this substance was anhydro-bisdiphenylsilicanediol. In another experiment, carried out in asimilar manner, the oily condensation product gave crystals melt-ing a t 112O, which were insoluble in a 5 per cent.solution ofpotassium hydroxide; this product was dianhydrotrisdiphenyl-silicanediol. The closed-chain condensation products were notobtained in these experiments.The crystals deposited from a mixture of chloroform and lightSystem : anorthic.petroleum were examined goniometrically.Sub-class : holohedral.a : b : c = 0.6536 : 1 : 1.868.a = 92 O 2 8 f ; j3 = 1 15O40' ; y = 86O24f.Forms developed : b(010), m(110), &TO),g(iT4).Table of Angles.Angle.010 : 110110 : 1x0110 : 010010 : 011li0 : 011110 : 011010 : 0011 l O : 001110 : 001011 : 001110 : Ti2010 : ii2 iio : 112No.10101020242418202011242424Limits.61'6' -62'1'60' 10'-6 1'33'56O40'-57O56'29'52'-30'42'52' 18'-53'14'105'44'-106'39'88O38'-89"29'66'41'-67'25'68'26'-68'44'58'8' -59'9'142'43'-143'19'119'5' -1 19'55'105'22'-106'38'Mean.61'36'60'58'57'26'30'22'52'49'106'8'89'4'67'3'68'37'58'38'143'1'119'31'106'1'Calculated. -57'26'52'53'106'14'89'0'69"8'143O2 1 '119'22'106'11'---The forms b, m, and p constitute a prism zone; the forms b andm are large and p small.The angles of the prism zone are:bm = 61O361, mp = 60°58/, p b l = 57O26'. They all approach thevalue 60°, and so the crystals are pseudo-hexagonal. Of the threeterminal forms, c(OO1) is much the large&, q and s being smallfaces on the edges bc and mlc rapectively492 KIPPING AND ROBISON :*4 c t io n, of Pip e ridin e on A n h.y dro b isdi ph e n y 1 silican e dio 1.When anhydrobisdipheuylsilicanediol is heated, it gives tetra-anhydrotetrakis- and some trianhydrotris-diphenylsilicanediol,together with a small proportion of a colourless powder, which isprobably a highly complex condensation product (loc. cit. p. 2133);when treated with acetyl chloride it is converted principally intotetra-anhydrotetrakisdiphenylsilicanediol.I n ethereal solution, in the presence of a small quantity ofpiperidine, it undergoes change, giving a mixture of products,which remains as an oil when the solution evaporates spontane-ously ; from this oil the transparent, rectangular crystals meltinga t about 184-186O (loc.cit., p. 2140) were obtained; this observa-tion eh0u.s that both trianhydrotris- and tetra-anhydrotetrakis-diphenylsilicanediol had been produced, the former, however, inrelatively small quantities. The action of the piperidine, therefore,is two-fold; it not only brings about the condensation of theanhydrobisdiphenylsilicanediol, but also1 hydrolyses some of thiscompound to diphenylsilicanediol, which then undergoes condensa-tion, giving a little trianhydrotrisdiphenylsilicanediol.A ction. of Hydrochloric .4 cid 07% ,.ln,hydrobisd~phe?~ylsilicanediot.When a very small quantity of hydrochloric acid is added to asolution of anhydrobisdiphenylsilicanediol in methyl alcohol, novisible change occurs immediately, but after some time a smallquantity of an oil is deposited.The solution, left to evaporatea t €he ordinary temperature, gives a residue which is partly crystal-line, partly oily. I f tho whole is treated again with methyl alcoholcontaining a little hydrochldric acid, and the solution is left t oevaporate, the residue then contains a larger proportion of thecrystalline product than before. This substance when freed fromoil with the aid of alcohol, separates from ethyl acetate in colourlessprisms, melting a t 188O. It consists of pure trianhydrotris-diphenylsilicanediol. The presence of the tetra-anhydro-derivativein the crude, crystalline product could not be detested.Hydrochloric acid, theref ore, under the conditions stated, doesnot bring about the condensation of anhydrobisdiphenylsilicanediol,but hydrolyses i t to the simple diol, which then undergoes con-densation.Dimhydro trisdip~,ei~,ykilicacrt