62
RALPHSPITZER, W. J. HOWELL, JR.,
[CONTRIBUTION FROM
THE
AXD
VERNERSCHOMAKER
Vol. 64
GATESAND CRELLIN LABORATORIES OF CHEMISTRY. CALIFORNIA INSTITUTE OF TECHNOLOGY, No. 8451
An Electron Diffraction Investigation of the Molecular Structures of Silicon Tetrabromide, Tribromosilane, and Dibromodifluorosilane BY RALPHSPITZER, W. J. HOWELL, JR.,
AND
VERNERSCHOMAKER
Introduction.-The bond lengths in the halides of the fourth, fifth, and sixth group elements are in general considerably shorter than the sums of the corresponding Pauling-Huggins covalent radii.l This discrepancy appears to be partially in which account is taken of the variation with s resolved by the introduction of a correction for of the relative scattering powers of the various the partial ionic character of the bonds?; how- atoms and of the effect of the harmonic vibrations ever, large disparities remain for the second row of the molecule. Z j is the atomic number of fluorides, while for the heavier halides the atom i; f i , the atom form factor5; lij, the interagreement is not precise. For the bromosilanes atomic distance; aijr one-half the mean square considered in this paper, the Si-Br distances are amplitude of vibration of atom i against atom j ; found to be 2.16 * 0.03 k., 0.06 8. shorter and s = &n/X sin cp/2 where X is the electron wave than the distance calculated by Schomaker and length and cp is the scattering angle. Since the Stevenson.2 The bond-angle values reported calculation of aij from spectroscopic data is not here are all approximately tetrahedral with feasible a t present for polyatomic molecules havthe exception of that for F-Si-F in SiFaBr2, ing large moments of inertia, an attempt was which seems to be decidedly less than tetrahedral. made to determine these temperature-factor conThis value may be in error, however, as noted stants by including them as parameters in the correlation treatment. asi- Br was arbitrarily below. set equal to zero, so that the quantities actually We have found approximate values for the ___ are - asi-Br; for brevity we call relative amplitudes of vibration &,* - ~ Z S ~ B ~ determined ? in silicon tetrabromide and tribromosilane which these quantities ai,. The values reported refer to should be helpful in connection with the spectro- the effective temperature of the scattering gas, scopic data for the consideration of the normal about 23" for SiHBra and SiF2C12, and about 100" modes of vibration and vibrational potential for SiBr4. The radial distribution formula used was functions of these molecules. D(1) = c, sin1 Experimental.-The apparatus and technique n SJ employed have been described by Brockway.3 The wave length of the electrons was determined with C, = Insne-asPn from transmission photographs of gold foil (a0 = and 4.070 k ) ,and was about 0.06 A. e-a2rua* = 0.1 We are indebted to Professor W. C. Schumb of the Massachusetts Institute of Technology for where I,z is a relative intensity, estimated for the supplying us with the samples used. various peaks so as to show no average dependInterpretation- In interpreting the plloto- ence on s, and,,,s is for the last observable feagraphs the correlation method was used in COIF ture oii the photograph. junction with the radial distribution method Silicon Tetrabromide.-The two parameters described by Schomaker. The simplified theo- to be determined were I, the Si-Br distance, and retical intensity curves for the various models u, the temperature-factor constant for Br-Br were calculated according to the formula vibration. The molecule was assumed to have the symmetry of a regular tetrahedron. ( 1 ) 1,. Pauling a n d .hl. I,. IIuggins. %. K I ~ (2) V. S:chs,m:tker a n d i 7 ? St~;zn.ou. The photographs showed a rather heavy back(1941). ground which tended to make the measurements ( 3 ) 1,. 0 . Urockwa::. K L ~ ,\ l u J . PI (4) V , Schomak~1- The-,i2, C.aliforuia Institute of Technology. less precise than usual. The difficulty of meas-
c
I meeting. Baltimore, R l d . , April, 1938; .lmerii.ari < ~ l i c i n i c : ~Society 1939.
