2108 KIPPING: ORGANIC DERlVATZVES OF SILICON. PART XVI.CCXXIII.-O~ganic Dwivatives of Silicon. Part X VLThe Preparation and Pmperties of L)@he&-s ilicanedio t .By FREDERIC STANLEY KIPPING.THE primary object of this research was to ascertain the relation’between the two dibenzyl compounds of the composition C,,H,,Si02,which had been obtained by Robison and Kipping (Trans., 1908,93, 439) by the hydrolysis of dichlorodibenzylsilicane (dibenzyl-silicon dichloride). For this purpose a comparative study of theproducts of hydrolysis of dichlorodiphenylsilicane seemed to bedesirable.A t that time, cjne such product, namely, diphenylsilicanediol( ( I dipbenylsilicol ”), had already been described by Dilthey andEduardoff (Ber., 1904, 37, 1139), who obtained it by decomposinghighly impure dicblorodiphenylsilicane with water ; they describedit as crystallising from benzene in needles, which melted a t138-139O. Quite recently the results of an investigation of“ dibenzylsilicol ” and (‘ diphenylsilicol ” have been published byMartin (Ber., 1912, 45, 403).*I n the experiments described in this paper, dichlorodiphenyl-silicane waa prepared by the interaction of silicon tetrachloride andmagnesium phenyl bromide, and the pure compound was decom-posed with water, ammonium hydroxide solution, or a dilutesolution of potassium hydroxide.A t first consideration it mightnot seem to be a very difficult task to hydrolyse the dichloro-* The examination of the dibenzyl compounds, referred to above, having beenleft unfinished, owing to the departure from Nottingham of Dr.Robison, I con-tinued the work in conjunction with Dr. Martin, with whom I also resumedthe study of diphenylsilicanediol. After many months of strenuous application,Dr. Martin came to the conclusion that he had obtained two isomeric diphenyl-silicols,” and had also devised methods for their conversion one into the other, butbefore the research was really finished he was obliged to abandon it, in order to takeup an appointment elsewhere. On his departure, Dr. Martin handed to me arecord of his work, which, a short time aftemard.s, I began to consider, with a viewto the publication of a joint paper. As some of the statements in this recordseemed to me to require verification, I repeated some of Dr.Martin’s more importantexperiments ; having failed to confirm them, I informed him of the results of myobservations and that I thought it necessary t o reviae the whole of the work beforesending the paper to be published. After some discussion, it was finally left to himto decide whether or not he should communicate his own results under his ownuame ; his decision led to the publication of the paper referred to above, in which,unfortunately, as will be shown later, he gave a very erroneous accouut of diphenyl-silicanediolKIPPING : ORGANIC DEHIVATIVES OF SILICON. PART XVI. 2109derivative by these processes, and to isolate the compound whichhad been formed, and yet the task proved to be ail exceptionallytroublesome one.Although it was easy enough to obtain a crystalline product,different preparations varied greatly in their behaviour, and undercertain conditioiis were 50 readily changed that the results weremost perplexing ; some preparations, for example, when warmedwith solvents, were converted into viscid, glue-like products ; someof these glues were readily soluble in alcohol, others almostinsoluble.A further difficulty was due to the fact that the pre-parations did not melt, but decomposed with effervescence, whenthey were heated, and although they liquefied in this prmess, thetemperature a t which decomposition occurred varied very greatlywith different preparations, even after they had been recrystal-lised.As illustrations of the behaviour here outlined, the followingsummary of the results of hydrolysing the dichloride under differentconditions may be given, accompanied by the statement that purediphenylsilicanediol usually decomposes a t about 128-13Z0, and isreadily and completely soluble in a 5 per cent.aqueous solution ofpotassium hydroxide.The decomposition of dichlorodiphenylsilicane with cold waterresults in the formation of a crystalline solid, which contains alarge proportion of the dihydroxy-compound ; after having beenwashed with cold chloroform, which extracts an oily mixture of a tleast three substances, this product begins to liquefy and decomposea t about 145-150O; but liquefaction is not complete until thetemperature has risen to about 165O. When the washed prepara-tion is heated with solvents in order to crystallise it, it may becompletely converted into an oil, which is readily soluble in alcohol,or it may give crystalline depcrsits, which vary considerably indecomposition point; all such deposits aze impure, as they do notdissolve completely in a 5 per cent.aqueous solution of potassiumhydroxide, and the impurity is not eliminated by recrystallisationfrom various anhydrous solvents.The decomposition of dichlorodiphenylsilicane with excess of aconcentrated solution of ammonium hydroxide results in the forma-tion of a viscid oil, which is practically free from diphenylsilicanediol, and is a mixture of a t least three compounds.The hydrolysis of the dichloride with excess of a 5 per cent.aqueous solution of potassium hydroxide results in the productionof a clear solution of the potassium derivative, SiPh,(OK), orSiPh(OH)*OK ; from this solution, diphenylsilicanediol is precipi-tated on the addition of acids or certain salts, but the properties7 A 2110 KIPPING : OKGIANIC DERIVATIVES OF HILICON.PART XVI.of the precipitate vary very considerably with the nature and thequantity of the precipitant.When the alkaline solution is just neutralised with acetic acid,the air-dried precipitate almost invariably decomposes, and liquefiescompletely a t about 105-llOo, and when heated with solvents,such as ether or ethyl acetate, it may be completely converted intoa glue-like substance, which is practically insoluble in