Jan., 1961
FREEENERGY OF ADSORPTION ON SILICA
A , the specilic hydroxylated surface area (unesterified silanol groups) as measured by the adsorption of methyl red dye from benzene, was taken as a measure of the activity of the surface groups. Kf and K, are the specific velocity constants for the esterification and hydrolysis reactions, A,, is the specific silanol surface of unesterified silica and (ROH) and ( K O ) are the activities of butanol and water in the heterogeneous system. Using a graph (Fig. 2) showing degree of esterification and wa,ter concentration as a function of reaction time, a smooth curve was fitted to the data points (see Table V). The individual values of Kf, K r and Ke then were correlated by plotting log K vs. P I , and leastsquares lines were drawn to fit the data points and serve as best estimates of the kinetics and equilibria of the esterification reaction. Slopes of these lines were used to compute the heats of activation and of reaction. Correlation of the rate constants by means of a simplified Amhenius equation gave an energy of activation of 21.9 kcal. per mole of ester groups for the esterifica,tionreaction and 9.0 kcal. per mole for the reverse (hydrolysis) reaction. In a similar manner, a simplified van't Hoff isochore equation applied to the equilibrium constants indicated an endothermic heat of reaction of 12.1 kcal. per mole of ester group Esterilication of Thermally Dehydrated Silica.The silanol groups on the surface of amorphous silica can be paxtially dehydrated by heating the powder a t 150 to 500O.B If the silica is thermally dehydrated before reaction with 1-butanol, the esterification reaction proceeds more rapidly.*g The degree of esterification obtained by refluxing the silica with 1-butanol a t 118' depends on the temperature a t which the silica previously has been (28) See p. 234 of ref. 6. (29) M . T.Goebd, LT. S. Patent 2,736,669 ( E I. du Pont de Nemoum Q Co., Inc.. 1956).
25
dehydrated. A possible explanation for this effect is that the dehydrated silica surface is much more reactive with alcohol, acting perhaps as an acid anhydride. Although the surface actually cannot be completely dehydrated, extrapolation of the line in Fig. 3 suggests that a completely anhydrous surface, if it could be attained, would become covered with butoxy groups (2.7 per square millimicron). Active Si-OSi bonds on the silica surface (formed by thermal condensation of adjacent SiOH groups) thus probably react with alcohols as follows, the linkages from silicon to the underlying particle not being shown. TABLE V VARIATION IN REACTION RATEAND EQUILIBRIUM CONSTANTS WITH TEMPERATURE Run
1 2 3 4 5 6 7 8
Kf
OC.
(hr.-l)
Kr (hr.)-1 (wt. % Hz0) - 1
118 160 160 220 220 245 260 320
0.012 0.14 0.60 18.4 7.7 11.2 27.9 116
0.078 0.17 0.22 2.2 0.5 0.6 1.9 3.6
Temp.,
Si-0-Si
K.
,.
Kr/Kr
(wt. % HxO)
0.15 0.82 2.7 8.4 15.4 18.7 14.7 32
+ ROH +SiOR + SiOH
Simultaneously, some of the silanol groups react
+
+
SiOH ROH +SiOR H20 H20 2SiOH Si-0-Si
+
In practice, thermal dehydration of finely divided silica to give fewer surface SiOH groups than about 2.0 per square millimicron is not possible, since temperatures in excess of 650' are required. Sintering, with loss of surface area, occurs above this point. However, by first dehydrating silica a t 450-500°, in air, definitely hydrophobic products can be obtained simply by boiling the dehydrated powder in an alcohol such as 1-butanol, and drying.
FREE ENERGY OF ADSORPTION. I. THE INFLUENCE OF SUBSTRATE STRUCTURE IN THE Si02-H20,Si02-n-HEXANE AND Si02-CH30HSYSTEMS BY R. L. EVERY, W. H. WADEAND NORMAN HACKERMAN Department of Chemistry, The University of Texas, Austin, Texas Received May 8, 1880
The free energies of adsorption for these systems were calculated from adsorption isotherms at 25' using the Gibbs adsorption equation. In particular, the effects of particle size and of outgassing treatments were investigated. Of special interest was the finding of maxima in the free energy of adsorption of all three adsorbates in the middle range of particle sizes studied. Specific surface areas varied from 0.070 t o 188 m.*/g.
