Phase Evolution in the CaZrTi2O7-Dy2Ti2O7 System: A Potential Host Phase for Minor Actinide Immobilization.

Zirconolite is considered to be a suitable wasteform material for the immobilization of Pu and other minor actinide species produced through advanced nuclear separations. Here, we present a comprehensive investigation of Dy3+ incorporation within the self-charge balancing zirconolite Ca1-xZr1-xDy2xTi2O7 solid solution, with the view to simulate trivalent minor actinide immobilization. Compositions in the substitution range 0.10 ≤ x ≤ 1.00 (Δx = 0.10) were fabricated by a conventional mixed oxide synthesis, with a two-step sintering regime at 1400 °C in air for 48 h. Three distinct coexisting phase fields were identified, with single-phase zirconolite-2M identified only for x = 0.10. A structural transformation from zirconolite-2M to zirconolite-4M occurred in the range 0.20 ≤ x ≤ 0.30, while a mixed-phase assemblage of zirconolite-4M and cubic pyrochlore was evident at Dy concentrations 0.40 ≤ x ≤ 0.50. Compositions for which x ≥ 0.60 were consistent with single-phase pyrochlore. The formation of zirconolite-4M and pyrochlore polytype phases, with increasing Dy content, was confirmed by high-resolution transmission electron microscopy, coupled with selected area electron diffraction. Analysis of the Dy L3-edge XANES region confirmed that Dy was present uniformly as Dy3+, remaining analogous to Am3+. Fitting of the EXAFS region was consistent with Dy3+ cations distributed across both Ca2+ and Zr4+ sites in both zirconolite-2M and 4M, in agreement with the targeted self-compensating substitution scheme, whereas Dy3+ was 8-fold coordinated in the pyrochlore structure. The observed phase fields were contextualized within the existing literature, demonstrating that phase transitions in CaZrTi2O7-REE3+Ti2O7 binary solid solutions are fundamentally controlled by the ratio of ionic radius of REE3+ cations.

Introduction

Zirconolite structured materials have been widely studied for the immobilization of actinide-rich radioactive waste streams, due to excellent chemical alteration resistance and radiation tolerance.[1−5] This includes use as a major constituent of the various SYNROC assemblages for the disposition of high-level actinide-rich wastes derived from nuclear fuel reprocessing.[6,7] The CaZrTi2O7 parent structure has been shown to accommodate U, Pu, Np, and Cm and is therefore a suitable host matrix for minor actinide (MA) species such as Am.[8−11] Natural analogue specimens have also been shown to retain ∼20 wt % U/Th over geological timescales.[12] The parent structure ABC2O7 is derivative of the anion deficient fluorite structure type and is closely related to the pyrochlore A2B2O7 family of minerals. The ideal zirconolite unit cell is composed of planes of corner-sharing CaO8 and ZrO7 polyhedra, interleaved by hexagonal tungsten bronze (HTB)-style motifs along the [001] direction.[13] Ti4+ cations are distributed across three distinct sites in the HTB plane: two Ti4+ sites are arranged in TiO6 octahedra and one Ti4+ is 5-fold coordinated in a 50% statistically coordinated trigonal bipyramidal site. In this idealized structure description, cation and HTB layers are integrated 1:1 along [001], related by a 180° rotation along the c* axis. Due to the two-layered structure, stoichiometric CaZrTi2O7 is referred to as the zirconolite-2M polytype, with reference to the monoclinic symmetry of the unit cell; this polytype has previously been demonstrated to crystallize over the compositional range CaZrTi3–O7 for 0.8 < x < 1.3.[14]

