Effects of coordinating heteroatoms on molecular structure, thermodynamic stability and redox behavior of uranyl(vi) complexes with pentadentate Schiff-base ligands

Uranyl(vi) complexes with pentadentate N3O2-, N2O3- and N2O2S1-donating Schiff base ligands, tBu,MeO–saldien–X2− (X = NH, O and S), were synthesized and thoroughly characterized by 1H NMR, IR, elemental analysis, and single crystal X-ray diffraction. The crystal structures of UO2(tBu,MeO–saldien–X) showed that the U–X bond strength follows U–O ≈ U–NH > U–S. Conditional stability constants (βX) of UO2(tBu,MeO–saldien–X) in ethanol were investigated to understand the effect of X on thermodynamic stability. The log βX decrease in the order of UO2(tBu,MeO–saldien–NH) (log βNH = 10) > UO2(tBu,MeO–saldien–O) (log βO = 7.24) > UO2(tBu,MeO–saldien–S) (log βS = 5.2). This trend cannot be explained only by Pearson's Hard and Soft Acids and Bases (HSAB) principle, but rather follows the order of basicity of X. Theoretical calculations of UO2(tBu,MeO–saldien–X) suggested that the ionic character of U–X bonds decreases in the order of U–NH > U–O > U–S, while the covalency increases in the order U–O < U–NH < U–S. Redox potentials of all UO2(tBu,MeO–saldien–X) in DMSO were similar to each other regardless of the difference in X. Spectroelectrochemical measurements and DFT calculations revealed that the center U6+ of each UO2(tBu,MeO–saldien–X) undergoes one-electron reduction to afford the corresponding uranyl(v) complex. Consequently, the difference in X of UO2(tBu,MeO–saldien–X) affects the coordination of tBu,MeO–saldien–X2− with UO22+. However, the HSAB principle is not always prominent, but the Lewis basicity and balance between ionic and covalent characters of the U–X interactions are more relevant to determine the bond strengths.


Introduction
Uranium is the most important element in nuclear engineering. The chemistry of uranium plays important roles in nuclear fuel fabrication and spent fuel reprocessing. Under ambient conditions, uranium is most commonly present as a hexavalent uranyl(VI) ion, UO 2 2+ , with a typical linear [O^U VI^O ] 2+ structure. The chemical separation of UO 2 2+ from various aqueous systems such as feed solutions of spent nuclear fuels and even seawater is one of the important research topics in nuclear chemistry. In the usual sense, coordination chemistry provides very powerful tools for chemical separation. Hence, the complexation between UO 2 2+ and organic ligands has been widely studied. [1][2][3][4][5][6][7][8][9][10][11][12][13] Pearson's Hard and So Acids and Bases (HSAB) principle is quite useful to describe preferential interactions between specic metal ions and coordinating atoms of ligands and to design organic molecules selectively coordinating with a target metal ion, 14,15 although this principle is rather empirical. In the HSAB principle, UO 2 2+ is classied as a hard acid, 14,15 and therefore, generally tends to more strongly interact with hard bases like N, O and F, compared with soer ones such as heavier congeners like P, S, and Cl. 14,15 Indeed, thermodynamic stability of a UO 2

2+
-halido complex in DMF follows the order of hardness of halide ligands, Cl À > Br À > I À . 16 In contrast, such a trend in complexation between UO 2 2+ and heteroatoms like N, O and S, seems not to be well understood systematically, although it would provide essential information to understand the fundamental nature of UO 2 2+ in more depth and to design molecular structures of ligands exclusively interacting with UO 2 2+ . Indeed, several extracting reagents have been successfully developed for separation of Am(III) and Cm(III) from Ln(III) on the basis of difference in coordinating affinities of these metal ions with so-donor atoms incorporated in the designed ligands. [17][18][19][20] In this study, we discuss strengths of U-N, U-O and U-S interactions formed in UO 2 2+ complexes having analogous coordination geometries. For this purpose, it is rst necessary to choose a suitable ligand system. Previously, we reported UO 2 2+ complexes with N 3 O 2 -pentadentate Schiff base ligands, UO 2 (R 1 ,R 2 ,-R saldien), shown in Fig. 1(a). 21,22 Its NR moiety can be substituted with O or S to provide the similar UO 2 2+ complexes, UO 2 (tBu,MeO-saldien-X) (UO 2 (L X ), X ¼ NH, O, S, Fig. 1(b)), where the U-O or U-S bond will be formed instead of U-NR. Here, we report synthesis and characterization of UO 2 (L X ) (X ¼ NH, O, S) to discuss effects of X to the U-X bond strength and thermodynamic stability as well as redox chemistry of this class of UO 2 2+ complexes.