7 1 I, I ' . i u l ~ ~ i g and J Sherman. Z K v t \ t . , 81, 1
(1933).
MOLECULAR STRUCTURES OF SOME SILICON POLYHALIDES
Jan., 1942
63
urement was enhanced by the slight, but deceptive, asymmetry of some of the peaks. In all, twenty-six features were measured and used in the radial distribution curve; of these, the first five and last six were rejected in the quantitative comparison with theoretical intensity functions since it was felt that these measurements were less reliable than the others. The observed values of s are given in Table I, together with values of I,, TABLE I Max. Min.
I,
-
12
3 2 2 7 -10 5 -5 4 -10 10 -7 4 -3 6 -10 10 -3 4 7 10 -10 5 1
13
-
1 1 2 2 3
3 4 4 5
5 6 6 7 7
8 8 9 9 10
10 11
11
-
-
12 13
5 6 9
Cn 2 2 -2 10 -17 10 -10 9 -23 24 -17 10 7 14 -22 21 6 7 -11 15 -13
-
-
-
6 1
5 5 7
SO
1.54 2.39 3.25 4.11 5.00 5.99 6.78 7.63 8.53 9.45 10.37 11 19 12.08 12.71 13.84 14.68 15.65 16.49 17.26 18.23 19.41 20.19 21.05 21.60 22.85 23.71 Average Average JSi-B?
sce.lod.
4.94 5.91 6.75 7.54 8.42 9.42 10.39 11.31 11.95 12.89 13.90 14.87 15.76 16.41 17.34 18.35 19.44 20.24 21.05 21.70 22.77 23.97
s/so
(0.988) ,987 ,996 ,988 ,987 .997 1.002
1.011 0.989 1.014 1.005 1.014 1.007 0.995 1.005 1.007 (1.002)’’ (1.002) (1,000) (1.005) (0.996) (1.011) 1.000 deviation 0.008 2.145 A.
Ratios enclosed in parentheses were not used in the quantitative comparison. a
and C,, used in the radial distribution function, and values of &cd. SO. The curves were calcuequal to 2.145 A. lated for The curves D, E and F (Fig. 1) correspond to a = 0, 0.001 and 0.003, respectively. None of the curves shows marked disagreement with the photographs; the best correlation would be attained with some curve between 13 and C. This fixes the value of a between 0.001 and 0.003. Figure la shows the radial distribution curve for silicon tetrabromide. The average value for the Si-Br distance obtained from fifteen features by the correlation method was 2.145 A. with an average deviation
10
15
20
S.
Fig. 1.-Radial distribution functions: simplified theoretical intensity curvesfor SiBr4.
of 0.9 per cent. This gives 3.50 A.for the Br-Br distance. The radial distribution function (Fig. 1) gives l
~
i
= - 2.16 ~ ~
A., ZB+B=
= 3.51
8.
Since the Br-Br distance is the more reliable, we base our value of I on it. The estimated best values are
* 0.03 A. l ~ i - ~2.15 ~ * 0.02 A. a = 0.002
~ B ~ - =B 3.51 ~
Tribromosilane.-Thirty features were measured, of which twenty were used for the quantitative comparison with the theoretical curves. The photographs were better than those for silicon tetrabromide. The values assumed for the BrSi-Br angle 6 were 113” (curve A), 111’ (B), 109.28’ (C), and 107” (D). The Si-Br distance was set a t 2.15K. in the calculations. The curves in Fig. 2 are divided into three g r o u p s w e s A, B, C, D, with a = 0, A’, B’, C’ and D’ with a = 0.001 and A”, B”, C”, D” with a = 0.003. In this molecule, as in silicon tetrabromide, it may be seen that the effect of changing the tem-
RALPHSPITZER, W. J. HOWELL,JR.,
64
AND
1
1
I
10
20
1
30
s.
Fig. 2.--Simplified
theoretical intensity curves for SiHBra.
