A multimodal X-ray spectroscopy investigation of uranium speciation in ThTi2O6 compounds with the brannerite structure

ThTi2O6 derived compounds with the brannerite structure were designed, synthesised, and characterised with the aim of stabilising incorporation of U5+ or U6+, at dilute concentration. Appropriate charge compensation was targeted by co-substitution of Gd3+, Ca2+, Al3+, or Cr3+, on the Th or Ti site. U L3 edge X-ray Absorption Near Edge Spectroscopy (XANES) and High Energy Resolution Fluorescence Detected U M4 edge XANES evidenced U5+ as the major oxidation state in all compounds, with a minor fraction of U6+ (2–13%). The balance of X-ray and Raman spectroscopy data support uranate, rather than uranyl, as the dominant U6+ speciation in the reported brannerites. It is considered that the U6+ concentration was limited by unfavourable electrostatic repulsion arising from substitution in the octahedral Th or Ti sites, which share two or three edges, respectively, with neighbouring polyhedra in the brannerite structure.

of XPS.Interestingly, a higher fraction of U 4+ was determined from XPS data acquired from a polished surface of (Th 0.55 U 0.30 Ca 0.15 )Ti 2 O 6 , compared to a fracture surface, which was attributed to enrichment of Ca 2+ and U 5+ and/or U 6+ at grain boundaries.
To further elucidate the incorporation and stability of U 6+ in the brannerite structure, we designed three solid solutions based on ThTi 2 O 6 , targeting U 5+ or U 6+ speciation, with charge compensation by co-substitution on the Th or Ti site: Two additional compositions were also examined, reflecting the substitution of U for Th in the same proportions as the above compositions, without charge compensating species: • (Th 0.95 U 0.05 )Ti 2 O 6 and (Th 0.90 U 0.10 )Ti 2 O 6 Recognising that ThTi 2 O 6 does not present redox flexibility when prepared under oxidising conditions, our intent was to control uranium speciation as the most significant variable in each solid solution, by judicious co-substitution of appropriate charge compensating species with known oxidation state.Charge compensating species were chosen based on their previously reported solid solubilities in the brannerite structure and/or similar titanate structures: Ca 2+ and Gd 3+ on the Th 4+ site [9][10][11] ; Al 3+ and Cr 3+ on the Ti 4+ site 19,23,24 .Note that compositions (Th 0.85 U 5+ 0.10 Ca 0.05 )Ti 2 O 6 and (Th 0.90 U 6+ 0.05 Ca 0.05 )Ti 2 O 6 are nominally identical to those previously investigated by Zhang et al., enabling a direct comparison, in principal, with this earlier study 16 .
Herein, we show that a minor fraction of U 6+ (2-13%) was indeed stabilised within the synthesised brannerite compositions, by High Energy Resolution Fluorescence Detected (HERFD) X-ray Absorption Near Edge Spectroscopy (XANES) at the U M 4 edge.However, both U L 3 edge XANES and HERFD U M 4 edge XANES evidenced U 5+ as the major oxidation state in all compounds.The evidence in support of uranate and uranyl speciation, and the factors limiting the concentration of U 6+ , are discussed.

Results
X-ray diffraction.X-ray diffraction was used to characterise the phases present in each product, as shown in Fig. 1, and summarised in Table 1 and Table 2.A compound with the brannerite structure (ThTi 2 O 6 ; PDF #04-007-2825) was the major phase formed in all compositions.Trace quantities of TiO 2 (rutile) and ThO 2 were also present in many compositions; trace ThO 2 was differentiated from UO 2 by electron microscopy observation of the microstructure and EDX analysis.No reflections characteristic of UO 2 , U 3 O 8 or UO 3 were observed; the estimated limit of detection of these oxides is around 0.5 wt% as determined by simulation of X-ray diffraction patterns.The diffraction patterns of compositions targeting Al 3+ or Cr 3+ substitution on the Ti 4+ site, exhibited reflections characteristic of trace ThO 2 .In contrast, the diffraction patterns of compositions targeting Ca 2+ or Gd 3+ substitution on the Th 4+ site did not generally exhibit reflections characteristic of trace ThO 2 .

