Electronic Structure of the d¹ Bent-metallocene Cp₂VCl₂: A Photoelectron and Density Functional Study

Abstract

The Cp₂VCl₂ molecule is a prototype for bent metallocene complexes with a single electron in the metal d shell, but experimental measure of the binding energy of the d electron by photoelectron spectroscopy eluded early attempts due to apparent decomposition in the spectrometer to Cp₂VCl. With improved instrumentation, the amount of decomposition is reduced and subtraction of ionization intensity due to Cp₂VCl from the Cp₂VCl₂/Cp₂VCl mixed spectrum yields the Cp₂VCl₂ spectrum exclusively. The measured ionization energies provide well-defined benchmarks for electronic structure calculations. Density functional calculations support the spectral interpretations and agree well with the ionization energy of the d¹ electron and the energies of the higher positive ion states of Cp₂VCl₂. The calculations also account well for the trends to the other Group V bent metallocene dichlorides Cp₂NbCl₂ and Cp₂TaCl₂. The first ionization energy of Cp₂VCl₂ is considerably greater than the first ionization energies of the second- and third-row transition metal analogues.

1. Introduction

The chemistry of bent metallocene complexes has been a major factor in the development of organometallic chemistry over the last 50 years [1–5], finding applications in diverse fields ranging from polymer catalysis [4,6,7] to cancer research [8–11]. The electron configuration and subsequent reactivity of the metal center can be fine-tuned with the choice of metal, cyclopentadienyl substituents, and additional coordinating ligands. Elucidating the electronic structure of bent metallocenes is important in understanding the chemical behavior and physical properties of these complexes. Photoelectron spectroscopy provides an experimental measure of the orbital energetics and interactions and at this point there is a general understanding of the photoelectron spectra of simple metallocenes [7,12,13]. One remaining point of contention has been the reported photoelectron spectrum of Cp₂VCl₂, which has been an outlier from all other reported metallocene photoelectron results. This has been unfortunate because Cp₂VCl₂ is a prototype for d¹ bent metallocene complexes, having a first-row transition metal, relatively simple ligands, and sufficient symmetry to be a foundational model for other studies.

The electronic structure of Cp₂VCl₂ has previously been investigated by electron paramagnetic resonance (EPR) [14–17] and photoelectron spectroscopy (PES) [15]. EPR experiments concluded that the unpaired metal electron resides in an orbital composed primarily of d_z2 character in the VCl₂ plane as depicted in the coordinates below, with some admixture of d_{x2 − y2} character. Photoelectron studies of Cp₂VCl₂ and (C₅H₄Me)₂VCl₂ [15,18] revealed spectra that were similar in structure to those of their Ti analogues, but with two low energy ionization features. Only one low energy ionization was expected for the single d electron in the vanadium complexes. The possibility of decomposition was investigated and considered less likely for the (C₅H₄Me)₂VCl₂ molecule. The lowest energy feature of the spectrum obtained from the (C₅H₄Me)₂VCl₂ sample at 6.60 eV was tentatively assigned to the metal d¹ ionization. The second feature at 7.20 eV was hypothesized to be a triplet component of the unpaired d¹ electron exchange splitting with the Cl pπ orbitals. The singlet component of this ionization could then contribute to broadening of the next band, rather than appear as a separate band. This assignment was called into doubt when it was found that the related Nb and Ta complexes had only one extra low energy ionization consistent with their d¹ configurations [12,18]. Subsequent photoelectron studies of Cp₂VCl [19], which has a high-spin d² configuration, showed bands at 6.80 and 7.40 eV. The spectrum of this monochloride complex appeared similar to the reported spectrum of the dichloride complex, suggesting that Cp₂VCl₂ may be decomposing to Cp₂VCl during the photoelectron experiment. Cp₂VCl₂ is known to undergo thermal- [15] and photoelectrochemical decomposition [20] to Cp₂VCl in other conditions.

