2372
MICHAEL L. HAIRAND WILLIAM HERTL
the liquid state, the ratio of Debye 0's for this compound was calculated using liquid volumes for CF4 and CzFs also. The second method lends added validity to the first as seen from the variance (at most, 2') between the two methods. The values of OE, calculated as the complement of the OD (best fit), for lattice rotational movement are given in column 3 and values of V ocalculated from eq 6 are given in column 4. The transition temperatures of the three compounds are given in the last column. The quantity Vo is a measure, a t the lowest temperatures, of the restrictive potential which hinders the molecule's free rotation in its lattice cavity. As previously explained, at higher temperatures when plastically crystalline behavior commences, these values of V o may not remain constant as for example in the case of CF,. The low-temperature values, however, are a measure of the relative ease with which the particular
molecule may begin, a t some higher temperature, a libration sufficient in magnitude to cause an anomaly in the observed heat capacity. As can be seen, the CgFe molecule experiences the highest potential barrier and will require the greater amount of thermal energy needed to cause the transition anomaly. The transition for the CJ?s molecule occurs slightly below that of the CzFB molecule because Vo is smaller, and finally the CF, molecule, having the lowest value of the restrictive potential, has also the lowest temperature of transition and exhibits a pretransition rise in the heat capacity,
Acknowledgments. Acknowledgment is made to the donors of the Petroleum Research Fund, administered by the American Chemical Society, for support of this research with Grant No. 1976-A5. The authors are also indebted to Dr. A. Plaush for the computer program used for the curve-fitting analysis.
Reactions of Chlorosilanes with Silica Surfaces
by Michael L. Hair and William Hertl Research & Development Laboratories, Corning Glass Works, Corning, New York 14880 (Received December 17,1068)
Reaction curves have been obtained spectroscopically for the reactions of (CH&SiCl, (CH&SiC12,(CH3)SiC13, and Sicla with the free hydroxyl groups on silica. Kinetic analysis of these curves shows that the monofunctional chlorosilane follows 1.0-order kinetics with respect to the number of surface bonding sites; the multifunctional silanes all follow 1.5 =k 0.2-order kinetics with respect to the number of surface bonding sites. The 1.5 overall reaction order is attributed to the presence of about 50% single OH groups on the surface and 50% geminal OH groups. The four chlorosilanes have an experimental activation energy of 22 & 2 kcal. A fast initial reaction, to the extent of 10-15%, takes place and is believed to be due in part to direct replacement of OH by C1. The validity of the kinetic analysis and confirmation of the replacement of OH by C1 reaction are shown by chlorine analyses of the treated silica. In the temperature range studied (200-400°) the reaction rates are independent of the pressure of chlorosilane. From high resolution spectra it is postulated that a band at 3747 cm-l is due to single OH groups, and that bands at 3751 and 3743 cm-l are due to geminal OH groups. The kiiietics of the chlorosilane reactions are compared and contrasted to the analogous methoxysilane reactions with silica.
Introduction The reactions between chlorosilanes and the surface of silica have been studied and investigated by many workers, primarily because of the utility of these reagents as coupling agents in polymer chemistry and surface deactivating agents in chromatography. Reaction with both freely vibrating and with adjacent Hbonded surface hydroxyl groups has been proposed,'P2 although spectroscopic evidence obtained in evacuated, water-free systems suggests that, a t temperatures below 500°, reaction is with the freely vibrating hydroxyl The Journal of Physical Chemistry
group only. Recent studies by Peris and Armistead and Hockey4indicate that the reaction of a silicon tetrachloride, trichloromethylsilane, or dimethyl dichlorosilane molecule with silica surfaces involves more than one silanol group, despite the fact that these OH (1) V. Ya. Davydov, A. V. Kiselev, and L. T. Zhuravlev, Trans. Faraday SOC.,60, 2254 (1964). (2) L. R. Snyder and J. W. Ward, J. Phys. Chem., 70, 3941 (1966). (3) J. B. Peri, ibid., 70, 2937 (1966). (4) C. G. Armistead and J. A. Hockey, Trans. Faraday Soc., 63, 2549 (1967).
