3395
J. Phys. Chem. 1992, 96, 3395-3403
Vapor Adsorption on Mica and Silicon: Entropy Effects, Layering, and Surface Forces D. Beaglehole Physics Department, Victoria University of Wellington, P.O. Box 600, Wellington, New Zealand
and H. K. Christenson* Department of Applied Mathematics, Research School of Physical Sciences, Australian National University, Canberra, A.C. T.2601, Australia (Received: August 14, 1991)
We have measured the adsorption isotherms of n-pentane, cyclohexane, tetrachloromethane, octamethylcyclotetrasiloxane (OMCTS), and water on muscovite mica and silicon. An angle-averaging, refractive-index-matchingellipsometric technique has been used to avoid problems due to anisotropy and back-reflection in the case of mica. With OMCTS on mica layering effects like those found with simple gases on graphite and single crystals at low temperatures are observed. OMCTS on silicon shows no layering, but the isotherm is qualitatively different from that of the other nonpolar liquids. The water isotherm on the slightly hydrophobic silicon surface is convex upward but smooth and shows only little adsorption (51 nm) even at saturation. Up to thicknesses equivalent to about half a statistical monolayer, all other isotherms are smooth and linear, showing the importance of entropy effects at low coverages. The isotherms of the nonpolar liquids except OMCTS on both mica and silicon are very similar when normalized by the molecular diameter. Close to saturation differences in wetting are seen. Cyclohexane and tetrachloromethane wet silicon, and n-pentane wets both mica and silicon. None of the isotherms agrees with the Lifshitz theory of van der Waals forces at separations below 2 nm. The results suggest that structural effects are present in many liquid adsorbed films at room temperature, provided that the substrates are smooth and homogeneous. The absence of layering does not imply agreement with the Lifshitz theory of van der Waals forces.
Introduction The influence of liquid structure on the interactions between solid surfaces across liquids has been experimentally well-documented in the past 10 years. The monotonic attraction predicted by the Lifshitz theory of van der Waals forces1 is not found at separations approaching those of molecular dimensions. Instead, oscillatory solvation forces are measured between smooth surfaces in most liquids, from near-spherical, nonpolar molecules such as tetrachloromethane to hydrogen-bonding liquids like methanol and 1,2-ethanediol.” Only in the case of water are the solvation (hydration) forces apparently absent under some circumstances, notably between mica surfaces in water a t low electrolyte concentrations. Here the force does superficially resemble a van der Waals attraction, although no quantitative information is available at separations below 2 nm.5 The experiments are in good qualitative agreement with the predictions of computer simulatied of hard-sphere and Lennard-Jones liquids between two walls and the results of integral equation’ or density-functional theories.* At separations beyond the range of the solvation forces the measured interaction is an attraction that appears to resemble the force predicted by Lifshitz theory. There are indications that even a very slight degree of surface roughness is sufficient to smear the solvation force and replace it by an attraction of a magnitude comparable to that expected from Lifshitz t h e ~ r y . ~ Measurements of the interaction between a solid surface and a fluid (gas-liquid) interface across liquids have been carried out in two principal ways. The first involves pressing a bubble against a solid surface and measuring the separation as a function of the gas pressure in the bubble. The force per unit area between the bubble and the plate is conventionally referred to as the disjoining pressure II and is in the case given by ( I ) Dzyaloshinskii, I. E.; Lifshitz, E. M.; pitaevskii, L. P. Adu. Phys. 1961,
= pb -p1
(1)
where pb is the pressure in the bubble and piis the pressure in the bulk liquid. Experimental setups to measure II have been described by many authors.I&’* Such measurements are only possible for relatively thick films, where the value of II is not too great (typically 11OOO N m-2, usually corresponding to thicknesses greater than 10 nm). The second method is more indirect and is based on determining the vapor adsorption isotherm of the liquid on the solid surface. The thickness of the adsorbed film on the solid surface depends on the “force” between the solid-film interface and the gas-film interface. A repulsive force implies a thickening film with a disjoining pressure defined byI2
II = - [ d G / d t ] , = -(kT/v,) In @/pol where G is the excess free energy of the film due to surface forces, t is the film thickness, p is the vapor pressure of the gas in equilibrium with the film, po is the vapor pressure at saturation, and v, is the molecular volume of the liquid in the film. Gas adsorption data yield information on the disjoining pressure for very large values of II, hence at relatively small intersurface separations or film thicknesses (55-10 nm). To a certain extent these measurements complement the plate-bubble method. Such experiments are in principle capable of yielding the same sort of information as direct force measurements between two solid sur faces. The adsorption of vapors to solid surfaces is not only a means of measuring the interaction across a thin film. It has been the subject of extensive experimental and theoretical study by physicists, chemical engineers, and physical chemists for diverse reasor~s.~*~~ The area is of great fundamental interest to physicists (IO) Derjaguin, B. V.; Kussakov, M. M. Acta Physicochim. URSS 1939,
10, 165.
10, 25.
(2) Christenson, H. K. J . Chem. Phys. 1983, 78, 6906. (3) Christenson, H. K. J. Chem. Soc., Faraday Trans. 1 1984,80, 1933. (4) Christenson, H. K. J . Dispersion Sci. Techno/. 1988, 9, 171. (5) Pashley, R. M. J . Colloid Interface Sei. 1980, 80, 153. (6) Lane, J. H.; Spurling, T. H. Chem. Phys. Letr. 1979. 67, 107. (7) Kjellander, R.; Sarman, S.J . Chem. Soc., Faraday Trans. 1991.87, 1869. (8) Freasier, B. C.; Nordholm, S. J . Chem. Phys. 1983, 79. 4431 (9) Christenson, H. K. J . Phys. Chem. 1986, 90, 4.
(11) Blake, T. D. J . Chem. Soc., Faraday Trans. 11975, 71, 192. (12) Derjaguin, B. V.; Churaev, N. V.; Muller, V. M. Surface Forces; Plenum: New York, 1987. ( I 3) Brunauer, S. The Adsorption of Gases and Vapours; Oxford University Press: London, 1943. (14) Dash, J. G.Films on Solid Surfaces; Academic Press: New York. 1975. ( I 5 ) Steele, W. A. The Interaction of Gases with Solid Surfaces; Pergamon: Oxford, 1974.