alcohol. Itmay be recrystallised from various solvents in the cold, and thusobtained in colourless anhydrous needles, decomposing a t about115O, but the decomposition point of such preparations often risesto about 140° when the specimens are merely kept for a short timeat the ordinary temperature. When ammonium chloride, or aconsiderable excess G f hydrochloric acid, is used to precipitate thediphenylsilicanediol from its alkaline solution, the air-dried precipi-tate liquefies with effervescence a t about 145-155O ; such precipi-tates may be recrystallised from hot solvents without their under-going much observable decomposition, but even then they areinvariably heterogeneous, and leave an insoluble residue when theyare treated with a 5 per cent.aqueous solution of potassiumhydroxide.Observations such as those summarised above, the undoubtedexistence of the two dibenzyl compounds of the compositionC,,H,,Si02 described by Robison and Kipping (Zoc.cit.), and thestatement in Martin's record that he had isolated two isomeric formsof diphenylsilicanediol, decomposing a t 140° and 1 60° respectively,all seemed to point in the one direction, and many attempts weremade to isolate the two or more compounds, which were presumablycontained in the above preparations.In the course of these attempts it was found that the diphenyl-silicanediol, precipitated from its alkaline solution with acetic acid,could be obtained in an apparently pure condition in colourlessprisms, which were completely soluble in a 5 per cent.potassiumhydroxide solution, and decomposed with effervescence at115-118O; on the other hand, the compound, precipitated withexcess of hydrochloric acid, was also obtained in needles, which,in spite of the fact that they were not quite pure, and were notcompletely soluble in potassium hydroxide solution, did not decom-pose and liquefy until about 140-145O.As various specimens of the preparations, decomposing a t115-1 1 8 O , gave on analysis results agreeing well with thoserequired for a diphenylsilicanediol, it had to be assumed provision-ally that they represented an isomeride or a polymeride of thecompound, SiP%(OH),, described by Dilthey and Eduardoff asmelting a t 138-139O, and also of that described by Martin aKIPPINa : ORGANIC DERIVATIVES OF SILICON, PART XVI.2111decomposing a t about 160O. It was further observed that thepreparations decomposing a t 1 1 5 - 1 1 8 O were transformed into thosehaving the higher decomposition point when they were recrystal-lised from cold solvents in the presence of a mere trace of hydro-chloric acid; this fact, of course, was compatible with the view thatthe acid had brought about some isomeric or polymeric change.There was, however, apart from all theoretical considerations,one experimental difficulty which, obviously, had to be overcomebefore a definite conclusion could be reached, namely, the removalfrom the preparations decomposing a t temperatures above 140°of the small proportion of impurity, which was insoluble in potass-ium hydroxide solution.It certainly did not Seem very probablethat these recognisably impure preparations could be an impureform of the apparently pure substance decomposing a t 115--118O,but it was a possibility which had to be considered, and whichcould only be decided by eliminating the impurity. For thispurpose the preparations in question were systematically fraction-ally crystallised from various anhydrous solvents and variousmixtures of such solvents, but the results were most unsatisfactory;although the deposits were well crystallised, and, as a rule, appearedhomogeneous, they invariably retained impurity, as shown by theirbehaviour towards potassium hydroxide solution ; moreover, thedecomposition points of the various fractions were most irregular,and ranged from about 140° to 1 6 5 O .It wits ultimately discovered that by the use of aqueous acetoneand chloroform alternately, in the manner described later (p.2121),the impurities in all such fractions could be removed, and that thepure diphenylsilicanediol which was thus obtained (usually)liquefied completely a t 128-132O.By the same method of purification, diphenylsilicanediol was alsoisolated from those apparently pure preparations which liquefiedcompletely a t about 1 1 8 O .The impurity which was thus removed from any preparation wasrelatively small in quantity, and consisted as a rule of an oilymixture of several compounds; as this mixture was readily solublein ether, ethyl acetate, and other anhydrous solvents, and yet wasnot removed when the impure diphenylsilicanediol was recrystal-lised repeatedly from these solvents, it would seem that the impuri-ties of which the mixture was composed had been adsorbed by thecrystals of the diol.Whatever may be the cause of the great difficulty of separatingdiphenylsilicanediol from relatively small quantities of certainsoluble impurities, innumerable observations have shown that allthose specimens of the diol which contain an appreciable propor2112 KIPPING: ORGANIC DERIVATIVES OF SILICON.PART XVI.tion of the substances insoluble in a 5 per cent. aqueous solution ofpotassium hydroxide, invariably decompose a t a much highertemperature than 128--132O, sometimes not until about 1 6 5 O ; purespecimens of diphenylsilicanediol, on the other hand, as statedabove, usually decompose and liquefy completely below 1 3 2 O ,although very occasionally the decomposition does not occur untilthe temperature has been raised to about 1 6 0 O .This irregular and unusual behaviour, in the author’s opinion,is not due to the existence of an isomeride of the compound(C,H,),Si(OH),, to which, indeed, it would be difficult to assign astructural formula, but is to be explained as follows: The crystal-line form of diphenylsilicanediol, which separates from solvents a ttemperatures within the limits of those employed (about 10-60°),decomposes and liquefies completely a t about 128-132O; but whennear its decomposition