2373
REACTIONS OF CHLOROSILANES WITH SILICASURFACES groups are sufficiently far apart that they do not hydrogen bond with each other. The following types of reaction can be envisioned
R +Si,-0-H
/
+ C1-Si-R
Experimental Section
--+
\
R R +Si,-0-Si-R
/ \
+ HC1
(1)
R
R >Si,-OH
/ + C1-Si-R + \ c1 R +Si,-0-Si-R
+Sis-O-H
+ \Si/ >Si,-O-H
/ + HC1 \ c1
(2)
R
C1
+
/\
C1
R
R
9Si,-0
+ Si,-0 OH
C1
R
OH
C1
R
geminal OH groups. A study of the kinetics of the interaction of mono-, di-, tri-, and tetrachlorosilanes with the freely vibrating surface silanol group has been undertaken to establish the mechanism of these interactions.
\/ Si /\
+ 2HCl
(3)
R
0
R
Silanes reacting via eq 1 and 2 react monofunctionally and should follow first-order kinetics with respect to the disappearance of the surface hydroxyl groups. If reactions 3 and/or 4 take place concurrently with reaction 2, then, if the energy of interaction of both hydroxyl groups is the same, one should observe a kinetic reaction order somewhere between 1.0 and 2.0, depending on the relative surface concentrations of single and
The apparatus and experimental procedures used have been described previously.6 The silica (Cab-OSill 150 m2/g, Cabot Co., Boston) was pressed into a disk. With the exception of the samples described in Table I11 and Figure 5, all silica samples were heated in air a t 800". This thermal pretreatment removes substantially all the molecular water (3500-cm-' band) and mutually hydrogen-bonded hydroxyl groups (3650cm-1 band) from the surface. The silica disk was then mounted in a cylindrical furnace fitted with IRtran end plates. The furnace was mounted in the spectrometer and connected to a vacuum rack. To carry out a reaction, the furnace was evacuated and heated to the desired temperature. After a spectrum of the silica had been taken, the gaseous reactant was introduced into the furnace from the vacuum rack and allowed to react for a given period of time. The furnace was then evacuated and a second spectrum taken. This procedure was repeated throughout the course of the reaction, allowing observation of the changes taking place on the surface of the silica. All spectra were taken with the silica a t reaction temperature. No changes were noted between the backgrounds of the hot silica and cold silica spectra (within 1%) in the region above 3000 cm-l so that any emission from the hot silica is not detectable in this region. Since the furnace is evacuated periodically and fresh reagent introduced, the gas-phase composition is essentially constant throughout the course of the reaction. By quantitatively following the changes in the intensity of the band due to the surface silanol group, a reaction curve is obtained. Eflect of Product HCl. I n previous work describing the kinetics of methoxysilane interactions with silica surfaces it was found that the methyl alcohol, produced by the reactions analogous to (1)-(4), reacted further with the surface, thus complicating the interpretation of the kinetics. Since the reaction discussed here produces HC1, it is important to know if HC1 interacts in any way with these silica surfaces. Previous work by Peria indicated that such reaction did not occur with his particular silica. I n order to confirm this observation, gaseous HC1 was allowed to stand in contact with both treated and untreated Cab-O-Si1 at reaction temperatures. I n no case was any significant amount (within 5%) of reaction with the surface observed to occur. Analysis of the Data. Reaction curves were obtained for the four chlorosilanes at various temperatures ( 5 ) W. Hertl, J . Phys. Chem., 72, 1248 (1968).
Volume 79, Number 7 July 1969
2374
MICHAELL. HAIRAND WILLIAMHERTL
and pressures of reactant. The reaction curves were analyzed by fitting the data to integrated rate equations of various orders, until a linear plot was obtained for at least 70% of the reaction. The form of the rate equation is
kt =
___
(m l 1){ (a
-
1 xp-1
.
' ' 1 7 5 % REACTION)
- L} (Up-1
where k is the rate constant, t is the time, rn is the order of the reaction, a is the initial concentration of reactant (free hydroxyl groups), and x is the amount of reactant consumed. Since the gas phase is kept constant during the course of the reaction, the quantities a and x refer to the concentration of surface sites. The quantity
-/
-I? 0.5
00
0
I
0 00
I
I
I
I
Results Figures 1, 2, 3, and 4 show the typical plots that were obtained. It is seen that the monochlorosilane follows 1 .O-order kinetics, whereas the dichloro, trichloro- and tetrachlorosilanes all follow approximately la6-1.7-order kinetics. I n Figure 5 some experimental data are plotted using 1.4-order kinetic plots, Deviations from the straight line are detectable at about 60-70% reaction. Kinetic plots at less than 1,4order show distinct curvature even earlier in the reaction, as do kinetic plots of greater than 1.7 order. Thus it can be concluded that these multifunctional silanes all follow about 1.5 0.2 order. Silica samples which were given a pretreatment of only 400" (Figure 5) did not give good plots.
