2190 JERUSALEM : THE MOlWHOTROPIC RELATIONSHIPSCCXXVI1.-The Morphotropic Relutionships betweerLSilicon and Carbon Cornpouds of Corr-espondinyCompositiorhs.By GEORGE JERUSALEM.CARBON occupies an altogether unique position amongst the elementsin crystallographio as well as in chemical respects; although largenumbers of compounds of the elements of group I V of the periodicclassification have been crystallographically examined, no case hasbeen yet observed in which a carbon atom is isomorphously replacedby one atom of any other element. It is true that both carbontetraiodide and silicon tetraiodide crystallise in the cubic system,but, since the crystal class is known in neither case, the informationrequisite for deciding as to whether these two compounds are is*morphous is lacking.Isomorphism might be expected as betweenthe metallic carbonates and metasilicates, but although comparisonis possible in a number of cases, no instance is on record in whichsilicon replaces carbon without profound modification of thecrystalline form ; thus lithium carbonate, Li,CO,, is monosymmetricwith a : b : c = 1.672 : 1 : 1.244, p= 114O25’ (Mallard, Zeitsch.Rryst. Mh., 1894, 2 3 , 483), whilst lithium metasilicate, Li,SiO,,is rhombohedra1 with a: c= 1 : 0.6681 (Friedel, Zeitsch. lilryst.Min., 1903,37, 204). Comparison of the crystalline forms exhibitedby the carbonates of the bivalent metals with those of thecompounds CaSiO,, MgCa( SiO,),, MgSiO,, MnSi’Os, MgTiO,,MnTiO,, and FeTiO,, fails to reveal any isomorphous relationshipbetween the carbonates and the latter compounds.Further, theobservation that carborundum, CSi, is a stable substance dissimilarin crystalline form from either component element does not favourthe view that the two elements are isomorphous.Whilst a considerable amount of evidence, such as the above,indicates that carbon is crystallographically quite unique, it isnoteworthy that ample evidence is available to show that silicon isdisplaceable by many bivalent elements without considerable changein crystalline form.From a discussion of the crystal data available for carbon andsilicon compounds, Barlow and Pope have been led to attributethe crystallographic-and also chemical-dissimilarity to a differencein the fundamental valencies of these two elements (Trans., 1908,93, 1554); carbon thus stands alone as the only element exhibitingthe fundamental valency of four, whilst silicon and many otherelements are fundamentally bivalent.Owing to the comparativBETWEEN SILICON AND CARBON COMPOUNDS. 2191ease with which, during the last few years, it has become possibleto exhibit the relation between chemical constitution and crystallineform, the question of the relative fundamental valencies of siliconand carbon has become one of great importance; the work describedin the present paper was undertaken as a step towards the solutionof the problem involved.A study of the crystal data already available for correspondingsilicon and carbon compounds indicates clearly that the two elementsexercise such different morphotropic effects that few positive resulhcould be hopefully anticipated unless some condition, hithertounconsidered, were introduced for the purpose of accentuating suchcrystallographic similarity of function as may exist between siliconand carbon.The directing condition, thus indicated as desira.ble,is found in the rule discovered empirically by Tscbermak (Tsch.Hin. Mitt., 1903, 22, 393; Ann. Reports, 1908, 5, 263) that, inthe crystalline form of a compound substance, the principal axes ofsymmetry tend to express numerically the atomic composition ofthe molecule; thus, if three similar atoms are present with othersin the molecule of a particular substance, the crystalline form ofthe compound will, in the majority of cases, include a threefoldaxis of symmetry.Since crystal structures are now regarded asclose-packed assemblages of atomic domains, a compound containingidentically located atoms, or groups of atoms, in the moleculewould tend to exhibit it particular element of symmetry displayingan n-fold repetition; Tschermak’s empirical rule has thus nowacquired theoretlcal significance and, simultaneously, a concretemeaning. The rule may be conveniently applied to the purpose ofaccentuating any possible morphotropic relationship between twoelements a and b, by preparing two substances containing in