Ligand Effects in Supported Metal Carbonyls: X-ray Absorption

The initial surface species were molecularly adsorbed [HRe(CO)s], which, upon heating to 80 OC in H2 or under vacuum, gave rhenium subcarbonyls with t...
0 downloads 6 Views 638KB Size
Langmuir 1992,8, 2716-2719

2715

Ligand Effects in Supported Metal Carbonyls: X-ray Absorption Spectroscopy of Rhenium Subcarbonyls on Magnesium Oxide A. Honji, L. U. Gron, J.-R. Chang, and B. C. Gates' Center for Catalytic Science and Technology, Department of Chemical Engineering, University of Delaware, Newark, Delaware 19716 Received December 27,1991. In Final Form: August 10, 1992

[HRe(CO)5]reacted with the surfacesof MgO powders, one being partially dehydroxylated (about 55% ) and the other almost fully dehydroxylated (about 93%). The initial surface species were molecularly adsorbed [HRe(CO)s],which, upon heating to 80 OC in H2 or under vacuum, gave rhenium subcarbonyls with three CO ligandsand three oxygen-containingligandsprovided by the MgO surface. Infrared spectra are consistent with CSUsymmetry in both structures, with bands at 2011 (w), 1895 (vs),and 1862 (sh) cm-1 for rhenium subcarbonylon the partially dehydroxylated MgO and at 2017 (w), 1908 (-1, and 1867 (sh) cm-1 for the rhenium subcarbonylon the almost fully dehydroxylated MgO. The average bond distances were determined by extendedX-ray absorption f i e structurespectroscopy: On the partiallydehydroxylatad is a surfaceoxygen) distanceswere 1.87, MgO, the Re-C, Reo*(0*is a carbonyloxygen), and Re-0,(0, 3.11,and 2.13 A,respectively,and on the almost fully dehydroxylated MgO, the Re-C, Re-0*,and Re-0, distances were 1.91, 3.12, and 2.13 A, respectively. These distances indicate chemical bonding of the rhenium carbonyl to oxygens of the MgO surface. They also demonstrate greater electron donation (backbonding)from the Re to the CO on the partially dehydroxylated MgO than from the Re to the CO on the almost fully dehydroxylated MgO surface, suggesting electron transfer from the MgO to the Re. The MgO surface is thus modeled as a rigid multidentate electron donor analogous to a molecular ligand. The X-ray absorption near edge data confirm a higher electron density on the Re atoms bonded to the oxygens of partially dehydroxylated MgO than on Re atoms bonded to almost fully dehydroxylated MgO.

Introduction The bonds between metal-containing adsorbates and surfaces of nonmetal supports influence the properties of materials including supported metal catalysts and microelectronic devices. The structures of metalaupport interfaces are poorly understood because they are almost always nonuniform. The goal of this research was to determine the influence of surface compositon on the structure and bonding of supported metal complexes that are structurally simple and nearly uniform. Metalcontainingspecies on metal oxide surfaces having different surface compositions were prepared and characterized to determine how the composition of the oxide surface influences the bonding in the supported metal complex. The supported species were chosen to be rhenium subcarbonyls on MgO because (1)the surface structures are robust and only moderately air sensitive and (2) they constitute a family with the same symmetry (suggested to be C%) over a wide range of compositions of the MgO surface, as shown by vibrational and X-ray absorption spectroscopies.l~sExtended X-ray absorption fiie structure (EXAFS)spectroscopy offers the opportunity to determine quantitative structural information characterizing the surface-bound complexes, and X-ray absorption near edge structure (XANES),combined with infrared spectroscopy, provides complementary measures of the electron transfer between the support and the adsorbed metal complex. For comparison, the literature provides

* Towhom corrmpondenceehould be addressed at the Department of ChemicalEngineering,University of California,Davis, CA 96616. (1) Kirlin, P. S.; KnBainger, H.; Gatee, B. C. J. Phys. Chem. 1990,94, 8451. (2)C u d , L.;Dobos,5.;Beck, A.; Vi-Oroez A. Catal. Today 1989,6, 91.

