40 ROBISON AND KIPPING :VI.-Organic Derivatives of Silicon. Part X X ,Some Condensation Products of Dibenx y bilicanediol.IN the last four papers on organic derivatives of silicon (T., 1912,101, 2108, 2125, 2142, 2156) it was shown that certain disub-stituted silicanediols of the type SiR2,(0H),, in which a t least oneof the radicles is aromatic, are capable of existence a t the ordinarytemperature, and, therefore, are mole stable than the correspondingderivatives of methane.Such compounds, however, readily undergo change when theyare heated alone, or when they are treated in solution with variousreagents, such as alkali hydroxides, ammonia, or mineral acids.The result, as a rule, is the formation of a mixture of open- andclosed-chain condensation products ; from such mixtures representa-tives of the following types have already been isolated:By ROBERT ROBISON and FREDERIC STANLEY KIPPING.HO*SiR,*O*SiR,-OH8iR,,<0.8iR 0- Si K 2>oHitherto very little attention has been paid to the conditionsunder which each of these different types of condensation productis formed; the mixtures of these compounds were obtained more orless by chance during the preparation or purification of the diols.Now the importance of gaining further information respectingthese condensation products need hardly be emphasised ; the studyof the reactions by which they are formed cannot fail t o throw somelight on the processes, doubtless of an analogous character, whichlead to the production of the complex mineral silicates.For this reason we have continued our investigations alongvarious lines, and Pave more particularly directed our attention toan examination of (a) the conditions under which the diols undergocondensation, ( b ) the conditions under which the open-chain areconverted into closed-chain compounds, and ( c ) the comparativereadiness with which the closed-chain compounds containing two,three, or four atoms of silicon respectively are formed.The resultsobtained from the study of dibenzylsilicanediol and its condensationproducts are described in this paper.Dibenzylsilicanedio: is very readily acted on by various reagents(p. 41), giving, as a rule, a mixture of anhydrobisdibenzylsilicane-diol and trianhydrotrisdibenzylsilicanediol.The former compounddoes not seem t o yield any further condensation product by thedirect elimination of the elements of water ; so far, we have not beeORGANIC' DERIVATIVES OF SILICON. PART XX. 41able to prepare from it either dianhydrobisdibenzylsilicanediol,0 Si(CH,Ph)<o> Si( CH,Ph),, or the analogue of tetra-anhydrotetra-kisdiphenylsilicanediol. In alcoholic solution, however, in presenceof hydrogen chloride, anhydrobisdibenzylsilicanediol is transformedinto trianhydrotrisdibenzylsilicanediol ; this change is doubtlessdependent on the hydrolysis of some of the anhydrobisdibenzyl-silicanediol to dibenzylsilicanediol, but whether the trianhydro-coinpound is formed by the subsequent condensation of the dibenzyl-silicanediol Q ith the unchanged anhydrobisdibenzylsilicanediol, ' orby the direct condensation of three molecules of the simple diol, isnot known.Dianhydrotrisdibenzylsilicanediol,HO*Si(CH2Ph),*O*Si(CH,Ph)2*O*Si(CH2Ph)2*OX3,the analogue of dianhydrotrisdiphenylsilicanediol, may be obtainedby carefully hydrolysing trianhydrotrisdibenzylsilicanediol withpotassium hydroxide or hydrogen chloride in acetone solution ; itis readily converted into trianhydrotrisdibenzylsilicanediol by tracesof hydrogen chloride in alcoholic solution.These facts show thatthe hydrolysis and formation of these closed oxy-silicon chains areeasily reversible reactions under the given conditions; it may alsobe concluded that the closed chain composed of three silicon andthree oxygen atoms alternately linked together, is more easilyformed than those containing either two or four atoms of both ofthese elements.EXPERIMENTAL.Dib ei.mylsilicamedio1.Dibenzylsilicanediol is converted into anhydrobisdibenzylsilicane-diol when it is heated at about l l O o (Zoc.cit., p. 2148), treated withhydrogen chloride in alcoholic solution (Zoc. cit., p. 2150), ortreated with acetyl chloride; in the last two cases trianhydrotris-dibenzylsilicanediol is also formed and may even be the principalproduct, but no other crystalline substance has been isolated fromthe oily