Introduction This work is an extension of that reported earlier on the effect of particle size of SiOeon calorimetrically determined heats of immersion in water.' A review of the literature reveals that although several workers have investigated surface thermodynamics, using a variety of adsorbents, few (1) W. H. Wade, R. L. Every and N. Hackerman, TAIE JOURNAL, 64, 355 (1960).
have tried to correlate adsorption properties to particle size. One group of investigators2 found that the heat of immersion of silica gels decreased with decrease in particle size and some previous work done in this indicated a varia(2) M. M. Egorov, K. G. Kraail'nikov and E. A. Syaoev, Doklady Akad. Nauk, S.S.S.R.,108, 103 (1966). (3) A. C. Makrides and N. Hackerman, THISJOURNAL, BS, 694 (1959).
Vol. 65
R. L. EVERY, W. H. WADEAND NORMAN HACKERMAN
26
TABLE Ia WATER P/PQ
Sample C Sample A5 Sample B OG 160' & Sample D Sample D Sample D Sample E OG 160° OG260° OG 160° 260° OG 115' OGb 160° OG 160'
V/Pc
V/P
VIP
1.575 1.299 1.083 1.358 1.126 0.854 .618 0.783 .10 .543 .20 .421 .465 .335 .30 ,374 .40 ,291 .339 .50 .264 .264 ,331 .60 .342 .70 .264 .398 .80 .272 .382 .531 .90 .681 .95 .472 ,815 .98 .571 a See Table I1 for specific surface 0.01 .025 .05
VIP
v/p
VIP
Sample F OG 160°
Sample G OG 115'
Sample G OG 260'
v/p
V/P
v/p
VIP
0.287 0.228 0.299 3.799 4.390 3.799 3.898 3.642 .197 .276 .260 2.835 3.091 2.835 2.677 1.378 .150 .256 .220 1.760 1.693 1.594 1.126 1.366 .173 ,130 1.181 ,201 1.122 0.902 1.146 0.957 .130 .114 0.776 .165 0.728 0.795 .665 .642 .lo6 .146 .114 .610 .571 .622 .539 .531 .lo2 .lo2 .516 .134 .496 .547 .472 ,484 .lo2 ,102 .469 ,134 .512 ,453 .453 ,437 .110 .lo6 .457 .142 .492 .441 .417 .441 .126 .114 .157 ,465 .457 .SO0 .457 .425 .146 .209 .150 .516 .591 ,512 .524 .449 .236 .236 .677 .315 .846 .697 .500 ,764 .307 .307 .866 .374 1.035 .866 ,563 ,906 .343 .343 1.032 .453 1.012 1.181 1.000 .614 Units = cc. a t S.T.P.per mS2per cm. area. b Outgassing temperature. 0
TABLEIb METHANOL Sample A OG 160'
P/PO
0.01 .025 .05 .10 .20 .30 .40 .50 .60 .70 .80 .90 .95 .98
Sample B OG 160"
Sample C OG 160"
Sample D OG 160'
Sample D OG 260'
Sample E OG 160'
Sample F OG 160'
Sample G OG 160'
v/p
v/p
v/p
v/p
VIP
VIP
v/p
v/p
0.685 .305 .172
0.515 .405 .255 .158 .091 .065 .054 ,046 .043 .040 .042 .049 .059 ,067
0.860 ,500 .297 .172 ,098 .070 .054 .047 .044 .043 ,044 .049 .053 .056
0.890 .480 .286 ,151 .os9 .065 .054 .046 .043 .043 .045 ,055 ,064 .072
0.300 .210 .156 ,103 .056 .041 ,035 ,032 ,030 ,030 ,035 .045 .053 .060
1* 100 0.825 ,470 ,204 .117 ,085 .067 ,058 .053 ,050 ,050 ,056 ,065 .075
0.425 ,305 .186 .lo9 ,061 .042 .036 .032 .030 ,030 ,032 ,038 ,045 .057
0.255 ,180 .122 ,077 .045 .035 .029 .025 ,023 ,022 .024 .031 ,041 ,050
.loo .055 .040 .033 .029 .027 .025 .025 .029 .032 ,035
TABLE IC TL-HEXANE PIP0
0.01 .025 .05 .10
.20 .30 .40 .50 .60 .70 .80 .90 .95 .98
Sample A OG 160'
Sample B OG 160'
Sample C OG 160 and 260"
