Interactions between Surface Hydroxyl Groups and Adsorbed Molecules.
I. The Thermodynamics of Benzene Adsorption by J. A. Cusumano and M. J. D. Low Department of Chemistry, New York University, New York, New York 10463 (Received January 20, 1,969)
Isotherms of benzene adsorption on porous glass were measured with a recording vacuum microbalance. Heats of adsorption and entropies derived from the data were compared with theoretical values computed for mobile and immobile adsorption. The thermodynamic considerations suggest a model for which benzene moJecules are localized and quite strongly bound at low degrees of coverage. The benzene molecules become mobile as the coverage is increased,the adsorbed benzene molecule lying flat with the plane of the ring parallel to the surface and rotating freely in the ring plane. Filling of the micropore structure begins at e = 0.6, increasing interaction between adsorbate molecules hindering some of the motion of adsorbed benzene.
Introduction
It is widely recognized that hydroxyl groups play an important role in determining the properties of silica surfaces, and consequently much work has been done to characterize the nature, number, and distribution of the Gas adsorption hydroxyl groups themselves.’ --3 studies have had a prominent role in such work, partly because in recent years it has been possible to obtain direct information not only about the amount and rate of gas take-up per se, but also about the influence of adsorbate and adsorbent upon each other. The adsorption of cyclic aromatics would appear to be of special interest for this purpose because the a-electron density, and hence the reactivity, of the adsorbate can be varied over an extended range. Using suitable techniques to measure the adsorbate-adsorbent interaction, the aromatic adsorbate could be taken as a variable probe of molecular dimensions. There have been a number of studies concerned with the adsorption of aromatics on pure silica^.^-'^ Infrared spectroscopicgJ’ or gravimetric6-*J2 methods were employed, but these two complementary techniques were rarely4s6used together. The adsorptions of various aromatics on porous glasses have also been examined1a-21 (much of the work was reviewed by Little’), but these studies were not at all extensive. Much of the infrared work was carried out in the overtone region and, as pointed out by Kiselev and LyginjZ2 consequently is of limited utility. Also, the modifying effects of boria and other contaminants on the propertiesZaof the porous glass mere not then known, and much of the work can be considered inadequate with respect to sample preparation and spectrometric techniques. Although the number of studies dealing with the adsorption of aromatics on siliceous surfaces is relatively large, additional information is desirable. We have therefore carried out detailed studies, using both The Journal of Physical Chemistry
gravimetric and infrared techniques, of the interactions of cyclic aromatics with highly degassed porous glass surfaces with the intent of providing qualitative and (1) L. H. Little, “Infrared Spectra of Adsorbed Species,” Academic Press, New York, N. Y., 1966. (2) E. A. Hauser, “Silica Science,” D. Van Nostrand and Co., Inc., Princeton, N. J., 1953. (3) W. Eitel, “The Physical Chemistry of Silicates,” University of Chicago Press, Chicago, Ill., 1954. (4) G. A. Galkin, A. V. Kiselev, and V. I. Lygin, Zh. Fiz. Khim., 41, 40 (1967). ( 5 ) V. Ye. Davydov, A. V. Kiselev, and B. V. Kuznetsov, R w s . J . Phys. Chem., 39, 1096 (1965). (6) J. A. Hockey and B. A. Pethica, Trans. Faraday Soc., 59, 2017 (1962). (7) L. H. Boulton, B. R. Clark, M. F. Coleman, and J. M. Thorp, ibid., 62,2928 (1966). (8) M. L. Hair and I. D. Chapman, J . Amer. Ceram. Soc., 49, 651 (1966). (9) A. Zecchina, C. Versino, A. Appiano, and G. Occhiena, J . Phys. Chem., 72,1471 (1968). (10) A. V. Kiselev and V. I. Lygin, Kolloid Zh., 23, 574 (1961). (11) G. A. Galkin, A. V. Kiselev, and V. I. Lygin, Russ. J . Phys. Chem., 36,951 (1962). (12) J. W. Whalen, J . Phys. Chem., 71,1557 (1967). (13) N. 