0022-3654/92/2096-3395%03.00/0 0 1992 American Chemical Society
3396 The Journal of Physical Chemistry, Vol. 96, No. 8,1992 because it allows the transition from two-dimensional to threedimensional phase behaviorl6~I7 and the transition from discrete to continuum (bulk) properties to be investigated. Gas adsorption is of practical importance in the study of powders and porous materials,Is where it is used for surface area determination.13 The area is related to other phenomena like crystal growth, surface melting,19 wetting, and spreading,20*21all of which ultimately depend on the nature of intermolecular and surface forces. It has long been recognized that the extent of adsorption of a vapor to a surface depends critically on the strength of the intermolecular interactions between the substrate and the vapor species. This forms the basis for the classification of isotherms by DashZZand others. Type I isotherms are found in the case of strong substratemolecule interactions and are characterized by complete wetting of the substrate by the liquid or adsorption that grows continuously as the vapor pressure approaches saturation. This is equivalent to a disjoining pressure which decreases monotonically with separation, and this is predicted by Lifshitz theory to occur for most common liquids on high-energy surfaces. Intermediate-strength substratemolecule interactions lead to type I1 isotherms or incomplete wetting (finite thickness at saturation)-they are found in the case when the liquid subtends a finite contact angle on the solid. Such isotherms are only directly predicted by Lifshitz theory in exceptional cases. Weak interactions lead to no adsorption at any pressure and show type 111 isotherms, or a negative disjoining pressure at small separations. Older classification schemes like that of Brunauer13 include many isotherms that show marked effects due to substrate heterogeneity and capillary condensation in porous materials. The strength of the substrate-molecule interaction is related to the surface energy of the solid, and high-energy solids such as ionic crystals and metals adsorb strongly and are wetted by most liquids when the surface is truly clean. Low-energy substrates such as organic polymers and surfactant monolayers usually show intermediate wetting behavior (finite contact angles) and only limited adsorption from vapor. In practice, many high-energy surfaces behave as low- or intermediate-strength substrates due to the presence of adsorbed material on the surface. There is by now a wealth of information on the adsorption of simple gases and hydrocarbons to smooth and homogeneous substrates at low These experiments have shown that the phase behavior of adsorbed films is complex. Structural effects due to the condensation of successive layers of gas molecules dominate at low temperatures, particularly below the triple point temperature where the adsorbed film is solidlike. These layer condensations are first-order phase transitions which terminate at layer critical temperatures, above which they become continuous. The adsorption isotherms show very sharp steps at lower temperatures but these become increasingly rounded with temperature, although structure is usually seen to persist to temperatures above the melting point. Complete wetting is often suppressed in solid films by the effect of the lattice of the underlying substrate, which prevents the adsorbed film from attaining the structure of the bulk crystal. Surface melting of the tap layers at temperatures below the melting point may in some cases counteract this effect. A number of two-dimensional phases have ~~
~
Beaglehole and Christenson been observed in monomolecular layers, in both the solid and the liquid state.30 Both lattice gas and density-functional models33 have gone some way toward rationalizing the phase behavior of adsorbed films in such ideal or near-ideal systems. Indeed, most work on homogeneous substrates like exfoliated graphite has focused on the phase behavior of the adsorbed films themselves, often in the solid state. Little attention has been given to the interaction between the vapor and solid across the adsorbed film. This is in stark contrast to the work on adsorption to more heterogeneous substrates, often near room temperature, where the isotherms are s m o o t h . ’ l ~This ~ ~ apparent lack of structure is not surprising (for discussion of a possible exception, see ref 40). The effect of surface heterogeneities would be to smear any layering transitions because they would occur at different pressures at different sites on the surface. Such an effect is responsible for the knee in the familiar BET adsorption isotherm.14 Even with reasonably homogeneous substrates adsorption of contaminants from the atmosphere would lead to an increased heterogeneity. Roughness of the substrate surface leads to uncertainties in the determined film thickness. Temperatures that are often greatly in excess of the triple point mean that thermal effects act to roughen the film-vapor interface and obscure any steps in the isotherm. Many of these latter investigations have addressed the question of the validity of various theories of intermolecular and surface forces to describe the determined isotherms. When electrostatic contributions such as double-layer forces between the solid surface and the liquid-vapor interface can be safely neglected, as in the case of nonpolar liquids, Lifshitz theory predicts a repulsive pressure across an adsorbed film in most cases. This repulsion will be proportional to t - j in the nonretarded limit (small t ) and proportional to t4 in the retarded limit at large thicknesses (this neglects the zerefrequency term which is not retarded but always small for nonpolar systems). The work of Sabisky and Anderson is perhaps the most convincing demonstration of the validity of Lifshitz theory for thin liquid This covered the thickness range of 1.2-30 nm, or four or more statistical layers of helium, but the films were obtained by allowing helium to climb up surfaces of alkaline earth fluorides and the liquid film was thus in direct contact with the reservoir of bulk helium. Other claims of agreement are less convincing, particularly in the case of adsorbed Good agreement with Lifshitz theory using the platebubble technique has been found in a number of systems.IlJ2 Many authors have sought to describe the observed isotherms using so-called potential theories.44 These assume that van der Waals forces are the dominant intermolecular interactions, but rather than considering the film, vapor, and solid as “bulk” media characterized by their dielectric frequency response (as in Lifshitz theory), the interaction between the surface and the adsorbing molecules is evaluated as a sum or integral of many individual interactions. These theories lead to a similar t - j relationship for the pressure across the film (or, equivalently, the chemical potential). Some models use semiempirical exponential and inverse-square contributions to obtain better agreement with experimental observations, and these are sometimes used to fit
~
(16) Dietrich. S. In Phase Transitions and Critical Phenomena; Domb, C., Lebowitz, J. L., Eds.; Academic: London, 1988; Vol. 12. (17) Sullivan, D. E.; Telo de Gama, M. M. In Fluid Interfacial Phenomena; Croxton, C. A., Ed.; Wiley: Chichester, UK, 1986. (18) Everett, D. H.; Haynes, J. M. In Colloid Science: Specialist Periodical Reports; Burlington House: London, 1973; Vol. l , p 123. (19) Dash, J. G. Contemp. Phys. 1989. 30, 89. (20) de Gennes, P. G. Rev. Mod.Phys. 1985, 57, 827. (21) Cambat, A. M. In Liquids ut Interfaces; Charvolin, J., Joanny, J . F., Zinn-Justin, J . , Eds.; Elsevier: Amsterdam, 1990. (22) Dash, J. G. Phys. Reo. 1977, 815, 3136. (23) Prenzlow, C. F.; Halsey. G . D. J . Phys. Chem. 1957, 61, 1 1 58. (24) Larher, Y.; Millot, F. J . Phys. (Paris) Colloq. 1977, 38, C4-189. (25) Hamilton, J . J.; Goodstein, D. L. Phys. Reu. 1983, 828, 3838. (26) Zhu. D.-M.; Dash, J. G. Phys. Reu. 1988, 838, I 1 673. (27) Nham, H. S.; Hess, G. B. Phys. Rev. 1988, 838, 5166. Volkmann, U. G.; Knorr, K. Surf. Sci. 1990,227,390. (28) Faul, J. W. 0.; (29) Gay, J. M.; Suzanne, J.; Coulomb, J. P. Phys. Rev. 1990, 841, 1 1 346. (30) Thomy, A.; Duval, X.; Reginier, J . SurJ Sci. Rep. 1981, I . I .
