Measurement of the Ligand Field Spectra of Ferrous and Ferric Iron

Jun 27, 2017 - Ligand field spectra provide direct information about the electronic structure of transition metal complexes. However, these spectra ar...
2 downloads 9 Views 2MB Size
Download PDF
Article pubs.acs.org/IC

Measurement of the Ligand Field Spectra of Ferrous and Ferric Iron Chlorides Using 2p3d RIXS Anselm W. Hahn,† Benjamin E. Van Kuiken,† Mustafa al Samarai,† Mihail Atanasov,†,‡ Thomas Weyhermüller,† Yi-Tao Cui,§ Jun Miyawaki,§ Yoshihisa Harada,§ Alessandro Nicolaou,∥ and Serena DeBeer*,† †

Max Planck Institute for Chemical Energy Conversion, Stiftstr. 34−36, D-45470 Mülheim an der Ruhr, Germany Bulgarian Academy of Sciences, Institute of General and Inorganic Chemistry, Akad. Georgi Bontchev Street 11, 1113 Sofia, Bulgaria § Institute for Solid State Physics (ISSP), The University of Tokyo, Kashiwanoha, Kashiwa, Chiba 277-8581, Japan ∥ Synchrotron SOLEIL, L’Orme des Merisiers, Saint-Aubin, Boîte Postale 48, 91192 Gif-sur-Yvette Cedex, France ‡

S Supporting Information *

ABSTRACT: Ligand field spectra provide direct information about the electronic structure of transition metal complexes. However, these spectra are difficult to measure by conventional optical techniques due to small cross sections for d-to-d transitions and instrumental limitations below 4000 cm−1. 2p3d resonant inelastic X-ray scattering (RIXS) is a second order process that utilizes dipole allowed 2p to 3d transitions to access d−d excited states. The measurement of ligand field excitation spectra by RIXS is demonstrated for a series of tetrahedral and octahedral Fe(II) and Fe(III) chlorides, which are denoted Fe(III)-Td, Fe(II)-Td, Fe(III)-Oh, and Fe(II)-Oh. The strong 2p spin−orbit coupling allows the measurement of spin forbidden transitions in RIXS spectroscopy. The Fe(III) spectra are dominated by transitions from the sextet ground state to quartet excited states, and the Fe(II) spectra contain transitions to triplet states in addition to the spin allowed 5Γ → 5Γ transition. Each experimental spectrum is simulated using a ligand field multiplet model to extract the ligand field splitting parameter 10Dq and the Racah parameters B and C. The 10Dq values for Fe(III)-Td, Fe(II)-Td, and Fe(III)-Oh are found to be −0.7, −0.32, and 1.47 eV, respectively. In the case of Fe(II)-Oh, a single 10Dq parameter cannot be assigned because Fe(II)-Oh is a coordination polymer exhibiting axially compressed Fe(II)Cl 6 units. The 5T → 5E transition is split by the axial compression resulting in features at 0.51 and 0.88 eV. The present study forms the foundation for future applications of 2p3d RIXS to molecular iron sites in more complex systems, including iron-based catalysts and enzymes.

■

INTRODUCTION Iron with its diverse range of accessible oxidation states is involved in a wide array of homogeneous,1−4 heterogeneous,5,6 and bioinorganic reactions.7,8 Toward the broader goal of a sustainable energy economy, iron-based catalysts have also attracted much interest in recent years as potential replacements for costly precious metal catalysts.9,10 A crucial aspect in understanding the mechanism of iron-based catalysts, and optimizing their efficiency, is the ability to understand the electronic structural changes, which occur at the catalytic active site. As the low lying excited states of an iron active site dictate the reaction energetics, the ability to experimentally measure the low lying d-to-d transitions is of particular interest. However, for the most common oxidation states of Fe in the high-spin configuration (i.e., Fe(III) S = 5/2 and Fe(II) S = 2), the measurement of d-to-d transition energies presents several experimental challenges. In the case of high-spin Fe(III), all dto-d transitions are both spin and parity forbidden. For high© 2017 American Chemical Society

spin Fe(II) in an Oh coordination environment, only the low lying 5T2g to 5Eg transition spin is allowed. Further, this transition often appears at energies less than 0.3 eV, and no assignments are affected. The incident beam at SEXTANTS is focused to a spot size of ∼2(v) × 100(h) μm. At BL07LSU, a linear horizontal polarization was selected for the measurements of the RIXS spectra. The beam was focused to a spot size of ∼4(v) × 30(h) μm. At both beamlines, the final resolution in the RIXS spectra was estimated to be ∼300 ± 30 meV when determined by the full width at half-maximum (fwhm) of a Gaussian fit of the elastic scattering features. The small beam spot size and high flux required for 2p3d RIXS measurements lead to high rates of beam damage for molecular spectra.26 For this reason, we applied a sample raster-scanning protocol previously used to obtain damage-free RIXS spectra of vanadium complexes.15 All samples were moved constantly in the beam during the course of the data collection. Scan speeds of 50, 100, 200, and 400 μm/s were tested in order to establish the required scan rate. In the present study, it was established that damage-free 2p3d RIXS spectra could be obtained at a scan speed of 100 μm/s at SEXTANTS and 200 μm/s at BL07LSU. Moreover, reproducibility of RIXS spectra collected at one beamline was verified at the other, which helps confirm the absence of beam damage despite differences in spot size and photon flux. At both beamlines, XAS spectra were measured via the total electron yield (TEY) from the sample drain current. The XAS spectra were used to choose incident energies for the emission measurements. Data were measured at a single sample spot because TEY measurements are surface sensitive, and due to the inability to obtain completely homogeneous coverage on the carbon tape, the powder

Figure 1. Schematic 1: Electron representation of the RIXS process. Black dashed lines correspond to the absorption and emission events giving rise to d-to-d transition energies in RIXS spectroscopy. In this illustration, the emission even corresponds to a spin-flip transition enabled by spin−orbit coupling of the 2p core hole. Emission leading to the same initial sextet ground state can also happen and result in the elastic line in the RIXS spectrum.