point, this form, A , is in a metastable condi-tion, and may change into another crystalline modification, B,which only decomposes a t much higher temperatures, namely, a tabout 1 6 0 O .The pure compound, as a rule, decomposes beforethe change in crystalline form occurs; in presence, however, of asufficiently large proportion of those substances which are insolublein a 5 per cent. aqueous solution of potassium hydroxide, itspartial or complete transformation into the crystalline form ofhigher decomposition point invariably takes place, probablybecause one or more of these imporities is isomorphous with thisform, B.Whether this suggested explanation of the observations is truein all particulars or not, there does not seem to be any evidenceof the existence of the isomeric diphenylsilicanediols described byMartin, and judging from the methods of preparation and decom-posing points of his supposed isomerides, both must be regardedas impure specimens of the diol.In justice to Martin, however, itshould, perhaps, be pointed out, that not only was he misled by apresumed analogy between dibenzylsilicanediol and diphenylsilicane-diol, which, in fact, did not exist, but that the difficulties presentedby the investigation of the last-named compound were altogetherexceptional, and might have led into error a much more experiencedchemist.The remlts of further experiments on diphenylsilicanediol, whichare described later, elucidate many of the observations which arerecorded in this paper. For the sake of clearness it may be brieflystated here that diphenylsilicanediol is very readily decomposedby traces of alkalis, and also, but not so readily, by traces of acids,giving compounds which in their turn are further changed by thesesame reagents ; consequently, the product obtained from thKIPPING : ORGANIC DERIVATIVES OF SILICON.PART XVI. 2113dichloride by the various methods indicated above may containvariable quantities of impurities, the nature of which depends onits method of preparation and subsequent treatment ; further, unlessthe product is completely freed from alkali or acid before it ishe,ated with solvents, it may be completely decomposed andconverted into oily or glue-like mixtures.EXPERIMENTAL.Preparation of Dichlorodiphenylsalica?~e, SiP&C12,.The action of different quantities of magnesium phenyl bromideon silicon tetrachloride was studied by Dilthey and Eduardoff(Zoc. cit.), who found that the product always eonsisted of a mixture,even when molecular proportions of the materials were employed ;they described their methods for the preparation of chlorotriphenyl-silicane and dichlorodiphenylsilicane, but they did not isolate thesecompounds ; they hydrolysed their crude reaction mixtures .withicecold water, and then purified the products of hydrolysis.The author's observations confirm those of Dilthey andEduardoff that the above interaction gives a mixture, no matterwhat proportions of magnesium phenyl bromide are used ; never-theless, with a suitable quantity of the Grignard reagent, a fairlysatisfactory yield of diphenyldichlorosilicane can be obtained.The following method was employed : Silicon tetrachloride(170 grams) is cooled to Oo in a, flask, provided with a stirrer, anda carefully prepared ethereal solution of magnesium phenyl bromide(2) mols.) is slowly added in the course of about one and a-halfhours.The contents of the flask are then left a t the ordinarytemperature during about twelve hours; it is advisable to shakethe flask vigorously a t intervals during this time, otherwise themagnesium salt, as it separates, may form a solid cake, whichcannot afterwards be removed. The flask is next heated duringabout three hours on a reflux apparatus; this is a very necessaryoperation, as the interaction is not complete a t the ordinary tem-perature, and, moreover, the magnesium salt is still in a gelatinouscondition, and, a t this stags, cannot be easily separated by filtra-tion.The ethereal solution of the silicon compounds is thenfiltered from the crystalline magnesium salt in absence of moisture(compare Ripping, Trans., 1907, 91, 216), the residue is washedwith pure ether, and the combined filtrate and washings are evapor-ated on the water-bath; towards the end of this operation a furtherquantity of magnesium salt is almost invariabry deposited, and theseparation gradually increases when the oily mixture is heated onthe water-bath during about half an hour. It would seem, there2114 KLPPING : ORGANIC DERIVATIVES OF SILICON. PART XVI.fore, either that the magnesium salt at first remains dissolved inthe ethereal solution, or that the reaction is not complete, even afterthe ethereal solution has been boiled during about three hours.I n consequence of this separation of magnesium salt, the oilyproduct must again be filtered and the residue washed with ether.The clear, brownish-yellow oil is distilled from an ordinary distil-lation flask, under a pressure of 50 mm.After traces of ether andbromobenzene have passed over, the thermometer rises and remainsfor some time in the neighbourhood of 110-120°, and the fractioncollected between these temperatures consists principally of tri-chlorophenylsilicane ; the thermometer then rises rapidly to about190°, and a very large fraction, consisting principally of dichloro-diphenylsilicane, is collected between this temperature and about225O.Another rapid rise then occurs, and the fractions collectedfrom about 260-300° depmit crystals of chlorotriphenylsilicane inthe course of some hours Above 300° little distils, but thereremains in the flask a considerable quantity of a very viscid, blackoil. The various fractions are systematically redistilled, thosecontaining the diphenyl compound preferably under diminishedpressure; in this way, from 170 grams of silicon tetrachloride thereare ultimately obtained about 30 grams of trichlorophenylsilicane,boiling a t 197-202O (atmospheric pressure), about 110 grams ofdichlorodiphenylsilicane, boiling