As the zirconolite structure has three distinct cation acceptor sites, the solubility of REE3+/Ac4+ species is extensive. The incorporation of Ce4+/U4+/Pu4+ species within the Zr4+ site in zirconolite is accommodated by several structural transitions, first from zirconolite-2M to the zirconolite-4M polytype.[8,9,15] The zirconolite-4M polytype is described by Coelho et al. as an admixture of zirconolite-2M and pyrochlore; four HTB-type layers interlaced with Ca/Zr polyhedra (zirconolite), and Ca/Ti polyhedra (pyrochlore), resulting in unit cell doubling along the c-axis from ∼11 to 23 Å.[16] Further isovalent substitution of cations within the Zr4+ site promotes a structural transformation from zirconolite-4M to pyrochlore, although it should be noted that this does not occur for the corresponding CaZr1–ThTi2O7 solid solution, for which the intermediate 4M phase does not form.[17] Cubic pyrochlore-structured materials (parent structure A2B2O7—space group Fd3̅m; Z = 8) have attracted significant interest in many areas of solid-state chemistry, with titanate and zirconate pyrochlores (A2Ti2O7 and A2Zr2O7, respectively) developed as potential wasteforms for actinides, due to high radiation stability.[18−24] The rare earth pyrochlore structure is derived from the fluorite (AO2) superstructure, with one-eighth of the oxygen atoms replaced by vacancies, and the A3+ and B4+ cations in 8- and 6-fold coordination with oxygen, respectively. These cations are ordered along the [110] direction, resulting in the unit cell adopting cubic symmetry. The phase stability of A2B2O7-type structures is dependent on the ionic radius ratio of the A and B cations; the ordered cubic pyrochlore structure is stable in the range 1.46 < rA/rB < 1.78. Compounds with rA/rB below this range adopt a disordered defect-fluorite structure, while compounds with rA/rB > 1.78 crystallize with monoclinic layered perovskite-related structure.[19]

The present study aims to systematically evaluate the phase transitions in the CaZrTi2O7–Dy2Ti2O7 system with the progressive accommodation of Dy3+, acting as a surrogate for actinide species such as Pu3+, Am3+, Cm3+, and Np3+. These data are expected to complement existing data for closely related zirconolite solid solutions Ca1–Zr1–Gd2Ti2O7, Ca1–Zr1–Y2Ti2O7, Ca1–Zr1–Nd2Ti2O7, Ca1–Zr1–Sm2Ti2O7, and Ca1–Zr1–Ce2Ti2O7+δ.[4,25−28] While the incorporation of Dy3+ within the zirconolite structure has not been previously reported, the Dy2Ti2O7 pyrochlore end member has attracted notable interest given its applications as a spin-ice compound, due to prominent geometric frustration.[29,30] Furthermore, Dy has been previously used as a surrogate for Am for the fabrication of AmN and (Am-Pu)N compounds, on the basis of ionic radii constraints and expediency (Am3+ = 1.09 Å; Dy3+ = 1.03 Å in 8-fold coordination).[31,32]

Experimental Procedure

Materials Synthesis

All materials used were fabricated by a conventional solid-state synthesis route from component oxides, targeting the solid solution Ca1–Zr1–Dy2Ti2O7 (0.10 ≤ x ≤ 1.00, Δx = 0.10). Precursors CaTiO3 (Sigma-Aldrich, 99.9%), ZrO2 (Sigma-Aldrich, 99.9%), TiO2 (anatase—Sigma-Aldrich, 99.9%) dried at 180 °C, and Dy2O3 (Alfa Aesar, 99.9%) dried at 800 °C were weighed according to the targeted composition, to yield 3 g batches. The oxide reagents were added to a ZrO2-lined milling jar and homogenized with Y-stabilized ZrO2 milling media and isopropanol for 20 min, using a Fritsch Pulverisette-23, operating at 25 Hz for 20 min. For each composition, the powder slurries were discharged, sieved to separate milling media, and dried at 80 °C overnight to evaporate excess solvent. Approximately 0.5 g of each composition was prepared for sintering by first compacting into the walls of a hardened steel die, under 3 tonnes of uniaxial pressure, forming powder compacts 13 mm in diameter. The pellets were then placed onto a zirconia crucible and sintered in air at 1400 °C (Δ5 °C min–1) for 24 h. Once cooled, the pellets were reground using a pestle and mortar, repressed into pellets, and subjected to a second sintering regime for a further 24 h at 1400 °C (Δ5 °C min–1) to promote phase purity. After the second sintering step, the pellets were recovered from the furnace for analysis.

Materials Characterization

Powder X-ray diffraction (XRD) was conducted using a Bruker D2 Phaser fitted with a Lynxeye position-sensitive detector. Data were acquired in the range 5° ≤ 2θ ≤ 80° (Δ0.02°) using Cu Kα radiation (λ = 1.5418 Å, Ni Filter), operating at 30 kV and 10 mA. Phase identification was achieved using the PDF4+ database. Quantitative phase analysis and unit cell dimensions were calculated by Rietveld analysis of powder XRD data, using the Bruker TOPAS software package. Prior to analysis by scanning electron microscopy (SEM), sintered pellets were mounted in cold setting resin and polished to a 1 μm optical finish. SEM data were collected using a Hitachi TM3030 operating with a 15 kV accelerating voltage at a working distance of 8 mm. Energy-dispersive X-ray spectrometry (EDS) for semiquantitative compositional analysis was conducted using a Bruker Quantax 70 spectrometer. EDS mapping was performed over an area of 140 × 105 μm2 for approximately 10 min.

