J . Phys. Chem. 1984, 88, 4127-4130
4127
Use of H,/H,O Gas Mixtures To Study Adsorption on Chromia-Promoted Magnetite at Water-Gas Shift Temperatures M. Tinkle and J. A. Dumesic* Department of Chemical Engineering, University of Wisconsin-Madison, Madison, Wisconsin 53706 (Received: November 7 , 1983)
The adsorption of hydrogen and water on chromia-promoted magnetite at 637 K was studied. Adsorption isotherms were collected with constant composition gas-phase mixtures of hydrogen and water. The H2/H20ratio was varied from 2.3 to 10.8 and the adsorption data were analyzed in terms of separate hydrogen and water isotherms. It was found that, at a given hydrogen partial pressure, a higher mole fraction of water in the gas resulted in a lower hydrogen uptake. Furthermore, hydrogen uptakes passed through a maximum as a function of hydrogen pressure, for constant H2/H20ratios greater than ca. 3. These trends indicate that hydrogen and water compete for adsorption sites and that hydrogen adsorption is dissociative on magnetite at 637 K.
Introduction A recent study',2 using gravimetric and volumetric methods probed the interactions of C O and C 0 2 with chromia-promoted magnetite at temperatures near 650 K. It was shown that the oxidation state of the magnetite surface could be varied with different CO2/CO gas mixtures (Le., by controlling the CO2/CO ratio). The effects of the surface oxygen content on the C 0 2 and C O adsorption isotherms were thus studied by using constantcomposition gas mixtures and thereby collecting two isotherms at once. The results of the study showed that, at a given C 0 2 pressure, an increase in C O pressure resulted in decreased C 0 2 and increased C O adsorption. Analogous results were obtained at constant C O pressure and varying CO, pressures. It was concluded that either the numbers of sites for C 0 2 and C O adsorption vary as the surface oxygen concentration varies or competitive adsorption between CO, and C O occurs. To further investigate the interaction of water-gas shift species with magnetite, we used mixtures of H2and H20in adsorption experiments. These results are presented here. Experimental Section The basic experimental procedure previously used for the CO2/CO adsorption isotherms' was used for collection of H2/H20 isotherms. The catalyst temperature of 637 K was chosen (a typical water-gas shift temperature), and the same equipment was used with the addition of heating tape and insulation to prevent the condensation of water. As for the CO2/CO studies, chromia-promoted magnetite (1.4 g, supplied by Haldor Topsere A/S) was the catalyst considered. To initialize the oxidation state of the catalyst surface, we reduced the sample overnight in a flowing 85% C02/15% CO gas mixture (Matheson, premixed) at 653 K. Following this, the catalyst was evacuated for 1 h at 653 K, as recommended elsewhere.' Following the evacuation, the catalyst was dosed with a C 0 2 / C 0 mixture having an oxygen activity equivalent to the H2/H20mixture under investigation according to the relationship Pco,/Pco = KwGs(PH~o/PHJ (1) where P, is the partial pressure of component i and KWGsis the water-gas shift equilibrium constant. According to gravimetric studies,2 the surface of chromia-promoted magnetite has the same oxygen content in a C 0 2 / C 0 mixture as it has in the corresponding H2/H20 mixture for the range of compositions studied. The initial pressure of this CO2/CO treatment was close to 20 kPa, and the equilibrium pressure (after 4-6 h) was ca. 7.5 kPa. Finally, the gas surrounding the catalyst was expanded into a 5000 cm3 bulb for 2 h. The final pressure after expansion was about 0.01 kPa. The temperature was reduced to 637 K following expansion. The cell was then isolated, the manifold evacuated, (1) Kubsh, J. E.; Chen, Y.; Dumesic, J. A. J. Caral. 1981, 71, 192. (2) Kubsh, J. E.; Dumesic, J. A. AICfiE J. 1982, 28, 793.
