Inner-shell excitation of η5-C5H5Co(CO)2 and related compounds studied by gas phase electron energy loss spectroscopy

Journal of Electron Spectroscopy and Related Phenomena, 57 (1991) 137-164 Elsevier Science Publishers

137

B.V., Amsterdam

Inner-shell excitation of $-C&H&o (CO), and related compounds studied by gas phase electron energy loss spectroscopy E. Riihl’, A.T. Wen and A.P. Hitchcock Department of Chemistry, McMaster University, Hamilton, Ont. L&S 4Ml (Canada) (Received 25 February

1991)

Abstract Oscillator strengths for carbon ls, oxygen 1s and cobalt 2p excitation of CpCo (CO )z (Cp = $‘cyclopentadienyl) have been derived from inner-shell electron energy loss spectra recorded under scattering conditions dominated by electric dipole transitions. The spectral featurea have been assigned on the basis of comparison with the spectra of the free ligands and with previously studied gas phase organocobalt complexes, Cog (CO), and Co (CP)~, along with extended Hiickel molecular orbital calculations. Comparisons of the core spectra of free and complexed ligands provide information on metal-ligand bonding. A simulation of the Cls spectrum of CpCo(CO).. a weighted sum of the Co,(CO), and Co(Cp), spectra, duplicates all features of the experimental spectrum, implying that the virtual orbit.& of the ligands (as sampled by core excitation) are relatively independent of those of the other ligands. Quantitative deviations from this “building block” model reveal that the virtual MOs associated with Cp are more influenced by ligand-hgand interaction than those associated with the carbonyl ligand. These results provide a first view of the extent to which core excitation spectroscopy will be useful in studying the electronic structure of complex organometallic compounds with mixtures of ligands.

INTRODUCTION

Organometallic complexes of 3d transition metals are involved in a large number of synthetic and catalytic processes [ 11. Broadly speaking, these complexes either have coordination sites available for interaction with incoming molecules or they have oxidation states readily available for oxidative additions, rearrangements, reductive eliminations and other important chemical transformations. Cobalt carbonyl, cobalt cyclopentadienyl and their derivatives have been receiving increasing attention recently regarding both preparative and catalytic processes. In the latter context, these molecules offer mo‘Permanent address: Institut fIir Physikalische W-1000 Berlin 33, Germany.

0368-2048/91/$03.50

Chemie, Freie Universitit

0 1991 Elsevier Science Publishers

B.V.

Berlin, Takustrasse

All rights reserved.

3,

138

lecular analogues of small organic molecules adsorbed on metal surfaces and thus they give information concerning heterogeneous catalytic processes. Inner-shell excitation spectra obtained by either X-ray absorption (NEXAFS) [ 21 or inelastic electron scattering [ 31 have proved useful in investigations of the geometric and electronic structures of chemisorbed organic molecules. Gas phase inner-shell electron energy loss spectroscopy (ISEELS) has long been used to study small molecules. Recently it has been applied to organometallic systems, such as transition metal carbonyls [4-71and metallocenes [ $1. With precise measurements and careful interpretation, core excitation energies and intensities can be used to investigate ground state bonding and electronic structure as well as relaxation effects associated with the creation of core excited states. In this paper we extend our studies to consider for the first time a molecule containing different types of ligands bound to the same metal atom. The goal is to determine whether the core excitation spectra of such complexes can provide information on how the electronic structure is affected by ligand-ligand interactions mediated by the central metal atom. If successful, such studies will help to understand the electronic interactions between small molecules and metal atoms both in organometallic complexes and in cases of co-adsorption of organic compounds on metal surfaces. In the present work, we have used ISEELS to record the gas phase core excit_ation spectra (CoSp, CoZp, Cls and 01s) for three organocobalt complexes Coz(CO)s, Co(Cp), and CpCo(CO), (Cp=$-C5Hs). The Cls spectra of cyclopentadiene (CpH) and its Diels-Alder dimer dicyclopentadiene WPHM have a1so been recorded to provide information concerning the Cls spectrum of the free cyclopentadienyl ligand. The Cls spectrum of Co (Cp), has been reported and interpreted in detail previously [8] while the Cls and 01s spectra of Coz (CO), have been presented briefly in comparison with those of other transition metal carbonyls [5].The present work emphasizes ligand core excitation spectra, in particular analysis of the Cls spectrum of CpCo (CO )z in relation to the spectra of Co (Cp) 2 and Co2 (CO) e The spatially-localized character of core excitation means that sums of model compound spectra (the “building block approach” [9] ) can be used to facilitate analysis of the core spectra of more complex species. Differences from additivity predictions may then be related to ligand-ligand interactions. The Co2p and 3p spectra of all three cobalt species have been discussed recently [lo]. An extension of the previous spectral interpretation is given. As in our previous work [8,10] extended Hiickel (EHT) calculations, carried out within the equivalent ionic core virtual orbital model (EICVOM [ 11 ] ), are used to assist spectral assignments. An interesting aspect of the discussion of the electronic structure of Co, (CO )8 has been the existence and correct description of the Co-Co single bond that is predicted for this compound by the 18 electron rule [ 121.The virtual molec-

139

ular orbit& of CpCo (CO), have been studied previously by semi-empirical quantum calculations [ 131 and electron transmission spectroscopy (ETS ) [ 141. The nature of inter-ligand charge transfer [ 151 and the distortion of the Cp ligand from five-fold symmetry [ 161 are important themes in electronic structure studies of CpCo(CO),. EXPERIMENTAL

The inner-shell electron energy loss spectrometer and the operating techniques have been described previously [ 17 1. The spectra were obtained by inelastic electron scattering with a high velocity incident electron beam (EO= 2.5 keV plus the energy loss), small scattering angle ( < 2 O) and 0.6 eV fwhm resolution. The organocobalt samples, purchased from Strem Chemicals, were used directly without further purification. In order to achieve adequate vapour pressure in the collision region of the spectrometer for these low volatility complexes, the sample cell, inlet line and collision chamber were heated to 50°C. The Cls and 01s spectra were converted to absolute optical oscillator strengths (f values) using a procedure described and tested previously [ 181. A precision of 5% was estimated from the reproducibility of analysing independent data sets while the absolute accuracy is about 20%, according to comparisons with other known optical oscillator strength data [ 181. The precision is sufficient to detect the rather small shifts in ls+ti oscillator strength per carbonyl group [ 4,5 1. The EHT calculations are carried out within the EICVOM model [ 111, i.e. using the “Z+ 1” equivalent core to account for the core hole potential and a unit valence charge to obtain the correct valence electron count. The calculations for CpCo(CO), were carried out with the Cls hole on each of the four symmetry inequivalent carbon atoms. The virtual orbital energies of the ground state of the cation are assumed to approximate the relative energies (term values ) of core+vaIence excited states, while the intensities of the core excitation to a given virtual orbital are assumed to be proportional to the sum of the squared coefficients of atomic orbitals on the core excited atom, which are electric dipole coupled to the excited core orbital [ 19 1. For example, Cc2 (N2p) in the EHT result for COAX+ is used to simulate the Cls spectrum while Cc2 (Ni3d) in the EHT result for CoNi (CO): is used to simulate the Co2p spectrum. This approach assumes that the overall transition matrix element is dominated by the atomic-like intra-shell terms, which is reasonable on the grounds of spatial overlap with the very compact core orbital [20]. As_ found in previous work [8,10] EHT with this approach seems to reproduce the shapes and relative positions of core excitation spectral features of organotransition metal complexes, giving some confidence in using the virtual MO character as a guide to spectral assignment. The Htickel parameters chosen by Albright et al. [ 211 have been used for consistency with our previous work

140

[8,10], in which the reliability of these parameters has been addressed. The

geometries used for Co (Cp), and CpCo (CO), are those determined experimentally [22,23 J. A non-bridged Dllhgeometry was used for the Co2 (CO) Bcalculations since this has been proposed to be the stable form in the gas phase

WI.