a t 2O0-21Oo/ 45 mm., and about15 grams of crystalline chlorotriphenylsilicane, together with smallerquantities of mixtures of these compounds, and possibly a littlediphenyl.The dichlorodiphenylsilicane boiling a t about 200-210° /45 mm.generally contains a very small proportion of diphenyl, but formost purpose8 this is of no consequence.When further purified,the dichloride is obtained ae a colourless liquid, boiling a t202--204O/45 mm., which fumes slightly in moist air; a sample ofsuch a preparation was analysed, with the following result:0.7414 gave 0'8443 AgC1. C1= 28.14.C,,H,,Cl,Si requires C1= 27.9 per cent.Decomposition. of Dichlorodiphenylsilicane with Water.Dichlorodiphenylsilicane is readily decomposed by cold water,with development of heat. I n preparing the dihydroxy-compoundby this method, the oily dichloride was slowly dropped into a largevolume of water, which was vigorously stirred and cooled duringthe whole operation. The somewhat pasty solid was washed withwater and dried in the air.It was first extracted with warm lightpetroleum t o remove any diphenyl which might be present, and theKIPPING : ORGANIC DERIVATIVES OF SLLICON. PART XVI. 2115with cold chloroform, which dissolved a considerable quantity ofa pale yellow oil. The perfectly colourless, crystalline residue thusobtained usually began to sinter at about 150°, but did not liquefycompletely until about 165O. When treated with various solvents,such as ether, ethyl acetate, acetone, etc., it sometimes gave orange-yellow solutions, which, however, gave colourless deposits when theywere completely evaporated a t the ordinary temperature. Thesecolourless deposits again gave the coloured solutions, and so on.Some preparations were completely or partly converted into oilyproducts. when attempts were made to recrystallise them from hotsolvents, such as ether or ethyl acetate.This behaviour was doubtless due to the presence of traces of hydrochloric acid, which hadbeen retained when the pasty solid was washed with water. Otherpreparations separated from hot ethyl acetate, etc., in colourlessneedles, and were fractionally crystallised ; although the variousfractions appeared to be homogeneous and identical, they decom-posed a t different temperatures, the most sparingly soluble at about158-l65Oy the most readily soluble a t about 140-150°.All thefractions were impure, and left a residue when they were treatedwith a 5 per cent. aqueous solution of potassium hydroxide, themost sparingly soluble portion giving the largest quantity ofinsoluble matter. From these impure, although repeatedlyrecrystallised specimens, pure diphenylsilkanediol was finallyisolated by the method described later (p. 2121).Decomposition of Dichlorod~ph'enylsilicnn e with A mmonkwmHydroxide Solution.As it seemed to be probable that the impurities contained inthe diphenylsilicanediol, prepared by the foregoing method, wereproduced by the action of the hydrochloric acid on the originalproduct of hydrolysis, experiments were made on the decompositionof the dichloride with a cold concentrated solution of ammoniumhydroxide.The results, however, were most unsatisfactory.When the dichloride was slowly dropped into a well-stirredsolution of ammonium hydroxide, it was completely converted intoa pale yellow oil, which after having been extracted with cold lightpetroleum and thus freed from traces of diphenyl, remained liquidduring many weeks. This oil was soluble in alcohol; it did notcontain m y appreciable proportion of diphenylsilicanediol, because,when treated with a 5 per cent. aqueous solution of potassiumhydroxide, it gave an extract from which acids precipitated onlysmall quantities of the dihydroxy-compound.The further iiivestigation of this oil will be described in thefollowing paper2116 KIPPING: ORGANIC DERIVATIVES OF SILICON.PART XVI.Decomposition of Dichlorod~phenylsilicalze with a Solution ofPo tassium Hydroxide.A t an early stage of this investigation it was observed that thecrude diphenylsilicanediol, precipitated from a solution of itspotassium derivative with dilute acetic acid, had a much lowerdecomposition point than specimens obtained by precipitation withhydrochloric acid, and generally contained a relatively smallquantity of matter which was insoluble in potassium hydroxidesolution; as this observation seemed to confirm the conclusion thatthe dihydroxy-compound was changed by hydrochloric acid, thefollowing method of preparation was tried. Dichlorodiphenyl-silicane (1 mol.) was slowly dropped into a 5 per cent.aqueoussolution of potassium hydroxide (more than 4 mols.); as a slightdevelopment of heat occurred, the solution was cooled with waterand well stirred during the operation.The pure dichloride gave a clear, colourless solution of thepotassium derivative, but unless the dichloro-compound had beenvery carefully fractionated, it contained traces of diphenyl, whichremained suspended in the alkaline solution and necessitated filtra-tion. During this operation the solution and the filtrate nearlyalways became turbid, owing to the absorption of atmosphericcarbon dioxide, and it was very necessary to carry out the oper*tion as quickly as possible, because the turbidity was difficult t oremove by further filtration.When the clear alkaline solution was cautiously neutralised, orrendered very slightly acid, with acetic acid, it gave a somewhatgelatinous precipitate, which wit8 50 bulky that the mixture becamea thick, white paste ; the precipitate, which under the microscopeappeared to consist of amorphous particles, was readily separatedby filtration and washed with water by the aid of the pump, buteven when it had been well pressed and left on porous earthenwareduring some hours, it retained an extraordinarily large proportionof water.The freshly precipitated substance was readily andcompletely dissolved by a 5 per cent. solution of potassiumhydroxide, but, as a rule, it changed so quickly that by the