6.Yaroslavsky, Zh. Fiz. Khim., 24, 68 (1950). (14) N. G. Yaroslavsky, “Methods of Investigating the Structure of Highly Dispersed and Porous Bodies,” USSR Academy of Science Press, 1953, (in Russian). (15) A. V. Kiselev and V. I. Lygin, Russ. Chem. Rev., 31, 175 (1962). (16) A. Ron, M. Folman, and 0. Schnepp, J. Chem. Phys., 36, 2449 (1962). (17) V. N. Abramov, A. V. Kiselev, and V. I. Lygin, Zh. F i z . Khim., 38, 1045 (1964). (18) V. N. Filimonov, Opt. Spectrosk., 1,490 (1956). (19) V. N. Filimonov and A. N. Terenin, Dokl. Akad. Nauk S S S R , 109,982 (1956). (20) A. N. Sidorov, Zh. Fiz. Khim., 30, 995 (1956). (21) A. N. Sidorov, Dokl. Akad. Nauk SSSR, 95,1235 (1954). (22) A. V. Kiselev and V. I. Lygin, Supplementary Chapter in ref 1, p 226. (23) M. J, D. Low and N. Ramasubramanian, J . Phys. Chem., 70, 2740 (1966) ; Preprints, Division of Petroleum Chemistry, 152nd Meeting of the American Chemical Society, New York, N. Y., Sept 1966, p 133.
INTERACTIONS BETWEEN SURFACE HYDROXYL GROUPS AND ADSORBED MOLECULES
793
quantitative information about the T-OH interaction. Specifically, part I of the present series of three papers is concerned with a detailed analysis of the thermodynamics of the adsorption of benzene on porous glass. Part I1 considers the same adsorbate-adsorbent system from the point of view of detailed infrared spectroscopic measurements. The results of additional, extensive gravimetric and spectroscopic measurements for the adsorption of a variety of suitably substituted benzenes, as well RS the mechanism of the T-OH interaction, are taken up in part 111.
Experimental Section Spectral grade benzene (Matheson Coleman and Bell) was triply distilled in uacuo over P205. Water used for hydroxylation was doubly distilled in vacuo. Both benzene and water were freed from gas by freeze-pumpthaw cycles. Oxygen was prepared by the thermal decomposition of KMn04 in vacuo. The adsorbent specimens were 1 X 2-cm pieces of a 1-mm thick sheet of Corning Code 7930 porous glass. Prior to adsorption experiments, a sample was alternately degassed and hydroxylated a t 700" seven or eight times in order to obtain a reproducible surface. The sample was then subjected to the following "standard" treatment: (a) heating in 100 Torr of oxygen at 600" for 4 hr; (b) degassing at 700" for l hr; (c) exposure to 4 Torr of water vapor at 25" for 0.75 hr; (d) degassing for 12 hr a t 500 or 750" ; (e) cooling in vacuo to the temperature at which the adsorption isotherm was to be measured. A Cahn RG recording vacuum microbalance (1 pg sensitivity, modified by the manufacturer for use with acetone and other condensable vapors) was used in conjunction with a conventional vacuum system capable of producing a dynamic vacuum of better than Torr. Sections of the system exposed to the adsorbates were fitted with Teflon high-vacuum stopcocks. The sample was suspended from the balance beam by quartz fibers. During an adsorption experiment the sample temperature was controlled to j=00.5" by immersing the sample "leg" in a thermostat. A calibrated iron-core, Nichrome-wound furnace held the temperature to k5" during the degassing steps. The gravimetric work showed (a) that benzene adsorption on a sample which had been subjected to the standard treatment was reversible and reproducible; (b) reproducible results were obtained after a sample had been subjected to several adsorption-desorptionstandard treatment cycles; and (c) identical isotherms were obtained with different samples which had been subjected to the same standard treatment. The infrared work, which will be described in part 11, showed that the optical density and integrated area data for the bands of the surface hydroxyl groups were reproducible, i.e., samples having the same surface hydroxyl group concentrations could be prepared by means of the standard treatment. Also, a series of
E O U i L l l R l U i P R E S S U R E Clair'
Figure 1. Adsorption isotherms. A, adsorption-desorption isotherm a t 31.5", sample degassed at 750"; B and C, isotherms at 4 2 O , samples degassed a t 500" (B) or 750" (C).