(3)Kirlin, P. S.;van Zon, F. B. M.; Koningsberger, D. C.; Gates, B. C. J. Phys. Chem. ISSO,94,8439. (4)Chang,J.-R.; Gron, L. U.;Honji, A.; Sanchez,K. M.; Gates, B. C. J. Phvs. Chem. 1991.95.9944. (@"Papile, C. J.; Gat&,B. C. Longmuir 1992,8,14.

characterization of molecular analogues of the surface species,including vibrational spectra that show how ligand donor strength influences the carbonyl stretching frequencies in metal carbonyls" and crystallographic data that quantify ligand effects in terms of bond distances.

Experimental Methods Material8 and Sample Preparation. The samples were prepared by chemisorption of [HRe(CO)6] on MgO powder (EM Science),as described previously? Samples were prepared and handled with exclusionof air and moisture on a double manifold Schlenk vacuum line or in a nitrogen-fiied Braun glovebox in which the concentrations of 0 2 and water were less than 1.0and 0.1 ppm, respectively. The partially dehydroxylated MgO support was pretreated by heating to 400 OC under vacuum at a rate of 3 OCImin, with the sample then held at 400 "C for 16 h followed by 1 h in flowing dry 0 2 and 1 h under vacuum at 400 OC. The surface area was approximately 75 m2/g. The sample, referred to as M g o ~is, inferred to be about 56% dehydroxylated.6a Another sample, denoted M@w, was about 93% dehydroxylateds18by a similar treatment at 700 instead of 400 OC; characterization of the rhenium subcarbonyl on this sample by EXAFS spectroscopy has already been reported.' Preparation of [HRe(CO)sl and its chemisorption on the support have been previously deecribed.' The MgO, incorporating molecularly adsorbed [HRe(CO)al,was held at 80 OC for 4 h in flowing Hz or in vacuum to partially decarbonylate the sample and form the surfacebound rhenium subcarbonyl; the same procedure had been used before.' The white materid wan stored in the drybox. T h e reference materials used for the EXAFS spectroecopy were [Os&O)ld (whichhasonlyterminal COligands)andboa. Their preparation is described elsewhere.a EXAFS parameters characteristic of the reference materials are summarkadin Table 1.

(6) Primet, M.; Basset, J. M.; Mathieu, M. V.; Prettre, M. J. Catal. 1973,29,213. (7) Lamb, H. H.; Gates, B. C. J. Am. Chem. Soc. 1988,108,Sl. (8) Anderson, P.J.; Horlock, R. F.; Oliver, J. F. Tram.Faraday Sm. 1966,61,2764.

0743-7463/92/2408-2715$03.00/0Q 1992 American Chemical Society

Honji et

2716 Langmuir, Vol. 8, No. 11,1992 Table I. Structural Parameten Characterizing Reference Compounds and Sampler Used in the EXAFS Analysis. crystallographic datab Fourier transform "Die shell N R,A Ak,A-1 &,A n 6 1.867 2.5-11.0 0.7-2.1 3 ReOsb 1st RS-0 [Oss(CO)1rl Os4sb 2 2.88 2.9-12.4 0.9-2.0 3 4 1.95 [OsS(CO)l2l os-cc 4 3.09 2.9-12.4 2.0-3.3 3 [0es(C0)121 os-o+ a Notation: N,coordination number for ahorbel-backacatterer pair; R,radial distance from crystal structure data, Ak, limita uae for forward transform (kis the wave vector); &, limits uaed for shell isolation(riadistance);n,power of k uaed for Fourier tramformation. Data for Re03 from Wyckoff, R. W. G. Crystal Structures ZZ, 2nd ed.;Wiler New York, 1963;p 52. Data for [Oes(CO)l2lfrom Corey, E. R.;Dahl, L. F. Znorg. Chem. 1962,1,521. f i r subtraction of the Os& contribution: N = 2, R = 2.88 A, the Debye-Waller factor A$ = -0.001 AS, and Eo = -3.3 eV, with an inner potential correction A E o = -4.4 eV on the difference file.

I

0

A

~

Zion

2onn

I Y 00

16

k, A-1 F'igure 2. Raw EXAFS data characterizing speciea formed by

adsorption of [HR~(CO)SIon MgOm followed by heating in H:, at 80 "C.

