3176
J . Phys. Chem. 1986, 90, 3176-3183
Low-Temperature CO Adsorption and State of Molybdena Supported on Alumlna, Titania, Ceria, and Zirconia. An Infrared Spectroscopic Investigation M. 1. Zaki,+ B. Vielhaber, and H. Knozinger* Institut fur Physikalische Chemie, Universitat Miinchen, 8000 Miinchen 2, West Germany (Received: October 16, 1985)
Supported molybdena catalyst samples were obtained by calcination of A1203,TiO,, CeO,, and ZrO, supports impregnated with paramolybdate anions (at loadings near the monolayer capacity of each support) at 773 K for 2 h. Oxidation (O,, at 673 K), reduction (H2,at 573-773 K), and sulfidation (H2S/H2, at 623 K) of the calcination products were carried out in situ. Then, IR carbonyl spectra were recorded for CO adsorption at 77-298 K on the various catalyst samples. The vMd4 vibrational frequencies observed were used in characterizing the state of Mo in each sample. Oxidized molybdena was shown to spread almost unrestrictedly on Al,O,; coordinatively unsaturated Mo sites are not exposed on this surface. On Ti0, and CeO,, the spreading of molybdena seemed to be notably restricted. Coordinatively unsaturated Mo(V) and Mo(1V) are respectively exposed on these materials and chemisorb CO. In comparison with A1203,Ti02 and CeO, were found to facilitate reduction of the supported molybdena to coordinatively unsaturated Mo sites in oxidation states less than or equal to +4. Upon sulfidation, however, reduction was very much improved on all the supports used (particularly on A1203)except ZrO,, on which molybdena exhibited a persistive H2- and sulfur-insensitive behavior. The potential of CO as a probe for the surface microstructure of supported molybdena catalyst was demonstrated.
Introduction Infrared vibrational frequency shifts of diatomic surface species (preferably CO, NO, and 0,) are extensively used to elucidate the nature of catalytic and/or adsorption sites of solid surfaces.'q2 A number of plausible arguments reported earlier3 have favored C O as a probe molecule for molybdate catalysts since it is more easily accessible to molecular spectroscopy than 0, and more selective and stable than NO. Its relatively weaker adsorption on these surfaces can be compensated for by applying a lowtemperature regime (Le., below room t e m p e r a t ~ r e ) . ~ The infrared-monitored vibrational frequency shifts of surface carbonyl species have often been interpreted as only being due to changes in the MC-0 bond strength. ( M stands for the coordinatively unsaturated adsorption site.) Two opposing electron interactions determine the C-0 bond strength and, hence, the carbonyl stretching frequency: namely the 0 donation (highfrequency shift) and the r back-donation from the adsorption site into the 27r antiboding orbital of CO (low-frequency shift). In addition, the observed carbonyl frequency can be influenced by mechanical coupling between the stretching vibrations of MC-0 and M-CO and by static lateral interactions' via which a "chemical" inductive effect is expected to reduce the electronaccepting power of the adsorption site as the adjacent sites get occupied. These two influences would lead respectively to highand low-frequency shifts of the MC-0 stretching mode. Their contributions to the observed carbonyl stretching frequency have not usually been considered quantitatively.*s9 In the present investigation, infrared carbonyl spectra of lowtemperature C O adsorption are used, after careful consideration of various possible competitive contributions to the observed frequency shifts of the MC-0 stretching vibration, in characterizing the state of molybdenum on the surface of oxidized, reduced, and sulfided A1203-supported catalysts. These materials develop extremely interesting catalytic properties'O and, hence, are of prime industrial importance. This investigation is also extended to similar catalyst systems with TiO,, CeO,, and ZrO, as supports in the hope of revealing any role undertaken by the support surface in modifying the surface state of the supported molybdena. Experimental Section Catalyst Preparation. Various molybdena-based catalyst samples were obtained by calcination of dried (293 K, 15 h) *To whom correspondence should be addressed. Permanent address: Chemistry Department, Faculty of Science, Minia University, El-Minia, Egypt.