*
t(rrec1
Figure 1. 1.0-Order kinetic plot for reaction of monochlorotrimethylsilane with free hydroxyl groups; P , 50 Torr; T,300". The Journal of Physical Chemistry
I
Ir
0
@ . I
I
10000
5000
I
15000
t (sec) Figure 3. 1.6-Order kinetic plot (as defined in text) for reaction of trichloromethylsilane with free hydroxyl groups; P , 50 Torr; T,300".
t(S0C)
Figure 4. 1.6-Order kinetic plot (as defined in text) for reaction of silicon tetrachloride with free hydroxyl groups; P , 50 Torr; T,300".
REACTIONS OF CHLOROSILANES WITH SILICA SURFACES
0.6
I-
I
I
0
1000
500
tlrecJ
Figure 5. 1.4-Order kinetic plot (as defined in text) for reaction of dichlorodimethylsilane with free hydroxyl groups on silica after various thermal pretreatments: 0, 950'; 0 BOO'; A, 600'; f, 400'.
This was due to the slow removal of hydrogen-bonded OH groups during the run, caused by the high reaction temperature and intermittent evacuation. All the kinetic plots show a positive intercept which correspond to a fast initial reaction of about 10-15%. The 1.0-order kinetic plots obtained with the monochlorosilane show that all of the free hydroxyl groups on the surface have about the same reactivity, and that these groups react independently of one another. This does not hold for the first 10-15ojO of the reaction, however, and this high initial reactivity will be discussed in detail below. The linearity of the kinetic plots is good evidence that this type of analysis is valid, and that it is justifiable t o interpret the order of the reaction as the average number of surface sites consumed when one silane molecule reacts. The di-, tri-, and tetrachlorosilanes all gave good 1.5 f 0.2-order plots. A 1.5-order reaction corresponds to 50% of the silane molecules reacting difunctionally and 50% reacting monofunctionally. Since the same stoichiometry is observed for all three polyfunctional silanes, this strongly suggests that this stoichiometry is determined by the nature of the surface hydroxyl groups. Peri and Hensley6have suggested that silica surfaces, when prepared at low temperature and fully hydroxylated, can be represented by the 100 face of cristobalite, in which each surface silicon atom holds t y o hydroxyls. Such a surface holds 7.9 hydroxyls/100 A2 and these are presumably hydrogen bonded. On dehydroxylation, it is suggested that two Si04tetrahedra are rotated to obviate abnormally long siloxane linkages. Computerized simulation of the random dehydroxylation of such a surface predicts that after drying at 800" there should be less than 1.2 OH/100 B2 and that all vicinal pairs should disappear; i.e., only geminal and
2375
single OH groups remain, both of which should be difficult to remove. Reaction with both SiC1, and AICls substantiated these figures. After drying at 800", samples holding 1.4-1.5 OH/100 ,k2 showed 63-71% of the hydroxyls to be paired. Accordingly, in our kinetic experiments, it would be expected that reaction would correspond to 63-71% of the silane molecules reacting difunctionally, i.e., 1.6-1.7 order. Within the experimental error, this agreement is good. Further confirmation of this stoichiometry was obtained by analysis of the residual chlorine. Weighed silica samples were allowed to react a t 400" with S i c 4 , MeSiCL, MezSiClz, and MesSiC1. When reaction was complete, the tube was evacuated at reaction temperature (400") to avoid any possibility of physically adsorbed silane remaining on the sample. The tube was then sealed under vacuum and then broken under acidic AgNOa. This procedure avoided any possibility of hydrolysis of the bonded C1 groups, due to the presence of atmospheric water vapor. (Samples which had been exposed to the atmosphere gave analytical results which were about 30% lower in chlorine content.) The results of the analyses are given in Table I. Table I: Residual Chlorine Content of Treated Silica and Calculated Reaction Stoichiometry
Wt % C1 found
Silane treatment
Monochlorotrimethyl (100% reaction) Monochlorotrimethyl (15% reaction) Dichlorodimethyl Trichloromethyl Untreated
Calcd stoichiometrya (% reacting monofunotionally)
0.3
.,.
