Correlating Valence and 2p3d RIXS Spectroscopies: A Ligand-Field Study of Spin-Crossover Iron(II)

The molecular spin-crossover phenomenon between high-spin (HS) and low-spin (LS) states is a promising route to next-generation information storage, sensing applications, and molecular spintronics. Spin-crossover complexes also provide a unique opportunity to study the ligand field (LF) properties of a system in both HS and LS states while maintaining the same ligand environment. Presently, we employ complementing valence and core-level spectroscopic methods to probe the electronic excited-state manifolds of the spin-crossover complex [FeII(H2B(pz)2)2phen]0. Light-induced excited spin-state trapping (LIESST) at liquid He temperatures is exploited to characterize magnetic and spectroscopic properties of the photoinduced HS state using SQUID magnetometry and magnetic circular dichroism spectroscopy. In parallel, Fe 2p3d RIXS spectroscopy is employed to examine the ΔS = 0, 1 excited LF states. These experimental studies are combined with state-of-the-art CASSCF/NEVPT2 and CASCI/NEVPT2 calculations characterizing the ground and LF excited states. Analysis of the acquired LF information further supports the notion that the spin-crossover of [FeII(H2B(pz)2)2phen]0 is asymmetric, evidenced by a decrease in eπ in the LS state. The results demonstrate the power of cross-correlating spectroscopic techniques with high and low LF information content to make accurate excited-state assignments, as well as the current capabilities of ab initio theory in interpreting these electronic properties.


Geometric Structure and Angular Overlap Considerations
The [Fe II (H2B(pz)2)2phen] 0 complex (1) is pseudo-Oh in geometry, with a strict C2 symmetry (Figure 1).All 6 N-ligands are sp 2 hybridized, being part of conjugated, delocalized rings.Therefore, all ligands participate in 1) s-donating, and 2) poopdonating (oop = out-of-plane) interactions with Fe, while there are no significant pip interactions (ip = in-plane).As a result, although seemingly pseudo-Oh, a C2 symmetric model is necessary to accurately simulate the electronic structure of this complex.In this model, the minimum descriptive geometry of 1, ligating N can be divided into 3 types, and summarized by 9 angles (Scheme 1).Scheme 1. Descriptive angles a of 1. a Under C2 symmetry, the coordination environment around Fe can be described in terms of 3 unique ligand atoms, each with a unique angle relative to the C2 axis (labeled qn, n = 1-3, shown in a).Each of these ligands also interacts asymmetrically with the central Fe, requiring further distinction by twist angle, jn (b), and torsion angle, yn (c).As j3 = 0, only j1 and j2 are shown.Axes x, y, z are provided along bonds in the pseudo-Oh coordinate frame.
To determine the ligand field parameters of 1, an Oh approximation was made by using q1,2,3 = 45°, 90°, 135°; j1,2,3 = 0°, 90°, 0°; y1,2,3 = 90°, 45°, 0°.While each of the six coordinating atoms of Fe can each interact through a sigma interaction with the dx2-y2 and dz2 orbitals, calculated as 6 s-interactions/2 orbitals = 3s.Meanwhile, significant p-bonding interactions between Fe and the surrounding ligands will only form between the out-of-plane p-orbitals of the H2B(pz) and phen ligands and the dxy, dxz, and dyz ligands, providing 6 p-interactions/3 orbitals = 2p.Therefore, the ligand field splitting parameter of 1 is defined as 10Dq = 3es -2ep.Using this definition together with the assigned LF states in Table 4 of the main text, AOMX was used to fit parameters 10Dq, es, and ep, along with Racah parameters B and C.    Experimental data are provided as circles, and the corresponding fit as solid lines.Spectra were collected over a range of 0-10 T at temperatures of 2 K (black), 5 K (red), 10 K (blue), and 20 K (green).Corresponding ZFS fitting parameters are summarized in the main text, Table 1.