.5 -

.o 7

b

S-

OIO 0

6 4

2 0

IO -

CP

cpco (CO)2

E

&_

s-

O-

~~

i....l....l....l.,.

260

290

310

300

Energy


Fig. 1. Oscillator strengths for Cls excitation of CO, Coz(CO), and CpCo(CO), derived from ISEELS spectra recorded using 2.5 keV final electron energy, 2 o scattering angle and 0.6 eV fwhm resolution. The hatched lines indicate the location of the Cls ionization potentials as determined by XPS [34].

141

RESULTS AND DISCUSSION

Cls spectrum of Co,(CO), and 01 s spectra of Co&O),

and CpCo(CO)2

The Cls and 01s oscillator strength spectra of CO, Co2(CO)s and CpCo (CO) 2 are shown in Figs. 1 and 2. In each case the spectra are plotted on common vertical scales with the oscillator strength divided by the number of carbonyls in the molecule in order to emphasize the systematics of the intensities of carbonyl core excitation. The energies and proposed assignments of

cpco X0) *

540 Energy

550

560

(eV>

Fig. 2. Oscillatorstrengthsfor 01s excitationof CO, Co2(CO), and CpCo(CO),. See Fig. 1 for furtherdetails.

-7.9

304.0

303.4

293.40 298.8

293.1

287.4b 268.3 269.3 299.4

EW)

co*ccoh

- 10.0

-5.4

0.3

6.1 5.1 4.1 3.0

TW)

297.3 300.2

292.73

289.2 290.3 290.64 291.4

- 6.6

-0.8

3.8

286.6 287.7’

-8.1

3.5 2.4

5.0

WI

W) 5.3

T(CO)

302.0

297.7

289.4 289.96 291.9

- 12.0

-7.7

-1.9

0.6

5.2 4.7 3.1

Shake-up

respectively.

bc(C-0)

IP Cod

Ir*&h?d *ddoclli R*dslod 3P

Ir(l(C=o)

co

(final orbital )

Double excitation a*(C=C)

Rrdberg IP Cpd o+(C-C)

4%,(Co3d) ~t(‘=d) n+(C=C)

Cp

E(eV)

284.8 285.3 266.8b

Assignment

and WCP),

CO(CPh TW)

of CO, Cop(CO)s. CpCWCO),

TWP)

265.4b

E(eV)

CPCO(CO)2

for featuresin the Clsspectra

*T=IP-E. bCalibration relative to CO,(Cls-r~=290.74 eV): LIE= -3.31 eV, -3.25eV and -3.92 eV for Cs(CO),, CpCo(CO)z and Co(@)2 DThe energies of these components of the main peak (maximum at 287.5 eV) have been derived from a fit to gaussian line&apes. dIPs from XPS [ 34 1. This is the energy of the main XPS line.

-4.7

2.8

293.3

296.1 300.8

8.7

T(eV)’

287.40

E(eV)

co

Energies,termvaluesand proposed assignments

TABLE 1

143 TABLE 2

Energies, termvalues and proposed assignments for features in the 01s spectra of CO, Co, (CO )8 and C~CO(CO)~

co E(eV) 534.11

Co2KJO)s UeV) 8.3

541.0 542.4

1.4

550.9

- 8.5

E(eV)

Assignment (final orbital )

CPWCOL

T(eV)

WeV)

T(eV)

534.1”

5.7

533.7”

5.3

7t*(GO)

536.7

3.5

538.6 539.78 545.6

1.1

535.6 536.4 537.6 538.97 544.7

3.4 2.6 1.4

?I? &b?al ti RJz?zg IPb

-5.8

548.9

-9.9

549.9

-5.8

- 10.2

dCalibrationrelativetoO,~Ols~Ir*=530.80eV):dE=3.26(8) and CpCo ( CO ) r respectively. bIP from XPS [ 341. This is the energy of the main XPS line.

eVand2.9118)

Double excitation o* (C-O) eVforCot(CO)a

TABLE 3

Oscillator strengths’for 1s+ tr* transitions derived from experimental spectra Species

01s z*WpIb

co CHICO), WCP), (=Pwco)z

0.045 0.077

OS40 0.145

#(co)=

*(cop

0.170 0.162d

0.076 0.061

0.143

0.065d

“Integrated peak areas. bIntensity per Cp group (symmetry label refers only tc Co (Cp)* ) . “Intensity per CO group. dThese values differ slightly from those reported in ref. 5 because of improved background subtraction.

the Cls and 01s spectral features for all species are presented in Tables 1 and 2. As found in previous studies of metal carbonyls [4-71, the Cls and 01s spectra of Co2 (CO ) 8 are dominated by ti and CPresonances similar to those of free CO, indicating that Cls and 01s excitations are localized on a single carbonyl. This is in sharp contrast to the Cp ligand where there is a major qualitative change in its Cls spectrum when it is bound to a metal atom (see ref. 8 and the following section). While the main features are similar, close inspection reveals that there are some differences between the spectra of cobalt carbonyl and CO, particularly

a*(M-CO), IC2p,,+Co&y) lc*(2p,) +Co(dzr -cL 1

x’a%)

bu

am

bz,

Co(d,--d,) Co(d,+dJ

d(M-CO), ti(M-CO),

Character

bti en

eg

b 1U

6% b, bu %

b lU

&h

Virtunl MO

I

CCo3d 0.360 0.372 0.000 0.000 0.086 0.086 0.002 0.002 0.002 0.036 0.036 0.201 0.081 0.081 0.210 0.089 0.008

Energy (eV) -4.25 -4.73 -7.79 - 7.91 - 8.28 - 8.28 - 8.52 - 8.69 - 8.69 -8.71 -8.71 -9.09 -9.11 -9.11 -9.21 - 9.23 -9.33

Ground state

0.059 0.053 0.145 0.143 0.229 0.012 0.133 0.103 0.150 0.083 0.155 0.065 0.091 0.122 0.082 0.096 0.116

Xzp

Properties (E, IF) of selected virtual MOs of the ground and excited states of Cs(CO), metry” derived from extended Hiickel calculations

TABLE 4

0.006 0.007 0.035 0.036 0.060 0.003 0.037 0.030 0.044 0.025 0.047 0.028 0.031 0.042 0.028 0.031 0.042

COZP

with Ddhsym-

- 4.46 -5.44 - 7.84 -8.11 -8.29 - 8.39 -8.63 . - 8.70 -8.79 - 8.97 - 8.98 -9.10 -9.17 -9.18 - 9.30 - 10.24 - 10.74

E(eV)

0.028 0.728 0.000 0.010 0.063 0.039 0.009 0.030 0.004 0.175 0.040 0.130 0.690 0.089 0.041 0.111 0.167

CCo3d

Cls excited state

0.058 o.os9 0.007 0.010 0.007 0.605 0.002 0.000 0.007 0.003 0.024 0.001 0.004 0.000 0.002 o.r91 0.702

EC2p 0.003 0.008 0.000 0.002 0.000 0.001 0.000 0.000 0.001 0.001 0.004 0.004 0.001 O.ooO 0.000 0.070 0.315

102P - 4.29 -4.79 - 7.82 - 7.96 - 8.28 -8.33 - 8.57 -8.70 - 8.70 -8.75 -8.83 -9.10 -9.15 -9.16 -9.27 -9.34 -9.46

E(eW 0.287 0.453 0.000 0.001 0.088 0.068 0.003 0.001 0.028 0.039 0.051 0.135 0.097 0.031 0.074 0.304 0.051