VERNERSCHOMAKER
Vol. 64
depth, maximum 6 too low compared with 7, maxima 8 and 9 too nearly equal, and minimum 12 too deep compared with 11 and 13. The curves C for 9 = 109"28', and B for 29 = 111 are almost identical. The only respect in which curves C disagree with the photographs is that they show minimum 8 slightly shallower than 6. Table I1 gives sobs./s&d. for models B and C. In both cases, the Br-Br distance agrees well with the radial distribution value of 3.55 A., but the tetrahedral model gives a long value for the Si-Br distance-2.18 A.compared with the radial distribution value of 2.16 A. The peaks of the radial distribution curve lead to the value 1101/20 for 6. For this reason, as well as the slight qualitative superiority of C, we choose our final model closer t o C than B, but assign limits of error which The best values are
perature factor is quite distinct from the effect of include both. changing the model. This is not true in general; lsi--~*= 2.16 * 0.03 A. for example, in difluorodibromosilane it is difficult LBrSi-Br = 1101/2 * l1/i0 IBr-Br 3.55 * 0.02 to decide on a model because the application of the aBr-Rr 0.001 temperature factor produces changes in the curves Difluorodibromosilane.-The photographs that cannot be distinguished from changes brought showed eleven measurable rings. The pattern about by varying the structural parameters is well represented by curve J (Fig. 3) except for slightly. peaks 3-4 and 9. The exact nature of these feaThe singly primed curves show the best agreement with the photographs with respect to the ef- tures is doubtful. The maximum labeled 3-4 is fect of the temperature factor, in that the doubly quite broad and appeared a t first glance to be primed group loses too much detail beyond s = rather more flat-topped than the figure indicates. 20, and the unprimed group shows many features For this reason two terms, numbers 3 and 4, were too sharply. This indicates that a = 0.001 is put into the radial distribution function to repreapproximately correct. The singly primed curves sent the extra width. The appearance of the are not shown below s = 14 because small M e r - ninth maximum and adjacent minima was also ences in the temperature factor have virtually no uncertain; hence s values for these features were effect in this region. The curve C' represents the not used in the quantitative comparison. Fourteen theoretical curves (Fig. 3) were calappearance of the photographs quite adequately. The curves D, D' and D" for 8 = 107" can be culated for symmetrical (C,,) tetrahedral models eliminated because they show minimum 9 deeper with Zsii-Br = 2.16 A.,IF+ ranging from 3.01 to than 10, maximum 11 higher than 12, and mini- 3.11 A.and lBr-Brfrom 3.49-3.61 A. as suggested mum 6 deeper than 8. These difficulties persist by the radial distribution function peaks (Fig. IC) for all values of a. at 2.15,3.07 and 3.56 A. Because the appearance The curves A, for 0 = 118O, are unsatisfactory of the theoretical curves is insensitive to variabecause they show minima 6 and 7 with the same tions in the Si-F distance, and the radial distribuE
MOLECULAR STRUCTURES OF SOMESILICON POLYHALIDES
Jan., 1942
65
TABLE I1 Max.
Min.
1,
C"
1
- 3 4
2
- 6
3
- 10
- 1 2 - 3 7 - 9 6 - 5 5 - 12 14 10 6 - 5 10 - 15 14 - 3 3 - 13 12 - 9 4 - 1 5 - 7 4
1
2
10
3 4 4 5
5 6 6 7
7 8 8 9 9
10 10 11
11 12 12 13 13 14 14 15 15
6 - 4 4 - 9 10 - 7 4 - 3 7 10 10 - 7 2 - 10 10 - 8 4 - 1 6 - 9 5 - 1 6 10 7
-
-
-
- 1 4
- 6 3
so
1.64 2.35 3.06 3.97 4.92 5.97 6.68 7.42 8.45 9.37 10.31 11.18 11.88 12.77 13.83 14.80 15.58 16.25 17.24 18.19 19.25 20.00 20.70 21.43 22.59 23.44 24.05 24.57 25.55 27.01
J
c"
5.91 6.78 7.52 8.38 9.47 10.45 11.33 11.92 12.85 13.89 14.92 15.86 16.51 17.31 18.38 19.42 20.39 21.04 21.84 22.85 23.87 24.89 25.65 26.21 27.25 Average Av. dev. h-Br
E B ~ B ~ a
Br-Si-Br = 109'28'.