experiments was carried out with the pure silica, Cab-0-Si1 (G. Cabot Co., Boston, Mass.), using the above standard treatment. As with porous glass, the results obtained with Cab-0-Si1 were reproducible, suggesting that the standard treatment was effective in permitting the preparation of reproducible surfaces.
Experiments and Results Isotherms. Porous glass sample's which had been subjected to a final degassing of 500 or 750" were prepared. Benzene adsorption-desorption isotherms were measured at each of three temperatures with each type of sample. The plots of Figure 1 are characteristic of the type of data obtained. The adsorption was entirely reversible, the amounts of benzene adsorbing at any particular relative pressure, PIPo, decreasing with increasing adsorption temperature, as expected. Individual isotherms exhibited Brunauer Type IV behavior;24 e.g., plot A, Figure 1. The samples degassed at 500' adsorbed somewhat more benzene than those subjected to more severe dehydroxylation by the 750" degassing; e.g., plots B, C, Figure 1. As the results obtained with the two types of samples were similar, only the data obtained with the 750" sample will be considered. The area of the adsorbed benzene molecule has previously been taken to be 32.3 A2, 25E 40 A2, 28b 41 A2 25a but more generally as 42 Az,7J6,25d-f so that it seems difficult to choose the best value. However, if the BETz6 method is applied to our adsorption data, (24) S. Brunauer, "The Adsorption of Gases and Vapours," Clarendon Press, Oxford, 1945. (25) (a) H.K. Livingston, J . Colloid Sei., 4, 447 (1949); (b) A. V. Kiselev and V. I. Lygin, Colloid J . USSR,23, 478 (1961); (c) P.G. Menon and P. Ramamurthy, KoZZoid-2. 2. Polym. 206, 159 (1965); (d) J. Smith, R. Pierce, and B. Cordes, J . Amer. Chem. SOC.,72, 5595 (1950); (e) J. M. Thorp and J. B. Woulf, Trans. Faraday Soc., 63, 2068 (1967); (f) ref 29,p 255. (26) S. Brunauer, P.H. Emmett, and E. Teller, J . Amer. Chem. SOC., 60, 309 (1938). Volume 7,JsNumber ,J February 19, 1970
J. A. CUSUMANO AND 11.J. D. Low
794
I 0
LO?
RELA-I E PrISSJRE
J
r3
++
Figure 2. Surface coverage vs. relative pressure. The apparent surface coverage 0 for adsorption a t 32" as function of the relative pressure; the sample was degassed a t 750".
