Synthesis and characterisation of Ca 1-x Ce x ZrTi 2-2x Cr 2x O 7 : Analogue zirconolite wasteform for the immobilisation of stockpiled UK plutonium

A series of Ca 1-x Ce ZrTi 2-2x Cr 2x O 7 zirconolite ceramics (0≤x≤0.35) were reactively sintered in air at 1350°C for 20h. Single phase zirconolite-2M was formed for x ≤ 0.15, with Cr 2 O 3 and an undesirable Ce-bearing perovskite phase present above x=0.20. Electron difraction analysis conirmed that the zirconolite-2M poly- type was maintained over the solid solution. X-ray absorption near edge structure (XANES) data determined that between 10–20% Ce was speciated as Ce 3+ , and Cr was present uniformly as Cr 3+ with near edge features consistent with occupation of octahedral sites within the zirconolite-2M structure. A sample corresponding to x=0.20 was processed by reactive spark plasma sintering (RSPS), with a rapid processing time of less than 1h. XANES data conirmed complete reduction to Ce 3+ during RSPS, promoting the formation of a Ce-bearing perovskite, comprising 19.3±0.4wt.


Introduction
The safe handling and conditioning of highly radioactive materials derived from nuclear fuel reprocessing operations is an issue of scientiic and socioeconomic importance. In the United Kingdom, a large stockpile of separated civil plutonium oxide (PuO 2 ) has amassed from spent fuel reprocessing via the PUREX (Plutonium-Uranium-Reduction-Extraction) process [1]. This stockpile is forecast to peak at 140 teHM (tonnes of heavy metal equivalent), following cancellation of the Fast Neutron Reactor programme in 1994, which has led to the need for other Pu management strategies, for example reuse as mixed oxide (U, Pu)O 2 (MOX) fuel in light water reactors [2]. Currently, MOX fuel presents a commercially unattractive prospect in the UK. The Nuclear Decommissioning Authority (NDA, UK) have assessed it to be a credible option, although appetite for oftake from commercial operators is weak [2]. A fraction of the Pu stockpile has also been declared unsuitable for MOX fabrication, due to chlorine contamination arising from long term storage in polyvinyl chloride (PVC) packaging [3]. An alternative approach is the conditioning of the Pu stockpile and waste material via immobilisation within a ceramic matrix.
Immobilisation within ceramic matrices is considered a viable option for the conditioning of actinides such as Pu, since crystalline materials generally ofer superior aqueous durability with respect to borosilicate glasses currently used for the vitriication of high level nuclear waste calcines [4][5][6]. Zirconolite is a naturally occurring titanate ceramic closely related to the pyrochlore family of minerals, with general structure A 2 B 2 O 7 [7]. Zirconolite has nominal stoichiometry CaZr x Ti 3-x O 7 , and crystallises in the zirconolite-2M polytype for 0.85 ≤ x ≤ 1.30, in space group C2/c [8]. This structure consists of ZrO 7 and CaO 8 polyhedra integrated between corner sharing TiO 6 /TiO 5 polyhedra arranged in a hexagonal tungsten bronze-style motif. This two-layer repeat constitutes the zirconolite-2M unit cell [9]. Zirconolite can exhibit extensive polytypical behaviours which are characterised by the stacking sequence between CaO 8 /ZrO 7 polyhedra and TiO 6 layers; the crystallisation of zirconolite polytypes occurs as a result of lanthanide/actinide incorporation and partial oxygen pressure. For example, the zirconolite-4M polytype was observed to form as an intermediate phase in Ca 1-x Zr 1-x Nd 2x Ti 2 O 7 system and is described as an intercalated sequence of zirconolite-2M and pyrochlore type modules [10,11]. The lexibility of zirconolite and its respective polytypes to incorporate a wide variety of lanthanide and actinides within its structure makes it an ideal candidate for the disposition of excess Pu. The aim of this work is to evaluate the efectiveness of Cr 3+ as a charge balancing species for the immobilisation of Ce; in this instance, Ce 4+ is applied as a surrogate for Pu 4+ . Cerium is justiied as a suitable surrogate on the basis of safety and expediency, however, the variations with respect to redox potential have been largely realised [12]. Compositions were fabricated by conventional solid state reaction -sintering of the compositional range Ca 1-x Ce x ZrTi 2-2x Cr 2x O 7 (0.00 ≤ x ≤ 0.35) and the phase evolution of zirconolite was characterised to determine the incorporation of Ce, using Cr 3+ to maintain electro-neutrality across the target zirconolite phase. The solid solution parameter x was limited at x = 0.35 as this would correspond to an upper Pu content of ∼ 20 wt. %, which is of interest for application. Reactive Spark Plasma Sintering (RSPS) [13] was successfully applied to produce Ca 0.80 Ce 0.20 ZrTi 1.60 Cr 0.40 O 7 (x = 0.20) ceramics of near theoretical density, however, this aforded complete reduction to Ce 3+ , with a deleterious impact on the phase assemblage, as a result of the use of graphite die and vacuum in the SPS system [14].