Materials and syntheses
All reagents used were of reagent grade and used as received, if not specied. 3-tert-Butyl-5-methoxysalicylaldehyde was synthesized as reported elsewhere. 23 UO 2 (tBu,MeO-Saldien-NH) (UO 2 (L NH )). To a solution of 3tert-butyl-5-methoxysalicylaldehyde (202 mg, 0.971 mmol) in ethanol (2 mL) was added 2,2 0 -diaminodiethylamine (52.1 mL, 0.480 mmol). This solution was heated to reux for 10 min. UO 2 (NO 3 ) 3 $6H 2 O (210 mg, 0.418 mmol) dissolved in ethanol (2 mL) was dropwise added to the solution. A red precipitate was formed within several minutes, and the suspension was stirred at 60 C for 1 h. Aer cooling to room temperature, a red precipitate was collected by ltration and rinsed with methanol. Recrystallization from CH 2 Cl 2 /ethanol yielded red crystals. Yield: 176 mg (48%). This compound was characterized by 1 H NMR, IR and elemental analysis. 1 3.40. The obtained crystals were also suitable for the X-ray crystallography.

Methods
The 1 H NMR spectra were recorded by using JEOL ECX-400 ( 1 H: 399.78 MHz) NMR spectrometer. The chemical shis of 1 H NMR were referenced to TMS (d ¼ 0 ppm). The IR measurements were performed by JASCO FT/IR4700 equipped with a diamond ATR attachment. Elemental analyses were carried out by Yanaco MT-6 CHN elemental analyzer. Cyclic voltammetry (CV) measurements of UO 2 (L X ) (1 mM) dissolved in DMSO containing 0.1 M tetra-n-butylammonium perchlorate (TBAP) were performed at 295 K under a dry Ar atmosphere by using BAS ALS660B electrochemical analyzer. A three-electrode system consisted of a Pt disk working electrode (diameter: 1.6 mm, surface area: 0.020 cm 2 ), a Pt wire counter electrode, and an Ag 0/+ reference electrode (0.1 M TBAP + 1 mM AgNO 3 / CH 3 CN). A ferrocene/ferrocenium ion redox couple (Fc 0/+ ) was taken as an external standard redox system. All samples were prepared under an inert Ar atmosphere. Dissolved oxygen gas in each sample solution was expelled by purging Ar gas for at least 10 min prior to starting the CV experiments. UV-vis-NIR spectroelectrochemical measurements in DMSO were performed with a JASCO V-770 spectrophotometer equipped with an optically transparent thin layer electrode (OTTLE) cell at 295 K. [24][25][26][27] Its optical path length was 1.0 Â 10 À2 cm, which was calibrated spectrophotometrically. 22 The three-electrode system was the same as that in the above electrochemical experiments with a replacement of the working electrode by a Pt gauze (80 mesh). The potential applied on OTTLE was controlled by BAS ALS660B. The absorption spectrum at each potential step was recorded aer equilibration of the electrochemical reaction at the applied potential on the working electrode, which completed within 3 min. The sample solution in the OTTLE cell was prepared in a similar manner to that for the CV measurements.

Crystallographic analysis
The X-ray diffraction data of the well-shaped single crystals of UO 2 (L NH )$(CH 2 Cl 2 ), UO 2 (L O )$(C 5 H 5 N) and UO 2 (L S )$(CH 2 Cl 2 ) were collected by a Rigaku XtaLAB mini II equipped with hybrid pixel array detector and graphite monochromated Mo Ka radiation (l ¼ 0.71073 A). Each sample was mounted on a MiTeGen Dual Thickness MicroMounts, and located in the temperaturecontrolled N 2 gas ow. Intensity data were collected by taking oscillation photographs. Reection data were corrected for both Lorentz and polarization effects. The structures were solved by the direct method and rened anisotropically by the SHELX program suite 28 for non-hydrogen atoms by full-matrix leastsquares calculations. Each renement was continued until all shis were smaller than one-third of the standard deviations of the parameters involved. Hydrogen atoms were located at the calculated positions. All hydrogen atoms were constrained to ideal geometry with C-H ¼ 0.95 A. The thermal parameters of all hydrogen atoms were related to those of their parent atoms by U(H) ¼ 1.2U eq (C). All calculations were performed by using the Olex2 crystallographic soware program package. 29 Crystallographic data of all complexes were summarized in Table  S1, † and deposited with Cambridge Crystallographic Data Centre as supplementary publication no: CCDC 2177295 (UO 2 (L O )$(C 5 H 5 N)), 2177296 (UO 2 (L NH )$(CH 2 Cl 2 )), and 2177297 (UO 2 (L S )$(CH 2 Cl 2 )).