-5;

zzno

12

I

1895

~~~

8

4

41.

woo

17511

Wavenumber, cm-1

'

'

'

2'

'

'

'

'

3'

'

'

'

'

' 4

R, A Figure 3. Imaginary part and magnitude of Fourier trmform (k*-weighted, Ak = 3.5-10.2 A-l, Re-0 phaae corrected) char-

Figure 1. IR spectra in the carbonyl stretching region charact e r i z i i the species formed by adsorption of [HRe(CO)bon (A) Mg0700followedbyheatinginHzat80OCand(B)MgOafollowed by heating in H2 at 80 "C.

acterizing the species formed by adsorption of [HRe(CO)sl on (A) MgOa followed by heating in Hz at 80 "C ( d i d line) and (B)MgO, followed by heating at 80 O C , under vacuum (dotted line).

Infrared Spectroecopy. The diffuse reflectance infrared experiment was performed with a Nicolet 51OM instrument having a spectral reaolution of 4 cm-l with a diffuse reflectance attachment deacribed previously! The supported rhenium subcarbonyl samples were loaded into the DRIFT cell in the drybox and then scanned in the 1750-2200-~m-~ region. X-Ray Absorption Spectroscopy (XAS). The XAS experiments were performed on X-ray beamline X-11A at the National Synchrotron Light Source at Brookhaven National Laboratory with a ring energy of 2.5 GeV and a ring current between 110 and 220 mA. The Si(ll1) double crystal monochromator was detuned 50% to reject higher harmonics; rem lution, AE/E, was 2.0 X 1V. The data were recorded with the aample in a cellthat allowed treatment under vacuum or in flowing gases prior to the measurements. The grade-6 H2 was used aa received. The powder samples were pressed into wafers in a Nrfiied glovebag, with the wafer thicknem chosen to give an absorbance of 2.5. The XAS experiments were done at liquid nitrogen temperature. Meaaurementswere made at energiea near the Re Lm absorption edge (10534 eV). Detaila and information about the measurements of the reference Compounds me preaented elsewhere.* The data analyeiswas carriedout as deacribed previo~aly.~

the rhenium subcarbonyl on the almost fully (93%) dehydroxylated MgO. The remaining shoulder, present at about 1980 cm-l in both spectra, is assigned to trace amounta of [RSZ(CO)S]~-. The variation in the band locations was typically 5cm-1 from one samplepreparation to another; the variations are attributed to small differences in the preparation conditions, including the period of storage in the drybox, where traces of water evidently reacted with the support. EXAFS oftheRheniumSubcarbony1. EXAFSdata from five scans were averaged and analyzed by the difference file technique of Koningsberger et al.11C'-12 as before.31~An EXAFS spectrum is shown in Figure 2. The signal to noise ratio is estimated to be about 401. A comparison of the Re-0 phase-corrected Fourier transforms of the EXAFS functions characterizing the rhenium subcarbonyls formed on partially dehydroxylated MgO by treatment in H2 and under vacuum are shown in Figure 3. A similar comparison for the rhenium subcarbonyls on partially dehydroxylated and on almost fully dehydroxylated MgO is shown in Figure 4. The imaginary parta of the Fourier transforme characterizing the rhenium subcarbonyls that were supported on MgO at a single calcination temperature are almost identical, independent of whether the rhenium subcarbony1 had been formed by heating the adsorbed [HRe(C0)bI in H2 or under vacuum; the structures of these rhenium subcarbonyls are inferred to be the same.

Results IR Spectra. The IR spectra of supported rhenium tricarbonyl on partially and fully dehydroxylated MgO (Figure 1) are consistent with spectra reported earlier for similarly prepared sample^.^^^ The peaks in the carbonyl stretching region were located at 2011 (vs), 1895 (vs), and 1862 (sh) cm-' for rhenium subcarbonyl on the partially dehydroxylated MgO and slightly shifted to higher frequencies, at 2017 (vs), 1908 (vs), and 1867 (ah) cm-', for (9) Maloney, S. D. Ph.D. Dissertation,University of Delaware, 1990.

(10) Duivenvoorden,F. B.M.;Koningsberger,D. C.; Uh,Y.S.;Gate, B. C. J. Am. Chem. Sa.1986,108,6264. (11)Lengebr, B.J. Phyr. (Paris) 1986,47, 76. (12) van Zon, J. B.A. D.;Koningsberger, D.C.; van't Blik, H. F. J.; Sayers, D. E. J. Chem. Phyr. 1986,82,6742.