Unit cell parameters.
The unit cell parameters of synthesised brannerite phases were determined by LeBail analysis of XRD data (Table 1); the weighted average cation radii (r w ) were estimated using appropriate Shannon radii (assuming as-batched compositions with U present as U 5+ only) 25 .When compared to previously reported unit cell parameters for ThTi 2 O 6

26
, the synthesised brannerites produced in this work had smaller overall unit cell volumes, in accordance with their reduced weighted average cation radii, as summarised in Table 2 and shown in Fig. 2 (e.g.228.96(1)Å 3 for nominal composition (Th 0.80 U 0.10 Gd 0.10 )Ti 2 O 6 with r w = 0.8506 Å; compared to 231.21 Å 3 reported for ThTi 2 O 6 with r w = 0.8567 Å).The unit cell parameters b and c, and the angle β, generally decreased with the weighted average cation radii, as shown in Fig. 2; the change in the a-parameter was comparably small and hence a clear trend was not apparent.These observations are consistent with previous systematic studies of the response of the brannerite crystal structure to substitution on the U/Th and/or Ti sites 13,21 .
U L 3 edge XANES.U L 3 edge XANES data were acquired to assess the bulk U oxidation state.The position of the U L 3 edge is dependent on the average U oxidation state, with some contribution from the local coordination environment of the U absorber.Initial examination of the acquired spectra, and U L 3 edge positon (see Fig. 3), suggested that U 5+ was the dominant oxidation state in all materials (edge position determined by the energy position of the maximum in the first derivative).The presence of a high concentration of Th in all compositions examined, with the Th L 3 edge at 16,300 eV, combined with the low U concentration (U L 3 edge at 17,166 eV), resulted in very small edge steps and incomplete normalisation of monochromator glitches in the pre-edge regions.
Linear regression of the U L 3 edge positions of the reference compounds (U 4+ Ti 2 O 6 , U 5+ 0.5 Yb 0.5 Ti 2 O 6 and CaU 6+ O 4 ) was used to estimate the average bulk U oxidation state of the brannerite materials reported here.All estimated average U oxidation states were in the range 4.8(2) + to 5.1(2) + , as shown in Table 2.The highest U oxidation states were determined for (Th 0.85 U 0.05 Gd 0.10 )Ti 2 O 6 , with sufficient Gd 3+ to charge balance 0.05 f.u.U 6+ (6Gd), and (Th 0.95 U 0.05 )(Ti 1.90 Cr 0.10 )O 6 , with sufficient Cr 3+ to charge balance 0.05 f.u.U 6+ (6Cr), both of which had estimated U oxidation states of 5.1(2) + .
As the individual contributions of U 4+ , U 5+ and U 6+ cannot be deconvoluted by conventional U L 3 edge XANES, due to the core-hole lifetime broadening of the spectra, HERFD U M 4 edge XANES spectra were also acquired to ascertain, conclusively, the contributions of the different U oxidation states to the overall average speciation.
HERFD U M 4 edge XANES.HERFD U M 4 edge XANES allows for a more definitive determination of average U oxidation state due to greater relative separation between the edge positions of U 4+ , U 5+ and U 6+ and Figure 1.X-ray diffraction patterns of ThTi 2 O 6 compositions adopting the brannerite structure, targeting U 5+ and U 6+ incorporation with appropriate charge compensation; nominal compositions are summarized in Table 1 and Table 2. Tick marks below show the reflection positions of ThTi 2 O 6 (PDF #04-007-2825).Diagnostic reflections of ThO 2 are marked with black circles; TiO 2 (rutile) by red circles.Composition identifiers generally use the nomenclature nEl, where n is the target U oxidation state and El is the charge compensating element.
Table 1.Unit cell parameters determined from LeBail analysis of X-ray diffraction patterns for ThTi 2 O 6 compositions, adopting the brannerite structure, targeting U 5+ and U 6+ incorporation with appropriate charge compensation, together with selected literature data.A weighted average cation radius (r w ) was calculated assuming nominal or reported compositions as appropriate, assuming U 5+ .R wp and χ 2 goodness-of-fit metrics from the Le Bail analysis are also included.Sample identification reference (ID) refers to Fig. 1 and generally uses the nomenclature nEl, where n is the target U oxidation state and El is the charge compensating element (or otherwise refers to a literature reference).*Calculated from literature reports of the unit cell parameters.