a soft X-ray photon causing a dipole allowed transition from the 2p63dn ground state to a 2p53dn+1 intermediate state. This state then decays via a dipole allowed transition to a 2p63dn′ final state. The difference in the incoming and outgoing photon energies corresponds to an excitation energy, and this quantity is called the energy transfer (ET). Because the 2p63dn′ state may differ from the ground state, 2p3d RIXS spectroscopy accesses d−d transition energies via two dipole allowed processes. In addition, 2p spin−orbit coupling (SOC) in the intermediate state enables spin flip transitions, and hence the final state spin multiplicity may differ from that of the ground state. While the illustration in Figure 1 is convenient, it is important to keep in mind the true nature of the states involved in the RIXS process. Specifically, transitions take place between the eigenstates of the many-electron relativistic Hamiltonian that includes SOC. These states can be represented as a linear combination of the Slater determinants that span the eigenspace of the Hamiltonian. Due to the presence of SOC, the states will have contributions from determinants with different spin multiplicities. The SOC in the initial and final states originates from unpaired 3d electrons. The magnitude of this interaction is typically small, and these states are usually characterized by a single spin multiplicity. This is not the case for intermediate states containing a 2p core hole, which may be of mixed character. For example, consider a d6 complex that possesses a quintet ground state; excitation of a 2p electron can result in an intermediate state that is characterized by a linear combination of quintet and triplet determinants. Consequently, a nonzero transition dipole can exist between the intermediate state and both quintet and triplet final states. This generates a virtual “breaking” of the conventional selection rules for d−d excitations, making 2p3d RIXS a uniquely powerful tool for investigating the electronic structure of first row transition metal complexes. Herein, we demonstrate the ability of 2p3d RIXS to map the multiplet structure of ferrous and ferric iron chlorides. The 8204

DOI: 10.1021/acs.inorgchem.7b00940 Inorg. Chem. 2017, 56, 8203−8211

Article

Inorganic Chemistry

Figure 2. Fe RIXS spectra of (a) Fe(III)-Td, (b) Fe(III)-Oh, (c) Fe(II)-Td, and (d) Fe(II)-Oh collected at the L3 edge. Experimental data are shown in black with individual fit components in blue. The sum of the fits is the orange dashed curve. samples lack macroscopic structural homogeneity. Consequently, continuously moving the sample results in measuring different Fe concentrations at different energies. Hence, undamaged XAS spectra were obtained by reducing the intensity of the incident X-ray beam. This was achieved by one or more of the following: inserting filters, translating the sample out of the beam focus, closing upstream apertures to attenuate the beam, or detuning the undulator. Last, the incident energy of the monochromator was calibrated by setting the L3 maximum of Fe2O3 to 708.5 eV in order to facilitate comparison with previously published L3-edge XAS measurements.27 Experimental Fitting Procedure. In order to identify spectral features in the RIXS spectra, the spectral profile was fitted by a series of Gaussian functions. A least-squared minimization utilizing the Levenberg−Marquardt algorithm was used to fit the position, width, and amplitude of the functions.28 Previous fits of 2p3d RIXS spectra have shown that the actual Lorentzian contribution to spectral lineshapes is quite small compared to the instrumental Gaussian contribution.29,30 Ligand-Field-Multiplet Simulations. The multiplet program Xclaim31 was used to simulate the transition RIXS spectra presented below. It is a multiplet approach based on the idea of Cowan’s atomic multiplet program RCN.32 This approach employs a model that only explicitly treats the 2p and 3d electrons of Fe and employs a Hamiltonian that includes the effects of a ligand field and spin−orbit coupling. Cubic symmetry is assumed unless otherwise stated. With these assumptions, the complete set of input parameters is comprised of the ligand field splitting (10Dq) and the set of two-electron integrals whose magnitudes are given by the Racah parameters B and C.31,32 It is conventional to reduce the magnitudes of the two-electron integrals linearly from their atomic values to account for covalency unless the C/B ratio strongly deviates from 3.73 (the C/B ratio in Orgel diagrams).33,34 The parameters used for each spectrum are described in detail in the text. The RIXS spectra were plotted with Gaussian and Lorentizian widths of 0.5 and 0.25 eV, respectively.

displayed under the data in Figure 2, with the summation of the fits in orange. The energies of the features are given in Table 1, Table 1. Energies (eV) of the Gaussian Peaks Fit to the Experimental Spectra in Figure 2a d-to-d transition energy (eV) fitted feature

Fe(III)-Td

Fe(III)-Oh

Fe(II)-Td

Fe(II)-Oh

1 2 3 4 5 6 7 8 9

1.66 1.89 2.32 2.79 3.26 3.69 4.37

1.13 1.56 2.37 2.00 3.36 3.68 4.09

0.32 2.0 2.35 2.6 2.8 3.09 3.46

0.51 0.88 1.51 2.02 2.39 2.56 2.87 3.43 3.85

a

It is noted that the maximum of the standard error of regression did not exceed 0.03 eV.

and the complete set of fitting parameters is provided in Table S1 of the SI. In all cases, the total fit tracks the experimental data quite well. The largest deviation is near 0 eV in Fe(II)-Td, but this is only due to the decision to employ only two Gaussian functions in the 0 to 1 eV region. This choice was made because, as discussed below, in Td symmetry there should only be the spin-allowed 5E → 5T transition. However, deviations from tetrahedral symmetry produced by the Jahn− Teller effect in Fe(II) can split this transition, giving rise to multiple features, and the crystal structure does show deviation from a perfect tetrahedron.35 Inspection of Figure 2 clearly shows that differences in the 2p3d RIXS spectra can be observed due to changes in both local site symmetry and oxidation state. It is instructive to first compare the Td and Oh and ferric complexes, shown in Figure 2a,b (top). As both complexes have a high-spin d5 ground state electron configuration, the observed features all correspond to spin forbidden transitions from the sextet ground state to states of lower spin multiplicity. Transitions to these states are spinforbidden in UV−vis spectroscopy but gain intensity in RIXS

