research papers
Tris-amidoximate uranyl complexes via η2 binding mode coordinated in aqueous solution shown by X-ray absorption spectroscopy and density functional theory methods
aShanghai Institute of Applied Physics, Chinese Academy of Sciences, Shanghai 201800, People's Republic of China, and bSchool for Radiological and Interdisciplinary Sciences (RAD-X), Soochow University, Suzhou 215123, People's Republic of China
*Correspondence e-mail: zhanglinjuan@sinap.ac.cn, wangjianqiang@sinap.ac.cn
The present study sheds some light on the long-standing debate concerning the coordination properties between uranyl ions and the amidoxime ligand, which is a key ingredient for achieving efficient extraction of uranium. Using X-ray absorption fine structure combined with theoretical simulation methods, the binding mode and bonding nature of a uranyl–amidoxime complex in aqueous solution were determined for the first time. The results show that in a highly concentrated amidoxime solution the preferred binding mode between UO22+ and the amidoxime ligand is η2 coordination with tris-amidoximate species. In such a uranyl–amidoximate complex with η2 binding motif, strong covalent interaction and orbital between U 5f/6d and (N, O) 2p should be responsible for the excellent binding ability of the amidoximate ligand to uranyl. The study was performed directly in aqueous solution to avoid the possible binding mode differences caused by crystallization of a single-crystal sample. This work also is an example of the simultaneous study of local structure and electronic structure in solution systems using combined diagnostic tools.
Keywords: uranyl speciation; X-ray absorption near-edge structure; XANES; extended X-ray absorption fine structure; EXAFS; local coordination structure; amidoxime ligand; DFT calculations.
1. Introduction
Extracting uranium from seawater, which contains ∼4.5 billion tons of U (Davies et al., 1964), is an effective way to meet the increasing demand for as advanced nuclear energy systems continue to develop. Extraction using adsorbent materials is a promising method for sequestering considerably low concentrations of uranium in seawater. Over 200 functionalized polymers exist as adsorbents; among them, the amidoxime-functionalized polymers are highly selective towards uranium (Schenk et al., 1982; Astheimer et al., 1983; Seko et al., 2004; Kim et al., 2013; Witte et al., 1984; Das et al., 2009; Ting et al., 2013; Vukovic & Hay, 2013) and have achieved capacities as high as 3.3 g uranium kg−1 adsorbent (Kim et al., 2014). Despite more than three decades of investigation, the only insight into how the amidoxime ligand binds to the uranyl ions in an aqueous environment has been accomplished mainly through computational investigations. In aqueous solution, the absence of long-range ordering rules out the diffraction method, while other typical spectroscopies such as IR provide limited structural information. Raman spectra can give useful information for investigating the molecules and chemical bonding in solution (Liu et al., 2017; McGrail et al., 2014), but cannot determine the local coordination structure. The identification of the coordination environment of the uranyl complex is critical, especially in aqueous solution, to understand the uranium extraction mechanism from seawater and even design an organic functionalized adsorbent under non-equilibrium conditions.
The coordination mode between amidoxime and UO22+ has been studied extensively, and three possible binding motifs have been reported as shown in Fig. 1. Witte firstly observed a monodentate oxygen-binding mode in a uranyl–amidoxime crystal (Fig. 1a), in which uranium is bound to the oxygen atoms of two amidoxime and two amidoximate ligands (Witte et al., 1984). Zhang and Katragadda suggested a chelating interaction (Fig. 1b) (Zhang et al., 2005; Katragadda et al., 1997), illustrating that U coordinates the oxygen atom of NOH and the nitrogen atom of NH2, but until now no representative single crystal has been reported. The η2 binding motif, which corresponds to the uranium bound to the oxygen and nitrogen atoms of the amidoxime ligand, was recently predicted to be the most stable structure on the basis of density functional theory (DFT) calculations and crystallography (Fig. 1c) (Barber et al., 2012; Vukovic et al., 2012; Guo, Wang et al., 2015; Wang et al., 2014). Using the available studies, several issues must be noted. First, most of the recent studies were based on the DFT method and direct experimental evidence remains relatively rare. The few existing experimental data were obtained through single-crystal X-ray structural analysis in non-aqueous media. Second, theoretical calculation is helpful in exploring the coordination structure of uranyl complexes in an aqueous environment. Almost all prior theoretical reports predicted that the η2 binding motif is thermodynamically preferred in aqueous media (Vukovic et al., 2012; Guo, Wang et al., 2015; Wang et al., 2014); however, a five-coordinate 1:2 uranyl complex with mixed coordination of two ligands is reported to be more stable than a pure η2 binding motif after applying a correction of the coupled cluster theory with simple, double and perturbative triple excitations [CCSD(T)] (Abney et al., 2016). Shi and his