Nominal Composition ID a (Å) b (Å) c (Å) β (°)
Volume (Å 3 ) R wp GOF r w (Å) , evidenced a small contribution at ca. 3727.5 eV, Fig. 4, attributed to the presence of U 6+28 .The HERFD U M 4 edge XANES of materials with a uranyl speciation exhibit additional post-edge features, observed at ca. 3730 eV for CaU 6+ O 4 , see Fig. 4. 28,29 These features were not observed in any of the HERFD U M 4 spectra of the materials studied here, demonstrating that, although detectable fractions of U 6+ were present (see Table 2), uranate, rather than uranyl, is the dominant U 6+ speciation.Our confidence in this statement is tempered by the signal to noise ratio of the data and, certainly, a minor fraction of uranyl speciation cannot be conclusively ruled out.Linear combination fitting (LCF) was utilised to estimate the proportions of U 4+ , U 5+ , and U 6+ and the average U oxidation state.Consistent with analysis of U L 3 edge spectra, LCF of the HERFD U M 4 edge spectra evidenced average U 5+ speciation, within the range 4.91(12) + to 5.12(12) + (detailed in Table 2).The compositions (Th 0.90 U 0.10 )Ti 2 O 6 and (Th 0.95 U 0.05 )Ti 2 O 6 were the least oxidised, with average U oxidation states of 4.93(12) + and 4.91(12) +, and U 4+ fractions of 14.9(20)% and 13.3(32)%, respectively.The most oxidised materials were: (Th 0.85 U 0.05 Gd 0.10 )Ti 2 O 6 and (Th 0.95 U 0.05 )(Ti 1.90 Al 0.10 )O 6 , with average U oxidation states of 5.12(11) + and 5.10(13) +, respectively.Fitting of the spectra of these compositions also necessitated a small but significant contribution from the U 6+ reference compound (CaU 6+ O 4 ), of 14.6(11)% and 10.2(14)% respectively.These observations are in keeping with the oxidation states estimated from the U L 3 edge spectra, as well as previous reports of a fraction of retained U 4+ in air-synthesised charge compensated brannerites 10,21 .
Iterative Target Transformation Factor Analysis (ITFA) was also performed to evaluate the individual contributions of U 4+ , U 5+ and U 6+ to the final spectra and average U oxidation state.Initial principal component analysis of the brannerite and reference compounds determined that only three spectral-like components were necessary to reproduce all experimental spectra (see Supplementary Information, Fig. S1).This suggested that the spectra of the materials studied here can be well described by the three reference compounds.
Following the principal component analysis, the ITFA procedure was continued, with each of the three theoretical components accurately describing one of the reference compounds (see Supplementary Information, Fig. S1).As such, each component was assigned a given U oxidation state, with the relative fractions of each component utilised to calculate an average U oxidation state for the materials under examination.The relative fractions of the synthetic spectra and the calculated U oxidation states were both in excellent agreement with those calculated from linear combination fitting of the spectra (see Table 2; (see Supplementary Information, Figs.S2-6).
This analysis was confirmed by repeating the ITFA procedure with a larger suite of U 4+ , U 5+ and U 6+ reference compounds, as well as only U 4+ Ti 2 O 6 and CaU 6+ O 4 reference compounds, to ensure the synthetic U 5+ spectrum was representative of the U 5+ contribution in these materials.In both cases the relative fractions of the synthetic spectra and overall U oxidation states were in very close agreement with both the linear combination fitting and the three reference compound ITFA.
The average U oxidation states determined by all methods (linear regression of the U L 3 edge position, and LCF and ITFA of the U M 4 edge) are in excellent agreement (see Table 2).The material shown to have the highest U oxidation state, (Th 0.85 U 0.05 Gd 0.10 )Ti 2 O 6 , contained 13.3(1)% U 6+ as determined by ITFA of the HERFD U M 4 edge spectrum (detailed in Supplementary Information Table S1).Assuming target stoichiometry, this Table 2. Tabulated information from characterisation of ThTi 2 O 6 compositions, adopting the brannerite structure, targeting U 5+ and U 6+ incorporation with appropriate charge compensation (see Table 1).The secondary phases identified in the X-ray diffraction patterns and U oxidation states, as-determined by linear regression (LR) of the U L 3 edge position, and linear combination fitting (LCF) and iterative target transformation factor analysis (ITFA) of HERFD U M 4 edge XANES spectra, are also included.www.nature.com/scientificreports/corresponds to only 0.0133 f.u. of U 6+ .No features relating to the presence of uranyl speciation were observed in the spectra of any of the materials produced here.