Materials synthesis
All compositions were synthesised by conventional solid state reaction -sintering between CaTiO 3 (Sigma Aldrich, 99.9 % trace metals basis), ZrO 2 (Sigma Aldrich, 99.9 % trace metals basis), Cr 2 O 3 (Sigma Aldrich, 99.9 % trace metals basis), TiO 2 (anatase, Sigma Aldrich, 99.9 % trace metals basis) and CeO 2 (Acros Organics, 99.9 % trace metals basis). Precursor materials were dried and weighed according to Ca 1- x Ce x ZrTi 2-2x Cr 2x O 7 where 0.00 ≤ x ≤ 0.35 (Δx = 0.05 increments). Batches were homogenised by planetary milling at 400 rpm for 20 min, using isopropanol as a milling agent. Slurries were discharged and left for 12 h at 90°C to evaporate excess solvent. Approximately 0.3 g of dried material was uniaxially pressed into 10 mm green body pellets under a force of 3 t (∼ 260 MPa), and reacted for 20 h in air at 1350°C. A specimen was also produced by reactive spark plasma sintering (RSPS).
Powder mixture with nominal composition Ca 0.80 Ce 0.20 ZrTi 1.60 Cr 0.40 O 7 (x = 0.20) was pressed into a graphite die (with graphite foil spacers) then sintered using a HP-D 1050 SPS system at 1320°C under vacuum with a dwell time of 30 min (100°C/min, 5 V and 1 kA at maximum sintering temperature). During the sintering process, the uniaxial pressure was maintained at 15 MPa.

Materials characterisation
Samples were prepared for powder X-ray difraction (XRD) by hand grinding the reacted specimens in an agate pestle and mortar using IPA. A Bruker D2 Phaser difractometer itted with Lynxeye position sensitive detector was used; difraction data were acquired in the range 5˚≤ 2θ ≤ 80˚, with step size of 0.02˚and 1 s per step, using Cu Kα radiation (λ =1.5418 Å, Ni ilter). Unit cell parameters for the zirconolite unit cell and quantitative phase analysis (QPA) were calculated using the Rietveld method, using the GSAS software with EXPGUI interface [15]. Samples were sectioned, mounted in cold setting resin, and polished to a 1 μm optical inish with diamond suspension prior to microstructure analysis using a Hitachi TM3030 scanning electron microscope (SEM), operating with a 15 kV accelerating voltage and a working distance of approximately 8 mm. Energy dispersive X-ray spectroscopy (EDS) was undertaken using a Bruker Quantax Energy Dispersive X-ray Spectrometer for compositional analysis.
