J. Phys. Chem. 1982, 86,3993-3996
3993
Interaction of CO, C02, and D2 with Rhodium Oxide: Its Reduction and Catalytic Stability P. R. Watson and 0. A. Somorjal' Materials and Molecular Research Division, Lawrence Berkeley Laboratory and Department of Chemistty, University of California, Berkeley, California 94720 (Received:January 6, 1982; In Final Form: March 23, 1962)
We have investigated the adsorption and interaction of CO, C02,and D2with Rh2O3-5H20 by thermal desorption spectroscopy (TDS).Both CO and D2 react strongly with the oxide lattice inducing decomposition during the desorption process to yield C02and D20. Adsorbed C02desorbs essentially intact. These results are compared with those found on rhodium metal formed by reducing the oxide to the metallic state. On this metallic surface C02dissociates while CO and D2 desorb without significant reaction. The use of C 0 2to maintain an oxidized metal surface in a reducing atmosphere during a catalytic reaction is discussed.
Introduction The oxides of the heavier group 8 transition metals, Pt, Pd, Rh, and Ir in general and rhodium in particular, are little studied and poorly understood systems; they have relatively poor thermodynamic stability as their oxygen dissociation pressures reach lo4 torr below loo0 K.lt2 The most common and relatively stable oxide of rhodium is the sesquioxide which is known in three anhydrous forms as well as the hydrated form Rh203.5H20. It is a yellow amorphous solid prepared by alkali precipitation from aqueous solution and may be more accurately represented by Rh(OH)3.rH20.1-3 In this laboratory we have made an extensive study of the surface properties of rhodium metal including the chemisorption of oxygen and the formation of ordered surface oxides? Most recently we have explored the use of rhodium catalysts for the hydrogenation of CO using both metallic rhodium5and bulk rhodium oxides.6~~The latter study has shown that both anhydrous Rho2 and Rhz03are readily reduced to the metal upon heating in vacuo or in an H2/C0 mixture. However, the hydrated sesquioxide, RhzO3.5H20,which had been dried at 450 "C for 1h, proved to be relatively stable upon heating in both vacuum and a reaction mixture of H2/C0. Furthermore, this oxide proved to have the unusual catalytic ability to produce oxygenated products during the CO hydrogenation reaction rather than methane and small concentrations of ethylene and propylene typically formed over the Rh metal c a t a l y ~ t . ~ ~ ~ The chemisorption of CO and H2 and their interaction with the lattice must be important processes during the hydrogenation of CO that also affect the stability of this oxide as a catalyst. In this paper we describe the chemisorption and reactions of CO, C02, and D2 on such an oxide surface that was investigated by means of thermal desorption spectroscopy (TDS) under ultrahigh vacuum conditions. (1)W. P. Griffiths, T h e Chemistry of the Rarer Platinum M e a Os, Ru, Ir, and Rh", Wiley-Interscience, New York, 1967. (2)F. A. Cotton and S. A. Hart, "The Heavy Transition Elements", Wiley-Halatead, New York, 1975. (3)S. K. Shukla, Ann. Chim., 1383 (1961). (4)D.G. Castner and G. A. Somorjai, Appl. Surf. Sci., 6,29 (1980). (5)B. A. Sexton and G. A. Somorjai, J . Catal., 46,167 (1977). (6)D.G. Castner, R. L. Blackadar, and G. A. Somorjai, J. Catal., 66, 257 (1980). (7)P.R. Watson and G. A. Somorjai, J. CataZ., 72,347 (1981). (8)A. Ueno, J. K. Hochmuth, and C. 0. Bennett, J. Catal., 49,225 (1977). 0022-3654/82/2086-3993$01.25/0
