- Email: [email protected]
The relation between electronic trends and the promoter effect on catalytic activity of hydrodesulfurization catalysts
Po/yMron Vol. 5, No. l/Z, pp. 151-155, Printed in Great Britain
1986 0
0277-5387/86 S3.00 + .OO 1986 Pergzamm Press Ltd
THE RELATION BETWEEN ELECTRONIC TRENDS AND THE PROMOTER EFFECT ON CATALYTIC ACTIVITY OF HYDRODESULFURIZATION CATALYSTS SUZANNE
HARRIS
Exxon Research and Engineering Co., Route 22 East, Clinton Township, Annandale, NJ 08801, U.S.A. (Received 1 July 1985) Abstract-The
origin of the promoting effect of certain first-row transition metals on the hydrodesulfurization (HDS) activity of MoS, is not well understood. We have carried out a systematic theoretical study of the electronic structure of simple clusters which model these catalyst systems and have used the results of these calculations to establish an electronic basis for the trendsin promotion effects observed in the MoS, systems. Both Co and Ni, which serve as promoters, have the ability to donate electrons to MO, while Cu, which serves as a poison, withdraws electrons from MO. The other first-row transition metals, which have little effect on the HDS activity of MO&, do not have the ability to donate to or accept electrons from MO. Thus promotion is associated with an increase in electron density on MO while poisoning is associated with a decrease in electron density. These results are consistent with previous results for binary catalysts which related particular electronic factors to the HDS activity of the sulfide.
Recently, we described a relation between trends in the calculated electronic structure of the transitionmetal sulfides (TMS) and trends in the measured hydrodesulfurization (HDS) activity of these materials.’ The experimental trends in the activities of the unsupported TMS had been previously measured using the HDS of dibenzothiophene as a model reaction,2 and a theoretical basis for understanding these experimental trends had been initiated by studying the electronic structure of the first- and second-row TMS.3 Calculations were carried out for a group of octahedral MS:- clusters, where the transition metal M was varied systematically across both the first and second transition series, and based on these calculations it was shown1 that a relation exists betwen the calculated electronic structure of the sulfides and their activity as HDS catalysts. Several electronic factors were identified which appear to be related to catalytic activity, and these factors were incorporated into a calculated activity parameter which correlates with the observed catalytic activity of the sulfides. We have now extended this combined theoretical and experimental approach to study the
promoted or “synergic” MoS2 catalyst systems, and in this paper we show that there is an electronic basis for the promotion effect observed in MoS2-based catalysts. Although it is well known that the addition of a second transition metal such as Co or Ni to a binary sulfide such as MoS, or WS2 brings about an increase in HDS activity, the basis for this promotion effect is certainly not well understood. Work from numerous 1aboratorieP has established, however, that the promoter atoms are associated with the edge planes of the layered compounds MoS, and WS2, in close proximity to the MO or W atoms. Several studies’** also suggest that the effect of this edge promoter is electronic rather than structural (i.e. the promoter influences the quality of the active site rather than altering the number of active sites). In order to further explore the idea that the basis for promotion is electronic in nature, a systematic study of the promoting effect of the first-row transition metals V-Zn on the HDS activity of MoS, was carried out in our laboratory.g In this study the activities of unsupported MS,-MoS, (M = V, Cr, Mn, Fe, Co, Ni, Cu or Zn) catalysts in the HDS of 151
152
S.
HARRIS
DBT were measured. It was found that : (1) V, Cr, Mn, Fe and Zn have little or no effect on the HDS activity of MoS, ; (2) Co and Ni, in agreement with previous literature, act as strong promoters ; and (3) Cu acts as a poison, serving to decrease the activity of MoS,. In order to understand these effects we have carried out a systematic theoretical study of the electronic structure of simple clusters which model these promoted catalyst systems. As mentioned above, the importance of the edge planes and the proximity of the promoter and MO atoms in MO& based catalyst systems has been previously established, and we have chosen the model clusters so that their structure is consistent with this established information. Calculations were carried out for a group of clusters MOM’S”,-, where the first-row transition metal M’ was varied systematically across the series from V to Zn. Each cluster carries a negative charge (n - ) because enough electrons were included in each cluster so that each sulfur is formally S2-, the molybdenum is formally Mo4+, and the first-row metal M’ has the formal oxidation state appropriate for the corresponding metal sulfide stable under reactor conditions. The geometry of the cluster can be seen in Fig. 1. Both metals are octahedrally coordinated by six sulfurs, three of which are shared between the two metals. Thus the model cluster is composed of two face-sharing octahedra. Within each cluster, all of the metal-sulfur bond distances were taken to be equal. These distances were determined for each cluster by averaging 2.42 A, the MO-S distance in MoS,, and the M’-S distance in the corresponding first-row transition-metal sulfide. The choice of octahedral coordination around MO allowed us to compare the Mo-S bonding and MO electronic contlguration in these model clusters with that in the simple binary clusters used previously. If the first-row transition metal M’ occupies sites around the edges of MO&, an examination of a model of MoS, suggests that M’ might occupy either octahedral or tetrahedral sites and share sulfurs with an adjacent MO. Once again we chose to use octahedral coordination for M’ because it allows us to make comparisons readily with our earlier
s _..._
M~‘&+M’.-.-S
hi’= V,Cr,Mn,Fe,Co,Ni,CuorZn
Fig. 1. Geometry of a model cluster MOM’S;- used for the molecular-orbital calculations.
