Cr²⁺ solid solution in UO₂ evidenced by advanced spectroscopy

Abstract

Advanced Cr-doped UO₂ fuels are essential for driving safe and efficient generation of nuclear energy. Although widely deployed, little is known about their fundamental chemistry, which is a critical gap for development of new fuel materials and radioactive waste management strategies. Utilising an original approach, we directly evidence the chemistry of Cr⁽³⁺⁾₂O₃–doped U⁽⁴⁺⁾O₂. Advanced high-flux, high-spectral purity X-ray absorption spectroscopy (XAS), corroborated by diffraction, Raman spectroscopy and high energy resolved fluorescence detection-XAS, is used to establish that Cr²⁺ directly substitutes for U⁴⁺, accompanied by U⁵⁺ and oxygen vacancy charge compensation. Extension of the analysis to heat-treated simulant nuclear fuel reveals a mixed Cr^2+/3+ oxidation state, with Cr in more than one physical form, explaining the substantial discrepancies that exist in the literature. Successful demonstration of this analytical advance, and the scientific underpinning it provides, opens opportunities for an expansion in the range of dopants utilised in advanced UO₂ fuels.

Introduction

Chromium (Cr) is an important additive used in uranium dioxide (UO₂) fuel fabrication. The doping of Cr, added as Cr⁽³⁺⁾₂O₃, into UO₂ fuels leads to enhanced grain growth, realising beneficial properties of higher fission gas retention and reduced swelling during reactor operations. During the subsequent long-term storage and disposal of such fuels post-fission, the dissolution behaviour in aqueous media, a two-stage mechanism dependent upon the oxidation of U⁽⁴⁺⁾O₂ to more soluble U⁽⁶⁺⁾O₂²⁺, will be integral to safety over millions of years^1,2. Since doping of the UO₂ lattice with aliovalent species will alter the valence state of U and, consequently, the local defect structure of UO₂, e.g. forming U^(4+/5+)O_2 + x, an understanding of the influence of Cr on the UO₂ lattice is required to support safety assessments for disposal. While the role of Cr on practical aspects such as fuel performance^3,4,5,6,7 and grain growth mechanisms^{8,9,10,11,12,13} is generally understood, agreement on the mechanism of Cr incorporation into the UO₂ structure and its consequent impacts on U chemistry has proven elusive^{14,15,16,17,18,19,20,21}. One reason for this is the prior focus on materials that have been sintered to mimic nuclear fuel production, on which a number of factors, such as pressure, temperature and atmosphere will influence the O diffusion kinetics in the material and, therefore, the final valence, distribution and coordination of U and Cr in the UO₂ structure. Since these sintered systems are highly complex, the lattice structures between studies tend to be incomparable due to the varying sintering conditions used.

A number of possible incorporation mechanisms and associated speciation of Cr within UO₂ have been proposed, implicating Cr³⁺ or Cr²⁺, and in one case, Cr¹⁺. A substitution mechanism is proposed, evidenced by a reduction in the lattice parameter of UO₂ upon doping, attributed to substitution of smaller Cr^2+/3+ (ionic radii of 0.73 Å and 0.62 Å, respectively²²) on the larger U⁴⁺ (ionic radius 1.00 Ų²) site. This mechanism requires charge compensation via positive defect formation; for example, U⁵⁺ species and/or oxygen vacancy (O_v) defects are expected to develop, causing further lattice distortion. Comparison of Cr K-edge X-ray Absorption Spectra (XAS), in particular, the X-ray Absorption Near Edge (XANES) spectral region of sintered Cr-doped UO₂ pellets with a range of Cr standards, has previously provided tentative evidence for the Cr³⁺ valence state^20,23. More recent ab initio atomistic simulations, based on qualitative assessment of XANES data, argue the plausibility of Cr²⁺ speciation²¹, with substitution onto the U⁴⁺ site and concurrent formation of one O_v, determined to be the most thermodynamically favourable configuration. Alternately, in density functional theory and empirical potential descriptions, where the valence state of Cr was not restrained to the trivalent state, modelling of defect concentrations in Cr-doped UO₂ at high temperature has predicted Cr¹⁺ interstitial defects would be the dominant species⁸.