0022-3654/84/2088-4127$01.50/0
TABLE I: Adsorption of Hydrogen and Water on Chromia-Promoted Magnetite initial
H2/H20 ratio
final H, press., kPa
OH;
final H,O press., (Pa
OH200
10.8
3.32 8.94 13.90 19.20 22.39
0.1087 0.1414 0.1730 0.1873 0.1993
0.435 0.849 1.249 2.004 1.900
0.0077 0.0104 0.0151 0.0150 0.0247
5.7
2.93 8.27 12.87 17.72 20.39
0.1179 0.1285 0.1290 0.1228 0.1153
0.848 1.115 1.408 2.195 2.730
0.0149 0.0243 0.0392 0.0531 0.0680
3.5
3.17 9.16 11.93 18.74 19.62
0.0915 0.0982 0.0995 0.0738 0.0667
0.861 1.556 3.264 3.613 4.968
0.0352 0.0561 0.0646 0.1071 0.1260
2.3
2.37 7.52 10.90 13.78 18.15
0.0842 0.0836 0.0752 0.0600 0.0560
1.477 1.873 2.657 4.213 6.704
0.0280 0.0495 0.0763 0.0968 0.1147
Moles adsorbed per mol of Nzin the BET monolayer and the first dose of H2/H20gas admitted to the cell. It was assumed that the amount of C 0 2 and C O in the system was negligible with respect to the hydrogen and water present for each isotherm. The H2/H20adsorption isotherms for mixtures with H 2 / H 2 0 ratios of 10.8, 5.7, 3.5, and 2.3 were measured at 637 K. The maximum dosing pressures used were about 25 kPa. The appropriate H2/H20mixtures were prepared by passing hydrogen (purified by passage through a palladium thimble purifier) through a heated saturator containing double-distilled water. The initial and final gas-phase composition were determined after each dose of H2/H20by the freezing method described by Kubsh et al.' Adsorption equilibrium for each point on the isotherm was attained in 5-7 h. Since the initial and final pressures and compositions of the gas mixtures were known, the H2and H20 uptakes could be calculated separately.
Results Data for the adsorption of hydrogen and water on chromiapromoted magnetite are shown in Table I. Coverages '6 are given in moles adsorbed per mole of N2 in the BET monolayer (357 pmol/g of catalyst). Total uptakes of hydrogen and water as a function of final pressure are plotted in Figures 1 and 2. At a given H2 partial pressure, the H2uptake decreases as the fraction of H 2 0 in the gas phase increases. Because the water partial 0 1984 American Chemical Society
4128
o'201 Y Nt Y l 0.15
0.10
0.05
0 0
12
8
4
20
16
24
P H 2 , kPa
Figure 1. Hydrogen adsorption on chromia-promoted magnetite a t 637 K. X, 0,A, and represent data collected at H 2 / H 2 0ratios of 10.8, 5.7, 3.5, and 2.3, respectively.
o.20 0.15
-
Tinkle and Dumesic
The Journal of Physical Chemistry, Vol. 88, No. 18, 1984
t1
1
c d
which are consistent with data from infrared studies. Hydrogen is believed to interact with surface anions to form O H groups. This can occur either through reductive adsorption on surface oxygen species*5 or through heterolytic adsorption on anion-cation pair sites.5-'0 This latter mode of adsorption also involves the binding of an equivalent number of hydride ions on coordinatively unsaturated cations. Both modes of adsorption can be associative or dissociative. Water may interact with anion vacancies on the surface to form a molecularly adsorbed species. Alternatively, it could interact with an anion-cation pair site to form either an associatively or a dissociatively adsorbed specie^.^,"-'^ For the case of hydrogen and water coadsorption studies, it must also be considered that dissociatively adsorbed hydrogen and water may, in fact, behave as equivalent species. Furthermore, more than one type of adsorbed hydrogen and/or water may exist. In the following discussion, Langmuir-type models are used to describe the adsorption data for H 2 and H 2 0 . Coverages fIi for each species i are given in moles adsorbed per mole of N, in the BET monolayer. Noncompetitive Model. The simplest model to be considered is analogous to the noncompetitive model which was found to be consistent with data for adsorption of CO2/CO mixtures.' For H2/H20mixtures, it is reasonable to assume that H 2 0 adsorbs on surface cations (Le., anion vacancies, "*") and H 2 on surface anions ("0* ") :
H20+ P I "
y
I
I
I
1
I
I
2
3
4
5
6
7
PHto, kPa
Figure 2. Water adsorption on chromia-promoted magnetite a t 637 K. X, 0,A, and 0 represent data collected a t H2/H20ratios of 10.8, 5.7, 3.5, and 2.3, respectively.