calculations. We verified with selected calculations using tetrahedrally coordinated M’ that a change in geometry would not change the major conclusions of our calculations using octahedrally coordinated M’. As in our previous work Is3 all the calculations on these clusters were carried out using the SCF-Xo! scattered wave method with tangent spheres.‘O The atomic sphere radii were chosen according to Norman’s criteria.” The atomic exchange parameter c1values for the regions within the metal and sulfur spheres are those of Schwarz.12 A weighted average of these atomic a values was used for the intersphere and outersphere region. In our previous reports we described in detail the electronic structure and bonding in the model octahedral first- and second-row transition-metalsulfur clusters. Figure 2(a) shows a schematic diagram of the valance energy levels calculated for those clusters. Each group of levels, except for those corresponding to the metal d-orbitals, is represented by a block which is labelled according to the atomic orbital making the major contribution to these levels. There is of course some mixing between the metal and sulfur orbitals, but this mixing is not important for this discussion. Lying at the lowest energy is a group of levels arising from the sulfur 3s orbitals. The next group of levels arises primarily from combinations of sulfur 3p orbitals. At the top of this group of sulfur levels is the It,, level, a nonbonding combination of sulfur 3p orbitals. Finally, at the highest energy (in most of the clusters) are the 2tze and 3e, levels. These levels correspond primarily to the metal 3d or 4d orbitals. We found that a major effect of varying the transition metal in these octahedral clusters is a shift in the energy of the metal d-orbitals relative to the energy of the sulfur 3p orbitals. As measured by the difference in energy between the 2tzs and It,, orbitals, it was found that from Ti to Ni in the first transition series the energy of the 3d orbitals shifts downward in energy (i.e. closer in energy to the sulfur 3p orbitals) by about 1.5 eV. This shift is expected when moving from left to right in the transition series and is con6rmed by available photoelectron spectra of the TMS.3 For the face-sharing octahedral clusters considered here, we can also draw a schematic energy level diagram showing the main features and patterns of the energy levels resulting from the calculations. It should be noted that the point group symmetry of the face-shared cluster is C,, (as compared to Oh in the octahedral clusters). Since in the C,, point group there are no triply degenerate representations, the set of 3d or 4d orbitals which in O,-symmetry transform as the t,, representation are no longer degenerate in these clusters. The remnants of the octahedral splitting between the tze- and e,-
153
Catalytic activity of hydrodesulfurization catalysts ] 4d “ei 3e, ]
3d I&”
M3d or4d 3 2bg
=I
nMS6 0)
4d ‘i2;
M’
3d “tag”
MoM’Sy@I
Fig. 2. Schematic valence energy level diagrams for :(a) an octahedral MS:- cluster, and(b) an MOM’S”,cluster made up of two face-sharing octahedra. The ordering of 3d and 4d levels is for M’ = Co or Ni.
