High Frequency Fe–H and Fe–H₂ Modes in a trans-Fe(η²-H₂)(H) Complex – A Speed Record for Nuclear Resonance Vibrational Spectroscopy

Abstract

Nuclear resonance vibrational spectroscopy (NRVS) and density functional theory (DFT) are complementary tools for studying the vibrational and geometric structure of specific isotopically-labeled molecular systems. Here we apply NRVS and DFT to characterize the trans-[⁵⁷Fe(η²-H₂)(H)(dppe)₂][BPh₄] complex. Heretofore, most NRVS observations have centered on the spectral region below 1000 cm⁻¹ where the ⁵⁷Fe signal is strongest. In this work, we show that state-of-the-art synchrotron facilities can extend the observable region to 2000 cm⁻¹ and likely beyond, in measurements that require less than one day. The ⁵⁷Fe–H stretch was revealed at 1915 cm⁻¹, along with the asymmetric ⁵⁷Fe–H₂ stretch at 1774 cm⁻¹. For a small fraction of the H₂-dissociated product, the ⁵⁷Fe–H stretch was detected at 1956 cm⁻¹. The unique sensitivity to ⁵⁷Fe motion and the isolated nature of the Fe–H/H₂ stretching modes enabled NRVS to quantitatively analyze the sample composition.

Graphical Abstract

High-energy vibrations of iron-bound hydride and molecular hydrogen were observed using synchrotron-based ⁵⁷Fe NRVS technique, assisted with ²H isotope labeling. The spectral bands from a mixed sample were rationalized by DFT calculations.

Since the discovery of the first transition metal dihydrogen complexes,¹ a rich chemistry has been developed for the production and interconversion of dihydrogen and hydride species.² This information is relevant for the development of better catalysts for hydrogen conversion,³ improved hydrogen storage media,⁴ and also for an understanding of hydrogenase enzymes.^5–7 However, characterization of metal-bound dihydrogen and hydride complexes remains a substantial challenge.^8–10 Due to their low electron-scattering cross-section, hydrogen atoms remain challenging to observe by X-ray crystallography, unless very high resolution data is available.¹¹ M(etal)–H and M–H₂ stretching modes are often quite weak in infrared (IR) spectra. Although they can sometimes be observed in concentrated materials, they are too weak to be seen in dilute enzymes via IR. Raman spectroscopy can sometimes be useful, but both metal hydride and dihydrogen are susceptible to photolysis.¹² Neutron diffraction (ND) and inelastic neutron scattering (INS) are powerful tools for hydrogen studies, but they require relatively large quantities of material (~1 mmol of hydrogen), and the latter is also subject to interferences from protons in the solvent or elsewhere in the sample.

Nuclear resonance vibrational spectroscopy (NRVS)^13–16 has become a popular technique for elucidating the element-selective normal modes of appropriate Mössbauer isotopes.^13–18 In previous work, we have shown that ⁵⁷FeNRVS can be utilized to observe Fe–H bending and Fe–H–Fe(Ni) wagging modes in various model compounds¹⁹ as well as intermediate species of [NiFe]²⁰ and [FeFe] hydrogenases.^21–23 However, to date the only reported Fe–H stretching modes were the bands around 1450 cm⁻¹ for a doubly-bridging LFe(μ-H)₂FeL complex²⁴ and at 1468/1532 cm⁻¹ in a singly-bridging LNi(μ-H)FeL complex.²⁰ To better gauge the feasibility for observing even higher frequency stretching modes expected for a Fe-terminal hydride and dihydrogen, we have applied NRVS to examine a well-studied trans-[⁵⁷Fe(η²-H₂)(H)(dppe)₂][BPh₄] complex,^{25, 26} hereafter called 1_H^H2. As shown in Figure 1, our DFT modeling of 1_H^H2 produced an optimized structure that overlaps well with the X-ray diffraction result.²⁶ Further we prepared the 1_H^D2 isotopologue with diatomic deuterium, which undergoes facile intramolecular exchange with the hydride yielding 1_D^HD, and then re-exchanges with molecular deuterium during a two week incubation time under the D₂ atmosphere, producing 1_D^D2.26, 27 Although a series of alternative deuterium isotopologues can result from this transformation (the sample collectively called 1-D), it nonetheless enables us to qualitatively identify any modes pertaining to the H and H₂ of the H-Fe-H₂ moiety relative to the non-deuterated NRVS sample (the sample called 1-H). These samples were synthesized according to procedures described in the Supporting Information.

Figure 1.

Molecular X-ray structure26 of 1_H^H2 (in black, with the BPh₄– counterion shown in green) superimposed with its optimized DFT model (colored elements). The H-to-D exchangeable hydrogens of the model structure are shown in light blue, and the rest of the hydrogens are omitted for clarity.