Sample D OG 160'
Sample G OG 160'
Sample D OG 260'
Sample E OG 160'
Sample F OG 160'
VIP
v/p
0.0123 .0117 .0110 ,0103 ,0095 .0095 ,0096 ,0099 ,0105 ,0118 .0155 ,0257 .0362 .0420
0.0123 .0117 .0110 ,0103 .0095 ,0094 ,0094 ,0095 .0098 ,0105 .0120 .0163 .0200 ,0223
VIP
v/p
VIP
VIP
v/p
v/p
0.0320 .m40 .0173 .0115 .0077 ,0065 ,0060 ,0060 ,0060 .0063 ,0072 .0097 .0126 ,0157
0.0360 .0230 .0168 .0142 ,0123 ,0117 ,0115 ,0116 .0121 .0132 .0153 .0198 .0260 .0350
0.0355 .0301 ,0262 .0218 ,0179 ,0155 .0138 .0130 ,0126 ,0128 ,0142 ,0190 ,0271 ,0356
0.0480 .0180 ,0153 .0134 .0129 ,0128 ,0127 ,0127 .0127 .0131 .0161 ,0216 .0300 .0405
0.0480 .m45 .0214 .OB0 ,0150 .0136 ,0131 ,0130 ,0134 ,0140 ,0162 .0216 .0300 .0405
0.0405 .0305 ,0185 .0155 ,0138 .0134 .0133 ,0135 .0140 .0151 .0170 .0232 ,0300 ,0375
tion of the heats of immersion with specific surface on porous solids, no effect due to particle size was area. Zhdanov4 noted a change in position of the found.6 In the present study, an attempt was made adsorption isotherms (normalized to unit surface to characterize systematically the adsorbent bearea) of water on quartz of varying specific areas, havior of seven different, non-porous SiO, samples. but made no mention of surface thermodynamics. The adsorbates (water, n-hexane and methanol) In another study Of the free energy Of adsorption ( 5 ) R. G. Craig, J. J. Van Voorhis and F. E. Bartell, T H I s JOURNAL, (4) S. P. Zhdanov,
Pwc. Acad. Sci., USSR, 120, 277 (1958).
60, 1225 (1956).
FREEENERGY OF ADSORPTION ON SILICA
Jan., 1961
27
TABLE IIa Crystal structure
Sample
A" B
&Quartz 8-Quartz p-Cristobalite p-Quartz
Surface area
Outgas temp., O C .
m.Vg.0
----?r(erg/cm.*)~-0.9~0 O-1.0~0
Water-------ASS erg/cm.+C.
AH8
erg/cd.z
o
0.070
160 93.1 105.7 729.1 2.09 20.4 .138 160 118.7 127.8 546.5 1.41 16.8 C .54 160 & 260 158.8 173.9 531.1 1.20 12.5 Db .91 115 176.0 199.8 341.5 0.482 12.2 160 203.5 227.1 388.5 .542 11.0 260 206.4 233.7 485.8 .845 10.8 E p-QuWtz 8.12 160 191.8 215.1 334.5 .401 11.9 F Amorphous (7) 162 160 39.5 49.6 -29.2 - .264 36.5 Gc Amorphousd 188 115 32.8 40.9 43.5 .0087 51.6 260 25.8 33.9 42.9 .0030 55.5 a Supplied by Cleveland Quartz Works, General Electric Company. b Supplied by C. A. Wagner, Inc. c Supplied by Godfrey L. Cabot, Inc. (Cab-0-Sil). Determined by Godfrey L. Cabot, Inc. e Surface areas were measured by Kr adsorption with the exception of samples F and G which were measured by Dr. J. W. Whalen of the Magnolia Petroleum Company Field Research Laboratory using Nr adsorption.
Sample
A B C D
Surfaae area, m.a/g.
0.070 .138 .54 .91
E F G
8.12 162 188
Outgas temp., OC.
160 160 160 160 260 160 160 160
-
--?r(erg/cm.Z)-0-0.9PO
75.1 107.3 122.7 121.4 54.1 154.6 77.3 55.9
TABLE IIb Methanol0-1.0PO
79.7 115.1 129.7 130.3 67.8 163.2 83.8 61 4
were chosen t o provide a polar compound, a nonpolar material, and one which was intermediate in character. Experimental
c
o
(A?)