Recently, we investigated the electronic structure of bent vanadocene dithiolates as minimum molecular models of the d¹ electron configuration of sulfite oxidizing enzymes using photoelectron spectroscopy and density functional theory [21]. This study prompted us to reinvestigate the electronic structure of the simplest d¹ bent metallocene, Cp₂VCl₂, in more detail to aid in the interpretation of the vanadocene dithiolate spectra. Improved gas-phase photoelectron spectroscopy techniques, comparison to the photoelectron spectra of Cp₂VCl and other Group V metallocene dichlorides, along with DFT calculations, have allowed us to isolate and reassign the photoelectron spectrum of Cp₂VCl₂ for determination of the ionization energy of the d¹ electron.

2. Results and Discussion

2.1. Photoelectron Spectroscopy

The low energy valence regions of the gas-phase photoelectron spectra collected from samples of Cp₂VCl and Cp₂VCl₂ are presented in Figure 1. The spectra obtained using typical data collection techniques are in agreement with spectra previously reported in the literature [15,19]. There is a large similarity between the two spectra obtained from the Cp₂VCl and Cp₂VCl₂ samples, Figures 1A and 1B respectively, which would not be expected given the anticipated relative ionizations from the high-spin d² metal center of Cp₂VCl versus the d¹ metal center of Cp₂VCl₂. When a spectrum of the Cp₂VCl₂ sample is recorded using sample entry procedures designed to minimize decomposition (see Experimental section), the overall shape of the spectrum changes dramatically (Figure 1C). Most importantly, the relative intensity of the ionization at 6.8 eV decreases substantially relative to the ionization at 7.4 eV. In addition, ionization intensity in the region labeled L3 in the spectrum of Cp₂VCl decreases while the ionization intensity in the region of L2 increases. The similarities in the spectra 1A and 1B and the differences between these and spectrum 1C are strong evidence that Cp₂VCl₂ is decomposing to Cp₂VCl in the spectrometer. The decomposition has been minimized, but not completely eliminated, as evidenced by the small amount of ionization intensity remaining at 6.8 eV in Figure 1C. The remaining ionization intensity in spectrum 1C from Cp₂VCl cannot be due to residual Cp₂VCl in the Cp₂VCl₂ sample because the sublimation temperature of Cp₂VCl is approximately 100°C lower than that of Cp₂VCl₂ (see Experimental) and would be observed first as the temperature of the cell rises. Also, this contamination cannot be due to photoelectrochemical dehalogenation [20] of the solid sample since the sample is in the direct path of the ionization source during the experiment and Cp₂VCl is not seen at lower temperatures, nor were ionizations found corresponding to the photochemical product Cl₂. Therefore, it is hypothesized that use of the quartz crucible and the short path to the photon beam reduces surface interactions of Cp₂VCl₂ with the aluminum sample cell, which may lead to reduction and dehalogenation of Cp₂VCl₂. It was noted that, during the experiment, a sharp doublet-ionization occurs at 12.75 and 12.82 eV, coincident with the first ionizations of HCl. The ionization intensity due to Cp₂VCl in Figure 1C is removed by subtracting the Figure 1A spectrum, appropriately scaled to the ionization of Cp₂VCl at 6.8 eV in Figure 1C, to yield the spectrum shown in Figure 1D. The ionizations shown in Figure 1D are taken to be those of Cp₂VCl₂. The subtraction of the Cp₂VCl spectrum leads to a shape (width and symmetry) for the first ionization of Cp₂VCl₂ at 7.40 eV that is indicative of a single ionization [22].

Figure 1.

Low energy He I spectra showing the extraction of the Cp₂VCl₂ spectrum from the data: (A) spectrum of Cp₂VCl, (B) spectrum of Cp₂VCl₂ sample loaded directly into ionization chamber, (C) spectrum of Cp₂VCl₂ sample effused directly into photon beam from a quartz crucible, (D) subtraction of A from C to give difference Cp₂VCl₂ spectrum.