(31) de Oliviera, M. J.; Griffiths, R. B. Surf. Sci. 1978, 71, 687. (32) Pandit, R.; Schick, M.; Wortis, M. Phys. Rev. 1982, 826, 5112. (33) Ball, P. C.; Evans, R. J . Chem. Phys. 1988,89, 4412. (34) Lando, D.; Slutsky, L. J. Phys. Rev. 1970, 8 2 , 2863. (35) Tadros, M. E.; Hu, P.; Adamson, A. W. J . Colloid Interface Sci. 1974, 49, 184. (36) Hu, P.; Adamson, A. W. J . Colloid Interface Sci. 1977, 59, 605. (37) Lee, W. Y.; Slutsky, L. J. J . Phys. Chem. 1982. 86, 842. (38) Gee, M . L.; Healy, T. W.; White, L. R. J . Colloid Interface Sci. 1989, 131, 18. Gee, M. L.; Healy, T. W.; White, L. R. J . Colloid Interface Sci. 1989, 133, 514. (39) Metsik, M. S.; Perevertaev, V. D. Kolloidn. Zh. 1966, 28, 99. (40) Gee. M . L.; Healy, T.W.; White, L. R. J . Colloid Interface Sci. 1990, 140, 450. (41) Anderson, C. H.; Sabisky, E. S. Phys. Rev. Lett. 1970, 24, 1049. (42) Richmond, P.; Ninham, B. W . J . Law Temp. Phys. 1971, 5 , 177. (43) Sabisky, E. S.; Anderson, C. H. Phys. Rev. 1973, A7, 790. (44) Adamson, A. W. Physical Chemisrry of Surfaces, 3rd ed.; Wiley: New York, 1976.
Vapor Adsorption on Mica and Silicon adsorption isotherms in much the same way as BET theory is employed. There has hitherto been a shortage of high-resolution measurements of adsorption isotherms on smooth and homogeneous substrates at room temperature. The extent to which structural effects in adsorbed films are important a t higher temperatures is unknown, and few serious quantitative comparisons with Lifshitz theory have been attempted. There is thus a real need for more data on vapor adsorption a t room temperature to attempt to increase our knowledge of the interactions in adsorbed films of common liquids. In this paper we present results of vapor adsorption measurements on muscovite mica and silicon. We have used a novel and very accurate ellipsometric technique4s to determine the thickness of adsorbed films of cyclohexane, tetrachloromethane, n-pentane, octamethylcyclotetrasiloxane,and water. Short accounts of some of these experiments have recently appeared e l ~ e w h e r e . ~ -We ~’ have compared our results to what is known about adsorption of simple gases on smooth and homogeneous substrates at low temperatures and to previous work with similar liquids on other well-defined substrates at room temperature. We have been particularly concerned with the disjoining pressure of the adsorbed film and with comparing this with theory and previous work. Our results show a number of features previously not seen at room temperature and suggest that many commonly held assumptions need to be critically reviewed. Above all, these measurements open the door to a number of far-reaching investigations of fundamental aspects of the phase behavior and surface forces of adsorbed films.
Materials and Methods Ellipsometry. With ellipsometry one measures the coefficient of ellipicity p which is the Im (r) at the angle of incidence where Re ( r ) = 0 (the Brewster angle), where r = rp/rsis the ratio of the amplitude reflectivities for p and s polarized waves. p provides the thickness of the absorbed layer t when that layer is thin and when its dielectric constant (equivalently the refractive index) is known. In particular, p is given by the expression derived by Drude in a continuum model of the media for small thicknesses
where is the dielectric constant of the liquid layer of thickness t on the substrate of dielectric constant c2, and light is incident through the vapor of dielectric constant E,. Note that p varies linearly with t for thin films. In analyzing the results the full Drude expression was used since particularly for the case of films on silicon significant nonlinearities occur even for small values oft. Standard values o f t taken from the Handbook of Chemistry and Physics, 45th ed., were used in the calculations. While the dielectric constant c of bulk liquids is easily measured, how it varies with thickness as the layer thins to submonolayer equivalent thicknesses has not been extensively studied. However, it is likely that the use of the bulk dielectric constant for such thin films will not lead to seriously misleading thicknesses. The reasoning is as follows: The (relative) dielectric constant of the layer is given by t = to P/E (4)
+
where P is the induced polarization per unit volume in the layer by the field E. If the dipole induced in a molecule by the incident electromagnetic wave is not affected by the fields of the induced dipoles on other nearby liquid or substrate molecules, then PIE will be just na, with n the number of molecules with polarizability a per unit volume. In the bulk liquid, interactions between the molecules modify this to become n a / ( 1 - n a / 3 ) , the Clausius(45) Beaglehole, D. Physica 1980, 1008, 163. (46) Beaglehole, D.; Radlinska, E. Z.; Ninham, B. W.; Christenson, H. K. Phys. Rev. Lett. 1991, 66, 2084. (47) Btaglchole, D.; Radlinska, E. Z.; Ninham, B. W.; Christenson, H. K. hngmuir 1991. 7, 1843.
The Journal of Physical Chemistry, Vol. 96, No. 8,1992 3397 Mossotti expression. When t has a value of about 2to, typical of most nonpolar liquids, then the interaction term provides a 30% renormalization to na. A submonolayer layer would thus be expected to be characterized by an effective dielectric constant which is smaller than the bulk by no more than about 15%. Note that this variation in the dielectric constant should occur quite quickly as the coverage 9 decreases below 1 monolayer, since n Qc
03.
In summary, in the submonolayer region we would expect the ellipticity to follow the above expression, with t fixed at a monolayer thickness d, but now with a surface coverage dependent t in the expression for p . The latter will be determined by the term (t - e l ) at low surface coverage, which will be proportional to n = nA/d,with nA the number of molecules per unit area, Le., to the surface coverage 0. Strachan4*and S h i ~ u k h i have n ~ ~ considered the effects of the coherent scattering by surface molecules spread over a continuum substrate, and derive more rigorously appropriate expressions for the ellipticity in the submonolayer region, through the relation to the bulk properties follows the same argument given above. Experimentally, the only attempt to check the prediction appears to be that of SmithSoin a study of the variation of the ellipticity with the adsorption of fatty acids on the surface of mercury. Within his accuracy of about 20% he finds p=aOd, t < d ; p=at, t > d (5) with the proportionality constant a having the same value in the two regimes. Also, Boinovich and Emelyanenkosl found no change from bulk in very thin layers (1-3 nm) of heptane adsorbed on free glass films. Mica as a Substrate. While mica with its feature of easy preparation of molecularly smooth surfaces has been much used in surface studies, its use as a substrate for ellipsometry measurements has caused difficulties. The top and bottom surfaces of the mica substrate are usually so parallel that severe interference occurs between the light waves reflected from the two surfaces. Interference can be reduced by approximate index matching at the back surface, and we have achieved this easily by using a thick layer of epoxy glue ( 5 min Araldite) to fix the mica substrate onto a black anodized aluminum holder. The index matching, however, cannot be perfect, since mica is optically anisotropic. With the approximation e, = ty # cz, it is best to choose t2 = e,(t, - t l ) / ( t z - e,) since this makes Im ( r ) = 0 at the Brewster angle for a thin anisotropic layer on the isotropic medium t2;see Lekner.s2 Measurements and calculations by one of uss3show that the residual signal, which has interference effects as the angle of incidence is varied, is shifted as a whole to larger values of Im ( r ) by an amount calculated for a thin film on an equivalent isotropic substrate, as shown in Figure 1. A measure of the angle-aueruged ellipticity signal thus removes the residual interference effects and leaves a signal that can be easily interpreted in terms of the adsorbed layer thickness. A variation in the angle of incidence of about 5 O will typically provide between 5 and 10 interference oscillations for a typical mica substrate of about 100-pm thickness, and so we perform the appropriate angle averaging by using a lens to provide this range of angles in converging incident light beam, followed by a second lens after the sample to focus the light beam onto the detector. One of the authorss4 studied drops spreading on mica using an imaging ellipsometer, and there the objective lens provided the simplifying angle averaging so that mica interference effects were also not important there. We have found that the ellipticity signal from clean mica depends also upon the rotational orientation of the anisotropic mica cleavage surface relative to the plane of incidence. Mica (48) Strachan, C. Proc. Cambridge Philos. SOC.1933, 29, 116. (49) Sivukhin, D. V. Sou. Phys. JETP 1956, 3, 269. (50) Smith, T.J. Opt. SOC.Am. 1968.58, 1069. (51) Boinovich, L. B.; Emelyanenko, A . M. SurJ Sci. 1990, 225, 206. (52) Lekner, J. Theory of Reflection; Martinus Nijhoff Amsterdam, 1987; p 144. (53) Beaglehole, D. Unpublished results. (54) Beaglehole, D. J . Phys. Chem. 1989, 93, 893.