27.9 19.6 18.6 18.6 28.9 14.1 28.4 31.9
---x(erg/cm.z) 0 0.9PO
-
12.51 20.47 24.11 21.93 23.91 23.41 17.17 15.52
--
n-Hexane-
0
- 1.0PO
14.62 24.99 28.80 27.61 28.53 28.40 23.05 18.95
o
('4.2)
168.9 70.6 54.8 51.9 48.8 49.8 71.4 76.2
were controlled by water-baths to =kO.0lo. Room temperature was maintained approximately 2' above sample temperature t o avoid spurious condensation. Pressures were read from large bore mercury manometers with traveling telescopes of 0.01 mm. readability. Calculations .-The free energies of adsorption for the systems indicated were determined from application of the
Material.-The seven samples of Si02 investigated are listed in Table IIa.3 Samples B and E were obtained from sample A by sedimentation from water. Sample B was drawn off in less than one minute and sample E was that portion which remained in suspension longer than 24 hours. Gibbs equation to adsorption isotherms measured at 25O.'-lo Sample C was prepared from sample D by heating to 1510" Here R is the free energy of adsorption per cm.2, R is the gas followed by a ra id water quench. The resulting material constant, T is the temperature, r denotes the excess surface was found by Z r a y diffraction to be approximately 90% concentration in moles per cm.2, and P is pressure. Assumcristobalite and lo%,quartz. Sample F was prepared from ing ideality in the gas phase, r is quite simply related to the sample G by heating for 8 hours at 800" and subsequently volume of gas adsorbed per em.* a t S.T.P., V . R was exposing i t to saturated water vapor for two weeks. evaluated graphically from plots of V / P vs. P/P to provide Prior to adsorption measurements, the samples were the numerical solution to the integral. The isotherms were vacuum outgassed for 36 hours a t an elevated temperature. difficult to reproduce near saturation vapor pressure so that The usual temperature was 160°, selected as the best compro- the isotherms and their derived quantities carry some unmise between the maximum removal of the physically certainty in the extrapolation to PO. However, the adsorbed water and the minimum loss of surface hydroxyl isotherms were satisfactory up to about 0 . 9 P and these values of the integral from zero pressure to 90% saturation groups. The water used was doubly-distilled and after being were also shown in Table 11. The former extrapolation to attached to the vacuum system was repeatedly frozen with PO realistically limits the accuracy of the R values to f 5%. Dry Ice and degassed under a vacuum of about 10- mm. Integrated heats of adsorption A H . for the water-Si02 until free of dissolved gases. The n-hexane was obtained system, as shown in Table 11, were calculated from previfrom Phillips Petroleum Co. and analyzed by them to be ously published work1J in which the heats of immersion were 99.98yo pure. The methanol was obtained from Baker reported. The difference between the heat of immersion Chemical Co. and analyzed by them to be 99.9% pure. and the integral heat of adsorption with liquid water as the The n-hexane and methanol each were refluxed under reference state is just the surface enthalpy of water which is vacuum with CaH2 for a period of two hours to reduce the given as 118.5 erg/cm.2.11 From the experimentally deterwater content to a minimum. The materials were dis- mined quantities, ir and AH., the entropies of adsorption tilled into receiver bulbs with standard break-off seals and (Table 11) were calculated for this system using the two these bulbs were sealed and removed from the condenser dimensional thermodynamic analog while under vacuum. The bulbs then were attached to the ir = A H a - T A S a adsorption apparatus. They were repeatedly frozen with liquid nitrogen and degassed under a vacuum of about 5 X It should be noted that for a more exact treatment, the entropies of adsorption should be normalized to constant 10-6 mm until free of dissolved gases. Apparatus.-The volumetric adsorption apparatus for (7) J. W. Gibbs, "Collected Works," Longmans, Green and Co., water has been described previously.6 The same general design was used in the construction of two other volumetric New York, N. Y., 1928,p. 315. (8) L. E. Drain and J. A. Morrison, Trans. Faradag Soc., 48, 316 adsorption systems for hydrocarbon studies. These were equipped with mercury cutoffs so as to prevent contamina- (1952). (9) D.H. Bangham, i b i d . , 33, 805 (19371. tion by stopcock lubricant. The sample bulb temperatures
.
(6) N. Hackerman and A. C. HaU, THISJOURNAL, 62, 1212 (1958).
(10) D.H.Bangham and R. I. Razouk, ibid., 38, 1463 (1937). (11) W. D. Harkins and G. Jura, J . Am. Chem. SOC.,66,919 (1944).
R. 1,. EVERY, W. H. ?VADE A M I NORMAN HACKERMAN
28
260 240 220
--
~
--
200 -WATER 180 -// 160 -: / \ 8 140 //-+// g120 -./ ./,3 100 -- - I/ 80-1 r 60 -1
-.