ediol.The crystals deposited from a mixture of chloroform and lightpetroleum were examinedORGANIC DERIVATIVES OF SILICON.PART XXI. 493System : anorthic. Sub-class : holohedral.a : b : c =0.5068 : 1 : 1.491.a = 90°42/; = 106'561 ; y == 86'30'.Forms developed : b(010), m(110), p(liO), c(OOl), r(012),I I <i 14).Table of Angles.Angle. No. Limits. Menn. Calculated.010: 1lO 6 66'42'-67'21' 66'58'110: 010 10 60'4'-6 1'53' 61'32' 61'2301O:OOl 16 89"7'-90'5 1' 90'19' -11O:OOl 15 74' 16'-Ri028' 75"l' -010:012 1 54'42' 54'42' -110: 012 1 63'1 1' 63"ll' 63'33'010:114 3 67'34'-69'11' 68'36' 69'25'110 : 114 3 120'28' -124'2' 122'47' 120'57'110 : Ti4 3 103'44' -105'14' 104"27' 203'0'-110: 1;o 7 51'4'-52"4' 51'39' -1JO:OOl 8 74"9'-74"55' 74O35' 74'34'110:012 1 94'50' 94'50' 93"54#The dominant forms are: b(010), p(l'iO), and ~(001).Bfanycrystals are simply four-sided prisms, terminated by c planes. Theprism face m is often missing, and when present it is only small.The face r(Ol2) was present on one crystal only. The form u(5'I 4 )was only developed on two crystals, and even then the faces werevery poor ones. This accounts for the divergence between thecalculated and measured angles. The crystals may be described asshort, stumpy prisms.The prism angles show some approach tohexagonal development : b m = 66O58/ ; mp = 51O39' ; pb' = 61O32'.If in the crystals of diphenylsilicanediol t is given the indices(011) instead of (013), then v will become (i12) instead of (116).Ths remaining indices would be unchanged. The axial ratioswould now be 0.5657 : 1 : 0.5666 instead of 0.5657 : 1 : 1.700.Similarly, if in the crystals of dianhydrotrisdiphenylsilicanediolr is made (011) instead of (012), then u would be (112) instead of(114), and the remaining indices would be unaltered. The axialratios would be:instead of 0.5068 : 1 : 1.491.To bring the axial ratios of the crystals of anhpdrobisdiphenyl-silicanedid into line, q would have to be made (031) instead of(Oll), and the ratios would then be 0.6536 : 1 :0*6227 instead of0.6536 : 1 : 1.868.'0.5068 : 1 : 0.7455,The indices of ~(1x2) would then become (332).Action.of Piperidine on Dianhydrotrisdiphenylsilicanediol.When an ethereal solution of dianhydrot#risdiphenylsilicanediol,to which a drop of piperidine had been added, was left to evaporateslowly at the ordinary temperature, i t gave a somewhat brown an494 KIPPING AND ROBISON :oily residue, from which, with the aid of ethyl acetate, there wasseparated a small proportion of a colourless, very sparingly solublepowder. The ethyl acetate solution gave a crystalline deposit, fromwhich pure tetra-anhydrotktrakisdiphenylsilicanediol was isolated.The mother liquors from this compound seemed t o contain tri-anhydrotrisdiphenylsilicanediol, but the latter was not identified withcertainty.It is obvious from these results that dianhydrotris-,like anhydrobis-diplienylsilicanediol, is hydrolysed by piperidine,the products of hydrolysis then undergoing condensation to theclosed-chain compounds.I n these circumstances the nature of the product or productsdoubtless varies with the conditions of the experiment, but acomparison of the results obtained by the action of piperidineon diphenylsilicanediol, anhydrobisdiphenylsilicanediol, and di-anhydrotrisdiphenylsilicanediol seems to show that the principalproduct is in all cases tetra-anhydrotetrakisdiphenylsilicanediol.The A ction of Hydrochloric A cid on Dianhydrotrisdiphenyl-siZicanedio1,The action of hydrochloric acid on dianhydrotrisdiphenylsilicane-diol in methyl-alcoholic solution is very different from that of thesame acid on anhydrobisdiphenylsilicanediol. After a few momentsthe solution becomes quite milky, owing to the separation of anoil, the quantity of which rapidly increases, and when the solutionis evaporated at the ordinary temperature it gives a crystallineresidue, which is free