■

RESULTS Figure 2 shows the 2p3d RIXS spectra for (a) Fe(III)-Oh, (b) Fe(III)-Td, (c) Fe(II)-Oh, and (d) Fe(II)-Td. All RIXS spectra are presented on the energy transfer (ET) axis, which is the energy difference between absorbed and emitted photon energy. The spectra shown in Figure 2 were collected with excitation energies set to the maximum of Fe L3 edge for each complex. The Gaussian fits to the experimental spectra are 8205

DOI: 10.1021/acs.inorgchem.7b00940 Inorg. Chem. 2017, 56, 8203−8211

Article

Inorganic Chemistry

deviation from Td symmetry, only a single feature is fit for each of the two low energy transitions. The RIXS features are identified at 1.66 and 1.89 eV, and both of these values are found to be between the reported axial and equatorial energies for each term, indicating good agreement with the optical experiment. While the feature at 2.18 eV is not readily identifiable in the RIXS spectra, this may again result from the limited experimental resolution. Further, we note that there are features at 2.32 and 2.79 eV that are in good agreement with transitions identified in the optical measurements. The simulations discussed below will assist in more clearly assigning the features observed in the RIXS spectra. The available optical data on the remaining complexes are limited, but we summarize briefly what is known here. In the case of Fe(II)-Td, there is some controversy regarding the assignments. Furlani et al. reported the 5E → 5T transition appears at ∼4500 cm−1 (0.56 eV) in the solid state and at ∼4000 cm−1 (0.50 eV) in solution.39 However, Solomon and co-workers stated that the 5E → 5T transitions were below 4000 cm−1 and could not be observed using their MCD instrument. As mentioned above the lowest energy transition in Fe(II) complexes must correspond the spin-allowed 5Γ → 5Γ transition. In the RIXS spectrum of Fe(II)-Td, the first inelastic feature appears at ∼0.32 eV (2800 cm−1). This supports the conclusions of Solomon and co-workers that the transition should be found below 4000 cm−1,40 and it highlights the ability of RIXS to measure transition energies that are out of the range of conventional spectrometers. It is possible that the features assigned by Furlani et al. were due to counterion or solvent transitions. The optical reflectance spectrum of Fe(III)-Oh was reported by Yamatera and Kato, who assigned features at 8900 and 12 800 cm−1 to 6A1g → 4T1g and 6A1g → 4T2g transitions, respectively.41 These transitions are in good agreement with the first two features observed in the RIXS spectrum at 1.13 eV (9110 cm−1) and 1.56 eV (12600 cm−1). Finally, to the best of our knowledge, there are no reported optical data for the ligand field spectra of Fe(II)-Oh. In order to quantitatively establish the connection between the measured RIXS spectra and ligand field theory, the spectra have been simulated using a ligand field multiplet model. The parameters that determine the valence excitation energies are the magnitude of the ligand field splitting and the electron repulsion parameters. The magnitudes of the electronic interactions can be described by the Slater−Condon parameters, F2 and F4, or equivalently the Racah parameters B and C. The actual input to the multiplet program is given in terms of the normalized Slater−Condon parameters, F2 and F4, which are related to the Racah parameters as follows.42

due to the large spin−orbit coupling associated with the intermediate state core hole. One notes that in Fe(III)-Td (Figure 2a) the lowest lying transition appears at ∼1.5 eV, while in Fe(III)-Oh the lowest lying transition appears at ∼1.1 eV. In addition, in the Fe(III)-Oh case, the d-to-d transitions are more clearly separated. The origins of the above observations will be discussed below in terms of ligand field theory. Similar trends are observed when comparing the Fe(II)-Td and Fe(II)-Oh ferrous complexes. In the Oh case, many wellresolved features are observed in the RIXS spectrum (d), which coalesce into a broad band in the case of Fe(II)-Td (c). These trends thus clearly establish the sensitivity of 2p3d RIXS to the local site symmetry. In addition to the changes due to local site symmetry, the spectra for the ferrous complexes are clearly distinct from the ferric analogues in that they contain an intense low energy feature at 0.32 eV for Fe(II)-Td and at 0.51 eV for Fe(II)-Oh. This low-energy feature corresponds to the spinallowed 5Γ → 5Γ transition between the e and t2 manifold in the Td case and the t2g and eg manifolds in the Oh case. No reasonable values of ligand field parameters allow for any other assignment of these features. The higher energy spectral features (>1.0 eV) correspond to excited states that possess a different spin state than the ground state. It is instructive to compare the RIXS spectra with previously published optical data. The most comprehensive study of the ligand field excitations in iron chlorides to date is the investigation of Fe(III)-Td by Solomon and co-workers.36 Both polarized single crystal absorption and MCD measurements were used to assign the d−d transitions in [PPh4][Fe(III)Cl4]. Table 2 compares the assignments from the optical Table 2. Comparison of Fitted RIXS Spectrum of Fe(III)-Td with Transition Energies from MCD and Polarized Absorption Measurements optical assignment 4

T1z 4 T1x,y 4 T2z 4 T2x,y 4 E 4 T2 4 E

optical (eV)

X-ray (eV)

1.55 1.71 1.83 1.93 2.18 2.26 2.64

1.66 1.89

2.32 2.79

measurements with the features fitted in the RIXS spectrum. The RIXS transitions were associated with the visible assignments that they most closely matched. We note that small deviations in the RIXS transition energies relative to the optical data likely arise from the use of the [NEt4]+ cation in the present study, as opposed to [PPh4]+ in the previous optical studies. The [PPh4]+ cation results in a crystal structure that exhibits a significant compression along the S4 axis. Hence, the resulting Fe(III)-Td possesses a distorted D2d symmetry with Cl−Fe−Cl bond angles of 107.5° and 113.5°.37 This distortion removes the degeneracy within both the t2 and e set of d orbitals. Consequently, the two lowest energy transitions are split by 0.16 eV (4T1) and 0.1 eV (4T2) into equatorial (x,y) and axial (z) components, giving rise to four features in the optical spectra at 1.55, 1.71, 1.83, and 1.91 eV. On the other hand, the low-temperature crystal structure of [NEt4][Fe(III)Cl]4 is less compressed with Cl−Fe−Cl bond angles ranging between 107.8° and 110.8°.38 Given the experimental resolution for the RIXS measurements and the smaller