Figure 2 shows that the resultant Cp₂VCl₂ spectrum bears strong similarities to the He I photoelectron spectra of Cp₂NbCl₂ and Cp₂TaCl₂ [12]. The ionization region of ~8–9.8 eV is particularly diagnostic, because this region contains ionizations that are symmetry combinations of the pπ orbitals of the two Cl atoms with the metal. The similar ionization pattern in this region indicates similar MCl₂ structure for the three molecules. The main difference between the spectra of Cp₂MCl₂ (M = V, Nb, Ta) (Figure 2) is the energy of the metal d¹ ionization, which can be ascribed to reduction of the effective nuclear charge on going from 3d to 4d to 5d, reflecting the shielding of the unpaired electron by the different atoms.

Figure 2.

He I photoelectron spectra of Cp₂VCl₂, Cp₂NbCl₂ and Cp₂TaCl₂.

Table 1 shows the analytical representation of the ionizations for Cp₂VCl₂ and Cp₂VCl. This table includes the change in band area with change in ionization source from He I to He II for Cp₂VCl. Green et al. have previously assigned the photoelectron spectrum of Cp₂VCl based on a He I/He II comparison and their assignments are reflected in Table 2 [19]. The predominant Cl pπ character mentioned above in ionizations labeled L1 and L2 is evidenced by their decrease in ionization intensity relative to the metal-based ionizations from He I to He II excitation.

Table 1.

Analytical representation of the ionization features of Cp₂VCl₂ and Cp₂VCl obtained with He I and He II photon sources.

Band	I.E.a (eV)	Band Widthb		Relative Areac
		High (eV)	Low (eV)	He I	He II
Cp2VCl2
M1	7.40	0.30	0.30	1	—
L1	8.47	0.47	0.38	5.42	—
L2	8.85	0.55	0.26	8.80	—
Cp2VCl
M1	6.78	0.33	0.23	1	1
M2	7.40	0.22	0.12	0.78	0.97
L1	8.33	0.39	0.39	3.67	1.41
L2	9.04	0.36	0.36	2.32	1.41
L3	9.45	0.54	0.49	5.92	4.22

^a: Vertical ionization energy defined as the position of the asymmetric Gaussian peak modeling the band,

^b: Widths of the asymmetric Gaussian peak modeling the high and low ionization energy sides of the band,

^c: Band areas are relative to an area of 1 for M1.

Table 2.

Experimental and calculated ionization energies (eV) for Cp₂VCl₂ and Cp₂VCl.

Band	I.E.a	Assignment	Kohn-Sham Orbital Energyb		ΔSCFc
Cp2VCl2
			α-spin	β-spin	Singlet	Triplet
M1	7.40	V dz2	−4.83 (7.40)		7.27
L1	8.47	Cl pπ	−5.81 (8.38)	−5.78 (8.35)	8.20	8.15
L2	8.85	Cl pπ	−5.96 (8.53)	−5.88 (8.45)	8.49	8.44
Cp2VCl
			α-spin	β-spin	Doublet	Quartet
M1	6.78	V dxz	−4.24 (6.78)		6.85
M2	7.40	V dz2	−4.87 (7.41)		7.56
L1	8.33	Cl pπ	−5.64 (8.18)	−5.59 (8.13)	8.33	8.18
L2	9.04	Cl pπ	−6.65 (9.19)	−6.34 (8.88)	9.27	8.91
L3	9.45	Cp pπ +Cl pπ	−7.17 (9.71)	−6.93 (9.47)	9.50	9.24

^a: Experimental vertical ionization energy.

^b: The values in parentheses are the orbital energies shifted by an amount such that the first orbital energy matches the first ionization energy.

^c: The difference in total energy between the molecular ground state and the molecule with an electron removed from the specified orbital.