timeit had been washed, it was no longer completely soluble.A great many different samples of air-dried preparations obtaindin the above manner were examined; when heated slowly from theordinary temperature, they all began to decompose at about 1 0 5 O ,and at about 107-1 loo they had completely liquefied, effervescencetaking place owing to the escape of steam.Some air-dried prepara-tions, which were heated a t looo, decomposed and liquefied com-pletely in the course of ten to twenty minutesRIPPING : ORGANIC DERIVATIVES OF SILICON. PART XVI. 2117When attempts were made to purify the precipitated diphenyl-silicanediol by recrystallisation, it behaved in a very curiousmanner. On being warmed with solvents such 88 ether, ethylacetate, etc., it seemed as a rule to be completely decomposed, andon subsequent maporation ah the ordinary temperature, the solu-tions gave a viscid, oily residue, which was very sparingly solublein alcohol.When dissolved in these solvents a t the ordinary tem-perature it ww sometimes deposited, by the spontaneous evapora-tion of the solutions, as a lustrous powder, which liquefied a t about140-155O; a t other times the solutions gave crusts of ill-definedprisms, whilst a considerable proportion of oily matter (readilysoluble in chloroform) remained in the mother liquor. These crustsusually decomposed with eff ervmence, and liquefied completely atabout 113-115q but even when well washed with chloroform theystill retained some impurity which was insoluble in a 5 per cent.potassium hydroxide solution. If such partly purified preparationswere recrystallised from cold ether or ebhyl acetate, colourlessneedles, decomposing at about 113-115O, might be obtained; onthe other hand, the decomposition point of the needlm mightsuddenly rise to about 135-140O.This change in properties was apparently not due to the removalof some impurity in the mother liquor, because a given sample,repeatedly dissolved in cold ether, nearly always changed, some-times after the first, sometimes after the second or third operation,even when the solvent was completely evaporated each time.Itwas also observed that those preparations, which decomposed at113--115O, sometimes changed spontaneously in the course of a fewhours at the ordinary temperature; they became white and opaque,and in such cases the decomposition point was found to have risento about 138-145O.I n spite of the fact that some preparations changed, apparentlyspontaneously, a t the ordinary temperature, it was sometimespossible to recrystallise small quantities of material from hot ethylacetate or chloroform without any rise in decomposition pointtaking place; the first time, in fact, that a preparation of this lowdecomposition point was obtained, no sign of instability wasobserved and no special precautions were taken in its purification,and yet i t was obtained in fairly large prisms liquefying a t about11 6O.As the addition of a trace of hydrochloric acid t o a solution ofthese preparations invariably brought about a rise in decomposi-tion point to about 145O, i t seemed possible that the behaviour justreferred to might be due to the presence of traces of acids in thesolvents employed.Some specimens were therefore recrystallisec2118 KIPPING: ORGANIC DERIVATWES OF SILICON. PART XVI.from ether a t the ordinary temperature, the ethereal solutionresting on water which contained a trace of potassium hydroxide;in these circumstances the diphenylsilicanediol was deposited inbeautiful, lustrous prisms, which were completely soluble in potass-ium hydroxide solution, and decomposed with effervescence a t115-118O. Similar results were obtained when the preparationswere recrystallised from ether in presence of a trace of piperidine,but neither method gave very satisfactory results, a ~ , unless theoperation wits conducted quickly, the diphenylsilicanediol wasdecomposed.A very simple modification in the method of obtaining diphenyl-silicanediol from its alkaline solution seemed t o give a pure pre-paration, or, a t any rate, one which was completely soluble inpotassium hydroxide solution; this was the addition of a littleether to the solution of the potassium salt before precipitating withacetic acid.I n presence of the ether, the precipitate lost entirelyits gelatinous character; it was far less bulky, quite distinctlycrystalline, and was readily freed from water when it was pressedon porous earthenware. As such preparations appeared to be homo-geneous under the microscope, were completely soluble in potassiumhydroxide, and had the same decomposition point as the prismsobtained from ether, they seemed t o be pure diphenylsilicanediol.The following analyses of three different air-dried specimensagreed with this view:C=66*7; H =5*7.11.0.1610 ,, 0.3932 GO, ,, 0.0838 H,O. C=66*6; H=5.8.111. 0.1743 ,, 0-4254 CO, and 0.0885 HzO. C=66*6; H=5*6.I. 0.1454 gave 0.3554 CO, and 0.0750 H,O.0.1956 ,, 0.0547 SiO,. Si=13*2.0.1489 ,, 0.0419 SiO,. Si=13.2.CI2H,,O2Si requires C = 66.6 ;, H'=5-6 ; Si = 13.1 per cent.Notwithstanding this evidence, there is little doubt that thesesamples contained a small proportion of impurity, which loweredthe decomposing point of the diol, but which could be removed bythe method giveq later.It would seem from subsequent observations that the behaviourof the diphenylsilicanediol, precipitated with acetic acid as describedabove, is due to the presence of traces of potassium hydroxidewhich are adsorbed from the alkaline solution by the somewhatgelatinous precipitate, and are not removed when the solutionappears to be neutralised, or when the precipitate is subsequentlywashed with water; as a result of the action of the alkali, smallquantities of products, some of which are insoluble in potassiumhydroxide, are formed, and, by the continued action of the alkalibefore or after recrystallisation, these products may be furtheKIPPIbTCI : ORGANIC DERIVATIVES OF SILICON. PART XVI.2119changed, and even the whole