Theoretical calculations
Density functional theory (DFT) calculations were performed using Gaussian 16 program (Revision B.01) 31 for characterization of UO 2 (L X ) and the one-electron reduced complexes, The atomic coordinates of UO 2 (L X ) were taken from those experimentally-determined and were used for structure optimization. Hybrid DFT functional B3LYP 32 was employed and solvent was modelled through a conductorlike polarized continuum model (CPCM) for DMSO (dielectric 3 ¼ 46.7). 33 For uranium, Stuttgart-type small-core effective core potential (ECP) and corresponding basis set has been used. 34 The most diffuse basis functions on uranium with the exponent 0.005 (all s, p, d, and f type functions) were omitted as in previous studies. [35][36][37][38][39] The 6-311G(d) basis sets were used for other elements (C, H, N, O, S). Vibrational frequency calculations at the same level of theory conrmed that no imaginary frequency was found to be present. Single-point calculations for energetic analysis were performed using the same condition.
NBO analysis were carried out by using the NBO 5.0 program. 40 The molecular structures of [UO 2 (L X )] À were taken from those of UO 2 (L X ) determined experimentally and were optimized aer addition of a single negative charge and doublet spin degeneracy to assume the one-electron reduction using the same condition. The Mulliken spin-density plots were illustrated by GaussView 6.1. 41

Results and discussion
Synthesis and structure determination of UO 2 (L X ) Each ligand was synthesized through a condensation reaction between a 3-tert-butyl-5-methoxysalicylaldehyde and the corresponding diamine in ethanol, and further reacted with one equivalent of UO 2 (NO 3 ) 2 $6H 2 O to afford UO 2 (L X ). These complexes were yielded as red microcrystalline solids, which were recrystallized from appropriate solvent mixtures to obtain single crystals suitable for X-ray structure determination. The IR peaks of [O^U^O] 2+ asymmetric stretching (n 3 ) and C]N stretching (n C]N ) of UO 2 (L X ) were observed at around 860-880 and 1630 cm À1 , respectively. The elemental analysis for UO 2 (L X ) well-agreed with the expected chemical formulae of them. The molecular structures of UO 2 (L X ) were determined by single crystal X-ray diffraction (SCXRD). The resulting molecular structures of UO 2 (L X ) are shown in Fig. 2 and S2. † The selected bond lengths of them are summarized in Table 1. As a general trend, UO 2 2+ in UO 2 (L X ) is ve-coordinated in its equatorial plane to give a pentagonal bipyramidal coordination geometry as expected in Fig. 1(b), which is typically found in UO 2