from any appreciable quantity of oil.Thecrystalline product seemed to be practically pure ; when recrystal-lised from ethyl acetate it gave colourless prisms of pure trianhydro-trisdiphenylsilicanediol, melting sharply a t 188O.Trianhiy dro trisdiph e n ylsilicanediol.This closed-chain compound is hydrolysed by alcoholic potassiumhydroxide, and finally gives a solution of the potassium derivativeof diphenylsilicanediol ; when the alcohol is evaporated and theresidue is treated with a slight excess of dilute acetic acid, the diolis precipitated in the usual form (Kipping, loc.cit., p. 2116).If, however, hydrolysis is carried out very cautiously and theprocess is interrupted a t a very early stage, the following reactionmay be realised:SiPh2*0 O<siph .O>SiPb, + H,O = HO*SiPh,*O*SiPh,*O*SiPh,*OH.2For this purpose the trianhydro-derivative (0.25 gram) wasdissolved. in a mixture of acetone and a little ether, and a 3 percent. aqueous solution of sodium hydroxide (0.05 gram) was added;a few seconds later the solution was acidified with dilute acetic acidORGANIC DERIVATIVES OF SILICON. PART XXI.495and the acetone and ether were rapidly evaporated a t the ordinarytemperature in a stream of air. The product wm extracted withether, the ether was evaporated, and the oily residue was treatedwith a little alcohol; the small proportion of insoluble matter,which doubtless consisted of unchanged trianhydrotrisdiphenyl-silicanediol, was then separated by filtration. The alcoholic solu-tion, when rapidly evaporated at the ordinary temperature, gavean oil, which separated from a mixture of chloroform and lightpetroleum in colourless crystals ; this product was pure dianhydro-trisd iphenylsilicanediol.Trianhydrotrisdiphenylsilicanediol, like the corresponding benzylderivative, may also be converted into the open-chain compound byhydrolysing it with hydrogen chloride.The trianhydro-compoundwas dissolved in a mixture of chloroform and acetone, and a fewdrops of concentrated hydrochloric acid were added ; after sixhours the solution was neutralised with ammonium hydroxide,and the solvents were rapidly evaporated at the ordinary tempera-ture. The somewhat pasty residue was extracted with alcohol, andthe solution was filtered from unchanged trianhydrotrisdiphenyl-silicanediol and evaporated ; the oily residue consisted essentiallyof dianhydrotrisdiphenylsilicanediol, which wits obtained in a stateof purity by recrystallisation from a mixture of chloroform andlight petrolgum.Since hydrogen chloride condenses dianhydrotrisdiphenylsilicane-diol, the above hydrolysis is a readily reversible reaction, just ain the case of the corresponding benzyl derivative (Robison andKipping, this vol., p.40). The action of piperidine on tri-anhydrotrisdiphenylsilicanediol was also examined in acetone solu-tion. The product was a mixture, from which a small proportion of apowder, practically insoluble in amtone, wit^ isolated ; the remain-der seemed to consist of a mixture of trianhydrotris- and tetra-anhy drotetrakis-diphenylsilicanediol.The crystals of trianhydrotrisdiphenylsilicanediol, deposited fromethyl acetate solution, were measured.System : orthorhombic. Sub-class : bisphenoidal.Forms developed : a(100), b(010), m(110), ~ ( 1 1 1 ) .a : b : c = 0.7750 : 1 : 0.4993.Table of Angles.Angle. No.100:010 19100: 110 19010: 110 19100: 111 17010: 111 18110:!