X-ray absorption spectroscopy (XAS) at the Dy L3-edge was conducted at the Photon Factory Synchrotron Facility (Tsukuba, Japan) using beamline BL-27B, in a conventional transmission configuration. Spectra were collected at the Dy L3-edge (7790 eV) for specimens corresponding to nominal composition x = 0.10, 0.30, 0.60, and 1.00. Spectra were collected between 7590 and 8540 eV at the following steps, with a count time of 1s/step: 5 eV (7590–7760 eV), 0.5 eV (7760–7860 eV), 1 eV (7860–7940 eV), and 2 eV (7940–8540 eV). Samples were finely ground and dispersed in poly(ethylene glycol), to a concentration corresponding to one absorption length, and pressed into 13 mm pellets. Spectra were collected alongside Dy2O3, Dy2TiO5, and Dy2Ti2O7 reference compounds, representing Dy3+ in 6-fold, 7-fold, and 8-fold coordination with oxygen, respectively. Calibration of all XAS data was performed by aligning the Dy2O3 reference compound from this dataset to a previously collected Dy2O3 standard that was calibrated using a Co foil (edge position 7709 eV). All edge positions were chosen as the peak of the first derivative. Data reduction and fitting of the Dy L3 edge EXAFS were performed using Artemis and Athena with FEFF6.[33] Best-fit models were informed by pyrochlore and zirconolite-2M structures from Farmer et al.[34] and Whittle et al.,[35] respectively. The F-test methodology for EXAFS was employed to test whether the addition of various shells was valid and statistically improved the fit, the results of which are denoted by α.[36] For α, >67% is equal to 1σ and >95% is equal to 2σ in terms of standard deviation.

Raman spectra were collected for each sample using a Horiba-XploRA Plus system, operating with a 532 nm air-cooled Ar+ laser line, with a laser power of 5 mW. Powdered specimens were flattened onto a glass slide and placed incident to the laser line, with spectra collected in the range 100–1100 cm–1. Transmission electron microscopy (TEM) was undertaken using either a JEOL F200 or an FEI Tecnai T20, both operating at 200 keV, with both micrographs and electron diffraction (ED) patterns taken using a CCD camera. Specimens were prepared for TEM via the crushed grain powdered route, whereby a small amount of powder was rapidly ground in isopropanol and the resultant solution pipetted onto a holey lined Cu mesh grid.

Results and Discussion

Systematic Examination of Phase Evolution in the Ca1–Zr1–Dy2Ti2O7 (0.10 ≤ x ≤ 1.00) System

The phase evolution of Ca1–Zr1–Dy2Ti2O7 ceramics was analyzed by powder X-ray diffraction (Figure ). Three distinct phase fields were identified, corresponding to mixtures of zirconolite-2M (C2/c), zirconolite-4M (C2/c), and cubic pyrochlore (Fd3̅m). The x = 0.10 composition was found to form single-phase zirconolite-2M, characterized by the doublet at 2θ = 30.5° corresponding to the (221) and (40-2) reflections, the (004) reflection at 2θ = 31.9°. Unit cell dimensions (Table ) as determined by Rietveld analysis were in agreement with those reported for closely related Ca1–Zr1–Gd2Ti2O7 solid solutions.[25] A structural transformation from zirconolite-2M to the zirconolite-4M polytype was observed in the compositional range 0.20 ≤ x ≤ 0.30, as characterized by the appearance of intense supercell reflections at 2θ = 7.8 and 31.1°, attributed to the (002) and (008) reflections in the zirconolite-4M structure, respectively. A representative section of the microstructure for the sample corresponding to x = 0.20 is given in Figure . Two distinct phases were distinguished by variation in backscattered electron contrast, identified to be zirconolite-2M and zirconolite-4M, in agreement with powder XRD data. The phase labeled A was determined by EDS analysis to be zirconolite-2M. The phase labeled B was consistent with zirconolite-4M, appearing brighter than the bulk matrix, given the expected higher solubility of Dy3+ in the 4M polytype. The average composition of both zirconolite phases was derived from semiquantitative EDS analysis and is in general agreement with targeted nominal stoichiometry (Table ). The zirconolite-2M phase accounted for just 19.0 ± 0.8 wt % of the overall phase assemblage when targeting x = 0.30, yet, a two-phase mixture of zirconolite-4M and cubic pyrochlore was observed in the range 0.40 ≤ x ≤ 0.50; hence, zirconolite-4M was not isolated as a single phase in the solid solution. This was unsurprising, as the 4M phase has only previously been reported to crystallize in single phase over a narrow compositional range, sensitive to preparation route, ionic radii of dopant/host site, and targeted solid solution regime of REE3+/Ac4+.[8,9,16] The cubic pyrochlore phase was evidenced by the appearance of reflections at 2θ = 15.1 and 29.2°, corresponding to reflections in the (111) and (113) plane, appearing at x = 0.40. The microstructure of the x = 0.50 sample is shown in Figure and clearly displays a microstructure dominated by two phases. The phases were confirmed by EDS analysis to be zirconolite-4M and cubic pyrochlore (labeled as A and B in Figure , respectively) Additional reflections were not observed for 0.60 ≤ x ≤ 1.00, indicating complete Dy2O3 substitution within the zirconolite-pyrochlore mixture, and confirmed by SEM analyses (Figure ).