Beaglehole and Christenson
3398 The Journal of Physical Chemistry, Vol, 96, No. 8, 1992
teflon stopper
Im 0.02
T
va in1
water bath
-
plane of incidence glass cell
leak
Imro2
: 0.2
n n U
U
reservoir cold trap
Figure 2. Schematic diagram of experimental setup showing cell (top) and gas-handling system (bottom). The light beam passes through the glass water jacket and cell walls normal to their surface.
Figure 1. Locus of rp/rqfor an anisotropic substrate of t, = 2.47, tz = 2.53, and d = 100 pm with approximate index matching t = 2.50. The angle of incidence varies from 52' to 63'. (top) No layer; (bottom) with surface layer t = 2.0 and t = 5.0 nm. The upward displacement of the angle average between top and bottom is just that calculated with the usual p-t relationship.
in fact is monoclinic. It cleaves in the 001 plane; the optic axis lies in the 010 plane, inclined at about 15' to the 010 normal; and ea 2.47, cd 2.56, cy 2.59, where p and y directions are 010, 100, respectively, and a! lies in the 010 plane.55 The theoretical description of this ellipticity from such an anisotropic crystal remains a challenge for the theoretician. We observe that the smallest signal is obtained when the mica optic axis lies in the p direction (in the plane of incidence), and we have used this orientation in all our measurements. Silicon as a Substrate. Silicon is weakly absorbing to visible light, having a dielectric constant for 600-nm radiation c = 15 + 0.156i. The absorption obviates any back-reflection problems. For a perfect surface it would also yield an ellipticity of about The large value for the real part of the dielectric constant gives a strong dependence of ellipticity upon adsorbed gas layer thickness. Wafers for the electronics industry have a high-quality finish, with typically a few angstrom root-mean-squaresurface roughness on the short scale. In the study here we used a wafer cut normal to the [ 1 1 11 direction. The nature of the surface depends upon the chemical treatment. We washed the substrate in a dilute H F / H N 0 3 wash and then placed the substrate in boiling H N 0 3 for 5 min, a procedure which leads to a thin well-formed SiOz layer on the surface.56 The resulting surface is moderately hydrophobic (the contact angle of water was not measured but should be similar to the angle measured on heat-dehydroxylated quartz, i.e. 2O-45O4O)and the ellipticity signal (in the absence of adsorbed gases) was about 3 X corresponding to an oxide layer of 3 nm. Adsorption Techniques. In earlier experiments we noticed that thick layers tend to adsorb onto metal surfaces in the measuring
-
-
-
(55) Deer, W. A,; Howie, R. A.; Zussman, J. In Rock Forming Minerals; Longmans: London, 1962; Vol. 3, Sheet Silicates. (56) Archer, R. J. J . Electrochem. SOC.1957, 104, 619.
cell, and it is essential to prevent these from draining onto the substrate. The mica surface was therefore mounted clear of the support, and the support was isolated from the metal sealing cap and glass walls by a Teflon stopper; see Figure 2. The temperature of the cell was kept below that of the plumbing and the surrounding room, by completely enclosing the cell in a secondary enclosure through which temperature-controlled water (A5 mK) flowed continuously. The gas-handling system is shown schematically in Figure 2. We used Datametrics capacitance pressure gauges, matched in sensitivity to the vapor pressure of the gas, and Nupro ultrahigh-vacuum valves. The system was heated to outgas, and checked to be leak tight using a helium leak detector, and also by monitoring the residual leakage rate when the valves were closed. The liquids were stored over Union Carbide 4A molecular sieve zeolite pellets to remove any residual water. The liquid was placed in the reservoir in an atmosphere of pure nitrogen. Residual water vapor in the cell was pumped by a liquid nitrogen trap, which was then closed off for the experiment. The reservoir liquid was frozen to liquid nitrogen temperature, and the whole system was pumped hard to remove the background nitrogen atmosphere. A fresh cleave of the mica surface was performed directly before mounting the substrate into the cell. The cell was then immediately pumped hard and liquid nitrogen trapped, and the background mica ellipticity was determined. Some vapor was then leaked into the cell, and the ellipticity recorded once equilibrium in both pressure and ellipticity was obtained, usually after a few seconds. A continuous recording of ellipticity versus pressure could be obtained on an X-Y recorder, leaking the gas in slowly. With this speedy response it was easy to cycle from low to high pressure quickly, thus monitoring the background mica value for contamination, and checking for pressure drifts. Near saturation with the high vapor pressure liquids the pressure in the cell would occasionally develop thermomolecular pressure oscillations, with the ellipticity following the pressure variations. The ellipticity p was measured with a sensitivity of 10-5,"5which corresponds to a thickness sensitivity of about 0.005 nm for liquids on mica and 0.002 nm for liquids on silicon. The data shown below are representative values read from the continuous experimental curves.
The Journal of Physical Chemistry, Vol. 96, No. 8, 1992 3399
Vapor Adsorption on Mica and Silicon
30
F
25
-
20
-
15
-
E 0
E
-
0
v
0
0.2
0.4
0.6
0.8
:
0 0 0
0.9
0.92
I
* I
A
2C 3
E
v
Y
1
1
6""0.6 0.8 1 PIP0 Figure 4. Ellipsometrically determined thickness f of adsorbed films on mica as a function of the relative vapor pressure p / p o at 18 OC of water (filled triangles), cyclohexane (open diamonds), tetrachloromethane (open squares),n-pentane (open triangles),and OMCTS (filled squares). 0
0.2
0.4
Results The adsorption isotherms measured on silicon at 18.1 OC are shown in Figure 3 for film thicknesses below 5 nm. Water shows very limited adsorption, as expected from the slight hydrophobicity of the silica-like surface of the silicon wafer. The amount of adsorption at saturation is about half that obtained by Gee et al.Q in a recent study of water vapor adsorption on heat-dehydroxylated quartz. Cyclohexane, tetrachloromethane, and n-pentane all appear to wet the silicon substrate (see below), and the two first isotherms are almost indistinguishable at vapor pressures below 0.9. Note the linear regime that extends to p / p o about 0.3, where the adsorbed amount equals about half the thickness of a statistical monolayer. OMCTS does not wet silicon and the shape of the isotherm is convex with respect to the pressure axis at low pressures, with an inflection point at p / p o 0.5, instead of concave as for the other liquids (except water). (In this paper we use "wetting" to indicate the growth of layers to thicknesses greater than 5 nm or so,since there seemsto be a clear distinction between liquids limiting at around 2 nm and those growing considerably thicker.) The adsorption isotherms for the same liquids on mica at 18.1 OC are shown in Figure 4. Cyclohexane, tetrachloromethane,
-
0.98
1
BO
"
0
0.96
.I
I
r
0.94
4
PIP0 Figure 5. Thickness of adsorbed films on mica and silicon for liquids that appear to wet the substrate for p / p o 1 0.9. The experimental points are n-pentane on silicon (filled diamonds) and mica (open diamonds), cyclohexane on silicon (open squares), and tetrachloromethane on silicon (filled squares).