To gain definitive experimental evaluation of the Cr incorporation mechanism, quantitative information relating to the lattice interatomic distances and near-neighbour coordination, through the use of Extended X-ray Absorption Fine Structure (EXAFS) region of XAS data is required. The only reported EXAFS fitting of Cr-doped UO_2, carried out on commercially available and chemically complex sintered nuclear fuel pellets, gave evidence that Cr is incorporated as Cr³⁺ in 6-fold coordination²³. The feasibility of Cr²⁺ valence and the presence of undissolved Cr⁽²⁺⁾O phases in the pellet was discussed, but not verified. This preliminary work reported a significant change in Cr-O bond length (2.02 ± 0.02 Å) when compared to the U-O distance (2.36 ± 0.02 Å) in UO₂, which was attributed to disruption in both the cation-cation and cation-anion lattices due to the presence of Cr.

In the current study, a high-flux, high spectral-purity beamline was utilised to fully resolve the XAS Cr K-edge, including both XANES and EXAFS regions, thus making it possible to fully resolve the local structure and valence state of Cr in UO₂. Additionally, high-energy resolution fluorescence detection (HERFD) XANES data were acquired at the U M₄-edge to evidence the oxidation state of U^24,25, complemented by Raman spectroscopy analysis of the Cr-doped UO₂ oxygen defect chemistry. Importantly, our study avoids the ambiguity induced in the interpretation of sintered UO₂ by focusing on simple heat treated (i.e., calcined, not sintered) materials. We thus develop an experimentally-evidenced, fundamental understanding of the mechanism of Cr₂O₃ incorporation within UO₂ oxide powder. This understanding is further applied to Cr-doped UO₂ sintered in a reducing atmosphere, where the unavoidable complexity introduced into the system, and the associated challenge in assigning incorporation mechanisms, is demonstrated.

Results and discussion

Cr-speciation in UO₂ calcined powder

Confirmation of successful Cr-doping of UO₂ powder upon calcination was given by ICP-MS analysis. All oxide powders were shown to contain the intended concentration of Cr when fully digested (in the range 300–2400 ppm) (Fig. 1a).

Fig. 1: Cr content and incorporation in UO₂.

a The measured Cr content of Cr-doped UO₂ calcined powder and sintered pellets. Total Cr content measured by ICP-MS following acid digest and matrix content of sintered samples measured by EPMA, excluding grain boundary precipitates; it highlights the proportion of Cr that is incorporated in the UO₂ matrix in solid solution, incorporated as precipitates and Cr that has been lost from the system due to volatilisation; b schematic representation and associated crystal structure of particles of calcined Cr-doped UO₂²¹ and; c schematic representation and crystal structures of the sintered matrix of Cr-doped UO₂^21,26, highlighting distribution of Cr in precipitates (green) at grain boundaries (white). In the crystal structures, grey atoms represent U, green atoms represent Cr and red represent O.

The approach to Cr K-edge EXAFS fitting considered a range of different structural models (a selection of which are shown in Supplementary Fig. 1 and Supplementary Table 1), including simple fits that only contained oxygen backscatterers, as well as fits based upon models from the DFT work of Sun et al.²¹. The best fit for the calcined material was obtained using a structure derived from ab initio atomistic models of Cr-doped UO₂, in which Cr²⁺ substitutes onto the U⁴⁺ cation site via the formation of O_v for charge balance, denoted by (1Cr²⁺ + 1O_v)²¹ (Fig. 1b). The EXAFS model (Fig. 2a, b, Table 1) indicates that the first O nearest neighbour environment is split, with 4 O atoms at a distance of 2.00 ± 0.01 Å, and 3 O atoms at 2.95 ± 0.02 Å, giving a total coordination of 7. Since this is less than the expected coordination of 8, it implies that one O_v is situated in the Cr coordination environment. Confirmation of Cr²⁺ speciation is given by the bond valence sum (BVS)²⁷ of 2.05. The second shell Cr–U environment was successfully fitted by the model, with 4 U atoms at a scattering distance of 3.58 ± 0.04 Å and 2 U atoms at 3.85 ± 0.06 Å. These latter Cr–U distances are in good agreement with U–U distances of UO₂ (3.83 Ų³) while the presence of shorter Cr–U distances can be attributed to distortion of U atoms adjacent to the O_v. Detailed fit results are displayed in Table 1. This data supports the hypothesis of Sun et al.²¹ who found that the most thermodynamically favourable structural arrangement of Cr in UO₂, through ab initio calculations, is a pair of associated Cr²⁺ and O_v, i.e. 1Cr²⁺ + 1O_v. Given the excellent statistics of the fit (R-factor and corroborating BVS), it is unlikely that any other Cr species exist in the calcined sample, with its bulk speciation being accurately described by the 1Cr²⁺ + 1O_v model.

