GRAVIMETRIC ADSORPTION STUDIES OF THORIUM OXIDE
tials. The phenomena related to the hole in organic solids are clearly rather more complex than those for the
1633
electron, and the mechanisms proposed are necessarily tentative.
Gravimetric Adsorption Studies of Thorium Oxide. 11.
Water
Adsorption at 25.OOo1
by E. L. Fuller, Jr., H. F. Holmes, and C. H. Secoy Reactor Chemistry Division, Oak Ridge National Laboratory, Oak Ridge, Tennessee
(Received December 14, 1965)
The complex nature of water adsorption on thorium oxide has been studied using a sensitive microbalance. High-temperature sintering appears to produce a material which predominantly presents the 100 cubic face in the surface, upon which there are three distinct modes of adsorption. There is a rapid chemisorption, forming surface hydroxyl groups which are slowly hydrated. I n addition and as a precursor for surface hydration, physical adsorption occurs.
The interaction of water with the surface of thorium oxide is somewhat complex and not well understood. Gravimetric adsorption studies have been made in an effort to interpret the effects of the past history on the nature of the surface. This work is in conjunction with the calorimetric2 and electrokinetic3 studies reported earlier. It is well known that even the specific surface area of thorium oxide (calculated from nitrogen adsorption isotherms) varies considerably, depending on the mode of preparation and calcining t e m p e r a t ~ r e . ~ , ~ We have studied the nature of water vapor adsorption on two well-characterized samples of thorium oxide in an attempt to evaluate the stoichiometry and thermodynamics of the adsorbed species. I n addition to the fundamental knowledge such data would offer, there is considerable interest in thorium dioxide as a heterogeneous ~ a t a l y s t . ~ ~ ~ Throughout this report we have assumed that a truly physically adsorbed water molecule occupies the theoretical 10.6 A2 rather than some value chosen to correlate the specific surface areas calculated from nitrogen and water isotherms (based on the 16.2 A2 occupied by a nitrogen molecule).'
The gravimetric apparatus and environmental controls used for this investigation have been described previously.s The two samples studied thus far are samples A and D as described in ref 2b and 3 (14.7 and 2.20 m2/g, resulting from 650 and 1200" calcination, respectively). When considered on a unit surface area basis, the two samples behaved identically when outgassed a t torr. The samples were held i n uucuo, and the temperature was incrementally increased with the ~
(1) Research sponsored by the U. S. Atomic Energy Commission under contract with the Union Carbide Corp. (2) (a) H. F. Holmes and C. H. Secoy, J. Phys. Chem., 69, 151 (1965); (b) H. F. Holmes, E. L. Fuller, Jr., and C. H. Secoy, ibid., 70,436 (1966). (3) H. F. Holmes, C. S.Shoup, Jr., and C. H. Secoy, ibid., 69, 3148 (1965). (4)V. D.Allred, S. R. Buxton, and J. P. hfcBride, ibid., 61, 117 (1957). (5) W. 5. Brey, B. H. Davis, P. G. Schmidt, and C. G. Moreland, J . Catalysis, 3, 303 (1964). (6) M.E.Winfield, Australian J. Sci. Res., 3, 291 (1950). (7) H. K.Livingston, J. Am. Chem. SOC.,66, 569 (1944). (8) E.L.Fuller, Jr., H. F. Holmes, and C. H. Secoy, V a c u u m Microbalance Tech., 4, 109 (1965).
Volume 70,Number 5
M a y 1966
E. L. FULLER, JR.,H. F. HOLMES,AND C. H. SECOY
1634
312
400
N
-5
340
300
309
rt 00
306 0
I00
200
300
TEMPERATURE
400
500
("C)
Figure 1. Weight loss on outgassing Tho:! at 10-5 torr.
a I v) a
305 304
resulting weight loss shown in Figure 1. It was found necessary to maintain the sample at each temperature for at least 16 hr to obtain a steady weight. The excellent correlation for these widely diverse specific surface areas points out that this desorption is truly related to the surface of the samples. Also, it is evident that a temperature of 500" is far from adequate to remove all of the tenaciously bound water to produce a "dry" surface at this pressure, since there is no indication of an approach to a limiting weight at this temperature. When the samples were heated to 500" in vacuo, a gray to black coloration began in the upper layers of the powder and penetrated more deeply with successive treatments at 500", whereas maintaining the 500" temperature did not propagate the color. This color is undoubtedly due to a small amount of carbon since it can be removed with oxygen treatment a t 400 to 500" with an accompanying weight loss. Also, raising the temperature to 1000" in vacuo will bleach the samples to their original white. We feel that the presence of this small amount of carbon (or its parent hydrocarbon) in no way invalidates our results since neither the surface area nor the energetics of water adsorption2 are altered by its presence. Immediately upon cooling from 500", the samples begin to gain weight slowly (6 to 7 pg/hr) even when exposed to the vacuum pumps, owing to the desiccating action of the activated surface. A series of successive adsorption and desorption isotherms for water vapor on sample A at 25.00" was obtained. The first two of these, shown in Figure 2, indicate the comdex nature of the Drocess. This min graph iS constructed from data Obtained l5 to after introduction or extraction of water vapor. For The Journal of Physical Chemistry
303 302
304
'
0
~
1
04
02
0.3
0.4
0.5
0.6 0.7
0.8
0.9
4.0
%/., Figure 2.