matter, which is usually produced from the diol in theseand other experiments.Dibenzylsilicsnediol is also condensed by alkalis ; when a solutidnof the diol in a 6 per cent.solution of sodium hydroxide is kepta t the ordinary temperature exposed to the air, in the course of aday o r two it deposits an oil, soinetimes a solid, which is insolublein a dilute aqueous solution of sodium hydroxide, but completelysoluble in cold ethyl alcohol; this product is t.herefore probably amixture of anhydrobisdibenzylsilicanediol and dianhydrctrisdi-benzylsilicanediol (p. 43). I f , however, the alkaline solution i42 ROBISON AND KIPPING :heated a t about looo during some hours, the precipitate is onlypartly soluble in cold alcohol, and contains a considerable propor-tion of trianhydrotrisdibenzylsilicanediol.The separation of these condensation products from a cold alkalinesolution of the diol certainly takes place more quickly in thepresence than in the absence of air; it is improbable, however,that the atmospheric carbon dioxide has any direct action exceptthat of diminishing the concentration of the sodium hydroxide.Apparently the solution contains the sodium derivative, orderivatives, of the diol in equilibrium with a very small quantityof one or more of the condensation products of the diol:Si(CH,Ph),(ONn), + H29 ++ Mi(CIT2Ph),(ONa)*OH + NaOHH,O + 2Si(CH2Ph),(ONa)*OH ++EO*Si(CH,Yh),*O*Si(CH2Ph)2*0 H + 2Na0 €1,and the conversion of the alkali into carbonate merely disturbs theequilibrium.The behaviour of all the other diols towards dilute solutions ofalkali hydroxides appears to be similar to that of the dibenzylcompound, but except in the case of diphenylsilicanediol, the con-densation products have not been isolated; it would also seem thatdibenzylstannanediol undergoes condensation under similar con-ditions (Smith and Kipping, T., 1913, 103, 2034).Dibenzylsilicanediol also undergoes condensation when its alcoholicsolution is treated with a drop or two of piperidine; trianhydro-trisdibenzylsilicanediol is formed, together with an oil, the com-ponents of which have not keen' isolated.Anhtydrobisdib enaylsilkcanediol.Several further attempts have been made t o convert this com-pound into dianhydrobisdibenzylsilicanediol, o<s' s, (C (cH H2Ph 2Ph), '2>0,or tetra-anhydrotetrakisdibenzylsilicanediol, but without success.Acetyl chloride, as already stated (Zoc. cit., p.2154), seems to havelittle or no action on the compound; in a recent experiment asolution of anhydrobisdibenzylsilicanediol in excess of the acidchloride was kept during some months and then evaporated, but theoily residue was readily and completely soluble in ethyl alcohol, andappeared to be free from higher condensation products.Thealcoholic solution subsequently deposited crystals of trianhydro-trisdibenzylsilicanediol, but this compound liad probably beenformed by the action of traces of hydrogen chloride on anhydrobis-dibenzylsilicanediol (see below).The behaviour of anhydrobisdibenzylsilicanediol towards alkaliORGANIC DERIVATIVES OF SILICON. PART XX. 43was also examined in alcoholic solution, but crystalline compoundscould not be isolated from the oily product which was formed inthe course of some days. The action of piperidine in alcoholicsolution was similar to that of potassium hydroxide, and led tothe formation of an oily condensation product.Anhydrobisdibenzylsilicanediol was also heated with phosphoricoxide ; the oily product was completely soluble in ethyl alcohol,and apparently did not contain trianhydrotrisdibenzylsilicanediol inappreciable quantity.When, however, an alcoholic solution ofanhydrobisdibenzylsilicanediol, to which a few drops of concentratedhydrochloric acid had been added, was kept at the ordinarytemperature, cryst.als of trianhydrotrisdibenzylsilicanediol weredeposited in the course of twenty-four hours. It must therefore beconcluded that the anhydro-derivative is hydrolysed by the hydro-chloric acid, giving dibenzylsilicanediol, which then undergoes con-densation under the influence of the same reagent.Dianhydro t risdi b enzy lsilicanedio 2,HO*Si(CH,Ph),*O*Si( CR,Ph),-O*Si(CH,Ph),*OH.Although this compound is very probably present in the oilswhich are formed by the action of dilute alkali hydroxides ondibenzylsilicanediol and on