Ruthenium 4d-to-2p X-ray Emission Spectroscopy: A Simultaneous Probe of the Metal and the Bound Ligands.

Ruthenium 4d-to-2p X-ray emission spectroscopy (XES) was systematically explored for a series of Ru2+ and Ru3+ species. Complementary density functional theory calculations were utilized to allow for a detailed assignment of the experimental spectra. The studied complexes have a range of different coordination spheres, which allows the influence of the ligand donor/acceptor properties on the spectra to be assessed. Similarly, the contributions of the site symmetry and the oxidation state of the metal were analyzed. Because the 4d-to-2p emission lines are dipole-allowed, the spectral features are intense. Furthermore, in contrast with K- or L-edge X-ray absorption of 4d transition metals, which probe the unoccupied levels, the observed 4p-to-2p XES arises from electrons in filled-ligand- and filled-metal-based orbitals, thus providing simultaneous access to the ligand and metal contributions to bonding. As such, 4d-to-2p XES should be a promising tool for the study of a wide range of 4d transition-metal compounds.

Introduction

n class="Chemical">Ruthenium is a 4d class="Chemical">n class="Chemical">transition metal (TM) that is found in formal oxidation states ranging from +8 to −2. It is a relatively rare element, being found at only 100 parts per trillion in the earth’s crust. Despite its scarcity, however, ruthenium has found use in numerous areas of research, spanning photochemistry and photophysics,[1,2] catalysis (including water oxidation,[3,4] metathesis,[5] and asymmetric hydrogenation of ketones[6]), and bioinorganic chemistry (including anticancer drugs[7,8] and as model species for heme metalloenzymes[9,10]).

To rationalize stn class="Chemical">ructure–fuclass="Chemical">nctioclass="Chemical">n relatioclass="Chemical">ns or to declass="Chemical">n class="Chemical">sign more efficient catalysis, a detailed understanding of the geometric and electronic structure of Ru and its coordination environment is highly desirable. In this regard, X-ray spectroscopic techniques are very attractive as element-selective probes of a metal of interest. For Ru, K-edge X-ray absorption spectroscopy (XAS) has seen a wide range of applications.[11−16] However, because of the intrinsic spectral broadening resulting from the short Ru 1s core-hole lifetime,[17] the Ru K-edge spectra are typically broad and rarely exhibit well-defined 1s-to-4d pre-edge features. Nevertheless, complemented with other techniques or computational tools, Ru K-edge XAS allows the characterization of Ru species under both ex situ(11−16) and in situ(18−20) conditions. Alternatively, one can utilize Ru L-edge XAS, which, due to the dipole-allowed nature of the 2p-to-4d transitions, provides a more direct probe of the empty 4d orbitals localized on the Ru.[16,21−26] As such, the Ru L-edge is a useful probe of the oxidation state and, to some extent, the site symmetry. To obtain information about the ligand environment, one can utilize the Ru extended X-ray absorption fine structure (EXAFS) (most typically at the Ru K-edge) to extract information about the ligands surrounding Ru and their distances. However, because EXAFS is a scattering-based method, similar mass atoms, such as C, N, and O (or P, S, and Cl), cannot be distinguished.

It is here that n class="Chemical">Ru X-ray emisclass="Chemical">n class="Chemical">sion spectroscopy (XES) in the valence region is potentially a very promising tool. In an XES experiment, one monitors the fluorescence that occurs when an electron from a higher energy level refills a core hole on the photoabsorber. These emission lines are referred to as K-emission lines when an electron fills a 1s core hole and L-emission lines when an electron refills a 2s/2p core hole. The former has been relatively intensely studied for 3d TMs, and limited applications for 4d TMs have been reported.[69−29] The weak valence-to-1s (or valence-to-core, VtC) XES region, which results from an electron in a ligand valence orbital refilling a 1s core hole at the metal, has been shown to provide a valuable probe of the ligand identity (C,N,O),[30−32] the protonation state (O2– vs OH–),[29,33−35] and the degree of bond activation (e.g., peroxo vs superoxo).[36−39] However, because valence-to-1s XES transitions gain intensity through a small amount of metal p mixing into the ligand orbitals (imparting the transition with dipole-allowed character),[27] the spectral transitions are intrinsically weak and, in most cases, provide little direct insight into the filled metal d orbitals (because d to s transitions are formally dipole-forbidden).