Xo3d

01s excited state

0.038 0.050 0.043 0.130 0.006 0.096 0.089 0.011 0.001 0.097 0.226 0.002 0.044 0.056 0.043 0.509 0.397

cc2p 0.004 0.006 0.006 0.018 0.000 0.013 0.012 0.001 0.000 0.014 0.034 O.OUO 0.007 0.009 0.006 0.083 0.066

co2p - 4.48 - 7.79 - 7.91 -8.06 - 8.43 - 8.43 - 8.58 -8.70 - 8.70 -a.79 - 8.79 -9.15 -9.39 - 9.39 -9.43 - 9.47 - 10.16

E(eVI 0.001 0.000 0.000 o.31gb 0.076 0.076 0.043 0.000 0.000 0.006 0.006 0.000 0.033 0.033 0.047 0.016 0.108

ECo3d

Co2p excited state

0.004 0.145 0.143 0.097 0.091 0.006 0.165 0.011 0.301 0.096 0.127 0.001 0.229 0.106 0.103 0.082 0.189

cc2p

0.028 0.085

0.000 0.035 0.036 0.021 0.024 0.002 0.046 0.002 0.090 0.029 0.039 O.OOO 0.082 0.038 0.039

cozp

“The non-bridged D4,, geometry taken from ref. 24 is the minimum energy configuration proposed to be the most stable gas phase conformer. bUnderlined values ( ,Zc* > 0.05) are expected to correspond to major transitions.

17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33

Virtual MO#

146

in the region between the a* and dc resonances. These differences include: small and opposite shifts in the transition energies for the carbon and oxygen ls-+~* excitations; additional weak bands 2-3 eV above the main lz* resonances; suppressions of the Rydberg structure; and a reduction of f( Z* ), the ls-+n? oscillator strength. We have shown previously that the reduction of f( K*) can be used to estimate the extent of dn-pn backbonding in metal carbonyls [ 51. The sensitivity of Is-, K* intensities to this aspect of metal-ligand bonding has been documented in the DV-Xar calculations of CO and NiCO clusters by Kojima et al. [25]. Oscillator strengths for the low energy ls-+ti transitions derived from the experimental spectra by curve fitting to gaussians are summarized in Table 3. The small shifts in the It* resonance, of opposite directions in the Cls and 01s spectra, are associated with differential relaxation effects [ 4-7,26,27]. The weak features 2-3 eV above the main ls-*~c* resonance in the Cls and 01s spectra of Co, (CO )8 are interpreted as transitions to delocalized n* states. According to MO calculations [ 28-311 the II*(CO) orbitals in the ground state of metal carbonyls can span an appreciable energy because of interactions among a* orbit& of adjacent CO ligands. Our EHT results predict that there are nine virtual MOs of strong n* (CO) character which span 1.5 eV in ground state Co2 (CO), (Table 4). Upon Cls excitation the energy spread increases to 3 eV but all of the n* oscillator strength is concentrated in excitations to the two lowest energy Ir* MOs, those localized on the core excited carbonyl. The C2p contribution to the LUMO changes from 0.12 in the ground state to 0.70 in the Cls excited state (Table 4). The other n* components in the EHT calculation of Cls (01s) excitation correspond to the ( ls-‘, K&,~~ ) states. The Cls and 01s continua of Co, (CO), are dominated by a broad a* (C-O) shape resonance which occurs about 1 eV lower in energy than in free CO. This shift to lower energy, which has been observed in the Cls and 01s spectra of other transition metal carbonyls [ 4-7 1, is associated with weakening and lengthening of the C-O bond upon metal complexation [4,5,25,31]. This is consistent with the conventional model of the linear M-C-O bond as one involving two interactions, a-donation and n-back-donation. The latter weakens the C-O bond which lengthens and approaches a double bond in character. Increased bond lengths have been deduced from vibrational spectra as well as from direct measurements of the structure. The shift in absolute Cls+ br (C-O) energy is consistent with the structural change within the bond length correlation [32 1, although there are complications in the application of this model to organometallics because of large differences in core hole relaxation with and without an adjacent metal atom [ 4 3. An interesting aspect of the Cls spectrum of Cop (CO), (Fig. 1) is the absence of any spectral features associated with Co-Co antibonding orbit&. Since the formal single bond between the two Co atoms has a rather low bond strength, the “weak bond” concept [ 331 predicts that there should be a 8 (Co-Co) or-

147

bital at low energy. DV-Xa MO calculations for Co2 (CO), [ 121 predict an a2” LUMO of Co-Co character which is somewhat isolated from the higher energy orbit& which are largely of It*(C-O ) character. In our EHT calculations for the ground state, the a2,,orbital has dt (Co-Co) antibonding character but it is 0.1 eV less stable than the Izt(CO) LUMO (see Table 4). Since EHT indicates this orbital also has moderate n*( C-O) character, one might expect Cls-+azu “8 (Co-Co)” transitions to produce a shoulder to the n* feature, perhaps on the low energy side if the weak-bond hypothesis is adopted. However, the Is-rti peak appears to be rather regular on the low energy side. Previously [ 41, a comparison of the core spectra of Mn, (CO) 10with several monometallic Mn carbonyls also did not reveal any evidence for a low-lying C 1s-+ dr (Mn-Mn ) transition which had been predicted to exist in that species [33]. It was tentatively concluded that such a transition was not observed in Mnz (CO),, because of spatial overlap limitations. The EHT results for Co, (CO), predict that the relatively localized d’(Co-Co) character of the azu orbital is lost with Cls excitation so that it is unlikely that a spectral feature assignable to Cls-+dr(Co-Co) can be identified. The corresponding Ols--+dc (M-M) transitions would be even less likely. On both symmetry and spatial overlap grounds, it is more likely that the metal np edges would contain spectral signatures of a low energy &(M-M) orbital. Although the EHT results suggest that the second virtual orbital retains a distinct b’ (Co-Co) character during Conp core excitation, no evidence was found for low-lying Conp-+# (Co-Co) excitations [lo] (see Co2p and Co3p spectra). A comparison of high resolution core spectra (Cls and Co2p) of Co,(CO), and HCo (CO), might be a more fruitful way of searching for evidence of core-b a* (Co-Co ) transitions. The 01s spectra of CO, CO~(CO)~ and CpCo(CO), are compared in Fig. 2. The main R* and a* resonances are found to have very similar shapes and energies in all three species, while the Rydberg features found in free CO are weaker or absent in metal-complexed CO and f ( TC*) is considerably reduced. The 01s spectra of Co2 (CO )8and CpCo ( CO)z are very similar, as expected from the spatially localized character of core excitation. The maximum of the a* (CO) resonance is about 1 eV lower in CpCo (CO), than in Co2 (CO),. This is probably associated with the chemical shift of -0.8 eV between the 01s IPs of CpCo (CO ) 2 and Co, ( CO )8 [ 341, which is associated with electron donation from the Cp- anion to the Co atom which, in turn, enhances back-donation into the z* (CO ) orbital, thereby transferring electron density from the CpCo fragment to the carbonyls, and decreasing the 01s IP [ 151. Cls excitation of Co(Cp), and models for the free cyckpentadienyl Eigand

In order to understand how x-bonding to a metal atom affects the Cls spectrum of a Cp ring it would be useful to compare the C 1s spectrum of cobaltocene