Raman spectroscopy.
As expected from the near single phase nature of the materials produced here, determined by XRD, the Raman spectra of the brannerite compounds, Fig. 5, are in excellent agreement with previously reported spectra of actinide brannerites 15,16 .An earlier investigation of (Th 0.85 U 0.10 Ca 0.05 )Ti 2 O 6 and www.nature.com/scientificreports/(Th 0.90 U 0.05 Ca 0.05 )Ti 2 O 6 (i.e. the same nominal compositions as reported here) determined the presence of uranyl species, from analysis of Raman spectra.This assessment was made by assignment of the ν 1 symmetric stretch, apparent as a weak and broad band in the range 780-820 cm −115 , deconvoluted from the comparatively strong and intense band at 765-770 cm −1 attributed to the A g symmetric stretch of the TiO 6 octahedra.The reported ν 1 stretch modes are within the wavenumber range previously determined for a diverse suite of uranyl compounds 30,31 .The ThTi 2 O 6 brannerite compounds investigated here all exhibited evidence of a weak and broad band, centred at 780 cm −1 , apparent as a shoulder on the comparatively intense and sharp band, centred at 765 cm −1 , which can confidently be assigned as the A g symmetric stretch of TiO 6 octahedra.However, we also observed the Raman spectrum of ThTi 2 O 6 to present such a shoulder on the A g symmetric stretch of TiO 6 octahedra at 765 cm −1 , in agreement with earlier investigation 15 .It is evident that this band, also observed in the spectrum of ThTi 2 O 6 , must have at least some contribution from other Raman active modes of the brannerite structure (potentially arising from distortion of the TiO 6 polyhedra).We believe caution should be exercised in attributing uranyl speciation in brannerite compounds by assignment of the weak and broad band deconvoluted at 780-820 cm −1 as the ν 1 symmetric stretch.This is further evidenced by comparison of the Raman spectra of (Th 0.85 U 0.05 Gd 0.10 )Ti 2 O 6 and (Th 0.80 U 0.10 Gd 0.10 )Ti 2 O 6 .These compositions were determined to incorporate 13.3(1)% and 6.8(1)% of U 6+ respectively (averaged from Table 1), but the weak and broad band centred at 780 cm −1 is not strongly modulated, which is not consistent with a dominant uranyl speciation.We consider Raman spectroscopy to be inconclusive with regard to determination of uranyl speciation in the brannerite compounds reported here.If present, the concentration of uranyl species in the compounds examined in this work are very low, with even the most oxidised material containing only 0.0133 f.u. of U 6+ as noted above.It should also be noted that as a result of the short excitation wavelengths utilised in standard Raman spectroscopy, it is a potentially surface sensitive technique with observations made from spectra not necessarily being representative of the bulk material, particularly in materials highly absorbing in the excitation laser regimes (532 nm) relevant to this work.