The ability to directly probe the filled ligand and the filled n class="Chemical">metal d orbitals is clearly of great iclass="Chemical">nterest iclass="Chemical">n obtaiclass="Chemical">niclass="Chemical">ng detailed iclass="Chemical">nformatioclass="Chemical">n about the boclass="Chemical">ndiclass="Chemical">ng, the coordiclass="Chemical">natioclass="Chemical">n eclass="Chemical">nvclass="Chemical">n class="Chemical">ironment, and the oxidation state of the metal site in the same experimental spectrum. This could be achieved by following 3d-to-2p or 4d-to-2p XES, for which the transitions are all dipole-allowed. Interestingly, however, valence to 2p XES lines have been relatively unexplored for both 3d or 4d TMs.[40] In the case of 3d TMs, this is largely due to the very low probability of a fluorescence event occurring in this energy regime, thus making nonresonant XES prohibitively challenging and instead requiring a resonant XES (or RIXS) measurement. For 4d TMs, the probability of a d-to-p fluorescence decay occurring is significantly increased relative to 3d TMs, making nonresonant 4d-to-p XES more readily accessible. Furthermore, because the valence region is composed of both filled ligands and, at higher energy, filled d orbitals, one can, in principle, map out the d character in both the filled ligand and filled metal d orbitals. To our knowledge, however, no systematic exploration of these spectra has been made. Herein we systematically explore the chemical information content of Ru 4d-to-2p XES spectra (also known as Lβ2 lines in the physics literature) for a series of molecular Ru complexes.

The chosen series of complexes includes n class="Chemical">RuII species, where the coordiclass="Chemical">naticlass="Chemical">ng ligaclass="Chemical">nds raclass="Chemical">nge from a σ doclass="Chemical">nor iclass="Chemical">n [class="Chemical">n class="Chemical">Ru(tacn)], to a π donor in [Ru(DMAP)], to the usually considered π acceptor[41] in [Ru(CN)] and a π acceptor in [Ru(bpy)] (Figure ). An oxidized analogue with a σ-donor ligand environment [Ru(NH)] was included to evaluate the effect in the 4d-to-2p transitions upon changing the oxidation state. Finally, a less symmetric example, [Ru(dmso)Cl], was utilized to examine the effects of lowering the site symmetry.

Figure 1

Structures of the investigated ruthenium model complexes.

Stn class="Chemical">ructures of the iclass="Chemical">nvestigated class="Chemical">n class="Chemical">ruthenium model complexes.

The experimental spectra are correlated to denn class="Chemical">sity fuclass="Chemical">nctioclass="Chemical">nal theory (DFT) calclass="Chemical">n class="Chemical">culated spectra. The strong agreement between experiment and theory grants access for clear assignments of the observed spectral features to be made. Importantly, we demonstrate that Ru 4d-to-2p XES allows for the filled ligand and metal 4d orbitals to be simultaneously probed, providing access to the nature of the ligand, the oxidation state of the metal, and the covalency between the metal and the ligand. We thus believe that 4d-to-2p XES has great potential for investigating the electronic structure of Ru-containing materials as well as all other 4d TMs.

Methods

Materials

n class="Chemical">Potassium hexacyanoruthenate(II) hydrate class="Chemical">n class="Chemical">K4[Ru(CN)6]·xH2O, hexaammineruthenium(III) chloride [RuIII(NH3)6]Cl3, and [RuII(bpy)3](PF6)2 (bpy = 2,2′-bipyridine) were purchased from Sigma-Aldrich and used without further purification. cis-RuII(dmso)4Cl2 (dmso = dimethyl sulfoxide) and [RuII(DMAP)6]Cl2 (DMAP = 4-dimethylaminopyridine) were prepared according to a published procedures.[42,43] [RuII(tacn)2](PF6)2 (tacn = 1,4,7-triazacyclononane) was synthesized following the reported procedure for [RuII(tacn)2](OTf)2,[44] employing the standard Ru precursor cis-RuII(dmso)4Cl2[42] and replacing the precipitating agent with NH4PF6.

XES Measurements

All samples were measured in the solid state at room temperature. The pure solids were ground to a fine powder and packed into 1 mm thick n class="Chemical">aluminum sample holders. The irradiated class="Chemical">n class="Chemical">side of the cell was covered with 8 μm Kapton film to maximize the detected X-ray photons. In the case of [Ru(bpy)3](PF6)2, an 0.8 μm aluminum window covered with 8 μm polypropylene was used to attenuate a beam-induced visible fluorescent signal which otherwise masked the desired L-emission lines.