colleague also reported monodentate and η2 coordination are the main binding modes for the amidoxime group in the presence of carbonate ligand (Wang et al., 2014). Third, the recent experimental results are mainly based on a single-crystal phase in a non-aqueous environment, where the single-crystal structure may be affected by many factors, such as the preparation conditions and the solvent medium. However, the structures obtained in non-aqueous media, although highly informative, are less relevant to the uranyl complexes in aqueous solution. The coordination structure of uranyl complexes can be affected by the concentration and temperature of the solution (Bailey et al., 2005; Gückel et al., 2013; Hennig et al., 2007). In aqueous solution, the participation of OH− or H2O has a great impact on the coordination configurations of uranyl (Neuefeind et al., 2004). Puckered-hexagon configurations of OH− or H2O about the uranyl group will tend to revert to plane-pentagon coordination (Evans, 1963). Additional water molecules can cause the change from bidentate to monodentate modes due to different proton transfer and thus lead to different electronic properties (Jackson et al., 2013). The relative complexation ability of ligands and water determines the uranyl species when strongly coordinating ions are present, and water may even be excluded entirely from the first coordination sphere. Thus, direct experimental determination of the coordination structure of uranyl complexes in the aqueous environment is essential. Unfortunately, the local structure in solution is difficult to obtain and thus there is great demand for an experimental technique combining an element-specific probe technique and accurate analysis. X-ray absorption spectroscopy is such a tool, and has been successfully applied in exploring the coordination structure in a solution.
The present work accurately investigated the coordination property of a UO22+–amidoxime complex in aqueous solution through X-ray absorption fine structure (XAFS) along with DFT calculations. Considering the extremely low U concentration in seawater, in the local space of the uranyl the dose of amidoxime is relatively high and thus the present research was performed in a highly concentrated amidoxime system. As a comparison, the uranyl–HClO4 solution was considered, which is believed to be pure hydrate complexes due to the weak coordinating ability of perchlorate ions (Sémon et al., 2001). Combined with the theoretical simulation, quantitative analysis is a powerful tool to investigate the local structural information. In this combined diagnostic method, DFT calculations were performed to optimize and identify the possible spatial arrangement from various binding motifs, as inputs for extended X-ray absorption fine structure (EXAFS) simulations. The corresponding theoretical spectrum that optimally matches the experimental pattern reflects the real spatial arrangement of the uranyl–amidoxime complex in solution. This work was directly performed in a solution to avoid the possible variations in binding mode caused by the crystallization process of a single-crystal sample.
2. Experimental and theoretical investigation
All samples were prepared from UO2(NO3)2·6H2O, which was dissolved in aqueous solutions containing the required concentration of HClO4 or CH3C(NH2)(NOH) (referred to as AO). In the sample, the concentrations of U and HClO4 were maintained at 0.04 M and 1 M, respectively, and the AO concentration was adjusted to 5 M. This choice of ion concentration was based on previous work (Ikeda-Ohno et al., 2009). The 5 M AO concentration was chosen to ensure the excess of AO ligands relative to the uranyl ions, which is similar to the local environment in the extraction of uranium.
The uranium L3-edge spectra were collected at beamline 14W1 of the Shanghai Synchrotron Radiation Facility with a Si (111) double-crystal monochromator in transmission mode. The electron beam energy of the storage ring was 3.5 GeV and the maximum stored current was ∼210 mA. The energy calibration was performed using a Zr foil (∼17 998 eV). Each sample was measured thrice and the obtained spectra were averaged. The U L3-edge data were analysed by the Demeter program (Ravel & Newville, 2005). The double-electron excitations affect the signal and can influence the results of data analysis (see Fig. S1 in the supporting information) (Hennig, 2007; Benfield et al., 1994; Gomilšek et al., 2003; Kodre et al., 2002; Zhang, Zhou et al., 2016). Thus, in uranium L3-edge experimental spectra the double-electron excitations were subtracted as a reflection of the data translated to the position in energy of the excitation using the standard procedures in Demeter. Fig. S1 shows the uranium L3-edge data before and after subtracting the double-electron excitation in k and R space, in which the feature at very short distances (R ≃ 1 Å) obviously improved. The theoretical phase and amplitude functions were calculated using FEFF 9.0 (Rehr et al., 2010). Fitting was performed on the k3-weighted FT-EXAFS (FT = Fourier transform) from 3.4 to 18.4 Å−1. An R window of 1–3 Å was used for the fitting. The amplitude reduction factor S0 2 was fixed at 0.9 in fits and the shifts in the ΔE0 were constrained to be the same value for all fitted shells.