themolecule three large atomic groups or radicles associated with ana and b atom, respectively, of a unique kind; any morphotropyexhibited as between a and b should then be easily traced bygoniometric examination.This particular development of the new mode of regarding crystalstructure is applied in the present paper to the examination oftribenzyl and triphenyl derivatives of silicol and carbinol; theintroduction of three benzyl or phenyl groups into the molecule is,of course, intended to secure t.he presence of a trigonal axis ofsymmetry and to ensure that the silicon and carbon analogues shallto a very great ext.ent exhibit identical marshalling of their com-ponent atoms.Trib enzylsilicol, (C,H,*CH,),Si*OH.A supply of this substance was kindly provided by Prof.F. S.Kipping, and melted at 1 0 6 O , as stated by Dilthey (Ber., 19052192 JERUSALEM : THE MOHPIIOI'ROPIC RELATIONSHIPS38, 4136).On slow evaporation of its solution in a mixture ofchloroform and petroleum boiling at 70°, it was obtained in smaI1,lustrous crystals suitable for measurement; the form (001) is thelargest, and {loo}, {101}, and { l l l } , although smaller, are welldeveloped. The pyramid, (122}, is always very small but quitebrilliant; no distinct cleavage was observed, but, as the faces of{ 111) are always much larger at one end of the c-axis, and its thefaces of { 122) are only observed at the same end of this axis, thesubstance is probably hemimorphous. No confirmation of the hemi-morphism was obtained by the study of the etch-figures producedby the action of alcohol on faces of the form (001).Crystalline form : Orthorhombic.a : b : c=1'7214: 1 : 2.1384.Forms observed: { l o o ) , {OOl), {101}, { l l l } , and (122).The following angular measurements were obtained :Angles.111 : Ti1111 : 111111 : 100101 : 001001 : 122122 : 122111 : 001101 : 111100 : 1221.01 : 100Number ofobservations.161433134919410Limits.55" 2'-55'49'43 46 -44 1561 31 -62 5450 59 -51 3565 31 -66 638 26-39 667 43 -68 1753 16 -53 2774 51 -75 2048 23Mean.55'30'20"44 1506213 051 13 3065 53 038 47 048 23 053 19 1075 7 5068 150Calculated.-62"14 50"51 10 038 50 048 22 2067 59 053 17 1075 15 3065 48 50It is very interesting to note that, although the substance doesnot exhibit a, trigonal axis of symmetry as would be anticipatedfrom Tschermak's empirical rule, it possesses a pseudetrigonal axis,as would be expected from the interpretation of the rule and themolecular composition in accordance with Barlow and Pope'smethod.Thus, a crystal presenting trigonal symmetry, referred torectangular axes, would exhibit as one axial ratio the value1 : d 3 = 1 : 1.7321, whilst an almost identical ratio, b : a = l : 1'7214,is actually observed on the crystals of tribenzylsilicol ; the pseudo-trigonal nature of the crystal structure is thus apparent. It willbe seen later that all the substances described below, in whichthree large groups are present in the molecular complex, conformt o the same rule, and betray the presence of a trigonal or a pseudo-trigonal axis, with the possible exception of triphenylsilicol, ofwhich the low crystal symmetry would naturally obscure the pseudo-trigonal character.Taking the valency volume of silicon as two, that of tribenzyl-silicol is W=110, and the equivalence parameters are obtained its:x : y : z=5.3418 : 3.1032 : 6.6358BETWEEN SILICON AND CARBON COMPOUNDS.21 93The density of the crystalline substance wits determined byRetgers' floating method in barium mercuri-iodide solution asd = 1.1772, whence the molecular volume, V = 270.66 ; the quotient,R = V / W = 2.4605, and the molecular distance ratios are calculatedas:x : J/ : w = 7 2116 : 4.1894 : 8.9586.Trib enzylcarbinol, (C,H,*CH,),C*OH.This substance was prepared by the method given by Klages andHeilmann (Ber., 1904, 37, 1456), and was obtained in crystalssuitable for goniometric measurement by spontaneous evaporationof its solution in a mixture of chloroform and petroleum.Thecrystals so closely resemble those of the preceding compound thatno separate description is necessary.Crystalline system : Orthorhombic.a : b : c=1*7166 : 1 : 2.1574.Forms observed: {loo}, {OOl}, {101}, { l l l } , and (122).The following angular measurements were obtained :Angles.1QO : 111001 : 111001 : 101100 : LO1111 : 111111 : 111.111 : T i 1100 : 122001 : 122Number ofobservations.26231018121316126Limits.6V52 '-62"20'68 