Preliminary experiments were done at the XAS beamline P64 at the PETRA III storage ring, capable of delivering an 8 × 1012 ph/s photon beam at 5 keV incident energy in a 40 μm × 200 μm (V×H) spot by means of a n class="Chemical">Si(111) moclass="Chemical">nochromator. The ficlass="Chemical">nal measuremeclass="Chemical">nts (aclass="Chemical">nd therefore all data showclass="Chemical">n here) were performed at the receclass="Chemical">ntly coclass="Chemical">nstclass="Chemical">n class="Chemical">ructed PINK tender X-ray beamline at BESSY II. Currently, the PINK beamline is operated in commissioning mode. A considerable gain in fluorescence signal could be obtained by using a multilayer monochromator (∼100 eV band pass), which provided 2 × 1013 photons/s and a lower excitation energy of 4 keV. The beam size on the sample was 30 μm × 500 μm (V×H).

All spectra were collected un class="Chemical">siclass="Chemical">ng aclass="Chemical">n iclass="Chemical">n-house-declass="Chemical">n class="Chemical">signed energy-dispersive vacuum spectrometer (10–5 mbar internal pressure) based on von Hamos geometry with a vacuum sample chamber environment (5 mbar working pressure). The analyzer was set up in a vertical dispersion direction, taking advantage of the small vertical beam size to improve the energy resolution. A cylindrically bent 25 mm × 100 mm α-quartz (101̅2) crystal (d = 2.282 Å) with a bending radius of R = 250 mm dispersed incoming fluorescence radiation onto a vacuum-compatible charge-coupled device (CCD) camera with a 26 μm × 26 μm pixel size. The 26.6 mm × 7 mm detector chip accepted an energy window of 2810–2850 eV (Bragg angles of θ = 75.3–72.6°).

Calibration of the energy scale was achieved by measuring reference X-ray emisn class="Chemical">sioclass="Chemical">n liclass="Chemical">nes iclass="Chemical">n the same iclass="Chemical">nstclass="Chemical">n class="Chemical">rumental configuration. The employed sample holder allowed for three samples as well as two references to be mounted for energy calibration: Pd foil and KCl powder. To establish the energy scale, the Pd Lα1, Pd Lα2, and Cl Kβ13 emission lines were calibrated to 2838.61, 2833.29, and 2815.6 eV, respectively.[45] Energies were translated into Bragg angles, and a linear fit was applied.

Damage scans were performed for all samples. Data corresponding to undamaged samples were selected through the comparison of sequential partial sum spectra. In all cases, the data were collected in continuous motion, employing the speed and the total irradiation dose per pass, as determined by the damage scans. The employed speed varied from 200 to 500 μm·s–1, and the total exposure was between 4 and 8 s per spot. With a longer exposure time, the spectra of n class="Chemical">[Ru(NH3)6]Cl3 showed class="Chemical">n class="Chemical">signs of damage, whereas no change in the spectra was seen for the other studied complexes for exposure times up to 200 s.

Theoretical Calculations

All caln class="Chemical">culatioclass="Chemical">ns were performed with the ORCA quaclass="Chemical">ntum chemistry software package.[46,47] For all class="Chemical">n class="Chemical">ruthenium species, the calculations employed DFT with the B3LYP[48−51] hybrid functional and the “old-zora-TZVP” recontracted scalar relativistic all-electron basis set, as well as the relativistic approximation ZORA. An auxiliary basis set for Coulomb fitting was employed: SARC/J.[52] The auxiliary basis set was automatically generated by the keyword AutoAux.[53] A tight self-consistent field (SCF) convergence threshold was chosen. An increased grid was used during the SCF iterations (Grid4), and no grid was used in the final energy evaluation after SCF convergence. To enhance the SCF convergence procedure, a combined method was utilized, employing an alternative algorithm to the direct inversion in iterative subspace (DIIS), KDIIS, and when the calculation was close to convergence, the program switched to a second-order SCF (SOSCF).[54,55] The conductor-like polarizable continuum model (CPCM) was used for charge compensation in all calculations. In particular, during the geometry optimizations, a more detailed cavity was specified to ensure that a minimal stationary point was reached, tested by computing the vibrational spectra. To this end, the number of points around each atom used to define the cavity with the GEPOL algorithm was explicitly defined to be 110 Lebedev grid points.[56] XES calculations were performed utilizing previously established protocols,[70] with optimized structures and the standard CPCM model. For [RuII(DMAP)6]2+, the crystallographic geometry was used for the XES calculations.[43] For the XES calculations, the same functional, basis set, and SCF convergence methods as those in the geometry optimizations were used. To include the spin–orbit in the calculations, the keywords CoreOrbSOC and DoSOC true were employed. A 2.2 eV broadening was applied to the calculated spectra, in agreement with the natural broadening of the tabulated L3 lines.[17] A constant 26.5 eV shift in energy was applied to the DFT calculated spectra. The molecular orbitals (MOs) were visualized with Chimera.[57]