Fig. 2: Crystal structure and local coordination analysis of Cr in Cr-doped UO₂ determined by EXAFS and XANES analysis.

k³-weighted Cr K-edge spectra, EXAFS model fits and Cr K-edge XANES spectra of Cr-doped UO₂ in comparison to known standards (Cr₂O₃, CrUO₄ and Cr⁰ foil), respectively, for (a–c) calcined powders; and (d–f) sintered pellets. For (a, b, d, e) orange lines are data and black dashed lines are the fits.

Table 1 Fitting parameters and results for 2400 ppm Cr-doped UO₂ calcined oxide powder and sintered pellet samples.

The substitution mechanism via vacancy formation is further supported by an increase in the O_v (U1) defects observed with increasing Cr-dopant concentration, according to Raman measurements (Fig. 3a). HERFD XANES data acquired at the U M₄-edge (Fig. 3b, Supplementary Fig. 2) showed, through principal component analysis of the data series, that only two components were required to accurately reconstruct each of the samples in the series. These were attributed to U⁵⁺ and U⁴⁺ oxidation states using the standards CrU⁵⁺O₄ and U⁴⁺O₂. Whilst the higher energy component matches well with the U⁵⁺ standard, U⁴⁺ shows some differences when compared to the UO₂ standard. This is likely due to the distortion around the U⁴⁺ coordination environment upon doping of UO₂ with Cr, possibly due to vacancy formation. Iterative target transformation factor analysis was used to indicate the approximate fraction of U⁴⁺ and U⁵⁺ in the Cr-doped oxide powders (Fig. 3c), with results indicating ~ 30% U⁵⁺ (and therefore ~70% U⁴⁺) for all samples. This corroborates the hypothesis that charge balance for Cr²⁺ within Cr-doped UO₂ is provided by both O_v and U⁵⁺ defect formation, with O_v potentially playing a more prominent role. The final stoichiometry is, therefore, $({({U}_{1-x}^{4+}{U}_{x}^{5+})}_{1-y}C{r}_{y}^{2+}){O}_{2-\frac{y}{2}}$.

Fig. 3: Raman Spectroscopy, U M₄-edge HERFD XANES and p-XRD analysis of Cr-doped UO₂ materials.

a Oxygen defect content realised by deconvolution of Raman spectra of calcined material (errors derived from the standard deviation of 10 measurements); b U M₄-edge HERFD XANES of calcined material. c Quantification of the U^4+/5+ ratio in calcined powder determined from fitting of U M₄-edge HERFD XANES data in b. d Oxygen defect content of sintered material realised by deconvolution of Raman spectra (errors derived from the standard deviation of 10 measurements). e U M₄-edge HERFD XANES of sintered material; and (f) lattice parameter values for calcined powder and sintered material, as determined by p-XRD (errors derived from the lattice parameter refinement).

Analysis of the UO₂ T_2g Raman band at 445 cm⁻¹, and other peaks relevant to defect structures in the matrix (Fig. 3a and Supplementary Fig. 3a, b), revealed a slight increase in the oxygen interstitial (O_i) (U3) concentration with increasing Cr-dopant concentration. Given that the concentrations of Cr in the calcined material exceed the generally agreed solubility limit of Cr in UO₂ (reported to be in the range of 700–1200 ppm^3,6,9,11,15, it could be assumed that excess Cr may be present as positive interstitial defects⁸. If this were the case, negative O_i defects would be required as a charge balance mechanism. Since the O_i defect concentration (Fig. 3a) broadly increases with increasing Cr content, this seems to be a plausible explanation. However, it is also possible that such defects arise from slight oxidation of the sample during analysis. Given that the EXAFS model did not require the presence of Cr interstitials to improve the fit, the latter explanation may be more reasonable.

