The polarized single-crystal ligand field (LF) absorption spectrum for [PPh4][FeCl4]1,2is given in Figure 1a; complex is near tetrahedral but has rigorousD2dsymmetry. These data, in conjunction with single-crystal MCD and transverse Zeeman polarized spectroscopies allowed assignment of the spin-forbidden 4 6A1 transitions in D2dsymmetry. From these data, the one-electron 3d orbital splitting pattern is derived2(Figure 2a) based upon Tanabe-Sugano matrices (to obtain 10Dq,B, andC) and splitting of the low-energy4T1and4T2 states (,); quantitative results are given inTable 1. The cubic LF splitting is large (10Dqẳ6,550 cm1) relative to the axial splittings ( and ). The LF analysis also reveals highly reduced Racah parameters, which reflect a very small electron repulsion in [FeCl4]1(ẳcomplex/Bfree ion0.40); this suggests high covalency in the Fe 3d-orbitals.2Low-temperatureQ-band EPR spectroscopy provide ground state information, 692 Spectroscopy and Electronic Structure of [FeX4]n(XẳCl, SR)
i.e., the molecular g-values and zero-field splitting (ZFS) for the axially distorted [FeCl4]1 complex (seeTable 1).2
The largest contribution to ZFS of the 6A1 ground state is second-order spin-orbit coupling to low-symmetry split excited states. Griffith3developed a model (Equation (1a)) that correlates the energy splitting of the lowest energy4T1excited state components in Figure 1awith D, the axial ZFS of the 6A1 in high-spin d5 systems. In Equation (1a), FeIII is the atomic spin-orbit coupling for FeIII and Ez and Ex,y are the energies of the 4T1(z) and 4T1(x,y) excited states, respectively (fromFigure 1a):
x 0.1
x,y z 2
1
0
7,500 10,000 12,500 15,000 17,500 Energy (cm–1)
(a ) 4T2
4E
4T2(x,y)
4T1(x,y)
4T2(z)
4T1(z)
0 2 4 3
1
4T1(x,y)
4T2(x,y)
4T2(z)
4T1(z)
4E
(b)
(M–1 cm–1 )ε
Figure 1 LF spectra for (a) [FeCl4]1and (b) [Fe(SR)4]1where Rẳ2,3,5,6,-(Me)C6H. The assignments are based upon polarized single-crystal absorption, transverse Zeeman, and MCD spectroscopies.1,2,5
Fe 3d
t2
e b1(3dx2–y2) a1(3dz2) 10Dq
e(3dxz,yz) b2(3dxy)
à
e(3dx z,y z) b2(3dxy) b1(3dx2–y2)
a1(3dz2) à
Fe 3d t2
e
10Dq
(a ) [FeC l4]1– (b ) [F e(S R )4]1–
δ
δ
Figure 2 Fe 3dorbital splitting diagrams for (a) [FeCl4]1and (b) [Fe(SR)4]1as derived from LF analysis of the data inFigure 1.5
Dẳ FeIII
2
5 1 Ez 1
Ex;y
ð1aị
Dẳ FeIII
2
5 k2z Ezk2x;y
Ex;y
!