Figure 1

Powder X-ray diffraction data for Ca1–Zr1–Dy2Ti2O7 specimens in the compositional range (0.0 ≤ x ≤ 1.0). The (002) and (008) zirconolite-4M reflections are highlighted by closed circles (●). Pyrochlore reflections are indexed by relevant (hkl) indices.

Table 1

Quantitative Phase Analysis and Unit Cell Parameters of Each Phase as Determined from Rietveld Analysis of Powder X-ray Diffraction Data

		unit cell parameters
nominal composition	phase assemblage (wt %)	a (Å)	b (Å)	c (Å)	β (deg)	V (Å3)
x = 0.10	zirconolite-2Ma	12.47925(30)	7.28070(17)	11.37370(28)	100.6002(13)	1015.752(42)
x = 0.20	zirconolite-2M (64.9 ± 0.2)	12.49392(54) - 2M	7.28442(32) - 2M	11.37380(51 ) - 2M	100.6038(23) - 2M	1017.463(77) - 2M
	zirconolite-4M (35.1 ± 0.2)	12.50991(76) - 4M	7.17395(43) - 4M	22.95916(81 ) - 4M	84.8136(48) - 4M	2052.04(19) - 4M
x = 0.30	zirconolite-2M (19.0 ± 0.8)	12.5102(23) - 2M	7.2721(13) - 2M	11.3692(20) - 2M	100.465(23) - 2M	1017.10(32) - 2M
	zirconolite-4M (81.0 ± 0.8)	12.45916(35) - 4M	7.19043(26) - 4M	22.97562(45) - 4M	84.8373(26) - 4M	2049.96(10) - 4M
x = 0.40	zirconolite-4M (47.8 ± 0.7)	12.4532(31) - 4M	7.1922(18) - 4M	23.0213(27) - 4M	84.790(17) - 4M	2053.39(77) - 4M
	pyrochlore (52.2 ± 0.7)	10.10079(87) - P				1030.54(27) - P
x = 0.50	zirconolite-4M (44.7 ± 0.6)	12.4531(22) - 4M	7.19229(90) - 4M	23.0140(21) - 4M	84.861(11) - 4M	2052.99(48) - 4M
	Pyrochlore (55.3 ± 0.6)	10.10033(86) - P				1030.40(26) - P
x = 0.60	pyrochlorea	10.10258(37)				1031.09(11)
x = 0.70	pyrochlorea	10.10736(15)				1032.555(47)
x = 0.80	pyrochlorea	10.11497(15)				1034.890(47)
x = 0.90	pyrochlorea	10.12357(13)				1037.530(40)
x = 1.00	pyrochlorea	10.12655(17)				1038.448(53)

Indicates phase purity.

Figure 2

Backscattered electron micrograph of the x = 0.20 composition, with EDS spectra of zirconolite-2M and zirconolite-4M polytype phases.

