Hydrothermal Synthesis and Structural Investigation of a Crystalline Uranyl Borosilicate

The relevance of multidimensional and porous crystalline materials to nuclear waste remediation and storage applications has motivated exploratory research focused on materials discovery of compounds, such as actinide mixed-oxoanion phases, which exhibit rich structural chemistry. The novel phase K1.8Na1.2[(UO2)BSi4O12] has been synthesized using hydrothermal methods, representing the first example of a uranyl borosilicate. The three-dimensional structure crystallizes in the orthorhombic space group Cmce with lattice parameters a = 15.5471(19) Å, b = 14.3403(17) Å, c = 11.7315(15) Å, and V = 2615.5(6) Å3, and is composed of UO6 octahedra linked by [BSi4O12]5− chains to form a [(UO2)BSi4O12]3− framework. The synthesis method, structure, results of Raman, IR, and X-ray absorption spectroscopy, and thermal stability are discussed.


Introduction
In the U.S., alkali borosilicate glasses are used as a primary nuclear waste form material for the immobilization of high-level waste. The glasses are durable and can host a wide range of elements in the disordered structures, which are composed of network forming SiO 4 and BO 4 tetrahedra as well as trigonal planar BO 3 anions that support compositional variety [1]. However, the low solubility of actinides in glasses with high actinide loadings has motivated researchers to study the crystal chemistry of actinide compounds, both to identify phases that may be forming under simulated conditions and to develop an understanding of the fundamental chemistry of waste form relevant phases [2,3]. Moreover, it has been recognized that for some waste streams, crystalline waste forms may offer an advantage over glasses such as higher waste loading capacities [4,5]. The structural complexity exhibited by multidimensional and porous crystalline materials can also be exploited, enabling the incorporation of specific radionuclides and, in some cases, ion exchange capabilities [6][7][8].
Among the many compound classes that have been studied for waste form applications, uranyl silicates and borates have received considerable attention. The variable arrangement of SiO 4 tetrahedra in uranyl silicate compounds has resulted in a multitude of different structure types, many of which are three-dimensional [9][10][11][12][13][14][15][16][17]. Likewise, the ability of BO 3 and BO 4 units to polymerize to form polyborate units has given rise to a rich structural chemistry in uranyl borates [3,[18][19][20][21][22][23][24]. It is therefore unsurprising that incorporation of mixed-oxoanions into the structures of uranyl compounds often facilitates further structural diversity. For example, a few uranyl borophosphate [25,26], borogermanate [27], aluminoborate [28], and aluminophosphate [29] compounds are known, most of which exhibit complex open-framework structures. No such uranyl compounds containing borosilicate units have been reported, although there are several existing non-uranyl compounds exhibiting a variety of different borosilicate units. While uranyl borosilicates are relevant to waste form applications, in the case of non-actinide borosilicates, the structural diversity afforded by the presence of the borosilicate units often promotes the formation of noncentrosymmetric structures, making some of these compounds potentially promising candidates as nonlinear optical materials [30]. Some of the known examples within these compound classes contain isolated BO 3 and SiO 4 structural units [31,32], condensed BO 4 and SiO 4 tetrahedra [33,34], trigonal planar BO 3 units condensed with tetrahedral BO 4 and SiO 4 [35], as well as HBO 4 and SiO 4 hydroxyborosilicate units [36].
Herein, we present the first example of a uranyl borosilicate compound, K 1.8 Na 1.2 [(UO 2 )BSi 4 O 12 ], which was synthesized using hydrothermal synthesis methods with supercritical water as the reaction medium. The structure of the compound was investigated, revealing residual electron density near the uranyl oxygen that was attributed to either the presence of a light atom in an unusual coordination or to an unidentified crystal defect, prompting further investigation of the compound using a combination of X-ray absorption and vibrational spectroscopy techniques, characterization of its thermal properties, as well as the employment of first principles calculations to determine the optimized geometry of the uranium coordination environment [37,38]. Using single crystal X-ray diffraction, the compound was found to crystallize in a centrosymmetric structure type with a complex three-dimensional crystal structure containing unique [BSi 4 O 12 ] 5chains composed of BO 4 and SiO 4 tetrahedra. Characterizations of the thermal and spectroscopic properties were performed using thermogravimetric analysis in addition to Raman, IR, and extended X-ray absorption fine structure spectroscopy. These data, in addition to the results of first principles calculations, suggest that the residual electron density in the structure is likely due to a crystal defect. The successful preparation of K 1.8 Na 1.2 [(UO 2 )BSi 4 O 12 ] serves to further expand on the diverse structural chemistry of uranium and moreover, highlights an under-explored set of synthetic conditions that may be exploited to obtain other novel actinide borosilicate compounds.