ð1bị
Rather unexpectedly, Equation (1a) fails to predict the correct magnitude and sign of D for [FeCl4]1 (Dcalcẳ ỵ0.24 cm1vs. Dexpẳ 0.04 cm1). It was determined that the failure results fromanisotropic covalency, i.e., the covalency of the axially split 3dxyand 3dxz,xyorbitals differ. The greater covalency of the 3dxyorbital in the flattenedTdstructure decreases the spin-orbit coupling to the4T1(z) component relative to the coupling to the 4T1(x,y): inEquation (1b),k2z<k2x,y where k2zandk2x,yrelate to the metal character in the 3dxyand 3dxz,xyorbitals, respectively. As a result, the axial ZFS is small and negative.1The importance of anisotropic covalency has been confirmed in a more recent treatment of the full Dtensor, which explicitly includes ZFS contributions from all excited states (not just the lowest energy quartet components) usingab initiomethods.4
There is a dramatic difference between the LF spectra for [FeCl4]1and [Fe(SR)4]1(Rẳ2,3,5,6- (Me)C6H)), as observed from Figure 1.5 The spin-forbidden 4 6A1 transitions are lower in energy for the tetrathiolate by nearly 8,000 cm1; this shift is much larger than would generally be expected from LF arguments. As with the tetrachloride, assignment of these transitions provides experimental LF parameters (Table 1) and a 3d orbital splitting pattern (Figure 2b).5 ZFS in [Fe(SR)4]1also requires the inclusion of anisotropic covalency to properly account for the sign and magnitude ofD. In this case, however,Dis much larger and positive, reflecting that the 3dxz,xy
orbitals are more covalent than the 3dxysuch thatk2z>k2x,yinEquation (1b). The spectroscopically derived axial orbital splitting is much larger in the thiolate and different from the chloride, as evidenced by the sign and magnitude ofand(seeTable 1). The larger axial distortion and higher 3dxz,yzorbital covalency results from the nature of the ligand orbitals involved in bonding with the metal ion (Figure 3). For the chloride, interactions occur through a-donor Cl 3porbital as well as two-donor Cl 3p orbitals. For thiolate ligands, one of the S 3p orbitals is involved in a-bond with C (SCinFigure 3b) thus modifying its spatial distribution and energetics relative to the other two S 3p orbitals, which are and relative to the metal. Most importantly, the SC interaction redirects the S 3porbital depending on the FeSC angle, which is>90. The-type S 3p orbital thus forms a pseudo-bond with the iron. This pseudo- interaction dominates the axial splitting and anisotropic covalency of the Fe 3dorbitals inFigure 2band results in a largely different orbital
Table 1 Compilation of LF and Fe 3dorbital splitting parameters for the [FeCl4]2,1and [Fe(SR)4]2,1redox couples. All values are in cm1except forandgvalues, which have no units.
LF splittings Racah parameters
Species 10Dq X B C D gz gx,y
[FeCl4]1 6,550 172 1,331 444 2,728 0.40 0.04 2.014,8 2.012,5
[FeCl4]2 4,100 23 <10 830 3,430 0.87
[Fe(SR)4]1 4,500 þ1,250 þ2,700 22 2,222 0.02 þ2.4 2.015,9 2.021
[Fe(SR)4]2 3,500 1,400 þ700 620 2,800 0.65 8.7 8.24
Figure 3 Important ligand fragment MOs for (a) chloride and (b) thiolate ligands involved in bonding with a transition metal. The many-atom thiolate orbitals are dominated by the strong SC -bond whereas the
chloride 3patomic orbitals interact solely with the metal 3dorbitals.
694 Spectroscopy and Electronic Structure of [FeX4]n(XẳCl, SR)
ordering than that observed for the chloride. Further, the splitting pattern and orbital covalencies, and hence the ZFS tensor in the tetrathiolate are highly dependent on the orientation of the C in the thiolate ligand (i.e., the FeSC angle as well as the dihedral angle between the FeSC and SFeS planes) as this significantly affects the orientation of the pseudo- bond with the iron.
Both [FeCl4]1and [Fe(SR)4]1exhibit very smallvalues indicating significant ligand character in the spin-forbidden LF transitions. In fact, the values ofare so small (<0.50) that they suggest that these ‘‘LF transitions’’ are actually best described as ligand-to-metal charge transfer (LMCT) transitions. This assignment provides a clear explanation for the large shift observed in the LF spectra between the tetrachloride and tetrathiolate complexes inFigure 1and is demonstrated by variable-photon energy photoelectron spectroscopy (VEPES) data (vide infra).