orbitals are still apparent in these lower-symmetry clusters, however, so that we continue to label “tlg” and “e#” sets of metal d-orbitals. A schematic level diagram for a MOM’S”,- cluster is shown in Fig. 2(b). Once again, the sulfur 3s levels lie below the broader group of levels resulting from the sulfur 3p levels. The top of the sulfur 3p group of levels is once again delineated by anonbonding combination of sulfur 3p orbitals, in this case the Za, level as shown in the diagram. The really notable feature of this level diagram is the presence, above the sulfur 3p levels, of levels resulting from both the MO 4d and M’ 3d metals. Each set of five d-orbitals forms “tZg”and “e#” sets of levels, so on the diagram we see a “tzg” and “e#” set of levels for both the metal 3d and 4d orbitals. The schematic diagram in Fig. 2(b) would correspond to the case where the M’ levels lie low enough in energy that the M’ 3d “tzg” and “eg” sets of levels each lie below the corresponding MO 4d sets of orbitals but not so low in energy that the entire 3d manifold of levels lies below the 4d levels. This relative energy separation applies to most of the clusters considered here. The relative energies of these levels and the numbers of electrons and their distributions within these levels both depend on M’. We will see below that it is these quantities which are related to the ability of the 3d metal M’ to serve as a promoter. As we noted earlier, the metal 3d orbitals become more stable upon proceeding from the left to the right side of the transition series, i.e. dropping in energy relative to the sulfur 3p orbitals. This means that in these clusters, which contain both a 3d metal and MO, the metal 3d orbitals also drop in energy
relative to the MO 4d orbitals. Proceeding across the 3d series from V to Zn this drop in energy is quite large. For V, Cr and Mn the 3d and Mo4d“tZg)) sets of levels are very close in energy. Moving to the right, however, the 3d levels shift lower in energy so that the 3d and 4d sets of levels lie in the order illustrated in Fig. 2(b). Finally, for Cu and Zn the shift in energy becomes large enough that the entire set of 3d levels lies below the MO 4d levels. At the same time that the 3d levels decrease in energy, the number of 3d electrons occupying the orbitals increases. For all the clusters, MO contributes two “d”-electrons while the number of “&‘-electrons contributed by the first-row metal depends on the metal. Thus the total number of “d”-electrons in the clusters varies from four in the VMO cluster all the way up to 12 in the Zn-Mo cluster. The distribution of electrons among the metal orbtitals depends on the relative energies of the orbitals, and an examination of the metal orbital occupations reveals an important difference between the clusters containing 3d metals which are known to promote or poison the activity of MoS, and the clusters containing those 3d metals which have no effect on the activity. In the clusters containing Co or Ni, the two metals which can act as promoters, we find that Co and Ni donate electrons to MO so that MO is formally reduced relative to the MO present in MoS,. This occurs because Co and Ni provide seven and eight “d”-electrons, respectively, to the cluster. In a pure Co-S or Ni-S cluster, the 3d “e: orbitals would be occupied. In these clusters containing MO, however, the lower energy MO “tZB” orbitals are available so
154
S. HARRIS
that a net transfer of charge from Co or Ni to MO occurs. This is a result of the MO “tZB”orbitals lying energetically between the 3d “tzg” and “eg” orbitals and Co and Ni having a high eneough d-electron count that the 3d “eg” orbitals would normally be partially occupied. The electron transfer of this type only occurs because both of these effects are present. In both the Co-MO and Ni-Mo clusters, the MO is thus formally reduced relative to the MO present in MoS,. In the cluster containing Cu, a metal which poisons the activity of MO&, we find that MO is formally oxidized relative to the MO in MoS,. This occurs because, while all five of the Cu 3d orbitals now lie well below the MO 4d levels, Cu contributes only nine “&‘-electrons to the cluster. As a result there is a net transfer of electrons from MO to Cu. This is not entirely surprising because it is known that in CuS13 (which contains nominally Cu2+ and S2-) the Cu 4d levels lie low enough in energy that some of the Cus are reduced to Cu+, while some sulfurs are oxidized to Sz -. In our Cu-MO cluster the MO electrons are even more available than the sulfur 3p electrons and thus the MO is oxidized relative to the MO in MoS2. In all the other clusters (those containing V, Cr, Mn, Fe and Zn) there is no such clear transfer ofelectrons between the MO and the 3d metal, and the electronic state of MO is similar to the MO in MoS,. For these metals the relative energies of the 3d and MO 4d orbitals and/or the number of 3d electrons are not appropriate for electron transfer. In summary, when the first-row transition metal M’ is varied in these Mo-M’ clusters, the changes in electronic structure depend on changes in both the relative energies of the MO 4d and M’ 3d orbitals and the number of “P-electrons which M’ contributes to the cluster. As the 3d levels drop in energy across the first transition series, the number of 3d electrons occupying these levels increases. It is only when M’ is Co, Ni or Cu, however, that these effects combine in such a way that the electronic state of MO is formally affected. Since it is these three first-row metals which have a measurable effect on the HDS activity of MoS,, we associate their ability to reduce or oxidize MO with their promoting or poisoning effect. Not only do these results identify an electronic origin for the promotion effect, but they are also consistent with our earlier theory relating catalytic activity to several electronic factors for the binary sulfides. One of those factors is the ability of the transition metal to bond covalently, in both a sigma and pi fashion, to the 3p orbitals of sulfur. This factor serves to differentiate the 3d from the 4d transition metals, since the 3d metalscovalently bond muchless effectively to sulfur than do the 4d metals. We defined a quantity, called B, which measures the covalent contribution to the metal-sulfur bond strength and