Synthesis
, and SiO 2 (300.4 mg) were added to a mineralizer solution of aqueous KOH (224.4 mg in 2 mL deionized water) in an approximate molar ratio of 1:10:10:8 of U:Na,B:Si:K. A 5% by mass excess of UO 3 ·xH 2 O was added to the reaction mixture to compensate for the unknown degree of hydration of the uranium starting material. The mixture was sealed in a 12.7 cm silver tube that was loaded into a high-pressure vessel containing approximately 20 mL of water to serve as counterpressure. The vessel was placed in a programmable oven and heated to 400 • C for 48 h, followed by a period of slow cooling to 350 • C at a rate of 3 • C per hour. The vessel was then allowed to naturally cool to room temperature. An estimated 30 MPa of pressure was generated during the reaction based on the pressure-temperature diagram of water.
Upon cooling, the silver tube was cut open and the mother liquor was decanted. The contents of the tube were sonicated in deionized water prior to vacuum filtering and thoroughly washing the products in water and acetone. The product mixture consisted of brown/black polycrystalline powders of unknown composition in addition to clusters of large yellow prismatic crystals ( Figure S1) in approximately 70% yield. The large yellow prismatic crystals were manually separated from the product mixture and ground to a powder to obtain a phase pure sample for bulk property measurements.
To obtain high quality single crystals suitable for single crystal X-ray diffraction measurements, efforts were made to optimize the reaction conditions by varying the hydroxide concentration. In order to have more precise control over the KOH concentration, subsequent attempts employed identical reaction conditions with substitution of KOH pellets for KOH solution diluted from a 45% w/w stock solution. Reactions containing 2 mL of 2M KOH solution (1:10:10:8 = U:Na,B:Si:K) resulted only in the formation of brown polycrystalline material, identified as a mixture of K(UO)Si 2 O 6 [17] and Na 7 UO 2 (UO) 2 (UO 2 ) 2 Si 4 O 16 [39] via powder X-ray diffraction. In comparison, reactions containing 2 mL of 1.5M KOH solution (1:10:10:6 = U:Na,B:Si:K) resulted in polycrystalline Na 7 UO 2 (UO) 2 (UO 2 ) 2 Si 4 O 16 along with an approximately 40% yield of high quality single crystals of K 1.8 Na 1.2 [(UO 2 )BSi 4 O 12 ] that were used for single crystal X-ray analysis.

Single Crystal X-ray Diffraction
Single crystal X-ray diffraction data were collected at 301(2) K using a Bruker D8 QUEST diffractometer equipped with an Incoatec microfocus source (Mo Kα radiation, λ = 0.71073 Å). Data were integrated and corrected for absorption effects using SAINT+ and SADABS programs [40]. An initial structure model was obtained with SHELXT and subsequently refined with SHELXL-2018 using the ShelXle interface [41,42]. The ADDSYM program implemented into PLATON software was used to check for possible missing symmetry, and no alternative symmetry was found [43]. Crystallographic data for K 1.8 Na 1.2 [(UO 2 )BSi 4 O 12 ] are provided in Table 1. A residual electron density peak of 4.17 e/Å 3 is located 1.34 Å from the uranyl oxygen and 1.82 Å from the uranium atom, along with a second peak of 2.25 e/Å 3 located trans to the 4.17 e/Å 3 peak across the uranium atom, suggesting the oxygen atoms of the uranyl group form a disordered crystallographic domain. No twinning was identified, and several different crystals were screened but the residual electron density was present in each structure solution. Single crystal X-ray data were also collected on the large crystals obtained from initial reactions utilizing KOH pellets rather than KOH stock solution, and refinement of the structure suggested a marginal difference in composition, K 1.7 Na 1.3 [(UO 2 )BSi 4 O 12 ], although with poor refinement statistics (Table S1). Similar to K 1.8 Na 1.2 [(UO 2 )BSi 4 O 12 ], residual electron density was also observed in the structure solution of K 1.7 Na 1.3 [(UO 2 )BSi 4 O 12 ]; however, only a single large peak of 10.35 e/Å 3 located 1.33 Å from the uranyl oxygen and 1.94 Å from the uranium atom was observed. Several different crystals were screened, and the same result was obtained in each case. The residual electron density persisted in data collected at low temperature as well as in structure solutions obtained in lower symmetry space groups, and no twinning was identified.