Table 2

Average Composition of Zirconolite-2M, Zirconolite-4M, and Cubic Pyrochlore as Determined by Semiquantitative EDS Analysis (Normalized to Seven Oxygen Atoms)

	average composition from EDS
nominal composition	zirconolite-2M	zirconolite-4M	pyrochlore
x = 0.10	Ca0.91(4)Zr0.76(9)Dy0.17(4)Ti2.16(7)O7
x = 0.20	Ca0.86(5)Zr0.69(8)Dy0.29(3)Ti2.16(7)O7	Ca0.76(9)Zr0.64(5)Dy0.52(7)Ti2.08(10)O7
x = 0.30	Ca0.78(10)Zr0.78(9)Dy0.37(4)Ti2.07(9)O7	Ca0.71(2)Zr0.65(3)Dy0.54(3)Ti2.10(4)O7
x = 0.40		Ca0.56(4)Zr0.59(7)Dy0.84(13)Ti2.01(15)O7	Ca0.48Zr0.41(5)Dy1.06(10)Ti2.08(9)O7
x = 0.50		Ca0.55(6)Zr0.72(3)Dy0.79(10)Ti2.14(7)O7	Ca0.46(6)Zr0.40(9)Dy1.11(9)Ti2.03(9)O7
x = 0.60			Ca0.39(5)Zr0.37(3)Dy1.09(6)Ti2.14(7)O7
x = 0.70			Ca0.29(1)Zr0.25(2)Dy1.35(8)Ti2.11(7)O7
x = 0.80			Ca0.18(3)Zr0.17(2)Dy1.51(11)Ti2.14(11)O7
x = 0.90			Ca0.10(2)Zr0.09(2)Dy1.72(7)Ti2.09(8)O7
x = 1.00			Dy1.86(13)Ti2.14(13)O7

Figure 3

Backscattered electron micrograph of the x = 0.50 composition, with EDS spectra of zirconolite-4M and cubic pyrochlore phases.

Figure 4

Backscattered electron micrographs of the x = 0.60, 0.70, 0.80, and 0.90 compositions, demonstrating phase-pure pyrochlore.

Further evidence of the structure transformation from zirconolite-2M to cubic pyrochlore was inferred by Raman data, collected in the range 100–1100 cm–1 (Figure ). The position and intensity of vibrational modes for zirconolite-2M (x = 0.10) are in excellent agreement with our previous observations for nominal CaZrTi2O7 synthesized under identical conditions.[37] The low symmetry of the monoclinic 2 M unit cell resulted in many active Raman vibrational modes. The dominant symmetric stretching vibration at 780 cm–1 was attributed to TiO6 octahedra.[25] Raman active modes in the range 100–700 cm–1 have not been previously deconvoluted and assigned to individual vibrational modes; however, it is accepted that the spectrum consists of internal vibrations of the TiO6, CaO8, and ZrO7 polyhedral groups.[26] As the nominal concentration of Dy3+ was increased for 0.10 ≤ x ≤ 0.40, a significant degree of broadening occurred in the spectral range 200–600 cm–1, attributed to the disorder induced by the accommodation of Dy3+ in the Ca2+ and Zr4+ sites, and associated polytype transformation to zirconolite-4M. A notable abatement of the dominant 780 cm–1 mode was also noted, an artifact we have recently observed in the Raman spectra of the CaZr1–CeTi2O7 zirconolite solid solution, for which a similar polytype transformation from zirconolite-2M to zirconolite-4M also occurred.[38] Factor group analysis has previously determined that the general cubic pyrochlore structure A2B2O(1)6O(2) gives rise to six Raman active vibrational modes: Ag, Eg, and 4F2g.[39] A reasonable fit for the end member Dy2Ti2O7 was achieved using a combination of pseudo-Voigt profile functions (Figure S1). Six modes were deconvoluted, corresponding to approximate wavenumbers: 206, 306, 331, 517, 547, and 697 cm–1. Bands for the Dy2Ti2O7 spectrum have also been assigned according to calculated wavenumbers for A2Ti2O7 pyrochlores (A = Y, Sm, Gd, Yb) by Gupta et al., accounting for slight variations in position, determined by rA/rTi.[40] The acquired Raman spectra are in excellent agreement with Y2Ti2O7, Gd2Ti2O7, Sm2Ti2O7, and Dy2Ti2O7 specimens formed by similar processing routes.[28,41,42]

Figure 5

Raman spectra for Ca1–Zr1–Dy2Ti2O7 in compositional range 0 ≤ x ≤ 1.00.