found that a large value of B, and thus a strong covalent contribution to the metal-sulfur bonding, correlates with high activity. Another important factor is n, the number of “P-electrons in the highest occupied orbital. We saw that a large value of n also correlates with high activity. Thus the most active catalysts have large values of n and/or B. We combined these quantities to form an activity parameter “AZ”, where : A, = nB. parameter provided us with a simple quantiative relationship between the calculated electronic structure and the HDS catalytic activity of the TMS. A metal sulfide with a large value of A2 will have a high HDS activity. Our model cluster calculations indicate that in the promoted CoMo and Ni-Mo systems one of these electronic factors for MO, n, the number of electrons in the HOMO, is increased by the presence of the promoter. Since these extra electrons occupy an antibonding orbital, the MO-S bonds are somewhat weakened. Thus the factor B which enters the activity parameter and measures the relative MO-S covalent bond strength decreases in the presence of the promoter. On the other hand, the presence of a poison such as Cu has just the opposite effect. To a first approximation, for all the other systems considered here the two electronic factors n and B for MO are unaffected by the presence of the 3d transition metal. We can calculate an activity parameter for the MO portion of the model clusters treated here. This activity parameter, along with the experimental HDS activities of the real catalyst systems, is plotted in Fig. 3, and from this figure we see that the trends in activity are reproduced by the This
-800
V
CI
Mn
Fe
Co
Ni
Cu
2x1
Metal(M) in Ms,-M0s,
Fig. 3. Calculated activity parameter for each mixed-metal sulfide system (right-hand scale). Shown for comparison, using the left-hand scale, are the measured HDS activities.g
Catalytic activity of hydrodesulfurization catalysts trends in the value of the activity parameters. The qualitative correlations obtained here thus indicate that the electronic factors identified earlier in our study of the binary sulfides are influenced by the presence of a promoter. It should be noted that the variations in the activity parameter are dominated by changes in n rather than B, and it appears that the dominant electronic factor related to promotion of MoS, is the increase in the number of “d”-electrons on MO. This result is consistent with the physical picture and model for thiophene binding which we described earlier for the binary sulfide systems. We suggested that these active catalysts can bind to the ring sulfur atom in thiophene and donate electron density into an empty thiophene pi* orbital. It is reasonable, then, that in the promoted MoS, catalysts an increase in the number of electrons occupying the MO 4d orbitals will result in an increase in catalytic activity, since an increase in delectron density on MO would strengthen the interaction between MO and the ring sulfur atom in thiophene. In conclusion, by combining measured trends in the HDS activity of promoted MoS, catalysts with calculated trends in the electronic structure of cluster models of these catalyst systems, we have established that there is an electronic basis for the promotion effects observed in MO&based HDS catalysts. The calculated electronic structure of the cluster models of the promoted catalyst systems indicates that Co and Ni have the ability to formally reduce MO in these systems, while Cu has the ability to formally oxidize MO. None of the other 3d metals have this
155
ability. The number of 3d electrons which Co, Ni or Cu contribute to the cluster and the energies of their 3d orbitals relative to the MO 4d orbitals make these metals unique when combined with MO. These results are consistent with our earlier identification of electronic factors which are related to the HDS activity of binary sulfide catalysts and with our model for thiophene binding to the catalyst.
REFERENCES 1. S. Harris and R. R. Chianelli, J. Catal. 1984,86,400. 2. T. A. Pecoraroand R. R. Chianelli, J. Catal. 1981,67, 430. 3. S. Harris, Chem.Phys. 1982,67,229. 4. R. J. H. Voorhoeve and J. C. M. Stuiver, J. Catul. 1971, 23,228,243. 5. A. L. Farragher and P. Cossee, In Proceedings of the 5th International Congresson Catalysis(Edited by J. J. Hightower), p. 1301. North-Holland, Amsterdam (1973). 6. C. Wivel, R. Candia, B. S. Clausen, S. Msrup and H. Topsrx, J. Catal. 1981,68,453. 7. S. Kasztelan, H. To&oat, J. Grimblot and J. P. Bonnelle, Bull. Sot. Chim. Belg. 1984,89,807. 8. R. R~Chianelli, T. A. Pecoraro, T. R. Halbert, W.-H. Pan and E. I. Stiefel, J. Catal. 1984,86,226. 9. S. Harris and R. R. Chianelli, J. Catal. (in press). 10. K. H. Johnson, Adv. Quantum. Chem. 1973,7,143. 11. J. G. Norman, Mol. Phys. 1973,31,1191. 12. K. Schwarz, Phys. Rev. B 1972,5,2466 ; Theor. Chim. Acta 1974,34,225. 13. J. C. W. Folmer, Ph.D. Thesis, Rijksuniversiteit te Groningen (1981).