B = F 2/49 − 5F 4 /441 C = 35F 4 /441

It is typically the case for simulations of X-ray spectra that a uniform percentage reduction is applied to both the free ion values of the F2 and F4 terms to account for covalency. However, in cases of highly covalent complexes, such as Fe(III) compounds, this is not justified because the large nephelauxetic reduction gives rise to reduced B values, meaning that a C/B ratio of ∼4 is an underestimate. Using the previously discussed MCD measurements, Solomon and co-workers determined a C/B ratio of 6.1 for Fe(III)-Td. Consequently, the F2 and F4 values must be adjusted independently for Fe(III) complexes, while a uniform reduction is employed in the case of the less 8206

DOI: 10.1021/acs.inorgchem.7b00940 Inorg. Chem. 2017, 56, 8203−8211

Article

Inorganic Chemistry

determined by averaging the energies of all microstates comprising a term. The two lowest energy features in Fe(III)-Td were associated with the transitions to 4T1 and 4T2 states at 1.66 and 1.89 eV, respectively. These energies are well reproduced by the simulation shown in Figure 3a, where features are found at 1.66 and 1.91 eV, and the multiplicities show that these are the two lowest-lying quartet states. The most obvious discrepancy between the experiment and simulation is that the features in the simulation are more well-separated. This could be attributed, in part, to the fact that a D2d compression occurs experimentally that broadens the spectrum, while the simulations employ a rigorous tetrahedron. The next set of transitions lies near 2.3 eV. In the simulated spectra there are three transitions visible at 2.12, 2.34, and 2.47 eV arising from transitions to quartet states. While only a single feature has been fit in the experimental spectra, multiple states could certainly contribute this feature. These states are either independent of or weakly dependent on the ligand field splitting. Consequently, one should expect a ∼2.3 eV transition in all tetrahedral Fe(III) complexes. It should be noted that there are numerous doublet states located between 2.5 and 3 eV, but these states do not give rise to intense absorption features. As a result, it is concluded that the dominant observable d−d transitions are all of 6Γ → 4Γ origin. In general, our calculations suggest that the intensities of transitions in RIXS spectra decrease with increasing ΔS (I(ΔS = 0) > I(ΔS = 1) > I(ΔS = 2)), but future studies will more rigorously test this observation. Finally, the ligand field parameters used to reproduce the RIXS spectrum are in reasonable agreement with the parameters derived by Solomon and co-workers. They reported a 10Dq value of 0.8 eV (6550 cm−1) and B and C values of 444 and 2728 cm−1, and the present simulations of the RIXS measurements utilized values of 0.7 eV and 399 and 2625 cm−1 The spectrum and corresponding simulation of Fe(III)-Oh is shown in Figure 3b. Here, the change to Oh symmetry results in nearly doubling the magnitude 10Dq value from −0.7 to 1.47 eV (Table 3). This value is in good agreement with the 10Dq of

covalent Fe(II) complexes. The simulation parameters determined for each complex are listed in Table 3. In the Table 3. Input Parameters for 2p3d RIXS Simulationsa Fe(III)-Td Fe(III)-Oh 10Dq (eV) F2(3d,3d) (eV) F4(3d,3d) (eV) 10Dq (cm−1) B (cm−1) C (cm−1)

−0.7 4.7 4.1 −6550 399 2625

1.47 5.630 4.349 11856 529 2784

Fe(II)-Td

Fe(II)-Oh

−0.317 7.512 (68.5%) 4.669 (68.5%) −2557 810 2988

0.51/0.88 8.225 (75%) 5.112 (75%) 4117/7137 886 3271

a

Two 10Dq values are given in the case of Fe(II)-Oh to account for the axially compressed structure. Percentages indicate reduction from atomic values.

case of Fe(II)-Oh, there are two values for the 10Dq, which accounts for the axial distortion in the geometric structure discussed in the Experimental Methods section. The LMCT features in the optical spectra of Fe chlorides typically appear at 20 000 cm−1 (∼2.5 eV) and above. Consequently, it is not expected that CT events contribute to the RIXS spectra at energy transfer values below ∼2.5 eV, and these effects are thus ignored in the simulations. As a result, the parameters for the simulations were determined by finding the best match between the experiment and simulation for the transitions found in the 0−2.5 eV energy window. Figure 3 shows the comparison between the simulations and experimental data for each complex, and individual term energies and multiplicities are given in Table S5 of the SI. The trends observed in the experimental spectra are generally reproduced in the simulations. The simulated spectra match the experiment within the first ∼2.5 eV; at higher transition energies there are also CT contributions to the RIXS spectra, consistent with previous MCD results of Neuenschwander et al.43 Previous studies have shown that multiplet calculations can reproduce ligand field spectra with reasonable accuracy.15,44 In addition to providing spectral profiles to correlate the simulated and experimental data, the simulations yield a fully diagonalized ligand field Hamiltonian. Consequently, term energies can be

Figure 3. Ligand field multiplet simulations of the 2p3d RIXS spectra of Fe(III)-Oh, Fe(III)-Td, Fe(II)-Oh, and Fe(II)-Td. Experimental RIXS spectra are presented in black together with fitted features (blue). The simulated spectra are plotted in orange. 8207

