May 20, 1955
237 1
SPECTRUM AND STRUCTURE OF DISILOXANE
0.50
250
I
I
2
4
I
1
I
6
8
IO
-
0.25 200
/
ij
0 00
-
I
-025
d
I50
t'
0
F'
x 100
K-0.50
n
1:
0
f
?
-0 75
-
3 -
50
IO0
-I 25
>
-1555
40
45
50
55
60
65
log M,.
I " "
0
Fig. 1.-Log [7] e's. log A f g for sodium polyphosphate in: (1) 0.35 Jf S a B r ; ( 2 ) 0.415 Jf S a B r .
12
I/m,.
Fig. 2.-Dependence of the molecular expansion coeffiWhen theoretical expressions for p 2 2 and p 2 3 become available, equation 14 should become useful cient, a,of sodium polyphosphate (XaPP-7) on the sodium in relating molecular dimensions to charge and bromide molarity, m3. electrolyte concentration.46 volume of the hypothetical uncharged polymer, is In the meantime we may look upon equation 11 equal to -G.9 X 10-3.47 This result confirms the as an empirical equation which in form reproduces solvent incompatibility of associated (NaP03) the experimental data. The intercept, f, which groups which has previously been hypothesized to may be considered as a measure of the excluded account for differences in both intrinsic viscosity (46) T h e l a s t t e r m i n equation 14 can be obtained rigorously from and salting-out behavior of polyphosphates in the membrane equilibrium d a t a , if par is known. F r o m literature d a t a presence of different alkali metal ions.25 (H. S. H a r n e d a n d C. C. Crawford, THISJOURNAL, 69, 1903 (1937)), ( 2 + p ~ m s )is found t o be equal t o 1.85 over t h e range investigated here. T h e n , with t h e membrane equilibrium d a t a , t h e expression in 0.025)/ms] (pu/P%)) braces in equation 14 simplifies t o { [(i2 f o r our system. Using a realistic value for i, e.g., one estimated from electrophoresis results, one finds t h a t t h e t e r m inside t h e brackets is a n order of magnitude larger t h a n t h e total experimental value of t h e expression inside t h e braces. Thus p22IP2 m u s t be of opposite sign t o t h e first t e r m a n d also very large. Finally, t o account for t h e slope of t h e line i n Fig. 2 b y equation 14, i t is necessary t h a t p22/P* shnuld have a large componenent which is inversely proportional t o ms.
-
[COYTRIBUTIOY FROM T H E
+
(47) Such a large negative value so f a r h a s been reported only for neutralized polyacrylic acid i n solutions of calcium salts ( P , J. Flory a n d J. E. Osterheld, J. Phys. Ckem., 56, 653 (1954)), b u t not for.anionic polyelectrolytes in solutions of sodium salts where t h e intercept usually is close t o zero. T h i s also explains why t h e difficulty which we encountered in a t t e m p t i n g t o extrapolate t h e molecular dimensions t o infinite ionic strength has not been observed in other polyelectrolytesalt systems. ?rTEW
BRUNSWICK, N E W JERSEY
DEPARTMENT OF CHEMISTRY,
UNIVERSITY O F CALIFORNIA, BERKELEY]
The Spectrum and Structure of Disiloxanel BY R. F. CURL,J R . , AND ~ KENNETH S. PITZER RECEIVED SOVEMBER 25, 1957 The infrared absorption spectrum of disiloxane in argon and nitrogen matrices a t 20°K. was investigated. T h e region covered was from 4000 to 600 cm.-'. The band a t 764 cm.-' in the gas phase is found t o have two components in the matrix. The Raman band a t 1009 an.-'appears in the infrared spectrum of the matrix. A Si-0-Si bond angle of less than 180' is proposed t o explain these results. The Si-OSi bond angle is estimated t o be near 155'.
It is now well established that the typical Si-0160" rather Si bond angle is relatively large-near (1) T h i s research was assisted b y t h e American Petroleum Ins t i t u t e through Research Project 50. (2) National Science Foundation Fellow, 1951-1957.
than the range of 110" characteristic of marly systems. The evidence includes the dipole moment measurements on methylsiloxanes by Sauer and Mead3 and the crystal structure studies of crys(3) R. 0. Sauer a n d D. J. M e a d , THISJOURNAL, 66, 1794 (1946).