All density functional calculations were performed using the B3LYP hybrid functional (Becke, 1988; Lee et al., 1988) implemented in the Gaussian09 program (Frisch et al., 2009), which is widely used and proven to be sufficiently accurate for extensive systems. The cc-PVDZ basis sets (Dunning, 1989) were adopted for hydrogen, carbon, nitrogen and oxygen atoms. The reliability of our theoretical methods in predicting the structural parameters was also evaluated by optimizing the geometrical structures of the hydrated uranyl ions at the B3LYP/cc-PVDZ level of theory, in which the predicted bond lengths in the aqueous solution are considerably close to the experimental results. Geometry optimizations were performed without symmetry restrictions. Frequency calculations were also performed to verify the structures at the energy minima. Solvation effects were taken into account using the conductor-like polarizable continuum model (CPCM) (Klamt & Schüürmann, 1993; Barone & Cossi, 1998; Cossi et al., 2003; Andzelm et al., 1995). Natural population analyses (NPA) and Wiberg bond indices were calculated to understand the bonding of uranyl–AO complexes based on the B3LYP results through the natural bond orbital (NBO) method (Weinhold & Landis, 2005) implemented in the NBO 3.1 program (Glendening et al., 2003). Charge decomposition analysis (CDA) developed by Franking et al. (Dapprich & Frenking, 1995; Frenking & Fröhlich, 2000) was performed as implemented in the Multiwfn 3.3 software (Lu & Chen, 2012).
3. Results and discussion
3.1. Uranium L3-edge X-ray absorption near-edge structure (XANES) and features of uranyl ions in HClO4 and amidoxime solutions
The quantitative local structural information can be extracted by R space, as shown in Fig. 2 and Table 1. The of the uranyl complex can be described as a bipyramid with the axial trans-dioxo atoms in the apical position and the equatorial coordination changed from tetragonal to hexagonal geometry. According to path analysis in R space, the peaks at R values of about 1.5 Å (A) and 2.0 Å (B) corresponded to the single scattering paths of the axial oxygen atoms (Oax) and the equatorial ligands (Leq), respectively. The fitting results revealed that the uranyl–HClO4 solution contains a uranyl unit exhibiting an R(U—Oax) of 1.77 Å and possesses five coordinated oxygen atoms derived from the water molecules at 2.41 Å along the equatorial plane. This result is consistent with the structure previously reported for hydrated uranyl ions (Neuefeind et al., 2004; Hennig et al., 2007; Sémon et al., 2001). By contrast, the distance of U—Oax (1.81 Å) in uranyl–AO solution is obviously longer than that in HClO4 solution, which indicates that the AO molecule has stronger interaction than the water molecule with the uranyl ions, thus affecting the relative positions of the uranium and the axial oxygen atom. At first glance, the Leq shell of the uranyl–AO sample shows a splitting (R + Δ ≃ 1.7 and 2.0 Å), similar to that of uranyl hydrate. However, previous investigations suggested that the FT peak (R + Δ = 1.7 Å) originating from the superposition of the Oeq shell with the side lobe of the Oax shell in uranyl–hydrate solution does not provide meaningful structural information (Ikeda-Ohno et al., 2008, 2009). A relative intensity reversal and position shift of the Leq shell are observed in the uranyl–AO sample compared with the hydrate. The stronger intensity at a peak of ∼1.7 Å indicates that the peak is no longer attributed to the superposition of the Leq and Oax shells, but contains the real coordination structure. Considering the asymmetry of peak B, two possible U—Leq shells can be used; the fitting distances at 2.32–2.42 Å are shown in Table 1. The average bond length of U—Leq is about 2.37 Å. It is worth mentioning that fits cannot differentiate between the coexisting atoms with close scattering ability, such as O and N atoms. Therefore, the result of the current fitting is insufficient to provide insights into the spatial arrangement of the uranyl–AO complex.