Ce L 3 XANES and Cr K XANES (X-ray absorption near edge structure) spectra for cold pressed and sintered samples (CPS) were acquired on beamline B18 at the Diamond Light Source (Oxford, UK) which operates at 3 GeV with a ring current of 300 mA. B18 is a bending magnet beamline producing X-rays in the energy range 2-20 keV. The optical arrangement consists of a water cooled vertically collimated Si mirror, a double crystal Si(111) monochromator, a double toroidal focusing mirror, and harmonic rejection mirrors. Samples were measured in QuEXAFS mode with simultaneous collection of transmission and luorescence data. Cr K-edge spectra were recorded between 5789 and 6550 eV in energy steps of 0.2 eV using an acquisition time of 35 ms step −1 . Ce L 3 -edge spectra were recorded between 5523 and 6154 eV in energy steps of 0.25 eV using an acquisition time of 50 ms step −1 . To improve data quality, the beam spot size was defocused to ca. 1.0 mm and multiple scans were acquired and averaged. Incident and transmitted beam intensities were measured using ionisation chambers which were illed with mixtures of He and N 2 gas. Fluorescence emission was detected using a 36 channel Ge solid state detector. In order to ensure energy reproducibility (± 0.05 eV) and allow absolute energy calibration, a Cr foil was measured in the reference measurement position; the edge position (measured as the irst inlection point in the derivative of the absorption) was aligned to the expected value of 5989 eV [16].
XANES data for the sample processed by RSPS was acquired at the Ce L 3 -edge on beamline BL-27B at the Photon Factory (PF) acceleratorbased light source facility (Tsukuba, Japan). The PF operates at 2.5 GeV with a ring current of 300 mA, and supplies beamline BL-27B with radiation energy in the energy range 4-20 keV via a bending magnet. The optical arrangement consists of a double crystal Si(111) monochromator and slits were used to reduce the size of the beam to 3 mm in the horizontal and 1 mm in the vertical. The incident and transmitted photon spectra were measured with ionisation chambers which were gas illed to absorb ca. 20 % of the X-rays in the front chamber and 90 % in the second chamber [17]. Spectra were recorded between 5590 and 5990 eV with energy steps of 5 eV (5590-5690), 0.5 eV (5690-5790), 1 eV (5790-5890) and 5 eV (5890-5990). An accumulation time of 1 s step −1 was used for all regions and multiple scans were acquired and averaged to improve the signal to noise. To ensure energy reproducibility (± 0.05 eV) a spectrum was collected from a CePO 4 standard between each sample.
Specimens were prepared for measurement by homogenising and suspended in polyethylene glycol (PEG) to produce samples with a thickness of approximately one absorption length. Spectra were also acquired from a selection of standards: CeO 2 , CePO 4 , KCrO 3 , Cr 2 O 3 and CrO 2 . Data reduction and analysis was performed using the Demeter software package [18]. A linear combination it of Ce L 3 -edge XANES spectra was performed with respect to standards, in order to determine the relative amount of Ce 3+/4+ species. The corresponding R-factor (mean square sum of the residual for each data point) was determined. The pre-edge region in the Cr XANES spectra were deconvoluted by an edge step (modelled by an arctangent function) and three pre-edge features (itted by Gaussian functions). To ensure itting quality, the Rfactor was optimised below 0.001. This methodology was based on the work of Wilke et al. [19].
Raman spectra of powdered specimens were collected using the Horiba-XploRA Plus system with 532 nm air cooled Ar + laser line and laser power of approximately 5 mW. Specimens for transmission electron microscopy were prepared via the powder route. Powders derived from the grinding of the sintered pellets were dispersed with isopropanol, with a small amount of the mixture pipetted from the top of the solvent line onto a holey lined Cu grid (Agar). Selected Area Electron Difraction patterns (SAED) were acquired for specimens with nominal composition x = 0.05 and x = 0.35, using an FEI Tecnai T20 operating at 200 keV. SAEDPs were captured using a CCD camera.