Langmuir, Vol. 8, No.11, 1992 2717

Ligand Effects in Supported Metal Carbonyls

c

E E

c!

; .a

Cil

-54 1

'

7

'

'

'

2

'

'

'

'

'

3

'

'

'

'

I 4

R, A acterizing the species formed by adsorption of [HRe(CO)sl on (A) M g 0 , ~followed by heating in Hz at 80 O C (solid line) and (B)M g O a followed by heating in Ht at 80 O C (dottad line). Table I1 A. EXAFS Reaulta4Characterizingthe Surface Species Formed after Heating of [HRe(CO)slon M&waqb shell N R A$,Az AE0,eV EXAFSreference 0.0031 6.65 OS4 R8-C 3.0 1.91 1.80 Re-0 2.8 2.13 0.0094 h-0-2.25 os-o* Re.-o* 3.2 3.12 4.0021

B. EXAFS Resulta Characterizing the Surface Species Formed after Heating of [HRe(CO)sI on MgOmo*b shell N R A$,A2 AEheV EXAFSreference 3.3 2.8 3.1

1.87 2.13 3.11

0.0025 0.0122 -0.0022

10.81 2.45 -2.62

os4 Re-0 OS-o*

Notation: N , coordination number for absorber-backecatterer pair, R, radial distance; A$, DebyeWder factor, difference with respect to reference compound;AEo, inner potential correction (correctionon the edge pition). Estimatedprecision: N , +15% (Re0-9

B

c'

Figure 4. Imaginary part and magnitude of Fourier transform (k'-weighted, Ak = 3.5-10.2 A-', Re-0 phase c o ~ ~ ~ t chared)

R8-C Re-0" Re-0'

e

E

*25%);R,*l%(Re-0n-fi*2%);

A$, *30%; AEo,+lO%.

However, the comparison of the rhenium subcarbonylson MgO samples with different degrees of dehydroxylation shows a significant difference between them. The nodes of the imaginary part of the Fourier transform characterizing the rhenium subcarbonyl on almost fully dehydroxylated MgO do not coincide with those of the rhenium subcarbonylon partially dehydroxylatedMgO (Figure 41, which implies a significant difference in bond distances and/or in the oxidation states of Re. The EXAFS analysis for the rhenium subcarbonyl on almost fully dehydroxylated MgO hae already been reported: with the results summarizedin Table IIA. The method of data analysis for the various samples is the same; it is also virtually identical to that used for the data characterizingthe MgO-supported rhenium subcarbonyls formed from [H&e3(CO)121.~ The EXAFS data were Fourier transformed over the useful range (3.54 < k < 10.33 A-1) with k3 weighting and no phase correction. The major contributionswere isolated by inverse Fourier transformation in the range 0.62 < r < 3.68 A. As before, only the R e 4 and Reo*contributions were used initially to fit the data (Figure 5A,B); as expected: their sum did not represent the observed data satisfactorily, since the ligands provided by the MgO surface were not yet accounted for (Figure 5A,B). In the next step of the analysis, a difference file was calculated by subtracting the initially estimated R e 4 and R e o * contributions from the experimental EXAFS function. The Reo phase corrected k3-weighted Fourier transform (3.54 < k < 10.33 A-l) of the difference file was calculated. The imaginary part of the Fourier transform

.-

L

a

2r.

.4.01 0

.

-0.81 0

.

. 1

.

,

2

.

R.A

V

.

,

I4

.

.

I4

3

e

9

I

E

.0)

a

,

I

.

2

.

3

R,

F'igure 5. Illustrationof stepsused in the differencefde technique for analpieof the EXAFS data characterizingthe sample formed by adsorption of [HRe(CO)sl on Mgqm followed by heating in HZat 80 O C : (A) imaginarypartofFouer transform (k*-weighted, Ak = 3.54-10.33 A-I, Re-0 phase correctad) of raw data (solid ' lien) and calculated ReQ*EXAFS (dashedline);(B)' partof Fouriertramform(ka-weightad,Ak = 3 . 6 4 - 1 0 . 3 3 ~ M phase corrected) of data (solid line) and calculated (Rsc+ Re0*)EXAFS (dashedline);(C) imeginarypart of Fouriertramform (k*-Weighted,Ak 5 3.54-9.5 A-', ReQ phase correctad) of data minus calculated (Re-C+ ReQ*)EXAFS (solid line) and calculated RB-oNpport EXAFS (dashed line).