For the comparative calclass="Chemical">culatioclass="Chemical">ns performed oclass="Chemical">n class="Chemical">n class="Chemical">iron complexes, all parameters were the same as those for the ruthenium species, except that in this case the def2-TZVP basis set was employed.[52]

Results and Discussion

Experimental and Calculated 4d-to-2p XES Spectra

The normalized 4d-to-2p XES spectra for the investigated series of n class="Chemical">Ru model complexes are showclass="Chemical">n iclass="Chemical">n Figure , where the experimeclass="Chemical">ntal spectra are located iclass="Chemical">n the left paclass="Chemical">nel aclass="Chemical">nd the DFT calclass="Chemical">n class="Chemical">culated spectra are shown in the right panel. All of the species bearing a RuII center are characterized by an intense main peak at ∼2837 eV (Table ) and the presence of less intense bands at lower energies of ∼2830–2833 eV, which are variable in number and overall shape depending on the species. The XES spectra of [Ru(tacn)] and [Ru(DMAP)] both consist of only one band at lower energy, although the spectra of [Ru(bpy)] and [Ru(CN)] show two features in this energy range, particularly well-defined for the latter. Additionally, for the one-electron oxidized complex, [Ru(NH)], the main peak splits into two, separated by ∼1.3 eV, whereas the band at lower energies shifts up in energy (by ∼0.4 to 0.6 eV) relative to the RuII complexes. To better understand the origins of the observed features, DFT calculations were utilized to calculate the spectra (Figure B,D). The calculated spectra are in good agreement with the experimental data and reproduce the general trends in both energy and intensity distributions. This thus allows us to use the computations to obtain further insight into the origins of the observed spectral features. In the sections that follow, we use a stepwise approach to provide a pedagogical deconvolution of the spectra.

Figure 2

Comparison between (A,C) experimental and (B,D) DFT calculated Ru 4d-to-2p XES spectra for [Ru(tacn)] (red), [Ru(DMAP)] (pink), [Ru(CN)] (blue), [Ru(bpy)] (gray), [Ru(NH)] (green) and (dmso)Cl] (orange). Both experimental and calculated spectra are normalized to the area. A broadening of 2.2 eV and an energy shift of 26.5 eV were applied to the calculated spectra.

Table 1

Energies Extracted from the Features Present in the 4d-to-2p XES Spectra

complex	main feature (eV)	less intense features (eV)
RuII Complexes
[RuII(tacn)2](PF6)2	2837.4	2833.2
[RuII(DMAP)6]Cl2	2837.8
[RuII(bpy)3](PF6)2	2837.6	2831.2, 2833.2
[RuII(CN)6]K4	2837.4	2830.9, 2833.0
cis-RuII(dmso)4Cl2	2837.2
RuIII Complex
[RuIII(NH3)6]Cl3	2837.1, 2838.4	2833.6

Comparison between (A,C) experimental and (B,D) DFT caln class="Chemical">culated class="Chemical">n class="Chemical">Ru 4d-to-2p XES spectra for [Ru(tacn)] (red), [Ru(DMAP)] (pink), [Ru(CN)] (blue), [Ru(bpy)] (gray), [Ru(NH)] (green) and (dmso)Cl] (orange). Both experimental and calculated spectra are normalized to the area. A broadening of 2.2 eV and an energy shift of 26.5 eV were applied to the calculated spectra.

We note that the following disn class="Chemical">cusclass="Chemical">n class="Chemical">sion focuses on a simple one-electron picture that can be interpreted with the aid of DFT calculations. This approach is sufficient for analyzing and assigning the main features of the XES spectra, as previously demonstrated for 3d TMs.[30−34,36−39,58] It should be considered, however, that the calculations do not capture the weak satellite features above ∼2840 eV (Figure ). These features likely arise from multielectron excitations, which have a higher probability due to the employed incident excitation energies, which are well above the 2p ionization threshold. We refer the interested reader to references (59−62) for a more detailed discussion of multielectron excitations.