0022-3654/86/2090-3176$01 S O / O
paramolybdate-impregnatedy-A1203,Ti02, CeO,, or ZrO, at 773 K for 2 h. Impregnation was carried out by the incipient wetness method from solutions (at pH 6) containing the amount of paramolybdate required for loadings near the monolayer capacity of each respective support [namely, 12.0 (for A1203),4.07 (TiO,), ' MOO, (ZrO,)]. Preparation pro6.78 (CeO,), and 4.76 wt % cedures, surface areas, and theoretical monolayer capacities for the supports have been given in detail earlier.]' In situ oxidation, reduction, and sulfidation of thin (20 f 2 mg/cm2) self-supporting wafers of the calcination products were carried out adopting the procedures given in Table I. The gases used in these pretreatment procedures, as well as CO, were (99.998%) Linde products. The gases were dried and deoxygenated by passing them through the appropriate Oxisorb traps before use. Wafers of the support materials were also subjected to the same pretreatments (Table I) for comparison purposes. For convenience, the catalyst and support samples are indicated in the text by the pretreatment experienced. Thus, e.g., Ox-MoAI and Ox-AI denote respectively oxidized A1203-supported molybdena and the oxidized A1203support at 673 K; Rd-MoTi (573) indicates TiO,-supported molybdena reduced at 573 K, whereas Sd-MoCe signifies Ce02-supported molybdena sulfidied at 623 K. Infrared Investigation. The procedure followed in carrying out (in situ) low-temperature C O adsorption experiments (at 77-298 K) has been reported earlier.3 IR carbonyl spectra (at 2300-1 900 cm-l) were recorded at a resolution of 5.3 cm-' on a Perkin-Elmer 580B spectrophotometer using an IR cell provided with CaF, (1) Peri, J. B. In Catalysis-Science and Technology; Anderson, J. R., Boudart, M., Eds.; Springer-Verlag: West Berlin, 1984; Vol. 5, pp 171-220. (2) Boehm, H.-P.; Knazinger, H. In Catalysis-Science and Technology; Anderson, J. R., Boudart, M., Eds.; Springer-Verlag: West Berlin, 1983; Vol. 4, pp 38-207. (3) Delgado, E.; Fuentes, G. A.; Hermann, C.; Kunzmann, G.; Knozinger, H. Bull. SOC.Chim. Belg. 1984, 83, 735. (4) Cotton, F. A.; Wilkinson, G.Advanced Inorganic Chemistry, 4th ed.; Wiley: New York, 1980; pp 82-86. (5) Sheppard, N.; Nguyen, T. T. In Aduances in Infrared and Raman Spectroscopy; Clark, R. J. H., Hester, R. E., Eds.; Heyden: London, 1978; Vol. 5, pp 67-148. ( 6 ) Okawa, T.; Soma, M.; Bandow, H.; Uchida, K. J. C u r d 1978,54,439. (7) Tsyganenko, A. A.; Denisenko, L. A,; Zverev, S. M.; Filimonov, V. N. J . Catal. 1985. 94. 10. (8) Lavalley, J.'C.; Saussey, J.; Rais, T. J . Mol. Caral. 1982, 17, 289 (9) Griffin, G. L.; Yates, J. T. J . Chem. Phys. 1982, 77, 3751 (10) Grange, P. Catal. Rev.-Sci. Eng. 1980, 21, 135. ( 1 1) Leyrer, J.; Vielhaber, B.; Zaki, M. I.; Zhuang, Shuxian; Weitkamap, J.; Knozinger, H. Mater. Chem. Phys. 1985, 13, 301.