0 -68 1751 18 -51 3738 21 -38 4455 32 -55 5143 31 -44 153 9 -53 3674 51 -75 2665 57 -66 7Mean.62" 8'50"68 9 2051 26 5038 32 5055 37 1043 40 2053 20 5075 9 4066 2 10Calculated - -51"28 '10"38 31 5055 42 2043 41 2053 19 3075 12 065 59 20The development of the faces indicates hemimorphous develop-ment as in the case of the corresponding silicol; the etch-figuresobtained do not reveal hemimorphism.The axial ratios calculatedshow that the substance is very closely related morphotropicallyto the corresponding silicol, and the present observations constitutethe first published evidence of morphotropy as between these twoelements.x : y : z =5*3482 : 3.1156 : 6.7215, with W = 112.The equivalent parameters are calculated ~ t s :The density of the crystals was determined as d = 1-1869, whencethe molecular volume, V = 258.99, and the quotient, I2 = V / TV =2.3124.The molecular distance ratios are therefore :x : $ : w * 7.0724 : 4.1200 : 8 8885.Tribensylmethyl Chloride, (C,H,*CH,),CCl.This substance was prepared by the method given by Schmerda(Monatsh., 1909, 30, 390), and exhibited the properties describe2194 JERUSALEM : THE MORPHOTROPIC RELATIONSHIPSby him; it separates from its acetone solution in very small butquite brilliant rhombohedron-shaped crystals.Crystalline system : Rhombohedral. Trapezohedral-tetatohedralclass.a : c = l : 0.3700.Forms observed : { 2fiO} and { lOil}.The following angular measurements were obtained :Number ofAngles, Observations. Limito. Mean. Calculated.2iTo : 1011 41 70” 0’-70”22’ 70” 6’20’’ -ioii : o i i i 18 39 34 -40 0 39 49 20 3 9 ‘4 7 ’2 0”The evidence that the crystals belong to the trapezohedral-tetartohedral class represented by quartz and cinnabar is, first, thatthe alternate faces of the form (2110) are very different in size,and, secondly, that the etch-figures produced on these faces by theaction of benzene are asymmetric with respect to the hexagonalplanes of symmetry normal to the faces.A good cleavage isobserved parallel to the form (2110).In order that the crystal form may be compared with those ofthe preceding compounds, it must first be stated with respect to thealternative hexagonal system and then ref erred to rectangular axesand the value, c / a , multiplied by five; the axial ratios are thusobtained in the form:a : b : c = 1.7321 : 1 : 2.1364, and the equivalence parameters as :x : y : z = 5*3,656 : 3.0977 : 6.6180, with W= 110.The use of the factor five in the multiplication of the ratio c / ais naturally just,ified by the very close similarity which theequivalence parameters of the three above substances exhibit afterthe multiplication has been performed in the case of tribenzylmethylchloride.Triphemjls&col, (C6H,)3Si*OH.Triphenylsilicol was prepared by Dilthey and Eduardoff’s method(Ber., 1904, 37, 1140), and showed the properties described bythese authors; the best crystals were obtained from solutions inmixtures of chloroform and petroleum, but even these were poor,and scarcely suitable for goniometric examination.The forms{loo}, {OlO}, and (001) show the largest faces, and are aboutequally well developed ; { 11 1 } and { 111 } are very poorly developed,and unsatisfactory in character.Crystalline system : Anorthic.a : b : c = 2,144 : 1 : 1.331, a = 5g030f, B = 113O291, y = 84O11’.Forms observed: {loo}, {OlO}, {OOl), ( l l l } , and (111)BETWEEN SILlCON AND CARBON COMPOUNDS.2195The following angular measurements were obtained :Angles.100 : 010 o_oi : 010100 : 111111 - : 001111 : 010111 : 001111 : 010111 : 100LOO : 001Number ofobservations.16810997542Limits.58" 1'--58'48'67 5 -67 4652 50 -53 2964 10 -54 4790 38 -91 1841 0 -41 5861 56 -62 3868 22 -68 5358 57 -59 24Mean. Calculated.58"20'50" -67 25 0 -53 6 10 -64 29 30 -90 57 40 _-41 39 10 42'10'20''6219 0 62 19 4068 36 30 69 2 2059 10 30 59 46 30The density of the crystals was observed as d = 1.1777, so that themolecular volume, V = 234.83 ; the quotient, R = V / W'= 2.5525No morphotropic relationship is immediately obvious between thisand the previously described substances, and, since triphenylsilicolbelongs t o the anorthic system, the symmetry affords no indicationas to the way in which a morphotropic relationship is to be sought;further, the ratio, V / W , is considerably larger than in the othercases referred to.It is consequently t,o be concluded that thissilicol does not fall into line with the series now under discussion.TriphenylcarbkoZ, (C,H,),C*OH.This substance has already been crystallographically examinedby Groth (Zeitsch.Kryst. .