Discussion
The aim of this study was to further investigate the stability of U 6+ in the brannerite structure, in four ThTi 2 O 6 solid solutions, targeting U 5+ or U 6+ speciation, stabilised by co-substitution and charge compensation on the Th or Ti site.The materials studied here were produced to examine the possibility of stabilising a significant fraction of U 6+ within the brannerite structure.HERFD U M 4 edge spectra provided direct and unambiguous evidence for a small fraction of U 6+ in all materials, however, U 5+ was the major U oxidation state present (exceeding  www.nature.com/scientificreports/77%, from ITFA).U 5+ was dominant whether sufficient charge balancing species were present to permit oxidation of all U present to U 6+ or not, or indeed when no charge balancing species were present (e.g.(Th 0.90 U 0.10 ) Ti 2 O 6 , 15.7(1)% U 4+ , 77.3(6)% U 5+ , 6.9(1)% U 6+ , from ITFA).However, it is interesting to note that the average U oxidation states of materials targeting U 6+ were generally characterised by a greater fraction of U 6+ speciation, compared to those targeting U 5+ speciation; for example, (Th 0.85 U 0.05 Gd 0.10 )Ti 2 O 6 and (Th 0.80 U 0.10 Gd 0.10 )Ti 2 O 6 , which were determined to incorporate 13.3(1)% and 6.8(1)% U 6+ respectively from ITFA.Compositions designed to target U 6+ speciation were determined to have a lower average oxidation sate, closer to U 5+ , which is likely to be realised by a low concentration of cation and/or oxygen vacancies, consistent with the detection of trace ThO 2 and TiO 2 impurities, and the known defect chemistry of the brannerite structure 27,32,33 .An earlier XPS study of (Th 0.55 U 0.30 Ca 0.15 )Ti 2 O 6 , with sufficient Ca 2+ to charge balance U 5+ , evidenced a higher contribution of U 4+ to the U 4f 7/2 peak when collected on a polished surface, compared to an unpolished fracture surface 10 .This was attributed to apparent Ca 2+ and U 5+ and/or U 6+ enrichment at the grain boundaries.Such enrichment of charge compensating elements in the grain boundaries of the materials studied here could also result in the determined bulk average oxidation state being lower than targeted, in addition to any cation and/or anion vacancies.U L 3 edge and HERFD U M 4 edge XANES are effectively bulk techniques, giving insight into the average U oxidation states and environments throughout a material.HERFD U M 4 edge spectra provided direct and unambiguous evidence for a small fraction of U 6+ in all materials, however, no features characteristic of uranyl speciation were observed, even in the most oxidised material, (Th 0.85 U 0.05 Gd 0.10 )Ti 2 O 6 , with 13.3(1)% U 6+ from ITFA.Although Raman spectra evidenced a weak and broad band centred at 780 cm −1 , potentially characteristic of a ν 1 symmetric stretch of uranyl species 15,16 , the evident presence of this band in the spectrum of ThTi 2 O 6 means that caution must be exercised in its diagnostic attribution to the presence of uranyl species in the brannerite structure.In comparison with X-ray spectroscopies, Raman spectroscopy using a 532 nm laser has a relatively low penetration depth in polycrystalline opaque materials such as those examined in this work, and may be more sensitive to uranyl speciation formed by surface oxidation.The data presented here directly and conclusively demonstrate the stabilisation of a small fraction of U 6+ in the brannerite structure, but the balance of evidence tends to support uranate, rather than uranyl, as the dominant U 6+ bulk speciation.Raman spectroscopy has proven more conclusive in substantiating the presence of uranyl speciation in metamict mineral brannerite, however, this may reflect past aqueous alteration of the specimen consistent with the observation of U-OH bending vibrations 34 .
A wide range of uranium (VI) oxometallates may be synthesised at high temperature in either air or oxygen atmosphere, so the partial pressure of oxygen is not thought to be the limiting factor in stabilisation of only a modest U 6+ concentration in the ThTi 2 O 6 compounds reported here 35 .The unit cell volume and weighted average cation radius of these compounds were shown to be well within the actinide brannerite stability field, as shown by Fig. 2, which suggests that ionic size effects are also unlikely to be limiting of the U 6+ concentration.The brannerite structure comprises corrugated sheets of BO 6 octahedra, in which each BO 6 octahedron shares three edges, with the sheets connected by chains of edge sharing AO 6 octahedra, in which each AO 6 octahedron shares two edges.In ThTi 2 O 6 , across the shared octahedral edges, the A…A cation distance is 3.823 Å, and the  1).The spectra of U 4+ Ti 2 O 6 , CrU 5+ O 4 and CaU 6+ O 4 reference compounds are included for comparison.The average U oxidation states determined from LCF and ITFA are detailed in Table 2.

Methods
Materials were prepared by solid state reaction and sintering.Stoichiometric amounts (see Table 2 1).The panel shows of the region from 680 to 860 cm −1 , showing a weak and broad band centred at 780 cm −1 as an apparent as a shoulder on the comparatively intense and sharp band centred at 765 cm −1 ; the latter can confidently be assigned as the A g symmetric stretch of TiO 6 octahedra (spectra are normalized normalised with respect to the intensity of the A g mode for comparison).