The adsorption and reaction of these gases on metal oxides has long been an area of active interest, particularly with regard to the H2 and CO oxidation and methanol synthesis reaction^.^.^ Relatively few of these studies, however, have been carried out on carefully characterized substrates under UHV conditions as were used here. While studies of the H2-0, and CO-O2 reactions over the platinum group metals abound,lWl5little if any such effort has been directed toward their oxides, although occasional mention of the effect of surface oxides is f ~ u n d . ~There ~J~ is mounting evidence of the importance of oxidized metal species for the catalytic activity of nominally reduced oxide-supported metal ~ a t a l y s t s . ~ J ~ J ~
Experimental Section The experiments were carried out in a stainless-steel ultrahigh vacuum chamber equipped with retarding field electron optics which were used to regularly monitor the cleanliness of the sample surface by Auger electron spectroscopy. A quadrupole mass spectrometer (UTI 100 C) was used to detect gases desorbing from the surface which was heated with a linear temperature ramp of about 20 K/s. The rhodium oxide (Rh2O3-5H20, 99.9%, Pfaltz and Bauer) was deposited from a methanol slurry onto a gold foil which provided both a physical backing for the powder sample and a means of heating. The foil was heated resistively, the temperature being measured by an alumelchrome1 thermocouple attached to the foil. The oxide sample was heated to 450 " C for 15 min to remove water of crystallization. Auger spectra taken after this treatment showed that only partial reduction of the oxide occur^.^ The surface area of the oxide, before desorption, was estimated at 10 m2/g by the BET method. The CO (Matheson 99.95%) was passed through a dry-ice/acetone bath to remove carbonyls; the D2 (Liquid Carbonic) was used without further purification. D2 was employed for the TDS experiments to avoid the compli(9)H.H.Kung, Catal. Rev., 22, 235 (1980). (10)F. P.Netzer and G. Kneriger, Surf. Sci., 51,526 (1975). (11)J. T. Yates, P. A. Thiel, and W. H. Weinberg, Surf. Sci., 82,45 (1979). (12)T. Engel and G. Ertl, J. Chem Phys., 69,1267 (1978). (13)J. L.Taylor, D. E. Ibbotsen, and W. H. Weinberg, Surf. Sci., 90, 37 (1979). (14)C. T. Campbell, S-K. Shi, and J. M. White, Appl. Surf. Sci., 2, 382 (1979). (15)M. A. Barteau, E. I. KO, and R. J. Madix, Surf. Sci., 104, 161 (1981). (16)M. Primet and E. Garbowski, Chem. Phys. Lett., 72,472 (1980). (17)C. A. Rice, S. P. Worley, C. W. Curtis, J. A. Guin, and A. R. Tarrer, J. Chem. Phys., 74,6487 (1981).
0 1982 American Chemical Society
3904
The Journal of Physical Chemistty, Vol. 86, No. 20, 1982
Watson and Somorjai
TABLE I: Temperature of the Maximum Rate of Desorption (T") for the Desorption of Various Species after Adsorbing D,, CO, and CO, on Rhodium Oxide (Fresh and Reduced) Compared with R h ( l l 1 )
T,,,
rhodium oxide
D,b
D,Ob
cob
CO,b
cob
CO,b
low
270 270
225 160, 225
190 24 5 300
high
40
170. 3 0 0 170; 300 165 170 250 140
165. 3 0 5 1 6 5 ; 270
low
115 115 < 60 < 50 75
low high
Rh( 111)' a Adsorbed gas.
CO,"
exposure high
reduced rhodium oxide
5. "C
c 0"
4 " substrate
f
Desorbed species.
References 19-24.
CO/CO/oxi de Amu 28
CO exposure (L)
, 3
200
400
0 T e m p e r a t u r e CQC)
200
400
400
Flgure 1. Thermal desorption spectra (TDS) of D, and D,O desorbed from dried Rh,O,.SH,O after exposure (measured In langmuirs) to D., The notation should be read desorbing gas/adsgaskubstrate in this and other diagrams. The mass spectrometer slgnal is in arbitrary units but is conslstent from one diagram to the next.
cation of residual H2 in the vacuum chamber. The mass spectrometer reading were corrected where necessary for differences in ionization efficiency, electron multiplier gain, and relative transmission.