Cr-speciation in sintered UO₂ pellets

The complexity induced during sintering (analogous to nuclear fuel synthesis) was investigated using calcined Cr-doped UO₂ powders pressed into pellets and heat-treated (sintered) at 1700 °C under a reducing 5%H₂:95%N₂ atmosphere. Loss of Cr, expected due to high temperature volatilisation^28,29, is observed through a reduction in the Cr concentration when compared with calcined powder (Fig. 1a). In contrast to the homogenous Cr incorporation in calcined powders, significant heterogeneity in the distribution of Cr in the sintered pellets was observed (Supplementary Fig. 4). This is exemplified by the precipitation of Cr-containing particles within grain boundaries, which has previously been observed for Cr-doped UO₂ sintered under a range of conditions^{6,9,11,12,15,20}.

Retention of a portion of Cr within the UO₂ matrix upon sintering is evidenced by a reduction in the lattice parameter with increasing Cr-dopant matrix concentration, as measured by powder X-ray diffraction (p-XRD, Fig. 3f). The absolute values of lattice parameter between the sintered and calcined oxide material are different, with the latter being lower than the former by ~0.004 Å. This is most likely due to the loss of some Cr²⁺ in the lattice due to volatilisation (Fig. 1a). Alternatively, the lattice parameter decrease could result from the reduction of Cr²⁺ to Cr⁰ (ionic radius 2.00 Å) in the reducing sintering atmosphere or, conversely, possible oxidation of pellets after sintering. Both alternative explanations are excluded on the basis of experimental data obtained: the absence of U⁵⁺ and U⁶⁺ in the HERFD-XANES data for the sintered material (Fig. 3e) and a lower O_i concentration than in the calcined materials (Fig. 3d), precludes enhanced oxidation. Moreover, analysis of the XAS data effectively excludes the presence of Cr⁰ within the EXAFS model (see below).

In accordance with the highly reducing sintering conditions, the white line position (E₀; when taken as the peak of the first derivative; Supplementary Data Table 2) of the Cr K-edge (~6000 eV, Fig. 2f) is shifted to lower energy than that of the Cr³⁺ standards (6006.0 eV) and the calcined Cr²⁺UO₂ material (6003.2 eV), and towards that of the Cr⁰ standard (5989 eV). The presence of Cr²⁺, rather than Cr⁰, in the sintered material is confirmed by the occurrence of the pre-edge feature at 5995 eV, which originates from the forbidden 1 s → 4 s transition of Cr^2+ 30. This transition becomes allowed for Cr²⁺ only when multiple Cr²⁺ environments are present, which induces s to p orbital mixing. The lower the symmetry of the Cr²⁺ environments, the more that such hybridisation can occur; consequently, it is proposed that multiple, low-symmetry Cr²⁺ environments exist for the sintered material. That this transition is not observed as strongly in the equivalent calcined Cr²⁺-containing UO₂ XANES spectra (Fig. 2c), is likely to be a consequence of the single Cr²⁺ coordination and high symmetry environment of those materials.

In comparison with the calcined material, fitting of the EXAFS data acquired from the sintered materials was challenging. As with the calcined material, several models were first explored; however, all produced unsatisfactory fits, both quantitatively and qualitatively. The best fit was achieved by utilising the 1Cr⁽²⁺⁾+1O_v structure²¹ in addition to a second Cr²⁺ environment of 2Cr⁽²⁺⁾+ 2O_v and also a Cr³⁺ environment, of Cr⁽³⁺⁾₂O₃ (Fig. 2e and Table 1; structures shown in Fig. 1c). The mixed Cr²⁺/Cr³⁺ oxidation state is confirmed by the resultant BVS, which gives an average Cr oxidation state of 2.35. The EXAFS fit was used to estimate that the sintered Cr-UO₂ contains ~65% Cr²⁺ and ~35% Cr³⁺. The presence of Cr³⁺ in such a reducing environment can be explained by the existence of U⁵⁺ species in the calcined precursor material, but none in the corresponding sintered material (Fig. 3b, d). We hypothesise that Cr³⁺ originates from the reduction of U⁵⁺ to U⁴⁺ during sintering and the associated redox couple with Cr²⁺/Cr³⁺.