R.F. CTJRI,,JR.,
3372
AND
tobalite by Nieuwenkamp.4 These investigations obtained the values 160 + 15" and 150", respectively. Also the study of octamethylspiro(S.5)pentasiloxane by Koth and Harkerj indicates that the unstrained Si-0-Si angle must be equal to or greater than the 130" value found in the planar rings of this compound. While these values arc all substantially less than 180",Lord, Robinson and Schumb6 conclude that their spectral studies of disiloxane are compatible with a strictly linear Si-0-Si structure and they advance qualitative arguments that the deviation from linearity, if any, must be small. In contrast, Emeleus, MacDiarmid arid Maddock7 concluded from the absence of resolvable structure in the 957 cm-' band of disiloxane that the molecule was significantly bent. There is no significant disagreement between the spectra found by the two groups of investigators, and indeed our results agree also ; nevertheless, the conclusions disagree sharply. We have sought to resolve this question by measurements of the spectrum of disiloxane in inert matrices a t 20"K., where the bands are much sharper. The results indicate a bent Si-0-Si structure.
Experimental A sample of disiloxane was prepared by the method used by Lord, el al.,@except that SiHaBr rather than SiH3C1 was the intermediate. The gas phase infrared spectrum of the sample in the sodium chloride region agreed with the gas phase spectrum given by Lord, et aZ.6 The melting point of the sample was -144'. The melting point reported by Stock, Somieski and Ll-intgen* is - 144.1'. T h e matrix isolation technique employed was developed in this Laboratory by Pimentel and Becker.g A gaseous mixture of disiloxane and argon was prepared with a mole ratio of 1 to 530, respectively. One hundredth of a mole of the mixture was sprayed onto a CsBr window maintained a t 20°K. by liquid hydrogen. The spray-on time was 30 minutes. I n a similar experiment with a nitrogen matrix, the ratio of moles of nitrogen to moles of disiloxane was 990; the amount of 0.01 mole; and the time of spray on was 1 hr. and 14 minutes. The matrix spectra are shown in Fig. 1. The inconsistency between extinction coefficients, say of the 1100 cm.-l band, calculated from the two
O l L
!
!
22110 2003
1
I800
I
I
I600
1400
1200
I000
800
C i h l
Fig. 1. ._
( 4 ) W.Xieuwenkamp. Z . K r t s i . , 9 2 , 82 ( 1 9 3 5 ) ; 96, 454 (19.48). ( 5 ) W.L. R o t h a n d D. Harker, . 4 h C v y s t , 1, 34 (1948). (6) R. C. Lord, D . W. Robinson and U'. C. Schumb, THISJOURIAL, 78, 1327 (1956). (7) H. J. Emeleus, A . G. MacDiarmid a n d A . G. Maddock, J . I n o r g . N7rrl. C h e m . , 1 , 194 ( 1 9 5 5 ) . (8) A. Stock, C. Somieski a n d R. Wintgen, B e r . , 60, 1754 (1017). (9) E. Becker a n d G. C. Pimentel. J . Chrm. P h r r . , 26, 224 (1956).
KEKXGTII S.PITZER
VOl. 80
spectra probably is due to a change of orientation of the spray-ciri tip with consequent different distributions of sample thickness. The frequencies of the bands observe(l in the gas phase and in both matrices are listed in Table I.
TABLE I INFRARED SPECTRA OF DISILOXANE (.lbsorptioii maxima, cm .-I) Gas phase
.. 764 957
.. 1107 1220 1700 2180
a
A Matrix
714 (?) 745 759 943 964 1011 1119
..
Nt
Matrix
Intensity
..
vvw vs vs vvs vvs
i46 762 947
95s 1007 1104
..
1711 .. 2213 2211 .. 2338 (?) .. There is a Ranian band a t 1009.
W
vs
w (2) v\v VS
vvw
An inconclusive effort was made to find the low frequency Si-0-Si bending mode. If the band is above 165 cm.-', it must be very weak, since no distinct absorption band was observed with a 70 cm. cell a t 41 mm. pressure. Our search at longer wave lengths (down t o 40 cm.-l) was less complete but no positive evidence was obtained. Some absorptions were noted a t very low frequency, but we felt they were too narrow to be the Si-0-Si bending motion. T h e search a t longer wave lengths was made possible by the very kind cooperation of Mr. Robert Ohlmann of the Physics Department. He carried out the search o n a grating instrument which he has constructed.
Discussion It is quite clear that most features of the disiloxane spectrum may be explained on the basis of a linear, Dsd, model. However, most of the features of the spectrum of dimethyl etheri0 also can be explained on such a basis even though the C-0-C bond angle is 111'. The two spectra are very similar in the small number of strong bands-a feature which indicates high symmetry. Thus, if the Si-0-Si bond angle is near 150" as indicated in other studies, the spectrum of disiloxane can be expected to differ only very slightly from that for the linear model. Two features appeared in the matrix spectra of disiloxane which conflict with the linear model. The band a t 761 cm.-l in the gas phase is assigned to the E, degenerate SiH3 rocking motion. The narrowing of the bands in the matrix shows this to be two separate bands separated by about 10 cm.-'. One is to be assigned to the in-plane motion and the other to the out-of-plane motion where the plane is defined by the non-linear Si0-Si structure. The other feature is the appearance of the weak band a t 1010 cin.-'. This is the totally symmetric SiHa deformation frequency which appears strongly in the Raman spectrum. In the gas phase infrared spectrum the very strong 957 cm.-' band is broad enough to conceal a weak band a t 1010 cm.-l. However, the matrix spectra show it unmistakably. Since gas phase selection rules do not hold rigorously in condensed phases, the appearance of this band is contributory evidence rather than proof that the Si-0-Si linkage is nonlinear. (IO) B. I,. Crawford and L. Joyce, ibid.,7 , 307 (1939).