of the diphenylsilicanediol may bedecomposed. When precipitation occurs in presence of ether, theprecipitate is not.colloidal, and does not adsorb alkali, or does soto a limited extent only, in consequence of which it is far morestable; such precipitates, however, contain a small proportion ofimpurity, which is soluble in solutions of alkali hydroxides.P r e cipi t a tion of Diph eny lsilican edio 1 with Hydrochloric A c id.When a solution of diphenylsilicanediol in potassium hydroxidewas treated with hydrochloric acid, it gave a bulky precipitate,which seemed to have much the same gelatinous character as thatobtained with acetic acid, but which, when air-dried, had a muchhigher decomposition point than the latter ; the decompositionpoint, moreover, seemed to vary with the quantity and concentra-tion of the acid which was used, and with the time which elapsedbefore the precipitate was separated and washed.If, for example,the solution was only just neutralised, and the precipitate wasimmediately separated, the decomposition point of the air-driedsample might be 130-135O; if, however, the solution was stronglyacidified and the precipitate was not separated until the next day,the sample inight only sinter slightly a t about 140°, and not liquefyuntil about 160-1 65O. Such preparations, especially those whichhad remained for some time in contact with hydrochloric acid,contained a variable quantity of matter which was readily solublein cold chloroform; when this had been extracted and the residuewas fractionally crystallised from ether, ethyl acetate, acetone, etc.,deposits, which decomposed from 160° to 170°, or from 135O to 145O,or a t intermediate temperatures, were obtained.All these fractionsappeared homogeneous, but left a variable quantity of an insolubleresidue when treated with a 5 per cent. potassium hydroxidesolution.Repeated attempts were made to obtain from them a pure sampleof the diol, but fractional crystallisation from the different mediaalready mentioned did not give the desired result. During theseexperiments it was noticed that when solutions in ether or ethylacetate were evaporated on the water-bath, crystallisation did nottake place, even when the solutions had become highly super-saturated, but when such solutione were then cooled, they gaveeither a transparent gelatinous mass or a cotton-wool-like solid,which then gradually changed to a crystalline powder.Preparations apparently very similar in character t o those justdescribed, but nearly free from matter soluble in cold chloroform,were obtained when alkaline solutions of the diol were treate2120 KIPPING: ORGANIC DERIVATIVES OP SILICON.PART XVI.with excess of ammonium chloride, which precipitated the whole ofthe diphenyl compound ; the fractional crystallisation of thesepreparations failed to give a product which was completely solublein pota.ssium hydroxide solution, and the decomposition points ofthe well crystallised fractions varied from about 135 to 160O.Isolation of Di~l~enylsiticanediol.In the course of the experiments recorded above there had accu-mulated a considerable quantity of recrystallised, impure diol,which had been produced by the decomposition of the dichloridewith water, or from the potassium derivative of the diol, by pre-cipitation with hydrochloric acid or ammonium chloride.Differentsamplm or fractions of this material showed very considerabledifferences in decomposition point, some liquefying a t about130-140°, others a t 140-150°, or even up to 160O. Preparationsof an apparently similar character had also been obtained fromthose decomposing at about 115O; when the latter were dissolvedin ethyl acetate with the addition of a trace of hydrochloric acid,the solutions gave on spontaneous evaporation colourless needlesor prisms, decomposing from about 145O up to 160O.Many of these samples had been repeatedly fractionally crystal-lised from ethyl acetate, ether, acetone, benzene, etc., and frommixtures of chloroform and ether, and chloroform and acetone, butin no case had a pure preparation been obtained; although thecrystals which separated from warm solvents consisted of beautiful,lustrous needles or prisms, and appeared homogeneous, they invari-ably left an insoluble residue when treated with potassiumhydroxide solution.The deposits obtained by the spontaneousevaporation of the solutions occasionally appeared to be hetero-geneous, and consisted of lustrous needles, together with white,cauliflower-like masses, which crept up and over the side of thebeaker ; these apparently different deposits had, however, practi-cally the same decomposition points, and their mechanical separa-tion led to no result.In nearly all cases the more sparingly solubleportions had the higher decomposition points (up to 170°), andseemed to be the more impure.It was also observed that specimens decomposing at about 140°,when left exposed t o the air at the ordinary temperature, sometimesunderwent some change, and then did not decompose until160-165O; this behaviour was not due to loss of solvent, and isreferred to later.Up to this time the use of aqueous solvents for the purificationof the diol had been avoided because it had been observed that hoKIPPINO: ORGANIC DERIVBTIVES OF SILICON. PART XVI. 2121aqueous alcohol and acetic acid decomposed the compound; as a lastresource, however, aqueous acetone was tried, with the followingresults.When any of the preparations of impure diphenylsilicanediol,which decompoaes above about 1 3 5 O , was dissolved in acetone, andthe solution diluted with a small proportion of water, the liquidbecame milky, and when left to evaporate a t the ordinary tempera-ture the milky fluid deposited colourless prisms.After a few days’time practically the whole of the diol had crystallised out, andwas easily separated from tho milky mother liquors, which passedunchanged through an ordinary filter; as a rule, in fact, thecrystals were so large that the mother liquor might