2.59.2.2 Charge Transfer Spectroscopy—Metal–ligand Bonding
The charge transfer spectrum for [FeCl4]1 (Figure 4a) had previously been assigned under the assumption that transitions in the near-UV correspond exclusively to Fe 3d L 3p or-type LMCT transitions. From valence photoelectron spectroscopy (PES) and DFT calculations (vide infra), however, it is clear that the complete CT manifold must occur within or near the experimentally accessible CT region (<40,000 cm1). The spectrum has, therefore, been reassigned as given inFigure 4a. This new assignment of the spectrum includes contributions from both- and-LMCT transi- tions. The polarized single-crystal CT spectrum for [Fe(SR)4]1 (Rẳ2-(Ph)C6H4) has also been obtained and assigned, as shown inFigure 4b.6The CT spectrum of [Fe(SR)4]1 is red-shifted by 8,000 cm1relative to [FeCl4]1, as also observed in the LF spectra (vide supra). Contrary to the LF region (Figure 1), however, this behavior is expected for CT spectra based on the lower valence state ionization energy (VSIE) of the thiolate relative to Cl ligands. An important feature in the CT spectrum of the tetrathiolate that is not observed in the tetrachloride is the presence of a weak, low-energy CT transition at13,000 cm1, which corresponds to a very weak-LMCT transition.
Lnb3dπ
Lσ 3dσ
Lnb3dσ
Lπ3dπ Lπ3dσ
Lπ3dπ
Lσ 3dσ
L L
x,y z
Energy (cm–1) ε(M–1cm–1)
10,000 20,000 30,000 40,000 50,000 10,000
5,000
0
10,000
5,000
0 (a)
(b)
Figure 4 CT spectrum of (a) [FeCl4]112and (b) [Fe(SR)4]16. The assignment for the former differs from that given in the literature (see text).
(This had originally been assigned as the spin-forbidden ‘‘LF’’ transitions which are now under- stood to be at8,000 cm1lower energy due to their CT character.) Further, the total intensity of -LMCT transitions for the thiolate is much lower than in the chloride, indicating that interactions are much less important to bonding in the tetrathiolate.
2.59.2.3 X-ray Absorption Edges—Covalency of Fe 3dOrbitals
Ligand X-ray absorption spectroscopy (XAS) edges provide a direct probe of the ligand character in the empty 3d orbitals in a transition metal complex, thus giving an experimental method for determining the covalency of metal–ligand bonds. The chlorine and sulfur K-edge XAS edges for [FeCl4]1and [Fe(SR)4]1(RẳPh) have been obtained; the intensity of their pre-edges is directly related to the amount of L 3p character in the Fe 3d orbitals (see Shadle et al.7 and Williams et al.8for details). There is significant ligand character in both species, although the S 3pcharacter in the tetrathiolate complex (30% per empty Fe 3dorbital) is greater than the Cl 3pcharacter in the tetrachloride complex (17% per orbital). These results demonstrate that the tetrathiolate complex is more covalent than the tetrachloride.
2.59.2.4 Variable Energy Valence Photoelectron Spectroscopy—Inverted Bonding Description PES directly probes the molecular orbital (MO) structure of a molecular complex in the valence region. High-resolution data of molecular species are normally only obtained in the gas phase, but under certain circumstances such data can be obtained in the solid state. High-resolution VEPES single-crystal data were obtained for [FeCl4]1 as shown in Figure 5. From the changes in the intensities of the VEPES ionization bands, the nature of the initial state MOs can be quantita-
Figure 5 Single-crystal variable photon energy PES data for [FeCl4]1(CsFeCl4).9Satellite peaks for this complex occur in the same energy region as the counterion Cs 5ppeak. Complementary data from RbFeCl4
(not shown) are clean in that energy region and have allowed detailed analysis of the satellite peaks in [FeCl4]1.
696 Spectroscopy and Electronic Structure of [FeX4]n(XẳCl, SR)
tively determined. The low-binding energy peaks (regions 1–2) decrease in intensity with photon energy; this is consistent with ionization of MOs containing predominantly ligand character.
Conversely, region 3 displays an intensity maximum at h45 eV (observable in Figure 5 and increased intensity of 3 at h= 40 eV relative to h= 25 eV), corresponding with strong Fe 3d character in these deeper-lying MOs. Resonance profiles of specific peaks over the Fe 3p ioniza- tion threshold (52 eV) probe the metal character and its energy distribution over the final (ionized) states in the PES experiment. Profiles for [FeCl4]1 (see Butcher et al.9 for details) indicate that region 3 is resonance enhanced, i.e., Fe 3d character remains in the same energy region in both the initial and final states. Since the Fe 3d character occurs in the same region in both the initial and final states, the electronic structure of the FeIV final state must be quite similar to the FeIII initial state. It has also been found that there is little off-resonance shake-up satellite intensity,9 which further indicates that the electronic structure does not change signifi- cantly on ionization (vide infra). (The satellite peaks are at about the same energy as the counter- ion Cs 3p peak in Figure 5 making it difficult to evaluate their intensity from these data.