2.2. Computational results

A central question concerns the ability of electronic structure calculations to account for the geometric structures and the electron energies and distributions of open-shell bent metallocene complexes of the early transition metals. The calculated structures of Cp₂VCl₂ and Cp₂VCl are compared with the reported structures determined from X-ray crystallography in Table 3 [23,24]. The structures agree reasonably well with the largest deviation being the V-Cl bond distance in Cp₂VCl₂ which is underestimated by only about 0.04 Å. The ground states of these molecules are calculated as an unrestricted doublet and triplet, respectively for Cp₂VCl₂ and Cp₂VCl. Spin contamination is minimal, giving a value of 0.78 compared to the ideal value for s(s+1) of 0.75 for Cp₂VCl₂ and 2.05 compared to the ideal value of 2.00 for Cp₂VCl.

Table 3.

Comparison of the calculated geometries with the crystallographic molecular structures for Cp₂VCl₂ and Cp₂VCl.^a

	Calculated	Experimental
Cp2VCl2b
CpC-C	1.403–1.428 (1.415)c	1.371–1.465 (1.418
		1.394–1.457 (1.424)
M-Cpcentroid	1.952, 1.948	1.983, 1.967
		1.971, 1.986
M-C	2.231–2.306 (2.269)	2.281–2.368 (2.314)
		2.284–2.342 (2.315)
M-Cl	2.371	2.409, 2.418
		2.410, 2.411
Cl-M-Cl	88.74°	86.6°
		87.1°
Cp2VCld
CpC-C	1.406–1.420 (1.413)	1.379–1.408 (1.394)
M-Cpcentroid	1.901, 1.901	1.946, 1.944
M-C	2.221–2.277 (2.250)	2.261–2.296 (2.278)
M-Cl	2.367	2.390

^a: All distances are given in Å,

^b: crystallographic data taken from Tzavellas et al.[23],

^c: average distance,

^d: crystallographic data taken from Fieselmann et al.[24].

DFT calculations were also used to model the electron density in the highest occupied molecular orbital (HOMO) for comparison with EPR studies. The trend of the Cl-M-Cl angle to decrease from Ti to V to Mo for the d⁰ to d¹ to d² configurations and the single crystal EPR studies on the d¹ complexes led to the conclusion that the HOMO of the V and Mo compounds is primarily along an axis normal to the plane bisecting the Cl-M-Cl angle. Lowering the symmetry of Cp₂VCl₂ from C_2v to C_s allows this axis to be labeled z, as illustrated in the Introduction. If the ligand contributions are neglected then |Ψ_HOMO> = a|d_z²> + b|d_{x²− y²}>. Previous Fenske-Hall [25] molecular orbital percent characters in the singly-occupied orbital were calculated to be 20.5/1 for d_z2/d_{x2 − y2} for Cp₂VCl₂, which agrees well with the ratio obtained from the EPR data for (C₆H₅Me)₂VCl₂ of 20.0/1 [15]. Using current DFT methods, the singly-occupied orbital of Cp₂VCl₂ is calculated to be 60.8% d_z² and 3.44% d_{x²− y²} giving a ratio of 17.7:1. This ratio is similar to the orbital percent characters calculated using the Fenske-Hall method and the observed orbital parameters calculated from the EPR experiment.

The energies of the ionizations observed for Cp₂VCl₂ and Cp₂VCl by photoelectron spectroscopy are compared with those calculated by the ΔSCF method and with the Kohn-Sham orbital energies (Table 2). The calculated Kohn-Sham orbital energies from the DFT calculations are related to the photoelectron ionization energies by applying a corollary to Koopmans’ theorem from Hartree-Fock calculations [26]. Koopmans’ theorem has intrinsic uncertainty including electron relaxation, electron correlation, relativistic effects, vibronic coupling, and differences in spin states [27–34], which can lead to poor correlation between the calculated orbital energies and the ionization potentials. A central principle of the Kohn-Sham orbital model is that the negative of the Kohn-Sham orbital energy for the HOMO, calculated with the correct functional and basis, will exactly match the first ionization energy of the molecule, but in current practice these energies differ substantially. Nonetheless, the relative energies of the Kohn-Sham orbitals can be useful in interpreting the pattern of ionizations. ΔSCF calculations were performed to account for the deficiencies in the frozen orbital approximation and to get information on the coupling of electron spins.