Results and Discussion TDS spectra for various exposures of D2 to the oxide surface at room temperature are shown in Figure 1. The temperature of the peak maxima is listed in Table I together with relevant data from desorption from rhodium The major desorbing species is D20which desorbs over a rather broad temperature range centered near 270 "C. A small amount of D2desorbs at 115 "C.As the desorption temperature of D2 does not seem to correlate with D20 desorption and the cracking fraction of D2from D20 in the mass spectrometer is very small, less than 1%,the Dz desorption peak represents D2desorbing from the surface. We can relate the areas under the Dz and D20 TDS peaks, AD, and AD 0,to the relative number of desorbed molecules N D 2 and NDzo byz6 ND~/NDzO = c(sD~/sD,O)(AD~/AD,O) (1) (18)C. T. Campbell and J. M. White, J. Catal., 64, 289 (1978). (19)R. A. Marbrow and R. M. Lambert,Surf. Sci., 67, 489 (1977). (20)D. G. Castner, B. A. Sexton, and G. A. Somorjai, Surf. Sci., 71, 519 (1978). (21)D. G. Castner and G. A. Somorjai, Surf. Sci., 83,60(1979). (22)D. G. Castner, L. H. Dubois, B. A. Sexton, and G. A. Somorjai, Surf. Sci., 103, L134 (1981). (23)P.A. Thiel, J. T. Yates, Jr., and W. H. Weinberg, Surf. Sci., 82, 22 (1972). (24)P.A.Thiel, E. D. Williams, J. T. Y a k , Jr., and W. H.Weinberg, Surf. Sci., 84,54 (1979). (25)W. Hotan, W. Gopel, and R. Haul, Surf. Sci., 83, 162 (1979).
200
400 T e m p e r a t u r e C'C)
200
400
Flgure 2. As the prevlous figure, but for CO and CO, desorbing after exposure to CO.
where S, and S are the relative pumping speeds for the two molecules. T h e ratio (SD,/SD,O) has been measured to be about 13.LZ6 The proportionality constant C describes the relative sensitivity of the mass spectrometer to the two species and was measured to be close to 0.7. From eq 1and the desorption data of Figure 1,we readily obtain the result that -40% of the Dz adsorbed reacts to form D20. The shape of the D2 desorption peak and the constant peak temperature, as the coverage is changed, suggest second-order kinetics that do not involve lateral interactions. As shown in Table I, the desorption temperature of D2is very high compared to that desorbed from rhodium meta1.20v21It is not likely, therefore, that this desorption arises from a small metallic component of the surface. Presumably, deuterium adsorbs dissociatively to form hydroxyl species, as is observed by infrared spectroscopy on many oxide system.25 On heating, the majority of the hydroxyl groups react to form D20,resulting in a reduction of the surface, while a small fraction re-form Dp The very different shapes of the desorption curves for D2and D20suggest that the Dz arises from a small number of favored sites that allow for easy dissociation of 0-H bonds. When D2 desorption has been completed, DzO desorption has only just begun, and the broad nature of its desorption curve extending over a range of 300 O C (26)S.Ferrer and G. A. Somorjai, Surf. Sci., 97, L304 (1980).
Interaction of CO, C02, and D2 wtttl Rh203-5H20
The Journal of phvsical Chemistty, Vol. 86, NO. 20, 1982 3005
I
'reduced' oxide
25
L CO
exposure
C02/C02/oxide
CO/COp/oxide
Amu 44
Amu 28
C 0 2 exposure (L)
Temperature ("C)
Flgure 3. TDS spectra for CO (amu 28) and CO, (amu 44) desorblng from a reduced sample of rhodium oxide after exposure to 25 L of CO and
cop
suggests that transport of the reacting species, rather than reaction or desorption kinetics, is the rate-limiting factor. In Figure 2 we present TDS spectra for CO adsorption with peak temperatures again listed in Table I. In this case, CO is detected desorbing at about 225 "C at low exposures with a second peak which grows in with increasing exposure at -160 "C. C02 desorbs in a single peak which has a maximum at about 190 "C at low coverages which shifts to about 245 "C at higher exposures. We can estimate the amount of CO that reacts to form COz by an analogue of eq 1. In the case of CO/CO2,we are helped considerably by the fact that the relative sensitivities (C)and the ratio of pumping speeds are both close to unity. Allowing for about 6% cracking of COPto CO in the ionizer of the maas spectrometer, we find that about 75% of the adsorbed CO reacts to form C02. The CO TDS spectra of Figure 2 resemble those reported for CO desorption from metallic rhodium foils5J8 and c r y ~ t a l s . 'In ~ ~these studies, CO was shown to desorb at -230 "C at low coverages and close to 200 "C at higher exposures. At very high exposures, a second peak appears below 100 0C.20i21The resemblance of the data we have obtained for CO desorption from rhodium oxide to the published data for the metal leads us to believe that much of the CO is desorbing from patches of metallic rhodium. In Figure 3 we show TDS spectra for CO and C 0 2that has been adsorbed on an oxide reduced to the metallic state by heating at 800 "C, in vacuo, for 30 min. This reduction was observed visually and by Auger electron spectroscopy. As shown in this figure and Table I, CO desorbs intact at about 200 "C, in good agreement with results obtained by other workers for metallic rhodium;5,1&24 a small CO desorption peak centered around 300 "C may be due to residual surface oxygen. On the other hand, on this reduced surface, C02 is completely dissociated, desorbing solely as CO, again in good agreement with other reports.22 Figure 4 displays the TDS spectra for COz adsorbed on a rhodium oxide sample. In contrast to the data of Figure 3, COz desorbs intact at 170 "C in a quite symmetric peak with a small shoulder at 280-305 "C. The small amount of CO that also desorbs is due to the dissociation of C02 on the rhodium metal present in the oxide and is equiv-
0
0 Temperature ("C)
Flgure 4. Same as Figure 2 but with C02 as adsorbate.