DOI: 10.1021/acs.inorgchem.7b00940 Inorg. Chem. 2017, 56, 8203−8211

Article

Inorganic Chemistry 1.37 (11 100 cm−1) determined from optical measurements.41 The first fitted band in the experimental RIXS spectrum is found at 1.13 eV. The simulated spectrum contains transitions at 1.13 and 1.25 eV that contribute to the first experimental feature. We assign these two transitions as 6A1g → 2T2g and 6A1g → 4T1g, respectively, with the dominant intensity contribution arising from the quartet state. The second and third features are fitted in the experimental spectrum at 1.57 and 2.36 eV, which are in good agreement with the simulated 6A1g → 4T2g transition at 1.63 eV and the 6A1g → 4A1g at 2.36 eV. It is worth noting that the 4A1g energy is independent of the ligand field strength. Above 2.5 eV, agreement between the experimental and simulated spectrum deteriorates as expected due to the presence of CT features. The multiplet simulations also reproduce the expected trends in the B values on going from Fe(III)-Td15,45with B = 399 cm−1 to Fe(III)-Oh with B = 529 cm−1, indicating a higher covalency for Fe(III)-Td, which is consistent with previous spectroscopic studies.27 Upon going from Fe(III) to Fe(II), there is the possibility of a spin-allowed transition between the t2g and eg orbitals. In the spectrum of Fe(II)-Td, the lowest energy feature in the RIXS spectrum is the 5E → 5T transition found in the simulated spectrum at 0.33 eV above the ground state. This transition sets the 10Dq value at −0.32 eV. The value of B then determines the energies of the dense band of triplet states that gives rise to the feature between 1.8 and 3.5 eV. The lowest energy triplet state is the 3T1 state found at 1.77 eV, which is in good agreement with the onset of the triplet band absorptions in the data. These triplet states are closely spaced, so that individual state assignments cannot be made. For example, triplet states are found at 1.94, 2.04, 2.10, and 2.23 eV. Nevertheless, the RIXS spectrum provides a map of the density of low-lying excited states. The 10Dq value determined here is lower than previously reported values by Furlani et al. As previously discussed, the RIXS data support the conclusions of Solomon and co-workers that the 5E → 5T transition occurs below 4000 cm−1 (∼0.5 eV). The last simulation is for the spectrum of Fe(II)-Oh (Figure 3d). The agreement between the experiment and simulation is remarkably good. The two lowest energy transitions at 0.54 and 0.91 eV are transitions to quintet states derived from the 5Eg state in Oh symmetry, where the lower and higher energy excitations are to the dx2−y2 (B1g) and dz2 (A1g), respectively. This splitting is due to an axially compressed geometry, and the geometric parameters discussed in the Experimental Methods section show that Fe(II)-Oh possesses D4h symmetry. These excitation energies are given by setting the dx2−y2 and dz2 orbital energies to dz2 = 0.88 eV and dx2−y2 = 0.55 eV. The remaining spectrum is comprised of excitations to triplet states. The lowest lying 5Γ → 3Γ transitions are found at 1.5, 1.8, and 2.0 eV. The B value of 886 cm−1 is greater than the value for Fe(II)-Td and reflects a reduction in covalency on going to Oh symmetry. To further explore the effects of the geometric distortion on the RIXS spectrum, Figure 4 displays simulations of the Fe(II)Oh spectrum in several different symmetries. First, a perfectly octahedral simulation was performed where the 10Dq value was set to the energy 0.51 eV to reproduce the first excitation energy (Figure 4, green line). It is clear from this simulation that an Oh geometry cannot be used to model this spectrum. The simulation fails to reproduce the feature at 0.88 eV, and it contains no triplet transitions below 1.75 eV. When a splitting is added to the eg orbital (Figure 4, orange line), the

Figure 4. Experimental spectrum of Fe(II)-Oh shown in black compared with simulated spectra to examine the effect of an axial distortion on the RIXS spectrum. The three symmetries considered are perfectly Oh (green), splitting of only eg orbitals (orange), and splitting of both t2g and eg (blue).

experimental spectrum is reproduced quite accurately. It is also the case that the t2g set of orbitals should also be split by the axially compressed geometry, and the dxy orbital should be found at a lower energy than the dxz and dyz. However, this splitting is expected to be much smaller than the splitting in the eg set (consistent with chloride being a much better σ than πdonor donor) and therefore is likely not visible with the present experimental resolution. In order to verify this, simulations have also been carried out with a t2g splitting that is 10% of the eg splitting (as shown in blue in Figure 4). As can be seen, this spectrum is almost identical (albeit broadened) when compared to the spectrum with no splitting in the t2g band. Larger splittings within the t2g band smear out the spectrum. Hence, we can conclude that the splitting in the t2g band is less than the 300 meV resolution of the experiment, and it can be safely ignored. An important detail that we have pointed out here is that the spectrum broadens when the symmetry deviates from cubic. This is due to the fact that the states split when the symmetry is lowered, and this effect was also present due to the axial compression in Fe(III)-Td. In general, the RIXS spectra of low-symmetry systems will be more complex, and ligand field models will require a great number of adjustable parameters. In these cases, interpretation of the spectra will likely require determination of the excited states using sophisticated ab initio quantum chemical approaches.

■

DISCUSSION Herein, we have demonstrated that 2p3d RIXS measurements may be utilized to extract ligand field parameters for a series of Fe chlorides. It is also crucial to discuss the advantages and limitations of RIXS compared to other techniques for investigating ligand field effects. As discussed above, RIXS can measure the d−d excitation spectrum without constraints as to the molecular symmetry or the spin states involved. This is a large advantage over conventional absorption or MCD measurements, which are bound by spin and parity selection rules. In addition, excitations at very low energy can be measured. Optical spectrometers rarely extend below 4000 cm−1. Below 4000 cm−1, electronic excitations may be difficult to distinguish from vibrational bands, which limits the use of IR techniques to study electronic excitations. For example, we were able to unambiguously assign the 5E → 5T2 transition in Fe(II)-Td at 0.32 eV. Both the specificity for d-to-d excitations and the broad energy range allows for a much more detailed element specific analysis of the complete electronic structure. Previous optical studies did not have the capability to observe 8208