Figure 2 .
Figure 2. Plots of unit cell parameters, angle β, and volume, as a function of weighted average cation radius of compounds with the brannerite structure; data for ThTi 2 O 6 and UTi 2 O 6 are shown as black crosses; data from this study for ThTi 2 O 6 compositions targeting U 5+ or U 6+ incorporation, with appropriate charge compensation, are shown as red (Th site) and blue (Ti site) symbols; data for U 5+ brannerites (derived from UTi 2 O 6 ) are shown as green symbols.SeeTable 1 for details.

Figure 3 .
Figure 3. U L 3 edge XANES spectra for ThTi 2 O 6 compositions, adopting the brannerite structure, targeting U 5+ and U 6+ incorporation with appropriate charge compensation (seeTable 1).The spectra of U 4+ Ti 2 O 6 , U 5+ 0.5 Yb 0.5 Ti 2 O 6 and CaU 6+ O 4 reference compounds are included for comparison.The corresponding average U oxidation states as-calculated by a linear regression are detailed in Table 2.

Figure 4 .
Figure 4. HERFD U M 4 edge XANES of ThTi 2 O 6 compositions, adopting the brannerite structure, targeting U 5+ and U 6+ incorporation with appropriate charge compensation (seeTable 1).The spectra of U 4+ Ti 2 O 6 , CrU 5+ O 4 and CaU 6+ O 4 reference compounds are included for comparison.The average U oxidation states determined from LCF and ITFA are detailed in Table 2.
) of UO 2 , ThO 2 (produced by decomposition of Th(NO 3 ) 4 •5H 2 O at 550 °C), TiO 2 (anatase), CaTiO 3 , Gd 2 O 3 , Al 2 O 3 , and Cr 2 O 3 were homogenised by high energy ball milling (Fritsch Pulverisette 23 reciprocating ball mill, 30 Hz, 5 min)utilising yttria-stabilised zirconia mill pots and milling media, with isopropanol as a carrier fluid.The milled slurries were dried at 85 °C, and the resulting powder cakes broken up by hand in a mortar and pestle.The milled powders were then pressed into 10 mm pellets under 2 t (approx.250 MPa).Pellets were heat treated in alumina crucibles at 1400 °C for 24 h in air.

Figure 5 .
Figure5.Normalised Raman spectra of ThTi 2 O 6 compositions, adopting the brannerite structure, targeting U 5+ and U 6+ incorporation with appropriate charge compensation (see Table1).The panel shows of the region from 680 to 860 cm −1 , showing a weak and broad band centred at 780 cm −1 as an apparent as a shoulder on the comparatively intense and sharp band centred at 765 cm −1 ; the latter can confidently be assigned as the A g symmetric stretch of TiO 6 octahedra (spectra are normalized normalised with respect to the intensity of the A g mode for comparison).
As with the U L 3 edge spectra discussed above, the low U and high Th concentrations, in the materials examined, resulted in low intensities of the HERFD U M 4 spectra.The spectra of U 4+ Ti 2 O 6 , CrU 5+ O 4 and CaU 6+ O 4 reference compounds were also acquired for comparison and to support further quantitative analyses.Consistent with analysis of U L 3 edge spectra, initial examinations of the HERFD U M 4 edge XANES supported an average oxidation state close to U 5+ , evidenced by the principle feature at ca. 3726 eV, as shown in Fig.4.Inspection of the HERFD U M 4 edge XANES of (Th 0.90 U 0.10 )Ti 2 O 6 and (Th 0.95 U 0.05 ) Ti 2 O 6 , also identified a small but distinct contribution at ca. 3725 eV, Fig.4, demonstrating the presence of an additional minor component of U 4+ .Inspection of the HERFD U M 4 edge XANES of some compositions targeting U 6+ , most obviously (Th 0.90 U 0.05 Ca 0.05 )Ti 2 O 6 and (Th 0.95 U 0.05 )(Ti 1.90 Al 0.10 )O 6

Table 1 )
. The spectra of U 4+ Ti 2 O 6 , U 5+ 0.5 Yb 0.5 Ti 2 O 6 and CaU 6+ O 4 reference compounds are included for comparison.The corresponding average U oxidation states as-calculated by a linear regression are detailed in Table2.