Hartree-Fock-Slater calculations on cation-induced changes in the adsorption of carbon monoxide on iridium tetrameric clusters

(1)

Hartree-Fock-Slater calculations on cation-induced changes in

the adsorption of carbon monoxide on iridium tetrameric

clusters

Citation for published version (APA):

Jansen, A. P. J., & Santen, van, R. A. (1990). Hartree-Fock-Slater calculations on cation-induced changes in the adsorption of carbon monoxide on iridium tetrameric clusters. Journal of Physical Chemistry, 94(17), 6764-6772. https://doi.org/10.1021/j100380a043

DOI:

10.1021/j100380a043 Document status and date: Published: 01/01/1990

Document Version:

Publisher’s PDF, also known as Version of Record (includes final page, issue and volume numbers)

Please check the document version of this publication:

• A submitted manuscript is the version of the article upon submission and before peer-review. There can be important differences between the submitted version and the official published version of record. People interested in the research are advised to contact the author for the final version of the publication, or visit the DOI to the publisher's website.

• The final author version and the galley proof are versions of the publication after peer review.

• The final published version features the final layout of the paper including the volume, issue and page numbers.

Link to publication

General rights

Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights. • Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain

• You may freely distribute the URL identifying the publication in the public portal.

If the publication is distributed under the terms of Article 25fa of the Dutch Copyright Act, indicated by the “Taverne” license above, please follow below link for the End User Agreement:

www.tue.nl/taverne

Take down policy

If you believe that this document breaches copyright please contact us at: openaccess@tue.nl

providing details and we will investigate your claim.

(2)

6764 J . Phys. Chem. 1990, 94, 6764-6172 duction by TEOA, the magnitude of ihu must decrease.

Invoking a pathway for the direct reduction of -[RuI3+ by the electrode is certainly reasonable in the derivatized polypyrrole films. An electrode hole transport channel from the -[RuI3+ sites to the electrode exists based on the semiconductor-like properties of the polypyrrole. In fact, the ultimate site of oxidation following the photochemical generation of -[RuI3+ may be as a hole in the surrounding polypyrrole, eq 26.

-[RuI3+

+

polypyrrole

-

-[RuI2+ + polypyrrole' (26)

In this case, the competitive reduction steps in Scheme I would be replaced by eqs 27-29. (27) polypyrrole+

-

polypyrrole (28) polypyrrole+

-

polypyrrole polypyrrole+

+

PQ+

-

polypyrrole

+

PQ2+ (29) It seems doubtful that hole transport through the polypyrrole plays a major role. Destroying the conductivity of the polypyrrole has an effect on the photocurrent, but only in decreasing it by -50%. A change a t this level could arise as a medium effect. The majority of the photocurrent experiments were carried out on films in which the polypyrrole conductivity had been destroyed. The electrochemical data in Figure 9 demonstrate that an independent - [ R U ] ~ + / ~ + electron-transfer hopping channel to the electrode also exists. This channel does appear to play a role in decreasing photocurrent production. From the data in Figure 1 I ,

jhu/r_IRU12+ increases with the thickness of an inner layer of

p ~ l y - ( p y r ) ~ - P Q ~ + but only up to the point (- 160

A)

where the outer -PQ2+/-[ RuI2+ layer no longer reaches the electrode by interpenetration. Past

-

160

A

there is no longer direct contact with the electrode by -[RuI2+ and the - [ R U ] ~ + / - [ R U ] ~ + electron transport channel no longer diminishes the photocurrent. If energy-transfer quenching of -[RuI2+* a t the electrode plays a role, the energy-transfer channel would be affected in the same way. In bilayer films the importance of the - - [ R U ] ~ + / ~ + electron- transfer channel diminishes as -[RuI2+ is diluted by -PQ2+. This is shown by the data in Figure 9. The -[RuI3+l2+ couple does not appear in films of ~ - , ~ ~ , 2 + = 0.21 in which the polypyrrole

-e

+TEOA

conductivity had been destroyed. The implied decrease in

-[ Ru] 3+/2+ electron-transfer hopping rate with dilution coincides

with the increase in photocurrent yield. It is, no doubt, a con- tributing factor.

Conclusions

Polymeric films which contain both -[RuI2+ and -PQ2+ have been prepared by oxidative electropolymerization of the appro- priate pyrrole derivatives. In these films the relative spatial dispositions and ratios of -[RuI2+ and -PQ2+ could be varied almost at will, which created a basis for the preparation of a variety of microstructures.

Photolysis of the films in the presence of the reductive scavenger TEOA led to sustained photocurrents. They arose by oxidative quenching of -[RuI2+* by -PQ2+ and electron transport to the electrode by -PQ2+/+ electron-transfer hopping.

The following conclusions can be reached based on the mech- anistic studies:

I . There are major disadvantages to polypyrrole as the structural framework for the films. Polypyrrole is a significant, competitive light absorber, and excited-state lifetimes are greatly reduced by energy- or electron-transfer quenching by polypyrrole.

2. Photocurrent efficiencies are greatly enhanced as -[RuI2+ is diluted b j -PQ2+.

3. There is no evidence for long-range energy transfer by energy-transfer hopping between the -[RuI2+ sites. If efficient quenching is to occur. the -PQ2+ sites must be intermingled with

4. Separating the -[RuI2+ sites from the electrode by an inner layer of poly-(pyr),-PQ2+ led to an enhancement in photocurrent efficiency.

-[ Ru]*+.

Acknowledgment. Acknowledgment is made to the Army Research Office for support of this work under grant no. DAAL03-88-K-0192. A.J.D. and N.A.S. acknowledge the National Science Foundation for support while visiting the University of Grenoble under grant no. INT-85 14009.

Registry No. (Pyr),-PQ2+*2PF[ (homopolymer), 127998-84-9; (Pyr)3[Ru]2+*2PFb- (homopolymer), 127998-90-7: TEOA, 102-71-6.

Hartree-Fock-Slater Calculations on Cation-Induced Changes in the Adsorption

of CO

on

Ir, Clusters

A.

**P. J . Jansen* and R.**

A .

van Santen

Laboratory for Inorganic Chemistry and Catalysis, Eindhoven University of Technology, P.O. Box 51 3, 5600 M B Eindhoven, The Netherlands (Received: September 29, 1989: In Final Form: February 28, 1990)

The adsorption of CO on various sites of a tetrahedral Ir4 cluster has been studied by using the Hartree-Fock-Slater method. The changes in the adsorption when a small metal cluster is deposited in a pore of a zeolite are modeled with a Mg2+ ion at the opposite side of the Ir, cluster. The formation of strong bonds of metal d and CO 2 r * orbitals favors the 2- and 3-fold adsorption geometry. I n the presence of a Mg2+ ion the distance between CO and the metal cluster increases, and the frequency

of the CO stretch vibration is shifted upwards. The changes can be ascribed to polarization of the Ir, clusters in the electrostatic field generated by the Mgz+ ion. As a result the repulsive interaction between the CO 5a orbital and the Ir, cluster is reduced, and back-donation into the unoccupied CO 2 ~ * orbital decreases. Implications of our results to catalysis are discussed.

1. Introduction

The CO adsorption on transition metals has been studied for many years because it is the initial step in several catalytic conversion reactions. It is also of interest in its own right, because it can be used to probe structural and electronic details of small metal particles and surfaces. From a quantum chemical point of view CO adsorption on a transition metal is very interesting because of the subtleties that occur in the change of the electronic structure of the transition metal, and the existence of very different

0022-3654/90/2094-6164$02.50 f 0

orbitals of CO (5a and 2**) that can interact with a substrate.' Extensive extended Huckel calculations have been done in our group on large clusters,2 and by Sung and Hoffmann on slabs.3 Most a b initio calculations have been dealing with XCO, where

( I ) Veillard, A. Quantum Chemistry: The Challenge of Transition Metals

( 2 ) de Koster, A. Thesis, Eindhoven University of Technology, Eindhoven.