148

with that of an isolated Cp ring - either as the Cp’radical or the Cp- anion, which are limiting electronic models for Cp bound to a metal atom [22 1. Unfortunately, such a direct comparison is not yet possible since the Cls spectrum of C,H, has not yet been obtained. Nevertheless, a reasonable estimate of the Cls spectrum of the C&H; anion can be made from its well-known MO structure [35] aIong with the experimental spectra of closeIy related species such as cyclopentadiene (Cp-H) and those of aromatic analogues such as benzene and pyridine. The latter are perhaps more appropriate analognes than Cp-H, since they are six-electron aromatic systems like the Cp- anion. This isoelectronic nature should lead to similar orbitals and thus similar carbon 1s excitation, although benzene does have one more Z* orbital than Cp-. In order to develop insight into the Cls spectrum of Cp- we first discuss the spectra of Cp-H and its Diels-Aider dimer, (Cp-H),. The Cls spectra of cyclopentadiene, cyclopentene and dicyclopentadiene ( [Cp-HI,) are presented in Fig. 3. The Cls spectrum of cyclopentene [36] has been reported earlier while this is the first presentation of the spectra of Cp-H and [ Cp-H] 2. The energies and proposed assignments for the spectral features of Cp-H and ( Cp-H), are listed in Table 5. The Cls excitation spectra

Energy

(eV)

Fig. 3. Comparison of the Cls oscillator strength spectra of cyclopentadiene (Cp-H), tene (C,H,) and dicyclopentadiene [ (Cp-H J2]. See Fig. 1 for experimental details.

cyclopen-

149

TABLE 5 Energies, term values and proposed assignments for features in the Cls spectra of C5H,-H, H)*, C,H, and C&H; (inferred) Feature

Cp-H E (*O.leV)

2(B)

284.7’ 281.4

3(C) 4(D)

290.5(e)= 291.0 297.0

1(A)

GHs

(CP-Hh

T(eV)

5.6 3.1

-0.5 -6.5

E (f0.1 285.1” 287.4 290.4 290.6 294.3

‘calibration relativeto CC&:AE= -6.02 36 for calibrationdata for C& bIP from XPS [ 341. %stimated from IPs of similar species.

T(eV) eV) 5.3 3.0

-0.2 -3.9

E (+_O.leV) 285.0 287.6 281.2 290.3 290.6 295.6

TteV)

5.3 2.7 3.1 -0.3 -5.3

(CsH5-

GH,_

AEkgnment

E(est)

(final orbital)

285.0(5)

n*(C-C) @(C-H) 3p Rydberg IPb ti(C-C) dc (C-C)

287.0(5) 290.5(e)” 291.0(5) 297(l)

eV and - 5.66 eV for C,H,-H and (CSHb-H)2 respectively.See ref.

of Cp-H and [Cp-H ] 2 both exhibit an intense low energy peak around 285 eV which is assigned to Cls-+a* (C-C) transitions. Relative to the Cls continuum this peak is more intense in Cp-H than in the dimer, consistent with the higher density of double bonds in Cp-H. Ir-conjugation in Cp-H shifts the Cls+ti transition 0.4 eV lower than that in the dimer and cyclopentene. The second peak around 287 eV in [Cp-H ] 2 and C5Hs is assigned to a mixed Rydberg/ valence state of predominantly d” (C-H) character associated with the saturated regions in these molecules. The corresponding second feature in Cp-H has contributions from Is-, 2ti transitions in addition to 1s + dc (C-H ) . The two broad features in the Cls continua of the spectra (around 291 and 297 eV respectively) are assigned to shape resonances associated with the C-C bonds. The intensity of the lower energy a* (C-C) shape resonance is highest in [ Cp-H ] 2, consistent with this molecule containing the greatest number of carbon-carbon single bonds. The higher energy d’ shape resonance is associated with the shorter carbon-carbon double bond in the unconjugated species. It is part of a characteristic two-peak pattern of multiple dc features in the conjugated species [ 37 1. We have estimated the Cls spectrum of the Cp- free ligand from those of Cp-H and aromatic analogues. This estimate is compared with the Cls spectra of Co ( Cp ) 2 and benzene in Fig. 4. Since the spectra of benzene and Cp-H are dominated by the lowest energy Cls+x* transition, we expect that the low energy R* (C=C ) region of the Cp- spectrum should be dominated by a strong single peak (A). Based on the spectra of benzene and pyridine, peak A is expected to occur around 285.0f0.5 eV. Above the Cfs IP the Cp- spectrum is expected to be characterized by two strong bands (C and D) associated with

,....l....l....‘.‘..‘....“...

280

290

m

Energy

310

ZKO

(eV>

Fig. 4. Cls oscillator strength spectrum of COG compared with that of benzene and that estimated for a free cyclopentadienyl anion (Cp’ ). See text for justification of the Cp’ spectrum.

the characteristic two-peak pattern of d” (C-C) shape resonances observed in conjugated systems. Recently Schwarz et al. [38] and Medhurst et al. [39] have argued that peak C in benzene is a doubly excited discrete state, perhaps overlapping with a one-electron Cls-+,d(C-C) transition [38]. Band C in Cp- is estimated to be located at 291.0 2 0.5 eV, a sharp peak with the second highest intensity. A second continuum dc resonance, band D, should occur around 297 + 1 eV. From comparison with the pre-edge region of benzene, a weak band (B) is inferred around 287.0 + 0.5 eV, just below the Cls IP, which would be associated with a transition to a state of mixed 3p Rydberg, ti (C-H) character. Clearly, there is a strong qualitative difference between the single IC*feature expected in free C, H< and the two discrete features observed in cobaltocene and other metallocene spectra [8 1. We now turn to the Cls spectrum of a Cp ring bonded to a metal atom as in cobaltocene and CpCo (CO) 2. The energies and assignments for the Co ( CP)~ spectral features are summarized in Table 1. The spectrum of COG has been discussed extensively earlier, in comparison with those of Fe(Cp), and Ni ( CP)~ [S]. A brief summary of our interpretation of the cobahocene spectrum is given here in order to assist the interpretation of the CpCo(CO), spec-

151

trum and to point out an additional feature not detected or assigned earlier. The Cls spectrum of cobaltocene, and other metallocenes [8], exhibits two sharp resonances in the pre-edge region and two broad bands above the Cls IP. Relative to the spectrum expected for the free Cp- ligand, there is an additional low-lying band (Fig. 4). This is attributed to Cls excitations to a 4e19 orbital of mixed R* (Cp ) and metal d,,& character. Curve fitting (Fig. 7a) clearly shows that there is a high energy shoulder on the first peak which may correspond to transitions to a second electronic state arising from the Cls-‘, 4e& ) configuration. The main peak in the cobaltocene spectrum is ascribed to Cls + 3e,, ( n* ) transitions, with minor contributions from Cls+ 4e2, (n* ) excitations. The ti (Cp) 3e2J4eZg peak occurs at 286.8 eV, higher than its counterpart in the inferred, free Cp- spectrum ( m 285 eV). The increased energy is attributed to the interaction between the metal 3d and Cp e$ (II?) orbitals. The two continuum bands around 292 and 298 eV are assigned to one-electron Cls+ ti (C-C) transitions. Cls

spectrum

of CpCo(Cc&

A molecular orbital diagram for the ground state of CpCo (CO )2, based on EHT calculations and consistent with previous theory [ 13 1, is presented in Fig. 5. The EHT results for the ground state and various core excited states are summarized in Table 7. The dominant interactions of the Cp- ligand with the Co (CO),+ fragment occur between the filled ey set of the Cp- ring and the filled or empty “molecular” orbitals of Co (CO)2 , arising from the appropriate linear combination of Co and CO MOs. The HOMO in CpCo(CO), is Cod,, and can be regarded as a non-bonding electron pair at the metal, stabilized by backbonding. In the ground state the LUMO and next-to-LUMO orbitals are heavily concentrated on the uncoordinated portion of the Co and have some CO character but very little contribution from n&. Both HOMO and LUMO, originating from the Co (CO),+ fragment, are destabilized by antibonding interactions with the Cp-ey orbitals (see Fig. 5 ). There are five virtual MOs above the LUMO and next-to-LUMO. Three of these five MOs are predominantly Co3d in character, one with some a*( C=O) character, the other two with some ti (M-CO) character; the other two MOs are ti (C=C ) orbitals of a’ symmetry which are the Cp-e; set interacting weakly with filled frontier orbitals of the Co (CO), fragment. Two factors complicate the application of the ground state MO scheme to the interpretation of the core spectra of CpCo ( CO)z. Firstly, the orbital character can change rather dramatically with addition of a localized core hole. Secondly, the Cls spectrum of CpCo (CO )2 is the overlap of transitions from Cp carbons and from CO carbons. Transitions to the same virtual orbital in the two sets of localized transitions should be separated by ~2 eV since the Cls (Cp) IP is 2.1 eV lower than the Cls (CO) IP [34]. This may be taken