of the contribution (Figure 5C), corrected for the R e 4 phase shift, is symmetrical and peaks positi~e1y.l~ Thie observation is consistent with the identification of the remaining scatterer as oxygen of the MgO support. The fmal fit and comparison with the data are shown in Figure 6A-C)both in k space and in r space. The comparison of the experimental and calculated Re-c and R e 4 * contributions is also shown in Figure 6D-E. The data are well represented by the fit. The parameter values are shown in Table IIB. The number of parameters is 12; the statistically justified14number is 14. XANES of the Rhenium Subcarbonyls. The Re Lm absorption edge data characterizing the rhenium subcarbonyls on MgO are shown in Figure 7. The intensity of the normalized threshold resonance of the absorption edge is virtually the m e for the two rhenium subcarbonyl sample8that were formed by decomposition of [Hb(CO)6] on the partially hydroxylated MgO; evidently, decomposition under vacuum and in H2 yields indistinguishable results (Figure 7B,C). There is, however, a marked difference between the results for these samples on partially dehydroxylatedMgO and the sample made with the almost fully dehydroxylatedMgO; the intensity of the absorption edge of the former is lower (Figure 7A,B). (13)van Zon, J. B. A. D. Ph.D. Dieeertation, Eindhoven University of Technology,T h e Netherlands, 1988. (14) Koningsberger, D. C.; Prim, R. X-ray Absorption, Principles, Applications, Techniques of EXAFS, SEXAFS, and XANES,Wiley New York, 1988,p 396.

2718 Langmuir, Val. 8, No. 11, 1992

Honji et 01.

e

noo

p

50

2s

(1I

'

25

'

50

1 5 % + & 4 5 Energy. cv

e

Figure 7. Structure of rhenium Lm absorption e characterizing the speciesformed by adsorption of [HRe(C )SI on (A) Mg0,00 followed by heating in HIat 80 OC, (B)Mgoww,followed by heating in H g at 80 O C , and (C) Mgoa followed by heating undervacuumat8O0C. CurveCisoffsethorizontallytofdtate

3

the comparison.

i

0

2

4

3

R. A

4.0 1

-4.0

I

0

2

1

3

4

3

1 4

R,A

-

4

0

. 1

0

2

R,A Figure 6. Resdta of EXAFS analysis obtained with the best calculated coordination parameters for the sample prepared by adsorption of [HRe(co)Slon MgOm followed by heating in H2 at 80 O C : (A) experimental EXAFS (solid line) and sum of the calculated (Rgc+ Re-O*+ R e - O d contributione (dashed

line); (B)imaginary part; (C) magnitude of Fourier transform (ks-weighted,Ak = 3.64-10.33 A-9 of experimentalEXAFS (solid h e ) and sum of the calculated (Rec+ Re-0+ R e - 0 4 contributions (dashed line); (D)imeginary part of Fourier traneform (ks-Weighted, Ak 3.64-10.33 A-', 0S-C phase corrected) of data minus calculated (Re-0 Re-O+) EXAFS (solid line) and calculated Rgc EXAFTdashed line); (E) imaginary part of Fourier transform (ka-weightad,Ak = 3.6410.33 A+, Os*+ phase corrected) of data minus Calculated Re-

+

+

O-+RecEXAFS(solidline)andCalcula~Re-O*EXAFS (dashed line).

Discussion The IR spectra (Figure 1) of the rhenium subcarbonyls on all the MgO supporta are consistent with the presence U of rhenium tricarbonyls with distorted C ~ symmetry.'-3 It is likely that various structures were present incorpo-

rating different surface ligands? but the resolution of the IR spectra is not sufficient to distinguish them. The vco bands of rhenium subcarbonyl supported on partially dehydroxylatedMgO are slightly shiftedto lower energies than those characterizing the rhenium subcarbonyl supported on almost fully dehydroxylated MgO. This comparison suggests a greater electron density on the rhenium in the former sample. The EXAFS data confirm the greater electron density on the Re bonded to the partially dehydroxylatedMgo, as follows: The Re-c bond distances are shorter and the C-O bond distances longer in these samples than in the rhenium subcarbonyl bonded to the almost fully dehydroxylated MgO (Table 11). The differences are explained by variations in the m e t a l 4 0 backbonding the greater electron density on Re on the partially dehydroxylated support can only result from greater electron donation to the Re by thissupport. The CO ligands donate electrons to the rhenium through a u-bonding interaction, and the rhenium backdonates electron density from ita fded d orbitals into the empty carbonyl r* (antibonding) orbital. The greater electron density on the rhenium on the partially dehydroxylated