( 3 ) Sung. S . S.; Hoffmann, R . J . A m . Chem. Soc. 1985. 107. 578. and Coordination Chemistry; Reidel: Dordrecht, 1986.

1989.

(3)

Adsorption of C O on lr4 Clusters The Journal of Physical Chemistry, Vol. 94, No. 17, 1990 6765 X stands for a transition-metal atom, and with carbonyl complexes.

The Hartree-Fock approximation, which has been used for the initial ab initio calculation^,^^ has been shown to give rather poor results due to its preference of a high occupation for the metal s orbitals. In general this high occupation gives a strong repulsion with C O so that no adsorption is found. Hence, inclusion of correlation has become necessary.'*2' Unfortunately, configuration interaction calculations on systems with two or more transition-metal atoms are extremely large so that only few of them have been done.22-24

An alternative class of methods is formed by the local density approximations. Many calculations have been done on carbonyl complexes using the Hartree-Fock-Slater (HFS) approximation,z5-33 which is the method we have used for this work. This method has been shown to give reasonable accurate results for diatomic molecules.26 In general, it overestimates the bond energy for carbonyl complexes, whereas bond lengths are predicted somewhat too short (see ref 1 , pp 159-177). This has to be compared, however, with Hartree-Fock calculations, which may not yield bonding at all, and configuration interaction calculations, which are hardly feasible for the systems we are interested in.