sdso
(0.990) 1.015 1.013 0.992 1.011 1.014 1.013 1.003 1.006 1.004 1.008 1.026 1.016 1.004 1.011 1.009 1.020 1.017 1.019 1.012 1.019 (1.035) (1.043) (1,027) (1.008) 1.012 0.006 2.175 3.55
SBb
5.93 6.70 7.58 8.31 9.36 10.38 11.18 11.72 12.75 13.79 14.83 15.55 16.31 17.15 18.23 19.22 20.09 20.69 21.63 22.69 23.75 24.61 25.08 25.98 (27.19)
SB/SQ
(0.993) 1.003 0.999 .983 .999 1.007 1.000 0.987 .998 .997 1.002 0.998 1.004 0.995 1.002 0.998 1.005 1.000 1.009 1.005 1.013 (1.023) (1.021) (1.017) (1 ,007) 1.000 0.005 2.15 3.54
Br-Si-Br = 111'.
tion function does not show a corresponding, well resolved peak, the value 1.545 A. was assumed, in agreement with the Si-F distance found in silicon tetrafluoride.6 This assumption may be partially justified by analogy with CFzClz and CF4,' in both of which the C-F distance is the same. It should be borne in mind that the reported values of the Br-Si-F and F S i - F angles are also essentially assumed values inasmuch as they depend on the assumed Si-F distance. The models investigated were systematized by locating them on a two-dimensional map of lF-Br against lBr-Br, All models lying outside a closed curve corresponding to the assigned limits of error are inacceptable, and, in general, become progressively worse as their distance from the selected points increases in any direction. Since it was impossible, as discussed above, to determine (6) L.0. Brockway, J . P h y i . Chew%.*,41,747(1937). (7) L. 0.Brockway and F. T. Wall, THISJOURNAL, 66,2372 (lY34).
the temperature factor experimentally, the value = 0.001, which is approximately equal to that obtained for the Br-Br vibration in SiBr4 and SiHBrs, was adopted here for both uBr-Br and aBr-F. si-^, u F - ~ , and uSivBr were all arbitrarily set a t zero. The curves disagreed with the photographs as follows: (a) Curves A, B and D show minimum 7 too deep, relative to minima 6 and 8. Curve D also reverses the relative depths of minima 3 and 5, and makes maximum 3 too high compared with 5. (b) In curves K and N, minimum 5 is deeper than 6, contrary to the appearance of the photographs. K also shows the fifth maximum higher than the second. C and G are also unsatisfactory in the latter respect. (c) L and H give maximum 3 too high relative to 5, and minimum 3 deeper than 5. None of the group E, F, I, J or M is definitely inacceptable, hence the assigned limit of error
uBr-Br
66
RALPH SPITZER,
w.J. BOWELL, JR.,
Vol. 64
AND VERNER SCHOMAKER
a 8
interactions, shows the effect of an infinite temperature effect on the Br-Br and Br-F vibrations. The selected best values are 2,q,-Br = 2.16 * 0.02 A. k , - u = 1.55 A. (assumed)
LBr-Si-Br = lIO"50' * 3" 3.56 * 0.05 A. LBr-Si-F = 111"20' * 3"
lRr--Br
0
E
E
G
H
I
J
K
ZF-B~ ~V-F
3.08 * 0.04b. LF-Si-F = 99' = 2.35 * 0.15A.
=
10"
Previous Work.-Wouters, De Hemptinne and Capron8have used the electron diffraction method to determine the structure of SiHBr3. They obtain the results 2.19
/s,-B~
=
ZBr-Br
= 3.63
* 0.05 b.