1986 0
0277-5387/86 S3.00 + .OO 1986 Pergzamm Press Ltd
THE RELATION BETWEEN ELECTRONIC TRENDS AND THE PROMOTER EFFECT ON CATALYTIC ACTIVITY OF HYDRODESULFURIZATION CATALYSTS SUZANNE
HARRIS
Exxon Research and Engineering Co., Route 22 East, Clinton Township, Annandale, NJ 08801, U.S.A. (Received 1 July 1985) Abstract-The
origin of the promoting effect of certain first-row transition metals on the hydrodesulfurization (HDS) activity of MoS, is not well understood. We have carried out a systematic theoretical study of the electronic structure of simple clusters which model these catalyst systems and have used the results of these calculations to establish an electronic basis for the trendsin promotion effects observed in the MoS, systems. Both Co and Ni, which serve as promoters, have the ability to donate electrons to MO, while Cu, which serves as a poison, withdraws electrons from MO. The other first-row transition metals, which have little effect on the HDS activity of MO&, do not have the ability to donate to or accept electrons from MO. Thus promotion is associated with an increase in electron density on MO while poisoning is associated with a decrease in electron density. These results are consistent with previous results for binary catalysts which related particular electronic factors to the HDS activity of the sulfide.
Recently, we described a relation between trends in the calculated electronic structure of the transitionmetal sulfides (TMS) and trends in the measured hydrodesulfurization (HDS) activity of these materials.’ The experimental trends in the activities of the unsupported TMS had been previously measured using the HDS of dibenzothiophene as a model reaction,2 and a theoretical basis for understanding these experimental trends had been initiated by studying the electronic structure of the first- and second-row TMS.3 Calculations were carried out for a group of octahedral MS:- clusters, where the transition metal M was varied systematically across both the first and second transition series, and based on these calculations it was shown1 that a relation exists betwen the calculated electronic structure of the sulfides and their activity as HDS catalysts. Several electronic factors were identified which appear to be related to catalytic activity, and these factors were incorporated into a calculated activity parameter which correlates with the observed catalytic activity of the sulfides. We have now extended this combined theoretical and experimental approach to study the
promoted or “synergic” MoS2 catalyst systems, and in this paper we show that there is an electronic basis for the promotion effect observed in MoS2-based catalysts. Although it is well known that the addition of a second transition metal such as Co or Ni to a binary sulfide such as MoS, or WS2 brings about an increase in HDS activity, the basis for this promotion effect is certainly not well understood. Work from numerous 1aboratorieP has established, however, that the promoter atoms are associated with the edge planes of the layered compounds MoS, and WS2, in close proximity to the MO or W atoms. Several studies’** also suggest that the effect of this edge promoter is electronic rather than structural (i.e. the promoter influences the quality of the active site rather than altering the number of active sites). In order to further explore the idea that the basis for promotion is electronic in nature, a systematic study of the promoting effect of the first-row transition metals V-Zn on the HDS activity of MoS, was carried out in our laboratory.g In this study the activities of unsupported MS,-MoS, (M = V, Cr, Mn, Fe, Co, Ni, Cu or Zn) catalysts in the HDS of 151
152
S.
HARRIS
DBT were measured. It was found that : (1) V, Cr, Mn, Fe and Zn have little or no effect on the HDS activity of MoS, ; (2) Co and Ni, in agreement with previous literature, act as strong promoters ; and (3) Cu acts as a poison, serving to decrease the activity of MoS,. In order to understand these effects we have carried out a systematic theoretical study of the electronic structure of simple clusters which model these promoted catalyst systems. As mentioned above, the importance of the edge planes and the proximity of the promoter and MO atoms in MO& based catalyst systems has been previously established, and we have chosen the model clusters so that their structure is consistent with this established information. Calculations were carried out for a group of clusters MOM’S”,-, where the first-row transition metal M’ was varied systematically across the series from V to Zn. Each cluster carries a negative charge (n - ) because enough electrons were included in each cluster so that each sulfur is formally S2-, the molybdenum is formally Mo4+, and the first-row metal M’ has the formal oxidation state appropriate for the corresponding metal sulfide stable under reactor conditions. The geometry of the cluster can be seen in Fig. 1. Both metals are octahedrally coordinated by six sulfurs, three of which are shared between the two metals. Thus the model cluster is composed of two face-sharing octahedra. Within each cluster, all of the metal-sulfur bond distances were taken to be equal. These distances were determined for each cluster by averaging 2.42 A, the MO-S distance in MoS,, and the M’-S distance in the corresponding first-row transition-metal sulfide. The choice of octahedral coordination around MO allowed us to compare the Mo-S bonding and MO electronic contlguration in these model clusters with that in the simple binary clusters used previously. If the first-row transition metal M’ occupies sites around the edges of MO&, an examination of a model of MoS, suggests that M’ might occupy either octahedral or tetrahedral sites and share sulfurs with an adjacent MO. Once again we chose to use octahedral coordination for M’ because it allows us to make comparisons readily with our earlier
s _..._
M~‘&+M’.-.-S
hi’= V,Cr,Mn,Fe,Co,Ni,CuorZn
Fig. 1. Geometry of a model cluster MOM’S;- used for the molecular-orbital calculations.