SPECTRUM AND STRUCTURE OF DISILOXANE
May 20, 1958
An estimate of the probable minimum bond angle may be obtained by comparing the intensity of the 1010 cm.-' band to that of the out-of-phase symmetrical SiH3deformation. For if one assumes that the dipoles generated a t each SiH3 group have the same magnitude in the in-phase and out-ofphase modes, then
2373
considered. The effect of these off -diagonal elements in the G matrix is calculated by use of secondorder perturbation. There are also first order perturbation elements in both the F and G matrices which can be included. The H-H repulsion between silyl groups is a first-order perturbation in the F matrix. It is estimated by using the potential function developed by Hirschfelder and Lin= cot2 nettll for the triplet state of hydrogen. This is divided by two as required by valence bond theory. where Also the effect of any mixing of the SiH3 wag with 01 = the Si-0-Si bond angle the E SiH3deformation is neglected. li = intensity of the in-phase deformation The bond angle calculated from the splitting in 10 = intensity of the out-of-phase deformation this manner is 157" in the case of argon and 150" in Both the out-of-phase symmetrical deformation and the out-of-phase unsymmetrical deformation the case of nitrogen. Thus we have a reversal bewere assigned by Lord, et al., to the gas phase band tween the estimates based upon the argon and I n the matrix this band is found to nitrogen matrix data from this splitting as coma t 957 cm.-'. be split into a doublet. Which component of this pared to the estimate of the bond angle obtained doublet should be identified with the out-of-phase from the intensity of the 1009 cm.-' band. The symmetrical deformation has not been deter- estimate of the bond angle obtained from the splitmined. Therefore, in order to estimate the mini- ting of the 764 cm.-l band can be called neither an mum bond angle the less intense component (964 upper or lower limit to the bond angle since the cm.-' in argon) was chosen. The estimate of the effects of the neglected force constants depend minimum bond angle obtained is 152" in argon, upon their sign. Nevertheless, the various esti158" in nitrogen. The peak intensity was used in mates fall in the range 150-158". These results apply directly to the disiloxane the estimate rather than the integrated intensity. This, together with other approximations already molecule in the matrix whereas other data refer to the gas molecule. One may ask whether the mentioned, limits the accuracy of these values. It should be noted that the band a t 1220 cm.-l matrix bends the molecule. If the matrix is crysin the gas phase is not observed in the matrix spec- talline, with cubic closest packing, linear cavities trum. I n our opinion no really satisfactory ex- appear to be easiest to form. Thus a straightening planation of this band has been found. We, of the molecule seems more likely than a bending. If the matrix is glassy, broadening of the bands and therefore, shall not discuss its disappearance. It is possible to estimate the Si-OSi bond angle not splitting is expected in analogy to liquid solufrom the splitting of the 764 cm.-' SiH3 wagging tion. Thus we believe there is evidence for a nonfrequency if certain simplifying assumptions are linear Si-0-Si structure in the free molecules ,aswell as in the matrices. made. The matrix isolation method promises to be a When the molecule is linear the F and G (povaluable tool for determination of vibrational astential energy and reciprocal kinetic energy) matrices may be factored into several blocks corre- signments. The experiment described here is one sponding to the representations of the D a d group. of the first applications of the matrix isolation The two components of the SiH3 E, wag are in two method for this purpose. At the dilution emseparate identical blocks. As the molecule is bent ployed interactions between the solute molecules off-diagonal elements appear in both the F and G should be quite small, and i t is hoped that the matrices which connect certain blocks. There solute molecules behave as a non-rotating gas a t still remain, however, two separate blocks by sym- liquid hydrogen temperature in so far as their inmetry, of which one contains one component of the frared spectrum is concerned. Thus vibrational '764 cm.-' band and the other block contains the bands are very sharp and overlapping bands may other component. It is assumed that the off- be resolved where they could not be in the gas diagonal elements which might appear in the F phase. Nevertheless the possible effects of solutematrix when the molecule is bent are zero since the matrix interactions must be kept in mind. forces which would yield such elements are ex- BERKELEY4, C.4LIF. pected to be very small. Some of the correspond(11) J. 0. Hirschfelder and J. W. Linnett, J . Chem. Phys. 18, 130 ing elements in the G matrix are non-zero and are (1 95 I).
2
(;)