Phase assemblage
It was determined from powder X-ray difraction data (  [20]. The unit cell parameters for the zirconolite phase over the range 0 ≤ x ≤ 0.35 are listed in Table 1. Zirconolite-2M was formed as a single phase in the compositional range 0 ≤ x ≤ 0.15, indicating substitution of Ce and Cr into the host phase was achieved. However, minor relections at 2θ = 33.1˚and 2θ = 33.7˚were evidenced for x = 0.20, which were attributed to CaTiO 3 and Cr 2 O 3 , comprising 1.2 ± 0.2 wt. % and 0.6 ± 0.2 wt. % of the total phase assemblage, respectively, as determined from the Rietveld analysis. With increased Ce/Cr content, 0.20 ≤ x ≤ 0.35, greater relative proportions of CaTiO 3 and Cr 2 O 3 secondary phases were found to be present, indicating the Ce and Cr were not fully substituted within the zirconolite-2M structure above x = 0.15. Despite the minor phase separation, zirconolite-2M remained present at approximately 94 wt. % when targeting x = 0.35, with perovskite accounting for ∼ 3 wt. % of the total phase assemblage. Reinement of the powder difraction data for the specimen corresponding to x = 0.35 is also displayed in Fig. 2.  Fig. 3c and d, respectively. In both instances, the patterns indexed to the zirconolite-2M polytype (space group C2/c), in agreement with XRD data. There was no indication of polymorphic transitions to the zirconolite-4M or other polytypes, which has been previously observed when partitioning Ce/U in the Zr 4+ site. For example, in the case of Vance et al., extensive "streaking" of crystalline relections was observed in difraction data for compositions of zirconolite-4M in the range CaZr 1-x U x Ti 2 O 7 (0.20 ≤ x ≤ 0.30) [21].

Microstructure analysis
Polished surfaces for each composition were analysed by SEM-EDS analysis. A representative micrograph for x = 0.05 is displayed in Fig. 4. A matrix with homogeneous BSE contrast was observed, conirming the formation of single phase zirconolite-2M, with an average composition of Ca 1.01(7) Ce 0.03(1) Zr 0.88(9) Ti 2.00(5) Cr 0.08(1) O 7 determined by EDS analysis (oxygen content was assumed to be stoichiometric, given the poor sensitivity of EDS towards oxygen determination). The compositional analysis was in good agreement with targeted stoichiometry, considering the associated error of EDS analysis (quoted in parentheses to 1 s.d.). It was observed that all specimens exhibited a sintered microstructure but with high internal porosity; Archimedes density measurements and EDS data for each composition are summarised in Table 2. The high internal porosity would be undesirable from a disposability perspective, if replicated in a corresponding Pu solid solution. The porosity would result in an increased contact surface area available for dissolution, and this may be reduced by increasing the sintering temperature and/or applying the external pressure. It should be noted that with reference to the prototypical CaZr x Ti 3-x O 7 composition, the site occupancy and distribution of [Ti]/[Zr] is known to be sensitive to processing temperature: ratios between 1.31 and 2.75 at 1500°C and 1300°C have been reported, respectively [8]. Considering the processing temperature used (1350°C), the observed [Ti]/ [Zr] ratio falls within the expected range of values, considering the associated error of EDS analysis (quoted in parenthesis to 1 s.d.).
In agreement with powder XRD, no other phases were observed in the microstructure for x = 0.10 and x = 0.15 compositions (data not shown). Phase separation was observed in the microstructure at x = 0.20, shown in Fig. 5. Cr 2 O 3 and perovskite were distinguished from the bulk matrix by BSE contrast. EDS analysis conirmed that Ce and Cr were also incorporated within the perovskite phase. The accommodation of Ce within the perovskite phase suggests partial Ce reduction to Ce 3+ , as the formation of Ce 3+ is known to promote and stabilise Ce-bearing perovskite [22,23]. A comparison of the EDS spectra of the perovskite phase and zirconolite-2M in the x = 0.35 composition is given in Fig. 6  displayed in Fig. 7. A minor fraction of unincorporated CeO 2 was evidenced in the x = 0.35 specimen by SEM analysis of the microstructure, yet, this was not detected by XRD. A ZrO 2 phase was also detected, however this constituted < 1 wt. % of the overall phase assemblage.