&fin., 1881, 5, 479), who found it to berhombohedra1 with a : c = 1 : 0.6984. Measurable crystals wereobtained from benzene solution, and these showed only the forms{lOTl} and { l l z O ) .Crystalline system : Rhombohedral.The following angular measurements were obtained :a : c = 1 : 0.6975.Number ofAngles, Observations. Limits. Mean. Calculated.2120 : lgll 13 56'46'-57"19' 57" 5'40" -1011 : 0111 5 65 35 -66 5 65 47 20 65O48'40''In order to render the crystal form comparable with those of thetribenzyl compounds described above, the ratio cia must be multi-plied by three and referred to a rectangular system of axes. Thefollowing ratios are thus obtained :a : b : c=1'7321: 1 : 2.0925.X : 9: ~=5*1271: 2'9601: 6.1939; W=94.The density of t'he crystals was determined as d = 1.1884, so thatthe molecular volume, V = 218.92, and t.he quotient, R = V / W =2.3289. The moIecular distance ratios are :x : $ : o = 6.7960 : 3.9236 : 8*2100.The value for R is slightly greater than that for tribenzylcarbinol2 196 JERUSALEM : THE MORPKOTROPIC RELATIONSHIPSnamely, 2.3124 ; this is in accordance with the indications obtainedin the case of the picrates and styphnates. Aniline picrate givesan R value of 2.464, slightly greater than 2.433, the value forbenzylamine picrate (Jerusalem, Trans., 1909, 95, 1290).For comparison with the above substances the following crystallineforms may be quoted.Triphenylmethane is described as ortho-rhombic by Hintze with a: b : c=0*5716 : 1 : 0.5867 (Zeitsch.Kryst.Min., 1884, 9, 546); on interchanging a and 6, and multiply-ing the resulting value of c / b by two, the axial ratios become:a : b : c =1*7495: 1 : 2.0528.o-Bromotriphenylmetliane, Ph,CBr, was found to be hexagonalby Hintze (Zoc. cit.) with the value (1; : G = 1 : 0.7843 ; on stating thisratio in the alternative hexagonal form as a : c = 1 : 0.6792, multiply-ing c / a by three, and referring the ratio to rectangular axes, thevalues are obtained as :a: b : c=1.7321: 1 : 2’0376.Triphenylacetic acid is monosymmetric (Groth, Zeitsch. Kryst.Min., 1881, 5, 483), and the axial ratios can be stated in the forma: b : c=0*5646: 1 : 0.6161, P=9O012’30’’ (Barlow and Pope,Trans., 1906, 89, 1719).On treating these axial ratios in themanner adopted with triphenylmethane, namely, multiplying c bby two, and then interchanging a and b, the values become:a : b : c=1*7712 : 1 : 2.1824, a=90°12’30”.The additive compound of triphenylmethane and benzene,(C,H,),CH,C,H,, is rhombohedra1 with a : c = 1 : 2.5565 (Hintze,Zoc. cit.) ; by referring this ratio to rectangular axes it becomes :a : b : c=1*7321: 1 : 2.5565.A consideration of the axial ratios, molecular distance ratios, andequivalence parameters for the above substances shows, first, thatalthough the morphotropic relationships are distinctly evident inthe axial ratios, they axe much more completely expressed by theequivalence parameters. Secondly, it is obvious that the moleculardistance ratios, although not greatly dissimilar in the instances inwhich they have been determined, differ much more amongst them-selves than do the equivalence parameters; the degree with whichthey correspond is, in fact, measured in the main by the degreeof approximation of the respective values of R = V / W .I n this,as in the majority of other cases, the molecular distance ratios lendthemselves less fruitfully to the discussion of morphotropic relation-ships than do the equivalence parameters ; the morphotropy musttherefore be considered merely in the light of the equivalenceparameters, and the axial ratios and molecular distance ratios maybe disregardedBETWEEN SILICON AND CARBON COMPOUNDS. 2197The following table states the equivalence parameters of all thesubstances dealt with above, calculated from the sets of axial ratiosfinally adopted.In the case of tribenzylsilicol, it is convenient tostate the equivalence parameters in accordance with the alternativesthat Si=2 and 4:1. (O,H,*CH,),Si*OH ......2. (CGH5'CH,)3Si'OH .. ...3. (CGH5*CH2)JC*OH ......4. (C,H;CH,),CCl .........5. (C, H,),C*OH . . . . . . . . . .6. (C,H,),CH . . . . . . . -. . . . . . .7. (C,H,)& Br , . . . . . . . , . . . . . ,8. (C,H,),C*C02H .........9. (CGH,),CH,C6HG ... ......5. Y. 6. 