Tuning Amidoximate to Enhance Uranyl Binding: A Density Functional

Oct 18, 2013 - This study investigates the electronic effects of amidoximate ligands bound to the uranyl cation, UO22+. Density functional theory ... ...
0 downloads 6 Views 1MB Size
Article pubs.acs.org/JPCA

Tuning Amidoximate to Enhance Uranyl Binding: A Density Functional Theory Study Carter W. Abney,† Shubin Liu,*,‡ and Wenbin Lin*,† †

Department of Chemistry, The University of Chicago, 929 E. 57th St., Chicago, Illinois, 60637 United States Research Computing Center, University of North Carolina, Chapel Hill, North Carolina, 27599 United States



S Supporting Information *

ABSTRACT: Amidoxime functionalized sorbents have shown great promise in extracting uranium from seawater, though the rationale for this affinity is not apparent. To enhance binding by amidoxime and to develop more selective sorbents, a detailed understanding of the electronic structure is necessary. This study investigates the electronic effects of amidoximate ligands bound to the uranyl cation, UO22+. Density functional theory calculations have been performed on a series of uranyl−amidoximate derivatives to investigate their structural, electronic, and thermochemical properties. The computational findings are in good agreement with available experimental data, with average error in bond length below 0.07 Å for all systems. Binding strength was observed to be directly related to electron donation, as evidenced by the plot of log(K/K0) vs the Hammett constant (σpara) of the substituent adjacent to the oximate function. From this observation, we propose and investigate two new imidazole-derived oximes, both of which possess greater binding strength than amidoximate derivatives.



INTRODUCTION The extraction of uranium from the ocean is a desirable means of sustaining the nuclear fuel cycle. It is estimated over 4 billion metric tons of uranium are available in seawater,1 more than 1000 times the amount available in terrestrial ores, providing adequate fuel for nuclear power generation through the next millennium.2 While over 50 years of research have focused on using adsorbent materials for uranyl (UO22+) sequestration, the ultralow concentration (estimated around 3.3 ppb),3 and the presence of numerous competing ions make efficient extraction a formidable challenge. Amidoxime-based sorbents have been studied extensively and remain the premier function for selective sequestration of uranium. Initially identified during systematic screening studies with organo-functionalized ion-exchange materials,4,5 field tests with amidoxime-functionalized polymers later demonstrated a sorption capacity of up to 1.5 g uranium extracted per kg of sorbent per 30 days.6−10 Potentiometric titrations of model ligands acetamidoxime and benzamidoxime revealed deprotonation prior to uranyl binding, resulting in the formation of neutral UO2(L)2 complexes.11,12 The binding motif for amidoxime chelation to uranyl was recently discovered through a combination of density functional theory (DFT) calculations and single-crystal X-ray diffraction studies, where an η2 coordination through the N−O bond of amidoximate was most thermodynamically favored.13 An identical binding interaction was observed between uranyl and an amidoximefunctionalized ionic liquid,14 and a survey of the Cambridge Crystal Structure Database revealed all related uranyl−oximate structures to adopt a similar η2 orientation.13 © 2013 American Chemical Society

While these previous studies have demonstrated how amidoxime coordinates to uranyl, questions remain as to why it exhibits such a strong affinity and, more importantly, how to improve upon the amidoxime functional group for uranium sequestration. It is anticipated improving the UO22+ binding strength of the sorbent as well as development of well-designed multitopic chelators would significantly improve the properties of sorbent materials. Several studies have revealed the tridentate chelating cyclic imide dioxime ligand to bond more strongly than the open chain diamidoxime,15−17 and complementary geometry for amidoxime-decorated hosts has very recently been investigated through de novo structure-based computational screening.18 However, detailed investigations of amidoxime− uranyl binding interactions have been largely neglected. We have performed DFT calculations to probe the structural, electronic, and thermochemical properties for a series of amidoxime-derived uranyl complexes. We observed bond strength correlates directly with the resonance electron donating properties of the substituent adjacent to the uranylbinding oximate group. From this observation, we propose two new promising amidoxime-derived sorbents for further investigation in extracting UO22+ from seawater.