dent to about 15% of the total desorbed species. Thus adsorbing CO on the rhodium oxide leads to the oxidation of the majority of the molecules to C02 (-75%) on heating, while the adsorption of C 0 2 leads to mainly desorption of C02with some 15% being dissociated to CO. This is due to the metallic content of the surface as rhodium metal dissociates COz and associatively adsorbs CO while the oxide adsorbs C02 and oxidizes CO. Thus we suggest that CO or C02adsorption could be used to titrate the degree of reduction of the oxide surface. Indeed, if many CO adsorption experiments are performed, the surface becomes progressively reduced as shown by a steady increase in the ratio of CO to C 0 2 produced on desorption. The C02TDS spectrum from CO adsorption (Figure 2 and Table I) has a very different shape and coverage dependence from the C02TDS spectrum from C02 adsorption of Figure 4. Thus the peak temperature in the C02/C02case is independent of exposure at 165 "C,while the peak of the CO2/COTDS spectrum varies from 190 "C to higher temperatures as the exposure is increased. This implies that the rate-determining step in the formation of C02from CO is not desorption of C 0 2from the surface as adsorbed C02 desorbs below the temperature of C02 production from CO at all exposures. The adsorption of CO on oxides has been interpreted to proceed through both adsorbed carboxylateB entites and carbonate species formed through a reaction with lattice oxygen.a Both species have been detected by IR spectroscopy depending upon the oxide and the conditions of adsorpti~n.~~~~ Implications for Catalysis We have shown elsewhere' that Rh203 prepared by drying the hydrated oxide is an effective catalyst for the conversion of mixtures of CO and H2 to oxygenated hydrocarbons. However, under reaction conditions (6 atm of 1:l H2/C0, 300 "C)the oxide is metastable, eventually reducing to the metal after many hours of use. The present work shows that reduction by both CO and H2 are likely pathways for this decomposition process. It is probable that the active catalyst surface consists of patches of oxidized rhodium that are necessary for the
J. Phys. Chem. 1982, 86, 3996-4003
3996
production of oxygenates and reduced metal.' Our results for CO adsorption indicate that if reaction to produce C02 is a chief mode of reduction of the oxide, then concomitant dissociation of CO, on metallic patches on the surface can serve to reoxidize the surface. Therefore, C02 formed during the synthesis reaction, or added with the reactants as in the case for the methanol synthe~is,~ may perform a valuable function in keeping at least a portion of the catalyst surface oxidized and active for the synthesis of oxygenated products.
Conclusions We have examined the adsorption and interaction of CO, C02, and D2 with an Rhz02.5Hz0sample that had been dosed at 450 "C. Our major findings are as follows: 1. Approximately 40% of adsorbed D2 reacts to form D20 which desorbs over a broad temperature range cen-
tered at 270 "C. The D, that desorbs does so at a temperature (115 OC) that is too high to be due to adsorption on metallic patches on the surface. 2. Approximately 75% of adsorbed CO reacts to form C02which desorbs between 190 and 245 OC depending on coverage. 3. Both CO and Dz desorb without significant reaction on an oxide sample that had been reduced to the metal in agreement with literature findings. 4. Adswbed C 0 2 desorbs intact at 170 OC regardless of coverage on the oxide while it dissociateson the metal and only CO desorbs as a result.
Acknowledgment. This work was supported by the Director, Office of Energy Research, Office of Basic Energy Sciences, Materials Sciences Division of the US. Department of Energy under Contract W-7405-ENG-48.