be merelydecanted from them.The milkiness of the original solutions appeared to be a roughmeasure of the impurity in the diol; according to this criterion,the preparations obtained by precipitating the alkaline solutionwith ammonium chloride were the least impure, whilst thoseobtained by decomposing the dichlonde with water were less impurethan those produced by precipitating the diol with hydrochloricacid. The crystals which separated from the milky fluid werecovered with an oily film, which was easily removed by washingthem with cold chloroform. When the washed crystals were againdissolved in acetone, and the solution wit^ diluted with a littlewater, a milky fluid was a.gain produced, but for any given samplethe milkiness was far less pronounced than it was a t the firsttreatment.The crystals which were subsequently deposited, whenagain washed with chloroform and then dissolved in acetone, usuallygave solutions which did not become milky on the addition ofwater, but which immediately deposited a crystalline precipitate ifsufficiently diluted.I n those cases in which a milkiness wasproduced, the purification was still incomplete, but after a repeti-tion of the operations a pure compound was obtained.The crystalline substance, isolated in this way and then finallyrecrystallised from hot ethyl acetate or acetone, was pure diphenyl-silicanediol ; the milky mother liquors, which often remained un-changed in appearance during many months, and the chloroformwashings of the crystals, contained the relatively small quantity ofimpurity, which apparently could not be removed by recrystallisa-tion from anhydrous solvents, and seemed so greatly to affectthe decomposition point of the pure diol.This impurity, so faras has been ascertained, is usually a mixture of three or fourcompounds.By the use of aqueous acetone and chloroform alternately, it wasalso possible to isolate diphenylsilicanediol from the recrystallise2122 KIPPING: ORGANIC DERIVATIVES OF SILICON. PART KVI.preparations decomposing a t about 115O, which had been originallyobtained by precipitation with acetic acid. Such preparations gavewith aqueous acetone, solutions which, as a rule, were not verymilky, but from which on spontaneous evaporation, there graduallyseparated both crystals and oil; the latter was easily removed withthe aid of cold chloroform, and the crystalline residue was redis-solved in aqueous acetone.The solutions then gave a deposit con-taining far less oil, and by a repetition of these processes the purediol was finally obtained. When two or three air-dried prepara-tions, which had been obtained from the potassium derivative byprecipitation with acetic acid and then kept during some weeks,were examined in this way, they were found to be highly impure,and apparently the longer a sample had been kept the greater theproportion of impurity.Diphen y ldican ediol, Si Ph,( 0 H),.The various samples of diphenylsilicanediol, isolated from thedifferent preparations by the method described, were carefullycompared; they all had the characteristics of a pure compound, andwere identical with one another:0-1480 gave 0.3605 CO, and 0,0743 H,O.C = 66.4 ; H = 5.6.0-1480 ,, 0-0412 SiO,. Si=13*1.CBHl20,Si requirw C = 66.6 ; H = 5.6 ; Si = 13.1 per cent.Diphenylsilicanediol crystallises from ether, ethyl acetate, acetone,etc., in long, colourless needles or prisms, which often exceed 20 mm.in length; when its solutions in ethyl acetate evaporate spontane-ously, it is sometimes obtained in large, well-defined crystals, suitablefor goniornetrical measurement.It is practically insoluble in water or light petroleum, onlysparingly soluble in cold chloroform, and very moderately so inboiling benzene, but it dissolves more readily in warm ether, andfreely in boiling ethyl acetate or acetone.All the freshly prepared, air-dried specimens of diphenylsilicane-diol, purified in the manner described above, and finally recrystal-lised from ethyl acetate, showed a t first the same behaviour whenthey were heated and directly compared; they began to sinterslightly a t about 125O, and if then the temperature was slowlyraised, they liquefied completely a t about 128-132O, and a vigorouseffervescence was observed owing to the escape of steam. In thecase of any given sample, however, the temperature at whichdecomposition set in and complete liquefaction took place, variedslightly with the conditions of the experiment, principally with therate of heating, but also with the state of division of the substanceKlPPlNG : ORGANIC DERIVATIVES OF SILICON.PART XVI. 2123The temperatures just given, and most of the decompositionpoints recorded in this paper, were observed when the rate ofheating (about looo) was a rise of about loo per minute.Although up to a certain time many freshly prepared samples ofthe pure diol had been heated, none had shown a decompositionpoint above 1 3 2 O , and it seemed that the behaviour of the purecompound was invariable.When, however, four such preparationswere examined again some two months after they had beenobtained, one of them seemed to have undergone some change, andfurther observations were made.One of these specimens, A , had been kept in a loosely-coveredbeaker, exposed to the laboratory air, and one of them, B, in aclosed weighing-bottle; the other two, C and D, had been kept onclock glasses in a desiccator which contained a little damp soda-lime.The specimens A and B had not changed in appearance, andconsisted of transparent needles. The specimen C seemed to havechanged very slightly in appearance as the needles seemed some-what opaque in parts. The specimen D, which consisted of rathersmall crystals, had obviously undergone some change, the previouslytransparent needles having become white and opaque ; althoughthis specimen sometimes decomposed completely below about 1 3 2 O ,i t usually showed no signs of change until about 150°, and decom-posed with effervescence a t about 155-160O; further, it