Oxidation state and local environment of Ce and Cr
Ce has been observed to exist as Ce 4+ and Ce 3+ in zirconolite wasteforms despite processing under oxidizing conditions. The oxidation state of cerium was determined through the use Ce L 3 XANES, acquired alongside Ce 3+ and Ce 4+ reference compounds (CePO 4 and CeO 2 , respectively). These data are displayed in Fig. 8. The CePO 4 (monazite) spectrum was comprised of a single intense white line feature characteristic of Ce 3+ speciation (in CePO 4 (monazite), Ce 3+ is coordinated as a CeO 9 polyhedron). The white line features for CeO 2 consisted of an asymmetric doublet with lower relative intensity than observed for CePO 4 (in CeO 2 , Ce 4+ is coordinated as a CeO 8 polyhedron). The initial and inal state electron conigurations associated with the white line features for both reference compounds are discussed elsewhere [26]. It was apparent from visual inspection of normalised XANES spectra that Ce was present mainly as Ce 4+ in Ca 1-x Ce x ZrTi 2-2x Cr 2x O 7 compositions, remaining analogous to Pu 4+ . However closer inspection of XANES data revealed a small shift in the edge position toward lower energy (with respect to the CeO 2 reference compound) and a small change in the relative intensity of the asymmetric doublet features, consistent with a minor fraction of Ce 3+ . Linear combination itting (LCF) of each composition was performed using the CeO 2 and CePO 4 reference compounds, the results are summarised in Table 3, which conirmed the presence of 15-30% Ce 3+ depending on the composition. Reduction of Ce 4+ to Ce 3+ results in an increase in average ionic radius (0.97 Å-1.143 Å for 8-fold coordinated species [27]), favoring partitioning within the Ca 2+ site in the perovskite phase, as was conirmed by SEM/EDS data.
XANES analysis at the Cr K-edge was undertaken on all compositions, and a series of Cr reference compounds representing various oxidation states and coordination environments. XANES spectra acquired at the Cr K-edge are displayed in Fig. 9. It was evident that all Crzirconolite specimens retained very similar edge features, consisting of an intense white line at approximately E 0 = 6002 eV (E 0 derived from the maxima of the irst derivative). For all zirconolite spectra a minor pre-edge feature was also observed. The E 0 position of the Cr 2 O 3 XANES spectrum was lower at approximately E 0 = 6000 eV, and the Cr 2 O 3 white line also featured a signiicant shoulder on the absorption edge and a moderately intense pre-edge feature, which were not observed in the measured zirconolite spectra. The E 0 position and white line features for YCrO 3 were more consistent with features observed in the zirconolite XANES data, inferring similar Cr speciation and chemical environment. It was evident, therefore, that Cr was present primarily in a single dominant Cr environment, for all zirconolite compositions similar to that in YCrO 3 , in which Cr 3+ adopts octahedra coordination by oxygen. Furthermore, the XANES spectrum for K 2 CrO 4 (in which Cr 6+ adopts tetrahedral coordination by oxygen) is characterised by an intense pre-edge feature, which exists at ∼ 15 eV prior to the main absorption edge. This feature was not observed for any Cr-containing zirconolite specimens, conirming the absence of Cr 6+ . Several isosbestic points were also revealed as the concentration of x was increased -these are labelled as A, B, C and D on Fig. 9. The presence of isosbestic points in a closed system is attributed to the contribution of the absorbing atom (i.e. Cr) in two or more chemical environments. Considering the reference spectra for Cr 2 O 3 , the formation of isosbestic points is a result of increased contribution of free Cr 2 O 3 present in the sample. Upon closer inspection, points A, B, C and D become apparent at some nominal composition in the interval 0.15 ≤ x ≤ 0.20. This was clearly distinguished by the appearance of the feature at 6012 eV (between points B and C in Fig. 9) which was common to both Cr 2 O 3 and zirconolite spectra. The intensity of this feature increased with elevated Cr content, x > 0.15, suggesting the presence of Cr 2 O 3 . This is in agreement with powder XRD data and microscopic analysis, in which a secondary Cr 2 O 3 phase was observed in the microstructure for compositions where x > 0. 15  hybridisation of Cr-3d energy levels [28]. These data are in agreement with Cabaret et al. in which three local electric quadrupole transitions were predicted for K pre-edge features of octahedrally coordinated Cr 3+ species from irst principles [29]. The normalised intensity and energy centroid of the itted Gaussian contributions to the pre-edge features of the x = 0.15 sample and YCrO 3 were determined to be similar. In the ideal YCrO 3 structure, CrO 6 octahedtra are arranged in a corner sharing motif, analogous to the hexagonal tungsten bronze layers of TiO 6 octahedra in the zirconolite structure. Cr 2 O 3 also contains Cr 3+ in octahedral coordination, however, the face and edge sharing of CrO 6 octahedra results in itted Gaussian contributions to the pre-edge feature that are dissimilar in normalised intensity and centroid energy, compared to those of the x = 0.15 and YCrO 3 . Thus, the Cr Kedge XANES analysis presented here is consistent with Cr substitution within the octahedral TiO 6 sites of the zirconolite structure.