2+
complexes. 21,22 The U^O ax bond lengths of UO 2 (2)) are 1.78-1.79 A, which is similar to those in UO 2 2+ complexes reported previously. 21,22 Herein, we introduced tertbutyl groups at the ortho-positions of the phenolate moieties in each system to control the structure of the ligand aer coordination to UO 2 2+ . To avoid steric collision between these bulky groups in UO 2 (L X ), two phenolate moieties are forced to be present in the opposite sides of the equatorial plane of UO 2 2+ ( Fig. 2 and S2 †). Such a twisted structure of a planar pentadentate ligand was also observed in the saldien-type ligands (e.g., Fig. 1(a)) we reported previously. 21,22 The bond angles around X are strongly affected by the difference in X. The mean bond angle of C (9) However, it is difficult at this moment to clearly describe in detail how the hardness/soness of X affects such a structural trend. Hence, we decided to focus on the bond lengths around X as another structural parameter directly affected by the coordination strength.
In the UO 2 (L X ) complexes studied here, the bond lengths between U and the phenolic O (U(1)-O(3), U(1)-O(4)) are 2.23-2.24 A regardless of difference in X. This is also the case for those between U and the imino N (U(1)-N(1), U(1)-N(2), 2.52-2.55 A). In contrast, the U(1)-X bond lengths depend on X. The U(1)-N(3) distance of UO 2 (L NH ) is 2.594(5) A, which is slightly longer than the corresponding interaction in UO 2 (L O ) (U(1)-O(5) ¼ 2.581(3) A). These bond lengths are commonly found in the previous reports. 21,22 The U(1)-S(1) distance in UO 2 (L S ) is signicantly longer than the others. However, the U(1)-S(1) distance of UO 2 (L S ) is still shorter than the sum of van der Waals radii of U and S (2.3 A + 1.8 A ¼ 4.1 A), 42 suggesting that chemical bonding interaction is certainly present between U(1)   (1) and S(1) in this complex. Indeed, the U(1)-S(1) distance of UO 2 (L S ) (2.981(2) A) is close to the U-S bond lengths in UO 2 2+thioether complexes reported previously (2.96-3.02 A). 2,43,44 The observed structural parameters of the U-X interactions in UO 2 (L x ) are quite common in uranyl complexes having X atom coordination reported so far. 2,21,22,43,44 Therefore, UO 2 (L X ) studied here are suitable for exploring impacts of X in the coordination chemistry of UO 2 2+ .
It could be misleading to discuss the strengths of the U-X bonding interactions solely on the basis of the observed bond lengths, because the sizes of N, O and S are different from each other. The bond strengths between two atoms can be normalized by reduction in an interaction distance (R UX ) derived from the sum of van der Waals radii and an actual bond length between U and X as shown in eqn (1). 45-47 where r U and r X , are van der Waals radii of U and X, respectively. d UX is the U-X bond length of UO 2 (L X ) determined by SCXRD. [45][46][47] Based on this denition, greater R UX implies weaker U-X bond (vice versa). As a result, R UX of UO 2 (L NH ) is 0.665, which is close to that of UO 2 (L O ) (0.670). This implies that the bond strengths of U(1)-N(3) in UO 2 (L NH ) and U(1)-O(5) in UO 2 (L O ) are similar to each other. In contrast, R UX of UO 2 (L S ) is 0.727, which is signicantly greater than those of UO 2 (L NH ) and UO 2 (L O ). Hence, the U(1)-S(1) bond strength of UO 2 (L S ) is supposed to be weaker than the U-X ones in UO 2 (L NH ) and UO 2 (L O ). Consequently, the bond strength of U-X interactions follows U-O z U-NH > U-S. As widely accepted in the HSAB principle, the hardness of X moiety follows O > NH > S. 14,15 Therefore, the trend of U-X bond strengths of UO 2 (L X ) cannot be explained only by the HSAB principle, while the bond strength of U(1)-S(1) of UO 2 (L S ) is certainly weaker than others. Note that all R UX of UO 2 (L X ) presented here are much smaller than those of noncovalent intermolecular interactions such as Cl/X and hydrogen bonds reported previously, where R ¼ 0.98-0.80. [45][46][47] Therefore, a coordination bond is certainly formed between U and X in each UO 2 (L X ).
Thermodynamic stability of UO 2 (L X ) (X ¼ NH, O, S) In the crystal structures of UO 2 (L X ) (X ¼ NH, O, S), the U-X bond strength depends on the difference in X. Therefore, there would also be some impact on the thermodynamic stability of UO 2 (L X ).
To conrm this issue, we investigated the complexation of UO 2 2+ and L X 2À in ethanol by spectrophotometric titration. Fig. 3 shows the UV-vis absorption spectra recorded at different total concentration ratios between UO 2 2+ and L X 2À represented by C U /C L . Note that these titration experiments were conducted under the presence of 0.4 mM NEt 3 employed as a H + scavenger aer the formation of UO 2 (L X ).
In all titration series shown in Fig. 3, the absorbance at 370 nm and 425 nm increased with an increase in C U /C L . Simultaneously, the absorption intensity at 350 nm decreased. The isosbestic points were clearly observed, indicating that the complexation equilibrium between UO 2 2+ and L X 2À only takes place in each system. As shown in the insets of Fig. 3(a) and (b), the absorbance at 370 nm tends to be saturated at C U /C L ¼ 1.0, indicating that UO 2 (L NH ) and UO 2 (L O ) are almost quantitatively formed. On the other hand, such a trend is equivocal for X ¼ S (Fig. 3(c)), implying that the weaker coordination of L S 2À .
To estimate the conditional stability constants containing NEt 3 (0.4 mM), the spectral series of Fig. 3 were analyzed by HypSpec program. 30 As a result, log b X of UO 2 (L X ) for X ¼ NH, O, and S are estimated to be 10 AE 1, 7.24 AE 0.02, and 5.2 AE 0.1, respectively. Since all the coordinating atoms except for X are common in the studied systems, the difference in log b X observed here can be ascribed to the difference in affinity of X with UO 2 2+ . As widely-accepted in the HSAB principle, the hardness of X atoms follows O > NH > S. 14,15 However, log b X of UO 2 (L X ) decrease in the order of UO 2 (L NH ) > UO 2 (L O ) > UO 2 (L S ), which is difficult to be rationalized only by the HSAB principle.
To understand this trend, we focus on difference in basicity of X in L X