support causes greater backbonding from the metal to the CO ** orbital. This backbonding strengthens and shortens the R e 4 bond and weakens and lengthens the C-O bond in the rhenium subcarbonyl on this support. One might at f i i t expect the electron donation by the Mgo to be correlated with the surface basicity of MgO. Researches have estimated base strengths of a series of MgO samples dehydroxylated at various temperaturealblB by titration (withbenzoic acid and appropriate indicators). In all cases, base strength increased with increasing pretreatment temperature (increasing fractional dehydroxylation), until a maximum was reached at 6-20 "C. But these results do not lead to a clear distinction between the basicities of the partially and fully dehydroxylated MgO samplestreated at 400 and 700 OC. There are several limitations to this characterization of the surface basicity of Mgo: First, a distribution of base strengths of the surface sites is expected, and this is not (16)Tanabe, K.Solid Acida and Bases; Academic €'rea: New York. 1970;pp 60-63. (16)Malinowrki, 9.; had,M. Catalyub;SpecialirtPeriodical Report, Royal Society of Chemism Cambridge,USS; Vol. 8, Chapter 4.

(17)K i j e ~ k iJ.; , Malinmki, S.J. Chem. Soc.,FaradayT".I 1978, 74,260. (18)Hattori.. H.:. Yoshii.. N.:. Tambe. K. Roc. 6th Znt. Conn. Catal. 1973,1,233. (19)Malinomki, 5.;S?czepaneka,S.; Bielanaki, A.; Sloczyneki,J. d. Catal. 1966,4, 324.

-

Langmuir, Vol. 8, No. 11, 1992 2719

Ligand Effects in Supported Metal Carbonyls

Table 111. EXAFS,Cryrtallographic,and Infrared Rerultr Characterizing Rhenium Subcarbonylr and Molscular Analoguer C a ( a v e r e) distance3 1.24 1.21 1.19 [(CO)sReb-OCHs)sRe(CO)sl1.27 ~~co~sRe~~-ocsHs~sRe~co~3l1.17 structural formulation [(CO)sRe(oMg~)z{HoW),1 [(CO)sRe{OMg7ao)~{HOMg~~l [(CO)sRebs-OH)IE

RgC(aver e) distance,y 1.87 1.91 1.89 1.79 1.88

Rs-o(aver e) distance,y 2.13 2.13 2.21 2.08 2.13

(vco)

ref

2011,1896,1862(sh) 2017,1908,1867(sh) 1990,1875 2016,1897

20 21,22 23

"The OH ligands are H-bonded to benzene solvent rings, decreasing the OH donor ability of [(CO)&(OH)3].

accounted for in the simple titrations. Furthermore, although calcination temperature influences the base strength significantly,other variables are also important, e.g., impurities, the atmosphere and duration of calcination, and the nature of the precursor from which the oxide is prepared (e.g., MgC03 or Mg(OH12). Rather than attempting to relate the EXAFS results for the rhenium subcarbonyls to the MgO basicity measured simply by titration, we interpret the EXAFS data as a basis for evaluating the electron donor properties of the MgO surface; we regard the rhenium subcarbonyl as a probe of the local surface sites at which it is bonded. We expect the surface ligands of the rhenium subcarbonyls on the almost fully dehydroxylated MgO to be predominantly 02-,whereas those of the rhenium subcarbonyls on the other MgO samples are expected to include a substantial number of OH- groups; infrared spectra of a series of MgO-supported rhenium subcarbonyls confirm this expectation.s The surface is thus viewed as a source of electron density though rigid 02-and OH- ligands, and the supported complexesare represented as virtual molecular analogues of [(C0)3Re(OR)31,where R = H, CH3, or CsH5. These OR groups have widely different electron donor tendencies, in the order OCH3 > OH 2 OC6HS. The differences in electron donor tendencyare illustrated by crystallographic data characterizing the bonding between Re and the CO ligands (Table 111); the results are consistent with the pattern of electron donation mentioned above. The infrared data of Table I11 are explained by the backbonding, but they are only a weak indicator in comparison with the crystallographic data. Analogously, the infrared data characterizing the supported complexes are also only a weak indicator in comparison with the EXAFS data. The electron transfer can also be probed by XANES, provided that the surface complexes all have the same symmetry; the infrared spectra are consistent with the suggestion that they