The aforementioned best fit model contained: 4.5 O atoms at a distance of 2.00 ± 0.01 Å; 1.5 O atoms at 2.44 ± 0.02 Å; 1 U atom at 3.34 ± 0.05 Å; 3 U atoms at 3.40 ± 0.07 Å; and 2 more U atoms at 3.91 ± 0.02 Å, indicating that a complex mixture of different Cr environments exists. Previous work has experienced some success in obtaining and analysing XAS data (both XANES and EXAFS^20,23) on systems similar to this sintered sample. While informative, without underpinning understanding from simplified systems like the calcined powder investigated in the present study, and the lack of corroborating analysis from other techniques, resolution of the coordination environment of Cr from these previous XAS studies has proven challenging. As such, this proposed model, with supporting data from other techniques and fitting models grounded in DFT simulations, provides insight to a possible structure for sintered Cr-doped UO₂. Given the complexity of the sample, including the likely multiple Cr oxidation states and coordination environments, further study of sintered materials developed under carefully controlled conditions should be performed, using this original analytical approach, to systematically evaluate the effect of sintering on the structure of Cr-UO₂ materials.

Mechanism of Cr incorporation in UO₂

Considering the data presented, the following mechanism of Cr incorporation in sintered UO₂ is proposed (Fig. 1c): Cr²⁺, already dissolved in the UO₂ lattice (substituted onto U⁴⁺ sites and associated with O_v) in the calcined starting material, is retained within the matrix upon sintering. For any given material where the proposed solubility limit of Cr is exceeded (> 700–1200 ppm Cr in UO₂^3,6,9,11,15), Cr is likely present in interstitial sites⁸ and is, thus, unbound and easily mobilised during sintering. This portion of Cr diffuses towards the grain boundaries allowing precipitation or volatilisation to occur (Fig. 1a). While it is not possible to confirm the oxidation state of this mobile fraction of Cr, it is likely that this reflects a combination of Cr²⁺, which was determined to exist in more than one environment, and some Cr³⁺, which was necessary to model the EXAFS data. The presence of Cr³⁺ is in agreement with µ-XANES analysis of an industrially-synthesised Cr-doped UO₂ pellet¹⁹, which confirmed that precipitates of Cr comprised of Cr³⁺.

This work provides direct experimental evidence that, prior to high temperature heat treatment, the fundamental lattice structure of Cr-doped UO₂ incorporates Cr²⁺ substituted on a U⁴⁺ site, charge balanced by the formation of both U⁵⁺ and O_v defects. The latter defects appear to remain upon sintering in a reducing atmosphere, while U⁵⁺ is reduced, promoting the oxidation of Cr²⁺ to Cr³⁺, although the main Cr oxidation state in these materials remains Cr²⁺. Were the materials to be sintered under an atmosphere with a higher partial pressure of oxygen, it is possible that Cr could be fully oxidised to Cr³⁺. Realisation of this mechanism not only provides detailed understanding of the defect structures in doped UO₂ materials but can also be utilised to predict their behaviour in reactor, long-term storage and disposal conditions. The difficulty in realising the full local structure of aliovalent doped-UO₂ materials, particular after the application of high temperature heat treatment, is highlighted. Such complexity can be, at least in part, resolved using high resolution spectroscopy techniques, which can be used to scientifically underpin the development of advanced doped uranium nuclear fuels. This work has shown that underpinning the dopant incorporation mechanism in calcined material is a fundamental step required to determining the evolution of crystal chemistry in Cr-doped UO₂ fuels.

Methods

Sample preparation

A suite of Cr-doped UO₂ samples were prepared via a wet synthesis method using uranium (VI) nitrate hexahydrate in solution (0.3 mol L⁻¹) and chromium (III) nitrate nonahydrate in solution (1.6 mol L⁻¹) (Cr(NO₃)₃∙9H₂O 99.99 %, Sigma Aldrich). Concentrated ammonium hydroxide solution (5 mol L⁻¹) was added at room temperature while stirring until a pH of 8–10 was reached, and a yellow precipitate was observed. Successful co-precipitation of U and Cr was confirmed by analysis of the supernatants by ICP-OES (ThermoFisher iCAP Duo6300), where 99.9 % precipitation for both U and Cr was achieved. Following vacuum filtration, precipitates were washed in deionised water and dried overnight at 90 °C to eliminate any remaining hydroxide. Thermal treatment at 750 °C for 4 h under a reducing (95% N₂ / 5% H₂) atmosphere allowed conversion of the precursor to oxide, confirmed by power x-ray diffraction analysis (employing LaB₆ standard), which showed the fluorite crystal structure in all of the pure and Cr-doped UO₂ samples (Supplementary Fig. 5). Homogeneity and powder reactivity were increased by milling at 35 Hz for 15 min. A stainless steel die was then used to uniaxially press 6 mm green pellets, at 500 MPa, before sintering in a reducing atmosphere (95% N₂/5% H₂) at 1700 °C for 8 h to obtain the final sintered pellets.