Fig. 5. Sketch of the frontier molecular orbit& of CpCo( CO), in terms of the overlap of the orbitals of Cp- and Co (CO ),’ fragments, based on EHT calculations for the ground state species (see also Table 7 ) .

into account in making correlations between EHT orbital energies and exper-

imental results either by comparisons via term values relative to each of the IPs, or by shifting the EHT predictions for Cp excitations several electronvolts below the EHT predictions of CO excitations prior to comparison with the experimental spectrum. The Cls spectrum of CpCo (CO), (Figs. I,6 and 7; energies, term values and proposed assignments in Table 1) is characterized by four strong resonances, two below the Cls IP and two in the continuum. These are assigned by a combination of spectral comparison and the EHT results. For the latter it is essential to use the equivalent core analogy. For example, when the core hole is placed on the Cp ligand the C2p (Cp) character of the LUMO increases considerably (0.024 to 0.158) and the orbital adopts a distinct M-7$(Cp) character, similar to that assigned for the Cls+LUMO transition in the metallocenes [B]. Thus we assign the first feature at 285.4 eV to Cls (Cp)-d&z,z,n* (Cp ) ) transitions, consistent with the spectrum of cobaltocene. The second intense peak is composed of a main transition at 287.7 eV and a

153

cpco cc012

l....l....l....l....~,

285

290

300

295

Energy

305

CeVI

Fig. 6. Comparison of the Cls oscillator strength spectrum of CpCo( CO jz with a simulation consisting of one-quarter of the Co, (CO), spectrum and one-half of the Co (Cp), spectrum. The two components of the simulation spectrum are shown in the lower part of the figure on the same vertical scale.

second peak at 286.8 eV which produces a distinct shoulder (see Fig. 7a). This band is attributed to the overlap of a lower energy Cls(C) -+a*(Cp) component and a higher energy Cls (CO)+ti(CO) component, based on comparison with the spectra of Co (Cp), and Co, (CO),. A simulation of the Cls excitation spectrum of CpCo (CO) 2 has been prepared by adding 0.25 times the Cls spectrum of Co2 ( CO)8 to 0.5 times that of Co ( CP)~ (Fig. 6). The simulated spectrum has essentially the same shape as the experimental one and reproduces all major features in the CpCo(CO), spectrum, consistent with the general observation of spatial localization of core excitation. Tracing the origin of the second feature in the simulation indicates that it is the overlap of Cls-+n*(Cp) and Cls+a*(CO) transitions with the Cp component at lower energy. The EHT calculations generally support this assignment except that, if the experimental separation of the CO and Cp Cls energies (2.1 eV [34] ) is employed, they suggest that the relative order of the Cp and CO n* components should be reversed. By analogy with cobalt carbonyl, the relatively weak feature at 289.2 eV is considered to be transitions to higher energy x&,ocd orbitals of carbonyls. It

Extondod

Hijckel

co‘Cp)a

L -10

-6

-6 -4 Terns Value

-2

WI>

0

-14

-12

Orbital

-10

-8 Energy

-6
-4

Fig. 7. (a; left) Expansions of the discrete portion of the Cls spectra of Co,(CO),, C~CO(CO)~ and Co (Cp )* on a term value scale with resolution into individual components by fitting to gaussian lineshapes. Only the most significant components are plotted. (b, right) Predictions of the Cls spectra of Coz(CO),, CpCo(CO), and Co(Cp>, based on extended Hiickel calculations (Tables 4, 6 and 7). Gaussians of 1.2 eV fwhm at the indicated orbital energy positions with areas given by Ic 2 (N2p) were added to generate the simulated spectra. The Cp and CO components of the simulated Cls spectrum of CpCo (CO ) 2 are shi%d by 3.0 eV and the orbital energy scale refers to that for CO excitation_ The measured chemical shift is only 2.1 eV 1341.

could also contain some contributions from Rydberg or metal 4p excitations, as suggested in previous discussions of core excitation [ ‘7 ] and ETS results [ 141. However, the EHT calculations predict negligible Co4p contributions to any of the negative energy virtual orbitals. The two strong bands at 291.8 and 297 eV in the continuum are ascribed largely to Cls-, @(C-C ) transitions at the Cp ligand. The broad band around 297 eV will also contain the Cls+dc (C-O) resonance which produces the shoulder around 301 eV. While the simulated and experimental Cls spectra of CpCo ( CO)z are in semi-quantitative agreement (Fig. 6)) there are quantitative discrepancies. One of the central goals of this study is to determine if deviations from an additivity model can provide information on ligand-ligand interaction in organometallic complexes. The most n&eworthy difference is that the first, 3d,,, resonance

155

is more intense in the experimental than in the simulated CpCo{CO), Cls spectrum. The EHT results are in qualitative accord with this observation (Tables 6, 7). The estimated intensity of the Cls (Cp) -+LUMO transition increases from 0.075 in Co (Cp), to 0.158 in CpCo (CO), (N.B. as noted earlier [ 8 1, the value for Co (Cp )2 includes a 50% intensity reduction relative to that indicated by the squared LCAO coefficient to correct for 50% occupancy and thus Pauli-excluded transitions). This is in reasonable agreement with the oscillator strength which increases change in the experimental Cls-+3d,,, from 0.045 to 0.077 between Co (Cp), and CpCo (CO Jz (Table 3). There are two possible interpretations of the increased intensity. The first is that it is associated with transitions into low-lying n? (CO) virtual orbitals. While the ground state EHT result indicates the LUMO of CpCo (CO), has z* (CO) character, the LUMO adopts a larger I? (Cp) character when a Cls(Cp) core hole is added (Table 7). The suggested Cls(Cp) +n*(CO) excitation would be of charge transfer character. These types of transition are generally weaker and shifted to higher energy relative to a frozen ground state orbital prediction. It is unlikely that the extra intensity in the first peak is associated with a direct Cls (CO) +n* (CO) component on account of the 2.1 eV chemical shift. The alternative and most likely explanation is that the dn*pz* orbital has a larger C2p character in COCOS than in CoCp,. If this. is the case, there would most likely be a larger C2p component in the counterpart dn-pa bonding orbital, thus implying a stronger Co-Cp bond in CpCo (CO )z than in CoCp,. Comparison of the EHT calculations for CpCo(CO), and Co (Cp) 2 suggests that the increased C2p (Cp) character in the dn-pz and dtiTABLE 6 Properties of virtual orbitals of the ground and excited states of Co ( Cp jz in DW symmetry derived from extended Hilckel calculations MO

de,

Ground state WV)

m3a

X2p

- 6.42

0.075 0.075

0.104 0.198

-6.55

0.000 0.000

- 9.67

0.646 0.646

(R*-ti)

3esu (n*+R*)

4el, d XZ.YZ

Cls excited E(eV)

CoZp excited

Xo3d

IC2p

- 6.42 - 6.48

0.075 0.043

0.000 0.001

0.108 0.205

-6.55 - 7.86

0.000 0.071

0.000 0.304

0.018 0.099

- 9-69 - 10.01

0.640 0.673

0.009 0.147b

E(eV)

CCo3d

1~2~

-6.55 - 6.55

0.000 0.000

0.108 0.205

- 6.69 -6.69

0.021” 0.021

0.105 0.200

0.207 0.207

0.018 0.153

- 11.17 -11.17

“The underlined values are believed to dominate the experimental spectra. bNote that the 4els orbitaI is singly occupied. Thus when using these results to prepare the predicted Cls spectrum this value was divided by two to correct for Pauli-excluded transitions.