Topological Analysis
The ToposPro software package (Blatov, Shevchenko, Samara, Russia) was used to perform crystal structure analysis [44,45]. The ADS program was used to obtain underlying nets using the standard structure simplification procedure [46].

Powder X-ray Diffraction
Powder X-ray diffraction (PXRD) measurements were performed on a Bruker D2 Phaser diffractometer (Cu Kα radiation, λ = 1.54184 Å) equipped with a LYNXEYE XE-T detector. Data were collected over a 2θ range of 5-65 • with a step size of 0.04 • . PXRD data confirmed the phase purity of the obtained samples.

Spectroscopy
Raman spectroscopy measurements were performed on a Horiba XploraPLUS Raman microscope (Horiba, Kyoto, Japan) that was used to acquire spectra over the range of 525 to 950 cm −1 with a 638 nm laser as the excitation source. Scans were performed at 25% laser power and were collected on approximately 15 mg of powder sample that was deposited between two microscope slides and sealed on all sides with tape as containment during the measurement. Fourier-transform infrared spectroscopy data were collected on a powder sample using a PerkinElmer Spectrum 100 FT-IR spectrometer (PerkinElmer, Waltham, MA, United States) in the range of 650 to 4000 cm −1 .
Extended X-ray absorption fine structure (EXAFS) measurements were made using the MRCAT beamline, sector 10-BM-A,B at the Advanced Photon Source [47]. Measurements were made using the uranium L III edge at 17.16 keV in transmission. Measurements were made for the new uranium borosilicate phase of interest and also for γ-UO 3 and α-U 3 O 8 as standards. Data analysis was performed using the Athena and Artemis programs from the Demeter package [48] with a k weight of 3 and usable data range to 12.5 Å −1 .

Thermal Properties
Thermogravimetric and differential thermal analysis (TGA/DTA) measurements were performed on a polycrystalline powder sample using a TA SDT Q600 TGA (TA Instruments, New Castle, DE, Unites States). The sample was heated from room temperature to 800 • C under a flow of nitrogen gas at a purge rate of 100 mL per minute, and the resulting powder was analyzed by PXRD for phase identification after heating.

First Principles Calculations
We performed first principles calculations to carry out geometry optimizations in the form of density functional theory (DFT), using the Vienna Ab-initio Package (VASP) planewave code (VASP Software GmbH, Vienna, Austria) [49,50] generalized gradient approximation of Perdew, Burke and Ernzerhof (PBE) [51] and projector augmented wave (PAW) method [52,53]. Spin-polarized calculations were performed with 520 eV cut-off energy for the plane wave basis set, 10 −4 eV energy convergence criteria and 3 × 3 × 3 k-point mesh. The ground state geometries at 0 K were optimized by relaxing the cell volume, atomic positions, and cell symmetry until the maximum force on each atom was less than 0.01 eV/Å. Implementation of Hubbard U correction of 4.0 eV for uranium f -orbitals resulted in similar optimized geometries.

Synthesis
Many factors are known to influence crystal growth in supercritical hydrothermal synthesis, including pH, temperature and pressure, mineralizer type, reactant solubilities, solution redox potential, reaction container type, etc. [54,55]. While there is valuable information to be gleaned from analysis of the reaction processes occurring in these conditions, such studies are often precluded by the presence of soluble intermediate species [56]. In the absence of the ability to perform mechanistic studies, part of our investigations of this system aimed to identify key factors dictating phase formation. Our findings indicated that the two predominating factors influencing the formation of K  [17]. Longer reaction times did not result in appreciable increase in yield or larger single crystals of K 1.8 Na 1.2 [(UO 2 )BSi 4 O 12 ]. The discrepancy between the products obtained from reactions beginning with KOH pellets versus KOH stock solution served as an indication of the relatively narrow range of conditions in which K 1.8 Na 1.2 [(UO 2 )BSi 4 O 12 ] will form, where phase formation did not occur in reactions with KOH concentrations greater than 1.5 M. The optimal concentration range was determined to be approximately 0.5-1.5 M. The asymmetric unit consists of one uranium atom, four potassium atoms, one sodium atom, two silicon atoms, one boron atom, and nine oxygen atoms. The structure is composed of isolated UO 6 octahedra that are coordinated through vertexsharing of the equatorial oxygen atoms to [BSi 4 O 12 ] 5− chains, forming a framework with channels in which disordered Na + and K + cations reside (Figure 1). The borosilicate chains are made up of BO 4 and SiO 4 tetrahedra that run parallel to the c axis, where each BO 4 tetrahedron shares all four vertices with SiO 4 tetrahedra within the chain and each SiO 4 tetrahedron shares vertices with two adjacent SiO 4 tetrahedra, one BO 4 tetrahedron, and a distorted UO 6 octahedron. A table of selected interatomic distances is provided in the supporting information in Table S2.  Figure 2a. The borosilicate FBB is an open-branched dreier-single ring that may be described as 5 :<4 > according to the notation proposed by Burns, Grice, and Hawthorne [57,58]. The simplified underlying net of the borosilicate chain topology is shown in Figure 2b. A variety of borosilicate units have been observed in non-uranyl compounds [31][32][33][34][35][36]; to the best of our knowledge, K 1  Within the larger voids of the [(UO 2 )BSi 4 O 12 ] 3− framework, K + cations are disordered over three sites that have a combined total occupancy of one, whereas the smaller channels contain disordered Na + /K + cations located on a single site staggered between UO 6 octahedra, with occupancies of Na + and K + refining to 0.61 and 0.39, respectively. It was expected that larger alkali cations could be incorporated into the structure considering the extensive disorder of the K + cations in the large channels of the framework; however, attempts to synthesize the Rb and Cs analogs were unsuccessful.