1
1
2
10
1
PIP0 Figure 3, Ellipsometrically determined thickness f of adsorbed films on silicon as a function of the relative vapor pressure p / p o at 18 OC of water (filled triangles), cyclohexane (open diamonds), tetrachloromethane (open squares), n-pentane (open triangles), and octamethylcyclotetrasiloxane (OMCTS) (filled squares).
-1
0
3400 The Journal of Physical Chemistry, Vol. 96, No. 8, I992 5 ,
1
I
I
Beaglehole and Christenson 0.8
0.7
I.
" t - 2
1
0
:o!
I
p a -
I@,
d;"
0
'
0.2
'
'
'
0.4
' '
0.1 '
'
0.6
' '
'
'
0.8
'
'
'
0 1
P/R Figure 7. Adsorption isotherms for OMCTS on mica as a function of temperature below and above the bulk melting point (11 (open diamonds), 14 (filled diamonds), 19 (open squares), and 24 "C (filled squares)). Note the lack of qualitative change at the triple point temperature (18.4 "C). The dashed line is a hypothetical isotherm at temperatures far below the triple point, where the layering transitions are first order. in all cases as is the sharp increase in adsorption at around p / p o * 0.8. For comparison, the adsorption isotherm of OMCTS on silicon is also shown. Adsorption data as a function of temperature below and above the melting point of bulk OMCTS (=triple point temperature, 18.4 "C) are shown in Figure 7 . There is no qualitative change in the isotherm at the melting point, as expected, but the sudden increase in adsorption at p / p o = 0.8 shifts to higher pressures with decreasing temperature. The total film thickness at saturation increases with temperature. The similarity of all the isotherms, on both silicon and mica, becomes readily apparent when the results are plotted on the same graph. By normalizing the film thicknesses by some quantity related to the mean molecular diameter of the molecules, the isotherms become virtually indistinguishable in many cases. In Figure 8 the measured film thicknesses have been normalized by the cube root of the molecular volume, which should give a good relative measure of the mean molecular diameter. With these units, one statistical monolayer corresponds to a normalized thickness of approximately 0.12. As can be seen,only the OMCTS isotherms are very clearly different from the others at low relative vapor pressures. The water isotherm on mica shows less adsorption at low pressures. Discussion The most striking features of the measured isotherms are (i) the very different shape of the OMCTS isotherms and in particular the structure shown by the isotherms of this substance on mica, (ii) the overall similarity of all the other isotherms below film thicknesses of 2 nm (except, to some extent, water on both mica and silicon), including the linear regime at low pressures on both mica and silicon, and (iii) the sharp distinction in wetting behavior on both mica and silicon between liqultls of very similar properties. We will discuss each of these points in turn and then turn our attention to comparisons with Lifshitz theory. We also discuss possible implications of the two main problems in vapor adsorption studies-temperature control and adsorption of contaminants. Layering: Structural Effects. The convex (upward) shape of the OMCTS isotherms and the sharp increase in adsorption at p / p o = 0.8 (on mica) are evidence of molecular layering in the adsorbed film. Such layering is commonly seen even in liquid films (above the triple point temperature of the bulk liquid) of simple The very rounded hydrocarbons and inert gases on shape of the isotherm suggests that the film is far above the roughening temperature, where the layering transitions are continuous. The similarity in shape of the OMCTS isotherm on silicon suggests that there are remnants of layering effects on this
0
0.2
0.6
0.4
0.8
1
P'PO Figure 8. Measured film thickness normalized by cube root of the molar volume for all liquids on both mica and silicon at 18 OC (arbitrary units). Note the similarity of all the isotherms except those of OMCTS on both silicon and mica and, to a lesser extent, those of water on mica and silicon. A value of about 0.12 corresponds to one statistical monolayer. OMCTS/mica, open squares; OMCTS/silicon, filled squares; water/ mica, open triangles; water/silicon, filled triangles; cyclohexane/mica, open diamonds; cyclohexane/silicon, filled diamonds; tetrachloromethane/mica, short dashes; tetrachloromethane/silicon, long dashes; pentane/mica, asterisks; pentane/silicon, crosses.