TABLE 7 Properties of virtual orbit.& of the ground and excited states” of CpCo (CO ) 2 with C, symmetry derived from extended Hiickel calculations Virtual MO

Ground state

C,

Character

E(eV)

Co3d

C(CPl2P

C(CO)2P

G2P

a’ a’

n*(C-C) A* (Cd)

-6.797 -6.799

0.050 0.037

0.199 0.389

0.004 0.005

0.007 0.001

a”

d,,,a*(M-CO)

-7.815

0.457

0.014

0.295

0.074

a’ a’

a*(M-CO), dx+,zta x*(C+),dW

- 8.563 -8.615

0.360 0.173

0.004 0.002

0.392 0.432

0.108 0.120

a” a”

dz,,x* (C-0)

-8.921

0.354 0.279

0.066 0.024

0.316 0.249

0.094 0.078

Virtual MO C*

Cls (Cp) excited stateb E(eV)

Co

C(CP)

C(CO)

0

a’ a’

- 6.47 - 7.97

0.046 0.000 0.106 0.121

0.004 Q.200

0.001 0.049

-6.80 -6.81

0.046 0.203 0.001 0.041 0.394 0.000

0.000 0.000

a”

- 7.62

0.415 0.142

0.269

0.059

-8.28

0.532 0.016 0.064

0.010

a’ a’

- 8.64 - 8.67

0.362 0.001 0.174 0.005

0.372 0.335

0.103 0.093

-8.78 -8.89

0.269 0.023 0.003 0.465 0.013 0.042

0.001 0.010

a” a”

-9.03 -9.46

0.297

0.288 0.278

0.089 -10.24 0.092 -10.60

0.478 0.075 0.409 0.260 0.004 0.587

0.140 0.226

Virtual MO C.

01s excited stats E(eV)

Co

C(CP)

C(CO)

0

a’ a’

-6.80 -6.80

0.050 0.200 0.038 0.389

0.003 0.003

0.000 0.000

-6.92 -6.93

0.010 0.382 0.011 0.009 0.203 0.001

0.002 0.000

a”

- 7.93

0.482 0.016

0.259

0.032

-8.49

0.034 0.001 0.521

3.138

a’ a’

- 8.68

-8.71

0.390 0.002 0.205 0.002

0.330 0.172

m 0.026

-9.08 -9.50

o.050 0.000 0.477 0.108 0.012 0.421

0.148 0.144

a” a”

-9.08 -9.30

0.362 0.069 0.253 0.023

0.578 0.403

o.090 -9.76 0.064 -10.62

0.115 0.054 0.341 0.296 0.034 0.063

0.126 0.023

-9.180

dyZ,dC(M-CO)

0.015 0.305 0.158

Cls(C0) E(eV)

excited state Co

C(CP)

C(CO)

0

Co2p excited state E(eV)

Co

C(CP)

C(CO)

0

“Excited states simulated by 2 + 1 species. In each case the square of the 2p, (for Clsor01s excitation) or Co3d (for Conp excitation) coefficients are listed. bEnergies and coefficients are weighted averages over values predicted with the Cls core hole on each carbon atom. The spread of energies and coefficients was quite small. ‘Underlined values should dominate the experimental spectra.

157

pti orbitals of CpCo(CO), is associated with charge transfer from the Cp ligand onto the Co atom and the CO ligands, as found in other theoretical work [ 151. Supporting evidence for the charge transfer is found in the Cls oscillator strength spectrum of CpCo (CO), since the Cls+lr* (CO) oscillator strength, decreases (relative to the per-CO value deduced from Coz ( CO)8, see Table 3)) as expected from increased occupancy of the x* orbital with increased backbonding, an effect that has been deduced previously from the decrease in the carbonyl stretching frequency from 2025 cm-’ in COAX to 1988 cm-’ in CpCo (CO), [ 401. This reduction in the a* oscillator strength is the explanation for the deviation at the rr*peak (285 eV) between the simulated and experimental Cls spectra of CpCo (CO ) 2 (Fig. 6). Overall this brief analysis suggests that detailed analysis of experimental core excitation spectra of mixed ligand organometallic compounds can give insight into ligand-ligand interactions. The EHT results (Tables 4,6 and 7) have been used to simulate the discrete portions of the Cls spectra of all three organocobalt species. Corresponding results for the Co2p spectra have been presented earlier [lo]. A sum of gaussian peaks of 1.2 eV fwhm located at the EHT eigenvalues with areas equal to the sum of the squares of the 2p A0 coefficients is plotted in Fig. 7b for comparison with the experimental spectra plotted on term value scales (Fig. 7a). The Cls(C0) and Cls(Cp) components of the Cls spectrum of CpCo(CO), were generated independently (with separate treatment of each of the three chemically distinct Cp carbons) and summed after a shift of 3.0 eV. This shift, which is somewhat larger than the 2.1 eV chemical shift in the Cls IPs [34], was chosen to optimize agreement with experiment. In general the IP is considered an adjustable parameter in matching semi-empirical calculations to experimental data. Given the crudeness of both the EHT calculations and the use of squared 2p coefficients to estimate spectral intensities, the trends in the simulation spectra, in terms of spectral shapes and relative intensities throughout the three molecules, are in remarkable agreement with those observed experimentally (compare Figs. 7a and ‘7b, which are on similar scales ) _ This gives some confidence in the validity of EHT for interpreting core spectra, although dif&ulties in its use for interpreting metal 2p spectra have been noted WI * Co2p and Co3p spectra The Co2p and Co3p core excitation spectra of Co2 (CO )*, CpCo (CO), and Co( Cp), are presented in Figs. 8 and 9 respectively. A smooth background extrapolated from the underlying continua at lower energy has been subtracted from each spectrum (indicated in Fig. 9 for the Co3p spectrum of Co, (CO )8) and the Co2p spectra have been converted to absolute oscillator strengths. No attempt was made to generate absolute Co3p optical oscillator strengths be-

158 t”“‘-‘-.‘.“‘““‘-“-‘--’ 86;

k

2h,

!o~~~ I

7 >0

4-

7

2

E

:

co2 (COba

k

z

cpco
0

jj

l....I....I....1.._.l....l

770

780

790

800

Energy

(PV)

e10

620I

Fig. 8. Co2p oscillator strength spectra derived from ISEELS spectra of Co2 (CO),+ CpCo (CO), and Co (Cp J2. The hatched lines indicate experimental IPs [ 34 1.

cause of uncertainties in the continuum shape associated with the difficult background subtraction. This is particularly challenging for the Co3p signal which is very weak and superimposed on a large valence-shell continuum (see top portion of Fig. 9). The energies, term values and proposed assignments of the observed features are presented in Table 8. The Co2p spectra of COAX and Co(Cp), are quite different from each other while the spectrum of CpCo (CO), has an intermediate shape. The metal 2p spectra of Mn [ 10 ] and Fe [ lo,41 ] carbonyl and metallocene complexes are remarkably similar to cobalt species with similar ligands. A simplified interpretation of the observed trends in terms of relative energies of dti and ddc virtual orbitals has been given recently [lo]. A more detailed interpretation of the cobalt 2p spectra, based on the EHT results, is given here. Excitations to MOs of large metal 3d character are expected to dominate Mnp core excitation spectra. The EHT results suggest that MOs constructed mainly from ligand rr*orbitals, which lie energetically above the 3d-dominated MOs, are also involved in the 2p excitations. Sine the Co2p IPs do not change very much among these three compounds ( K 0.5 eV [34] ), the large changes in the Co2p spectra must be associated with the Co3d energy levels as well as those of e-related MOs. The basic pattern is a shift from domination of the M-Cp spectrum by lower energy (Co2p- ‘,3d) states to domination by higher