Influence of the Ligand Framework

To understand the influence of the ligands on the n class="Chemical">Ru 4d-to-2p XES spectra, it is useful to begiclass="Chemical">n by calclass="Chemical">n class="Chemical">culating the 4d-to-2p XES of a hypothetical singlet (S = 0) Ru2+ atom in the absence of any ligands (Figure ). In this case, only a single feature at ∼2838 eV is observed in the XES spectrum, which corresponds to dipole-allowed transitions from the filled Ru 4d orbitals to the 2p core hole.

Figure 3

DFT calculated 4d-to-2p XES spectra for Ru2+(S = 0) (dark yellow) and [Ru(tacn)] (black). The spectra are not normalized.

DFT calclass="Chemical">culated 4d-to-2p XES spectra for class="Chemical">n class="Chemical">Ru2+(S = 0) (dark yellow) and [Ru(tacn)] (black). The spectra are not normalized.

Upon adding the σ-only donating n class="Chemical">tacn ligaclass="Chemical">nds, oclass="Chemical">ne observes a maiclass="Chemical">n feature at ∼2837 eV aclass="Chemical">nd a class="Chemical">new feature at ∼2833 eV. The maiclass="Chemical">n feature correspoclass="Chemical">nds to traclass="Chemical">nclass="Chemical">n class="Chemical">sitions from the filled t2g orbitals of [Ru(tacn)] to the 2p core hole, as was the case for the isolated Ru2+ atom. However, both the energy and the intensity of this feature have decreased. The loss of intensity reflects a decrease in the percent of 4d character in the donor t2g orbitals, which has been reduced from 100% in the isolated Ru2+ atom to ∼85–91% in [Ru(tacn)] due to covalent mixing with the ligands. This is an interesting observation for a σ-only donor such as tacn, where generally the t2g covalency is assumed to be negligibly small[63,64] due to the fact that the metal t2g orbitals are nonbonding with the ligand orbitals in rigorous O symmetry. The ∼6–9% tacn contribution to the filled t2g orbitals thus reflects the mixing that becomes allowed due to deviations from ideal symmetry. The decrease in the energy of the ∼2837 eV feature results from the net decrease in the effective charge on Ru upon complex formation, which destabilizes the 2p core orbital to a greater extent than the 4d orbitals, resulting in a transition at lower energy. This is observable by inspection of the respective energies of the 2p and 4d orbitals for both considered systems. Whereas upon coordination the 2p orbitals are destabilized at ∼8.8 eV due to the presence of the new point charges of the ligand environment, the 4d orbitals show, on average, a destabilization of only ∼6.4 eV upon coordination.

Finally, the trann class="Chemical">sitioclass="Chemical">n at ∼2833 eV may be asclass="Chemical">n class="Chemical">signed as transitions arising from filled tacn orbitals, which σ-interact with the empty set of Ru eg orbitals. Hence, the intensity of this feature reflects the extent of Ru 4d character in the filled tacn orbitals, which amounts to ∼20%. These findings are also illustrated qualitatively in the MO diagram in Figure . Thus these results highlight that 4d-to-2p XES simultaneously probes both filled-metal- and filled-ligand-based orbitals. Because the selection rule is based on a dipole-allowed d-to-p transition, these spectra provide a far more direct probe of metal–ligand bonding than valence-to-1s (or VtC) XES, where the intensity mechanism is derived from dipole-allowed p-to-s transitions and hence in most cases provides no direct information on the mixing of metal nd orbitals with the ligands.[31]

Figure 4

Molecular orbital analysis derived from the orbital origin of the bands (left). Qualitative molecular orbital diagram for [Ru(tacn)] (right).

Moleclass="Chemical">cular orbital aclass="Chemical">nalyclass="Chemical">n class="Chemical">sis derived from the orbital origin of the bands (left). Qualitative molecular orbital diagram for [Ru(tacn)] (right).

To more quantitatively understand the intenn class="Chemical">sity mechaclass="Chemical">nism of the observed class="Chemical">n class="Chemical">Ru 4d-to-2p spectral features for a σ-only donor ligand, we have also performed a series of hypothetical calculations in which the Ru–N(tacn) bond distance is systematically varied through the symmetric breathing vibrational mode of the species (Figure ).

Figure 5

Calculated spectra for [Ru(tacn)] and structures obtained along its symmetric breathing vibrational mode, drawing the ligand environment closer to and further away from the metal. The Ru–N average bond length is shown for each structure, being 2.17 Å for the relaxed optimized structure. The spectra are not normalized.

Caln class="Chemical">culated spectra for [class="Chemical">n class="Chemical">Ru(tacn)] and structures obtained along its symmetric breathing vibrational mode, drawing the ligand environment closer to and further away from the metal. The Ru–N average bond length is shown for each structure, being 2.17 Å for the relaxed optimized structure. The spectra are not normalized.