In this paper we show the results of our work on the C O adsorption on a tetrahedral Tr4 cluster, and the influence of a Mg2+ ion on this adsorption. Calculations on such small clusters, though it is debatable how well they can be used to represent single-crystal surfaces, are of interest because they exist in some catalytic systems. This is especially the case for metal particles that have been deposited in the pores of zeolites. These zeolites often contain channel cations that compensate for the negative zeolite lattice potential. They are located between the deposited metal cluster and the wall of a pore. Cations of high charge close to metal particles have been shown to give remarkably enhanced activities in hydrocarbon conversion reactions as well as increased sulfur r e ~ i s t a n c e . ~ , - ) ~ We have chosen to study the adsorption of C O

~~~~ ~ ~

(4) Bagus, P. S.; Hermann, K.; Seel, M. J . Vac. Sci. Technol. 1981, 18, ( 5 ) Hermann, K.; Bagus, P. S. Phys. Rev. E 1977, 16, 4195.

(6) Bauschlicher, Jr., C. W.; Bagus, P. S. J . Chem. Phys. 1984.81, 5889. (7) Demuynck, J. Chem. Phys. Lert. 1977, 45, 74.

(8) Faegri, Jr., K.; Almlof, J. Chem. Phys. Lett. 1984, 107, 121. (9) Hillier, I . H.; Saunders, V. R. Mol. Phys. 1971, 22, 1025. ( I O ) Bagus, P. S.; Roos, B. 0. J . Chem. Phys. 1981, 75, 5961. ( I I ) Basch, H.; Cohen, D. J . A m . Chem. SOC. 1983, 105, 3856. ( 1 2 ) Bauschlicher, Jr., C. W. J . Chem. Phys. 1986,84, 260.

(13) Bauschlicher, Jr., C. W . ; Bagus, P. S.: Nelin, C. J.; Roos, B. 0. J . (14) Bauschlicher, Jr., C. W.; Siegbahn, P. E. M. J . Chem. Phys. 1986, (15) Blomberg, M. R. A.; Brandemark, U. B.; Siegbahn, P. E. M.; (16) Ha, T.-K.; Nguyen, M. T. J . Mol. Srrucr. (THEOCHEM) 1984,109, (17) Kao, C. M.; Messmer, R. P. Phys. Reu. E 1985, 31, 4835. (18) Kouteckg, J.; Pacchioni, G.; Fantucci, P. Cbem. Phys. 1985, 99, 87. ( I 9) Luthi, H. P.; Siegbahn, P. E. M.; Almlof, J. J . Phys. Chem. 1985,89,

(20) Pacchioni, G.; Koutecky, J.; Fantucci, P. Cbem. Phys. Lett. 1982, 92,

(21) Rives, A. B.; Fenske, R. F. J . Chem. Phys. 1981, 75, 1293.

(22) Bagus, P. S.; Hermann, K.; Bauschlicher, Jr., C. W. J . Chem. Phys. (23) Bagus, P. S.; Nelin, C. J.; Bauschlicher, Jr., C. W. J . Vac. Sci. (24) Pacchioni, G.; Koutecky, J. N A T O ASI Ser., Ser. C 1986, 176, 465. ( 2 5 ) Baerends. E. J.; Ros, P. Mol. Phys. 1975, 30, 1735.

(26) Baerends, E. J.; Ros, P. Int. J . Quantum Chem. 1978, 12, 169. (27) Post, D.; Baerends, E. J. J . Chem. Phys. 1983, 78, 5663.

(28) Dunlap, B. 1.; Yu. H. L.; Antoniewicz, P. R. Phys. Reu. A 1982, 25, (29) Johnson, J. B.; Klemperer, W . G. J . Am. Cbem. SOC. 1977.99, 7132. (30) Jorg. H.; Rosch, N. Chem. Phys. Lert. 1985, 120, 359.

(31) Rosch, N.; Jorg, H.; Kotzian, M . J . Chem. Phys. 1987, 86, 4038. (32) Ziegler, T.; Rauk, A. Inorg. Chem. 1979, 18, 1755.

(33) Ziegler, T.; Tschinke, V.; Ursenbach, C. J . Am. Chem. SOC. 1987, 435.

Chem. Phys. 1986, 85, 354. 85, 2802.

Mathisen, K. B.; Karlstrom, G. J . Phys. Chem. 1985, 89, 2171. 331. 2156. 486. 1984, 81, 1966. Technol. A 1984, 2, 905. 7. 109. 4825.

8

s

₈

I A I \

6

0

6

Figure 1. Structures of the three Mg2+-Ir4C0 systems that have been studied. From top to bottom are shown an oxygen atom, a carbon atom,

four iridium atoms, and a magnesium ion. The structures of the three

Ir,CO systems that have been studied are the same except for the Mg" ion that is absent.

TABLE I: ExDonentin1 Coefficients for Basis Functions

Ir C 0 ME*+ - 0

Is" 26.55 Is" 5.40 Is" 7.36 Is" 10.00

2s' 26.55 2s 1.24 2s 1.70 2s 2.70 3s" 22.40 2s 1.98 2s 2.82 2s 4.35 4s" 8.85 2p 0.96 2 p 1.30 2p 2.35 5s" 6.95 2p 2.20 2p 3.06 2p 3.90 2p" 35.24 3d 2.50 3d 2.00 2p 7.10 3p" 18.47 3s 0.75 5p" 4.57 3s 1.75 3d" 22.15 4d" 10.65 4 P 7.59 5d 1.55 5d 2.75 5d 4.75 6s 1.30 6s 2.35 6p 1.81 4p" 9.99 3s 1.10

"

Additional function for core orthogonalization.

with the cation on the opposite side of the cluster. We have used Mg2+ as cation because it has a higher charge than the usual cation Na+ and K+, thus increasing the effect on the C O adsorption. Iridium has been used because it is representative for a transition metal active in hydrocarbon conversion ~ a t a l y s i s . ~ '

2. Computational Details

We have performed restricted nonrelativistic Hartree-Fock- SIater38 LCAO calculations on C O adsorbed on tetrahedral Ir, clusters and looked a t the influence of a Mg2+ ion on the adsorption. The distance between two iridium atoms was taken equal to the nearest-neighbor distance in the bulk, Le., 2.72

A.

The distance between the carbon and oxygen atom was taken equal to 1.13

A,

Le., the experimental distance for the free CO molecule. No geometry optimization was done for the Ir4 cluster or for the C-0 distance. Only the distance between C O and Ir4 was optimized in all adsorption geometries. We have studied only end-on

(34) Rabo, J. A.; Schomaker, V.; Pickert, P. E. In Proceedings of the 3rd

International Congress on Catalysis; North-Holland: Amsterdam, 1965; Vol. 2.

(35) Gallezot, P. Catal. Re&-Sci. Eng. 1979, 20, 121.

(36) Boudart, M.; Djega-Mariadassou, G. Kinetics of Heterogeneous Catalyfic Reacrions; Princeton University Press: Princeton, NJ, 1984.

(37) Ravenek, W.; Jansen, A. P. J.; van Santen, R. A. J . Phys. Chem. 1989, 93, 6445.

(38) Slater, J. C. The Self-Consistent Field for Molecules and Solids: In

Quanrum Theory of Molecules and Solids; McGraw-Hill: New York, 1974; Vol. 4.

(4)

6766 The Journal of Physical Chemistry. Vol. 9 4 , N o . 1 7 , 1990 Jansen and van Santen

TABLE 11: Exponential Coefficients for Fit Functions

I r C 0 Mg2+ I C 2s 3< 4$ 5s 5s 6s 1 s 7s 8 5 95 9s 95 9s 9s I os 1 l a 1 Is ‘ 9P 9P 9P IOP 1 OP 1 I P I I P 9d 9d 9d 1 Od I Od I Id 9f 9f 1 O f I O f 9g 9g 1 og 23.10 53.10 48.95 46.00 41.85 30.60 27 85 24. I O 15.18 21.80 14.40 1 1

.oo

8.10 6 00 3.90 5 , 3 5 3.65 2.60 11.52 8.474 6.233 5.1 1 1 3.8 I6 3 144 2.319 9.14 6.723 4.945 4.055 3.028 2.495 9.482 6.918 4.607 12.16 7.93 5 . i 8 6 14.54 Is 2s 2s 3s 3s 3s 3s 3s 4s 2P 2P 3P 3P 3P 4P 4P 3d 3d 3d 4d 4d 5d 10.80 I s 14.72 1s 11.75 2s 16.01 2s 7.78 2s 10.60 3s 7.57 3s 10.32 3s 5.35 3s 7.29 3s 3.78 3s 5.15 4s 2.66 3s 3.63 4s 1.88 3s 2.57 5s 1.76 4s 2.40 5s 7.60 4s 1.77 5s 5.03 2p 10.42 5s 4.89 2p 6.90 3.46 3p 6.71 2.44 3p 4.74 2.28 3p 3.35 1.68 4p 3.13 7.90 4p 2.30 5.58 3d 9.36 3.94 3d 6.61 3.68 3d 4.67 2.70 4d 4.36 2.48 4d 3.21 5d 2.94 20.00 12.70 1 1

.oo

7.80 5.40 5.10 3.80 3.50 2.50 1.85 1.50

adsorption with the carbon atom closest to the Ir4 cluster. The I-fold (atop), 2-fold (bridge), and 3-fold (hollow) adsorption have been studied. I n all geometries the Mg2+ ion was put on the opposite side of the cluster; i.e., when C O was I-fold (2-fold/3- fold) absorbed then Mg2+ was 3-fold (Zfold/l-fold) adsorbed (see Figure 1 ) . The distance of the Mgz+ ion to the nearest iridium atom was taken to be equal to the ionic radius (0.66

A)

plus half the Ir-lr distance. With this distance we tried to compromise between a maximum influence on the CO adsorption and still retaining the ionic character of Mg2+.

We have used the HFS-LCAO method implemented by the group of Baerends of the Free University in Am~terdam.~”’ The programs allow us to calculate directly the adsorption energy via the Ziegler transition-state and to decompose this adsorption energy into a steric part, consisting of an electrostatic part and an exchange repulsion part, and an interaction part, which can be split up into various symmetries. We have been using fractional occupation numbers for degenerate orbitals at the Fermi level, which are to be interpreted as yielding the statistical energy of a multideterminantal wave function assuming that eq 3.10 in ref 43 holds. The ionization potentials that we will show have been calculated by using the Slater transition method.38

The basis sets (see Table I ) that we have used are essentially of double {quality (the importance of the triple { d for iridium is explained in ref 44). Polarization functions have been added for iridium, carbon, and oxygen. This has been necessary as some

(39) Baerends, E. J.; Ellis, D. E.; Ros, P. Chem. Phys. 1973, 2, 41.

(40) Baerends, E. J.; Ros, P. Chem. Phys. 1973, 2, 52.

(41) Baerends, E. J.; Ros, P. Chem. Phys. 1975, 8 , 412. (42) Ziegler, T.; Rauk, A. Theor. Chim. Acta 1977, 46, I .

(43) Ziegler. T.; Rauk, A,; Baerends, E. J . Theor. Chim. Acra 1977, 43, (44) Snijders, J . G.; Vernooijs, P.; Baerends. E. J. At. Nucl. Data Tables 261.

1981, 26, 483.

TABLE 111: Adsorption Energy of CO on Ir, and Its Decomposition (in eV)

geometry I-fold 2-fold 3-fold

A E , -1.08 (-1.56) 0.63 (-6.29) 0.37 (-6.24) ~ ~ S I , , , , 1.7 1 7.48 7.94 -1.56 (-1.08) -10.42 (-3.51) -7.47 (-4.05) 0.00 (0.00) -0.09 (-0.08) -3.20 (-0.01) W”, -2.64 -9.89 -10.30 AEads -0.93 -2.42 -2.39