It has previously been observed that LIESST can occur in both directions
(converting LS to HS, and HS back to LS) depending on the energy of the incident radiation utilized. 3While irradiation of the LS state in the energetic range of the 3 T1 and 1 T2 states (typically at 500-700 nm) induces trapping of the HS state at low temperature, irradiation in the range of the 3 T1 and 5 E (800-1000 nm) of the HS state has been also shown to induce reformation of the LS state. 3MCD measurements of 1 prior to irradiation at 5 K, 5 T still exhibit considerable spectral intensity, despite not containing any trapped HS complex as demonstrated via 57 Fe Mössbauer, Figure S1.
Further monitoring of MCD signal intensity as a function of time exposed to the MCD light source itself at 640 nm demonstrated that the incident beam was sufficiently intense to induce LIESST (Figure S5).Therefore, it is possible that the intensity of the ~11,700 cm -1 feature, assigned to the 5 T ® 5 E transition, is partially diminished by the probing radiation of the MCD spectrometer.

Figure S1 . 2 Figure S3 .
Figure S1.57Fe Mössbauer spectrum (1.6 K, 0 T) of 1 prepared as a polysiloxane mull (circles, black) along with corresponding fit (red, dashed) using a single doublet with d = 0.51 mm/s, DEQ = 0.38 mm/s, and an asymmetric linewidth of 0.31 mm/s (1.09 asymmetry factor).Residual shown above in blue, solid.These results are consistent with previously published data of LS 1.1

Figure S4 .
Figure S4.MCD signal dependence on sample irradiation by fiber optic (red) and the MCD light source itself (black).Measurements were performed at -5 T, 5 K.

Figure S5 .
Figure S5.Temperature dependent MCD of HS 1 imbedded in a polysiloxane mull.Spectra were collected at an applied field of 5 T. The high-spin state was generated by exposure of the sample to light irradiation (640 nm) until signal intensity at 590 nm was saturated.Re-exposure to light irradiation was performed between individual scans of field and temperature.Poorer S/N is observed in the 8000-13000 cm -1 region as this constitutes the overlap of the respective very low and very high energy regimes of the PMT and InGaAs detectors employed.

Figure S6 .
Figure S6.VTVH-MCD saturation curves of 1 collected at 590 nm/16,950 cm -1 .Experimental data are provided as circles, and the corresponding fit as solid lines.Spectra were collected over a range of 0-10 T at temperatures of 2 K (black), 5 K (red), 10 K (blue), and 20 K (green).Corresponding ZFS fitting parameters are summarized in the main text, Table1.

Figure S7 .
Figure S7.(top) Irradiation-dependent Fe L2,3-edge spectrum of 1. Spectra indicated by Nx were collected on a single sample spot N number of times.The top spectrum is the pure HS spectrum collected at 220 K, and the pure LS spectrum collect at 65 K with additional Co filters is shown on bottom.(bottom) Comparison of RIXS collected at 220 K (solid lines) with that collected at 15 K under equivalent scanning/beam attenuation conditions (dotted lines).

Figure S8 .
Figure S8.(top) Fe L3-edge of HS 1 (black) compared with that calculated at the SA-CASCI/NEVPT2(22,13) level (red).An energy shift of -0.7 eV is applied to the calculated spectrum.A dashed vertical line indicates the incident energy used to collect the spectra in the bottom graph.(bottom) Experimental (black) vs. SA-CASCI/NEVPT2(22,13) calculated (red) Fe 2p3d RIXS at an incidence energy of 706.25 eV.The calculated Fe 2p3d RIXS represents the calculated spectrum at a 706.95 eV incident energy.

Figure S9 .
Figure S9.(top) Fe L3-edge of HS 1 (black) compared with that calculated at the SA-CASCI/NEVPT2(22,13) level (red).An energy shift of -0.7 eV is applied to the calculated spectrum.A dashed vertical line indicates the incident energy used to collect the spectra in the bottom graph.(bottom) Experimental (black) vs. SA-CASCI/NEVPT2(22,13) calculated (red) Fe 2p3d RIXS at an incidence energy of 707.25 eV.The calculated Fe 2p3d RIXS represents the calculated spectrum at a 707.95 eV incident energy.

