Article pubs.acs.org/JACS
Cite This: J. Am. Chem. Soc. XXXX, XXX, XXX−XXX
Energy-Degeneracy-Driven Covalency in Actinide Bonding Jing Su,† Enrique R. Batista,*,† Kevin S. Boland,† Sharon E. Bone,† Joseph A. Bradley,†,‡ Samantha K. Cary,† David L. Clark,*,† Steven D. Conradson,†,∥ Alex S. Ditter,†,‡ Nikolas Kaltsoyannis,⊥ Jason M. Keith,†,# Andrew Kerridge,△ Stosh A. Kozimor,*,† Matthias W. Löble,† Richard L. Martin,† Stefan G. Minasian,†,§ Veronika Mocko,† Henry S. La Pierre,† Gerald T. Seidler,‡ David K. Shuh,§ Marianne P. Wilkerson,† Laura E. Wolfsberg,† and Ping Yang*,† Downloaded via UNIV OF RHODE ISLAND on December 13, 2018 at 06:16:31 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.
†
Los Alamos National Laboratory, P.O. Box 1663, Los Alamos, New Mexico 87545, United States University of Washington, Seattle, Washington 98195, United States § Lawrence Berkeley National Laboratory, Berkeley, California 94720, United States ⊥ School of Chemistry, The University of Manchester, Oxford Road, Manchester M13 9PL, United Kingdom ‡
S Supporting Information *
ABSTRACT: Evaluating the nature of chemical bonding for actinide elements represents one of the most important and long-standing problems in actinide science. We directly address this challenge and contribute a Cl K-edge X-ray absorption spectroscopy and relativistic density functional theory study that quantitatively evaluates An−Cl covalency in AnCl62− (AnIV = Th, U, Np, Pu). The results showed significant mixing between Cl 3p- and AnIV 5fand 6d-orbitals (t1u*/t2u* and t2g*/eg*), with the 6d-orbitals showing more pronounced covalent bonding than the 5f-orbitals. Moving from Th to U, Np, and Pu markedly changed the amount of M−Cl orbital mixing, such that AnIV 6d- and Cl 3p-mixing decreased and metal 5f- and Cl 3p-orbital mixing increased across this series.
■
INTRODUCTION Covalency1 is a fundamental concept for rationalizing many chemical and physical phenomena. In contrast to transition delements where metal−ligand orbital mixing and overlap is well established, covalent bonding for the actinides (An) has been debated for decades. There is a body of evidence showingin certain systemssubstantial 5f- and 6d-covalency exists.2−20 These data are juxtaposed with numerous studies suggesting that An−ligand bonding is primarily ionic.21 The later conclusions are fueled by observables that include (but are not limited to) An−ligand bond distances being predictable using ionic radii and spectroscopic data, mostly for actinides in the +3 oxidation state, that show 5f → 5f optical transitions are marginally impacted by the ligand field. Note, in higher oxidation states appreciable spectroscopic changes are observed optically.9,12,22,23 Reconciling these observables and quantifying 5f- versus 6d-participation in covalent bonding represents a long-standing and important fundamental problem in actinide science. To directly address these challenges, we have characterized covalency for PuIV and the other early actinides (ThIV, UIV, NpIV) within an Oh−AnCl62− framework using ligand K-edge X-ray absorption spectroscopy (XAS)24 and spin−orbit density functional theory (DFT) transition dipole moment calculations.17,25−27 This duo has emerged as one of the most powerful approaches for characterizing metal−ligand covalency. Ligand K-edge XAS probes dipole-allowed bound-state transitions from the ligand 1s-orbitals to molecular orbitals that © XXXX American Chemical Society
contain ligand p-character resulting from metal−ligand orbital mixing. As a result, the transition intensity in the ligand K-edge XAS measurement is directly related to the metal−ligand orbital mixing coefficient. For several reasons, we elected to evaluate electronic structure and bonding in the highly symmetric AnCl 6 2− dianions. For many elements the octahedral metal complexes provide a foundation upon which understanding of metal−ligand bonding exist. The ubiquity of the Oh-MCl6x− structure also enables M−Cl covalency to be quantitatively evaluated across the periodic table as a function of the metal identity. Finally, within the AnCl62− platform 5f- and 6d-hybridization is formally forbidden, owing to selection rules governing octahedral symmetry. Hence, 5f- versus 6d-orbital mixing can be discretely probed. We observed that AnIV 6d-orbitals participated in covalent bonding to a larger extent than the 5f-orbitals and that 5f-orbital mixing increased across the actinide series (from Th to Pu). The results are presented in the context of an energy-degeneracy-driven covalency concept, which provides new insight into 5f- and 6d-electronic structure and bonding.