0.12
...
1.5 3.1 0,001
70% 65%
...
a Stoichiometry calculated assuming 15% initial reaction as per eq -5.
It should be noted that the monofunctional silane shows a residual chlorine content not present in the original Cab-0-Sil. This is attributed in part to an initial fast reaction (eq 5 ) which is discussed later. +Si,-0-H
+ C1-Si--Me
=
+Si,-Cl
+ HO-Si-Mes
(5)
From these C1 analyses the reacted stoichiometry of the chlorosilanes with the surface can be calculated. The values are given in Table I. I n the calculations it is assumed that the first 15% of the reaction takes place via eq 5 and the remaining 85% via eq 1 to 4. For the latter 85% of the reaction, the analyses give a (6) J.
B.Peri and A. L. Hensley, J. Phys. Chem., 72, 2926 (1968). Volume 7% Number 7 July 1969
2376
MICHAELL. HAIRAND WILLIAMHERTL
value of 65-70% of the OH groups reacting monofunctionally, Le., 1,30-1.35 order, which agrees reasonably well with the kinetic result of 1.5 + 0.2 order. Armistead and Hockey4found that silica which had been pretreated at 500" and had reacted with methylchlorosilanes gave a reaction stoichiometry corresponding to -1.3-order reaction. For silica pretreated at 300" they obtained 1.0-order stoichiometry. Temperature Dependence. The temperature dependences of the reactions were measured by carrying out part of the reaction at 300", changing the temperature, and then carrying out the remaining portion of the reaction at the new temperature. A kinetic plot was made and the slope of the latter part of the plot was compared to the initial part of the plot. From this a relative rate constant, measured with respect to the datum temperature (300"), was obtained. This method obviates any differences in the surface reactivity of the silica. An Arrhenius plot of the relative rate constants is given in Figure 6. The experimental activation energies measured from the slopes obtained from each compound are given in Table 11. The activation energies are essentially the same, thus suggesting identical mechanisms.
Table I1 : Experimental Activation Energies of the Chlorosilane Reactions Eaet,
Silane
kcal
Monochlorotrimethyl Dichlorodimethyl Trichloromethyl Tetrachloro
22.0
22.6 21.0 22.0
Pressure Dependence. A number of runs were carried out at various temperatures in which the pressure of the silane in the furnace was changed part way through the reaction. The pressures were varied from 15 to 80 Torr. I n no case was any deviation from linearity noted in the kinetic plot. Thus, the rate of the reaction is completely independent of the pressure of the gaseous chlorosilane. The heats of adsorption and frequency shifts observed during the physical adsorption of these chlorosilanes have been described previously.' At the temperatures at which chemical bonding takes place with the silica surface (i.e., greater than about 150") the fraction of the hydroxyl groups which are covered with these physically adsorbed chlorosilane molecules is very small. At 150" it is just detectable, but at 200" it is not detectable and must be less than 1%. At the reaction temperatures used for most of the experiments (300-400"), p / p o for the chlorosilanes can be Although it is not estimated to be about 3 X usual to speak of physical adsorption at these temperaThe Journal of Physical Chemistry
t
*.O
2,ot
,
,
1.5
I.6
I
I.7
,
,
1.8
I .a
IOOOIT
Figure 6. Arrhenius plot of relative rate constants for reactions of chlorosilanes with free hydroxyl groups. The relative rate constant is the ratio of the slope of the kinetic plot at the given temperature to the slope of the kinetic plot a t the datum temperature (300'). W, (CH&SiCl; a, (CH&SiC12; 0, CH3SiCla; OISiClr.