The concentration of Cr in the doped UO₂ was verified by complete digest in concentrated nitric acid. Approximately 20 mg of calcined material was dissolved in 2 M HNO₃ (ultrapure) and each vessel was magnetically stirred at 90 °C to accelerate the dissolution. An aliquot was analysed for the Cr concentration using inductively coupled plasma mass spectroscopy (ICP-MS, ThermoFisher iCAP RQ). Triplicate samples were measured to obtain an average Cr concentration for each sample. This method was repeated for sintered pellet samples, which were crushed to a powder using a pestle and mortar before being dissolved. The average Cr concentration is reported as Total Cr concentration in Fig. 1a where the difference between Calcined powder and Sintered pellet concentration is considered loss of Cr due to volatilisation during heat treatment.

Electron probe microanalysis

Sintered pellets were ground using SiC paper, polished to 1 μm using diamond suspension and measured at the EPMA facility, School of Geosciences, University of Edinburgh, UK. A Cameca SX100 micro analyser with a 15 keV accelerating voltage, 160 nA beam current and a beam diameter of 1 μm was used with standards of FeCr₂O₄ and UO₂. Elemental composition was analysed by wavelength dispersive X-ray analysis using crystals of PET and LPET to measure U Mα and Cr Kα signals. Measurements were taken at 15 points across the pellet surface and an average wt. % Cr within the UO₂ matrix calculated as well as maps of 100 × 100 μm. The average wt. % Cr is reported as the Matrix Cr content in Fig. 1a and is considered the concentration of Cr in solid solution. The limit of detection was 25 ppm Cr hence the first two samples in the series could not be measured and are displayed as 0 ppm Matrix Cr content in Fig. 1a. The difference between Total Cr in Sintered pellets (as measured by ICP-MS and described above) and the Matrix Cr is considered the proportion of Cr present in precipitates (Fig. 1a and Supplementary Fig. 4).

MATLAB software was used to process images and adjust the threshold, allowing any areas of high Cr concentration to be observed. Back scattered electron images were overlaid with EPMA Cr elemental maps to highlight location of Cr.

Raman spectroscopy

The crystal structure of calcined and sintered Cr-doped UO₂ was analysed by Raman spectroscopy using a Renishaw inVia Reflex confocal spectrometer equipped with a Leica DM2500 microscope. Pressed green pellets of calcined powders and sintered samples polished to 1 μm using diamond suspension were measured using a 514.5 nm green argon laser with a holographic grating of 1800 lines mm⁻¹ to acquire a spectral acquisition between 200 and 800 cm⁻¹. The acquisition time was 30 s per spectrum and the laser power was set to 5 mW to limit sample oxidation. These configurations allowed a 2–3 cm⁻¹ spectral resolution. An average of 10 point measurements across the surface of each sample, avoiding grain boundaries, was taken to confirm their reproducibility and homogeneity of the composition. Information from the Raman spectra was extracted using Lorentz function fitting where the intensity ratios of the triplet defect peaks (U1, U2 & U3) relative to the main UO₂ T_2g band (445 cm⁻¹) was carried out. These defect peaks correspond to O_v (U1) (~527 cm⁻¹), LO mode (U2) (~574 cm⁻¹) and O_i (U3) (~634 cm⁻¹)³¹.