TABLE IV: Gross Populations for I r L O

Ir S 0.87 0.64 0.56

(nearest to CO) p -0.14 0.01 0.04

d 8.09 8.21 8.27

co

5u 1.79 1.46 1.50

2a* 0.32 0.69 0.75

test calculations have shown (see also ref 45). No polarization functions have been used for Mg2+, the reason being that we were not interested in the detailed interaction of the cation with Ir4; the Mg2+ serving only as a model perturbation. Moreover, the calculations gave no reason to suspect that polarization functions for Mg2+ were necessary. The fit functions for the electron density that are used by the HFS implementation are shown in Table 11.

The fit functions for iridium have been used satisfactorily in a previous p ~ b l i c a t i o n . ~ ~ The fit functions for carbon and oxygen have been taken from ref 45. Actually, we could use, as can be inferred from Table 2 and Table 1 of ref 45, a somewhat smaller set. An estimate of the numerical error in the adsorption energy, which includes the fit of the electron density, is given in section 3B. In all calculations we have used an exchange parameter a

= 0.7. All integrals have been computed n ~ m e r i c a l l y . ~ ~ 3. Results and Discussion

A . CO Adsorption on Ir,. The results of the CO adsorption in the geometries of minimal energy are shown in Tables 111 and I V . The adsorption energy is defined as

3Eads = E(Ir4CO) - E(Ir4) - E ( C 0 )

Hence the negative values for LIE,& show that we have bonding. If we compare AEads for the three adsorption geometries, we see that there is a clear preference for the 2-fold or 3-fold adsorption. We have decomposed the adsorption energy in a number of contributions.

AEads = AE,,,,,,

+

AE,,

+

AEr

+ AE6

The first term AEslcric on the right-hand side is the energy change when we superimpose Ir4 and CO without changing their electronic configuration. This may be interpreted as what is normally called the steric rep~lsion.~’ The other three terms form the interaction energy AEinl split up into various symmetries. We use the des- ignations o, T , and 6 as generic names. The point group symmetry

is lower that C,,, so with u we really mean a , , with T we mean

e (in C3J or b, plus b2 (in C,,), and with 6 we mean a*. We see from Table 111 that the steric repulsion as well as the interaction energy is much larger for the 2- and 3-fold adsorption, yielding a net larger adsorption energy. This means that the interaction is stronger for the 2- and 3-fold adsorption. The decomposition of the interaction energy poses a problem due to the change in the electronic configuration. If we take for example the 3-fold adsorption then the valence electronic configuration for infinite separation of Ir, and CO should be a!6a2e28. (The

Td symmetry,37 and 8ir4 for CO in C,, symmetry.) Two electrons electronic configurations for the fragments are a:e8tyt,8

f

for Ir, in

(45) Baerends, E. J.; Vernooijs, P.; Rozendaal, A.; Boerrigter, P. M.; Krijn,

M.; Feil, D.; Sundholm, D. J . Mol. Struct. (THEOCHEM.) 1985, 133, 147. (46) Boerrigter, P. M . ; te Velde, G.; Baerends, E. J . I n t . J . Quuntum

(5)

Adsorption of C O on Ir, Clusters The Journal of Physical Chemistry, Vol. 94, No. 17, I990 6767 T A B L E V: Adsorption Energy of CO on Ir (in eV) and Gross

PoDulations for IrCO

electronic configrn

a'nW uSn764 a%%'

A Eadr -6.9 1 -5.85 -6.64 S 0.27 0.43 0.57 P 0.03 0.07 0.05 d 8.64 8.45 8.34 50 I .59 1.69 1.64 2K* 0.42 0.3 1 0.35

have been transferred however from a a , orbital to a (0.97

electron) and e (1.03 electron) orbitals, giving aj4a:.a7e29.03. Consequently the total energy of all electrons in a , orbitals with respect to lr, and C O at infinite separation increases, whereas the energy of a2 and e orbitals decreases, only because the number of electrons in orbitals with a certain symmetry changes. This is shown in particular for the a2 symmetry in 3-fold adsorption. Although these orbitals do not actively participate in bonding, their contribution to the interaction energy is large. We can get estimates for the various symmetry contributions to the interaction energy without the change in electronic configuration by trans- ferring the electrons back to their original orbitals assuming that the molecular orbital energies do not change. If there has been no complete crossing of molecular orbital levels, but only a new degeneracy at the Fermi level, then we can correct for these electron transfers by adding to each of the contributions to the interaction energy the number of electrons that are transferred out of the corresponding symmetry times the Fermi energy. If we take for example again the 3-fold adsorption then we have a Fermi energy of -3.30 eV, so that the correction for the u symmetry is 2 X -3.30 = -6.61 eV, for a it is -1.03 X -3.30 = 3.41

eV, and for 6 it is -0.97 X -3.30 = 3.19 eV. The sum AEo

+

AE,

+

AEd is invariant. (Actually, there has been a complete crossing of the Fermi level of an a2 orbital in the 3-fold adsorption geometry. However, its molecular orbital energy is only slightly higher than the Fermi energy so that the error that we are making by applying the procedure above is negligible.) We thus get the numbers in parentheses in Table Ill. These are quite reasonable if we compare them with calculations for larger Ir4-C0 separations where no electron transfer between orbitals of different symmetries has taken place. We find for example in the 1-fold adsorption with adsorption distance 2.17

A

(see footnote in Table X ) that AE, = -1.23 eV and AE, = -0.86 eV.

We see that the u orbitals are stabilized more than the a orbitals. Inspection of the molecular orbitals shows that the interactions are predominantly with the 5 0 and 2 a * orbitals of CO. Hence in the synergic a-a mechanism proposed by Dewar, Chatt, and D ~ n c a n s o n ~ ~ . ~ ~ the u donation from the 5u seems to be more important than the a back-donation into the 2a*. The contribution of the 6 orbitals is seen to be very small. The reason is that they do not actively participate in the bond formation. Their energy change is entirely due to the change in the effective field that they feel. However, we have to be careful in our analysis of the various contributions to the adsorption energy. Using the constrained space orbital variation method Bagus et al. reached the conclusion that the a back-donation is more important than the u dona ti or^.^^,^^ As they distinguish between metal and C O polarization on the one hand and charge transfer on the other, their and our results are not in contradiction. The u interaction energy we find corresponds to the sum of their u donation and a large part of their polarization energy.

Before discussing the Ir,CO results further we will address Ir-CO interaction in the simplest possible system IrCO, in order to study the relative importance of the u and a interactions. We assumed a linear geometry and Ir-C and C - O distances as in the I-fold adsorption of minimal energy for Ir,CO. The C O molecule

Ir

co

(47) Dewar, M. J. S. Bull. SOC. Chim. Fr. 1951, 18. C71. (48) Chatt, J.; Duncanson, L. A . J . Chem. SOC. 1953, 2939.

/

I

Figure 2. Contour plot of the electron density difference p(IrC0)

-

p(1r)

- p(C0) for the a ' ~ * 6 ~ state of IrCO. Dashed lines show a decrease and solid lines an increase of the electron density, except for the solid lines next to dashed lines which depict nodal surfaces. Subsequent contours correspond to f0.001, f0.003, f0.007, f0.013, f0.027, f0.054, 10.108, f0.218, and f0.439 electrons/A'.

has (valence) electronic configuration u6a4. For iridium we consider only states densn. In C,, these become, considering only those with the lowest occupation of u symmetry, a ' ~ ~ 6 ~ , u2r4@, and u2a3?j4. If we combine these with CO we get U ' T ~ ~ ~ , a8a8S3,

and a8a764. Results of calculations for these electronic configurations are shown in Table V. We see that a low occupation of the u orbitals is preferred. The overlap of the C O 5u orbital and the iridium 6s orbitals is 0.4. If iridium would remain in its atomic ground-state configuration d7s2, then there will be a strong repulsion. The changes in the a part of the wave function will strive to minimize this repulsion. We also note that variations in the occupation of the a orbitals changes the adsorption energy almost as much as changes in the occupation of the a orbitals. The gross populations of the most important orbitals reflect the electronic configuration changes. The u electrons on iridium prefer the more compact d, orbitals to the s orbital. The repulsion is thus reduced by stabilization of an excited state of the iridium atom,6,'8v2' in which the 6s orbital is not occupied. We have studied the way in which the reduction of the u repulsion is accomplished by calculating an electron density difference plot. The plot for a7n8S4