DOI: 10.1021/acs.inorgchem.7b00940 Inorg. Chem. 2017, 56, 8203−8211

Article

Inorganic Chemistry the lowest lying d−d feature at 0.32 eV in Fe(II)-Td. On the other hand, when d-to-d excitations can be measured, optical experiments provide much higher resolution data than RIXS. In the case of Fe(III)-Td, Solomon and co-workers were able to assess the effects of lowering symmetries from Td to D2d, which resulted in intraband splittings of 1331 and 172 cm−1 for t2 and e, respectively. These effects are below the 300 meV resolution of the present set of experiments. That said, the newest generation of RIXS spectrometers offers resolution on the order of 50 meV (∼400 cm−1). Consequently, some of the resolution limitations will be overcome with newly designed instrumentation.46 Another approach for extracting ligand field parameters is the use of L-edge XAS. Wasinger et al. previously reported 10Dq values for iron chloride complexes.27 Their measurement contains the same series of complexes studied here with the exception of Fe(II)-Oh, which was synthesized as a monomeric [NaK3](Fe(II)Cl6) salt. The value of 10Dq was reported to be −0.5, 1.2, −0.3, and 0.6 for Fe(III)-Td, Fe(III)-Oh, Fe(II)-Td, and Fe(II)-Oh, respectively. While our numbers are in relatively good agreement for the 10Dq in terms of trends and magnitudes, the L-edge absorption spectra do not appear to be as sensitive to the B and C values as the RIXS. In their ligand field multiplet simulations, the authors used a C/B ratio of 3.73, which we found to be much too small to accurately reproduce the RIXS spectra, and it is much smaller than the ratio derived from optical spectroscopy for the tetrahedral molecules. 2p3d RIXS spectra of Fe chlorides are more highly featured than the same L-edge absorption. This helps to decrease ambiguities in the assigning of features and the extraction of parameters. While this is not to suggest that L-edge absorption measurements lack utility, a more detailed picture of the valence electronic structure may be attainable from a 2p3d RIXS measurement.

■

CONCLUSION

■

ASSOCIATED CONTENT

■

SQUID and Mössbauer data for each of the synthesized complexes and a complete list of all fitting parameters from the fits shown in Table 1 (PDF)

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. ORCID

Thomas Weyhermüller: 0000-0002-0399-7999 Yi-Tao Cui: 0000-0001-7104-0059 Yoshihisa Harada: 0000-0002-4590-9109 Serena DeBeer: 0000-0002-5196-3400 Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS The authors acknowledge the Max Planck Society for funding. S.D. acknowledges the DFG for financial support (project DE 1877/1-1, within the SPP 1927 “Iron−Sulfur for Life”). We acknowledge SOLEIL for use of synchrotron radiation facilities under proposals 20140793 and 20151219 at the SEXTANTS beamline. We also acknowledge SPring-8 for beamtime and technical support. Experiments at SPring-8 BL07LSU were performed jointly by the Synchrotron Radiation Research Organization and the University of Tokyo (Proposal No. 2016A7506).

■

REFERENCES

(1) Bauer, I.; Knölker, H.-J. Iron Catalysis in Organic Synthesis. Chem. Rev. 2015, 115, 3170−3387. (2) Bolm, C.; Legros, J.; Le Paih, J.; Zani, L. Iron-Catalyzed Reactions in Organic Synthesis. Chem. Rev. 2004, 104, 6217−6254. (3) Castle, S. L. Organic synthesis: Better chemistry through radicals. Nature 2014, 516, 332−3. (4) Lo, J. C.; Gui, J.; Yabe, Y.; Pan, C.-M.; Baran, P. S. Functionalized olefin cross-coupling to construct carbon−carbon bonds. Nature 2014, 516, 343−348. (5) Okamura, M.; Kondo, M.; Kuga, R.; Kurashige, Y.; Yanai, T.; Hayami, S.; Praneeth, V. K.; Yoshida, M.; Yoneda, K.; Kawata, S.; Masaoka, S. A pentanuclear iron catalyst designed for water oxidation. Nature 2016, 530, 465−8. (6) Kandemir, T.; Schuster, M. E.; Senyshyn, A.; Behrens, M.; Schlögl, R. The Haber−Bosch Process Revisited: On the Real Structure and Stability of “Ammonia Iron” under Working Conditions. Angew. Chem., Int. Ed. 2013, 52, 12723−12726. (7) Kowalska, J.; DeBeer, S. The role of X-ray spectroscopy in understanding the geometric and electronic structure of nitrogenase. Biochim. Biophys. Acta, Mol. Cell Res. 2015, 1853, 1406−1415. (8) Lancaster, K. M.; Roemelt, M.; Ettenhuber, P.; Hu, Y.; Ribbe, M. W.; Neese, F.; Bergmann, U.; DeBeer, S. X-ray Emission Spectroscopy Evidences a Central Carbon in the Nitrogenase Iron-Molybdenum Cofactor. Science 2011, 334, 974−977. (9) Chirik, P. J. Iron- and Cobalt-Catalyzed Alkene Hydrogenation: Catalysis with Both Redox-Active and Strong Field Ligands. Acc. Chem. Res. 2015, 48, 1687−1695. (10) Bedford, R. B. How Low Does Iron Go? Chasing the Active Species in Fe-Catalyzed Cross-Coupling Reactions. Acc. Chem. Res. 2015, 48, 1485−1493. (11) Smith, A. T.; Pazicni, S.; Marvin, K. A.; Stevens, D. J.; Paulsen, K. M.; Burstyn, J. N. Functional Divergence of Heme-Thiolate Proteins: A Classification Based on Spectroscopic Attributes. Chem. Rev. 2015, 115, 2532−2558. (12) Solomon, E. I.; Brunold, T. C.; Davis, M. I.; Kemsley, J. N.; Lee, S.-K.; Lehnert, N.; Neese, F.; Skulan, A. J.; Yang, Y.-S.; Zhou, J.

In summary, we have demonstrated that 2p3d RIXS can be used to provide detailed insight into the complex electronic structure of d5 and d6 iron chloride complexes. The ability to observe both spin and parity forbidden transitions over a wide energy window is a major advantage of this method over standard optical approaches. The combination of RIXS spectroscopy with simulations based on ligand field theory allows for the extraction of ligand field parameters. Together the spectroscopy and model parameters give a qualitative picture of the chemical bonding (in terms concepts such as covalency) and a thorough mapping of the excited state spectrum. By providing information on metal−ligand bonding and identifying excited states that may participate in chemical reactions, RIXS is an ideal tool to probe reactivity in transitionmetal-containing systems. While improved experimental resolution is still required to assign all ligand field transitions, the present study provides a foundation for future 2p3d RIXS studies of iron active sites, in both biological and chemical catalysis.