■
RESULTS AND DISCUSSION The Z2AnCl6 (An = Th, U, Np, Pu; Z = Me4N1+ and PPh41+),28,29 hereafter referred to as AnCl62−, were characReceived: September 11, 2018
A
DOI: 10.1021/jacs.8b09436 J. Am. Chem. Soc. XXXX, XXX, XXX−XXX
Article
Journal of the American Chemical Society terized by Cl K-edge XAS. The identity of the NMe41+ and PPh41+ cation had no substantial impact on the Cl K-edge XAS spectra. The Cl K-edge XAS spectra from AnCl62− (Figure 1)
edge features observed by ligand K-edge XAS spectroscopy. We remind the reader that the pre-edge transition intensities are directly related to the M−Cl mixing coefficient.24 Hence, the existence of pre-edge peaks in Cl K-edge XAS provided unambiguous evidence for covalency in the Np−Cl and Pu−Cl bonds. The series also highlighted diversity in electronic structure from Th to Pu. To quantify differences in An−Cl orbital mixing in AnCl62−, the spectra were deconvoluted. Curve fitting models (Table 1 and Figure 2) showed that peak
Figure 1. Normalized and background-subtracted Cl K-edge XAS spectra from AnCl62− (AnIV = Th, U, Np, Pu).
were similar in that they contained a sharp edge-peak at high energy (∼2828 eV) superimposed on an absorption threshold. The energies, intensities, and line shapes for these edge peaks were similar. For UCl62−, these data were consistent with previous reports,11 which highlighted the reproducibility of these measurements. Although the edge region of the Cl K-edge XAS spectra was primarily invariant upon moving from Th to U, Np, and Pu, substantial changes were observed in the pre-edge spectral region, 200 μm) of the low Cl tape prevented transmission to the Kapton backing. After shipping to the Stanford Synchrotron Radiation Lightsource (SSRL), the samples were
orbital overlap intergral (Table 2 and Figure S3), our results suggested that the extended AnIV 6d-orbitals were available for orbital mixing to a large extent than the more contracted AnIV 5f-orbitals. Hence, the coupling term (HML), which relates to the orbital overlap integral, seemed most important in directing AnIV 6d- vs 5f-orbital mixing with Cl 3p-orbitals. These observations were consistent with numerous accounts suggesting that limited 5f-orbital radial distributions cause small HML, which in turn can limit 5f-orbital participation in covalent bonding.7,13,38 The HML term decreases from Th to Pu, owing to the 5forbital radial density decrease (Figure 7). This decrease in orbital overlap should accompany decreased An−Cl orbital mixing. However, the experimental and computational results showed that AnIV 5f- and Cl 3p-orbital mixing increased when moving from ThIV to PuIV. We rationalized the observed increase by considering the denominator of eq 1 (E0M − E0L).11,39 For example, moving from Th to Pu decreased the energy of the 5f-orbitals and slightly increased the energy for the Cl 3p nonbonding orbitals (t1g) (Figure 4). The end result is a more favorable energy degeneracy orbital term (E0M − E0L) for the actinide 5f- and Cl 3p-orbitals. In this case, it seems that the positive effects from (E0M − E0L) outweigh negative impacts E
DOI: 10.1021/jacs.8b09436 J. Am. Chem. Soc. XXXX, XXX, XXX−XXX
Article
Journal of the American Chemical Society
with an exponent of 0.97 for Th and of 0.88 for U, Np, and Pu were removed to avoid negative Mulliken population of atomic orbitals in the molecular orbitals (MOs). The transition-dipole method (see below) was employed to calculate the core electron excitations at the experimental geometry of AnCl62−. The scalar relativistic effects and spin−orbit coupling effects were both taken into account by the zeroorder regular approximation (ZORA).53 The molecular orbital overlap matrix is printed using the subkey SFO with the arguments eig and ovl to the EPRINT key. Orbital overlap of the An−Cl bonds was printed using the input keyword TRANSFERINTEGRALS, and in the calculation the AnCl62− molecule was built from two fragments, i.e., An4+ and Cl66−. DFT Simulation of Cl K-Edge XANES Spectra. The Cl K-edge XANES spectra of AnCl62− were simulated as the Kohn−Sham orbital energy differences from SO-DFT/B3LYP calculations, i.e,. the energy difference between an occupied orbital and a virtual orbital of the ground state. For a specific core excitation, the oscillator strength was calculated using the transition-dipole approximation between this occupied MO and the virtual MOs. The Cl K-edge spectra were simulated by calculating core electron excitations originating from Cl 1s dominated MOs to virtual MOs at the experimental crystal structure. All other excitations from orbitals between the Cl 1s and the highest occupied MOs were excluded by restricting the energy range of the occupied orbitals involved in the excitation, so that only the excitations from the Cl 1s core level to virtual MOs were allowed. All the