The ground state of neutral Cp₂VCl₂ is s = ½ (V⁴⁺, d¹, ²A′) while the ground state of Cp₂VCl is s = 1 (high-spin V³⁺, d², ³A″). Figure 3 shows the contour plots for the spin α and β orbitals of Cp₂VCl₂ that correspond to the ionizations evaluated in this study by PES. For Cp₂VCl₂, ionization from the singly-occupied HOMO will lead to a singlet state, while ionization from the doubly-occupied ligand-based orbitals leads to both singlet and triplet state configurations with coupling to the d¹ electron. For Cp₂VCl, ionization from the singly-occupied HOMO and HOMO-1 lead to doublet states, and ionization from the doubly-occupied ligand-based orbitals will lead to either doublet or quartet state configurations. Table 2 shows the calculated values for the singlet/triplet states of Cp₂VCl₂ and doublet/quartet states of Cp₂VCl using the ΔSCF method. For Cp₂VCl₂, the calculated singlet/triplet state separation is only 0.05 eV for both ionizations L1 and L2, and therefore ionizations to these individual states are not likely to be resolved in the photoelectron spectrum. For Cp₂VCl the calculated doublet/quartet state separation for band L1 is 0.15 eV, still less than the 0.4 eV width of the unresolved vibrational broadening of the band. For bands L2 and L3, the calculated doublet/quartet state separations are 0.36 and 0.26 eV, respectively; however, these bands are now in a region of overlapping ionizations that preclude observation of individual states.

Figure 3.

Spin correlation diagram for Cp₂VCl₂ showing α- and β-spin molecular orbital character and electron occupations.

ΔSCF calculations for Cp₂VCl₂ predict the singly-occupied HOMO ionization to be at 7.27 eV, consistent with the first ionization of Cp₂VCl₂ at 7.40 eV (Table 2). The calculated ΔSCF ionizations for L1 and L2 are also similar with the HOMO-1 and HOMO-2 ionizations, and follow the ionization trend. The shifted Kohn-Sham orbital energies for L1 and L2 are also similar to the observed ionization energies. ΔSCF calculations were also performed on Cp₂VCl to confirm that removing an electron from the triplet ground state d orbital configuration will give a metal ionization that coincides with removing an electron from the HOMO of Cp₂VCl₂. Table 2 shows that removing an electron from the HOMO-1 of Cp₂VCl gives a ΔSCF calculated ionization energy of 7.56 and a shifted Kohn-Sham orbital energy of 7.41, coincident with the HOMO ionization of Cp₂VCl₂. Both the ΔSCF method and shifted Kohn-Sham orbital energies follow similar trends to the observed ionization energies for Cp₂VCl. These calculations support that subtraction of the Cp₂VCl spectral contribution from the mixed Cp₂VCl₂/Cp₂VCl spectrum is effective for revealing the Cp₂VCl₂ spectrum. The first ionization of Cp₂NbCl₂ and Cp₂TaCl₂ was also calculated by the ΔSCF method to be 6.65 and 6.37 eV, almost exactly coincident to the observed ionization energies of 6.69 and 6.39 eV [12].

3. Conclusions

We have revisited the gas-phase photoelectron spectrum of Cp₂VCl₂ and used improved collection techniques to reduce the Cp₂VCl decomposition product from the Cp₂VCl₂ He I spectrum. Subtraction of ionization intensity due to the Cp₂VCl decomposition product leads to a Cp₂VCl₂ spectrum that is similar in appearance to other Group V bent metallocene dichlorides. Density functional theory calculations support the assignments made for the Cp₂VCl₂ spectrum and comparison to the calculated ionization energies for Cp₂VCl also confirm that the d¹ ionization of Cp₂VCl₂ coincides with one of the metal ionizations of Cp₂VCl, further supporting the superposition of the Cp₂VCl and Cp₂VCl₂ spectra. The open-shell calculations on these Group V bent metallocenes show good reliability in calculating the geometric structures and electron distributions and give good account of the ionization energies using either the shifted Kohn-Sham orbital energies or the ΔSCF method.