2À
. The pK a values of protonated diethylamine ((CH 3 -CH 2 ) 2 NH 2 + ), dimethyl ether ((CH 3 ) 2 OH + ), and dimethyl thioether ((CH 3  of UO 2 (L X ) described above. Therefore, the basicity of X atom would also provide some contribution to the thermodynamic stability of UO 2 (L X ). At this moment, it is still too early to verify linear free energy relationship between log b X and pK a .
To further elucidate the nature of U-X bonds in UO 2 (L X ), we carried out DFT calculations of UO 2 (L X ), followed by the natural bond orbital (NBO) analysis. The molecular structures of UO 2 (L X ) were taken from those of UO 2 (L X ) determined by X-ray crystallography, and were optimized with B3LYP method. 32 The optimized structures of UO 2 (L X ) are shown in Fig. S4, † and selected bond lengths are summarized in Table S2. † All the bond distances well agree with those determined crystallographically (Table 1). Table S3 † summarizes natural charges and Wiberg bond indices (WBI) 17,18,40 of center U and coordinating atoms in the optimized structures.
No signicant differences were found in the natural charge on the axial and equatorial coordinating atoms except for X. Both N(3) in UO 2 (L NH ) (À0.646) and O(5) in UO 2 (L O ) (À0.565) have negative natural charges, indicating that the center U and X atoms interact electrostatically. In contrast, the natural charge of S(1) in UO 2 (L S ) is positive (+0.326), implying that the electrostatic attraction between U and S is little expectable despite signicant penetration between these atoms in UO 2 (L S ) within the sum of van der Waals radii as described above. To provide a rationale for the U-S bonding interaction experimentally observed, bond orders of U-X interactions were estimated in terms of WBI. As a result, some covalency was detected in the U-S bond of UO 2 (L S ) as pronounced by WBI ¼ 0.471, which is signicantly larger than those of the other U-X bonds (WBI ¼ 0.277-0.345). Therefore, the bonding interaction between U and S of UO 2 (L S ) is rather covalent, while it is somewhat weakened by the electrostatic repulsion between these positively charged atoms. The signicant covalency of the U-S interaction compared with the electrostatic characters in U-NH and U-O would be a typical manifestation of the HSAB principle. In connection with this, N is usually considered to be soer than O, while the stability of UO 2 (L NH ) is greater than UO 2 (L O ) despite the hardness of UO 2