is shown in Figure 2. We see an increase of the electron density in the 2r* region and a decrease in the 5 u region. There is

moreover a decrease a t the iridium end and an increase beside iridium. The electrons of iridium are not polarized away from CO but rather pushed aside. This has been found also for PdC0.'8 It is, however, different from a polarization away from C O which has been found for FeCO, NiCO, and C U C O . ' ~ ~ ~ ~ Inspection of the molecular orbitals reveal that sd hybridization, with a dzz orbital participating ( z axis along the molecular axis), is causing the increase. There is also a slight increase due to the fact that d,, and d,, orbitals get more electrons, which can be seen by the four maxima next to iridium.

We now discuss the adsorption of C O on Ir,, starting with the

1 -fold adsorption geometry. Apart from the three additional iridium atoms we have the same structure as for IrCO. The results of the calculations are however very different. We calculated a heat of formation for the Ir, cluster of 28.35 eV. If we compare this t o the adsorption energy of CO, we see t h a t the interactions between the iridium atoms are much stronger than the interaction with CO. Hence the substrate for CO becomes very different, and differences in the adsorption results are only to be expected. We first note that the adsorption energy is much higher for IrCO than for Ir,CO. Actually the adsorption energies for IrCO seem excessively high. HFS poorly describes the configuration of a single iridium atom, because of the high spin polarization. We

have therefore fixed the configuration to d7s2 and taken this as reference, though HFS prefers d9, which is 4.43 eV lower in

(6)

6168 T h e Journal of Physical Chemistry, Vol. 94, No. 17, 1990 Jansen and van Santen

Figure 3. Contour plot of the electron density difference p(lr,CO) -

p(Ir4) - p(C0) for I-fold adsorption. Values at the contours as in Figure 2.

energy. The IrCO probably shows this same preference, yielding high adsorption energies with respect to iridium d7s2. With respect to iridium d9 the adsorption energies are -2.48, -1.42, and -2.21 eV for a7s864, a8ir764, and aas863, respectively. For Ir, these problems are not to be expected as the electronic configuration is determined by the interatomic interactions. The difference is

also shown by the electron density difference plots Figures 2 and 3. The shifts in the electron density are much larger for IrCO than for Ir,CO I-fold (note the exponential scale for the contours). In both cases we see a decrease of the electron density between iridium and carbon, and a t the iridium end of the Ir-CO molecule/fragment. There seem to be some differences in the s

back-donation, which shows an increase in the s region between iridium and carbon for I r 4 C 0 which is split up for IrCO. The reason for this is probably the decrease in 50 population that

overlaps the 2 s * .

The repulsion between the 6s and the 5a orbitals for Ir,CO 1-fold will be much smaller than for IrCO. The reason is a depopulation of the s orbital in Ir4 with respect to the iridium atom.

I n Ir, there is only one occupied orbital of the right symmetry (a, of C3J that has any appreciable s character. This is a totally symmetric combination of Td of the four 6s orbitals. Due to the small weight of each of these 6s orbitals in the complete molecular orbital, the overlap with the 5a orbital is much smaller for Ir, than for the iridium atom (0.13 and 0.40, respectively). On the other hand, the overlap with the 5d9 orbital remains at about 0.1 3. This means that the influence of d orbitals is much'larger for the Ir,

cluster. We also note that this depends on the precise position of the molecular orbital with respect to the Fermi level and hence on the cluster geometry.

From Table 111 we have noted that the adsorption energy for the 2- and 3-fold adsorption is higher than for the I-fold and that both contributions, steric repulsion and interaction energy, are much larger. I t is clear that the steric repulsion should be higher as the carbon atom is in contact with more iridium atoms than for the I-fold adsorption. This increase is enhanced by the larger interaction that pulls the C O molecule closer to the Ir, cluster.

As for the interaction, inspection of the molecular orbitals shows again that there is little participation of the iridium s orbitals. The difference with the 1 -fold adsorption is that the 50 and 2 s * orbitals of C O can interact with many more cluster orbitals and that there is also a substantial interaction with the 4a orbital. The main reason why the interaction is so much stronger for the 2- and 3-fold adsorption is shown in the electron density difference plots, Figures 4 and 5 . We see again a decrease of the electron density between the carbon atom and the Ir, cluster, and a decrease in the region near the iridium atoms closest to CO, pointing away from the molecule. Thus the mechanism for decreasing the repulsion is the same as for IrCO. The most important observation, however, is a strong increase just above the Ir-C lines. This

Figure 4. Contour plot of the electron density difference p(Ir4CO)

-

p(lr,) - p(C0) for 2-fold adsorption. Values at the contours as in Figure

2.

Figure 5. Contour plot of the electron density difference p(lr,CO) -

p(lr4) - p(C0) for 3-fold adsorption. Values at the contours as in Figure

L .

TABLE VI: Ionization Energies for CO Orbitals (in eV)

40 5 a I n free CO 20.2 14.2 1 -fold 17.5 13.9 2-fold 16.4 14.1 3-fold 16.4 14.0 ref 49 17.2 14.1 ref 50 16.7 12.6"/13.9 ref 51 16.5 12.6a/13.8 17.4 14.7 14.2 (b,) 13.7 14.1 13.9 13.8 14.0 (b2) (I See text

increase is due to the very favorable interaction between the 2s*

orbitals of C O and a d orbital of an iridium atom (this can be inferred from the cloverleaf structure at the iridium atom position). This increase points to the formation of the Ir4-C0 bond and also yields the large back-donation shown in Table IV.

We want to end this section by making some comparisons between our results and experimental results. In a recent paper Gtlin et al.49 have collected published results for adsorption energies and give some new ones themselves. The adsorption energies we have calculated are in general somewhat larger than the ones given by Gtlin et al. There may be two reasons for this. First, HFS in general overestimates bond energies. Second, we have used smaller clusters than have been investigated by G6lin et al.,

(49) Gelin, P.; Auroux, A.; Ben Taarit, Y.; Gravelle, P. C. Appl. C a r d .

(7)

Adsorption of C O on Ir, Clusters

TABLE VII: Adsorption Energy of CO on Mg2+-Ir4 and Its Decomposition (in eV)

geometry

1 -fold 2-fold 3-fold

AEsteric 0.18 5.62 4.78 AEo -0.86 (-0.86) 21.91 (-4.83) 5.53 (-3.96) PEb 0.00 (0.00) -2.69 (-0.14) -0.01 (-0.00) AE,d, -0.98 -2.35 -2.42 -0.30 (-0.30) -27.14 (-2.95) -12.71 (-3.23) G", -1.16 -7.93 -7.19

TABLE VIII: Gross Populations for Me2+-IrrC0

Ir (nearest to C O ) s 0.60 0.47 0.46

d 7.95 7.93 7.96

co

5U 1.83 1.50 I .56

27~* 0.04 0.53 0.56

P -0.03 -0.01 0.04

whereas the adsorption energy increases when the cluster size decreases. More useful comparisons can be made with results from electronic spectroscopy experiments. We have calculated ionization and excitation energies using the Slater transition-state method. The results are shown in Table VI. According to Plummer et aLso a small cluster such as we have used should show the same spectra as a semiinfinite solid. This means that our results should agree with experimental results even if the experimental system is larger. We see that this is indeed the case for the UPS experiments. We find again a clear difference between the l-fold adsorption on the one hand and the 2- and 3-fold adsorption on the other. The ionization energy of the 4u orbital is at higher binding energies for the 1-fold adsorption. The 5u

orbital is a t higher binding energies for the 2- and 3-fold ad- sorption. The shift in the ionization energy of 5u is so large that it ends up very near 1 A. Peaks in UPS spectra of adsorbed C O

are generally ascribed to 5u that shifts to higher binding energies due to the adsorption and 40 and 1 A that are not affected. The