However, complementary data for RbFeCl4 are clean in this area and clearly demonstrate the weak intensity of the off-resonance satellites in [FeCl4]1.) Overall, the VEPES data provide a rather unusual electronic structure description for [FeCl4]1: the highest energy MOs (region 1) have mostly ligand character, whereas the Fe 3d character is at much deeper binding energy (region 3). This is opposite from the normal behavior of transition metal complexes where the metal 3dorbitals are above the ligand valence orbitals. This ‘‘inverted’’ bonding scheme can be understood from DFT calculations.
2.59.2.5 Density Functional Calculations—Electronic Structure Description of Ferric Complexes DFT calculations (using the VWN-BP86 functional) have been used to assist in the interpretation of the spectroscopic data that are available for [FeCl4]1 and [Fe(SR)4]1. Our results are generally in good agreement with other theoretical calculations on these systems. It has become customary in recent years to use hybrid functionals (e.g., B3LYP) to investigate transition metal systems; however, our investigations indicate that the BP86 pure DFT provides a more reasonable description of the electronic and geometric structure of high-spin iron systems. Spin-restricted DFT calculations on [FeCl4]1result in a typical transition metal MO scheme (Figure 6a) but are in poor agreement with spectroscopic data (vide supra). Removing the spin restriction (Figure 6b) allows the majority and minority spin orbitals in the Sẳ5/2 system to differ spatially and energetically, allowing for spin polarization of the Fe 3d manifold. These results are in much better agreement with experiments. Spin polarization splits the Fe 3d manifold into a filled 3d and an empty 3d manifold that behave very differently. The five 3d orbitals are so stabilized in energy that they drop below the Cl 3p orbitals creating the inverted bonding scheme that was observed in the VEPES data inFigure 5. Conversely, the 3d orbitals remain above the Cl 3p
orbitals. DFT results for [Fe(SCH3)4]1 are qualitatively very similar to those from [FeCl4]1, although the high energy of the S 3porbitals (relative to the Cl 3p) increases the covalency of the metal–ligand bonds, which are dominated by the interactions in the set of minority-spinorbitals.
The 3d orbital splitting pattern for [FeCl4]1 obtained from DFT (Figure 6, right) is quite similar to that observed from analysis of the low frequency (LF) and CT spectra (seeFigure 2a).
However, the DFT results overestimate the importance of-contributions, affecting the magni- tude of the splitting of the 3dz2 and 3dx2y2 orbitals (). DFT results for [Fe(SCH3)4]1 (not shown) are also in good agreement with the experimentally derived energy splitting, although the results are highly sensitive to the orientation and identity of the thiolate ligands. In any case, the bonding is dominated by the pseudo-S 3p orbital (see Figure 3) with only minor contributions from-interactions from the other S 3porbital and the SC -bond.
A major result from the spin-unrestricted DFT calculations is the dramatic difference between the - and-spin orbital structures, which help explain the VEPES data and the CT nature of the spin-forbidden4 6A1 LF transitions. The deep binding energy Fe 3d character in the valence PES data are the inverted 3d orbitals. Also, we see that the spin-forbidden LF transitions are LMCT transitions between the highest occupied -MOs (L 3p) and the lowest-unoccupied-MOs (Fe 3d). However, the inverted bonding scheme does not exhibit itself in the spin-allowed CT spectra and in the nature of the ligand–metal bonding, both of which depend predominantly on the -spin manifold. Therefore, it is of importance to note that although the inverted bonding scheme dominates much of the spectroscopy of [FeCl4]1 and [Fe(SR)4]1, it does not play a significant role in defining the bonding within the complex. The bonding is dominated by high
covalency due to the low energy of the 3d manifold relative to the ligand valence orbitals, which are higher in energy in the thiolate complex.