Water isotherms for sample rl at 25.00".
clarity, only about half of the experimental points are shown. The sharp discontinuities correspond to overnight stops in the isotherm construction. Although the data do not represent equilibrium isotherms, the curves have the general shape of type-I1 isotherms (Brunauer classificationg) with high-pressure hysteresis due to capillary condensation. It appears that underlying a reversible physical adsorption process is a slow irreversible binding of water to the surface. The term irreversible, as used in this paper, applies to the bound water which cannot be removed at the temperature of the isotherm, and does not imply thermodynamic irreversibility. The kinetic nature of this binding i s revealed in the overnight discontinuities in the lower isotherm and in the increasing vacuum weight after each excursion to higher pressures of water vapor. Although only the first two are shown in Figure 2, each successive isotherm failed to close upon the preceding one by a decreasing amount. The experiment was terminated prior to achieving complete closure of the adsorption and desorption branches as was done with sample D (see below). (9) s. Brunauer, "The Adsorption of Gases and Vapors," vel. 1, Princeton University Press, Princeton, N. ,J,, 1945.
GRAVIMETRIC ADSORPTIONSTUDIES OF THORIUM OXIDE
The specific surface area calculated from these water isotherms (BET methodlo) decreased from 28.8 to 9.6 m2/g: The initial values are unquestionably high due to the inability of the BET theory to differentiate between chemical and physical adsorption and the comparatively large amount of water involved in the chemical process. A similar decrease in the specific surface area with higher surface water content has been observed for aluminum oxide." This diminution of surface area from the 14.7 m2/g is probably real and due to the fact that irreversible binding of water effectively decreases the surface area. Classical treatment12 of the capillary hysteresis desorption closure at a partial pressure of 0.35 shows quite a number of pores of radii as low .zs 10 A. A similar irreversible binding has been observed underlying the physical adsorption of water on silicon oxide by Barrett.13 His isotherms were normalizable (identical when each is referred to its own vacuum weight), whereas it is quite evident that thorium oxide isotherms are riot normalizable. Water vapor adsorption on sample D gave the same behavior as sample A in that each successive adsorption occurred at higher sample weight, owing to the slow underlying irreversible binding. Figure 3, which shows the initial and final isotherms only, indicates that the only schematic difference lies in the capillary hysteresis region (closure at pip, = 0.52) where the higher calcination temperature has sintered out all pores of radii less than 50 A. Since each successive isotherm more closely approximates the preceding one, an attempt was made to obtain an isotherm involving only physical adsorption. Six months of repeated adsorption and desorption was required to construct the final reversible isotherm (the upper ciirve in Figure 3). The high-pressure hysteresis, due to capillary condensation, still persists, and the entire isotherm is reproducible to 1 pg over the entire pressure region. The final vacuum weight is only slightly dependent on temperature as observed experimentally a t 18 and 32". The rate of irreversible binding is a complex function of both the water vapor pressure and 1,he amount of water previously bound. The rate is increased markedly with the first increments of prewne, but higher pressure increments have less accelerating effects. The rate diminishes appreciably as more water is bound under the physically adsorbed water. Incremental increases in temperature to 1000" indicate that a "dry" weight of thorium oxide is achieved at 1000" i n vacuo. Duval14 reports that a temperature of 950" must be achieved to obtain a tga (thermogravimetric analysis) curve for stoichio-
1635
500 6
500 5 500.4
500 3 5002
I
500.f
3
500.0 499 9
499.5
! iO0OT
+ ---I--' I I
I
0
metric thorium oxide. When the sample was again exposed to water vapor at 25.00", the same slow irreversible binding process began, and the isotherms essentially reproduce the initial ones. The process was not followed to completion for a second time. The irreversible binding of water to the surface of thorium oxide must certainly be oriented to specific surface sites in light of the energetics of adsorption2 and in view of the fact that it cannot be removed in vacuo without raising the temperature. These two facts preclude the concept that this water is held by physical forces of the van der Waals type. I n attempting to calculate the area occupied by a chemisorbed molecule one must know which crystalline faces are present in the surface. No conclusive evidence has yet been forwarded to answer this problem for thorium oxide. The chemisorption capacity for water on thorium oxide is 192, 135, and 221 pg/m2 for the 100, 110, and 111 surface planes, respectively, based on the (10) S. Brunauer, P. H. Emmett, and E. Teller, J . .4m. Chem. Soc., 60, 309 (1938). (11) N. Hackerman and W. H. Wade, J . Phys. Chem., 68, 1592 (1964). (12) J. H. deBoer, Proc. Symp. Colston Res. Soc., 10, 68 (1958). (13) H . M. Barrett, A. W. Birnie, and M. Cohen. J . Am. Chem. Soc., 62, 2839 (1940). (14) C. Duval, "Inorganic Therrnogravimetric Analysis," 2nd ed, Elsevier Publishing Co., New York, N. Y., 1963.