Structure Description
The U-O bond lengths within the distorted UO 6 octahedra are typical, with the equatorial bond distances ranging from 2.233 (2) (Figure 3). Using this model that takes into account the coordination of a peroxo group to the uranium atom, the BVS at the uranium site was calculated to be 6.67. Selected interatomic distances for K 1.7 Na 1.3 [(UO 2 )BSi 4 O 12 ] are provided in Table S3. While many uranyl peroxo compounds are known [59][60][61][62][63][64], reported examples are characterized by peroxo interactions existing within the equatorial plane; none that exhibit peroxo interactions along the axial direction have been reported. Furthermore, O-O bond lengths in uranyl peroxo compounds typically range between about 1.45 to 1.52 Å [65], and thus, the significantly shorter peroxo bond in this model (1.33 Å) warrants some skepticism. As an alternative to a peroxo group, the observed residual electron density might simply be an artifact from an unknown crystal defect, possibly relating to the incorporation of slightly more Na + on the disordered Na + /K + site in K 1.7 Na 1.3 [(UO 2 )BSi 4 O 12 ] than in K 1.8 Na 1.2 [(UO 2 )BSi 4 O 12 ]. Given the novelty of such a coordination environment for uranium, as well as this possibility of a crystal defect, further analysis was needed to either confirm or reject the peroxo structure model.

Spectroscopy
In order to probe for a peroxo group in the structure of K 1.7 Na 1.3 [(UO 2 )BSi 4 O 12 ], Raman spectroscopy was performed. Shown in Figure 4, the predominating feature in the Raman spectrum is a band at 776 cm −1 , corresponding to the symmetric stretching of the uranyl moiety. This is consistent with the uranyl bond lengths determined from the single crystal X-ray data [66]. A few weak bands are observed in the spectrum between 548 to 744 cm −1 and are attributable to BO 4 and SiO 4 bending modes [34,67]. For η 2 -peroxo complexes, the O-O symmetric stretching is normally observed as a strong band near 870 cm −1 [68], although this band has been observed to occur anywhere in the range of approximately 830 to 880 cm −1 in the spectra of uranyl peroxo compounds [69,70]. This region is bare in the Raman spectrum of K 1.7 Na 1.3 [(UO 2 )BSi 4 O 12 ], with the exception of a weak band centered at approximately 830 cm −1 . Even in the case of a partially occupied peroxo group, it is expected that a band corresponding to the symmetric stretching mode would be fairly intense, thus we attribute the weak band at 830 cm −1 to the antisymmetric stretching of the uranyl moiety [16,71].
The IR spectrum of K 1.7 Na 1.3 [(UO 2 )BSi 4 O 12 ] is complex, with significant overlap of the absorption bands of BO 4 , SiO 4 , and UO 2 2+ observed. The spectrum, shown in Figure 5, exhibits two bands at 686 and 750 cm −1 that are attributed to the BO 4 and SiO 4 bending modes, in addition to antisymmetric and symmetric stretching absorptions of BO 4 and SiO 4 in the region of 750 to 1134 cm −1 [35,72]. The vibrational band at 901 cm −1 is assigned to the uranyl antisymmetric stretching [72,73].  EXAFS was used to further probe the presence or absence of the peroxo oxygen in a 'side-on' configuration. The first coordination shell if the peroxo group is present has two oxygen ions at a distance of~1.8 Å from the central uranium, with four additional oxygen ions at~2.25 Å. The peroxo group oxygen is present at 1.95 Å, approximately halfway between the other oxygen distances. The structure model without the peroxo group has the same first-shell configuration but without the peroxo oxygen. Both structure models have nearly identical atom locations, with the primary difference being the presence of the peroxo oxygen, so it was only necessary to fit the first shell of the EXAFS data to distinguish between them.
First-shell fits were successful for the γ-UO 3 and α-U 3 O 8 materials, and the fit to the new uranium borosilicate was good with or without the peroxo group, with R-values of 0.026 with the peroxo group and 0.022 without ( Figure 6). For the model without the peroxo ion, the 4-parameter fit yielded sensible values with shifts of the oxygen ions of less than 0.01 Å and mean square relative displacement of 0.0020 (7). However, the model for the peroxo group moved the peroxo oxygen to 2.13 Å while the initial model placed the oxygen ion at 1.95 Å. This result indicates that the peroxo oxygen is not a feature of the first coordination shell and that the model with only 6-fold coordination of the uranium ion is correct.