THEORETICAL METHODS Calculations were executed with the Gaussian 09c01 package19 using DFT at the B3LYP level of theory.20,21 The Stuttgart RSC Received: August 23, 2013 Revised: October 17, 2013 Published: October 18, 2013 11558

dx.doi.org/10.1021/jp408460x | J. Phys. Chem. A 2013, 117, 11558−11565

The Journal of Physical Chemistry A

Article

Figure 1. Optimized structures for [UO2(H2O)5]2+ (left) and [UO2(AO)2(MeOH)2] (right), with the latter displaying η2 binding of amidoximate (AO) to uranyl. Red, white, gray, dark blue, and light blue spheres represent O, H, C, N, and U, respectively. An identical color palate is used in all subsequent figures.

ments39 of [UO2(H2O)5]2+ and bond lengths obtained by single crystal X-ray analysis13 of [UO2(AO)2(MeOH)2] reveal that the computational method reliably predicted the structural parameters (average errors of 0.062 and 0.049 Å, respectively). For both systems, bond lengths were generally overestimated with the largest discrepancies for the equatorially coordinated solvent molecules. Average distances for U-solvent coordination differed from experimental results by 0.082 and 0.142 Å for [UO2(H2O)5]2+ and [UO2(AO)2(MeOH)2]. All other structural discrepancies were less than 0.05 Å in deviation. In an effort to obtain more accurate structural values, the calculation for [UO2(H2O)5]2+ was repeated with solvation effects taken into consideration as experimental values for [UO2(H2O)5]2+ were obtained by EXAFS while in water. Consideration of solvent effects improved the predicted structural parameters relative to the gas phase calculations, with average errors decreasing to 0.036 Å. The influence of solvation significantly altered the axial UO distance for [UO2(H2O)5]2+, bringing the bond length within 0.002 Å of the experimentally determined value and the U-solvent distance to within 0.05 Å of the experimental value. Selected average bond lengths are displayed in Table 1. The mother liquor used for the crystal preparation of [UO2(AO)2(MeOH)2] was a mixture of unspecified volumes of methanol, nitromethane, and dichloroethane, making mean-

1997 relativistic effective core potential (ECP) was used for uranium, replacing the 60 core electrons and representing valence electrons by a contracted [8s/7p/6d/4f] basis.22 For all light atoms (carbon, nitrogen, oxygen, fluorine, and hydrogen), a 6-311+G* basis set was used.23,24 Spin−orbit interactions were not considered explicitly, and the self-consistent field was set to tight, quadratically convergent and with an extra step in the event the first order SCF did not converge (XQC). These computational parameters are known to yield accurate geometries and energetics for actinyl complexes.13,25−28 Structures were first optimized and then frequency calculations performed to confirm geometries and obtain thermochemical data. Optimization calculations were performed both in gas phase and in the solvated state, though discussion focuses on values obtained for the solvated complexes. The influence of solvation was modeled by performing calculations with the integral equation formalism model (IEFPCM)29 using a polarizable conductor calculation model (CPCM).30,31 Previous investigations demonstrated solvation effects to be accurately represented by continuum models, assuming the first coordination sphere was filled, with explicit inclusion of additional solvent molecules having modest effects.27,32 Binding enthalpies (ΔH) and Gibbs free energies (ΔG) were calculated with zero-point energy (ZPE) and thermal corrections. Wiberg bond indices (WBIs)33 were determined by natural bond orbital (NBO) analysis34−36 for solvated systems at the same level of theory. Energy decomposition analysis was performed with the AOMix software platform, version 6.81, using the FO execution option.37,38 AOMix calculates molecular orbital compositions in terms of constituent chemical fragments. Following initial structural optimization, uranyl−amidoximate complexes were separated into two fragments: (1) uranyl and equatorial aquo ligands and (2) amidoximate or amidoximate-derived ligands. All energy decomposition calculations were performed including influence of aqueous solvation by IEFPCM using the same basis set and to the same level of theory as all aforementioned calculations.

Table 1. Selected Average Bond Lengths (Å) for [UO2(H2O)5]2+ and [UO2(AO)2(MeOH)2] gas phase 2+

[UO2(H2O)5] UO U−OH2 avg. error

Δ

calcd 1.748 2.492

calcd

0.012 0.082 0.062 gas phase

Δ

1.758 2.460

0.002 0.050 0.036 solvated (MeOH)

exptlb

calcd

Δ

calcd

Δ

UO U−OMe U−N U−O O−N CN avg. error

1.789 2.458 2.398 2.383 1.409 1.290

1.789 2.600 2.440 2.357 1.369 1.288

0.000 0.142 0.042 0.026 0.040 0.002 0.049

1.806 2.579 2.429 2.334 1.378 1.292

0.017 0.121 0.031 0.049 0.031 0.002 0.042

a

11559

1.76 2.41

solvated

[UO2(AO)2(MeOH)2]



RESULTS AND DISCUSSION The reliability of the proposed theoretical method was tested first by gas phase calculations on uranyl pentahydrate [UO2(H2O)5]2+ and uranyl diacetamidoximate with two coordinated methanol molecules [UO2(AO)2(MeOH)2] (AO = acetamidoximate), shown in Figure 1. EXAFS measure-

exptl

a

Experimental data from ref 39. bExperimental data from ref 13. dx.doi.org/10.1021/jp408460x | J. Phys. Chem. A 2013, 117, 11558−11565