situation seems to be, however, somewhat different. The I T level is indeed not affected, but for the 2- and 3-fold adsorption there are two molecular orbitals that have both about an equal amount of 4a and 5a character. This means that the 4a level does par- ticipate somewhat in the bonding, which can also be inferred from the change of its ionization energy. UPS spectra have been reported for C O adsorbed on reconstructed Ir( 100)51352 and for the Ir,(C0)12 complex.s0 The adsorption geometry for C O / l r - (100) is not known. We see in Table VI that the agreement with experiment is very good. BrodEn et al.51952 interpreted a shoulder of the peak at 13.8/13.9 eV as due to the 5u orbital. Plummer et al. reinterpreted these spectra and ascribed the peak at 13.8/13.9 eV to 5 u and 1 A. From our calculations we can conclude that CO is adsorbed in a 2-fold position on Ir( 100). The agreement for the Ir4(CO)12 complex is also very fair. The I-fold adsorption geometry is recognizable by the high ionization energy of the 4 0

orbital that particpates less in the bonding than in the 2- and 3-fold adsorption. The splitting that we calculate for the 5u and I n orbitals may be too small to be resolved in the spectra. We also calculated some excitation energies for the 3-fold adsorption finding 13.9 eV for 40

-

2 7 * , 11.6 eV for 50

-

2 ~ * , and 1 1 . 2 eV for 1 A

-

2 ~ * . Electron energy loss spectra show a broad band

around these energies.s3 However, as the resolution is poor we have not investigated these energies further.

The C-0 stretch frequency has been studied with infrared spectroscopy for various iridium catalyst^.^^^^^^^ These studies

(50) Plummer, E. W.; Salaneck, W . R.; Miller, J. S. Phys. Reo. E 1978,

(51) Brodtn, G.; Rhodin, R. N . Solid Srare Commun. 1976, 18, 105. (52) BrodCn, G.; Rhodin, T. N.; Brucker, C.; Benbow, R.; Hurych, 2. Surf.

(53) Netzer, F. P.; Mack, J. U.: Bertel. E.; Matthew, J . A . D. SurJ Sci.

18. 1673.

Sci. 1976, 59, 593.

1985. 160, L509.

The Journal of Physical Chemistry, Vol. 94, No. 17, 1990 6769

- 2 . 3 9

Ir4 b Ir4CO

Mg2+-Ir4

-

MgZ+-Ir4C0

Figure 6. Adsorption energies for Mg2+ and CO. The 3-fold adsorption geometry is used with the CO a t the optimized distance for the H F S calculation without Mg2+.

1 - 3 9 4

1

-4.08

- 2 39

r

-.

, \

Figure 7. Contour plot of the electron density difference p(Mg2+-lr4) - p(Ir4) - p(Mg2+) for 3-fold adsorption of Mg2+. Solid and dashed lines are to be interpreted as in Figure 2. Subsequent contours correspond to f0.005, fO.O1l, f0.020, f0.035, f0.058, f0.096, f0.159, 10.262, and f 0 . 4 3 2 electrons/A3.

TABLE IX: Decomposition o f the Steric Repulsion (in e V )

geometry system G l s t a t AExrep

1 -fold I r 4 C 0 -4.12 6.43 Mg2+-Ir4C0 -1.80 1.98 2-fold I r 4 C 0 -12.37 19.85 Mg2+-I r 4 C 0 -9.66 15.28 3-fold Ir4C0 -12.88 20.82 Mg2+-Ir4C0 -9.47 14.25 indicate that CO adsorbs 1-fold. We have estimated this frequency using the 2 ~ * occupation and Figure 4 of ref 26. The discrepancy with our results predicting a preference for 2- or 3-fold adsorption is probably due to a cluster size effect. HFS calculations by Post and Baerends on Cu,-CO showed the same di~crepancy.~' Recent calculations using the same method of C O on Cu slabs showed, however, a preference for the I-fold a d s o r p t i ~ n . ~ '

B. CO Adsorption on M$+-lr,. The results of the C O ad- sorption on MgZ+-Ir4 in the geometry of minimal energy are shown in Tables VI1 and VIII. The adsorption energy in this case is defined as

AEads = E(Mg2+-Ir,CO) - E(Mg2+-Ir4) - E ( C 0 ) The adsorption energy is decomposed just as in Table 111. If we compare the adsorption energies with and without the cation we see only small differences. Actually these differences are well within the numerical accuracy of our calculation. In order to get an estimate for the numerical error in our results we calculated the adsorption energy of C O and Mg2+ in two steps: adsorbing first CO and then Mg2+, and adsorbing first Mg2+ and then CO. The results are shown in Figure 6 . As the sum of the separate adsorption energies should be the same, we see that the numerical error in our calculations is about 0.1 eV. The geometry we used in these calculations was the optimized geometry for Ir,CO without

Mgz+ with C O 3-fold adsorbed. (Unless explicitly stated oth-

(54) Solymosi, F.; Rask6, J . J . Catal. 1980, 62, 253.

( 5 5 ) Tanaka, K.; Watters, K . L.; Howe, R. F. J . Catal. 1982, 75, 23. (56) Toolenaar, F. J. C. M.; Bastein, A. G. T. M.; Ponec, V. J . Cafal. (57) Baerends, E. J. Private communications.

(8)

6770 The Journal of Physical Chemistry, Vol. 94, No. 17, 1990 Jansen and van Santen

TABLE X: Structural (in A) and Vibrational (in em-') Prowrties

geometry system adsorpn distance' wlrX WC43*

I -fold Ir,CO 2.09 228 2180 Mg2+-lr4C0 2.36 158 2420 2-fold Ir,CO 1.48 404 1870 Mg2*-lr4C0 1.52 364 2000 3-fold Ir,CO 1.43 369 1810 Mg2+-lr,C0 1.51 306 1980

ODistance from C to the plane through the nearest Ir atoms and perpendicular to the CO axis. *Calculated by using Figure 4 of ref 26. erwise, with n-fold adsorption we will mean n-fold adsorption of CO.)

Although the adsorption energies do not change when Mg2+ is added, we see that there are changes in the decomposition. I n

all cases both the steric repulsion and the interaction energy become less. We can split up the steric repulsion further in an electrostatic and an exchange repulsion part.27

AEsteric = AEe~stat

+

AExrep

We see in Table IX that the electrostatic part increases (becomes less negative) and the exchange repulsion decreases. The change in the electrostatic part is due to the positive charge of the cation and the dipole of C O that is pointing toward the cation, which is electrostatically unfavorable. We show in Figure 7 why the exchange repulsion decreases. The cation polarizes the Ir4 cluster so that there is a decrease of electrons at the C O end of the cluster that reduces the repulsion. The cloverleaf structure a t the top iridium that shows an increase in the electron density is small and too far away from the C O to have much effect at normal adsorption distances. We shall see, however, that this area of increased electron density does show up in the distance dependence of the exchange repulsion. We see that the electrostatic interaction increases about 3 eV for all adsorption geometries. This is reasonable as the position of the C O with respect to the cation is for all geometries about the same. The decrease of the exchange repulsion is larger in all cases; hence we always find a decrease for the steric repulsion.

The changes in interaction energy counteract the changes in the steric repulsion. The corrections for the change in electronic configuration are much larger than for Ir,CO, the reason being that the Fermi level is lowered by approximately 9.8 eV due to the uncompensated charge of the cation. (The electronic configurations for Mg2+-Ir4 are a!4a:e28, a~8a~~s'b~ob:0.19, and a14a2e28 1 2

for Mg2+ coordinated I-, 2-, and 3-fold, respectively.) If we take the corrected values we see that the interaction of the u orbitals parallels the steric repulsion. This is to be expected as the changes in the u orbitals mainly strive to reduce the repulsion, as will become clear from the change in the 5u occupation. The a orbitals show less back-donation. This can be ascribed to the positive charge of the cation which makes it harder for electrons to move toward CO. The relative importance of the u donation and 7

back-donation seems to be about the same as for I r 4 C 0 . The 6

orbitals do not play a significant role. We also note that there is again a great similarity between the 2- and 3-fold adsorption. The only change on might wish to observe is that the steric repulsion for h 4 C 0 is somewhat higher for the 2-fold as for the 3-fold adsorption but that this is reversed for Mg2+-lr4C0. Such change is also found for the interaction energy.

Gross populations shown in Table VI11 show the some picture for the effects of the cations as (the decomposition of) the adsorption energies. The iridium atom(s) farthest away from the cation have become slightly positive due to polarization. The decreased number of electrons in the s and d orbitals yield the reduction of the repulsion and also (as far as the d orbitals are concerned) a decreased back-donation. There is also a reduction in the number of electrons in the 27* and an increase in the 5a

orbitals. The reduction in the 2 a * orbitals also agrees with the reduced back-donation that has been inferred from AEr. The effect is largest for the 1-fold, less for the 2-fold, and least for the 3-fold adsorption. The reason for this is probably that the electrons have to move farther away from the cation when being

TABLE XI: Fit Parameters A (in eV) and a (in A-') for the Exchange Repulsion geometry system A N 1 -fold Ir,CO 6050 3.28 Mg2+-lr4C0 10400 3.61 2-fold Ir,CO 3320 2.90 Mg2+-lr,C0 ' 5490 3.21 3-fold Ir,CO 2820 2.83 Mg*+-lr,CO 6040 3.28