anhydrobisdibenzylsilicanediol, it hasnot yet been isolated from these products.Attempts were thereforemade to prepare it by the hydrolysis of trianhydrotrisdibenzyl-silicanediol in accordance with the equation :The trianhydro-derivative, although practically insoluble inalcohol, slowly dissolves in an alcoholic solution of potassiumhydroxide a t the ordinary temperature, owing t o the occurrence ofhydrolysis ; from freshIy prepared solutions water precipitates asolid, and the filtered solution gives with acetic acid a precipitate ofdibenzylsilicaiiedioL An examination of the solid obtained in thisway from the alkaline solution showed that it consisted of anhydro-bisdibenzylsilicanediol, or of a compound melting a t 8 2 O , or of amixture of these two substances, and that its nature varied withthe conditions of the experiment, owing to the occurrence of pro-gressive hydrolysis. I n order t o find a suitable method for thepreparation of the compound melting a t 82O, which was the desireddianhydro-derivative, various experiments were made under differentconditions; the results showed that with the aid of acetone, i44 ROBISON AND KIPPING :which trianhydrotrisdibenzylsilicanediol is readily soluble, theprimary product of hydrolysis is readily obtainable as follows :A solution of trianhydrotrisdibenzylsilicanediol in cold acetoneis treated with a 3 per cent.solution of potassium hydroxide(2 mols.), and thirty seconds afterwards a slight excess of diluteacetic acid is added. The solution is then further diluted withwater, and the Qily precipitate is separated by filtration or extractedwith ether. The ethereal solation of the oil is mixed with lightpetroleum and left t o evaporate; the crystals which separate arethen recrystallised several times from a mixture of the two solventsjust mentioned.Dianhydrotrisdib enzylsilicanediol is thus obtained in large,transparent crystals, melting at 8 2 O ; the samples for analysis weredried over sulphuric acid:0.1899 gave 0,5019 CO, and 0.1088 H,O.C = 72.2 ; H =: 6.4.0.1733 ,, 0.4600 CO, ,, 0.0993 HZO. C=72*4; H=6*4.C42H4404Si3 requires C = 72'3 ; H = 6.3 per cent.Molecular-weight determinations were made by the cryoscopic0.253, in 13.55 grams of benzene, gave At - 0'12O.0.446, ,, 13'55 ,, >, ,, At -0.21'. M.W. =771.These results agree only moderately with the calculated molecularweight (C,,H,,O,Si, requires M.W. 696), and indicate a consider-able degree of association in benzene solution ; in this respect theycorrespond closely with the values obtained under similar conditionsin the case of other condensation products of this character, suchas anhydrobisdiphenylsilicanediol and dianhydrotrisdiphenyl-silicanediol (Kipping, Zoc.cit., p. 2132, 2134); they may thereforebe regarded as satisfactory.Dianhydrotrisdibenzylsilicanediol is readily soluble in ether oralcohol, and in all the other common organic solvents with theexception of cold light petroleum, in which it is only sparinglysoluble; it is practically insoluble in water, and also in a colddilute solution of potassium hydroxide.It is obvious from the above-descriked method of preparation ofdianhydrotrisdibenzylsilicanediol that the rupture of the closedchain contained in the trianhydro-derivative takes place veryrapidly under the influence of the potassium hydroxide, whereas thefurther hydrolysis, which results in the formation of anhydrobis-dibenzylsilicanediol and dibenzylsilicanediol takes place more slowly.Although the above-mentioned quantity of the alkali was generallyused, the proportion of this substance seemed to be of much lessimportance than the time during which hydrolysis was allowedmethod in benzene solution :M.W.