0 1986 American Chemical Society
The Journal of Physical Chemistry, Vol. 90, No. 14, 1986 3177
Supported Molybdena Catalysts
TABLE I: Procedures and Conditions Adopted in Pretreating the Catalyst Samples and Supports atmosphere period, Drocess (1 atm) temp, K min
oxidation reduction
stream of O2 at a rate of 50 cm'/min stream of H2 at a rate of 50 cm'/min
sulfidation stream of 1:lO mixture of H 2 at a rate of 50 cm3/min
+ HIS
673 573, 673, 773 623
subsepuent treatment
evacuation (10" Torr) at 673 K for 30 min evacuation (10" Torr) at the reduction temp for 30 min and at 298 K for further 30 min evacuation (10" Torr) at 623 K for 30 min, followed by H 2 reduction at 573 K for 30 min and then evacuation at 573 K for 60 min
60 120 90
TABLE II: Frequencies and Assignments of MoC-0 Species Produced during CO Adsorption on Molybdena-Based Catalysts Obtained under Various Pretreatment Conditions As Reported in the Literature
support
pretreatment conditions
H2 reduced H, reduced H i reduced H2 reduced H2 reduced
Si02
H z reduced at 823 K CO photoreduced at 298 K H 2 reduced at 973 K H2S/H2 sulfided at 523 K H2S/H2 sulfided at 523 K CO photoreduced at 298 K
2204 2203 2194-21 87 2190-21 85 2190-21 85 2175, 2070 2190 2181 2140, 2108 2177-2172 2100 21 10 2128, 2080, 2040
~-A1203
H2S/H2 sulfided at 523 K
21 20-2 100
7-A1203
SO, y - ~ i ~ o ~
at at at at at
770 823 770 823 973
K K
IR bands, cm-'
K K K
H2S/H2 sulfided at 673 K Si02
SO2 YmA1203
assignment
ref
MoS+(CO) MoS+ (CO) MO~+(CO)' MO~+(CO) Mo~+(CO)~ MO~+(CO) MO~+(CO) MO~+(CO) Mo4+(CO) MO'+(CO) Mo2+(CO) Mo2+(CO) Mo2+(C0)2 MoZ2+(C0) Mo*+(CO)"
3 13 3 13 1 1 3 13 13 14 1 3 13 13 14
Mob+ sites represent Moo adjoining Mo(1V) or other lower valent Mo ions.
windows, the design of which will be described in a future communication.I2 The wavenumbers obtained from the spectra are accurate to within f l cm-I.
Results and Discussion Enumeration and Analysis of Reported v M d IR ~ Vibrational Frequencies. Reported frequencies and assignments of infrared MoC-O stretching vibrations of carbonyl species observed during CO adsorption on A1203- and Si02-supported molybdena are summarized in Table 11, where the pretreatment conditions applied and the source of data are also given. The oxidation state of some of the Mo adsorption sites listed has been confirmed by means of direct surface analysis techniques, namely M O ( V ) , ' ~ JMo(1~ V),"915J6 and Mo(III)" by X-ray photoelectron spectra (XPS) and Mo(V) by ESR spectra.I8 The existence of the divalent state (Mo(I1)) has been recently suggested on the basis of I R nitrosyl spectra.I9 As can be seen from Table 11, linear carbonyl species coordinated to Mo sites of valency 1 + 3 exhibit MoC-0 stretching frequencies higher than that of the free CO molecule (2143 cm-I). In contrast, those held by Mo of valency 2143 cm-I as being due to partially positively charged CO adsorbed species. They have pointed out20that the postulation of species such as (12) Kunzmann, G.; Ertl, G.; Kn6zinger, H., to be submitted for publication. (13) Guglielminotti, E.; Giamello, E. J . Chem. Soc., Faraday Tram. 1, in press. (14) Peri, J. B. J. Phys. Chem. 1982, 86, 1615. (15) Jagannathan, K.; Srinivasan, A.; Rao, C. N. R. J . Catal. 1981,69, 418. (16) Li, Chung Ping; Hercules, D. M. J . Phys. Chem. 1984, 88, 456. (17) Cimino, A,; DeAngelis, B. A. J . Catnl. 1975, 36, 11. (18) Petrakis. L.: Kiviat. F. E . J . Phvs. Chem. 1974. 78. 2070. i19j Rosen, R. P.; Segawa, Koh-Ichi; Millmann, W.' %;Hall, W. K.J . Catal. 1984, 90, 368. (20) Harrison, P. G.;Thornton, E. W. J. Chem. SOC.,Faraday Trans. 1 1978. 74, 2707 _
61)Gardner, R. A,; Petrucci, R. H. J . Am. Chem. SOC.1960,82, 5051.