5'3418 : 3.1032 : 6.63585.3740 : 3'1219 : 6.67585,3482 : 3.1156 : 6.72155'3666 : 3.0977 : 6.61805,1271 : 2.9601 : 6.19355'1572 : 2'9479 : 6.05155.1359 : 2.9651 : 6.04165.2384 : 2.9576 : 6.45435.2314 : 3.0203 : 7.7213W=110 Si=2W=112 S i = 4 w= 112 w= 110lV= 94IV= 92lV= 92lt'= 100TI.'= 122Considering the parameters 2 and 3, calculated on the basis thatsilicon and carbon both exhibit the fundamental valency 4, it isnoticed that in passing from the silicol to the carbinol, the x-valuediminishes by about one-half per cent,., whilst the z-value increasesby rather a larger fraction.If crystal structure is to be regardedas a question of close-packing, it is difficult to conceive that inthe large tribenzylsilicol molecule the displacement of the siliconby a carbon atom of approximately the same valency volume fourcan lead, without change of symmetry, t o such it notable changein dimensions of the packed structure; it seems thus indicated thatcarbon and silicon have not the same fundamental valency of four.On considering next the values 1 and 3, calculated on the basisof Si=2 and C=4, it is seen that the differences for x and y aresmall, and that practically the whole weight of the displacement of2.2 and 3 Differences 0.02581 , 9 3 ,, 0.00643 7 9 4 ,, 0'01743 9' 5 ,) 0'22115 9' 6 ,) 0'03015 ,, 7 ,, 0.00875 Y , 8 ,, 0.11136 1 , 9 ,, 0'0742!I* x.0-0063 0'04570.0121 0-08570*0179 0.10350.1555 0'52800'0122 0'14200-0050 0'15190.0025 0.15190'0524 0.6698silicon by carbon falls on the z-parameter, which thus changes bynearly 2 per cent..; this is in complete agreement with what wouldbe anticipated by an increase of volume of one constituent atomicdomain from 2 to 4.Such a substitution, provided that themechanical operation of Tschermak's rule conserves the marshalling,could well lead to the expansion of the assemblage in one of thethree rectangular directions of principal symmetry. An identicaleffect is, in fact, observed in operation in other cases, notably inthe passage from tribenzylcarbinol to tribenzylmethyl chloride,where a similar change of valency volume by two units occurs a2 198 JERUSALEM : THE MORPIIOTROPIC RELATIONSHIPS, ETC.the result of the displacement of the hydroxyl group by the chlorineatom; here again small changes occur in the values of two dimen-sions, x and y, and most of the weight of the substitution is thrownon to the third or z-parameter, which alters by about 1.5 per cent.These considerations suggest that silicon and carbon have notthe same fundamental valency, but that, whilst that of carbon isfour, that of silicon is two.The available evidence is, however,not sufficient in amount to enable such a decision to be arrived atwith certainty, but it must be concluded that the quantitativeevidence, just i ~ s in the case of the humite series, points to thevalue of 2 rather than 4 as representing the fundamental valencyof silicon.The table of differences quoted makes it clear that the passagefrom tribenzylcarbinol to triphenylcarbinol, differences 3 and 5, isaccompanied by a marked contraction of the structure in all threerectangular directions, but that this effect is much more markedin the direction of the z-dimension than in those of x and y. Fromdifferences 5 and 6, and 6 and 7, it is seen that the substitutionof the hydroxyl group in triphenylcarbinol by hydrogen or bromineaffects the crystal Structure almost entirely in the direction of thez-axis; this is precisely what takes place in the correspondingoperation of passing from tribenzylcarbinol to tribenzylsilicol, inwhich the valency volume was diminished by two units on the viewthat Si=2, The differences 5 and 8 show that the displacementof the hydroxyl group in triphenylcarbinol by carboxyl also producesa maximum effect in the z-dimension, although the dimension of xis also appreciably affected. The differences 6 aad 9 indicateclearly that, in the passage from triphenylmethane to its additioncompound with benzene, the dimensions of LG and y are increasedto an equal and small extent, whilst the main change in dimensionsfalls on the z-axis.It is proposed to extend the application of Tschermak’s rule tothe investigation of morphotropy in later communications.I desire to express my heartiest thanks to Prof. W. J. Pope,F.R.S., for having suggested this work, and for his kind help duringits elaboration.UNIVERSITY CHEMICAL LABORATORY,CAMBRIDQE