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.inorgchem.7b00940. 8209

DOI: 10.1021/acs.inorgchem.7b00940 Inorg. Chem. 2017, 56, 8203−8211

Article

Inorganic Chemistry

prediction of metal ion photoreduction. J. Electron Spectrosc. Relat. Phenom. 2015, 198, 31−56. (27) Wasinger, E. C.; de Groot, F. M. F.; Hedman, B.; Hodgson, K. O.; Solomon, E. I. L-edge X-ray Absorption Spectroscopy of NonHeme Iron Sites: Experimental Determination of Differential Orbital Covalency. J. Am. Chem. Soc. 2003, 125, 12894−12906. (28) Press, W. H.; Teukolsky, S. A.; Vetterling, W. T.; Flannery, B. P. Numerical Recipes 3rd ed.: The Art of Scientific Computing; Cambridge University Press, 2007; p 1256. (29) Ghiringhelli, G.; Matsubara, M.; Dallera, C.; Fracassi, F.; Gusmeroli, R.; Piazzalunga, A.; Tagliaferri, A.; Brookes, N. B.; Kotani, A.; Braicovich, L. NiO as a test case for high resolution resonant inelastic soft x-ray scattering. J. Phys.: Condens. Matter 2005, 17, 5397. (30) Wang, H.; Butorin, S. M.; Young, A. T.; Guo, J. Nickel Oxidation States and Spin States of Bioinorganic Complexes from Nickel L-edge X-ray Absorption and Resonant Inelastic X-ray Scattering. J. Phys. Chem. C 2013, 117, 24767−24772. (31) Fernández-Rodríguez, J.; Toby, B.; van Veenendaal, M. Xclaim: A graphical interface for the calculation of core-hole spectroscopies. J. Electron Spectrosc. Relat. Phenom. 2015, 202, 81−88. (32) Cowan, R. D. The theory of atomic structure and spectra. University of California Press, 1981; Vol. 3. (33) Griffith, J. S.; Orgel, L. E. Spectra of Transition-Metal Complexes of the Type [Co(NH3)5X]2+. J. Chem. Soc. 1956, 4981− 4983. (34) Orgel, L. E. Spectra of Transition-Metal Complexes. J. Chem. Phys. 1955, 23, 1004−1014. (35) Zhou, B.; Ogura, S.; Kato, N.; Idobata, Y.; Kobayashi, A.; Kobayashi, H. Single-component Magnetic Molecular Conductor [Fe(dmdt)2] (dmdt: dimethyltetrathiafulvalenedithiolate). Chem. Lett. 2013, 42, 977−979. (36) Deaton, J. C.; Gebhard, M. S.; Solomon, E. I. Transverse and longitudinal Zeeman effect on [PPh4][FeCl4]: assignment of the ligand field transitions and the origin of the 6A1 ground-state zerofield splitting. Inorg. Chem. 1989, 28, 877−889. (37) Herber, R. H.; Nowik, I.; Kostner, M. E.; Kahlenberg, V.; Kreutz, C.; Laus, G.; Schottenberger, H. Mössbauer Spectroscopy and X-ray Diffraction Study of 57Fe-Labeled Tetrachloroferrate(III)-Based Magnetic Ionic Liquids. Int. J. Mol. Sci. 2011, 12, 6397. (38) Lutz, M.; Huang, Y.; Moret, M.-E.; Klein Gebbink, R. J. M. Phase transitions and twinned low-temperature structures of tetraethylammonium tetrachloridoferrate(III). Acta Crystallogr., Sect. C: Struct. Chem. 2014, 70, 470−476. (39) Furlani, C.; Cervone, E.; Valenti, V. Electronic structure and spectra of tetrahalogenoferrates (II). J. Inorg. Nucl. Chem. 1963, 25, 159−163. (40) Pavel, E. G.; Kitajima, N.; Solomon, E. I. Magnetic Circular Dichroism Spectroscopic Studies of Mononuclear Non-Heme Ferrous Model Complexes. Correlation of Excited- and Ground-State Electronic Structure with Geometry. J. Am. Chem. Soc. 1998, 120, 3949−3962. (41) Yamatera, H.; Kato, A. The d-d Bands of Hexachloroferrate(III) Ion. Bull. Chem. Soc. Jpn. 1968, 41, 2220−2220. (42) Brorson, M.; Schaeffer, C. E. Orthonormal interelectronic repulsion operators in the parametrical Dq model. Application of the model to gaseous ions. Inorg. Chem. 1988, 27, 2522−2530. (43) Neuenschwander, K.; Guedel, H. U.; Collingwood, J. C.; Schatz, P. N. Electron-transfer transitions in hexachloroferrate(III). Singlecrystal absorption and MCD spectra. Inorg. Chem. 1983, 22, 1712− 1718. (44) Liu, B.; Wang, R.-P.; Glass, E. N.; Hill, C. L.; Cuk, T.; Okamoto, J.; Huang, D.-J.; van Schooneveld, M. M.; de Groot, F. M. F. Distorted Tetrahedral CoII in K5H[CoW12O40]·xH2O Probed by 2p3d Resonant Inelastic X-ray Scattering. Inorg. Chem. 2016, 55, 10152− 10160. (45) Maganas, D.; Kristiansen, P.; Duda, L.-C.; Knop-Gericke, A.; DeBeer, S.; Schlögl, R.; Neese, F. Combined Experimental and Ab Initio Multireference Configuration Interaction Study of the Resonant