The intenn class="Chemical">sity of both features chaclass="Chemical">nges upoclass="Chemical">n the variatioclass="Chemical">n of the class="Chemical">n class="Chemical">Ru–ligand distances. In particular, as the ligand is drawn away from the metallic center along its symmetric breathing vibrational mode, the intensity of the spectrum increases. This reflects the loss of covalent mixing of the metallic orbitals with the ligands, slightly increasing the 4d character in the MOs that generate the transitions. As a result, the features gain intensity, showing that the 4d character in the originating MOs drives the intensity mechanism of the features. It should be noted that the 4d-to-2p XES spectra are rather sensitive to the average Ru–N bond length, with the intensity of the spectra increasing 23% over only a 1.4% increase in bond length and decreasing 9% over a 0.9% decrease in bond length. Additionally, although more clear in the σ-bonding eg(σ) feature, both features show a shift in energy to higher values upon elongation of the metal–ligand bond lengths. This behavior arises from the destabilization of the contributing MOs upon the loss of covalent mixing with ligand-based orbitals.

Having analyzed the pure σ-donor ligand case, a more complicated case can be tackled, where both σ and π contributions need to be conn class="Chemical">sidered. Iclass="Chemical">n this respect, a kclass="Chemical">nowclass="Chemical">n σ- aclass="Chemical">nd π-doclass="Chemical">naticlass="Chemical">ng ligaclass="Chemical">nd is employed: class="Chemical">n class="Chemical">DMAP,[43] affording the octahedral [Ru(DMAP)] species. The 4d-to-2p experimental XES spectrum of this species (Figure A) shares, in principle, the characteristics of the spectrum of [Ru(tacn)]. In the same way as previously done, the origin of the features can be analyzed in combination with DFT calculations (Figure ). From a closer analysis of the DFT-calculated spectrum, two components in the main feature can be detected, labeled A and B in Figure (top).

Figure 6

Molecular orbital analysis derived from the orbital origin of the bands (top). Qualitative molecular orbital diagram for [Ru(DMAP)] (bottom).

Moleclass="Chemical">cular orbital aclass="Chemical">nalyclass="Chemical">n class="Chemical">sis derived from the orbital origin of the bands (top). Qualitative molecular orbital diagram for [Ru(DMAP)] (bottom).

The caln class="Chemical">culatioclass="Chemical">n shows that the maiclass="Chemical">n feature arises from class="Chemical">n class="Chemical">t2g orbitals, which are now partially mixed with π orbitals of the DMAP ligands, originating two combinations: bonding and antibonding. The highest energy MOs arise from the antibonding t2g–π(DMAP) combination, and at deeper binding energy (by ∼1.3 eV), the bonding combination appears. The two features are, however, not resolvable within the experimental resolution. Additionally, at lower energies and with less intensity, the ligand-based eg(σ) bonding interaction, originating from the mixing of Ru 4d eg and σ(DMAP) orbitals, is observed.

In contrast with the previously analyzed spectra, the spectn class="Chemical">rum of [class="Chemical">n class="Chemical">Ru(CN)] shows two distinct features at lower energies (Table ). Owing to the diatomic nature of the ligands, the complexity of the spectrum increases. With the aid of DFT calculations, the origin of the observed features can be interpreted. The main feature arises from t2g orbitals that appear to be antibonding or nonbonding (Figure ), in contrast with the classical picture of CN– as a π-acceptor. With less intensity and lower energy, feature B is found to be composed of two components: a σ-bonding interaction of eg orbitals with σ(CN–) orbitals (B1) and a π-bonding interaction of t2g and π(CN–) orbitals (B2). At lower energies than B, feature C originates from a σ-bonding interaction of eg orbitals with σ(CN–) orbitals.

Figure 7

Molecular orbital analysis derived from the orbital origin of the bands for [Ru(CN)].

Moleclass="Chemical">cular orbital aclass="Chemical">nalyclass="Chemical">n class="Chemical">sis derived from the orbital origin of the bands for [Ru(CN)].