Along with the H–H stretching mode (ν^⇄), it is common to assign five normal modes for a metal complex with a side-on (η²) dihydrogen ligand, involving symmetric (ν^⇈) and asymmetric (ν^⇅) stretching, in-plane (δ) and out-of-plane (δ′) bending, and rotation about the M-H₂ axis (τ^↻).²⁸ For 1_H^H2 these Fe–H₂ modes will be split by coupling with Fe–H motions, as shown in Chart 1 for a minimal C_2v symmetry DFT model 1_H^H2-C_2v. Of course, a complete analysis needs to consider interactions with the dppe ligands. The modes illustrated in Chart 1 nevertheless provide a good starting point for interpretation of our NRVS spectra.

Chart 1.

Normal modes expected for a trans coordination of dihydrogen and hydride ligands in an idealized Fe(II) octahedral complex. The 9 modes are based on C_2v-symmetry trans-[Fe(η²-H₂)(H)(PH₃)₄]⁺ = 1_H^H2-C_2v model calculations as described in the Supporting Information and shown in Figure S1. Rows top to bottom: arrow-style mode representations; calculated frequencies (cm⁻¹); C_2v character names; arrow-based character names presently employed for 1_H^H2 and its derivatives.

Solid state NRVS spectra for samples 1-H vs. 1-D in the high-frequency range are compared with each other and with corresponding DFT ⁵⁷Fe-PVDOS simulations in Figure 2a,b. All of the modes expected for 1_H^H2 in this region are readily assigned (Table S1), as well as bands attributed to dissociation (loss of H₂ forming 1_H^*) or substitution (exchange of H₂ by N₂ called 1_H^N2) by-products, see Figures S2 and S6–S7 for their DFT structures and individual spectra.^{26, 27} Starting at the low-frequency end, an isotopically sensitive band at 1053 cm⁻¹ can be assigned to symmetric Fe–H₂ stretching (ν^⇈), see arrow-style representations in Figure 2; animations showing DFT-calculated normal modes with significant (>10%) either ⁵⁷Fe or Fe-bound proton PVDOS characters are available as part of the Supporting Information.

Figure 2.

(a) Experimental NRVS data for high-energy (1000–2000 cm⁻¹) bands of samples 1-H and 1-D; (b) collective DFT ⁵⁷Fe-PVDOS simulations for 1-H including 52% 1_H^H2, 26% 1_H^*, and 22% 1_H^N2, and 1-D using equivalent contributions from 1_D^D2, 1_D^*, and 1_D^N2; (c) DFT stick H(+H₂)-PVDOS intensities for the Fe ligand hydride (and dihydrogen in 1_H^H2) nuclei in the three color-coded species of 1-H, weighted with their sample composition factors. The composition (%) of the mixed sample was obtained based on the relative intensities of the experimental spectra bands in the 1000–2000 cm⁻¹ region, with individual contributions from the three pure calculated species shown in Figures S3 and S6–S7. Top: predicted atomic motions from DFT simulations for normal modes with H(+H₂)-PVDOS>10%. The mode-to-band associations are highlighted by broken lines.

There is a shoulder in the data at ~1080 cm⁻¹, identified as ν^⇈ mixed with ligand Ph–H bending. The corresponding asymmetric Fe–H₂ stretching (ν^⇅) bands are seen at much higher energies – 1774 and 1276 cm⁻¹ in 1_H^H2 and 1_D^D2 respectively. The Fe–H stretching (ν_↑) band for 1_H^H2 is clearly observed in the spectrum at 1915 cm⁻¹. The modes from 1350–1410 cm⁻¹ in 1-D are the ν_↑-modes of the possible deuterium isotopologues and their decay products 1_D^D2, 1_D^*, and 1_D^N2. There is a weaker feature at ~1956 cm⁻¹ for the 1-H sample, consistent with 1_H^*. An admixture of 1_H^N2 possibly results in a shoulder at ~1890 cm⁻¹, yet our calculations indicate that both ν_↑(1_H^H2) and ν_↑(1_H^N2) modes contribute to the same 1915 cm⁻¹ band. Here we note that the ν_↑(1_H^*) band at 1956 cm⁻¹ represents the highest observed NRVS frequency to date, and this augurs well for future studies aimed at characterization of iron-hydride interactions. Further, it is important to note that the NRVS specificity for Fe motion, high energy, and isolation of Fe–H_x modes affords a strong handle on the quantitative composition of our NRVS spectrum with respect to adventitious subspecies 1_H^* and 1_H^N2, that were anticipated by the IR data and supported by its DFT simulation (Figures S4 and S5). Finally, we note that compared to the IR spectrum, identification of hydrogen-based modes at these energies in NRVS (Figure 2a) is unambiguous.