monolayer coverage occurred at 5.13 X mol of benzene/g of sample. The nitrogen BET surface area of the samples was 135 m2/g. Then, assuming a crosssectional area of the benzene molecule of 42 A2 for benzene lying "flat" on the surface (other data including the spectra of part I1 show this configuration to be the most probable one), the theoretical monolayer (e = 1) is estimated to occur at a surface concentration of 5.33 X mol of benzene/g of sample (0 is the conventional fractional surface coverage). The choice of 42 A2 would thus seem reasonable. It should be noted that, as shown in Figure 2, the amount of benzene adsorbed was almost a linear function of PIPo up to the monolayer point. Experimental ThermodynamicFunctions. The methods of de Boer2' were used to estimate the thermodynamic functions from the adsorption data. For these calculations, the adsorbate gas phase at a pressure of 760 Torr was chosen as the initial standard state. Corresponding values for the equilibrium pressures P1and P2 relative t o the same degree of surface coverage were then selected from isotherms obtained at temperatures T I and Tz. The isothermal change in the Gibbs free energy, AG1, in going from the three-dimensional gas standard state at a pressure Poof 760 Torr to the adsorbed state with equilibrium pressure PI at temperature TI, is given by AGI = -RTI In (P0/Pl). The value of AG2 was calculated in a similar manner. Values of the differential heat of adsorption, AHa, were obtained from the temperature variation of the adsorbate vapor pressure at constant coverage, and those of the molar differential entropy of adsorption, AX,, were then obtained from the relation AGl = AH, - TIASa. The values of these functions and their relations are shown in Figure 3. Experimental AS, Values for Two Limiting Models. In considering the thermodynamic aspects of a gasThe Journal of Physical Chemistry
5.0
1o
t
h G ' l i = 2 5 3 C4L/ :LE
o-bG7:*+ F R A C T I C I W L S ~ R F ~ SmERirE E 9
Figure 3. Thermodynamic functions. H indicates the surface coverage a t which adsorption hysteresis begins; 750"-degassed sample. The values for liquid benzene are also shown.
solid interaction, it is frequently beneficial to compare the experimentally derived vaIues of the adsorption entropy with theoretical values for both mobile and immobile adsorption. To permit the comparison with the two limiting models t o be made, it is necessary that identical initial and final standard states be used in both sets of data. In the present work, the adsorbed state at half-monolayer coverage was arbitrarily but conveniently chosen as the adsorbed standard state for the immobile model, and that state which corresponds t o a distribution of the adsorbate on the surface with an intermolecular distance equal to that of the three-dimensional gas phase at 0" and 760 Torr was chosen as the adsorbed standard state for the mobile model. The experimental entropy changes in going from the initial standard state (gas phase, 760 Torr) t o the adsorbed standard state are,27then
-AXoi, = - A S - R l n [e/(l - e)] (immobile model) (1) -AXo,,
= -AX
- R In [ A" / A] (mobile model) (2)
where A S is the entropy of adsorption derived from the isotherm data. Values of A , the area available to each molecule &ta given surface coverage e, could readily be obtained from the surface area, cross section, and 8. The value of A", the area available to each molecule in (27) J. H.de Boer and S . Kruyer, Proc. Koninkl. Ned. Akad. Wettenschap, 55,451 (1952);dbid.,56,67 (1953);ibid.,56,238(1953);ibid., 57,92 (1954);ibid.,58,61 (1955);ibid.,65,17 (1961).
HYDROXYL GROUPS AND ADSORBED n!fOLECULES
INTERACTIONS BETWEEN SURFACE
,Qrot
the adsorbed standard state for the mobile model, was obtained by applying the two-dimensional equation of state, FA = kT, to the standard-state conditions. The average distance between gaseous benzene molecules a t 00 and 760 Torr is 3.339 X lo-' cm, so that
F"
+
gSotr
gS"rot
,Soconfig
-
+
gSovib
dorot
-
+
-
=
,&rot
+
gSotr
d"tr
gS"rot
-
+
sSovib
,Sorot
-
+
gSolvib
aSovib.