analysis in
‡Debye–Waller factor. Error: σ2 ≤ ±0.0005 Å2. |
As we know, the XANES feature is sensitive to the spatial structure. Fig. 3(a) shows the experimental L3-edge XANES spectra of uranyl ions in HClO4 and AO solutions, which are similar with both having a strong white line (WL) (A), a shoulder (B) at about 15 eV and a feature (C) at about 35 eV above the WL maximum. In terms of the scattering theory, the peaks in the spectrum could be ascribed to the backscattering of photoelectrons from different coordinated shells, and thus the intensity and shape of the peaks reveal the structure of the uranyl–AO complex. The most intense resonance A mainly originates from the dipole that allowed U 2p3/2 → 6d electronic transitions. Two main observations are drawn from Fig. 3(a): (i) the line shapes in AO and HClO4 solutions are consistent with each other, implying similarity in the coordination structure; (ii) the relative positions of peaks B and C slightly intensified in AO aqueous solution compared with those in HClO4 environment. As previously mentioned (Sémon et al., 2001), in the perchlorate acid solution the uranyl ion is surrounded by pure hydrate molecules. By contrast, an obvious pattern improvement indicated that the existence of AO ligands can induce the adjustment of the local structural environment around the uranium atom. In fact, in the XANES spectra of uranyl compounds the energy positions of peaks B and C depend on the U—Oax and U—Oeq distances, respectively. Specifically, 1/R2 was proportional to ΔE, the difference between the multiple-scattering resonance and WL. Thus a shorter U—Oax distance and a longer U—Oeq distance should be obtained in the HClO4 solution than in the uranyl–AO solution, which is exactly consistent with the analysis. Although very informative, the XANES analysis is too complicated to obtain the quantitative spatial structural information independently, especially for actinide chemistry.
To explore the spatial arrangement of the uranyl–AO complex, attention was paid to the (b) shows the experimental oscillation function for the U L3 edge in k space with good quality up to k = 19 Å−1 in both HClO4 and AO solutions. A couple of features are worth noting. First, the main oscillation of the U L3 edge becomes closer together in k space in the uranyl–AO complex, which originates from a longer U—Oax distance and has been verified by fits in R space. Second, the spectra of uranyl ions in the HClO4 and AO solutions present markedly different characteristics, especially in the k range of 6–10 Å−1. In order to clarify the origin of peak characteristic changes and obtain the spatial structure information, an assisted theoretical method is necessary.
oscillation pattern. As we know, the oscillation is a superimposition of individual scattering paths from ligand shell atoms; thus, pattern changes directly reflect the local structure of the uranyl system. Fig. 33.2. Metal–ligand binding motif in the uranyl–AO solution
The possible isomeric structures of the uranyl–AO species in the aqueous solution were investigated by DFT calculations. DFT-optimized geometrical structures served as models and were used in et al., 2014; Dunning, 1989; Guo, Huang et al., 2015; Di Santo et al., 2011; Vetere et al., 2003). Table 2 shows that the average bond length of U—Oax/U—Oeq (1.758 Å/2.435 Å) based on the cc-PVDZ basis sets is much closer to the experimental values (RU–Oax = 1.77 Å, RU–Oeq = 2.41 Å). We then used the same basis sets to optimize the structures of all possible uranyl–AO species. Table 2 shows the optimized structure of the possible [UO2(AO)x]2−x models, including the cationic compounds with mono-AO ligand, the neutral compounds with di-AO ligands and the anionic compounds with tris-AO ligands. The three possible binding motifs including monodentate, bidentate and η2 modes were also considered. The DFT calculations revealed that the distances between uranium and the axial oxygen (U—Oax) increased with the increase in AO coordination. The U—Oax distance in a uranyl complex with mono-AO ligand is 0.02 Å longer than that in the UO2(OH2)52+ complex. Similarly, in uranyl complexes with tris-AO ligands the distances between U and the axial oxygen atoms are longer by 0.01–0.02 Å than those in complexes with two AO ligands and longer by 0.03–0.04 Å than those in mono-AO species. These results indicate that AO coordination weakens the U—Oax bonds considerably compared with water molecules. Observing the uranyl complex binding with the same number of AO ligands, different binding motifs have a negligible effect on the U—Oax distance, although obviously different characters of U—Oeq bond length were observed, which may be attributed to varying binding strengths of the oxime oxygen atoms and oxime nitrogen atoms as well as of amide nitrogen atoms to uranyl ions (Wang et al., 2014; Guo, Huang et al., 2015). A comparison between the distances estimated by DFT and assists in eliminating some of the proposed models with larger structural deviation. For example, most of the di-AO species exhibiting a bidentate model involving the oxime oxygen atom and the amide nitrogen atom are rather unlikely because the U—Oeq distances obtained by DFT are considerably longer compared with the values obtained by fits (2.37 Å). All possible structural models exhibiting U—Oeq distances of ∼2.34–2.40 Å, which are fairly close to the experimental value (2.37 Å), which are marked in bold in Table 2, and their corresponding geometries are displayed in Fig. 4.