As previously noted, the resulting MO diagram appears to contradict the traditional bonding picture, where n class="Chemical">cyanide is coclass="Chemical">nclass="Chemical">n class="Chemical">sidered to be a strong π-acceptor ligand. In general, it is expected that the metal t2g orbitals interact with the CN– via a strong π-back-bonding interaction, which is not present in the calculated MOs (Figure ). If a stepwise construction of the diagram is considered, then, first, only the π interactions can be addressed (Figure ). Both π(CN–) and π*(CN–) can interact with the filled t2g orbitals, affording a bonding combination, which is mainly π(CN–) in character, an antibonding combination, which is mainly π*(CN–) in character, and an intermediate in energy combination, which is mainly t2g in character. Because of the interplay of the energies of these different orbitals, the CN– can be considered as having more π-acceptor or π-donor character, and in this case, the t2g orbitals are effectively nonbonding, with some mixing with CN– orbitals, where the π-donor character predominates. We note that the non-π-acceptor character of CN– has already been reported for other hexacyanide metal complexes.[41] In addition to the π interactions, because of the σ-donor character of CN–, there is also mixing of the eg orbitals with CN– orbitals with the appropriate symmetry, namely, σ*(2s–2s) and σ(2p–2p), originating three eg(σ) MOs.

Figure 8

Qualitative molecular orbital diagram for CN– (left) and [Ru(CN)] (right).

Qualitative moleclass="Chemical">cular orbital diagram for Cclass="Chemical">n class="Chemical">N– (left) and [Ru(CN)] (right).

The last complex to analyze in the series of n class="Chemical">RuII species is [class="Chemical">n class="Chemical">Ru(bpy)] (Figure A,B). The spectrum of this species is similar in the number of features and energies to that of [Ru(CN)]. The relevant point group to consider in this case is D3, where, first, the octahedral symmetry is lifted and a small trigonal perturbation appears.[65] In this point group, the 4d orbitals split into one a1 and two e sets. The DFT analysis shows that the main feature of the spectrum originates from a1 and e orbitals, which are mostly 4d in character (Figure ). The lower energy features comes from different MOs from the e set that possess Ru 4d character and both π and σ mixing with ligand orbitals.

Figure 9

Molecular orbital analysis derived from the orbital origin of the bands for [Ru(bpy)].

Moleclass="Chemical">cular orbital aclass="Chemical">nalyclass="Chemical">n class="Chemical">sis derived from the orbital origin of the bands for [Ru(bpy)].

The derived MO diagram for [n class="Chemical">Ru(bpy)] is iclass="Chemical">n agreemeclass="Chemical">nt with previously reported diagrams.[65,66]

Oxidation State

In addition to the series of n class="Chemical">RuII species, a oclass="Chemical">ne-electroclass="Chemical">n oxidized complex, [class="Chemical">n class="Chemical">Ru(NH)], was studied to assess the influence of the oxidation state of the photoabsorber in the L-emission lines. Because of the nature of the amino ligands, a similar electronic structure, and therefore similar spectral features, is expected for [Ru(NH)] and [Ru(tacn)], analyzed previously, since they both possess a σ-only ligand environment. As shown in Figure C and Table , the spectrum of [Ru(NH)] is composed of a main feature at ∼2838 eV and a less intense feature at ∼2834 eV. However, the spectrum of [Ru(NH)] is characterized by broadening of the main feature, arising from transitions from the filled t2g orbitals, which splits into two components at 2837.1 and 2838.4 eV. This difference can be interpreted with the aid of DFT calculations (Figure and Figure S1).

The overlay of the experimental and caln class="Chemical">culated spectra for [class="Chemical">n class="Chemical">Ru(NH)] in Figure S1 shows a correlation between the broadening of the experimental feature and the formation of two distinct groups of calculated transitions separated by ∼ 1.3 eV. Furthermore, Voigt deconvolution of the spectrum clearly shows two components in the main peak (Figure S6), whereas only one is present for the studied RuII species (Figures S2–S5). To rationalize these differences, a closer inspection of the orbital origin of the features is required. Figure shows the MOs involved in each region of the spectrum.

Figure 10

Molecular orbital analysis derived from the orbital origin of the bands for [Ru(NH)].

Moleclass="Chemical">cular orbital aclass="Chemical">nalyclass="Chemical">n class="Chemical">sis derived from the orbital origin of the bands for [Ru(NH)].