calculations. We verified with selected calculations using tetrahedrally coordinated M’ that a change in geometry would not change the major conclusions of our calculations using octahedrally coordinated M’. As in our previous work Is3 all the calculations on these clusters were carried out using the SCF-Xo! scattered wave method with tangent spheres.‘O The atomic sphere radii were chosen according to Norman’s criteria.” The atomic exchange parameter c1values for the regions within the metal and sulfur spheres are those of Schwarz.12 A weighted average of these atomic a values was used for the intersphere and outersphere region. In our previous reports we described in detail the electronic structure and bonding in the model octahedral first- and second-row transition-metalsulfur clusters. Figure 2(a) shows a schematic diagram of the valance energy levels calculated for those clusters. Each group of levels, except for those corresponding to the metal d-orbitals, is represented by a block which is labelled according to the atomic orbital making the major contribution to these levels. There is of course some mixing between the metal and sulfur orbitals, but this mixing is not important for this discussion. Lying at the lowest energy is a group of levels arising from the sulfur 3s orbitals. The next group of levels arises primarily from combinations of sulfur 3p orbitals. At the top of this group of sulfur levels is the It,, level, a nonbonding combination of sulfur 3p orbitals. Finally, at the highest energy (in most of the clusters) are the 2tze and 3e, levels. These levels correspond primarily to the metal 3d or 4d orbitals. We found that a major effect of varying the transition metal in these octahedral clusters is a shift in the energy of the metal d-orbitals relative to the energy of the sulfur 3p orbitals. As measured by the difference in energy between the 2tzs and It,, orbitals, it was found that from Ti to Ni in the first transition series the energy of the 3d orbitals shifts downward in energy (i.e. closer in energy to the sulfur 3p orbitals) by about 1.5 eV. This shift is expected when moving from left to right in the transition series and is con6rmed by available photoelectron spectra of the TMS.3 For the face-sharing octahedral clusters considered here, we can also draw a schematic energy level diagram showing the main features and patterns of the energy levels resulting from the calculations. It should be noted that the point group symmetry of the face-shared cluster is C,, (as compared to Oh in the octahedral clusters). Since in the C,, point group there are no triply degenerate representations, the set of 3d or 4d orbitals which in O,-symmetry transform as the t,, representation are no longer degenerate in these clusters. The remnants of the octahedral splitting between the tze- and e,-
153
Catalytic activity of hydrodesulfurization catalysts ] 4d “ei 3e, ]
3d I&”
M3d or4d 3 2bg
=I
nMS6 0)
4d ‘i2;
M’
3d “tag”
MoM’Sy@I
Fig. 2. Schematic valence energy level diagrams for :(a) an octahedral MS:- cluster, and(b) an MOM’S”,cluster made up of two face-sharing octahedra. The ordering of 3d and 4d levels is for M’ = Co or Ni.