The Journal of Physical Chemistry A

Article

Figure 2. Select molecular orbitals for [UO2(AO)(H2O)3]+1. Subscript LP represents a lone pair for orbitals 55, 57, and 58. Orbitals 44, 54, 55, 57, and 58 comprise the majority of bonding interactions between AO and uranyl. The absence of bonding in 56 between π(O−N) and 3σu is due to symmetry considerations, which is similar for orbital 50 (not shown). Geometries were optimized with solvation effects.

with the uranium 5f orbital 1φu (orbital 57), the oxygen lone pair with UO22+ 3σu (orbital 55), and the oxygen lone pair and CN π-bonding orbital with the 5f orbital 1δu (orbital 58). Several important orbital interactions are displayed in Figure 2, and molecular orbital (MO) energies and character are summarized in Table 2. Additional MOs for AO, uranyl, and the uranyl−AO complex are available in the Supporting Information.

ingful consideration of solvent effects difficult. Calculations of [UO2(AO)2(MeOH)2] were performed including effects of solvation in methanol, yielding a smaller deviation in the Usolvent distance than calculations performed in the gas phase and an average error of 0.042 Å. These bond lengths are displayed in Table 1. Similar calculations performed with dichloroethane and nitromethane further improved the Usolvent distance, but at the expense of the amidoximate O−N bond length, resulting in average errors of 0.068 and 0.067 Å, respectively. Similar bond lengths and average errors were obtained from varying the dielectric coefficient using water as the solvent, or from averaging the dielectric coefficients for methanol, dichloroethane, and nitromethane. These data are available in the Supporting Information, Tables S1−S3. In aqueous solutions, UO22+ is coordinated by five aquo ligands in the equatorial plane. This has been demonstrated both computationally40 and experimentally,39 though prior studies also demonstrate a proclivity to hexacoordination in the gas phase.27 Previous work performed on AO−uranyl complexes demonstrates η2 coordination displaces two equatorial aquo ligands to obtain the thermodynamically favored structure.13 Calculations were performed for one AO ligand coordinating to UO22+, with aquo ligands completing the inner coordination sphere. The valence orbitals for UO22+ in aqueous solution are σ(f), π(f), σ(d), and π(d) at −11.1, −11.9, −12.2, and −12.2 eV.41 A recent DFT study of the uranyl−peroxo system revealed the σ and π bonding orbitals of a simple peroxo ligand to be lower in energy than the uranyl σ(d) and π(d), with covalent interactions occurring between the peroxo lone pair orbitals and the uranyl 5f orbitals.28 As the oximate function is partly composed of two adjacent π-conjugated Lewis bases, similar to the peroxo ligand, we expected similar electronic structure in uranyl−AO complexes. However, the incorporation of the adjacent amine and methyl group result in electron donation, raising the energy level of the oximate orbitals. While the σ bonding orbitals of N−O have no contribution to bonding between the oximate function and uranium, similar to peroxide, the π bonding orbitals were observed to be higher in energy than the uranyl σ(d) and π(d) and thus had significant bonding interactions. Most notably, the conjugation of O−N π-orbitals overlap constructively with the 1πg and 2πu orbitals of UO22+ (orbitals 44, 54). Other major bonding contributions come from interaction between the lone pairs on oxygen and nitrogen

Table 2. Energies and Characters of the MOs for [UO2(AO)(H2O)]+1 MO

energy (eV)

character

58 57 56 55 54 53 52 51 50 49 48 47 46 45 44

−9.330 −11.185 −11.618 −12.418 −12.747 −13.142 −13.435 −13.466 −13.524 −13.726 −13.794 −14.011 −14.220 −14.404 −14.411

OLP, π(CN)/1δu OLP, NLP/1φu π(O−N)/3σu OLP/3σu π(O−N)/2πu 2πu 2πu 2πu π(O−N−C−N)/1πg 3σg 3σg 1πg OLP, π(N−C−N)/3σg σ(H3C−C) π(O−N)/1πg