substrate too. The greater surface roughness and/or some slight surface heterogeneity related to the Occurrence of silanol groups on the silica-like surface may be responsible for the absence of any distinct steps in the adsorption isotherm. It is even possible that surface heterogeneity of the mica substrate is responsible for the very smoothed appearance of the first layer condensation. Any surface heterogeneities would have less influence as the second layer condenses because the surface is already covered by OMCTS molecules. The fact that the isotherms are plotted as a function of the relative vapor pressure does, however, expand the features at lower pressure compared to those at higher pressures. This expansion would be even more noticeable if the results were plotted as a function of chemical potential (kT In [p/pO]).If we assume that the adsorption is governed by van der Waals interactions that decay as r 3(that this is reasonable is clear from results discussed below), we can give roughly equal space to each layer by plotting the film thickness as a function of (1- c ( o ) - ' / ~(cf. ref 25) to obtain the isotherms shown in Figure 9. As can be seen, the 0-1 and 1-2 transitions are now seen to be much closer in shape, although the second one still appears to be slightly sharper than the first (compare the slopes of the isotherms in the vicinity of the transitions). The fact that the 0-1 transition is less sharp than the 1-2 may be indicative of a higher critical temperature for the second layering transition, as predicted by lattice gas theories and seen previously in some experiments. For comparison, the results of using the same "linearization" for cyclohexane on mica and OMCTS on silicon are also shown (these isotherms have been displaced for clarity). These isotherms are qualitatively different and appear completely smooth. The shift in the 1-2 transition toward lower pressures (and chemical potentials) with increasing temperatures is characteristic of a liquid whose wetting of the substrate increases with temperature. Such a shift has been observed with gases adsorbed on graphite.25 The maximum adsorption of OMCTS is also seen to increase with temperature (see Figures 6 and 9). As expected, the isotherms show a smooth evolution with temperature that is not affected by the bulk triple point. Disregarding any significant decrease in adsorption at still lower temperatures, a hypothetical isotherm at temperatures far below the triple point temperature, where the layering transitions are first-order, might appear as the dotted line in Figure 7 . The other liquids show no sign of layering effects. Jt is likely that the convex shape of the water isotherm on silicon is related
The Journal of Physical Chemistry, Vol. 96, No. 8,1992 3401
Vapor Adsorption on Mica and Silicon
1o9
4 I
A
m
3
1OB
A A
-E
0
9
h
E 5 2
lo7
t
= lo6
CI
1
1o5
t
\
\
1o4
0
0
0.01
0.02
0.03
-113
(1 -10)
Figure 9. Adsorption isotherms of OMCTS on mica (1 1 (open diamonds), 14 (filled diamonds), 19 (open squares), and 24 O C (filled squares)), cyclohexane on mica at 18 O C (filled triangles), and OMCTS
on silicon at 18 OC (open triangles) plotted as a function of ( p This gives equal weight to the different layers of OMCTS on mica on the assumption that the adsorption is governed by some van der Waals-like potential decaying as PI3. The points for cyclohexane on mica and OMCTS on silicon have been displaced by 0.01 to the right for the sake of clarity. to surface heterogeneities (silanol groups would have a larger affinity for water than other sites) rather than layering, given that the isotherm on mica (which has no surface hydroxyl groups) appears completely smooth. The other liquids are all much further above their respective triple point temperatures than OMCTS; cyclohexane, where T,is 5.4 OC, is closest. A liquid-vapor interface is rough due to capillary waves, and the density profile is believed to decay from liquid to vapor over a length of some two molecular diameter^.^' In a thin film of a few molecular layers on a solid surface the potential of the solid substrate may have the effect of increasing order in the adsorbed film and therefore layering is commonly seen to persist above the melting point. The higher the temperature above the melting point, the less the ordering effect of the solid substrate. Layering in OMCTS compared to the other liquids is also favored because its polarizability and consequently attraction for the substrate are larger. By contrast, the forces measured between two mica surfaces in cyclohexane, tetrachloromethane, and OMCTS2 are virtually identical when normalized by the mean molecular diameter. In this case the important factor is the hard-core repulsion between the individual molecules and the molecules and the surface, and this is much less dependent on temperature than the smoothness of a liquid-vapor interface. The solvation forces in both OMCTS and cyclohexane have been shown to be virtually independent of temperature in the range 14-30 0C.2,58 Similarity of Isotherms. The adsorption laws for all films on uniform surfaces should tend to Henry’s law in the limit of low pressures. If the adsorbing species are sufficiently far apart, only the substratemolecule interaction is important and the behavior is dominated by entropy effects related to the coverage of the surface. Both the BET and Langmuir isotherms tend to Henry’s law at low pressures. Our isotherms show linear behavior up to unusually high degrees of coverage, up to almost 1/2 monolayer in the case of cyclohexane, n-pentane, and tetrachloromethane on mica and slightly less for the same liquids on silicon. The presence of surface heterogeneities normally acts to reduce the linearity, particularly at very low pressures where the most energetic sites are covered very quickly. Typically, linear behavior is normally seen only up to about 1/ 10-monolayer coverage.I4 Our results are thus evidence of good substrate homogeneity (note the (57) Beaglehole, D. Phys. Reo. Lett. 1987, 58, 1434. (58) Christenson, H. K.; Israelachvili, J . N . J . Chem. Phys. 1984, 80, 4566.
0.1
On
10
1
100
t (nm)
Figure 10. Disjoining pressure of adsorbed films on mica surfaces plotted on a logarithmic scale. Symbols show experimental points (squares, pentane; triangles, water; diamonds, cyclohexane);the solid line is the prediction of the Lifshitz theory of van der Waals forces for n-pentane. The dashed line shows the total disjoining pressure from the van der Waals interaction and an extra contribution caused by the dissolution of 1/4 monolayer of soluble material in pentane, assuming the solute behaves ideally and has the same molecular volume as pentane (see text). The Lifshitz interaction for cyclohexane is almost the same as for pentane, and that for water is only slightly smaller at a given separation,but has the same shape.
exception of water on silicon, discussed above). Even in the case of water on mica intermolecular interactions are obviously not of major importance in the submonolayer regime. (Note that we have argued earlier that a nonlinearity of less than 15% might be expected between p and t , occurring in the vicinity of t d . The observed linearity in the experimental curves extends to about 0.4d.) The similarity of the isotherms of pentane, cyclohexane, and tetrachloromethane on both mica and silicon as well as water on mica extends to larger thicknesses (see Figure 8). In the case of the nonpolar liquids the interactions would be expected to be very similar on both substrates. That the water isotherm on mica is as close as it is to the other liquids is perhaps more surprising. The adsorption of water in terms of molecular layers, however, is larger than the other nonwetting liquids-possibly a reflection of increasing mperativity due to hydrogen bonding once the water film is thicker than a few molecular layers. Wetting Isotberms and Lifshitz Tbeory. If the disjoining pressure of the films is calculated using eq 2 and the results are compared with the predictions of the Lifshitz theory of van der Waals forces, the results appear as in Figures 10 and 11. Note that the calculations are straightforward for micas9v60and are expected to be quite accurate as the dielectric data used have previously been shown to yield values in good agreement with experimental results for benzene4 and water6’ between two mica surfaces at separations in the range 3-20 nm. Silicon is more difficult to treat-the fact that the surface is like fused silica necessitates a complete triple-film calculation and there is some uncertainty about the precise thickness and dielectric properties of the surface layer of silica. We have chosen to calculate the nonretarded Hamaker constants from Lifshitz theory for the liquids on silicon and silica and then use a combining relationship2’ to approximate the disjoining pressure in the composite system. Note that (a) for small thicknesses (12 nm) only the silica layer is important and (b) the uncertainties will not change the overall
-
(59) Richmond, E. P. In Colloid Science: Specialisr Periodical Reports; Burlington House: London, 1975; Vol. 2, p 130. ( 6 0 ) Christenson, H. K. Ph.D. Thesis, Australian National University, 1983. (61) Israelachvili, J. N.: Adams, G.E. J . Chem. SOC., Faraday Trans. I 1978, 74, 975.
3402 The Journal of Physical Chemistry, Vol. 96, No. 8, 1992
log
A
9
E
10’
K
1o6
,,,/
\ ,
,
,,n,o,,j
0.1
1 10 t (nm) Figure 11. Disjoining pressure of adsorbed films on silicon surfaces.