back-donated in the I-fold than in the 2- or 3-fold adsorption geometry (see data on the geometry; Table X). The decrease of the 2a* occupation implies that although the cation does not change the adsorption it very probably will make it harder for the CO molecule to dissociate. The increase of the occupation of the 5u orbital is small, but it is found for all three geometries. One would expect the cation to increase the u donation by pulling the electrons away from CO. As this does not happen, we conclude that the change in the u orbitals on adsorption is mainly polarization to reduce repulsion.

Although the results of Tables VI1 and VI11 have been explained as if the cation represents a point charge, this is by no means the case. This situation is best approximated when C O is I-fold adsorbed, Le., 3-fold adsorption of Mg2+. In this geometry there is only a small charge transfer of 0.26 electrons to the cation. The charge transfer is 0.36 electrons for the 2-fold and even 0.71 electrons for the 3-fold adsorption of CO. Thus the charge transfer is less when the coordination of the cation is higher. If we look a t the molecular orbitals we see that for the 3-fold adsorption, where the charge transfer to Mg2+ is largest, there is one (partially) occupied orbital with appreciable 3s character of Mg2+. It is this orbital that is responsible for the charge transfer. It also has some contribution from an unoccupied I r 4 C 0 orbital with some 5 0 character. In this way the occupation of the So orbital is thus increased. The changes that the cation induces in the electronic structure of the Ir, cluster are, however, too small to alter the interactions with the CO molecule in a fundamental way. The fact that a crude electrostatic picture cannot explain all of the influence of the cation can also be inferred from Table IX which shows that the electrostatic repulsion between the cation and the dipole moment of CO is overcome by the reduction of the exchange repulsion.

The parameters that define the adsorption geometries with and without the cation are shown in Table X. The cation increases in all cases the adsorption distance. As the adsorption energy does not change, one expects that the 1r-C vibrational stretch frequency decreases. This is indeed the case as is shown also in Table X.

In the calculation of the frequency we have assumed that the CO molecule is rigid and that the Mg2+-Ir4 cluster is fixed. We do not know of any measurements of these frequencies, but they seem to be about in the normal range one finds for other systems. The estimates for the C 4 stretch frequency are also displayed in Table X. The frequency for the I-fold adsorption is very likely too high as it is obtained by extrapolation of the relation shown in Figure

4 of ref 26. We think, however, that the trend to higher fre- quencies is significant. Of interest is also the distance dependence of the exchange repulsion. This can be reasonably fitted by using an exponential function

AExrep = nA exp(-aR)

where R is the distance between the carbon and the nearest iridium atoms and n is the number of nearest iridium atoms. The results of the fits are given in Table XI. For small adsorption distances the exchange repulsion is larger with Mg2+, but it decreases much faster than without the cation. This correlates nicely with Figure

7 where there is an increase of the electron density near the top iridium atom.

Some calculations have been done previously on the influence of cations on CO ad~orption.~~"' Different from our studies, these

(58) Wimmer, E.; Fu, C. L.; Freeman, A . J. Phys. Reo. Lett. 1985, 55, 2618.

(9)

Adsorption of C O on Ir, Clusters

b

C

-.,-._

1 ' 1 ' 1 ' -12.00 - 8 . 0 0 - 4 . 0 0 ( MO

e n e r g y

0

_..

77%-

? V )

Figure 8 . Density of states plots: (a) total density of states, (b) local density of states of the CO 5a orbital, (c) local density of states of the CO 27' orbitals. T h e solid curves are the results for h 4 C 0 and the dashed curves for Mg2+-lr4C0. In all cases CO is 3-fold adsorbed. The local density of states in (b) and (c) are magnified by 5 with respect to the total density of states. The Fermi level is set to zero.

calculations have been done to model promoter effects on metals, putting an alkali-metal atom on a metal slab or cluster next to a C O molecule. This difference in position of the cation leads to opposite results. Wimmer et aLs8 show that the C O levels are lowered with respect to the metal levels, whereas we find just the opposite as is shown in Figure 8. This can be explained from an electrostatic point of view. The positive cation stabilizes the orbitals that are close to it. In the case of Wimmer et al. the C O orbitals are closest to the cation, whereas in our case the metal orbitals are closest. All studies show that the influence of the cation is dominantly electrostatic, although Wimmer et al. find, just as we do, that there is some covalent interaction as well. BonaEiE-Koutecky et al.s9 show that the adsorption energy is increased most when the cation is close to the CO molecule. Recent calculations have shown, however, that changes in adsorption energies may be different for other absorbates.62 The overall result is that changes of the frequency of the C-0 stretch vibration by the presence of alkali-metal ions depend on the position of the cation. If the metal cluster is in between CO and the cation, one expects the CO frequency to shift upwards. If C O and the cation are on the same side of the metal cluster the reverse is expected.

There exists a considerable literature on the behavior of chemisorbed CO to metal particles in the channels of a zeolite as a function of c a t i 0 n . 6 ~ ~ Also reports on competitive adsorption

(59) Bonafit-Kouteckv, V.; Kouteckq, J.; Fantucci, P.; Ponec, V. J . Caral.

(60) Nsrskov, J . K.; Holloway, S.; Lang, N. D. SurJ Sci. 1984, 137, 65.

(61) Lang, N. D.; Holloway, S.; Nsrskov, J . K . Surf Sci. 1985, 150, 24.

(62) Sanchez Marcos, E.; Jansen, A. P. J.; van Santen, R. A. Chem. Phys. ( 6 3 ) Naccache, C.; Primet, M.; Mathieu, M . V. Adu. Chem. Ser. 1973, 121, 266.

(64) Gallezot, P.; Datka, J.; Massardier, J.; Primet, M.; Imelik, B. In

Proceedings of the 6th International Congress on Catalysis; Bond, G . C., Wells, P. B., Tompkins, F. C . , Eds.; Letchworth: London, 1977; Vol. 2. (65) Figueras, F.; Gomez, R.; Primet, M. Adu. Chem. Ser. 1973,121,480.

(66) Chukin, G . D.; Landau, M . V.; Kruglikov, V. Ya.; Agievskii, D. A,; Smirnov, B. V.; Belozerov, A . L.; Asrieva, V. D.; Goncharova, N . V.; Rad- chenko, E. D.; Konovalcherov, 0. D.; Agafonoy, A . V. In Proceedings of the 6th International Congress on Catalysis; Bond, G. C., Wells, P. B., Tompkins, F. C . Eds.; Letchworth: London, 1977; Vol. I .

(67) Besoukhanova, C.; Guidot, J.; Barthomeuf, D. J . Chem. Soc., Faraday Trans. I 1981, 77, 1595.

1988, 111, 409.

Lett. 1990, 167, 399.

The Journal of Physical Chemistry, Vol. 94, No. 17, 1990 6771

of toluene and benzene have a p ~ e a r e d . ~ ~ , ~ ~ Studies on zeolites X and Y as well as zeolite L have been reported. Whereas Pt particles in CeY and NaHY behave as Lewis acids?O they behave as Lewis bases in zeolite Le7' Basicity increases with size of the earth alkali-metal cation. This basic behavior agrees with the observed lowered stretch frequency of C O adsorbed atop to Pt particles in zeolite L next to alkali-metal cation.68 Experimental results consistently indicate increased basic behavior of transition-metal particles occluded in zeolite L. The situation is less clear for the zeolites X and Y. Very high catalytic hydrogenation activities have been measured for metal particles next to highly charged ~ a t i o n s , 6 ~ ~ ~ ~ , ~ ~ indicating increasing Lewis acid character. Some C O adsorption studies indicate an increased C O frequency in such a situation,63d6 but also limited data are available that contradict this.68 It therefore seems that the experiments on these zeolites warrant closer study. Our work agrees with increased Lewis acidity of metal particles in contact with cation of high charge, if in that situation the metal particle is located in between adsorbate and cation, as is probably the case for zeolites X and Y. Experimental evidence seems to indicate that in zeolite L this situation does not exist. In zeolite L the adsorbate appears to be located between metal particle and cation.

4. Conclusions

We have performed Hartree-Fock-Slater calculations on the adsorption of CO on Ir, and studied the influence of a Mg2+ ion at the opposite end of the Ir, cluster on the Ir4-C0 bond. These calculations serve as a model study for the adsorption on small metal clusters deposited in the pores of a zeolite in which part of the silicon has been replaced by aluminum. Charge-compen- sating cations are present in the pores close to the metal particles. We have studied the adsorption energy and analyzed it in terms