simulations. A comparison between the experimental spectra and the simulated oscillation spectra may reveal the intrinsic coordinate structural information. To guarantee that the DFT-optimized structure is nearly similar to the experimental value, we first simulated the structure of the hydrated uranyl ions with typical basis sets for actinyl complexes such as 6-31++G*, 6-31+G(d,p) and 6-311G(d,p), as well as cc-PVDZ for oxygen and hydrogen atoms (WangTo provide more direct evidence for determining the accurate binding mode, we calculated the corresponding . Fig. 5(a) shows comparisons between the experimental and theoretical spectra. For comparison, the pure hydration situation in the uranyl–HClO4 complex is also displayed, in which the model with five water ligands was considered, and simulated theoretical spectra are shown in Fig. 5(a). A similar overall oscillation pattern in the experimental and theoretical case of HClO4 solution can be observed particularly within 6–10 Å−1, which again confirmed the reliability of the optimized structural model. In Fig. 5(a), by carefully comparing the pattern between the experimental spectra and the simulated spectra of the optimized structures of the uranyl–AO species, we found that model No. 10 matched the experimental result very well, especially with regard to the appearance of an oscillation peak at approximately k ≃ 8.5 Å−1. This result showed that the η2 motif with tris-AO ligands is the main binding mode in the uranyl sample in solution containing high AO concentration. The corresponding theoretical XANES spectra of the No. 10 model are also similar to those of the experimental spectra as shown in Fig. 5(b). This binding motif is consistent with that of the previously reported (Barber et al., 2012; Vukovic et al., 2012).
oscillation spectra based on the selected structural models optimized by DFT as shown in Fig. 43.3. Metal–ligand bonding nature in uranyl–AO complexes
The bonding nature including bond orders and atomic charges has been investigated through NBO analysis at the B3LYP level of theory for the uranyl–AO complexes with different binding motif and different AO number; results are summarized in Table 3. In all uranyl–AO complexes, the Wiberg bond indices (WBIs) of U—OAO, U—N and U—Ow bonds show the following bond order trend: U—OAO > U—N > U—Ow, indicating that the U—O/N bonds derived from the AO ligand have more covalent character than those from the H2O molecule. The electron-donating ability of the ligands towards uranium can also be evaluated through NPA. Table 3 shows that the natural charges in the uranium atoms of these complexes are much lower than those of the free uranyl cations (2.736), demonstrating the significant charge transfer from the ligands to uranium (Wang et al., 2014). As for the uranyl–AO complex with η2 binding mode, the net charges on uranium atoms are lower than those in other complexes with monodentate or bidentate modes. This supports the larger electron-donating ability of the AO ligand to uranium in complex with η2 binding mode. We also explore the bonding interaction in uranyl–AO complexes with different AO number. The WBIs of the U—Oax bonds slightly decreased as the coordinated AO ligand increased, consistent with the slight increase in U—Oax bond lengths in Table 2, which is well understood as equatorial orbital interaction slightly weakening the U—O triple bond in [UO2(AO)x]2−x (x = 1 to 3) complexes (Xu et al., 2013; Guo, Wang et al., 2015; Guo, Huang et al., 2015). On the basis of the NPA, with the increasing replacement of water molecules by the AO ligands, the natural charges on the uranium atoms decreased, indicating the stronger electron donation from the AO ligand to the uranium than from the water molecule. This observation is also supported by the considerable difference between ΔQ(AO) and ΔQ(H2O), whose values show AO and H2O coordination, respectively, transfer ∼0.7–0.9 e− and ∼0.2 e− to uranium.