E8 t 2050
,
,
-01
,
,
0
,
,
0.1
E1 /a u
,
,
0.2
,
,
03
, I
Figure 1. Observed carbonyl stretching frequencies ( 0 )of CO adsorbed on the various Mo"+ sites indicated vs. those (0)calculated considering only the electric field strength (1 au = 51.475 V A-l) generated at the center of the CO molecule by the adsorption site (a, observed stretching frequency of the stretches of the free CO molecule (gas phase); 8, observed frequencies of vM.,mtc4 after subtracting the corresponding mechenical coupling).
CO+or C02+has been made largely without regard to the electron donor-acceptor properties of the adsorption site and that C O adsorption on ZnO was inconsistentz2with an electron transferadsorption mechanism in the formation of such charged C O species. As an alternative, they have considered the effect of strong electric fields, as was well expounded by Hush and Williams,23 on the equilibrium nuclear configuration and, hence, the vibrational frequency of CO held by a number of supported cations other than Mo ions. We here extend their analysis to supported Mo ions by plotting t h e observed stretching vibrational frequencies of linear MoC-O species (Table I) vs. those calculated considering solely the effective electric field (Ef) at the center of the CO molecule generated by the adsorption site (Figure 1). The calculations were carried out adopting the mathematical derivations reported by Hush and Williams.23 Accordingly, Ef was obtained by calculating Qe/4?rR, where Qe is the charge of the adsorption site and R is the distance (22) Taylor, J. H.; Amberg, C. H. Can. J . Chem. 1961, 39, 535. (23) Hush, N. S.; Williams, M.L. J . Mol. Spectrosc. 1974, 50, 349.
3178 The Journal of Physical Chemistry, Vol. 90, No. 14, 1986
from the metal nucleus to the center of mass of the CO molecule (approximated by the metal ionic radius24 plus half the collision diameter of C O ( u = 3.7 A)). Figure 1 shows that the values plotted do not fall on one and the same line. For MoZ+,in particular, a significant deviation is shown. This may imply the possible existence of partaking parameters other than Ev If the oxidation states assigned to the various Man+ sites were correct (Table 11), the disagreement between the observed and the calculated CO frequencies may be due to additional frequency shifts induced by the mechanical coupling between the stretching vibrations of the MoC-0 and Mo-CO bonds. Frequency shifts due to mechanical coupling were calculated by using the mathematical formulations detailed by Okawa et aL6 assuming various values near the frequency observed2sfor the uMoCO vibrations (Le., 520 cm-I). Then, frequency shifts of YhtoC-0 vibrations free of mechanical coupling contributions were obtained (by subtraction) and plotted vs. the Efvalues as also shown in Figure I . The revised linear relationship considerably reduces the disagreement between the observed and calculated frequencies except for Mo2+. The insignificant deviation still observed for Mo sites of valencies 1 + 3 (Figure 1) could well be ascribed to an opposing effect caused by static lateral interaction' between the adsorbed molecules, particularly since most of the observed frequencies (Table I) were recorded from highly CO-covered surfaces. Such a static lateral interaction, as considered by Tsyganenko et al.' to cause a chemical inductive effect, would weaken the electron-accepting power of the adsorption site after occupation of the adjacent sites, thus causing a low-frequency shift. Indeed, this is shown (Figure 1) to be true for all Man+ surface sites concerned except for Mo2+. Finally, if the results of the foregoing analysis (Figure 1) are to be considered as a further confirmation for the assignment of spectral data reported for C O adsorption on Mo sites of valencies L+3, certainly this cannot be considered to be the case for the carbonyl species formed on the sites claimed for Mo(I1). This conclusion arises from two aspects: (i) the invalidity of the purely electrostatic model as the contribution of the electron T backdonation to the bonding between the C O molecule and the adsorption site increases and (ii) doubt as to the formation of M O sites having a definite +2 oxidation state. As far as the second aspect is concerned, to the best of our knowledge no independent physical evidence has been advanced suggesting the presence of Mo sites in a divalent state on reduced or sulfided catalysts obtained under conditions similar to those summarized in Table I, particularly on A1203and Si02. In fact, the values reported for the vibrational frequency of the carbonyl species held by adsorption sites claimed to be Mo(I1) (Table 11) lie within or close to the range of frequencies (2120-2100 cm-') interpretedI4 as being due to C O held by Mo6+sites. These Mo6+sites are suggestedi4to have indefinite oxidation states in the range