= 761ORGANIC DERIVATIVES OF SILICON. PART XX. 45to proceed. If, for example, more than a few minutes elapsebetween the addition of the alkali and that of the acetic acid, theproduct always contains a considerable quantity of dibenzylsilicane-diol, even when only two molecular proportions of potassiumhydroxide are present.Hydrolysis of Trianhydrotrisdibenz.?llsilicanediol with HydrogecnC it Eorid e.It has already been shown that dibenzylsilicanediol is slowlyconverted into trianhydrotrisdibenzylsilicanediol when it is treatedwith concentrated hydrochloric acid in alcoholic solution (loc. cit.,p.2150), and that under similar conditions anhydrobisdibenzyl-silicanediol is aiso transformed into the same condensation product.I n the latter case it is clear t,hat hydrolysis of the anhydro-derivative must precede condensatibn ; the reaction brought aboutby the hydrochloric acid is therefore a reversible one, and the pro-duction of trianhydrotrisdibenzylsilicanediol in almost theoreticalquantities is determked merely by the insolubility of the latter inalcohol. Experiments were therefore made to ascertain whetherunder suitable conditions trianhydrotrisdibenzylsilicanediol couldbe hydrolysed with hydrochloric acid; it was thus found that thetrianhydro-derivative could be converted into dianhydrotrisdibenzyl-silicanediol in the following manner.Trianhydrotrisdibenzylsilicanediol is dissolved in acetone andone drop of concentrated hydrochloric acid is added; after abouthalf an hour's time the solution is diluted with water and stirredvigorously.The precipitated solid is separated by filtration, andextracted with cold alcohol, which leaves a residue of unchangedtrianhydrotrisdibenzylsilicanediol ; the alcoholic solution is thendiluted with water and the oily precipitate extracted with ether.The ethereal solution contains dianhydrotrisdibenzylsilicanediol,which is purified by recrystallisation from a mixture of ether andlight petroleum. I n this way only about 15 per cent.of thetrianhydro-derivative is transforzed into the dianhydro-compound,but by submitting the unchanged substance to the same treatmentagain, further quantities of the dianhydro-compound may beobtained.It might be concluded therefore that in acetone solution thehydrolysis of trianhydrotrisdibonzylsilicanediol by hydrogen chlorideis a reversible reaction 46 ROBISON AXD KIPPING :and that under the conditions described above, equilibrium isattained when the proportion of dianhydrotris- to trianhydrotris-dibenzylsilicanediol is about 1 : 5 or 1 : 6. Probably the aboveequation expresses only one of several reversihle reactions whichoccur, others being the following :2Si(C M,Yh),(O H )2 t+ HO*Si(C H2Ph),*0.Si(r;’H,Ph)2*OH + H,O.Conversion of Di- into Tri-anhydrotrisdib enzylsilicanediol.If the formation of dianhydrotrisdibenzylsilicanediol from thetrianhydro-derivntive in the manner just described is really areversible reaction, then in alcoholic solution the change expressedby the above equation should proceed almost completely from rightto left, owing to the insolubility of the trianhydro-compound inalcohol.When an alcoholic solution of dianhydrotrisdibenzylsilicanediol,to which a drop of concentrated hydrochloric acid has been added,is kept a t the ordinary temperature, it soon deposits a crystallineprecipitate, melting a t 98O, which consists of pure trianhydrotris-dibenzylsilicanediol.A quantitative experiment, made in the following manner, showedthat the change is practically complete.A known quantity ofdianhydrotrisdibenzylsilicanediol, contained in a weighing bottle,was dissolved in alcohol, t o which one drop of concentrated hydro-chloric acid had been added; after the solution had been kept forabout twelve hours, during which time it deposited crystals oftrianhydrotrisdibenzylsilicanediol, it was evaporated in a desiccatorover sulphuric acid and potassium hydroxide, and the residue wasdried a t about 90° until constant in weight. The loss amountedto 2.48 per cent., against a theoretical loss of 2.58 per cent.Dianhydrotrisdibenzylsilicanediol also undergoes condensation inalcoholic solution in presence of a trace of sodium hydroxide, evenmore readily than in presence of hydrogen chloride; within anhour, crystals of trianhydrotrisdibenzylsilicanediol are depositedfrom the solution.Acetic anhydride seems to nave little action on dianhydrotris-dibenzylsilicanediol ; a small quantity of the compound was heatedwith this reagent for a short.time, but was afterwards recoveredunchanged.I n one experi-ment the residue obtained by evaporating an acetyl chloride solutionof the dia nhydro-compound a t the ordinary temperature was treatedThis conclusion was fully confirmed by experiment.The action of acetyl chlorida was also studiedORGANIC DERIVATIVES OF SILICON. PART XX. 47with alcohol ; i t yielded crystals of trianhydrotrisdiberizylsilicane-diol. I n another experiment, a weighed quantity of the substancewas