pressures are small, the measurements of H 2 0uptake are subject to greater error. However, the general trend that H 2 0 uptake increases as the fraction of water in the gas phase increases is apparent. The decreased adsorption of H2at higher pressures is an important finding of this investigation. In an attempt to explain this effect, it should be mentioned that although the final compositions along each isotherm are not exactly constant, no general trend in compositional change exists. After collection of the isotherm for a given H 2 / H 2 0 gas mixture, the cell was isolated and the manifold evacuated. The gas content of the cell was then allowed to expand into a 5000-cm3 volume for 2 h. A second isotherm using the same gas mixture was then collected. For all four compositions studied, the water adsorption was found to be reversible within experimental uncertainty. However, a significant portion of the hydrogen uptake (0.05 moles per mol of N2in the BET monolayer) was found to be irreversibly adsorbed. In contrast, CO, and CO adsorption on magnetite are almost completely reversible at this temperature.' Finally, it should be noted that the surface coverage by reversibly adsorbed hydrogen passed through a maximum with increasing hydrogen pressure (at constant H2/H20ratio), as was described above for the behavior of the total hydrogen coverage.
*
2
HzO**
This model for adsorption does not explain the data shown in Figures 1 and 2. Specifically, it does not explain the decreased adsorption of hydrogen at higher water pressures. For CO2/CO mixtures, the apparent competition between C 0 2 and CO was explained by a variation in the number of anion and cation sites with changes in the CO2/CO gas-phase ratio.' As the CO2/CO ratio increased, the surface oxygen content increased, providing more sites for C 0 2 adsorption; thus, enhanced CO, and suppressed CO adsorption occurred at higher CO2/CO ratios. For H 2 / H 2 0 mixtures, however, this argument does not explain the data. According to gravimetric data, as the H 2 / H 2 0 ratio decreases, the surface oxygen content increases. If H2 adsorbed on surface oxygen, then the H2adsorption uptake should be greater at lower H 2 / H 2 0 ratios. This is contradictory to the observed effect; therefore, it can be concluded that H2/H20adsorption does not follow the simple model analogous to the noncompetitive model for CO,/CO adsorption. Competitive Models. The decrease in hydrogen uptake at higher water contents implies that competition may be occurring between these species. The simplest competitive adsorption would assume that both H2and H 2 0 adsorb on the same site. Physically, this would occur if, for example, H, and H 2 0 each adsorbed on anion-cation pairs. If the adsorption site is represented by LY and adsorption occurs as
(4)
Discussion Models for Adsorption of Hydrogen and Water. The adsorption of hydrogen and water on oxides in general has been studied e~tensively.~-'~ Modes of adsorption have been suggested
(5)
(3) Burwell, R. L., Jr.; Haller, G. L.; Taylor, K. C.; Read, J. F. Adu. Catal
(9) Griffin, G. L.; Yates, J. T., Jr. J. Chem. Phys 1982, 77, 3744. (10) Ito, T.; Murakami, T.; Tokuda, T. J . Chem. Soc., Faraday Trans. I 1983, 79, 913. (11) Knazinger, H. Adu. Catal. 1976, 25, 184. (12) Morimoto, T.; Yokota, Y.; Nagao, M. J . Colloid. Sci. 1978,64, 188. (13) Kurtz, R. L.; Heinrich, V. E. Phys. Reu. B: Condens. Matter 1982, 26, 6682. (14) Debnath, N. C.; Anderson, A. B. Surf. Sci. 1983, 128, 61. (15) Udovic, T. J. Ph.D. Thesis, University of Wisconsin-Madison, 1982.
1969, 20, 1.
(4) Burwell, R. L., Jr. NES Spec. Publ. 1970, No. 455, 155. (5) Yamazoe, N.; Fuchigami, J.; Kishikawa, M.; Seiyama, T. Surf. Sci. 1979, 86, 335.