Symbols show experimental points (open squares; cyclohexane; open triangles, OMCTS; filled diamonds, water; filled square, tetrachloromethane); the solid line is the predictions of the Lifshitz theory of van der Waals forces for tetrachloromethane (see text). The dashed line shows the disjoining pressure caused by the dissolution of 1/6 monolayer of soluble material in tetrachloromethane, assuming the solute behaves ideally and has the same molecular volume as tetrachloromethane. The dotted line is the sum of the solute and the van der Waals contribution. appearance of the theoretical curves. The nonwetting liquids (including OMCTS) all show similar behavior on mica-the experimental points cross the Lifshitz force at submonolayer thicknesses and then show considerably greater than expected film thicknesses until adsorption levels off at or close to saturation. Pentane wets mica and the films grow to very large thicknesses-in excess of 20 nm at saturation. On silicon only water and OMCTS appear to show thicknesses below about 5 nm close to saturation, whereas the isotherms of pentane, cyclohexane, and tetrachloromethane are very similar and grow to large thicknesses (=lo nm). The disagreement with Lifshitz theory for the thin films of the nonwetting liquids is not surprising. Lifshitz theory is a continuum theory and uses bulk dielectric data to predict the behavior of thin films on substrates. It is clearly of questionable validity in films of only a few molecular diameters. Theoretically, wetting behavior is expected for all studied liquids on both mica and silicon. The fact that no liquids except n-pentane wet mica, is however, in accord with the observations that these liquids all show small but finite contact angles on mica. Even if Lifshitz theory is accepted as being valid for thick films, this does not imply that it can correctly predict wetting or the lack there0f.l The positive disjoining pressure calculated assumes that the film will thicken because the free energy decreases monotonically from its value at zero film thicknesses uo to the value at effectively infinite thicknesses usL uLv. If, however, the system has an energy minimum at some small thickness, where the theory does not hold, this would prevent further thickening and the result is nonwetting, as observed. Such a situation would conceivably lead to hysteresis in the behavior of the films. If the film were thinned from large thicknesses down to contact, it might be possible to access the regime that is inaccessible on thickening. The behavior on silicon is also in qualitative agreement with contact angle data. The nonwetting liquids are OMCTS and water, and water has a rather high contact angle on the hydrophobic silica-like surface. The wetting liquids all show similar S-shaped curves on the log-log plot of Figure 11 and much larger than predicted film thicknesses, as for n-pentane on mica. The fact that some of the liquids wet the substrate and others with very similar properties do not warrants careful consideration. The wetting films are of such thicknesses that one might expect Lifshitz theory to hold. There are a number of previous investigations showing that Lifshitz theory provides an accurate representation on the behavior of thicker films. Examples include
+
the work of Sabisky and Anderson on helium films on alkaline earth the work of Blake on decane films on alumina,” and various Soviet studies of films on glass and polished steel.’* None of these, however, has involved actual vapor adsorption measurements. Most work showing good agreement with Lifshitz theory for solid-liquid-vapor systems has been carried out with the bubbleplate method. Gee et al. unfortunately did not provide any meaningful comparison for large film thicknesses in their recent publications on adsorption of alkanes and alcohols38 to silica. We note also that the work of Blake” on vapour adsorption of n-decane (film thicknesses 1 5 nm) appears to show exactly the same deviation from Lifshitz theory as that observed by us. A recent study by Tidswell et a1.62using X-ray techniques finds good agreement with the Lifshitz theory for cyclohexane on silicon. Their measurement was made at 30 OC which, other than their technique, appears to be the major difference with our study. Effect of Surface Contamination, The thick-film region presents two very great experimental problems. The first is temperature uniformity and control, and close to saturation temperature gradients of a few hundreths of a degree can lead to substantial differences between actual and measured relative vapour pressure. We do not believe that this is a serious concern in our system. The second problem is related to the Occurrence of adsorbed species other than the vapor under investigation. Any high-energy surface that is handled in an atmosphere will adsorb molecules to the surface. Particularly mica has a very high surface energy, but on cleavage in air at normal pressures it is rapidly coated with an adsorbed layer that reduces its surface energy c o n ~ i d e r a b l y . ~ ~ . ~ ~ This has been amply discussed in various papers on direct force measurements between mica surfaces. The precise chemical composition of the adsorbed material is not known, but it is certainly composed mainly of water vapor and carbon dioxide. There are definite indications that some water-soluble compound, possibly a potassium salt, also forms on the surface through a slow leaching p r o c e ~ s . ~ ~There , ~ ~ is thus no doubt that the mica substrate is covered by some adsorbents which are not removed under vacuum. Freshly cleaved mica is wetted by water, but an instant of exposure is sufficent to give a small contact angle.66 Small variations in the amount and composition of adsorbed material may be responsible for the variability of the isotherms from sample to sample. The adsorption of spurious contaminants during an experiment (e.g., pump oil from the vacuum system) would invariably lead to hysteresis in the measured isotherms, and this was never observed in our experiments. The effect of thin (subnanometer) adsorbed layers on the calculated van der Waals forces would be marginal in all but the thinnest films. Dissolution of any adsorbed species in the liquid films would, however, lead to an additional disjoining pressure due to the reduction in chemical potential of the adsorbed liquid. This can be a large effect close to saturation. It has been shown that the very thick wetting films of water found on hydrophilic quartz and silica6’ can be quantitatively accounted for by the dissolution of soluble species in the water films.68 A similar calculation for our wetting isotherms shows that such an effect would lead to isotherms similar to those observed on both mica and silicon. As can be seen in Figure 10, a quantity of soluble material equal to 1/4 monolayer is sufficient to give a good fit to the pentane isotherm on mica. Slightly smaller quantities are required to give (approximate) fits to the silicon isotherms (see Figure 11). The deviations for films I 1 0 nm may be explained by gravitational thinning.
7
lo8
5
Beaglehole and Christenson
(62) Tidswell. I . M.; Rabedeau, T’. A.; Pershan, P. S.;Kosowsky, S. D. Phys. Reu. Lerr. 1991, 66, 2108. (63) Bailey, A. 1.; Kay, S.M. Proc. R . SOC.(London) 1967, A301, 47. (64) Christenson, H . K.; Israelachvili, J. N. J . Colloid Interface Sci. 1987, 1 1 7 . 576.
(65) Bhattacharyya, K. G.Lnngmuir 1989, 5, 1155. (66) Ross, S.;Morrison, 1. D. Colloidal Sysrems and Interfaces; WileyInterscience: New York, 1988; p 99. (67) Pashley. R. M.;Kitchener, J . A . J . Colloid Inrerjace Sci. 1979, 7 1 , 491. (68) Pashley, R. M . J . Colloid Interface Sei. 1980, 78, 246.
J . Phys. Chem. 1992, 96, 3403-3408 The possible presence of adsorbed contaminants in the films is obviously of great concern. However, despite the fact that their presence leads to a possible explanation of the experimental curves for those liquids and substrates forming thicker layers near po, we do not believe that the contamination is a problem in the present experiments. The overall features are reproducible from experiment to experiment, and since any species that is soluble in pentane would also be soluble in both tetrachloromethane and cyclohexane it is clear that its presence is not sufficient to cause wetting-it may merely determine the extent to which the wetting film grows at a particular vapor pressure. Further, the quantities of solute required to give fits vary widely from experiment to experiment without any qualitative change in the isotherms themselves. The fact that water does not wet mica is a good indication that if any water-soluble material is present the quantity is not enough to seriously affect the results. Finally, it is also important to remark that even if contaminants were present, they would have no more than a very marginal influence on the behavior of the thinner films. Implications. These measurements together with previously published short a c c o u n t ~show ~ , ~ the ~ wealth of information that can be obtained by careful and accurate measurements of adsorption isotherms on well-defined substrate surfaces. The extent of the linear regime at low vapor pressures, the Occurrence of
3403
layering at room temperature, and the definite (but not unexpected) nonagreement with Lifshitz theory in thin films are all features that have not previously been seen or sufficiently welldocumented. A model of adsorption that can successfully account for these features is a challenge to the theoretician. Perhaps the attempt of Mahanty and to incorporate finitesize effects into Lifshitz theory is a first step in the right direction. On the experimental side the field is open. The increase in adsorption of OMCTS with temperature suggests the possibility of a wetting transition at some higher temperature. Studies of solidlike films of substances with melting points well above room temperature should provide further interesting comparison with low-temperature work. A method of protecting the mica from the atmosphere prior to evacuation may solve the problem of (possible) surface contamination. Acknowledgment. We thank B. W. Ninham, R. M. Pashley, and E. Z. Radlinska for helpful discussions. Registry No. OMCTS, 556-67-2;H20,7732-18-5;Si, 7440-21-3; CCI4, 56-23-5;pentane, 109-66-0;cyclohexane, 110-82-7. (69) Mahanty, J.; Ninham, B. W. J. Chem. Soc., Faraday Trans. 2 1974, 70, 637.