_11 16110: 111 121 1 1 : l ~ ~ sLimits.89'50'-90'7'37O2 1'-38' 12'51'55'-52'29'59O54'-6Oo 17 '66'58'47'27'50' 34'-5 1 ' 1 3'80'30'-81'27'7 808' -7 ~ 2 9 'Mean.Calculated.90"o' 90'0'37'48' 37O48'52'12' -60'7' 60'3'67'14' -50"48+' 50'50'78O15' 78'23'81'3' 80'57496 KIPPING AND ROBISON :The crystals occur in two well-marked habits. One of these isprismatic, u and b being large faces, whilst the m faces are small.I n other cases the crystals are markedly tetrahedral.7'rianhydro t e t rali isdiphen y Zsilicnnediol,HO*SiPh,*O*SiYh,*O* SiPh,*O*SiPh,*OH.This condensation product of diphenylsilicanediol, the mostcomplex open-chain silicon compound so far obtained, may beprepared by cautiously hydrolysing tetra-anhydrotetrakisdiphenyl-silicanediol,SiPh2*o'SiPh2>0 + H,O = HO~SiPi~;O*SiP~,~O~SiPh~~O~SiPb,.OH O'Si Ph 2*0 S I PI)When the closed-chain compound is warmed with excess ofalcoholic potassium hydroxide, it seems to be completely hydro-lysed, giving the potassium derivative of diphenylsilicanediol, butowing t o the fact that tetra-anhydrotetrakisdiphenylsilicanediol isalmost insoluble in alcohol, the reaction takes place rather slowly.I n acetone solution hydrolysis takes place more rapidly, as thetetra-anhydro-derivative is more soluble in this liquid, but it isdifficult to interrupt the process a t the desired stage.Hydrogenchloride seems to have little effect on tetra-anhydrotetrakisdiphenyl-silicanediol ; when solutions of the substance in acetone or ether,or acetone and chloroform, are treated with a little concentratedhydrochloric acid, and kept during periods varying from thirtyminutes to one day, most of the original compound is recoveredunchanged, and only traces of a product soluble in cold alcoholare obtained.Under the following conditions, however, the partial hydrolysisof tetra-anhydrotetrakisdiphenylsilicanediol may be accomplished.The closed-chain compound is dissolved in chloroform, and a con-siderable excess (5-6 mols.) of an alcoholic solution of sodiumethoxide is added, care being taken to use sufficient chloroform toprevent the precipitation of the anhydro-derivative. After aninterval of not more than one minute, a slight excess of very diluteacetic acid is added, the liquids are well shaken together, the chloro-form solution i s separated, and the aqueous solutmn is extractedwith a little chloroform.The combined chloroform extracts arerapidly evaporated at the ordinary temperature, and the solidresidue is extracted two or three times with cold ethyl alcohol,which leaves undissolved a considerable proportion of the originaltetra-anhydrotetrakisdiphenylsilicanediol. The filtered alcoholicextract is diluted with water and vigorously stirred; the precipi-tated trianhydrotetrakisdiphenylsilicanediol is then separated bORGANIC DERIVATIVES OF SILfCOK. PART XXI. 497filtration, dried in the air, and crystallised several times froma mixture of chloroform and light petroleum.The yield by this method is rather poor, but the crude substanceis not contaminated to any great extent with other products ofhydrolysis ; moreover, the unchanged tetra-anhydrotetrakisdiphenyl-silicanediol, which is recovsred, may be treated again, and theoperations may be repeated until practically the whole of theoriginal substance has been transformed into trianhydrotetrakis-diphenylsilicanediol. When attempts are made to increase the yieldby alloying a longer time to elapse between the addition of thesodium ethoxide and the neutralisation with acetic acid, althougha larger proportion of the tetra-anhydro-compound is hydrolysed,the product is a mixture, from which it is difficult to isolate thecomponents; this is so even when the theoretical quantity ofsodium ethoxide is used.Trianhydrot etra~:isd~~hennylsilicanediol separates from a mixtureof ether and light petroleum in short, colourless prisms; whenheated moderately quickly it melts sharply at 128*5O, but whenheated very slowly it sinters a t about 127O, probably owing toincipient decomposition.The samples for analysis were dried over sulphuric acid:0.1572 gave 0,4091 CO, and 0.0736 H,O.C= 71.0 ; H = 5.2.0.1965 ,? 0.5124 CO, ,? 0.0932 H,O. C=71*1; H=5*3.The molecular weight was determined by the cryoscopic method0.262 in 12.1 benzene gave At -0'12O.C,8H4,0,Si4 requires C = 71.0 ; H = 5.2 per cent.in benzene solution :M.W.=884.0-494 ,) 12.1 ,, ,, A t -0.23'. M.W.