Thermal Properties
TGA measurements were performed to assess the thermal behavior of K 1.7 Na 1.3 [(UO 2 ) BSi 4 O 12 ]. It was expected that if the peroxo oxygen in question were present in the structure, a distinct weight loss step associated with loss of an oxygen would be observed. Based on a 40% occupancy of the peroxo oxygen as determined from structure refinements, a weight loss of approximately 0.9% was calculated. As shown in Figure 7, a total weight loss of only 0.45% was observed over the measurement range of 25 • C to 800 • C. PXRD patterns collected on TGA samples after heating revealed the material to be thermally stable up to 800 • C, with no significant changes observed in the patterns before and after heating ( Figure 8). To verify the weight loss observed was not attributable to the loss of oxygen from a peroxo group, single crystal X-ray data were collected on a single crystal sample that was synthesized and subsequently heated to 800 • C. The residual electron density was still present in the structure model, and no other differences were observed in the structure model obtained for the sample heated post-synthesis.

Energy Optimization Calculations
First principles calculations were used to optimize the geometry of the structure in order to determine the optimized U-O bond lengths of the uranyl oxygens and the peroxo oxygen. The results of the calculations are provided in Table 2. Three sets of calculations were performed in which the structure contained a fully occupied peroxo oxygen, a halfoccupied peroxo oxygen, and no peroxo oxygen, allowing for a comparison of uranyl bond distances as a function of peroxo oxygen occupancy. The results of the calculations revealed the optimal U(1)-O peroxo and U(1)-O(1) bond distances for the structure containing a half-occupied peroxo oxygen should be significantly elongated at approximately 2.17 and 2.16 Å, respectively. Only nominal differences in calculated bond lengths between the structure containing half-and fully occupied peroxo oxygen were determined, suggesting that any peroxo oxygen occupancy would result in elongated U-O bonds. In comparison to the experimentally determined bond lengths, the calculated bond lengths for the structure containing no peroxo oxygen are still slightly elongated but are more consistent with the structure model obtained from single crystal X-ray data. Thus, in combined consideration of the structure, thermal, and spectroscopy data, the calculations better support the structure model exhibiting no axially coordinated peroxo.

Conclusions
In summary, hydrothermal synthesis methods were employed for the preparation of the first example of a uranyl borosilicate phase, K 1 4 and SiO 4 tetrahedra, which constitutes a new borosilicate FBB. This highlights the structural diversity that is accessible in this new class of compounds, warranting further exploration of crystalline uranyl borosilicate compounds.
Successful synthesis of the phase was determined to occur under a relatively narrow set of conditions, where slight deviations in the KOH concentration of the reaction was initially thought to produce a related but distinct secondary phase with a possible partially occupied peroxo group in place of one of the uranyl oxygen atoms. The composition of this phase-with no peroxo accounted for-was determined to be K 1.7 Na 1.3 [(UO 2 )BSi 4 O 12 ] from single crystal X-ray data. No evidence of O 2 2− symmetric stretching was observed in the Raman spectrum and first-shell fits of the EXAFS spectrum were consistent with a UO 6 coordination lacking a peroxo group. No loss of peroxo was observed upon heating the sample in TGA measurements, and in fact, no thermal decomposition was found to occur over the measurement range up to 800 • C. Furthermore, first principles calculations indicated the presence of a peroxo group should result in significantly longer U-O bonds than those determined experimentally. We therefore conclude the residual electron density in the structure models are attributable to a crystal defect and not a peroxo group.

Conflicts of Interest:
The authors declare no conflict of interest.