Volume 70, .\lumber
5
M a y 1966
1636
face-centered-cubic lattice parameter for the bulk oxide (ao= 5.597 A). The right ordinate in Figure 3 is the equivalent monolayer capacity of sample D based on the exposure of the 100 plane in the surface and the specific surface area of 2.20 m2/g. The excellent correlation of the final vacuum weight to that of the equivalent three chemisorbed layers lends credence to this model. The simplest and most straightforward interpretation of this result is that of a dissociative chemisorption to form two surface hydroxyl groups per surface thorium, which in turn are each hydrated with a water molecule in the final equilibrium state. This is the surface analog of a hydrated bulk hydroxide and satisfies the observed stoichiometry. It is hoped that concurrent infrared spectroscopic studies of these same materials will shed some light on the nature of the bonding of these nondissociated hydrating molecules on the surface. This model receives considerable support from the three distinct energetic adsorptions noted for high-temperature calcined thorium oxide.2b I n addition, this stoichiometric amount of water is just that required to truild up one completed face-centered-cubic lattice unit on the surface above the 100 ThOz plane, with the hydroxide and water oxygen occupying image positions of the substrate oxide ions. Recent nmr studies15 have also shown that there is considerable rigid orientation of water on the surface of thorium oxide, which involves a slow hydration process. l6 Furthermore, electrical conductivity studies on quite a number of oxides conclusively show that the hydration process does not involve electron transfer.” Antoniou18 has observed that there is the equivalent of three chemisorbed layers of water in an immobile state on the surface of silica. The model presented here proposes a third distinct type of adsorption: an immobile, associative type (much akin i,o bulk hydrogen bonding) in addition to the truly physical and chemical adsorption. Since physical adsorption occurs in addition, the additional nomenclature seems warranted. The slow kinetics of adsorption cannot normally be associated with a
The Journal of Physical Chemistry
E. L. FULLER, JR.,H. F. HOLMES,AND C. H. SECOY
true physical adsorption process, and the energetics2$ of this surface hydration are near that normally associated with hydrogen bonding.lg It appears that the hydrolysis of the surface oxide is relatively rapid but that the hydration of the surface hydroxyls is a slow process. This hydration involves physical adsorption as a precursor with an orientation to each individual hydroxyl group. These extremely slow kinetics of associative adsorption from the gas phase should not be confused with the well-defined first-order kinetics observed calorimetrically2 in liquid water. The former is present for both high and low surface area materials, regardless of porosity, whereas the latter is due to diffusion into pores and is observed only when small pores (dimensions near that of a water molecule) are present. The small crystallite size (large specific surface area) and low firing temperature of sample A thwart any idealized quantitative treatment. Qualitatively, however, this material exhibits the same slow, irreversible behavior but the interactions are much more complex. This nonideal behavior is also noted in the net differential heats for this material.2b It appears that one must deal with a material which has a low specific surface (resulting from a high-temperature calcination) to obtain data indicative of an ideal crystal plane surface. The infeasibility of attempting to evaluate thermodynamic data from these isotherms is shown by the low vapor pressure of the initial adsorbed species and the prohibitive kinetics in arriving a t equilibrium states. Equally impracticable is the practice of attempting to predict the amount of adsorption from predetermined isotherms.
(15)W.S. Brey and K. D. Lawson, J . Phys. Chem., 68, 1474 (1964), (16) K.D.Lawson, Thesis, University of Florida, 1963. (17)A. Bielanski, “Catalysis and Chemical Kinetics,” Academic Press Inc., New York, N . Y., 1964,p 104. (18) A. A. Antoniou, J. Phys. Chem., 68, 2754 (1964). (19) G.C. Pimentel and A. L. McClellan, “The Hydrogen Bond,” W. H. Freeman and Co., San Francisco, Calif., 1960,pp 206-225.