calculated transition intensities were evenly broadened with a Gaussian function of fwhm of 0.5 eV (i.e., peak width) to emulate the experimental spectra. The calculated spectra were shifted by a constant 43.0 eV, which aligned the experimental and calculated preedge peaks. The above DFT transition dipole moment method has been applied successfully in the simulation of ligand XAS and metal Ledge XANES spectra of actinide and lanthanide compounds, giving good agreement with experimental results.17,44 To verify the higher accuracy of the SO-DFT approach over SRTDDFT54 and SR-DFT methods in calculating the Cl K-edge XAS spectra of actinide compounds, we make a comparison of the simulation results among these three methods with the use of the B3LYP functional shown in Figure S2 in the Supporting Information. At the scalar relativistic level, the TDDFT method gives almost the same results as the DFT transition dipole method in simulating the Cl K-edge XAS spectra, particularly in the pre-edge energy range. As the AnCl62− (An = U, Np, and Pu) compounds have a 5f open-shell ground state, and spin−orbit coupling will mix the scalar relativistic 5f a2u, t2u, and t1u orbitals and decrease the orbital energy degeneracy as shown in the AmCl63− system,17 the ground-state electron occupations at the SO level greatly differ from those at the SR level. Furthermore, SO effects have significant influence on the transition probabilities and energies for transitions that involved the forbitals, which has also been demonstrated in the AmCl63− system.17 This is also evidenced by the comparison between SR and SO-DFT simulated spectra in Figure S2. The SO-DFT simulated spectra show an excellent agreement with the experimental results as shown in Figure 5. That is to say, compared to SR-TDDFT and SR-DFT methods, the SO-DFT method greatly improves the agreement with experimental spectra in terms of peak position and intensity. That is why the SO-DFT method was employed in this work.
inserted into the LANL Tender X-ray chamber. The chamber was separated from the beam pipe by a beryllium window and continually flushed with He gas. Sample fluorescence was monitored using a partially depleted series (PIPS) charged particle detector with a 5000 mm2 active area (PD5000−75−500AM). Sample fluorescence was monitored against the incident radiation (I0) that was measured using an ionization chamber (30 cm) downstream from the beryllium window through which He gas was continually flowed. The X-ray absorption spectra were measured at two synchrotrons. Experiments conducted at SSRL were made under dedicated operating conditions (3.0 GeV, 500 mA) at Beamline 4-3. This beamline was a wiggler side-station equipped with a liquid nitrogen double-crystal Si(111) monochromator. The beam was collimated and unfocused to allow for high-energy-resolution measurements on homogeneous samples. The crystals were run detuned by 30% at 3200 eV to minimize harmonics. Spectra were collected with 1 mm vertical slits and 5 mm horizontal slits. At SSRL radiation damage experiments were conducted by repeatedly collecting spectra rapidly (acquisition time of 10 s). Next, the spectra were reproduced with 15 min data acquisition times. No evidence of spectral change was observed over the course of 2 h. Data collected at the Institute for Nuclear Waste Disposal (INE) beamline for Actinide Research at the Angströmquelle Karlsruhe (ANKA) synchrotron were obtained under operating conditions of 2.5 GeV and 120 mA. The beamline was equipped with a 1.5 T bending magnet and a water-cooled Si(111) doublecrystal monochromator. The crystals were detuned by 30% from the rocking curve maximum. Two mirrors were utilized. The first mirror was vertically collimating, and the second mirror was vertically and horizontally focusing. For all Cl K-edge measurements, energy calibrations were conducted externally using the maxima of the first pre-edge features in Cs2CuCl4 (2820.20 eV).24 Spectra were initially obtained at SSRLreproduced at SSRL during multiple experimental scheduling periodsand at ANKA. The pre-edge intensities were reproducible between 3% and 5%. CL K-Edge XAS Data Analysis. Data manipulation and analysis was conducted as previously described by Solomon and co-workers.24 Data were analyzed by fitting a line to the pre-edge region, 2702.1− 2815.0 eV, which was subsequently subtracted from the experimental data to eliminate the background of the spectrum. The data were normalized by fitting a first-order polynomial to the post-edge region of the spectrum (2836−3030 eV) and setting the edge jump at 2836 eV to an intensity of 1.0. This normalization procedure gave spectra normalized to a single Cl atom or M−Cl bond. A deconvoluted model for the Cl K-edge XAS data was obtained using a modified version of EDG_FIT47 in IGOR 6.0. The pre-edge regions (