Figure S10 .
Figure S10.(top) Fe L3-edge of HS 1 (black) compared with that calculated at the SA-CASCI/NEVPT2(22,13) level (red).An energy shift of -0.7 eV is applied to the calculated spectrum.A dashed vertical line indicates the incident energy used to collect the spectra in the bottom graph.(bottom) Experimental (black) vs. SA-CASCI/NEVPT2(22,13) calculated (red) Fe 2p3d RIXS at an incidence energy of 708.25 eV.The calculated Fe 2p3d RIXS represents the calculated spectrum at a 708.95 eV incident energy.

Figure S11 .
Figure S11.(top) Fe L3-edge of HS 1 (black) compared with that calculated at the SA-CASCI/NEVPT2(22,13) level (red).An energy shift of -0.7 eV is applied to the calculated spectrum.A dashed vertical line indicates the incident energy used to collect the spectra in the bottom graph.(bottom) Experimental (black) vs. SA-CASCI/NEVPT2(22,13) calculated (red) Fe 2p3d RIXS at an incidence energy of 709.25 eV.The calculated Fe 2p3d RIXS represents the calculated spectrum at a 709.95 eV incident energy.

Figure S12 .
Figure S12.(top) Fe L3-edge of LS 1 (black) compared with that calculated at the SA-CASCI/NEVPT2(22,13) level (red).An energy shift of -0.55 eV is applied to the calculated spectrum.A dashed vertical line indicates the incident energy used to collect the spectra in the bottom graph.(bottom) Experimental (black) vs. SA-CASCI/NEVPT2(22,13) calculated (red) Fe 2p3d RIXS at an incidence energy of 708.25 eV.The calculated Fe 2p3d RIXS represents the calculated spectrum at a 708.8 eV incident energy.

Figure S13 .
Figure S13.(top) Fe L3-edge of LS 1 (black) compared with that calculated at the SA-CASCI/NEVPT2(22,13) level (red).An energy shift of -0.55 eV is applied to the calculated spectrum.A dashed vertical line indicates the incident energy used to collect the spectra in the bottom graph.(bottom) Experimental (black) vs. SA-CASCI/NEVPT2(22,13) calculated (red) Fe 2p3d RIXS at an incidence energy of 709.25 eV.The calculated Fe 2p3d RIXS represents the calculated spectrum at a 709.8 eV incident energy.

Figure S14 .
Figure S14.(top) Fe L3-edge of LS 1 (black) compared with that calculated at the SA-CASCI/NEVPT2(22,13) level (red).An energy shift of -0.55 eV is applied to the calculated spectrum.A dashed vertical line indicates the incident energy used to collect the spectra in the bottom graph.(bottom) Experimental (black) vs. SA-CASCI/NEVPT2(22,13) calculated (red) Fe 2p3d RIXS at an incidence energy of 709.25 eV.The calculated Fe 2p3d RIXS represents the calculated spectrum at a 709.8 eV incident energy.

Figure S15 .
Figure S15.Pseudo-Voigtian band deconvolution analysis of the RIXS of 1 collected at 200 K. Experimental data are plotted in black, individual fit bands in blue (dotted), dark red (dash dot), and green (solid), and the sum of the fit in red (dashed).Incident energies used for the collection are indicated in the upper right corner of each individual spectra.Standard deviations (s) for each spectrum are provided as a shaded gray area.

Figure S16 .
Figure S16.Pseudo-Voigtian band deconvolution analysis of the RIXS of 1 collected at 50 K.Experimental data are plotted in black, individual fit bands in blue (dotted), dark red (dash dot), and green (solid), and the sum of the fit in red (dashed).Incident energies used for the collection are indicated in the upper right corner of each individual spectra.Standard deviations (s) for each spectrum are provided as a shaded gray area.