159

1

J....l....l....l....1....1

i0

60

70

Energy

cpco (CO>2

co (Cp)2

60

Loss

90

10

(eV1

Fig. 9. ISEELS spectra of COAX, CpCo(CO), and Co(Cp), in the region of CoSp excitation. A large background has been subtracted from each spectrum. The raw data and subtracted background are indicated at the top for Co, (CO),.

energy (C02p-~,3d) states in M-CO. The unoccupied M3d atomic orbital character contributes mainly to a doubly degenerate e-type (be,,) orbital in Co (Cp), and probably to two closely spaced counterparts in the other two species which are of lower symmetry. Thus, focussing solely on a ligand-field type of picture of the Co3d orbitals does not appear adequate to explain the relatively large spread between the two main bands observed in each CoBp,,, (and CoZp,,,) spectrum. Instead, the two-band aspect of each spectrum may consist of a lower energy metal 2p-,metal3d band and a higher energy metal 2p+ligand 2p “charge transfer” band. Multiplet effects (exchange splittings) which are very important in the metal 2p spectra of open shell species [42], are unlikely to be involved in Co, ( CO)8 and CpCo (CO& since they have closedshell ground state and only singly degenerate virtual orbit&. The invocation of charge transfer contributions is suggested by the EHT analysis which predicts (on energetic grounds at least) that the lower energy features in the 2p spectra arise largely from M2p+M3d transitions, whereas the higher energy ones are almost completely due to M2p-+ n* (Cp/CO) charge-

160

161

transfer type transitions. In Co (CP)~, the lower energy peak is assigned to Coap +LUMO 4eIp (Cod,,, ) transitions,while the higher energy peak is attributed to promotions from Co2p to the 3e,, (and perhaps 4e2,) MOs which are the x* (C-C) MOs of the Cp ring (Table 6)+ Experimentally the two peaks are well resolved. One expects the virtual orbitals accessed in Co2p excitation to have strong metal 3d character and only small ligand 2p character. The weakness of the higher energy band in the Co2p spectrum of Co (Cp ), reflects the fact that the Co2p core hole does not have substantial overlap with MOs localized on the Cp ligands. The cobalt 2p spectrum of Co2 (CO), differs dramatically from that of cobaltocene. The intensities of the two peaks are reversed and the lower energy peak shrinks into a shoulder on the more intense higher energy peak, with an intensity ratio of approximately 1; 3. The two peaks are not fully separated in Co, (CO&, suggesting that there is a wide distribution of Co3d density associated with strong interactions between Co and CO of both dx-pn and do-pa character. The EHT calculations for Co2p excitation of Co, (CO ) 8 (Table 4) suggest that the low energy shoulder is associated with Co2p excitations to a group of frontier MOs (#29-33). The lowest energy b,, MO of this group (the LUMO ) is basically CO2p, but it has a remarkably enhanced CoSd,, character on Co2p excitation - the Co2p coefficient rises from 0.008 in the ground state to 0.108 in the Co2p excited state. The high energy Co2p band may be related to another group of virtual MOs (#20-23). This includes a b,, MO (#20) which is almost a pure K*(2p,,2pY) MO of the CO ligand in the ground state but becomes a Co3d,p_,z-dominated MO upon Co2p excitation. The ratio of the squared Co3d coefficients of these two bl, MOs (0.108:0.319) is close to l/3,in good agreement with the experimental intensity ratio. Compared to Co (Cp )z there is a greater change in the Co3d coefficients between the ground and Co core-excited states of COAX This is consistent with stronger interaction between Co and CO in Co2 (CO) 8than between Co and Cp in Co (Cp ) 2. The energies of the two b,, MOs with large Co3d character in Co2 (CO), are much closer to each other than their counterparts in cobaltocene (Table 6). In addition, there are several CO-localized MOs (#24-2’7) filling the gap ( w 2.1 eV ) between the two b,, levels (Table 4). Both these factors help explain why the two major peaks are less well resolved in the Co2p spectrum of Co2 (CO), than in that of Co (Cp),. The EHT results for CpCo (CO), (Table 7) also help to explain its Co2p spectrum. In this mixed ligand species the two Co2p features are even less well resolved, perhaps reflecting more complicated bonding between Co and the two types of ligand. The CpCo (CO), and Co, (CO), spectra look similar - the energies of the two major features are almost identical (Table 8). This suggests that the two carbonyls play more important roles than the Cp ring in determining the Co3d content of the frontier unoccupied MOs of CpCo (CO) 2, consistent with stronger M-CO than M-Cp interactions. Co2p excitations to the

162

dJd,, components of the LUMO and next-to-LUMO, which are M-Cp antibonding in character, enhance the first peak compared with its counterpart in Co&CO)& Based on these EHT results, a logical interpretation of the Co2p spectrum of CpCo (CO), would be to assign the first feature to Co2p excitations into the three lowest energy virtual MOs, which are two a” MOs followed by an a’ MO (Table 7). All three contain substantial Co3d components, especially the a” LUMO. Thus the first feature in the CoZp spectrum is chiefly a Co2p+3d excitation. The second peak can be attributed to Co2p excitations to the four remaining upper unoccupied MOs which have smaller Co3d character. In this second group, two MOs are mainly associated with the carbonyl ligands, and the rest are largely localized on the Cp ring. Co core excitations to the two highest energy, Cp-based virtual MOs may contribute on the high energy side of the second spectral feature. The Co3p spectra (Fig. 9) are much less detailed than the Co2p spectra partly because of the small spin-orbit splitting which leads to complicated spectral overlap, and partly because of the large natural linewidth associated with the rapid Coster-Kronig decay of metal 3p states. Even so, the trends observed in the 2p spectra are also discernible in the 3p spectra. Cog ( CO)8 shows a low energy shoulder on the main part of the CoSp-, 3d resonance while the Co3p spectrum of Co (Cp )z is strongest at lowest energy. It is possible that the second peak in the Co (Cp), spectrum is stronger than the first feature only because of overlap with a strong, lower energy C03p~,~--*Co3d component. There does not appear to be any evidence for a Conp+dt (Co-Co) transition in either the 2p or 3p spectra of Co, (CO)* Indeed, the shift from a high energy to a low energy dominated spectrum between Co,(CO), and Co( Cp), is opposite to the predictions of the weak bond model [ 33 1. Comparison oft he Co2p spectrum of Co,(CO), with the Fe2p spectrum of Fe(CO), [ 10,411 does not reveal any evidence for an additional d” (Co-Co) feature in the former. SUMMARY

The Cls, Ols, Co2p and Co3p core excitation spectra of Coz (CO)a, CO(Cp)z and CpCo (CO), have been analysed with the aid of extended Htickel MO calculations and spectral comparisons. Simple spectral additivity gives a reasonable first order description but it is inadequate for quantitatively reproducing the Cls spectrum of CpCo (CO),. The differences between the additivity simulation and experiment have been discussed in terms of ligand-ligand interactions. The core excited states associated with carbonyls are found to be much more independent of their local environment than those associated with the z-bonded Cp ligand. The Conp spectra are found to be extremely sensitive to the nature of the ligand.

163 ACKNOWLEDGEMENTS

This research was supported financially by the Natural Sciences and Engineering Research Council of Canada. A.T.W. acknowledges support of McMaster graduate research fellowships and teaching assistantships.