|
To further understand the nature of the metal–ligand bonding in uranyl–AO complexes, we performed CDA and some of the relevant uranyl–AO bonding molecular orbitals (MOs) are shown in Fig. 6. The MO plots provide a graphic description of the U—O and U—N σ-bonding, which mainly originates from the interaction of U 5f, 6d orbitals and N, O 2p orbitals. Firstly, we compared the compositions of the MOs involving the metal–ligand bonding in the uranyl–AO complex with different binding motifs, which are significantly different. For the complex No. 3 with a monodentate motif, the U—O σ-bonding MOs contain 4% U 5fδφ and 86% O 2p character. For model No. 2 with a bidentate motif, HOMO-3 contains 2% U 5fφ, 4% U 6dδ and 86% (N, O) 2p atomic orbitals and HOMO-1 contains 2% U 5fφ, 3% U 6dδ and 82% (N, O) 2p character. As for model No. 1 with η2 motif, the U—O σ-bonding MOs contain 7% U 5fφ, 7% U 6dδ and 74% N/O character. These results indicate that introduction of the η2 binding motif increased the uranium 5f orbital character during metal–ligand bonding. To further probe the bonding interaction between uranyl and the AO ligand in uranyl–AO complexes with different AO number, we also performed CDA for compounds No. 4 and No. 10, and present the main uranyl–AO bonding MOs in Fig. 6. Interestingly, the number and shape of MOs associated with the metal–ligand bonding change with the number of coordinated AO ligands. In the UO2(AO)(H2O)3+ compound, only one bonding MO between uranyl and AO was found, i.e. HOMO-1, which is composed of 7% U 5fφ, 7% U 6dδ and 74% (N, O) 2p atomic orbitals. When two AOs coordinated to uranyl ions, in the UO2(AO)2(H2O) compound HOMO-2 and HOMO-3 are formed by the interaction of (N, O) 2p with U 5fφ, 6dδ orbitals, respectively, while in the UO2(AO)3− compound (Zhang, Su et al., 2016) two MOs, i.e. HOMO-3 and HOMO-4, are both composed of 6% U 6dδ and ∼80% (N, O) 2p orbitals. In addition, HOMO-5 mainly comes from the interaction of U 5fφ and (N, O) 2p orbitals. The contributions of U 5fφ, 6dδ orbitals to the uranyl–AO bonding MOs obviously increased as the coordinated AO ligands increased, probably because of a better symmetry and energy match with the (N, O) 2p orbitals. The strong binding ability of the AO ligand to uranyl is caused by the strong covalent interaction and orbital between U 5f, 6d and (N, O) 2p as shown in Fig. 6.
4. Conclusions
In summary, we explored the geometry and electronic structure of aqueous UO22+–amidoxime complex by performing experimental and theoretical analyses of data combined with DFT calculations. fitting in R space directly provides information on bond length in the uranyl–amidoxime solution, which then serves as a reference to identify the most probable spatial arrangement of the complex. DFT calculations were performed to simulate the spatial arrangement of the possible binding motif, as well as the bonding nature between the amidoxime ligand and uranyl ions. As a comparison, uranyl contains five water molecules on the equatorial plane and forms a pure hydrate species in uranyl–HClO4 solution. In the highly concentrated amidoxime solution, the hydrate water molecules were replaced by the tris-amidoximate ligands, which coordinate with uranyl ions via an η2 coordination motif. CDA results show that the strong binding ability of the amidoximate ligand to uranyl is caused by the strong covalent interaction and orbital between U 5f, 6d and (N, O) 2p. It is worth mentioning that this study provides an approach to directly determine the instinctive coordination properties in uranyl complexes, even in other actinide complexes; this method is particularly suitable in solution systems. In future studies, we will further investigate the detailed coordination geometry structure of the uranyl–amidoximate complex under different environmental conditions (pH, temperature and concentration) using combined diagnostic tools.
Supporting information
Supporting figures, tabes and text. DOI: https://doi.org/10.1107/S160057751800067X/rv5073sup1.pdf
Acknowledgements
We thank Christoph Hennig and Ling-Yun Jang for hints and fruitful discussions about the double-electron excitations.
Funding information
This work was partly supported by the National Natural Science Foundation of China (grant No. 11405254); Joint Funds of the National Natural Science Foundation of China (grant Nos. U1532259 and U1532111); Knowledge Innovation Program of Chinese Academy of Sciences; and Youth Innovation Promotion Association (2014237), Chinese Academy of Sciences.