orbitals are still apparent in these lower-symmetry clusters, however, so that we continue to label “tlg” and “e#” sets of metal d-orbitals. A schematic level diagram for a MOM’S”,- cluster is shown in Fig. 2(b). Once again, the sulfur 3s levels lie below the broader group of levels resulting from the sulfur 3p levels. The top of the sulfur 3p group of levels is once again delineated by anonbonding combination of sulfur 3p orbitals, in this case the Za, level as shown in the diagram. The really notable feature of this level diagram is the presence, above the sulfur 3p levels, of levels resulting from both the MO 4d and M’ 3d metals. Each set of five d-orbitals forms “tZg”and “e#” sets of levels, so on the diagram we see a “tzg” and “e#” set of levels for both the metal 3d and 4d orbitals. The schematic diagram in Fig. 2(b) would correspond to the case where the M’ levels lie low enough in energy that the M’ 3d “tzg” and “eg” sets of levels each lie below the corresponding MO 4d sets of orbitals but not so low in energy that the entire 3d manifold of levels lies below the 4d levels. This relative energy separation applies to most of the clusters considered here. The relative energies of these levels and the numbers of electrons and their distributions within these levels both depend on M’. We will see below that it is these quantities which are related to the ability of the 3d metal M’ to serve as a promoter. As we noted earlier, the metal 3d orbitals become more stable upon proceeding from the left to the right side of the transition series, i.e. dropping in energy relative to the sulfur 3p orbitals. This means that in these clusters, which contain both a 3d metal and MO, the metal 3d orbitals also drop in energy
relative to the MO 4d orbitals. Proceeding across the 3d series from V to Zn this drop in energy is quite large. For V, Cr and Mn the 3d and Mo4d“tZg)) sets of levels are very close in energy. Moving to the right, however, the 3d levels shift lower in energy so that the 3d and 4d sets of levels lie in the order illustrated in Fig. 2(b). Finally, for Cu and Zn the shift in energy becomes large enough that the entire set of 3d levels lies below the MO 4d levels. At the same time that the 3d levels decrease in energy, the number of 3d electrons occupying the orbitals increases. For all the clusters, MO contributes two “d”-electrons while the number of “&‘-electrons contributed by the first-row metal depends on the metal. Thus the total number of “d”-electrons in the clusters varies from four in the VMO cluster all the way up to 12 in the Zn-Mo cluster. The distribution of electrons among the metal orbtitals depends on the relative energies of the orbitals, and an examination of the metal orbital occupations reveals an important difference between the clusters containing 3d metals which are known to promote or poison the activity of MoS, and the clusters containing those 3d metals which have no effect on the activity. In the clusters containing Co or Ni, the two metals which can act as promoters, we find that Co and Ni donate electrons to MO so that MO is formally reduced relative to the MO present in MoS,. This occurs because Co and Ni provide seven and eight “d”-electrons, respectively, to the cluster. In a pure Co-S or Ni-S cluster, the 3d “e: orbitals would be occupied. In these clusters containing MO, however, the lower energy MO “tZB” orbitals are available so
154
S. HARRIS
that a net transfer of charge from Co or Ni to MO occurs. This is a result of the MO “tZB”orbitals lying energetically between the 3d “tzg” and “eg” orbitals and Co and Ni having a high eneough d-electron count that the 3d “eg” orbitals would normally be partially occupied. The electron transfer of this type only occurs because both of these effects are present. In both the Co-MO and Ni-Mo clusters, the MO is thus formally reduced relative to the MO present in MoS,. In the cluster containing Cu, a metal which poisons the activity of MO&, we find that MO is formally oxidized relative to the MO in MoS,. This occurs because, while all five of the Cu 3d orbitals now lie well below the MO 4d levels, Cu contributes only nine “&‘-electrons to the cluster. As a result there is a net transfer of electrons from MO to Cu. This is not entirely surprising because it is known that in CuS13 (which contains nominally Cu2+ and S2-) the Cu 4d levels lie low enough in energy that some of the Cus are reduced to Cu+, while some sulfurs are oxidized to Sz -. In our Cu-MO cluster the MO electrons are even more available than the sulfur 3p electrons and thus the MO is oxidized relative to the MO in MoS2. In all the other clusters (those containing V, Cr, Mn, Fe and Zn) there is no such clear transfer ofelectrons between the MO and the 3d metal, and the electronic state of MO is similar to the MO in MoS,. For these metals the relative energies of the 3d and MO 4d orbitals and/or the number of 3d electrons are not appropriate for electron transfer. In summary, when the first-row transition metal M’ is varied in these Mo-M’ clusters, the changes in electronic structure depend on changes in both the relative energies of the MO 4d and M’ 3d orbitals and the number of “P-electrons which M’ contributes to the cluster. As the 3d levels drop in energy across the first transition series, the number of 3d electrons occupying these levels increases. It is only when M’ is Co, Ni or Cu, however, that these effects combine in such a way that the electronic state of MO is formally affected. Since it is these three first-row metals which have a measurable effect on the HDS activity of MoS,, we associate their ability to reduce or oxidize MO with their promoting or poisoning effect. Not only do these results identify an electronic origin for the promotion effect, but they are also consistent with our earlier theory relating catalytic activity to several electronic factors for the binary sulfides. One of those factors is the ability of the transition metal to bond covalently, in both a sigma and pi fashion, to the 3p orbitals of sulfur. This factor serves to differentiate the 3d from the 4d transition metals, since the 3d metalscovalently bond muchless effectively to sulfur than do the 4d metals. We defined a quantity, called B, which measures the covalent contribution to the metal-sulfur bond strength and