-
aSolvib
uS"tr
(3)
(4)
(5)
The partition functions were calculated by conventional techniques,2sand found to be gQtr
=
aQtr
gQrbt
=
[ h ( iCW Z ~ ) ~ T / ~ ~ ] " / ' V "(6) = [ 2 r ( C mt)kT/h2]LAAo i
1
(7)
- [8r3(I~/12~)1'8kT/h2]P'a (8)
TU
= R In [M'/'T"/")
+ = R{ln ( l / r u ) [8n3(I~)kT/h2]'''+ R { h (l/a)[83(L~I~)1'zkT/h2] + 1)
(12) (13) (14)
'/2)
(15)
,Sorot a8"vib
+ IC In Q
(11)
R ( l n (l/au) [8g3(I~12~)"*kT/h2]1/z a/2)
,Sorot
where the subscripts g and a signify gas and adsorbed state, respectively, and the superscript " signifies the standard state. Also, t r stands for translation, rot for rotation, config for configuration, and vib for vibration. Concerning the vibrational contributions, the superscript (01) refers to the internal modes, whereas (") refers to the symmetrical vibrations of the entire adsorbed molecule with respect to the surface plane. The assumption is made that gSolvib = *8"vib, i.e., that adsorption does not affect the internal vibrations of the benzene molecule. That assumption is not unreasonable, because infrared spectra showed (part 11) that the perturbation of the C-C and C-H stretching modes caused by adsorption was very small. All entropies were calculated for the mean of two isotherm temperatures. Each of the contributions to the total entropy change was calculated using the ensemble approach of statistical mechanics where the general expression relating the partition function Q of a system to its entropy can be shown to be 28
S = kT(b In Q/bT),,
- 2.30 = 2/3gSotr + 1.62 log T - 2.38
gSotr
-
- exp(-hv/kT)]
In these equations V" is the volume of a mole of gaseous benzene, L Ais Avogadro's number, I A and I B are molecular moments of inertia, u is the symmetry number, and the rest of the symbols have their usual meaning. v in eq 11 was taken to be 135 cm-I as suggested by Ron, et al.l8 The application of eq 5 to eq 6-11 gives the final expressions
gSOrot
gSovib
(for one axis of rotation) 1 = - [ ~ T ~ ( I A I B ) ' / ' ~ T / ~ ~ (10) ] TU
= exp[--hv/2kT]/[l
,&b
-
and for the mobile model, by -AS"mt
(9)
(for two axes of rotation)
gSolvib
a8"vib
= 1 [ 8 r 3( I ~ ) k T / h ~ l ' / ' TU
= kT/A" = 0.338dyn/cm
and the standard molecular area is 4.08T X cm2, T being taken as the average of the two isotherm temperatures. I n the present study, A" = 1.265 X cm2. Figure 4 shows the values of the experimental entropy changes, obtained by means of eq 1 and 2, as functions of coverage. Theoretical AS, Values for Two Limiting Models. The theoretical value of the entropy of adsorption for the immobile model is given by -ASoit =
795
=
R( (hv/kT) [exp(hv/kT) - I]-' In [l - exp(-hv/kT)])
(16) (17)
was calculated from Boltzmann's equation, S = k In W , where W = M ! / N ! ( M - N ) ! for M sites and N molecules. By the usual statistical techniques this leads to aS'aonfig = -R In [8/(1 - e)] (18) aS"oonr ig
The theoretical entropy changes for the two limiting physical models in their standard states were calculated using eq 12-17; the values are given in Table I. Table I: Theoretical Contributions to the Entropy of Adsorption by Various Degrees of Freedom Entropy mode gSotr BSOIOt
-a S Ooonfig -,Sot,
Immobile model (eu)
39.2 20.9 0.0
...
-s o r o t
-3.0 -4.0 (-11.1)"
- ABOt
53.2 (42.1)"
-aSovlb
Mobile model (eu)
39.3 20.9
...
-29.6" -3.0 -4.0 ( - l l . l ) b
23.6 (12.5)b
" The theoretical int,egral and differential entropy terms are of equal value for the immobile model, but - a X d t r differs by R for the mobile model. To permit comparison wit,h experimental values, which are differential ones, -&Y'tr has been corrected and all terms in the table can be taken as differential entropies. Values in parentheses refer to two modes of surface rotation as described by eq 16. (28) T. L. Hill, "Introduction to Statistical Thermodynamics,"Addison Wesley, New York, N. Y . ,1962,p 19.