X-ray absorption spectroscopy

XAS measurements were performed, at room temperature, on both calcined powder and sintered Cr-doped UO₂ at several beamline facilities. All samples diluted with poly-ethylene glycol (PEG) to 1 absorption length for the desired element and edge (Cr K- or U M₄-edges). Fluorescence mode was employed at beamline I20-scanning at DLS to measure the Cr K-edge (5989 keV) using a 64-element Ge detector with Xspress4 signal processing. The beam size was 400 × 300 µm (FWHM). The wiggler-sourced I20 beamline utilises a Si (111) four-bounce crystal monochromator, which results in high flux and excellent energy resolution. As such, I20-scanning is uniquely suited to determine the local structure of Cr-doped UO₂, especially where Cr was in low concentration in the heavily absorbing UO₂ matrix. Multiple scans were taken to improve data quality with energy steps of 5 eV (5889–5985 eV), 0.2 eV (5985–6010 eV) with a time step of 1 s step⁻¹ and 0.04 Å⁻¹ (6010-6700 eV) with a time step of 1–6 s step⁻¹ in this regions. High-energy resolution fluorescence detection X-ray absorption near edge spectroscopy (HERFD XANES), at the U M₄-edge (3728 eV), was carried out at the BM20 beamline at the European Synchrotron (ESRF)³², France, with a help of X-ray emission spectrometer³³. Surfaces of as-sintered pellets were measured in fluorescence mode, while calcined powders were mixed with poly-ethylene glycol (PEG) and pressed into 6 mm pellets for analysis, also in fluorescence mode. Standards of known U and/or Cr valence state and coordination, including UO₂, CrUO₄, Cr₂O₃ and Cr Foil were also measured, in transmission and fluorescence mode to account for self-absorption effects.

A standardised element foil was measured simultaneously with Cr-doped UO₂ samples and used as a reference to which the E₀ positions of all raw data was aligned. A cubic spline background subtraction and normalisation procedure³⁴ was carried out on all aligned data before multiple scans were merged using Athena³⁵. The EXAFS region, measured in eV, was then converted to wavenumber, k, with units of Å⁻¹. The Fourier Transform of this data was then fit using models generated from applying possible scattering paths of the central absorber (Cr) to surrounding atoms (O and U). Scattering paths were generated from appropriate structural CIF files, using the FEFF 6 algorithm³⁶ in Artemis. During the process of trialling a range of possible EXAFS models, a number of parameters; range of k (Å⁻¹) and radial distance r (Å), and the amplitude reduction factor (S₀²) were optimised for the data and the following parameters allowed to refine; degeneracy (N), fitted bond length (R)(Å), shift from Cr K-edge position (5989 keV) (∆E₀) and the Debye–Waller factor (σ²). Ultimately, the final fits fixed the degeneracy (N) to the values from the relevant CIF files, or a manual refinement was performed. The F-test was applied to each fitted path, the result (α) indicated the confidence of adding the path to improve the fit (>67% gives a confidence of 1σ, >95 % gives a confidence of 2σ)³⁷. The bond valence sum (BVS) returns a summation of bond distances in a coordination shell, which is equal to the formal oxidation state of the cation absorber³⁸. The bond valence of Cr-O coordination was calculated for both calcined and sintered material and used to determine the proportion of Cr²⁺/Cr³⁺ in the sintered material.

Despite efforts during collection of Cr K-edge data, reduced resolution of the XANES region at the lowest Cr concentrations resulted in inconsistency in the background removal and normalisation, most prominently in the 300 ppm sample for both oxide powder and sintered samples (Fig. 2c, e and Supplementary Data Fig. 6). Nevertheless, the same features were observed for all materials and, as such, only the samples with the highest dopant concentrations were utilised for Cr K-edge EXAFS analysis.

Principal component analysis was carried out on U M₄-edge data (raw data was pre-processed in Athena as described above) as part of the Iterative Target Transformation Factor Analysis³⁹ to calculate the relative concentration of each component of the absorption spectra; UO₂ and CrUO₄ standards were used.

X-ray diffraction

Powder XRD (p-XRD) characterisation was performed using a PANalytical Xpert3 diffractometer in reflection mode with a 45 keV/40 mA generator. To avoid oxidation, the calcined oxide samples were measured immediately post-heat treatment (i.e. within 10 min of removal from the furnace). Similarly, sintered samples were crushed using a pestle and mortar upon removal from the furnace, within a controlled inert atmosphere (N_2(g)), and immediately measured. Data were collected between 5° and 100° 2θ with a step size of 0.013° and a step time of 40 s, a fixed slit size of 0.5 was used. An internal standard of LaB₆ (20–30 wt. %) was used for data alignment, and p-XRD patterns corrected in the WinXPow software. Le Bail refinements carried out in the Topas software, allowing accurate determination of the lattice parameter as a function of Cr content in UO₂.