found that a large value of B, and thus a strong covalent contribution to the metal-sulfur bonding, correlates with high activity. Another important factor is n, the number of “P-electrons in the highest occupied orbital. We saw that a large value of n also correlates with high activity. Thus the most active catalysts have large values of n and/or B. We combined these quantities to form an activity parameter “AZ”, where : A, = nB. parameter provided us with a simple quantiative relationship between the calculated electronic structure and the HDS catalytic activity of the TMS. A metal sulfide with a large value of A2 will have a high HDS activity. Our model cluster calculations indicate that in the promoted CoMo and Ni-Mo systems one of these electronic factors for MO, n, the number of electrons in the HOMO, is increased by the presence of the promoter. Since these extra electrons occupy an antibonding orbital, the MO-S bonds are somewhat weakened. Thus the factor B which enters the activity parameter and measures the relative MO-S covalent bond strength decreases in the presence of the promoter. On the other hand, the presence of a poison such as Cu has just the opposite effect. To a first approximation, for all the other systems considered here the two electronic factors n and B for MO are unaffected by the presence of the 3d transition metal. We can calculate an activity parameter for the MO portion of the model clusters treated here. This activity parameter, along with the experimental HDS activities of the real catalyst systems, is plotted in Fig. 3, and from this figure we see that the trends in activity are reproduced by the This
-800
V
CI
Mn
Fe
Co
Ni
Cu
2x1
Metal(M) in Ms,-M0s,
Fig. 3. Calculated activity parameter for each mixed-metal sulfide system (right-hand scale). Shown for comparison, using the left-hand scale, are the measured HDS activities.g
Catalytic activity of hydrodesulfurization catalysts trends in the value of the activity parameters. The qualitative correlations obtained here thus indicate that the electronic factors identified earlier in our study of the binary sulfides are influenced by the presence of a promoter. It should be noted that the variations in the activity parameter are dominated by changes in n rather than B, and it appears that the dominant electronic factor related to promotion of MoS, is the increase in the number of “d”-electrons on MO. This result is consistent with the physical picture and model for thiophene binding which we described earlier for the binary sulfide systems. We suggested that these active catalysts can bind to the ring sulfur atom in thiophene and donate electron density into an empty thiophene pi* orbital. It is reasonable, then, that in the promoted MoS, catalysts an increase in the number of electrons occupying the MO 4d orbitals will result in an increase in catalytic activity, since an increase in delectron density on MO would strengthen the interaction between MO and the ring sulfur atom in thiophene. In conclusion, by combining measured trends in the HDS activity of promoted MoS, catalysts with calculated trends in the electronic structure of cluster models of these catalyst systems, we have established that there is an electronic basis for the promotion effects observed in MO&based HDS catalysts. The calculated electronic structure of the cluster models of the promoted catalyst systems indicates that Co and Ni have the ability to formally reduce MO in these systems, while Cu has the ability to formally oxidize MO. None of the other 3d metals have this
155
ability. The number of 3d electrons which Co, Ni or Cu contribute to the cluster and the energies of their 3d orbitals relative to the MO 4d orbitals make these metals unique when combined with MO. These results are consistent with our earlier identification of electronic factors which are related to the HDS activity of binary sulfide catalysts and with our model for thiophene binding to the catalyst.
REFERENCES 1. S. Harris and R. R. Chianelli, J. Catal. 1984,86,400. 2. T. A. Pecoraroand R. R. Chianelli, J. Catal. 1981,67, 430. 3. S. Harris, Chem.Phys. 1982,67,229. 4. R. J. H. Voorhoeve and J. C. M. Stuiver, J. Catul. 1971, 23,228,243. 5. A. L. Farragher and P. Cossee, In Proceedings of the 5th International Congresson Catalysis(Edited by J. J. Hightower), p. 1301. North-Holland, Amsterdam (1973). 6. C. Wivel, R. Candia, B. S. Clausen, S. Msrup and H. Topsrx, J. Catal. 1981,68,453. 7. S. Kasztelan, H. To&oat, J. Grimblot and J. P. Bonnelle, Bull. Sot. Chim. Belg. 1984,89,807. 8. R. R~Chianelli, T. A. Pecoraro, T. R. Halbert, W.-H. Pan and E. I. Stiefel, J. Catal. 1984,86,226. 9. S. Harris and R. R. Chianelli, J. Catal. (in press). 10. K. H. Johnson, Adv. Quantum. Chem. 1973,7,143. 11. J. G. Norman, Mol. Phys. 1973,31,1191. 12. K. Schwarz, Phys. Rev. B 1972,5,2466 ; Theor. Chim. Acta 1974,34,225. 13. J. C. W. Folmer, Ph.D. Thesis, Rijksuniversiteit te Groningen (1981).