Raman spectra in the wavenumber range 100 -1000 cm −1 are presented in Fig. 11. Many active Raman vibrational modes are observed due to the low symmetry of the monoclinic unit cell. A prominent Raman mode at 780 cm −1 was observed for all samples, and a variety of intense modes were evidenced in the 150 -400 cm −1 range.

Composition
Zirconolite-2M Unit Cell Parameters At low concentrations e.g. x = 0.05, the spectrum is in good agreement with the active Raman modes observed in our previous studies for the zirconolite-2M structure [30]. As the substituted concentration is elevated, many of these features undergo broadening, in particular those at a lower Raman shift. This is attributed to disorder induced by Cr 3+ and Ce 4+/3+ partitioning in Ti 4+ and Ca 2+ sites respectively. The degree of broadening does not appear to change beyond x ≥ 0.20, consistent with the observations that Cr 3+ was not fully incorporated into the zirconolite-2M structure beyond this concentration. Whilst it is known that the dominant vibrational mode at 780 cm −1 is attributed to the symmetric vibrations of TiO 6 octahedra, we are not aware of any studies that have directly assigned other prominent modes to speciic bonds in the zirconolite structure [31]. However, it is generally accepted that vibrational modes below 700 cm −1 are attributed to CaO 8 and ZrO 7 polyhedral vibrations, and TiO 6 bending and stretching vibrations [30]. Jafar et al. performed Raman spectroscopic studies on the CaZrTi 2 O 7 -Y 2 Ti 2 O 7 system, where a structural transition from zirconolite-2M to cubic pyrochlore was observed, with the formation of the zirconolite-4M polytype as an intermediate phase [32]. A characteristic feature of the transition to the zirconolite-4M and pyrochlore structures was the abatement of the TiO 6 peak at 780 cm −1 . This feature remains prominent in this instance, inferring zirconolite crystallised in the 2M coniguration throughout the solid solution.

Reactive spark plasma sintering
In order to overcome poor densiication observed in reactively sintered specimens, a sample of milled precursor with nominal composition x = 0.20 was processed by reactive spark plasma sintering (RSPS). It should be noted that this is not the same as consolidation via SPS, as in this instance, the reaction and consolidation occur simultaneously. Rather, here, we report the formation of heavily-doped zirconolite with a rapid processing time of less than 60 min. We have previously reported the formation of nearly single phase CaZrTi 2 O 7 using this technique [13]. Whilst zirconolite wasteforms with incorporated Ce were reported by Clark et al., these were only consolidated by SPS subsequent to zirconolite formation via solid state reaction [22,33]. Powder XRD data comparing the phase assemblage of the RSPS and CPS product are displayed in Fig. 12. The sample processed by RSPS formed zirconolite-2M as the major phase, comprising 64.9 ± 0.2 wt. % of the overall phase assemblage. It was noted that relative phase fraction of perovskite (labelled with closed circles on Fig. 12) signiicantly increased in comparison to the CPS sample, from 1.2 ± 0.2 wt. % to 19.3 ± 0.4 wt % of the phase assemblage. There was also an increase in the quantity of ZrO 2 (labelled by open circles on Fig. 12), however this was still < 3 wt. % of the overall phase assemblage. A relative increase in Cr 2 O 3 was also observed, from 0.6 ± 0.2 wt. % to 2.1 ± 0.3 wt. %. An accessory pyrochlore-structured phase with prototypical composition Ca 2 Ti 2 O 6 (JCPDS No. 04-0103, space group Fd-3m) was formed, with an intense relection at 2θ = 31.1˚ [34]. It was not possible to resolve the Ca 2 Ti 2 O 6 by SEM, presumably due similar backscattered electron coeicient to the accompanying perovskite and/or zirconolite-2M phase, yet the phase comprised 10.9 ± 0.5 wt. %. It is likely that partial incorporation of Ce occurred within both CaTiO 3 and Ca 2 Ti 2 O 6 phases, producing similar BSE contrast. Our previous attempts to form CaZrTi 2 O 7 by RSPS resulted in approximately 2.1 wt. % CaTiO 3 , with partial Ti 3+ speciation conirmed