References
Abney, C. W., Mayes, R. T., Piechowicz, M., Lin, Z., Bryantsev, V. S., Veith, G. M., Dai, S. & Lin, W. (2016). Energ. Environ. Sci. 9, 448–453. CSD CrossRef CAS Google Scholar
Andzelm, J., Kölmel, C. & Klamt, A. (1995). J. Chem. Phys. 103, 9312–9320. CrossRef CAS Google Scholar
Astheimer, L., Schenk, H. J., Witte, E. G. & Schwochau, K. (1983). Sep. Sci. Technol. 18, 307–339. CrossRef CAS Google Scholar
Bailey, E. H., Mosselmans, J. F. W. & Schofield, P. F. (2005). Chem. Geol. 216, 1–16. Web of Science CrossRef CAS Google Scholar
Barber, P. S., Kelley, S. P. & Rogers, R. D. (2012). RSC Adv. 2, 8526–8530. Web of Science CSD CrossRef CAS Google Scholar
Barone, V. & Cossi, M. (1998). J. Phys. Chem. A, 102, 1995–2001. Web of Science CrossRef CAS Google Scholar
Becke, A. D. (1988). Phys. Rev. A, 38, 3098–3100. CrossRef CAS PubMed Web of Science Google Scholar
Benfield, R. E., Filipponi, A., Bowron, D. T., Newport, R. J. & Gurman, S. J. (1994). J. Phys. Condens. Matter, 6, 8429–8448. CrossRef CAS Web of Science Google Scholar
Cossi, M., Rega, N., Scalmani, G. & Barone, V. (2003). J. Comput. Chem. 24, 669–681. Web of Science CrossRef PubMed CAS Google Scholar
Dapprich, S. & Frenking, G. (1995). J. Phys. Chem. 99, 9352–9362. CrossRef CAS Google Scholar
Das, S., Pandey, A. K., Athawale, A. A. & Manchanda, V. K. (2009). J. Phys. Chem. B, 113, 6328–6335. CrossRef CAS Google Scholar
Davies, R. V., Kennedy, J., McIlroy, R. W., Spence, R. & Hill, K. M. (1964). Nature (London), 203, 1110–1115. CrossRef Web of Science Google Scholar
Di Santo, E., Santos, M., Michelini, M. C., Marçalo, J., Russo, N. & Gibson, J. K. (2011). J. Am. Chem. Soc. 133, 1955–1970. CrossRef CAS Google Scholar
Dunning, T. H. (1989). J. Chem. Phys. 90, 1007–1023. CrossRef CAS Web of Science Google Scholar
Evans, H. T. (1963). Science, 141, 154–158. CrossRef CAS Google Scholar
Frenking, G. & Fröhlich, N. (2000). Chem. Rev. 100, 717–774. Web of Science CrossRef PubMed CAS Google Scholar
Frisch, M. J. et al. (2009). Gaussian09. Gaussian Inc., Wallingford, CT, USA. Google Scholar
Glendening, E. D., Reed, A. E., Carpenter, J. E. & Weinhold, F. (2003). NBO, Version 3.1. Gaussian Inc., Pittsburgh, PA, USA. Google Scholar
Gomilšek, J. P., Kodre, A., Arčon, I. & Hribar, M. (2003). Phys. Rev. A, 68, 042505. Google Scholar
Gückel, K., Tsushima, S. & Foerstendorf, H. (2013). Dalton Trans. 42, 10172–10178. Web of Science PubMed Google Scholar
Guo, X., Huang, L., Li, C., Hu, J., Wu, G. & Huai, P. (2015). Phys. Chem. Chem. Phys. 17, 14662–14673. CrossRef CAS Google Scholar
Guo, X., Wang, Y., Li, C., Huai, P. & Wu, G. (2015). Mol. Phys. 113, 1327–1336. CrossRef CAS Google Scholar
Hennig, C. (2007). Phys. Rev. B, 75, 035120. Web of Science CrossRef Google Scholar
Hennig, C., Schmeide, K., Brendler, V., Moll, H., Tsushima, S. & Scheinost, A. C. (2007). Inorg. Chem. 46, 5882–5892. Web of Science CrossRef PubMed CAS Google Scholar
Ikeda-Ohno, A., Hennig, C., Rossberg, A., Funke, H., Scheinost, A. C., Bernhard, G. & Yaita, T. (2008). Inorg. Chem. 47, 8294–8305. Web of Science PubMed CAS Google Scholar
Ikeda-Ohno, A., Hennig, C., Tsushima, S., Scheinost, A. C., Bernhard, G. & Yaita, T. (2009). Inorg. Chem. 48, 7201–7210. Web of Science PubMed CAS Google Scholar
Jackson, V. E., Gutowski, K. E. & Dixon, D. A. (2013). J. Phys. Chem. A, 117, 8939–8957. CrossRef CAS Google Scholar
Katragadda, S., Gesser, H. D. & Chow, A. (1997). Talanta, 45, 257–263. CrossRef CAS Google Scholar
Kim, J., Tsouris, C., Mayes, R. T., Oyola, Y., Saito, T., Janke, C. J., Dai, S., Schneider, E. & Sachde, D. (2013). Sep. Sci. Technol. 48, 367–387. Web of Science CrossRef CAS Google Scholar