Gallosilicate Zeolite Catalysts: Structural Features of [Si,Ga]-ZSM-5 Xinsheng Liu and Jacek Klinowski* Department of Chemistry, University of Cambridge, Lensfield Road, Cambridge CB2 I E W,U.K. (Received: May 16, 1991; In Final Form: October 28, 1991)
A series of gallosilicateswith the ZSM-5 structure was synthesized in an alkaline medium using tetrapropylammonium bromide (TPABr) as a template, and characterized by XRD, IR, TGA, SEM, MAS NMR, and chemical analysis. The results clearly demonstrate that the number of defects, the strength of interaction between the template molecules and the framework, the distribution of particle sizes, and the size of the crystallites, as well as the content of template, water molecules, and sodium cations, are all correlated with the gallium content.
Introduction Isomorphous substitution of Si atoms in zeolitic frameworks by heteroatoms such as Al,’ B,2 Fe,3-7 Ge,* Ti,”2 and Ga’3-30is
of considerable importance in view of the interesting adsorptive and catalytic properties of the products in reactions such as
(1) Keijsper, J. J.; Post, M. F. M. ACS Symp. Ser. 1989, No. 398. (2) Coudurier, G.;Auroux, A.; Vbdrine, J. C.; Farlee, R. D.; Abrams, L.; Shannon, R. D. J . Catal. 1987, 108, 1 and references therein. (3) Szostak, R.; Thomas, T. L. J . Catal. 1986, 100, 555. (4) Marosi, L.;Stabenow, J.; Schwarzmann, M. German Patent N M 2 831 61 1, 1980. ( 5 ) Borade, R. 9.; Halgeri, A. B.; Prasada Rao, T. S.R. In New Developments in Zeolite Science and Technology. Proceedings ofthe 7th International Zeolite Conference. Murakami, Y . , lijima, A., Ward, J. W., Eds.; Kodansha: Tokyo, 1986;p 851. (6) Meagher, A.; Nair, V.; Szostak, R. Zeolites 1988, 8, 3. (7) Calis, G.; Frenken, P.; de Boer, E.; Swolfs, A.; Hefni, M. A. Zeolites
1987, 106, 287.
1987, 7, 319.
(8) Szostak, R. Molecular Sieves; Van Nostrand Reinhold: New York, 1989. (9) Perego. G.;Belussi, G.; Corno, C.; Taramasso, M.; Buonomo, F.; Esposito A. In ref 5 , p 129. (IO) Guth, J. L.;Kessler, H.; Wey, R. In ref 9, p 121. (11) Reddy, J. S.;Kumar, R.; Ratnasamy, P. Appl. Catal. 1990,58, LI. (12)Thangaraj, A.; Kumar, R.; Ratnasamy, P. Appl. Catol. 1990.57, L1. (13) (a) Inui, T.; Matsuda, H.; Yamase, 0.;Nagata, H.; Fukuda, K.; Ukawa, T.; Miyamato, A. J. Catal. 1986.98.491. (b) Inui, T.; Makino, Y.; Okazami, F.; Miyamoto, A. J . Chem. Soc., Chem. Commun. 1986. 571. (14) (a) Kitagawa, H.;Sendoda, Y.; Ono,Y. J . Catal. 1986,101, 12. (b) Shibata, M.;Kitagawa, H.; Sendoda, Y.; Ono, Y. In ref 5, p 717. (c) Sirokman, G.; Sendoda, Y.; Ono,Y.Zeolites 1986, 6. 299. (15) Ono, Y.;Kanae, K. J. Chem. Soc.. Faraday Trans. 1991, 87, 669.
(16)Simmons, D. K.; Szostak, R.; Agrawal, P. K.; Thomas, T. L.J . Catal.
(17) Scurrell, M. S.Appl. Catal. 1988, 41, 89. (18) la) GneD. N. S.:Dovemet. J. Y.: Guisnet. M. J . Mol. Catal. 1988. 45,’281. ‘ (b) Gnkp, N. S.;Doyemet, J. Y .; Seco, A. M.; Ramoa Ribeiro, F.f Guisnet, M. Appl. Catal. 1988, 43, 155. (19) Petit, L.;Bournonville, J. P.; Raatz, F. In Zeolites: Facts, Figures, Future; Jacobs, P. A.; van Santen, R. A,, Eds.; Elsevier: Amsterdam, 1989; p 1163. (20) Chen, N. Y.;Degnan, T. F.; McCullen, S.B. U S . Patent 4,861,932 (1989). (21) Kanai, J.; Kawata, N. Appl. Catal. 1989, 55, 115. (22)Nemet-Marrodin, M. U S . Patent 4,861,933(1989). (23) (a) Berti, G.; Moore, J. E.; Salusinzky, L.; Seddon, D. Ausf. J. Chem. 1989, 42, 2095. (b) Seddon, D. PCT Int. Appl. WO 89 10,190 (1990). (24) Khodakov, A. Yu.; Kustov, L. M.; Bondarenko, T. N.; Dergachev, A. A.; Kazansky, V. 9.; Minachev, Kh. M.; Borbely, G.; Beyer, H. K. Zeolites 1990, 10, 603. (25) Herbst, J. A.; Owen, H.; Schipper, P. H. U S . Patent 4,929.337 (1990). (26) Le Van Mao, R.; Jao, J.; Sjiariel, B. Catal. Lett. 1990, 6, 23. (27) Price, G. L.; Kanazirev, V. J. Coral. 1990, 126, 267. (28) (a) Petit, L.;Bournonville, J. P.; Guth, J. L.; Raatz, F.; Seive, A. Eur. Pat. 351,311 (1990). (b) Petit, L.; Bournonville, J. P.; Guth, J. L.;Raatz, F.; Kessler, H. Eur. Pat. 351,312 (1990). (c) Seive, A,; Guth, J. L.; Raatz, F.; Petit, L. Eur. Pat. Appl. 342,075 (1990). (29) Klazinga, A. H.;Seelen, K. J. M. E.; Minderhoud, J. K. Eur. Pat. 380,I80 ( 1 990).
0022-3654/92/2096-3403$03.00/00 1992 American Chemical Society