A.
LBr-Si-Br = 110"
These values for the Br-Br distance and the Si-Br distance are M significantly higher than ours although the bond angle values N agree well. We believe the present determination to be the more 0 reliable because Wouters and coworkers measured only the first I I I I I I five maxima, whereas we measured 5 15 2Z fifteen maxima and fifteen minima. s. Fig. 3.-Simplified theoretical intensity functions for SiFpBrz. The models are It has been our experience that determinations based on a few described by the following parameters: measurements extending to only i ~ r - ~ r IP-B~ 18 -r' L! Br-Si-Br f F-Si-& L F-Si-F relatively small values of s are A 3.53 8.05 2.07 114O10' 113 "55' 93 "50' likely to be inaccurate, particuB 3.83 3.09 2.23 114"lO' 112 92 "25' c 3.63 3.06 2.37 114'10' 110°20' 100"10' larly if the minima are neglected. D 3.59 3.12 8.13 112O10' 113"55' 87 '5' Moreover, it may be significant E 3.59 3.09 2.29 112"10' 112"O' 95"15' that the above authors have reF 3.59 3.06 2.41 112"lO' 110"20' 102 "40' ported several other interatomic G 3.59 3.02 2.55 1 1 ~ ~ 1 0 ' 108'10' 112 "5' distances which appear to be too H 3.55 3.12 2.19 110"lO' 113'55' 90"0' 1 3.55 3.09 2.33 llOolO' 112O 0 ' 97'40' high by about their assigned experiJ 3.55 3.06 2.45 110"10' 110"20' 104'50' mental error. Their determination K 3.55 3.02 2.59 110"10' 108O10' 114"0' of the size of the SiHC1s9molecule L 3.51 3.09 2.37 108"30' 112"O' 99 "50' gives a result 2 to 3% higher than M 3.51 3.06 2.48 108'30' 110"20' 106"45' that obtained by Brockway3 and N 3.51 3.02 2.61 108"30' 108°10' 115"30' by Pirenne.l0 For SiBrClagb De %-B~ = 2.16 A., ki-F = 1.547b. for all models. Hemp tinne and Wouters again reallows the range of parameters covered by these port 2.19 A.for the Si-Br distance, and give 2.05 A. The Of Scalcd.Sob.) which vary ( 8 ) Wouters, De Hetuptinne and Capron, Amr. POL. sci. Brurrlles. little from one curve to another, are given in 574, 26 (1937). (Y) (a) De Remptinnc a n d Wouteru, h-ulure, 1.38, 884 {l!+.'Hi); Table 111for curve J. ( b ) 139, Y28 (1937). Curve 0, which includes only Si-Br and Si-F (10) pirenut, J . ckpm.PI,^,., 7 , 144 ( ~ 9 3 ~ ) . 1.
OO'
TETRAHYDROXYQ~JINONE AND RHODIZONIC ACID
Jan., 1042 TABLE I11 Max. Min.
1 1 2 2 3 3 4 5 5 6 6 7 7 8 8 9 9 10 10
11 11
12
I,
- 5 6 - 8 10 - 7 5 5 -11
10 - 5 3 - 7 10 -10
6 - 2 2 -6 10 -8 8 -10
C*
-4 4 -6 9 8 7 7 -17 16 -8 4 -11 16 -16 9 -3 3 -8 12 -8 7 -7
-
[CONTRIBUTION FROM THE
12
so
1.61 2.49 3.20 4.21 5.14 6.23 7.08 7.89 9.03 10.10 10.91 11.74 12.78 13.74 14.90 15.33 16.72 17.12 18.18 19.67 21.25 22.71
SJ
SJ/SO
2.45 3.22 4.18 5.28 6.47"
(0,984) 1.005 0.993 1.027 (0.972)
7.87 9.07 10.19 10.90 11.75 12.75 13.84 14.86 15.74 16.71 17.27 18.13 19.62 21.21 22.61
,997 1.003 1,010 0.999 1.000 0.998 1.008 0.999 (1.027) (1.000) (1.009) 0.997 ,997 1.000 0.996
10
6
23.96 23.61 Average Av. dev. ki--Rr
e
67 (0.969) 1.002 0.003 2.16
See text for discussion.