Kim, J., Tsouris, C., Oyola, Y., Janke, C. J., Mayes, R. T., Dai, S., Gill, G., Kuo, L.-J., Wood, J., Choe, K.-Y., Schneider, E. & Lindner, H. (2014). Ind. Eng. Chem. Res. 53, 6076–6083. CrossRef CAS Google Scholar
Klamt, A. & Schüürmann, G. (1993). J. Chem. Soc. Perkin Trans. 2, pp. 799–805. CrossRef Web of Science Google Scholar
Kodre, A., Arcon, I., Gomilsek, J. P., Preseren, R. & Frahm, R. (2002). J. Phys. B, 35, 3497–3513. Web of Science CrossRef CAS Google Scholar
Lee, C., Yang, W. & Parr, R. G. (1988). Phys. Rev. B, 37, 785–789. CrossRef CAS Web of Science Google Scholar
Liu, Q., Zhang, Q., Yang, S., Zhu, H., Liu, Q. & Tian, G. (2017). Dalton Trans. 46, 13180–13187. CrossRef CAS Google Scholar
Lu, T. & Chen, F. (2012). J. Comput. Chem. 33, 580–592. Web of Science CrossRef PubMed Google Scholar
McGrail, B. T., Sigmon, G. E., Jouffret, L. J., Andrews, C. R. & Burns, P. C. (2014). Inorg. Chem. 53, 1562–1569. CrossRef CAS Google Scholar
Neuefeind, J., Soderholm, L. & Skanthakumar, S. (2004). J. Phys. Chem. A, 108, 2733–2739. CrossRef CAS Google Scholar
Ravel, B. & Newville, M. (2005). J. Synchrotron Rad. 12, 537–541. Web of Science CrossRef CAS IUCr Journals Google Scholar
Rehr, J. J., Kas, J. J., Vila, F. D., Prange, M. P. & Jorissen, K. (2010). Phys. Chem. Chem. Phys. 12, 5503–5513. Web of Science CrossRef CAS PubMed Google Scholar
Schenk, H. J., Astheimer, L., Witte, E. G. & Schwochau, K. (1982). Sep. Sci. Technol. 17, 1293–1308. CrossRef CAS Google Scholar
Seko, N., Katakai, A., Tamada, M., Sugo, T. & Yoshii, F. (2004). Sep. Sci. Technol. 39, 3753–3767. CrossRef CAS Google Scholar
Sémon, L., Boehme, C., Billard, I., Hennig, C., Lützenkirchen, K., Reich, T., Rossberg, A., Rossini, I. & Wipff, G. (2001). ChemPhysChem, 2, 591–598. Google Scholar
Ting, C. F., Jie, X., Wei, H. J., Mei, G., Sheng, H. & Lin, W. X. (2013). J. Dispersion Sci. Technol. 34, 604–610. CrossRef CAS Google Scholar
Vetere, V., Maldivi, P. & Adamo, C. (2003). J. Comput. Chem. 24, 850–858. CrossRef CAS Google Scholar
Vukovic, S. & Hay, B. P. (2013). Inorg. Chem. 52, 7805–7810. Web of Science CrossRef CAS PubMed Google Scholar
Vukovic, S., Watson, L. A., Kang, S. O., Custelcean, R. & Hay, B. P. (2012). Inorg. Chem. 51, 3855–3859. Web of Science CSD CrossRef CAS PubMed Google Scholar
Wang, C.-Z., Lan, J.-H., Wu, Q.-Y., Luo, Q., Zhao, Y.-L., Wang, X.-K., Chai, Z.-F. & Shi, W.-Q. (2014). Inorg. Chem. 53, 9466–9476. CrossRef CAS Google Scholar
Weinhold, F. & Landis, C. R. (2005). Valency and Bonding: a Natural Bond Orbital Donor–Acceptor Perspective. Cambridge University Press. Google Scholar
Witte, E. G., Schwochau, K. S., Henkel, G. & Krebs, B. (1984). Inorg. Chim. Acta, 94, 323–331. CSD CrossRef CAS Web of Science Google Scholar
Xu, C., Su, J., Xu, X. & Li, J. (2013). Sci. China Chem. 56, 1525–1532. CrossRef CAS Google Scholar
Zhang, L., Su, J., Yang, S., Guo, X., Jia, Y., Chen, N., Zhou, J., Zhang, S., Wang, S., Li, J., Li, J., Wu, G. & Wang, J.-Q. (2016). Ind. Eng. Chem. Res. 55, 4224–4230. CrossRef CAS Google Scholar
Zhang, A., Uchiyama, G. & Asakura, T. (2005). React. Funct. Polym. 63, 143–153. Web of Science CrossRef Google Scholar
Zhang, L., Zhou, J., Zhang, J., Su, J., Zhang, S., Chen, N., Jia, Y., Li, J., Wang, Y. & Wang, J.-Q. (2016). J. Synchrotron Rad. 23, 758–768. CrossRef CAS IUCr Journals Google Scholar
© International Union of Crystallography. Prior permission is not required to reproduce short quotations, tables and figures from this article, provided the original authors and source are cited. For more information, click here.