4.0 Experimental methods

4.1 Photoelectron Spectroscopy

Samples of Cp₂VCl (97%) and Cp₂VCl₂ (95%) were purchased through Aldrich and used as received. Photoelectron spectra were recorded using an in-house instrument built around a 36-cm hemispherical analyzer (McPherson), custom-designed sample cells and detection and control electronics. The electron detection and instrument operation are interfaced to a National Instruments PCIe-6259 multi-function data acquisition card and custom software. Argon was used as an internal calibrant during data collection and the instrument resolution (measured using FWHM of the argon ²P_3/2 peak) was 0.020–0.030 eV. The sublimation temperatures (at 10⁻⁵ Torr, monitored using a “K” type thermocouple passed through a vacuum feed-through and attached directly to the sample cell) were 195–205° for Cp₂VCl₂ and 90–120° for Cp₂VCl. For Cp₂VCl₂, a crushed crystalline sample was placed in a quarter-inch diameter quartz crucible that was inserted into the ionization chamber of the sample cell directly below the photon beam [35,36].

In the photoelectron spectra, the vertical length of each data mark represents the experimental variance of that point. The valence ionization bands are represented analytically with the best fit of asymmetric Gaussian peaks [22]. The number of peaks used in a fit was based solely on the features of a given band profile. The peak positions are reproducible to about ±0.02 eV (≈3σ). The parameters describing an individual Gaussian peak are less certain when two or more peaks are close in energy and overlap.

4.2. Theoretical Methods

The Amsterdam density functional theory suite (ADF 2006.01, using the standard parameters except for the options given in parentheses) was used to study the electronic structure of Cp₂MCl₂ (M = V, Nb, and Ta) and Cp₂VCl. The optimized geometries of Cp₂MCl₂ and Cp₂VCl were constructed in C_s symmetry using the crystal structures as a starting point. In C_s symmetry, the Cp rings were staggered and the mirror plane is coincident with the xy plane, which bisects and each Cp ring and the Cl-V-Cl angle for Cp₂VCl₂ or lies along the V-Cl bond for Cp₂VCl. Calculations on the ground state and excited state ions were conducted in the spin unrestricted mode (Unrestricted) since Cp₂VCl and Cp₂MCl₂ contain two and one unpaired electrons, respectively. A generalized gradient approximation, with the correlation of Perdew, et al. [37] and exchange correction of Handy and Cohen [38] (GGA OPBE), was used. This functional has been found by us [21,39–41] and others [42–45] to give reasonable geometry parameters, spin states, and ionization and oxidation potentials. The calculations employed Slater-type orbitals for basis sets with double-zeta valence plus polarization functions for main group elements and triple-zeta valence plus polarization functions for the metal (DZP for C, H, and Cl, and TZP for V, Nb and Ta). Geometry optimizations were also carried out using integration to six significant figures (integraton = 6.0), using the zeroth-order relativistic approximation (ZORA), using tighter convergence criteria for the energy, bond distances and angles (E= 0.0005, grad=0.001, rad=0.0005), using the smoothing of gradients (Aggressive) option, and the convergence of the self-consistent field was tightened (Converge 1e-6 1e-6).

ΔSCF calculations of the ionized states were performed at the fixed geometry of the neutral molecule, with one electron removed from the relevant orbital. For the ligand-based ionizations both singlet and triplet ionized states for Cp₂VCl₂ and doublet and quartet ionized states for Cp₂VCl were calculated. The ΔSCF estimate of the ionization energy is the difference between the calculated total energy of the ionized state and that of neutral ground state molecule. As is customary for larger polyatomic molecules with low frequency vibrational modes, a semi-classical treatment of the ionization intensity is assumed that neglects zero-point vibrational energies (ZPEs). This semi-classical interpretation is reasonable since the low-frequency vibrational modes give an essentially continuum, vibrationally unresolved band in the photoelectron spectrum.