In Figure 3 we present NRVS spectra for 1-H vs. 1-D in the range below 1000 cm⁻¹, again compared with each other and with DFT simulations. As mentioned above, the Fe–H₂ bending modes are split by coupling with Fe–H motions in 1_H^H2. Thus, different frequencies and intensities are observed for in-plane (δ) and out-of-plane (δ′) Fe–H₂ bending depending on whether the associated Fe–H bending is in-phase (_←^⇇) or out-of-phase (_→^⇇), as depicted in Chart 1. Starting from high energy, an isotope-sensitive feature is observed at 823 cm⁻¹ in 1-H. Our DFT calculation for 1_H^H2 indicate a coupled in-plane in-phase Fe–H₂ and Fe–H bending (δ_←^⇇) mode at 821 cm⁻¹, which has sufficient ⁵⁷Fe motion to be observable by NRVS. The experimental feature at 733 cm⁻¹ in 1-H is consistent with Fe–H₂ and Fe–H out-of-plane in-phase bending (δ′_←^⇇) motion. In our calculations this feature is derived from four modes with similar character predicted in the 726–732 cm⁻¹ range, but with varying degrees of coupling to the outer dppe ligands. Due to this coupling with the outer coordination shell, these modes are sensitive to the packing of the solid-state environment which may account for the slight shoulder at 728 cm⁻¹ in the experimental spectrum of 1-H. The next two isotopically-sensitive features at 558 and 584 cm⁻¹ in 1-H are identified as H–Fe–H₂ out-of-phase bending modes, out-of-plane (δ′_→^⇇) and in-plane (δ_→^⇇) respectively.

Figure 3.

(a) Experimental NRVS data for low-energy (0–1000 cm⁻¹) bands of samples 1-H and 1-D; (b) DFT ⁵⁷Fe-PVDOS simulations for 1-H and 1-D; (c) DFT stick H(+H₂)-PVDOS intensities in the three color-coded species of 1-H, weighted with their sample composition factors. The ×4 intensity insets in (a-b) display spectra >370 cm⁻¹. Top: predicted atomic motions for selected normal modes of 1_HH2 with H(+H₂)-PVDOS>10%. Mode-to-band associations are highlighted by broken lines. Further figure details follow the legend of Figure 2.

The most prominent NRVS feature is located at 284/273 cm⁻¹ for 1-H/D (Figure 3a,b), with its intensity approximately 100 times higher (Figure S6a) than that of the high-energy Fe–H stretching band at 1915 cm⁻¹. This global maximum band results from vibrations reminiscent of the intense transaxial bending modes of 6-coordinate Fe complexes with a four-fold axis of symmetry.^{29, 30} In all the calculated species, the DFT modes that produce the most intense ~250–300 cm⁻¹ region reveal entire [H/D-Fe-(H₂/D₂)/(N₂)] axial fragment displacements either parallel to the equatorial plane formed by the Fe-P bonds, or perpendicular to this plane, illustrated by animations in the Supporting Information. The high intensity and lack of splitting of the 284 mode in 1_H^H2 (and 278 mode in 1_D^D2) indicate the transaxial ligands (H₂, H) and slight distortion of the P-Fe-P angle in the Fe-H₂ plane have no resolvable effect on this vibrational mode.

Our analysis indicates that modes with significant contribution (>10%) from the H^– and H₂ ligands motion do not supply 1-H spectral intensities at energies below ~570 cm⁻¹ (Figure S6a), while the corresponding low-energy limit for the D⁻ and D₂ ligands in ⁵⁷Fe-PVDOS of 1-D is ~260 cm⁻¹ (Figure S7a). For example, the strongest mode of rotational ν^↻(1_H^H2) character is predicted at 384 cm⁻¹ but conveys only negligible Fe motion (Figures 3 and S6b). However, at 392 cm⁻¹ there is a satellite mode calculated with similar, but much weaker H₂ rotational character, that contains ⁵⁷Fe-PVDOS and agrees well with the shoulder at the same position in the experimental spectrum of 1-H. Inspection of the 392 cm⁻¹ normal mode displacements indicates it acquires intensity through coupling with equatorial Fe–P motion. The nearby isotopically-sensitive intensity in 1-H at 316, 332 (shoulder), and 361 cm⁻¹ are calculated at 314, 333, and 364 cm⁻¹ containing varying Fe–P stretching motions and, similarly to the 284 cm⁻¹ mode, near-concerted translation of the H-Fe-H₂ moiety within the Fe-P plane. The second greatest intensity in the spectrum is at 259 cm⁻¹ in 1-H, and it is primarily correlated with P–Fe–P out-of-plane bending motion where H-Fe-H₂ is collectively translating through the equatorial plane.

In summary, we have characterized trans-[⁵⁷Fe(η²-H₂)(H)(dppe)₂][BPh₄] with ⁵⁷Fe NRVS combined with DFT calculations. The Fe-H stretching mode in the compound (and the H₂-loss byproduct) represent the highest energy vibrational features ever observed by NRVS. Moreover, our combined approach developed a complete picture of the H-Fe-H₂ vibrational structure even in the congested low energy regime. This implies that hydrogen motion in this low energy region can serve as a surrogate probe in systems that are not yet capable of being explored with high energy NRVS. There are clear accessible applications to the study of hydrogen storage materials - for example, Fe-containing metal–organic frameworks (MOFs) where the presence of framework hydrogen limits neutron-based approaches.³¹ Extension of this work to dilute biological systems is challenging, but with continued improvements in synchrotron radiation X-ray production it will eventually be possible to characterize Fe-H/H₂ vibrational modes in enzyme intermediates.