2+
. The stronger basicity of NH provides an additional effect to strengthen the U-NH bond compared with that of U-O.
Electrochemistry and spectroelectrochemistry of UO 2 (L X ) (X ¼ NH, O, S) As mentioned above, X strongly affects the thermodynamic stability of UO 2 (L X ). Recently, we have reported that the redox potential of UO 2 (R 1 ,R 2 -Me saldien) ( Fig. 1(a)) is signicantly governed by substitution at R 1 and R 2 positions. Therefore, we expect that the difference in X may also vary the redox potentials of UO 2 (L X ). To clarify this point, the electrochemical measurements of UO 2 (L X ) in DMSO were carried out. Fig. 4 shows the obtained cyclic voltammograms of UO 2 (L X ), where a couple of cathodic (E pc ) and anodic peaks (E pc ) has been observed. These redox waves are reproducible even in multiple scanned cyclic voltammograms recorded at the potential sweep rate (n) of 100 mV s À1 , indicating that the reduction product at E pc undergoes no successive reactions, and is fully reoxidized to UO 2 (L X ) at E pa (Fig. S6 †). The peak potential separation (E pc À E pa ) tends to increase (111-490 mV) with increasing v from 50 mV s À1 to 500 mV s À1 (Fig. S7 and Tables S4-S6 †), implying that these redox systems of UO 2 (L X ) are quasireversible. Even aer careful survey of the DFT calculations described later, we, however, could not nd any critical rationales for the differences in the electrochemical reversibility of these systems. Anyway, the peak potential separations (E pa À E pc , see Table S6 in ESI †) of UO 2 (L S ) were also much greater than the theoretical value of a reversible system (59 mV). Therefore, all the systems studied here are regarded to be electrochemically irreversible. Although we do not have unequivocal explanation for the above points at this moment, solvation structures around these uranyl complexes could be largely modied through the electron transfer. Note that all the redox reactions are chemically reversible as demonstrated by occurrence of the isosbestic points in the spectroelectrochemical experiments shown in  in these systems at 295 K were estimated as 1.6 Â 10 À6 , 1.8 Â 10 À6 and 8.1 Â 10 À7 cm 2 s À1 , respectively, where the redox reactions observed in Fig. 4 were assumed to be electrochemically irreversible. 49 As summarized in Tables S4-S6, † the formal potential E 0 (¼(E pc + E pa )/2) of each UO 2 (L X ) is around À1.60 V vs. Fc 0/+ with regardless of v, and also seems not to be largely affected by X. The E 0 value of UO 2 (L NH ) well agrees with that of its analogue, UO 2 (tBu,MeO-Me saldien), we reported previously (À1.60 V vs. Fc 0/+ in DMSO). 21,22 Therefore, the coordinating L X 2À would not have strong contribution to the redox events of UO 2 (L X ). From these results, we assume that the redox centers of all UO 2 (L X ) are the UO 2 2+ moiety. However, cyclic voltammograms does not provide any detailed information about the reductant of UO 2 (L X ). Hence, we carried out the spectroelectrochemical measurements and theoretical calculations to further understand the redox chemistry of UO 2 (L X ).
To determine the electron stoichiometry (n) in the reduction of UO 2 (L X ), the spectroelectrochemical measurements were performed. The UV-vis-NIR spectra of each system were recorded at different potentials (E). Fig. 5, S8 and S9 † show the obtained spectral variations at X ¼ O, NH, and S, respectively. As a general trend, the absorption bands of UO 2 (L X ) around 30 000 and 20 000 cm À1 gradually decreased with decreasing E, while new absorption bands appeared around 25 000 and 15 000 cm À1 . Moreover, isosbestic points were clearly observed, indicating that the redox equilibria of UO 2 (L X ) only take place in the current potential ranges. Using the absorbance at 24 630 or 24 876 cm À1 , the concentration ratio (C O /C R ) between the oxidant (UO 2 (L X )) and its reductant at each E was calculated.
The relationship between C O /C R and E should follow the Nernstian equation, eqn (2).
where E 0 , R, T, and F are the formal potential, the gas constant (8.314 J mol À1 K À1 ), the absolute temperature, and the Faraday constant (96 485 C mol À1 ), respectively. The slope and intercept of the linear relationship between E and ln(C O /C R ) (insets of Fig. 5(b), S8(b) and S9(b) †) allow to determine n and E 0 of the redox reactions of UO 2 (L X ). The estimated n values of UO 2 (L X ) are close to unity (Table S7 †), indicating that the reduction of UO 2 (L X ) affords [UO 2 (L X )] À . The E 0 values estimated from the spectroelectrochemical measurements (Table S7 †) agree with  those observed in the CV measurements (Tables S4-S6 †). UV-vis-NIR spectra of UO 2 (L X ) and [UO 2 (L X )] À in DMSO were summarized in Fig. 6. The spectral features of all UO 2 (L X ) are quite similar to each other. All UO 2 (L X ) showed characteristic bands around 28 000 and 24 000 cm À1 . These absorption bands were also observed in UO 2 2+ complexes with Schiff base ligands, and can be assigned to the p-p* transition bands of Schiff base ligands. 21,22 Therefore, the difference in X leads to no signicant differences in the electronic structures of UO 2 (L X ).
Even aer the reduction, [UO 2 (L X )] À with different X commonly have the intense bands at around 25 000 cm À1 with 3 $ 10 4 M À1 cm À1 and weak bands at 16 400, 14 500, 12 200 and 7200 cm À1 with 3 $ 10 2 M À1 cm À1 (Fig. 6). Note that [UO 2 (L NH )] À has a characteristic band at 5200 cm À1 , although this absorption is not clearly observed in [UO 2 (L O )] À and [UO 2 (L S )] À (Fig. 6). The absorption bands at around 25 000, 16 400, 14 500, 12 200 and 7200 cm À1 are generally observed in U V O 2 + complexes with Schiff base ligands as reported previously. 21,22 The intense absorption at 25 000 cm À1 is assigned to a p-p* transition in the Schiff base ligands and/or a ligand-to-metal charge transfer (LMCT). 21,22 In accordance with TD-DFT calculation, 21,22 the absorption band at 16 400 cm À1 is attributable to a metal-toligand charge transfer (MLCT) from a 5fd u orbital of the U 5+ center to the p* orbital of the coordinating ligand. Finally, those at 14 500, 12 200 and 7200 cm À1 are ascribed to the f-f transitions arising from the 5f 1 electron conguration of U 5+ . 21,22,35 [UO 2 (L NH )] À only exhibited the absorption band at 5200 cm À1 attributable to another f-f transition, 21,22,35 while this is not the case for the others studied here. As a matter of fact, this transition is not always clearly observable as we reported previously. 22 To theoretically support occurrence of U V O 2 + in each [UO 2 (L X )] À , we further performed DFT calculations of [UO 2 (L X )] À . Initially, the molecular structures of [UO 2 (L X )] À in DMSO were taken from those of UO 2 (L X ) determined by the X-ray crystal structure and were optimized aer addition of a single negative charge and doublet spin degeneracy to assume the one-electron reduction. The optimized structures of [UO 2 (L X )] À were shown in Fig. S10 † and the selected structural parameters are summarized in Table S2. † The U^O ax bond lengths of [UO 2 (L X )] À are 1.86 A, which are ca. 0.08 A longer than those of the corresponding UO 2 (L X ) determined by SCXRD ( [UO 2 (L X )] À is exclusively localized in the center U as shown in the Mulliken spin density surfaces (Fig. 7), clearly indicating that these reduced complexes are of U V O 2 + regardless of difference in X. Consequently, the X moiety does not largely affect the redox chemistry of [UO 2 (L X )] À/0 .