Ionic Volumes of Electrolytes in Propylene Carbonate from Densities, Ultrasonic Vibration Potentials, and Transference Numbers at 25 OC Raoui Zana,+ Jacques E. Desnoyers,$ Gerald Perron,$ Robert L. Kay,'§ and Kuoshin Lee§ CNRS. Centre de Recherches sur les Macromolecules, 67083 Strasbourg-Cedex, France; Facult6 des Sciences, Unlversit6 de Sherbrooke, Sherbrooke, Quebec, Canah JIK 2R 1; and Department of Chemistry, CarnegkcMellon Universm, Pittsburgh, Pennsylvania 15213 (Recelved: March 12, 1982; In Final Form: May 27, 1982)
Measurements of densities, ultrasonic vibration potentials (uvp's), and transference numbers of electrolytes in propylene carbonate (PC)at 25 "C were combined to give standard partial molar volumes V(ion) of individual ions in PC. Cations and anions of the the same size have approximately the same V(ion), which is in contrast with the ionic values that would be obtained from an extrapolation of the iio(R,NI) to zero cation molecular mass, which yields much more negative values for anions than cations. The electrostriction in PC is close to that in water, ethylene glycol, ethanol, and dimethyl sulfoxide and about the same for cations and anions.
Introduction Propylene carbonate (PC) is an extremely valuable solvent for e1ectrolytes.l It has a relatively high dielectric constant (64.9), has good solvent properties for most electrolytes, and, being aprotic, can sustain metals as reactive as alkali metals. As such it is used in recently developed batteries involving lithium metal.2 It is often suggested as a reference solvent for solvation studies of electrolyte^.^ A survey of the literature shows that, with the exception of enthalpies of solution,4 the solution properties of electrolytes in PC are poorly known. For instance, the limiting equivalent conductances A,, of ions in P P 7differ in some cases by more than 15%, whereas a precision of better than 0.1% is usually obtained on these quantities. Also, few reliable partial molar volumes are found in the literature.B-10 Finally, essentially no precise data on partial molar heat capacities and compressibilities are available. Our current interest in the properties of ions in nonaqueous solvents, together with the peculiar structure of the PC molecule
CNRS.
* Universite de Sherbrooke. 8 Carnegie-Mellon
University. 0022-3654/82/2086-3996$01.25/0
and the paucity of reliable volumetric and transport data on ionic solutions in PC, led us to undertake a thorough investigation of ionic volumes in a joint study involving three laboratories. The transference numbers of K+ in KClO, solutions and of Et,N+ in Et4NC104solutions were accurately determined at Carnegie-Mellon University. These data provided a basis from which the A,, of ions in PC could be obtained from the reported equivalent conductances of electrolytes. The partial molar volumes of a number of electrolytes in PC were determined at the (1) W. H. Lee, 'Chemistry of Non-Aqueous Solvents", Vol. 4, J. T. Lagoski, Ed., Academic Press, New York, 1976,Chapter 6. (2)T. Sakai and K. Teratsukaea, Jpn. Kokai Tokkyo Koho 78,101, 628 (1979);Ray-0-Vac Corp. Jpn. Ltd.,ibid., 81,08,386 (1981),Chem. Abstr., 94,216584g,(1981);Daini Seikosha Co.Ltd. Jpn. ibid., 81,01,468 (1981),Chem. Abstr. 94,216682e,(1981). (3)H.L. Friedman and C. V. Kriahnan in 'Water: A Comprehensive Treatise", Vol. 3,F. Franks, Ed., Plenum Press, New York, 1973,p 1. (4)C. V. Krishnan and H. L. Friedman in "Solute-Solvent Interactions", Vol. 2,J. F. Coetzee and C. D. Ritchie, Eds., Marcel Dekker, New York, 1976,Chapter 9. (5)L. M. Mukherjee, D. P. Boden, and R. Linduer J. Phys. Chem., 74, 1942 (1970). (6)M. Jansen and H. L. Yeager, J. Phys. Chem., 77,3089(1973);78, 1830 (1974). (7)N. Matsura, K. Unemoto, and Y. Takeda, Bull. Chem. SOC.Jpn., 48,2253 (1975). (8)M.Dack, K. J. Bird, and A. J. Parker, Aust. J . Chem., 28, 955 (1975). (9)R. Gopal, D. K. Agarwal, and R. Kumar, Z. Phys. Chem. (Frankfurt am Main), 84, 141 (1973). (IO) K. Sing, D. K. Agarwal, and R. Kumar, J. Indian Chem. SOC.,52, 304 (1975).
0 1982 American Chemical Society