(6) Kokes, R. J.; Dent, A. L. Adu. Catal. 1972, 22, 1. (7) Eischens, R. P.; Plisken, W. A,; Low, M. L. D. J. Catal. 1962, 1 , 180 (8) Coluccia, S.;Bocuzzi, F.; Ghotti, G.; Morterra, C. J . Chem. Soc., Faraday Trans. 1 1982, 78, 21 11.
The Journal of Physical Chemistry, Vol. 88, No. 18, 1984 4129
Adsorption on Chromia-Promoted Magnetite
150
3001t
125
250
x
2 100
/
*0°
"a a= cis
75
50
25
Figure 3. Langmuir-type isotherms for competitive adsorption of H, and H20(eq 8) on chromia-promoted magnetite at 637 K. X, 0, A, and 0 represent data collected at H2/H20 ratios of 10.8, 5.7, 3.5, and 2.3, respectively. The slope and intercept of the line correspond to K H =~ 0.5 kPa-' and a saturation coverage of 0.19 mol adsorbed per mol of N2 in the BET monolayer.
Figure 4. Langmuir-type isotherms for competitive adsorption of H2 and H,O (eq 9) on chrcmia-promoted magnetite at 637 K. X, 0, A,and 0 represent data collected at Hz/H20 ratios of 10.8, 5.7, 3.5, and 2.3, respectively. The slope and intercept of the line correspond to KH20= 0.3 kPa-' and a saturation coverage of 0.23 mol adsorbed per mol of N2 in the BET monolayer.
(all measured in moles per mol of N2 in the BET monolayer), then Langmuir-type expressions can easily be derived. For example
0
70 'Ob
A
_\N 0
4 40
30 x
20
' 0
-PHzO =-
1
OH1O
KH20%at
+-1
%at[
2
4
6
8
x
10
12
14
16
18
PH~o~H~ OH20
+
pH.')
(')
The data are plotted according to eq 8 and 9 in Figures 3 and 4. The plots show a good fit, and the slopes and intercepts give consistent positive values for asat,K H ~and , KH~o.Similar results are obtained if the reversible amount of hydrogen adsorption is plotted according to eq 8 and 9. In general, the above model explains the observed competition between adsorbed hydrogen and water, e.g., the hydrogen coverage decreases with increasing water pressure at constant hydrogen pressure, as seen in Figure 1. However, this model cannot explain the maxima in hydrogen uptake with increasing pressures seen in Figure 1. The maximum in the hydrogen isotherm can be described by a competitive Langmuir model where hydrogen adsorption and desorption on some site a are second order:
where kiand k{ are the rate constants for adsorption and desorption of species i. The hydrogen uptake OH1 is described by
where Ki is the adsorption equilibrium constant for species i.
Figure 5. Langmuir-type isotherms for competitive adsorption of H2 and H 2 0 (eq 13) on chromia-promoted magnetite at 637 K. X, 0, A, and tl represent points collected at H2/H20 ratios of 10.8, 5.7, 3.5, and 2.3, respectively. The slope and intercept of the line correspond to KH2 = 2 kPa-' and a saturation coverage of 0.32 mol adsorbed per mol of N2 in the BET monolayer.
Linearized forms of the hydrogen and water isotherms are as follows:
rIt can be seen from eq 12 that, at higher pressures, the KH#H~O term would increase faster than the (KH2PH2)"' term, thereby giving a decrease in hydrogen adsorption. Plots of the data according to eq 13 and 14 are given in Figures 5 and 6. A similar result is obtained if only reversible hydrogen adsorption is considered. The isotherms for the four different H2/H20ratios do not superimpose completely in the plot of eq 13, giving a range for the value of asatfrom 0.23 to 0.33 mol/mol of N2in the BET monolayer. In addition no site a can be imagined which is physically reasonable and which would exactly fit this model. However, mobile dissociative adsorption by hydrogen on anioncation pairs accompanied by immobile water adsorption, also on anion-cation pair sites, would probably follow a similar isotherm.16
4130
The Journal of Physical Chemistry, Vol. 88, No. 18, 1984
Tinkle and Dumesic o,20
r
0
-/e
----_-_-----
I
I
I
I
I
1
I
1
2
3
4
5
6
7
KH2 pH2
Figure 6. Langmuir-type isotherms for competitive adsorption of H2and H20 (eq 14) on chromia-promoted magnetite at 637 K. X, 0, A, and 0 represent p i n t s calculated for data collected at H2/H20ratios of 10.8, 5.7, 3.5, and 2.3, respectively. The slope and intercept of the line correspond to KHlo = 0.11 kPa-’ and a saturation coverage of 0.46 mol adsorbed per mol of N2 in the BET monolayer.