gently warmed with a small amount of acetyl chloride untilthe latter had evaporated, apd was then heated t o 1 1 0 O ; no loss inweight, however, could be detected. The same sample was againtreated with acetyl chloride, and after the latter had beenevaporated, the residue was heated to about 125O in the courseof forty-five minutes.The total loss in weight was 2.9 per cent.,the theoretical loss for one molecule of water being 2.6 per cent.When the residue was warmed with alcohol, it gave a sparinglysoluble powder, which separated from a mixture of ether and lightpetroleum in crystals melting a t 98O. These results prove thatdianhydro- may be converted into trianhydro-trisdibenzylsilicanediolby the treatment described, but they do not show whether it isthe- acetyl chloride itself which brings about condensation, orthe traces of hydrogen chloride which are doubtless formed in thecourse of the experiment.Effect of Hent o n Dirtnh,ydrotrisdibenzylsilica~ed~ol.Although dianhydrotrisdibenzylsilicanediol is so readily convertedinto the closed-chain compound by acids or alkalis, this changecannot be satisfactorily accomplished by merely heating thedianhydro-compound (in the air).Quantitative experiments showedthat a t 130° only a very slow loss in weight took place; a t highertemperatures, the loss was more rapid, but the residue had a strongodour of benzaldehyde, probably owing to atmospheric oxidation.In one case, during one hour a t 175O, 0.3125 gram of substancelost 5.4 per cent., but there was no indication of a constant weighthaving been attained; as the theoretical loss for one molecule ofwater is only 2.58 per cent., more than one-half of the observedloss was doubtless due to the volatilisation of the benzaldehydeformed by atmospheric oxidation.The residue yielded less than0.1 gram of trianhydrotrisdibcnzylsilicacediol, together with an oilfrom which crystah of dianhydrotrisdi benzylsilicanediol (m. p. 82O)were isolated.Trianh ydro trisLil: b enzy lsilican ediol.Crystals of this compound deposited from a mixture of chloroformand light petroleum have been examined for us by Mr. VernonStott under the direction of Mr. A. Hutchinson, M.A., of themineralogical laboratory, Cambridge ; we are indebted to thesegentlemen for the following report 48 ORGANIC DERIVATIVES O F SILICON.PART XS.System : oblique. Sub-class : holohedral.Forms developed: a{100}, c{OOl), m{110}, s{101}, T{TOl).Table of angles :a : b : ~ = 2 ' 7 9 7 : 1 : 1.643 ; 6 =94'38'.Angle.100 : 101101 : 001100 : $01001 : lo1To1 : loo100 : 110101 : 110110 : i i o101 : 110- 031 : 110Number ofmeasurements. Limits.9 59"dO'-6O01 1'9 25 14 -26 139 85 2 -85 4011 27 2 -28 49 66 31 -67 423 70 0 -70 3633 80 0 -80 3316 39 11 -39 4342 88 13 -88 5447 96 59 -97 58Mean.59"43'25 3885 2227 3866 5170 1 480 1539 2888 3097 33Calculated.59'29'25 5327 4770 1680 388 2i97 38---The crystals are prismatic in habit, being elongated in a directionThe ?ii faces are in general muchAt our request, Dr. I. M. I-Teilbroa has very kindly photographedperpendicular t o t.he diad axis.larger than the a faces.Oscillatioit frequcncies.32 31 36 38 400042 44 46Pull curve : Dibenzylsiticanediol i?z alcohol, with or wilhoztt alknli.Dot curve : Anhydrobisdibenzylsilieancdiol in alcohol.Dash curve : Dianhydrotrisdibemylsilieanediol in nlcoitol.Dash-dot curve : l'riunhydrotrisdibenzylsilica~~ediol in chloroforin.the absorption spectra of dibenzyl.silicanedio1 and of the variouscondensation products of this diol, which, so far, have been preTHE OPTICAL ROTATORY POWER, ETC. 4 9pared. From the curves, which are given in the accompanyingdiagram, i t will be seen that the spectra of all the four compoundsare very similar, and show one fairly broad band, which covers theregion of the first four or five bands shown by benzene (Baly andCollie, T., 1905, 87, 1332). In the case of those compounds whichcontain four or six benzyl groups, this band is found at lowermolecular concentrations than in the case of dibenzylsilicanediolitself. It will also be seen that the addition of alkali t o thealcoholic solution of the last-named compound brings about noalteration in the absorption spectrum.The authors are indebted to the Government Grant Committee ofthe Royal Society for a grant in aid of this investigation.UNIVERSITY COLLEGE,NOTTINQH AM