do, although they are not a precise indication. The Re LIIIabsorption edge data (Figure 7) show a lower absorbance by the sample on the partially dehydroxylated MgO. The intensities of the threshold resonance of the L absorption edge are related to transition probabilitiesof exciting inner-core 2p electronsinto vacant d valence levels. The higher the oxidation state of the metal, the greater the number of vacancies in the valence level, and hence the higher the probability of the transition. Assuming that the area of the normalized LIDthreshold resonance line is proportional to the number of vacant d states, the oxidation s t a b of metal is indicated qualitatively by the normalized LIII threshold resonance line.% The results indicate a greater electron density on the rhenium (20) Nuber, B.; Oberdorfer,F.; Ziegler, M. L. Acta Crystallogr., Sect. E 1981,37, 2062. (21) Ciani, G.; Sironi, A.; Abinati, A. Garr. Chim. Ital. 1979,109,615. (22) Beringhelli, T.; Ciani, G.;DAlfonso, G.;Sironi, A.; Freni, M. J. Chem. SOC.,Dalton Trans. 1986, 1507. (23)Ciani, G.; DAlfonso, G.;Freni, M.; Romiti, P.; Sironi, A. J. Organomet. Chem. 1978,152,85. (24) Lytle, F. W.; Wei, P. S. P.; Greegor, R. B.; Via, G.H.; Sinfelt, J. H.J. Chem. Phys. 1979, 70,4849.

on the partially dehydroxylatedMgO than on rhenium on the almostfully dehydroxylated MgO. A semiquantitative correlation between oxidation state and the intensity of the white line absorbance has been noted by Meitzner et al.= and Fung et al.26 The loading of Re in the samples impliesthat the average distance between Re centers was about 10 A; consistent with this estimate, there was no evidence in the EXAFS spectra of Re-Re interactions (althoughthese were evident in samples prepared from [H3Rea(C0)121~).Thus, to a good first approximation, the rhenium subcarbonylswere isolated from each other on the surface, justifying the interpretation given above. The rhenium subcarbonylsreported here are among the simplestand most thoroughly characterized surfacespecies that involve metals interacting with metal oxides. They are almost unique in terms of the quantitative structure data provided by EXAFS spectroscopy. Thus these samples would be of interest for characterization by the methods of theoretical chemistry, which have been applied previously, for example, to less well characterized Si02supported complexes formed from [Rh(allyl)al.27

Conclusions [HRe(CO)s] reacted with the surfaces of MgO samples that were about 55 and 93% dehydro~ylated,2~ being converted into surface-bound rhenium subcarbonyls, (CO)~OH)3-,{0)x, where the surface donor ligands are largely OH in the partially dehydroxylated sample and are largely 02-in the highly dehydroxylatedsample. The results provide strong evidence for electron transfer from the MgO through surface ligands to the supported Re, as shown by the C-O bond distances and confirmed by the white line absorbance. The surface complexes are analogous to compounds in the group [(CO)aRe(OR)sl. The electron-donor OR groups in these compounds explain the differences in bond lengths in the Re-C-O moiety; similarpatterns explain the bond lengths in the analogous complexes on the surface, which are attributed to the different electron donor properties of the OH- and 02surface groups.

Acknowledgment. We gratefully acknowledge the support of the U.S.Department of Energy, Office of Energy Research, Office of Basic Energy Sciences (ContractFGO287ER13790); we also acknowledge the U.S.Department of Energy, Division of Materials Science, under Contract Number DE-FG05-89ER45384 for its role in the operation and development of Beamline X-11A at the National Synchrotron Light Source. The NSLS is supported by the Department of Energy, Division of Materials Science and Division of Chemical Sciences under Contract Number DE-AC02-76CH00016. (25) Meitzner, G.;Via, G. H.;Lytle,F. W.;Sinfelt, J. H. J. Chem.Phya. 1987,87,6354. (26) Fung, A. S.;Tooley, P. A.; Kelley, M. J.; Koningeberger, D. C.; Gatea, B. C. J. Phys. Chem. 1991, 96,225. (27) Halet, J.-F.; Hoffmann, R. J. Am. Chem. SOC.1989, 111, 3548.