tures, clearly some kind of surface complex must be formed at one stage of the reaction. If the reaction rate is a function of the surface coverage of this complex (as it is in the methoxysilane reactions with silica) then the degree of coverage must be almost independent of the ambient pressure of the chlorosilane, in the temperature and pressure ranges used here. Adsorption isotherms have been reported in the literature which do show pressure-independent portions well before saturation coverage is reached.8 The reaction with methoxysilanes proceeds at a satisfactory rate between 100 and 200", whereas with the chlorosilanes a temperature between 300 and 400" must be used to obtain a similar reaction rate. I n this respect the chlorosilane reactions are different from the methoxysilane reactions. I n the latter case, the reaction is strongly pressure dependent, and the physically adsorbed material can be observed spectroscopically as the reaction proceeds.6jD Effect of Hydrogen-Bonded Silanols. I n studying the kinetics of the methoxysilane-silica reactions it was noted that the rate of reaction was very strongly dependent on the number of nonreactive, hydrogenbonded hydroxyl groups present on the surface. A series of analogous runs was carried out with the chlorosilanes in which the concentration of hydrogen-bonded hydroxyl groups on the silica surface was varied by means of the thermal pretreatment. No great difference in reaction rates was noted between samples containing large amounts and samples containing virtually no hydrogen-bonded groups. Some rate (7) W. Hertl and M. L. Hair, J . Phys. Chem., 72, 4676 (1968). (8) See, for example, 8. J. Gregg end K. S. W. Sing, "Adsorption, Surface Area and Porosity," Academic Press, London, 1967,p 9,113. (9) W.Hertl, ref 5 and J. Phys. Chem., 72, 3993 (1968).
REACTIONS OF CHLOROSILANES WITH SILICA SURFACES constants obtained wit'h these different silica samples are given in Table 111. The variation in the surface reactivity varies only over a factor of about 2. This is to be compared to the variation of two orders of magnitude which was observed with the methoxysilanes. No direct correlation between the thermal pretreatment and the surface reactivity was noted. Table 111: 1.6-Order Rate Constants at 200' for Reaction of (CH3)2SiC12 with Silica Containing Different Amounts of H-Bonded Hydroxyls k
0.100 0.043 0.072 0.065 0.049
Thermal pretreatment
800°, 800°, 400°, 800°, 200 O
3 hr 3 hr 3 hr 3 hr
Several reactions were carried out for extended periods of time using silica samples which contained large amounts of hydrogen-bonded hydroxyl groups. For reaction times up to 30 hr no significant changes were noted in intensities of the bands due to these groups. (The reaction with the freely vibrating hydroxyl groups is virtually completed within about 6 hr.) A significant decrease in the concentration of the hydrogen-bonded hydroxyl groups was noted, however, for reaction times of 60 hr. Thus, although these groups do react with the chlorosilane, they do so at a rate which is an order of magnitude slower. Positive Intercepts, Free OH Groups. All the kinetic plots obtained with the chlorosilanes show a positive intercept, which corresponds to about 10-15% reaction (Figures 1-6). Methoxysilanes do not show this effect. Snyder and Ward2 have noted a high initial reactivity in reactions of chlorosilanes with silica, particularly samples with small pores. I n their case, the fast reaction was ascribed to some hydrogen-bonded silanol groups. Such groups are not present on our silica surfaces and thus this explanation is not applicable in our system. The presence of chloride on Cab-0-Si1 which had reacted with monochlorosilane suggests that the fast reaction could be of the type shown in eq 5, in which a surface hydroxyl group is simply replaced by a chlorine atom. Such substitution of surface hydroxyl groups by halides is well known,lO and an analogous reaction between silica and carbon tetrachloride is known to take place between 350 and 600°.3 In that case, gaseous COCL and HCl are the products. If the initial fast reaction is caused solely by the replacement of -OH by -C1, then a silica sample reacted only for the first 15% with monochlorosilane should contain the same amount Of a which has completely reacted. The data given in Table 1 show that a sub-
2377 stantial fraction of the chlorine does appear on the surface during the first 15% of reaction, and therefore suggests that reaction 5 does account for at least part of the initial fast reaction. I n order to further investigate this effect, the band ascribed to the freely vibrating hydroxyl group on silica was expanded and resolved on an electronic curve resolver (Figure 7), Three bands were observed a t 3751, 3747, and 3743 em-'. To determine if the initial fast reaction could be ascribed to one of these bands, a kinetic run was carried out using 1OX abscissa expansion on the spectrometer. This was sufficient to allow area measurements of the individual OH bands. The measured areas and peak extinctions are plotted in Figure 8 as a function of
t
FREQUENCY (ern-')
Figure 7. Expanded spectrum of band due to freely vibrating hydroxyl group. The upper curve is the observed spectrum; the lower curves are the resolved bands and are composed of symmetrical, Lorentzian shaped bands.