>/---
4,
//
fi
METHSNOL,-Eb--,\\\
/'
/
-: - Y-7
20 -1
\\
'>\
\ \ \\
'j
\ I
\J
____----------__ ---4 "-HEXANE
,yCL--3-y
-4
\ \
'
40
\
1 0
Vol. 65
adsorbed water and surface hydroxyl groups a t the highest outgassing temperatures. Since hysteresis was observed, any siloxane bridges formed during outgassing must slowly rehydrate on exposure to water vapor as previously noted. Any methylation of the siloxane groups formed a t the higher outgassing temperatures must be a reversible process.1s This is suggested by absence of hysteresis a t low relative pressure and by a decrease in methanol adsorption with increased temperature. The latter behavior, unfortunately, is complicated by the loss of the remnants of physically adsorbed water which might promote methanol adsorption via enhanced hydrogen bonding. In contrast to the water behavior on crystalline
Jan., 1961
FREEENERGY OF ADSORPTION ON SILICA
pected a simple monotonic decrease in T as was observed for the heats of immersion, the results obtained over the particle size range would be quite unexpected. The observed T vs. area behavior can be understood only in the light of a previously advanced hypothesis, namely, that the large variation in immersional heats with particle size is due to a direct correlation between crystallinity and particle size. A qualitative explanation based on low specific area samples having crystalline surfaces and high specific area samples having amorphous surfaces has been offered, and in previous work the amorphous character of the SiOz surface has been discussed a t some length3. The entropy data obtained from the present investigation serve to demonstrate the gradual transition from a crystalline to an amorphous surface with decreasing particle size. Unless the adsorbed layer is completely mobile in character, adsorption on a periodic crystalline substrate will produce an entropy change more negative than that for adsorption on an aperiodic amorphous surface. If once again, one assumes a direct proportionality between specific area and crystalline-amorphous character of the surface, then Table IIa pirovides verification of this in the entropy data for the system SiOz-water, The comparable magnitudes of the AH, and TAS, contributions to the free energy explain the existence of the maximum in the water-Si02 system. Undoubtedly, the same interpretation should be taken for the methanolSiOz and the n-hexane-Si02 systems, although there are no experimental enthalpy data available at the present time. One would expect that as bonding forces become progressively weaker in going from water to methanol to n-hexane, the free energy would be less dependent on details of the surface structure, and it is noted that the maxima do become less pronounced. The only data in disagreement with this interpretation are for sample F-water system which gives a negative A&. Previous anomalous results have been obtained for this sample,l but all data taken have been consistently reproducible. Although the data for the p-cristobalite fit the T us. area curves, this may be fortuitous since the material was converted to 6-cristobalite from pquartz subsequent to any grinding operation. Since bulk recrystallization occurred, it is reason-
29
able that any surface amorphicity was simultaneously removed. Unfortunately, it was impossible to obtain the entire spectrum of particle sizes studied from a single sample so that the entire mode of surface generation could be laid to a single type of operation, Le., grinding. The three samples, A, B and E, which were obtained by a single grinding operation, are consistent with the remaining samples. I n the actual grinding operation, the largest particles will have had the shortest life. I n other words, the smaller the particle, the longer will have been its grinding history. This means that the smallest particles obtained have been subjected to both the highest pressures and highest local grinding temperatures. Here we have two compensating effects: (a) the crushing of the quartz during which the sample is plastically deformed and the Si+4ions are replaced in the surface by the polarizable 0-2 ions-an amorphous layer being produced’kand (b) an annealing process due to the high local temperatures which partially restores the crystal strucstructure.19 It is impossible from the present data to assign quantitative values for processes (a) and (b). Qualitatively, process (a) seems to overshadow (b) for the samples studied. It is interesting to note that w for the adsorbate molecules on the various samples (Table 11)parallel the free energies of adsorption, the minimum value of w occurring in the middle range of particle size studied, coinciding with the maximum value of T. Two independent reasons should be given for the low adsorbate density in the first layer for both the lowest and highest specific area samples. For the crystalline low area samples, the area is governed by periodicities of the lattice substrate, whereas, for the amorphous high area samples, the packing is governed by the concentration of sufficiently high energy adsorption sites. Acknowledgment.-This work is a contribution from the Department of Chemistry, The University of Texas. The authors wish to take this opportunity to express their appreciation to the American Petroleum Institute (Project 47D) for their continued support and interest in this project. (18) R. Gomer and C. S. Smith, “Structure and Properties of Solid Surfaces,” The Univ. of Chicago Press, Chicago, Ill., 1953, Chap. IV. (19) D.DEustachio and S. Greenwald, Phys. Rev., 69, 532 (1946).