=870.C4,R,O,Si, requires M.W.= 811.These results, like those obtained under similar conditions inthe case of the other open-chain condensation products of diphenyl-silicanediol, are higher than the theoretical values, but the extentof association, as indicated by the experimental results, rapidly fallsas the molecular weight of the condensation product rises; this isshown by the following data:MW. M.W. Difference,(Found). (Calculated). per cent.Anhydrobisdiphenylsiicanediol ......... 573 414 38Dianhydrotrisdiphenylsilicanediol . . . . . . 755 613 23Trianhydrotetrakisdiphenylsilicanediol 877 81 1 8Trianhydrotetrakisdiphenylsilicanediol resembles the other twoopen-chain condensation products of diphenylsilicanediol. Itseparates from a mixture of ether and light petroleum in mmsive,transparent prisms, and is readily soluble in chloroform, acetone,and most of the ordinary solvents, but only sparingly so in col498 KIPPING AND ROBISON :ethyl alcohol, and practically insoluble in cold light petroleum.Itdoes-not dissolve appreciably in a 5 per cent. solution of potassiumhydroxide.Conversion of Trianhy dro- into T e tra-anhydro- t e t rak: isdiph eny l-silicanediol.The conversion of trianhydro- into tetra-anhydro-tetrakis-diphenylsilicanediol is easily accomplished. When an alcoholicsolution of the former is warmed with a trace of sodium hydroxide,a crystalli_ne precipitate is soon formed, and in a short time mostof the dissolved trianhydro-derivative has undergone the desiredtransformation ; the air-dried precipitate melts a t 200°, and whenrecrystallised once from ethyl acetate i t melts sharply a t 201°, andconsists of pure tetra-anhydrotetrakisdiphenylsilicanediol.When an alcoholic solution of trianhydrotetrakisdiphenylsilicane-diol is warmed with a few drops of concentrated hydrochloric acid,a precipitate forms much more slowly than when sodium hydroxideis used, and the tetra-anhy dr ot etra kisdip henylsilicanediol which isdeposit.ed is impure; the air-dried precipitate melts from about187O t o about 200°, but when recrystallised from ethyl acetate itgives a deposit of pure tetra-anhydrotetrakisdiphenylsilicanediol.The results of t h s action of sodium hydroxide and of hydrogenchloride vary, however, with the conditions of the experiment; if,for example, an alcoholic solution of trianhydrotetrakisdiphenyl-silicanediol is treated with a drop of concentrated hydrochloricacid and left at the ordinary temperature for a day o r two, itdeposits crystals melting a t about 185*, which probably consistof a mixture of trianhydrotris- and tetra-anhydrotetrakis-diphenyl-silicanediol (Kipping, Zoc.cit., p. 2140). This is due, no doubt, tothe fact that the acid causes both hydrolysis and condensation tooccur, just as it was shown to do in the case of dianhydrotrisdi-benzylsilicanediol (Robison and Kipping, Zoc. c i t . ) .Trianhydrotetrakisdiphenylsilicanediol may also be convertedinto the tetra-anhydro-derivative with the aid of heat. The open-chain compound seems to begin to decompose at about 1 3 5 O , butthe action does not become rapid until the temperature has risen toabout 180-190O; a t this stage the escape of bubbles of steam maybe clearly observed, but when the liquid maw has been' heated a t190-200° during about fifteen minutes, decomposition seems tocease.The cooled product is a vitreous solid, which crystallises onthe addition of a little ether, and from which cold alcohol extractsonly a small proportion of soluble matter; the residue meltsindefinitely a t about 185O, but when repeatedly recrystallised froORGANIC DERIVATIVES OF SILICON. PART XXI. 499hot acetone i t gives pure tetra-anhydrotetrakisdiphenylsilicanediol(m. p. 200O).Although it was thus proved that the open-chain is convertedinto the closed-chain compound by the action of heat, the reactiondoes not take place quantitatively ; the crude tetra-anhydro-deriv-ative is mixed with a certain proportion of some substance whichlowers it0 melting point, and which is only removed with dSculty.This other product is in all probability trianhydrotrisdiphenyl-silicanediol, because the mother liquors from the pure tetra-anhydro-derivative yield a residue which separates from cold ethylacetate in large, transparent rhombs, and these crystals have theproperties of the crystalline mixture of trianhydrotris- and tetra-anhydrotetrakis-diphenylsilicanediol previously described (Zoc.cit.,p. 2140). The formation of trianhydrotrisdiphenylsilicanediolcould be easily accounted for on the assumption that some of thetrianhydrotetralusdiphenylsilicanediol is partly hydrolysed todianhydrotriscliphenylsilicanediol and diphenylsilicanediol by thesteam which is liberated durixg the formation of the tetra-anhydro-compound.I'e traanh ydro t e trakisdiph erylsilicanediol.The well-defined, almost rectangular plates in which thiscompound is sometimes deposited from ethyl acetate solution a tthe ordinary temperature were measured.System : anorthic.Sub-class : holohedral.u.: 6 : c =0.5614 : 1 : 0.5770.a=83O56/; P = 103O52/; y=96O44/.Forms observed: tx{lOO}, b{010}, c{OOl}, m{110}, q{O11},${ l5Of.Table of A ngles.Angle. No. Limits. Mean. Calculated010: 110 3 56'23'-57' 16' 56'56' 57'6g110: 100 4 27'13'-27'48' 27'30' 27"25'010:001 5 94'26'-94'53' 94'37' -110: 001 6 80'39'-8 1'57' 81'8' 50'58'120:OOl 12 7 6' 2 9'-7 7 ' 1 8' 76'544' 76'57a - 64' 5' 010: 011 - -110:011 1 65'27' 65'27' -12O:Oll 1 98'31' 98'31' 99'3'100: 120 10 50'32'-50'46' 50'38' -120: 010 6 44"46'-44'59' 44'504' -100:001 11 7 6 ' 3 1'-7 6' 47 ' 76"41' -190: 011 1 74'38' 74'35' 75O9'The crystals are flat, almost square plates, the pinacoid a{100}being predominant, and the forms b{010} and c{OOl} beingdeveloped along the edges of the plates.VOL.cv. L 500 ORGANIC DERIVATIVES OF SILICON. PART XXI.As diphenylsilicone could not be obtained by heating diphenyl-silicanediol, the action of heat on tetra-anhydrotetrakisdiphenyl-silicanediol was studied; it seemed possible that a t a high tempera-ture the last named compound might decompose, yielding diphenyl-silicone, just as paraformaldehyde, for example, gives formaldehyde.I n order to avoid atmospheric oxidation, a small quantity of thepure substance was heated in a bent tube in an atmosphere ofcoal gas; it distilled at a very high temperature, giving a colourlessdistillate, and no appreciable charring took place. The distillatesolidified immediately to a hard mass, which was free from anyappreciable quantity of oily matter, and melted from about180-185O. On crystallisation from ethyl acetate the product wasdeposited in the well-defined, rectangular plates described above,but these melted indefinitely from about 185-190°, and it wasonly after three or four further recrystallisations that pure tetra-anhydro t&r akisdip henylsilicanediol (m. p. 200-20 1 O) was ob-tained. I n spite of this fact, the original product of distillationseemed to consist almost entirely of the one substance. The smallproportion of impurity which was certainly contained in it, andwhich was so difficult to remove, was not definitely identified, butin all probability it consisted of trianhydrotrisdiphenylsilicanediol.The most soluble deposit obtained from the last ethyl acetatemother liquors melted a t about 180--185O, and contained a verysmall proportion of crystals, which did not become opaque whenheated, but melted at 186-188O; this substance was doubtlesstrianhydrotrisdiphenylsilicanediol. It is obvious from these resultsthat diphenylsilicone is not formed in appreciable quantities underthe above conditions.The authors gratefully acknowledge the financial assistance forwhich they are indebted to the Government Grant Committee ofthe Royal Society.UNIVERSITY COLLEGE,NOTTING 11 AM