time. The changes in peak extinctions agreed well with the areal changes of the 3747-cm-1 band. However, the band at 3751 cm-' dropped initially in integrated intensity much faster than the 3747-cm-I band (cf. Figure 8). Thus, one cause for the initial fast reaction can be ascribed to the faster initial disappearance of the OH groups causing the 3751-cm-l band. However, even with the main band (3747 cm-l) there is an initially faster reaction. The reason for this is not entirely clear. Examination of the integrated intensities of the three resolved bands shows that the central band (3747 cm-l) accounts for about 60% of the total, and the low- and high-frequency bands about 20% each. It is tempting, therefore, to speculate that the central band is due to the single groups on the surface while (10) M, I,. Hair, "Infrared Spectroscopy in Surface Chemistry," Marcel Dekker, New York, N. Y., 1987.
Volume 73, Number 7
July 1969
G. AVITABILE, P. GANIS,AND V. PETRACCONE
2378
H
\
H 0
/
v Si
0
/\
0
ZOO0
4000
6000
8ooo
A large body of experimental evidence has now been accumulated showing that on a silica surface that has been pretreated at 500" or more, about half the surface hydroxyl groups are sufficiently close to interact bimolecularly with chlorosilanes, despite the fact that they do not hydrogen bond with each other. The chlorosilanes interact in a somewhat different manner to the methoxysilanes. Reaction in both cases is with the freely vibrating hydroxyl group, but whereas with the chlorosilanes the rate is not dependent upon the number of hydrogen-bonded groups, with the methoxysilanes the rate of reaction is slower as the number of hydrogen-bonded hydroxyls increases. With long reaction times the adjacent groups will partially interact with the chlorosilanes. Moreover, in the case of the chlorosilanes, there is an initial fast reaction and the reaction is independent of pressure.
10000
tfSCCI
Figure 8. Concentration-time curve for reaction of CHIISiCIJ with free hydroxyl groups. Concentrations (i.e., fraction remaining) were measured using: X, peak extinction; 0, integrated area under 3747-cm-1 band; 0, integrated area under 3751-cm-l band.
the other two bands are due to geminal groups. This, then, would account for the observed 1.4order kinetic reactions, i.e., 60% single groups and 40% geminal groups. The occurrence of the low- and high-frequency bands could be due to a removal of the degeneracy due to a slight coupling through the silicon atom.
Acknowledgment. The authors wish to thank Mr.
J. A. Gabel for assistance in the experimental work.
The Crystal and Molecular Structure of
1,3,5,7-Tetrarnethylcycloocta-cis,cis,cis,cis-1,3,S,l-tetraene by G. Avitabile, P. Ganis, and V. Petraccone I s t i l t h Chimico, Univeraitb d i Napoli, Naples, Italy
(Received December 18, 1908)
The crystal structure of the title compound hw been determined: space group I41/a, a = b = 7.53 i 0.03 , c = 18.54 f 0.10 A. The structural problem was solved by using the close-packingprinciple. The molecule has S4 symmetry. Carbon-carbon bond lengths in the ring are alternatively 1.33 and 1.48 A; valence angles are 122". These parameters are not significantly different from those of cyclooctatetraene and of most crystalline derivatives of cyclooctatetraene. The conformational parameters of the molecule are discussed in connection with its variable-tcmpcraturo pmr behavior.
Introduction Structural studies on a series of cyclic alkanes and alkenes are in progress in our laboratory with the aim of relating their structures to their physicochemical properties. The investigation of the crystal structure of 1,3,5,7t e t r a m e t h y 1cy cloo c t a-cis,cis,cis,cis,-l13,5,7-tetraene The Journal of Physical Chemistry
(henceforth called TAICOT), which is reported in this paper, was undertaken, in particular, for the following reasons. Many variously substituted cyclooctatetraenes have been studied by X-ray methods' but in most instances as ligands in transition metal complexes. It is (1) (a) P. Hoppe, Naturwissenschaften, 54, 67 (1967) ; (b) B. Dickens and W. N. Lipscomb, Inorg. Chem., 3, 1529 (1964).