for the Si-C1 distance, a value 2% higher than that (2.01 A.) which has been reported for the Si-Cl bond in other molecules. l1
Summary The molecular structures of SiBr4, SiHBrl and SiFzBrz have been studied by the electron diffraction method. In all of these molecules the Si-Br distance is 2.16 =t0.03 A., and the valence angles are but little distorted from the tetrahedral value. The relative temperature factors have been estimated for SiBrr and SiHBrs from the appearance of the electron diffraction photographs. (11) Lister and Sutton [Trans. Faraday SOL. 87, 393 (1941)l report from an electron diffraction study that the Si-Br distance in silicon tetrabromide is 2.14 * 0.02 A,, in substantial agreement with our value. PASADENA,
CALIFORNIA RECEIVEDSEPTEMBER 13, 1941
DEPARTMENT OF BIOLOGICAL CHEMISTRY, WASHINGTON ST.LOUIS,MISSOURI]
UNIVERSITY SCHOOL OF h'fEDICINE.
Preparation of Tetrahydroxyquinone and Rhodizonic Acid Salts from the Product of the Oxidation of Inositol with Nitric Acid* BY PAULW. PREISLER AND LOUISBERGER Some confusion exists in the literature concerning the identification of the salts of rhodizonic acid (dihydroxy-diquinone) and of tetrahydroxyquinone. The quantitative analytical procedures usually employed for this purpose do not distinguish with certainty between substances of such close elementary composition, especially when they are hydrated or cannot be properly dried.' These materials have now been conclusively identified by application of electrometric oxidationreduction potential measurements. Modification of the methods for the preparation of these substances from the products of the oxidation of inositol (cyclohexanehexol) by nitric acid have been devised and crystalline materials of high purity have been prepared. The quantities of rhodizonate or of tetrahydroxyquinone salt
* Presented before the American Chemical
Society at St. Louis,
have been found to be aiTected by the rate of the addition and the basic properties of the salt added to the oxidized inositol, and by the degree of agitation with air or oxygen of the mixture so formed. When potassium acetate is used, pure potassium rhodizonate is obtained, whereas with potassium carbonate mixtures with tetrahydroxyquinone, salts usually result. Salts of rhodizonic acid and of tetrahydroxyquinone are utilized as indicators in the volumetric determination of sulfates with barium salt solution.2 A comparison of the properties of the pure substances indicates that the color changes when used as indicators are probably due to salts of rhodizonic acid rather than of tetrahydroxyquinone. Preparation of Oxidation Products of Inositol.-To ensure uniform reproducible results the procedure is given in
Mo., April, 1941.
(1) E. Bartow and W. W. Walker, I n d . Eng. Chem., SO, 300 (1938); 0.Gelormini and N. E. Artz, THISJOURNAL, 62,2483 (1930); F . A. Hoglan and E. Bartow, ibid.. 64, 2397 (1940); B . Homolka, Bcr., 64, 1393 (1921); M.Maquenne, Conpl. rend., 104, 297 (1887); R.Nietzki and T. Benckiser, Bcr., 18, 409 (1885).
(2) G . Gutzeit, Helv. China. Acta, 12, 725 (1929); W.A. Peabody and R. S. Fisher, I n d . Eng. C h c n . , Anal. E d . , 10, 651 (1938), W. C. Schroeder, I n d . Eng. Chsm., Anal. E d . , 5, 403 (1933); 8, 127 (1936); 9, 69 (1937); 10, 206 (1938); R. Strebinper and L. Zombory, 2. otaal. Chrm., 79, l(1929).