Volume 74, Number 4 February 19, 1970
J. A. CUSUMANO AND M. J. D. Low
796
0 400
I 0 E33
0 83:
FRACTIGlWL S U i F C C E L 3 \ € R S C E 8
Figure 4. Entropy changes. Experimental entropy changes estimated for immobile ( AS”i,) and mobile (AS”,,) surface species; 750”-degassed sample.
Discussion The hysteresis loops observed are of the general type
E of de Boer’s classifi~ation~~ and could be caused by the presence of tubular capillaries with short necks and wide, sloping bodies or by chains of similar pores which would form longer tubular capillaries having wide parts of various radii. However, as pointed out by de Boer,2gtype E loops could be caused by other pore assemblies, and the sloping part of the desorption branch in the low-pressure region (e.g., -40-20 Torr of trace A, Figure 1) suggests that a group of pores of widely varying neck dimensions makes a minor contribution to loops such as that shown in Figure 1. In view of the nature of the adsorbent and its method of manufacture, however, it is not unlikely that the majority of the pores active in benzene adsorption are of shape-group XV, i.e., tubular capillaries with widened parts, shaped much like a string of uneven-sized beads. Such an interpretation is compatible with a model of porous glass as a random packing of approximately uniform spheres recently proposed by Cadenhead and Everett on the basis of benzene adsorption on heattreated porous glass.30 The weak, reversible adsorption process itself is taken as a physical adsorption, in part caused by an interaction of the benzene with surface hydroxyls. The occurrence of this interaction is merely suggested by the decreased adsorption of the more dehydroxylated 750”degassed samples-e.g., traces B, C, Figure 1-but is definitely established by the infrared data (part 11). Multilayer formation and condensation occurred at the higher pressures, as indicated by the isotherms. With increasing adsorption it is apparent that and values decreased rapidly and, after passing through a small maximum near e = 0.75, approached the values for pure liquid benzene (Figure 3). In be noted Of these changes it judging the than an error of *l% in the pressure measurements, The Journal of Physical Chemistry
larger than the error normally experienced, would lead to a corresponding error of f0.30 kcal/mol for AH, and A 1 cal/mol deg for AS,. Such maxima in heats of adsorption have been observed previously, as with benzene adsorption on porous Aerosil.6 For example, Boulton, et u L . , ~ attributed a maximum in AH, for benzene adsorption on porous silica gel to a compression of the liquid within the filled pores resulting from a decrease in curvature of the menisci near saturation. In the present case, the AH, maximum near 0 = 0.75 occurred near the onset of adsorption hysteresis (the broken line marked H in Figure 3), so that it seems probable that it was brought about by enhanced cooperative interactions of adsorbed benzene within the micropore system of the glass. The behavior of AS, paralleled that of AH.. As might be expected, the entropy loss was very high at low degrees of coverage, indicating a high degree of localization of the adsorbate. AS, values decreased rapidly with increasing coverage, however, suggesting an increase in the mobility of the adsorbed benzene molecules. The AS, maximum near the onset of hysteresis can be explained by a change in molecular mobility caused by a variation in the degree of surface-adsorbate interaction in the pore system. Conversely, the Gibbs free energy AGa decreased continuously with increasing coverage, no maximum arising because of compensation between the AH, and AS, terms of the free energy expression. Such compensation of enthalpy gain by entropy loss is well known for hydrogen-bonded system~.~~ Complementary to these interpretations is the fact that the maxima of the thermodynamic functions occurred at surface coverages at which there were distinct changes in the values of the various spectroscopic parameters, e.g., frequencies and optical densities of the hydroxyl bands, and C-H band extinction coefficients. Those changes, which have not been found with adsorption on nonporous silicas, will be considered in part 11. In contrast to the behavior found with the present porous adsorbent, AH, and AS, values have been found to decrease monotonically with increasing coverage for adsorption on nonporous systems as, for example, with Kiselev’saz data on the adsorption of benzene on nonporous Aerosil silica. Comparison of the experimental and theoretical entropy values est*imatedfor the mobile and immobile models provides further information about the adsorbed benzene, although, as pointed out by Everett,a3the (29) J. H. de Boer, in “The Structure and Properties of Porous Materials,” Vol. X, Colston Papers, D. H. Everett and F. 8. Stone, Ed., Butterworth and Co., Ltd,, London, 1958, p 68 ff. (30) D. A. Cadenhead and D. H. Everett, J. Phys. Chem., 7 2 , 3201 (1968). (31) G. C. Pimentel and A. L. McClellan, “The Hydrogen Bond,” W. H. Freeman and Co., San Francisco, Calif., 1960, p 220. (32) A. V. Kiselev and D. P. Poshkus, Dokl. Akad. Nauk SSSR,120, 834 (1958).