For the open-shell 4d5[n class="Chemical">Ru(class="Chemical">n class="Chemical">NH)] species, the t2g orbitals are partially filled. Unlike the previously explored low-spin d6 configurations, in the case of the open-shell low-spin d5 configuration, there is significant spin-polarization contribution that accounts for the separation of the valence orbitals into separate spin orbitals with different spatial densities α and β. As a result, different emission channels contribute to the spectrum, and the energy separation between them accounts for the broadening of the spectrum. Although this does not accurately capture the true multiplet structure, previous studies have shown that a simple DFT approach can roughly approximate the expected splitting.[67] What becomes clear from Figure and the energy ordering of the spin orbitals is that the emission from the spin orbital 48a generates the 2837.1 eV component of the main feature, whereas the transitions from the other four filled t2g spin orbitals (48a, 49a, 49b, and 50a) generate a group of transitions closer in energy that give rise to the 2838.4 eV feature. As a result, the broadening of the main peak reduces the intensity of the feature, changing the intensity ratio between the main feature A and the less intense C, which arises from the bonding σ combination of the eg and σ orbitals from the ligands. Furthermore, the d5 configuration, in comparison with the d6 configuration, possesses one fewer allowed emission event, which is responsible for an overall reduction of the intensity of feature A. The oxidation state influence in the A/C intensity ratio can also be clearly distinguished in the overlap of the DFT calculated spectra for [Ru(NH)] and its reduced analog [Ru(NH)] (Figure S7).

Site Symmetry

For the (pseudo) octahedral or symmetric species analyzed here so far, the electronic stn class="Chemical">ructure of the complexes caclass="Chemical">n be ratioclass="Chemical">nalized iclass="Chemical">n terms of MO diagrams with the combiclass="Chemical">natioclass="Chemical">n of experimeclass="Chemical">ntal XES L-emisclass="Chemical">n class="Chemical">sion lines and DFT calculated spectra. Regardless of the oxidation state of the photoabsorber or the nature of the bound ligands, the spectra consist of a main feature (originating from the filled t2g orbitals) and one or two less intense features ∼4 eV lower in energy. Nevertheless, this is not the case anymore for the asymmetric [Ru(dmso)Cl] (Figure C). The loss of symmetry and the inclusion of different ligating atoms in the ligand environment have strong effects in the spectral profile of the species. The enhanced mixing of orbitals results in less defined features. The main feature and those originating from mostly ligand-based orbitals overlap, resulting in a broad feature with less structure. The main orbitals from which the emission lines originate are shown in Figure .

Figure 11

Orbital origin of the bands for (dmso)Cl].

Orbital origin of the bands for (n class="Chemical">dmso)Cl].

Comparison with VtC 1s

The n class="Chemical">Ru 4d-to-2p XES allows the uclass="Chemical">nique posclass="Chemical">n class="Chemical">sibility to simultaneously probe both metal- and ligand-based orbitals. This is in contrast with the better known valence-to-1s (VtC) XES, which dominantly originates from metal p character mixing into ligand-based orbitals. Whereas collecting VtC XES can be challenging due to the weakness of the lines, the dipole-allowed nature of the 4d-to-2p XES lines greatly enhances the observed spectral intensities. A comparison of calculated emission spectra (Figure ) shows that the 4d-to-2p spectrum for [RuII(CN)6]4– presents ∼30 times increased intensity with respect to the VtC spectrum of [FeII(CN)6]4–.

Figure 12

Comparison of intensities of the calculated spectra of [Ru(CN)] (left) and [Fe(CN)] (right).

Comparison of intenclass="Chemical">sities of the calclass="Chemical">n class="Chemical">culated spectra of [Ru(CN)] (left) and [Fe(CN)] (right).

Hence, 4d-to-2p XES spectroscopy holds promise as a unique tool to investigate n class="Chemical">Ru species. Of particlass="Chemical">n class="Chemical">cular interest is the possibility of studying Ru-based catalysts, widely spread in diverse areas of chemistry.[5,6,68]

Conclusions

4d-to-2p XES is shown to be a promin class="Chemical">siclass="Chemical">ng tool to achieve detailed iclass="Chemical">nclass="Chemical">n class="Chemical">sight into the electronic structure of Ru complexes. Because of the dipole-allowed nature of the lines, intense emission lines arise. The spectra of symmetric octahedral species consist of a main feature at ∼2837 eV, originating from the filled t2g orbitals, and two or fewer less intense feature(s) at a lower energy of ∼2833 eV. Whereas the main feature is characteristic of the oxidation state of the metal center, the features at lower energies and lower intensity are dependent on the ligand identity. DFT calculations complemented the analysis and allowed for detailed insight into the electronic structure origins of the observed spectral features. A unique characteristic of 4d-to-2p XES is the ability to observe dipole-allowed transitions arising from both filled-np-ligand- and filled-4d-metal-based orbitals. The energies and intensities of the features are a direct reflection of the strength of metal–ligand bonding via covalency. As such, 4d-to-2p XES should serve as a very useful electronic structure probe for Ru and other 4d TMs.