by paramagnetic electron resonance measurements [13]. Furthermore, in the case of Clark et al., a sample of CaZr 0.5 Ce 0.5 Ti 2 O 7 was sintered in air, and subsequently consolidated by SPS [22]. During this process, a perovskite fraction with stoichiometry Ca 0.40 Ce 0.40 TiO 3     was stabilised due to the reducing conditions imposed by the SPS graphite die, with 90 % reduction to the Ce 3+ species determined by XANES. Therefore, it was understood that the SPS environment caused the partial reduction of Ce 4+ to Ce 3+ promoting the formation of a Ceperovskite phase. XANES spectra for the sample corresponding to x = 0.20 processed by air sintering and RSPS, with respect to Ce 3+ and Ce 4+ reference compounds, are displayed in Fig. 13. The XANES features for the RSPS sample consisted of a single intense feature consistent with the Ce 3+ reference compound. Linear combination analysis determined that a complete reduction to the Ce 3+ species was achieved. In consideration of the reduction during SPS, the reduction of Ce 4+ or Pu 4+ and formation of undesirable perovskite phase may be suppressed by use of a boron nitride die [35], instead of graphite die used in this work. SEM-EDS analysis of the polished RSPS surface was in good agreement with XRD data, demonstrated in Fig. 14

Conclusions
The incorporation of Ce within CaZrTi 2 O 7 zirconolite, with Cr acting as a charge compensation species, was investigated for the Ca 1-   x Ce x ZrTi 2-2x Cr 2x O 7 system. The densiication achieved by a cold-press and sinter method was poor, and therefore this is not a practical choice of solid solution for Pu 4+ immobilisation from the perspective of the ceramic microstructure, despite the eicacy of Cr as a charge compensator. High porosity was evidently overcome by the application of reactive spark plasma sintering, yet a signiicant disruption to the phase assemblage was observed, attributed to the total reduction of Ce to Ce 3+ and subsequent increase in the Ce-perovskite phase fraction. Electron difraction analyses were consistent with zirconolite adopting the 2M polytype at all levels of Ce incorporation for the CPS solid solution. Powder XRD and SEM analysis conirmed that zirconolite-2M was formed as a single phase for 0.00 ≤ x ≤ 0.15. Cr 3+ was fully incorporated within the zirconolite phase up to x = 0.15, after which phase separation occurred. Ce incorporation within the perovskite phase was observed for x > 0.15, yet even at maximum targeted wasteloading, the phase assemblage was comprised of ∼ 94 wt. % zirconolite-2M. Ce L 3 XANES data and linear combination itting with reference compounds conirmed ∼ 80 % Ce 4+ . It was concluded from Cr K-edge XANES data that Cr was incorporated into the zirconolite phase as Cr 3+ . Deconvolution of the Cr K XANES pre-edge region for zirconolite produced features consistent with YCrO 3 reference compound, inferring octahedral coordination in the zirconolite phase as expected. The potential behaviour of a Pu bearing zirconolite ceramic charge balanced with Cr 3+ could be hypothesised using this data. Pu can be present as trivalent and tetravalent cations, similar to Ce. It was inferred from Ce L 3 XANES that Ce was distributed mainly as Ce 4+ . Accounting for discrepancies in redox behaviour, Pu is expected to be uniformly present as Pu 4+ , as Ce has a greater propensity to undergo auto-reduction at high processing temperatures. At values for which 0.05 ≤ x ≤ 0.15, Pu 4+ could be expected to fully partition within zirconolite-2M. However, these data imply that a conventional sintering route using Cr as a charge compensating cation may produce a microstructure with elevated porosity. This paper also highlights an important caveat for the deployment of reactive spark plasma sintering for Pu-bearing zirconolites, demonstrating the need for careful compositional design and atmosphere selection to control redox reactions.

Declaration of Competing Interest
The authors declare that they have no known competing inancial interests or personal relationships that could have appeared to inluence the work reported in this paper.