Geometric and Electronic Structure/Function Correlations in NonHeme Iron Enzymes. Chem. Rev. 2000, 100, 235−350. (13) Ament, L. J. P.; van Veenendaal, M.; Devereaux, T. P.; Hill, J. P.; van den Brink, J. Resonant inelastic x-ray scattering studies of elementary excitations. Rev. Mod. Phys. 2011, 83, 705−767. (14) Butorin, S. M. Resonant inelastic X-ray scattering as a probe of optical scale excitations in strongly electron-correlated systems: quasilocalized view. J. Electron Spectrosc. Relat. Phenom. 2000, 110−111, 213−233. (15) Van Kuiken, B. E.; Hahn, A. W.; Maganas, D.; DeBeer, S. Measuring Spin-Allowed and Spin-Forbidden d−d Excitations in Vanadium Complexes with 2p3d Resonant Inelastic X-ray Scattering. Inorg. Chem. 2016, 55, 11497−11501. (16) Wernet, P.; Kunnus, K.; Josefsson, I.; Rajkovic, I.; Quevedo, W.; Beye, M.; Schreck, S.; Grubel, S.; Scholz, M.; Nordlund, D.; Zhang, W.; Hartsock, R. W.; Schlotter, W. F.; Turner, J. J.; Kennedy, B.; Hennies, F.; de Groot, F. M. F.; Gaffney, K. J.; Techert, S.; Odelius, M.; Fohlisch, A. Orbital-specific mapping of the ligand exchange dynamics of Fe(CO)5 in solution. Nature 2015, 520, 78−81. (17) Zhang, Y.; Wang, S.; Learmonth, T.; Plucinski, L.; Matsuura, A. Y.; Bernardis, S.; O’Donnell, C.; Downes, J. E.; Smith, K. E. Electronic excitations in vanadium oxide phthalocyanine studied via resonant soft X-ray emission and resonant inelastic X-ray scattering. Chem. Phys. Lett. 2005, 413, 95−99. (18) Golnak, R.; Bokarev, S. I.; Seidel, R.; Xiao, J.; Grell, G.; Atak, K.; Unger, I.; Thürmer, S.; Aziz, S. G.; Kühn, O.; Winter, B.; Aziz, E. F. Joint Analysis of Radiative and Non-Radiative Electronic Relaxation Upon X-ray Irradiation of Transition Metal Aqueous Solutions. Sci. Rep. 2016, 6, 24659. (19) Kunnus, K.; Zhang, W.; Delcey, M. G.; Pinjari, R. V.; Miedema, P. S.; Schreck, S.; Quevedo, W.; Schröder, H.; Föhlisch, A.; Gaffney, K. J.; Lundberg, M.; Odelius, M.; Wernet, P. Viewing the Valence Electronic Structure of Ferric and Ferrous Hexacyanide in Solution from the Fe and Cyanide Perspectives. J. Phys. Chem. B 2016, 120, 7182−7194. (20) James, B. D.; Bakalova, M.; Liesegang, J.; Reiff, W. M.; Hockless, D. C. R.; Skelton, B. W.; White, A. H. The hexachloroferrate(III) anion stabilized in hydrogen bonded packing arrangements. A comparison of the X-ray crystal structures and low temperature magnetism of tetrakis (methylammonium) hexachloroferrate (III) chloride (I) and tetrakis (hexamethylenediammonium) hexachloroferrate (III) tetrachloroferrate (III) tetrachloride (II). Inorg. Chim. Acta 1996, 247, 169−174. (21) Styczeń, E.; Pattek-Janczyk, A.; Gazda, M.; Józw ́ iak, W. K.; Wyrzykowski, D.; Warnke, Z. Thermal analysis of bis(tetraethylammonium) tetrachloroferrate(II). Thermochim. Acta 2008, 480, 30−34. (22) Han, J.; Nishihara, S.; Inoue, K.; Kurmoo, M. On the Nature of the Structural and Magnetic Phase Transitions in the Layered Perovskite-Like (CH3NH3)2[FeIICl4]. Inorg. Chem. 2014, 53, 2068−2075. (23) Sacchi, M.; Jaouen, N.; Popescu, H.; Gaudemer, R.; Tonnerre, J. M.; Chiuzbaian, S. G.; Hague, C. F.; Delmotte, A.; Dubuisson, J. M.; Cauchon, G.; Lagarde, B.; Polack, F. The SEXTANTS beamline at SOLEIL: a new facility for elastic, inelastic and coherent scattering of soft X-rays. J. Phys.: Conf. Ser. 2013, 425, 072018. (24) Chiuzbaian, S. G.; Hague, C. F.; Avila, A.; Delaunay, R.; Jaouen, N.; Sacchi, M.; Polack, F.; Thomasset, M.; Lagarde, B.; Nicolaou, A.; Brignolo, S.; Baumier, C.; Luning, J.; Mariot, J. M. Design and performance of AERHA, a high acceptance high resolution soft x-ray spectrometer. Rev. Sci. Instrum. 2014, 85, 043108. (25) Yamamoto, S.; Senba, Y.; Tanaka, T.; Ohashi, H.; Hirono, T.; Kimura, H.; Fujisawa, M.; Miyawaki, J.; Harasawa, A.; Seike, T.; Takahashi, S.; Nariyama, N.; Matsushita, T.; Takeuchi, M.; Ohata, T.; Furukawa, Y.; Takeshita, K.; Goto, S.; Harada, Y.; Shin, S.; Kitamura, H.; Kakizaki, A.; Oshima, M.; Matsuda, I. New soft X-ray beamline BL07LSU at SPring-8. J. Synchrotron Radiat. 2014, 21, 352−365. (26) van Schooneveld, M. M.; DeBeer, S. A close look at dose: Toward L-edge XAS spectral uniformity, dose quantification and 8210

DOI: 10.1021/acs.inorgchem.7b00940 Inorg. Chem. 2017, 56, 8203−8211

Article

Inorganic Chemistry Inelastic X-ray Scattering Spectrum of CO2. J. Phys. Chem. C 2014, 118, 20163−20175. (46) Baker, T. M.; Mako, T. L.; Vasilopoulos, A.; Li, B.; Byers, J. A.; Neidig, M. L. Magnetic Circular Dichroism and Density Functional Theory Studies of Iron(II)-Pincer Complexes: Insight into Electronic Structure and Bonding Effects of Pincer N-Heterocyclic Carbene Moieties. Organometallics 2016, 35, 3692−3700.

8211

DOI: 10.1021/acs.inorgchem.7b00940 Inorg. Chem. 2017, 56, 8203−8211