REFERENCES 1

2 3

4

5 6 7 8 9 10 11 12 13 14

15 16 17 18 19

P.A. Dowben, DC. Driscoll, R.S. Tate and N.M. Boag, Organometallics, 7 (1988) 305. J.D. Morrison (Ed.), Asymmetric Synthesis, 2 Vols., Academic Press, New York, 1984. J-K. Stille, Mod. Synth. Methods, 3 (1983) 1. J.T. Mague, J. Organomet. Chem., 242 ( 1983) 241. S.G. Davies, Organotransition Metal Chemistry Application to Organic Synthesis, Pergamon Press, Oxford, 1982. C. Masters, Homogeneous Transition-Metal Catalysis, Chapman and Hall, New York, 1981. R. Mason and M-W. Roberts, Inorg. Chim. Acta, 50 {1981) 53. C.K. Rofer-DePoorter, Chem. Rev., 81 (1981) 447. D. Forster, A. Hem&man and D.E. Morris, Catal. Rev., 23 (1981) 89. J. Stiihr, and D.A. Outka, J. Vat. Sci. Technol. A, 5 (1987) 919. A.P. Hitchcock, J. Electron Spectrosc. Relat. Phenom., 25 (1982) 245 (updates of this core excitation bibliography are available from the author). T. Tyliszczak, F. Esposto and A.P. Hitchcock, Phys. Rev. Lett., 62 (1989) 2551. E. Riihl and A.P. Hitchcock, J. Am. Chem. Sot., 111 (1989) 2614. A.P. Hitchcock, A.T. Wen and E. Rtihl, J. Electron Spectrosc. Relat. Phenom., 51 (1990) 653. D. Sondericker, Z. Fu, J. Bradley and W. Eberharch, J. Chem. Phys., 92 (1990) 2203. G. Cooper, K.H. Sze and C.E. &ion, J. Am. Chem. Sot., 111 (1989) 5051. G. Cooper, K.H. Sze and C.E. Brion, J. Am. Chem. Sot., 112 (1990) 4121. E. Rtil and A.P. Hitchcock, J. Am. Chem. Sot., 11 (1989) 5069. D. Outka and J. Stijhr, J. Chem. Phys., 88 (1989) 3539. A.P. Hitchcock, A.T. Wen and E. Riihl, Chem. Phys., 147 (1990) 51. W.H.E. Schwarz, Chem. Phys., 11 (1975) 217. W. Heijser, E.J. Baerends and P. Ros, Faraday Symp. Chem. Sot., 14 (1980) 211. R. Benn, K. Cibura, P. Hofmann, K. Jonas and A. Rufinska, Organometahics, 4 (1985) 2214. P. Hofmann, Angew. Chem., Int. Ed. Engl., 16 (1977) 536. - A. Modelli, G. Distefano, M. Guerra and D. Jones, J. Am. Chem. Sot., 109 (1987) 4440. M. Guerra, D. Jones, G. Distefano, A. Foffani and A. Modelli, J. Am. Chem. Sot., 110 (1988) 375. A. Modelli, A. Foffani, F. Scagnolari, S. Torroni, M. Guerra and D. Jones, J. Am. Chem. Sot., 111 (1989) 6040. J-C. Giordan, J.H. Moore, J.A. Tossell and J. Weber, J. Am. Chem. Sot., 105 (1983) 3431. D.A. Brown, H.L. Clarke and N.J. Fitzpatrick, J. Organomet. Chem., 47 (1973) Cll. L.B. Byers and L.F. Dahl, Inorg. Chem., 19 (1980) 277. A.P. Hitchcock, S. Beaulieu, T. Steel, J. Stahr and F. Sette, J. Chem. Phys., 80 (1984) 3927. A.P. Hitchcock, Phys. Ser., T31 (1990) 159. R. McLaren, S.A.C. CIark, I. Ishii and A.P. Hitchcock, Phys. Rev. A, 36 (1987 ) 1683. I. Ishii and A.P. Hitchcock, J. Chem. Phys., 87 (1987) 830.

164 20 21 22 23 24 25 26

27

28 29 30 31 32 33 34 35 36 37 38 39 40 41 42

F.M.F. de Groot, M. Grioni, J.C. Fuggle, J. Ghijsen, G.A. Sawatzky and H. Petersen, Phys. Rev. B, 40 (1989) 5715. T.A. Albright, P. Hofmann and R. Hoffmann, J. Am. Chem. Sot., 99 I.1977) 7546. A. Haaland, Act. Chem. Res., 12 (1979) 415. B. Beagley, C.T. Parrott, V. Ulbrecht and G.G. Young, J. Mol. Struct., 52 (1979) 47. V. Bellagamba, R. Ercoli, A. Gamba and G.B. Suffritti, J. Organomet. Chem., 190 (1980) 381. I. Kojima, A.K. Srivasta, E. Miyazaki and H. Adachi, J. Chem. Phys., 84 (1986) 4445. E.W. Plummer, W.K. Ford, W. Eberhardt, R.P. Messmer and H.J. Freund, Surf. Sci., 158 (1985) 58. W. Eberhardt, E.W. Plummer, C-T. Chen and W.K. Ford, Aust. J. Chem., 39 (1986) 853. W. Eberhardt, C.T. Chen, W.K. Ford, E.W. Plummer and H.R. Moser, Springer Ser. Surf. Sci., 4 (1985) 50. R. Murphy, E.W. Plummer, CT. Chen, W. Eberhardt and R. Carr, Phys. Rev. B, 39 (1989 ) 7517. B. Gumhalter, K. Wandelt and Ph. Avouris, Phys. Rev. B, 37 (1988) 8048. W. Wurth, D. Coulman, A, Puschmann, D. Menzel and E. Umbach, Phys. Rev. B, 41 (1990) 12933. E.J. Baerends and P. Ros, J. Electron Spectrosc. Relat. Phenom., 7 (1975) 69. J.C. Giordan, J.H. Morre and J.A. Tossell, J. Am. Chem. Sot., 103 (1981) 6632. I.H. Hillier and V.K. Saunders, Mol. Phys., 22 (1971) 1025. P.S. Bagus, C.J. Nehn and C.W. Bauschlicher, J. Vat. Sci. Technol. A, 2 (1985) 905. Ph. Avouris, P.S. Bagus and A.R. Rossi, J. Vat. Sci. Technol. B, 3 (1985) 1484. F. Sette, J. Stiihr and A.P. Hitchcock, J. Chem. Phys., 81 (1984) 4906. R. McLaren, I. Ishii, A.P. Hitchcock and M.B. Robin, J. Chem. Phys., 87 (198’7) 4344. W.L. Jolly, K.D. Bomben and C.J. Eyermann, At. Data Nucl. Data Tables, 31 (1984) 433. F.A. Cotton, Chemical Applications of Group Theory, Wiley, New York, 1971. A.P. Hitchcock, DC. Newbury, I. Ishii, J. Stijhr, J-A. Horsley, R.D. Redwing, A.L. Johnson and F. Sette, J. Chem. Phys., 85 (1986) 4849. J.A. Horsley, J. Stshr, A.P. Hitchcock, D.C. Newbury, A.L. Johnson and F. Sette, J. Chem. Phys., 83 (1985) 6099, W.H.E. Schwarz, T.C. Chang, U. Seeger and K.H. Huang, Chem. Phys., 117 (1987) 73. L. J. Medhurst, T.A. Ferrett, P.A. Heiman, D.W. Lindle, S.H. Liu and D.A. Shirley, J. Chem. Phys., 89 (1988) 6096. H.W. Chen, W.L. Jolly, J. Kopf and T.H. Lee, J. Am. Chem. Sot., 101 (1979) 2607. A.T. Wen, E. R&l and A.P. Hitchcock, unpublished work. G. van der Laan, G.A. Sawatzky, R. Karnatak and J.M. Esteva, Phys. Rev. B, 33 (1986) 4253.