INTERACTIONS BETWEEN
SURFACE HYDROXYL GROUPS AND ADSORBED MOLECULES
difficulties in differentiating between mobile and immobile adsorption are severe when complex molecules are adsorbed on very heterogeneous surfaces. However, the good agreement between the gravimetric and infrared effects, i e . , the occurrence of maxima in the thermodynamic functions at the surface coverages a t which distinct changes were observed in infrared spectra, suggests that some use can be made of the thermodynamic treatment. The theoretical values of 53.2 and 42.1 eu are much too high in comparison to the experimental value of 28.0 eu for an immobile species, and the theoretical value of 12.5 eu (for two modes of surface rotation) is too low for the mobile model (Figure 4 and Table I). For the mobile model, however, the theoretical value of 23.6 eu, estimated on the basis of one degree of rotational freedom about an axis perpendicular to the surface and to the plane of the benzene molecule, is close to an experimental value of 22.6 eu (taken from Figure 4). The 1 eu difference can easily be accounted for by a small amount of additional librational freedom and/or a small degree of vibration of the adsorbed molecule perpendicular to the surface, or by the 1 eu experimental error. The mobile model is less suitable at coverages lower than 8 = 0.5. At 8 = 0.1, for example, -ASoit = 51.2 eu, -AXoie = 52 eu, and -AS",, = 43.4 eu, whereas - A S o m t = 23.6 or 12.5 eu, depending on the number of degrees of rotational freedom, so that the immobile model appears to provide a better description a t low coverages. The thermodynamic considerations therefore lead to the following model for the state of benzene adsorbed on hydroxylated porous glass. At low degrees of coverage the benzene molecules are localized and quite strongly bound, as suggested by the high enthalpy and entropy values. As the coverage is increased, the
797
benzene molecules become mobile, the adsorbed benzene molecule lying flat with the plane of the ring parallel to the surface and rotating freely in the ring plane. In diffusing over the surface, the spinning adsorbed molecule may move slightly toward and away from the surface and may occasionally have an additional 180" rotation, much like a flat stone skipping over water and occasionally flipping over onto its back. Filling of the micropore structure begins a t 8 = 0.6, increasing interaction between adsorbate molecules and hindering some of the motion of the adsorbed benzene. The Frenkel equation34 r =
rOexp(-AH,/RT)
where r ois of the order of 10-la sec and r is the residence time of the adsorbed molecule, is also useful in considering the present system. For benzene at low coverages, AH, =20 kcal/mol so that r = 100 see, the benzene adsorption approaching the behavior expected in a chemisorption. However, a t increasing coverages, AH, falls rapidly to about 8 kcal/mol, at which point T FZ sec. This change in residence times of over 10 orders of magnitude implies a change in adsorbate from a tightly bound to a quite loosely bound state. This is expected for a physical adsorption and is thus congruent with the model of adsorbed benzene based on the thermodynamic parameters. Acknowledgment. Support by means of grants from the Communicable Disease Center and the National Center for Air Pollution Control is gratefully acknowledged. (33) D. H. Everett, Proc. Chem. Soc., 38 (1957). (34) J. H. de Boer, " The Dynamical Character of Adsorption," Clarendon Press, Oxford, 1945.
Volume 74, Number 4 February 10, 1970