To obtain further insight into the bonding nature of the uranyl−AO complex, Wiberg bond indices (WBIs) were calculated using natural bond orbital (NBO) analysis. WBI values between 0.1 and 0.5 are regarded as largely ionic, whereas covalent bonds have WBIs with values in the vicinity of 1. For example, WBIs for bonds between UO22+ and phosphine oxides used in nuclear fuel reprocessing were calculated to be below 0.393, indicative of ionic bonding and electrostatic interactions,27 while a phosphoryl−urea complex used to extract UO22+ from seawater simulant yielded a higher WBI of 0.629, suggestive of some covalent character.42 These phosphine-based ligands often coordinate through the phosphoryl oxygen in a monodentate fashion, yielding only one WBI value. In contrast, AO binds through both the oximate 11560

dx.doi.org/10.1021/jp408460x | J. Phys. Chem. A 2013, 117, 11558−11565

The Journal of Physical Chemistry A

Article

O and N, yielding WBIs of 0.885 and 0.576, respectively, and a combined WBI of 1.461. NBO analysis revealed bonds between U and hybridized orbitals from light elements were 85.6% O and 86.1% N character, whereas the interaction between U and the O porbital was 92.8% localized on O. The localization of these orbitals suggests that modulating energy levels could result in more equal electron contribution and a stronger bond. A conjugated system of π orbitals connects the oxime function to the adjacent amine, providing a means for electron donation through mesomeric effects. This observation lead us to speculate uranyl−amidoximate bond strength could be tuned by varying the adjacent substituent. By increasing electron donation, the energy level of the bonding MOs should increase to afford a greater contribution from uranyl and to yield a more thermodynamically stable bond. To investigate this hypothesis, we calculated the thermodynamics for a series of oximatederived ligands coordinated to UO22+ and probed the bonding characteristics by NBO analysis. Adjacent substituents on the oximate-based ligands were selected on the basis of their electron donating character via resonance. The series of ligands is displayed in the inset of Figure 3.

Observing the optimized structures of the uranyl−oximate complexes, several key bond lengths and angles are strongly influenced by the strength of the uranyl−oxime bond. A systematic weakening of UO bonds are observed as πdonation is increased, lengthening from 1.772 Å for an adjacent nitro group to 1.789 Å for dimethyl amino. The decrease in UO bonding can also be quantified by WBIs, with values of 2.25, 2.24, and 2.22 for uranyl complexes with nitrooximate, oximate, and amidoximate, respectively. Additionally, the O UO bond angle appears to generally decrease with increased electron donation, though there are several exceptions to this trend. As expected, U−O and U−N bond lengths also decrease as a function of bond strength and thus electron donation.15 A summary of key bond lengths and angles is presented in Table 3. Inspection of the molecular orbitals involved in uranyl− oximate binding reveals several important interactions to be affected by the nature of the adjacent substituent. First, electron withdrawing functions lowered the energy of the MOs involved in bonding with uranyl. Substituting −NO2 for −NH2 lowered the energy of the bonding orbitals by an average 1.34 eV, corresponding with weaker bonding as a result of mismatched orbital energies. Second, the contribution from the oxime oxygen lone-pair to the uranyl 5f basis functions is directly related to the extent of electron donation, as can be seen by comparing orbitals for uranyl−amidoximate, uranyl−oximate, and uranyl−nitrooximate. While minor interaction can qualitatively be observed for all ligands with electron donating functions, NBO analysis suggests bonding occurs between the oxygen p-orbital and uranyl on ligands with strong electron donating groups (−NH2, N(H)CH3, and N(CH3)2). Finally, electron donation affects the delocalization of the oximate N− O π orbitals into those of the uranyl oxygen lone pairs. This secondary interaction, most clearly observed in orbital 44 of the uranyl−AO complex (Figure 2), can be probed by examining the WBIs for the oximate N−O interactions with the uranyl oxygen. Values of 0.0171, 0.0190, and 0.0192 were obtained for uranyl−oxygen bonding with nitrooximate, oximate, and amidoximate complexes, respectively. Relevant MOs and corresponding energies for AO, oximate, and nitrooximate are provided in Figures S2, S4−S5 in the Supporting Information, while a comparison between MOs is provided in Figure S6 in the Supporting Information. WBI values for all uranyl−oximate complexes are provided in Table S5 in the Supporting Information. To elucidate the relationship between electron donation and bond strength, ΔG values obtained by DFT calculation were used to calculate equilibrium constants from the following formula:

Figure 3. Plot of log(K/K0) vs Hammett σpara constants for the adjacent substituents on oximate-derived ligands binding to uranyl. K0 is determined from the uranyl−oximate complex, where H is the adjacent substituent. Values for K were obtained from calculated values of ΔG. The red line is the linear regression with R2 = 0.9773. The point corresponding to the dimethyl amine substituent (N(Me)2) was not included in calculating the regression due to the sterically induced distortion in orbital conjugation. Inset: The scope of oximate-derived ligands, as bound to UO22+ in an η2 fashion. Three aquo ligands complete the coordination sphere for all complexes.

ΔG = −RT (ln K )

Table 3. Representative Bond Lengths (Å) and UO Angle (°) for Uranyl−Oximate Complexes substituent

UO