was nolonger readily and completely soluble in a 5 per cent. aqueoussolution of potassium hydroxide, but gave a very small proportionof a flocculent substance, which did not dissolve in the course ofabout ten minutes.The other three specimens, A , B, and C, usually decomposedconipletely a t 128-13Z0, but occasionally one of them did notsinter until about 150°, and decomposed with effervescence a t about155--159O; they were all completely soluble in a 5 per cent. aqueoussolution of potassium hydroxide.The results of a great many more experiments with the samplesA , 23, and G, and with various freshly prepared pure specimens,proved that diphenylsilicanediol might show one of two verydifferent decomposition points ; when six to eight melting-pointtubes, all containing the same finely powdered sample, were heatedsimultaneously, the contents of one, or possibly two, of the tubesmight not decompose until about 145O or even 1 5 8 O , those of allthe others having completely liquefied below 1 3 2 O .Now as it is most improbable that the observed difference inbehaviour between identical samples under as nearly as possibleidentical conditions could be due merely to a difference in the rateof decomposition, i t seems necessary to conclude that diphenyl-VOL.CI. 7 2124 KIPPING: ORGANIC DERIVATIVES OF SILICON. PART XVI.silicanediol is dimorphous; that the crystalline form which isdeposited from solvents and which decomposes below 1 3 2 O may passinto a more stable one, which does not decompose until about 160O.This conclusion, which appears to be fully established, affordsan explanation of many of the apparently anomalous facts recordedin this paper.In the first place, it is obvious from the behaviourof the pure compound, that the change in crystalline form whichsometimecs occurs is due to some inappreciable and fortuitous differ-ence in the experimental conditions; consequently, i t is not onlypossible, but highly probable, that the presence of impurity of aparticular kind might condition the change in crystalline form,especially if the impurity were isomorphous with the form whichis stable at the higher temperature. The fact that all specimensof diphenylsilicanediol containing a certain proportion of thosesubstances which are insoluble in potassium hydroxide solutionhave always, and those containing traces of such impurities havegenerally, the high decomposition point, would thus be explained ;as these impurities are probably trianhydrotrisdiphenylsilicanedioland tetra-anhydrotetrakisdiphenylsilicanediol (compare followingpaper), neither of which would be melted a t 1 3 2 O , the presenceof even a trace of either substance might cause the crystallinetransformation of the diol to take place.The fact that those impure specimens of the diol which containsmall quantities of substances soluble in potassium hydroxidesolution, and which decompose a t about 1 1 8 O , have a much higherdecomposition point after they have been crystallised in presenceof an acid, may also be accounted for; the impurity which lowersthe decomposition point of the diol is some compound or mixtureof low melting point which is changed by the acid into one of theabove-named substances of high melting point ; experimentalevidence which strongly supports this assumption is given later(p.2141).The spontaneous rise in decomposition point which was oftenobserved in the case of impure samples of the diol might, of course,have been apparent only, that is to say, the sample might not havechanged until its decomposition point was being taken; in any case,unless the specimen were free from alkali and acid, a change in thedecomposition point, might well occur as the result of chemicalchanges in the impurity present, or in the diol itself; such changesmight lead to the formation of the substance or substances whichfavour the crystalline transformation of the diol.So far the only instance in which a highly purified sample appearsto have changed is that of the specimen D, referred to above;whatever may be its cause or its nature, this change was only verKIPPING : ORGANIC DERIVATIVES OF SILICON. PART XVII. 2125superficial, and the opaque crystals immediately became trans-parent when they were moistened with acetone.One further point remains to be considered, namely, how Martinwas led to believe that the preparations which he regarded asisomeric diphenylsilicols could be transformed one into the otherby the methods which he describw (Zoc. cit.). According to hisstatements, when the I‘ isomeride ” decomposing a t about 160° wasdissolved in an aqueous solution of potassium hydroxide, and thesolution then treated with acids, the “ isomeride ” decomposinga t about 1 4 4 O was obtained, but if alcoholic potassium hydroxidewas used, the original ‘(isomeride” decomposing a t about 160°was precipitated. These statements are doubtless incorrect in sofar as any question of isomerism is concerned, but it is possiblethat the actual observations are to be accounted for as follows:The crude preparation decomposing a t about 160° contained a con-siderable proportion of impurity which was insoluble in an aqueoussolution of potassim hydroxide ; when aqueous alkali was usedand this impurity became visible, it was separated by filtrationbefore the solution was acidified, so that the precipitated diol wasless impure than the original sample, and decomposed a t a ratherlower temperature. When, however, the original sample wasdissolved in alcoholic potassium hydroxide, the impurity was notprecipitated, or was only partly precipitated, and consequently thediol finally obtained on the addition of an acid was just as impureas before, and decomposed at approximately the same temperatureas the original sample.The author is indebted to Mr. T. A. Smith, B.Sc., for someassistance in the preparation of diphenylsilicanediol,Government Grant Committee of the Royal Societyin aid of this research.UNIVEHSWY COLLEGE,NOT‘l‘ ING HAM.and to thefor a gran