Conclusions
In this study, UO 2 (tBu,MeO-saldien-X) (UO 2 (L X ); X ¼ NH, O, S) were synthesized and structurally characterized to discuss impacts of the heteroatoms (X) to the coordination chemistry of UO 2

2+
. The crystal structures of UO 2 (L X ) showed the vecoordinated UO 2 2+ with L X 2À in the equatorial plane. The U^O ax bond length of UO 2 2+ and the bond length between U and phenolic O are not affected by the difference in X. On the other hand, the U-X bond length increases in the order of UO 2 (L O ) < UO 2 (L NH ) < UO 2 (L S ). Aer taking into account the differences in the atomic size of X, the normalized U-X bond strength in UO 2 (L X ) was found to follow U-O z U-NH > U-S. While the U-O and U-NH bond strengths are similar to each other, the weaker U-S interaction can be explained by the HSAB principle. The logarithmic conditional stability constant (log b X ) of UO 2 (L X ) in ethanol containing 0.4 mM NEt 3 decreases in the order of UO 2 (L NH ) (log b NH ¼ 10) > UO 2 (L O ) (log b O ¼ 7.24) > UO 2 (L S ) (log b S ¼ 5.2). This trend cannot be explained only by the HSAB principle, but rather follows the order of basicity of X. The theoretical calculations of UO 2 (L X ) suggested that the ionic character of U-X bonds decreases in the order of U-NH > U-O > U-S. In contrast, the covalency increases as U-O < U-NH < U-S.
No signicant differences were found in the electrochemistry of UO 2 (L X ) with different X in terms of E 0 and U-centered redox reaction. As demonstrated in this work, a UO 2

2+
-ligand bond strength does not always follow the HSAB principle, but is also affected by other factors such as Lewis basicity and balance between ionic and covalent interactions of donating atoms to the center metal. These points should be more carefully considered to design molecular structures of ligands suitable for hydrometallurgical separations of metal ions of interest.

Author contributions
T. T. devised the main conceptual ideas, carried out all experiments, and wrote the manuscript in consultation with K. T. K. T. supervised this project, discussed all experimental results with T. T., and edited the manuscript. All authors have given approval to the nal version of the manuscript.

Conflicts of interest
There are no conicts to declare.