Another model using dissociative hydrogen adsorption can also explain the decreased hydrogen adsorption at higher pressures. Consider the associative adsorption of water on anion-cation pair sites accompanied by dissociative hydrogen adsorption on two anion sites. If the water adsorption is associative, the following expressions can be used to develop a Langmuir-type expression:
Figure 7. Simulated isotherms for water adsorption obtained from eq 21. For both curves, Oo*sa, = 0.3 and Oamt = 0.4. ab = 3 for the solid line and a@ = 6 for the dashed line.
0.05
0.02 0.01
t~
1 t-
-I
n
0
H20
+ O* + * % ’k”p H20.0*.*
(16)
The non-Langmurian deviations introduced by requiring that the anion and cation of the anion-cation pair site be adjacent for associative adsorption will not be considered.16 For this model OO*
=
e.
- O H 2 0 - 20H2
(17)
eH10
(18)
= 08,, -
1
2
3
4
5
6
7
Figure 8. Simulated isotherms for hydrogen adsorption obtained from eq 22. For both curves, Ooasat = 0.3 and O.,, = 0.4. a0 = 3 for the solid line and a@ = 6 for the dashed line.
Equations 16 and 17 can be rearranged, solving for OHIO and as follows:
OH2
where Bo. and 0, represent surface anion and cation sites, respectively, and OH10 and OH2 the measured uptakes of H20 and H2 (all in moles per mol of N2 in the BET monolayer), the following Langmuir-type expressions are obtained: PH200Hl P H ~ ~
-
KH2IJ2
1 (pH200H2)
+- ~2KH200‘,,t ~ ~ H ~ o p ~ 0*8at
~
(19) ~ /
KH2 and KH20are the equilibrium constants for adsorption of H2 and H20, respectively. The data plotted according to these expressions for either reversible or irreversible hydrogen and total water adsorption give OOqmt as a function of gas-phase composition; but the highest value of Bo.,t (Le., smallest slope) was obtained for the isotherm with the lowest mole fraction of water. According to the gravimetric studies: the surface under such an atmosphere would have the lowest surface oxygen content. An important note, however, is that, as for other models including dissociative hydrogen adsorption and desorption, a maximum in the hydrogen adsorption isotherm vs. pressure is successfully predicted. (16) Tinkle, M.; Dumesic, J. A. J . Phys. Chem. 1983, 87, 3557.
where y = PH20/PH, and 0 = KHlo/KHl.These relationships are plotted in Figures 7 and 8. ~ Conclusions The application of Langmuir-type models to volumetric data for coadsorption of hydrogen and water on chromia-promoted magnetite adds to the understanding of these complex adsorption processes. Although none of the models presented here completely describes the system, several general conclusions can be drawn. First, since hydrogen adsorption is suppressed in the presence of water, adsorbed hydrogen and water must compete for adsorption sites on the surface. Second, dissociative hydrogen adsorption and competition with water for adsorption sites results in a maximum in the hydrogen adsorption isotherm. Finally, the data suggest that the differences in hydrogen and water adsorption at different H2/H20 compositions are not directly related to the variation in surface oxygen content as a function of gas-phase composition. Acknowledgment. We acknowledge the National Science Foundation for providing a Graduate Fellowship to M.T. and also for partial financial support of this work. Registry No. H2, 1333-74-0; HzO, 7732-18-5; chromia, 1308-38-9; magnetite, 1317-61-9.