of various contributions, the gross populations of the most important metal and C O orbitals, and electron density difference plots. Where possible we have made comparisons with experimental results.

We have found that the mechanism of adsorption is more or

less the same for all adsorption geometries. When the metal cluster, with or without the cation, and the CO molecule are brought together with a frozen electronic structure, there is a strong repulsion between the 5u orbital of C O and some metal orbitals, Le., the 6s orbital in the case of IrCO and the 6s and 5d orbitals in the case of Mg2+-Ir4C0. The interaction of the u orbitals strive to reduce this repulsion, whereas the interaction of the K orbials causes a bond to be formed between the metal d orbitals and the 2 ~ * orbitals of CO. The change in the gross population of the 5u orbital shows that polarization of the u orbitals is more important than u donation. The electrons on the nearest metal atom(s) are pushed aside by the C O molecule via an sd hybridization on the metal atom, thus minimizing the repulsion.

For small metal clusters the details of the adsorption depend on the particular electronic structure of the cluster. For a single iridium atom the stabilization of excited states with low s occupation is important for reducing the repulsion. On the other hand, in lr, the metal-metal interaction has already yielded such a low occupation. Two further general observations are that bonding with the d orbitals for iridium is more important than with the s orbital and that energetically the u interaction is more important

( 6 8 ) de Mallman, A,; Barthomeuf, D. In Studies in Surface Science and Catalysis; Karge, H. G.; Weitkamp, J. Eds.; Elsevier: Amsterdam, 1989; Vol. 46.

(69) Sheu, L. L.; Knozinger, H.; Sachtler, W. M . H. Catal. Left. 1989, 2, 129.

(70) Tri, T. M.; Massardier, J.; Gallezot, P.; Imelik, B. In Studies in Surface Science and Catalysis; Imelik, B., Naccache, C., Coudurier, G., Praliaud, H., Meriaudeau, P., Gallezot, G., Martin, G. A., Vedrine, J. C. Eds.; Elsevier: Amsterdam, 1982; Vol. 1 1.

( 7 1 ) Larsen, G.; Haller, G . L. Catal. Lett. 1989, 3, 103.

(72) Dalla Betta, R. A,; Boudart, M . I n Proceedings of the 5th Inrerna- tional Congress on Catalysis; Hightower, J . W., Eds.; North-Holland: Am- sterdam, 1973; Vol. 2.

ACS Symp. Ser. 1977, 40, 5 3 8 .

(10)

6772 J . Phys. Chem. 1990, 94, 6112-6180

than the A interaction; Le., reduction of repulsion is more important

than attractive bond formation.

There is a clear distinction between the I-fold adsorption on the one hand and 2- and 3-fold adsorption on the other. All interactions seem to be much weaker for the I-fold adsorption. Due to the bend in the connection Ir-C-0 for the 2- and 3-fold adsorption, there is a very favorable overlap between the 2a*

orbitals of CO and metal d orbitals. This gives rise to relatively strong bonds as can be seen in Figures 4 and 5 . For the I-fold adsorption this overlap is small and hardly any bond is formed. The preference for 2- and 3-fold adsorption is probably cluster size dependent. The strong repulsion for the 2- and 3-fold adsorption causes the 4a level to participate. Two molecular orbitals are formed with about equal amounts of 4a and 5a character and little metal character. The participation of the 4a orbital can clearly be seen in UPS spectra, e.g., compare the spectra of Ir4(C0),2 and CO-Ir( 100).

The Mg2+ ion has two opposing effects on the CO adsorption. It polarizes the metal cluster pulling electrons away from the side where CO is to adsorb. This reduces the repulsion with the 50 orbital but also makes the a back-donation more difficult. The net effect is that there is no change in the adsorption energy. The a back-donation is hampered most for the I-fold adsorption where the electrons have to move farthest away from the cation. The change is most clearly shown by the occupation of the 2a* orbitals.

The occupation of the 5u orbital changes little. The reduced overlap results in a slightly higher 5 u orbital occupation. The decrease of the occupation of the 2r* orbitals will make it harder for CO to dissociate.

The Mg2+ ion does cause some changes in the structure and the vibrational properties. The distance between the CO molecule and the metal is increased. As the adsorption energy does not change, the potential minimum becomes flatter. This is reflected by a decrease of the Ir-C stretch frequencies. The C - 0 stretch frequencies increase due to the depopulation of the 2n* orbitals.

There is an important difference between the geometries we have studied and geometries in which the cation sits next to CO on a metal surface. Electrostatic considerations show that in the latter case the CO orbitals are stabilized with respect to the metal orbitals. Via this mechanism the 27r* orbitals can become more populated and dissociation possible. The reverse happens in our case. We note that, although the electrostatic effect of the cation is most important, there is some covalent interaction as well. The position of the cation may be inferred from vibrational spectra as cations close to CO decrease the C - 0 stretch frequency, whereas when the metal cluster is between the cation and CO the frequency is increased. The position of the cation, as determined by the zeolite structure and the cation size, may explain the catalytic properties of metal clusters in zeolites X, Y , and L as a function of the cation type.

Effect

of

the Electrical Double Layer

on Voltammetry at Microelectrodes

John D. Norton, Henry S. White,*

Department of Chemical Engineering and Materials Science, University of Minnesota, Minneapolis, Minnesota 5545.5

and Stephen W . Feldberg

Brookhaven National Laboratory, Upton, New York I 1 973 (Received: October 2 , 1989;

In Final Form: April 2, 1990)

An analysis of transport of charged and uncharged species associated with a steady-state faradaic process at a spherical microelectrode is reported. We examine systems comprising various relative concentrations of a redox species and, if charged, its counterion and an inert electrolyte. Of particular interest is the behavior of these systems when the thickness of the diffuse double layer (characterized by the Debye length, K - I ) and the radius of the electrode (ro) are comparable. Transport of each

species is assumed to be governed by the Nernst-Planck equation. A generalized solution obtained by using finite-difference simulations demonstrates that significant enhancement or inhibition of the steady-state flux can occur and will depend upon the dimensionless parameter rOK, upon the relative values of the applied potential (E,,,), the formal redox potential ( E O ' ) ,

and the potential of zero charge (E,,,), upon the charges and relative concentrations of the species in solution, and upon the distance of closest approach of the reactant to the electrode surface. Analytic solutions for several limiting cases are discussed and serve as simple expositions of the phenomena as well as a verification of the simulations. In infinitely dilute ionic solutions, Le., roK

-

0, the limiting flux of ionic species may be computed directly from the Smoluchowski-Debye theory for ionic bimolecular reaction rates. Computation of theoretical voltammograms in the limit of infinite dilution (Le., roK

-

0) reveals the surprising result that under certain conditions the steady-state current-voltage curve will be peaked rather than sigmoidal, giving the appearance that electrochemical activity occurs only within a small (several hundred millivolt) potential window. The effect is easily explained when the electric field and the charge of the reacting species are considered.

Introduction

Development of electrochemical applications of microelectrodes has flourished during the last decade, expanding considerably beyond early applications as biological simulators and sensors.',* It is now possible to produce a microelectrode or microelectrode array whose critical dimension, ro (e.g., the half-width of a band electrode or the radius of its hemicylindrical approximation; the

(1) Adams, R. N . Anal. Chem. 1976, 48, I 128A.

( 2 ) Wightman. R . M . Anal. Chem. 1981, 53, 1125A

radius of a disk electrode or its hemispherical approximation), is of the order of several

nanometer^.^!^

Since the size of an electrode dictates the spatial (as well as the time) domain that is probed, the measurement of faradaic currents a t an electrode with nanometer dimensions may reveal fundamental information about the solvent structure and potential distribution within a few

(3) Morris, R. B.; Franta, D. J.; White, H. S . J . Phys. Chem. 1987, 91, (4) Seibold, J . D.; Scott, E. R.; White. H . S. J . Electroanal. Chem. 1989, 3559.

264, 281