Phase Field Confirmation by High-Resolution Transmission Electron Microscopy

The phase fields identified by powder XRD and SEM analyses were further evidenced by high-resolution transmission electron (HR-TEM) microscopy, coupled with selected area electron diffraction (SAED). As electron diffraction analysis allows variations in stacking sequence to be distinguished, polymorphic transitions in REE3+/Ac4+-doped zirconolites can be reconciled. As presented in Figure a, high-resolution TEM analysis of the x = 0.3 composition shows the layered structure of zirconolite-4M, as has been previously reported.[16] The bright-field HRTEM image presented in Figure a was captured with the beam orientated down the [110] zone axis, as is shown by the indexed electron diffraction pattern in Figure b. “Streaked” reflections can be observed in Figure b, as identified by the left-pointing green arrows, and clearly discernible ordered reflections, identified with right-pointing blue arrows. Such reflections are the direct consequence of the stacking sequence observed in Figure a, induced through the varied layer spacings indicated in this micrograph (i.e., ∼3 and ∼7 Å). This area shows a highly ordered stacking sequence in what is a grain of the zirconolite-4M structure, leading to the strong reflections observed, while the varied layer spacings produced the “streaked” reflections. The reduced contrast is likely induced by a layer rich in Dy, similar to Nd-doped zirconolite studies that have been reported, although high-angle annular dark-field imaging would be required to confirm this hypothesis. The dark-field micrograph presented in Figure a shows a two-layered band structure down the [110] zone axis, imaged with the objective aperture positioned over the diffuse reflection indicated by the right-pointing red arrow. The two-layered bands have spacings of ∼23 and ∼11 Å, representing a doubling of the unit cell along the c-axis. As described by Coelho et al., these imperfections are commonplace throughout an indexed zirconolite-4M structure and indicate the presence of both 4M and 2M spacings within a single-crystal grain. In contrast to Figure , this area contained nonuniform domains of varied spacing, suggesting variations in the level of 4M ordering within each crystal for the x = 0.30 sample. Analysis of the x = 0.60 composition through electron diffraction confirmed the formation of the pyrochlore structure, as presented in Figure a,b for the [111] and [211] zone axis patterns, respectively. No evidence of a zirconolite-2M or 4M phase was detected throughout the grains observed, confirming the phase transition to a pure pyrochlore phase at x = 0.60, in agreement with powder XRD and SEM observations.

Figure 6

(a) Bright-field TEM micrograph of the x = 0.3 sample, showing the zirconolite-4M structure, with the electron beam positioned down the [100] zone axis and (b) a [100] zone-axis electron diffraction pattern indexed to the zirconolite-4M structure. Streaked reflections are identified by left-pointing green arrows.

Figure 7

(a) Dark-field TEM micrograph of the x = 0.3 sample, showing the zirconolite-4M structure, with the electron beam positioned down the [110] zone axis and the objective aperture over the diffuse reflection indicated by the arrow in (b) a [110] zone-axis electron diffraction pattern indexed to the zirconolite-4M structure.

Figure 8

Zone-axis electron diffraction patterns of the x = 0.6 sample with the electron beam positioned down the (a) [111] and (b) [211] zone axes. Both patterns are indexed to the pyrochlore Fd3̅m structure.

XAS Investigations of Dy Oxidation State and Coordination

Dy L3-edge X-ray absorption near edge structure (XANES) spectra were collected for zirconolite-2M, zirconolite-4M, and cubic pyrochlore (corresponding to x = 0.10, 0.30, and 0.60, herein referred to as Dy-titanates for clarity) alongside Dy2O3, Dy2TiO5, and Dy2Ti2O7 reference compounds, containing Dy3+ in 6-, 7-, and 8-fold coordination, respectively (Figure ). Experimental XANES spectra at the Dy L3-edge of all reference compounds and Dy-titanates were characterized by three distinct features (labeled as A, B, and C in Figure and Table S2). Primarily, the white line crest (feature A) was composed of a single intense feature for all compounds at the overlapping edge position of 7792.5 ± 0.3 eV. The major contribution to this feature arises from dipole allowed 2p3/2 → 5d3/2 electronic transitions.[43] Theoretically, absorption spectra at the Dy L3-edge also comprise a weak pre-edge feature; however, this cannot be resolved by conventional XAS due to 2p core-hole lifetime broadening. Nevertheless, this feature can be observed with complementary techniques such as resonant inelastic scattering spectroscopy (RIXS).[44] Second, a weak yet discernible feature (labeled as B in Figure and Table S2) was also clearly distinguished (we note that this was most prominent in Dy2Ti2O7, Dy2O3, and the sample corresponding to x = 0.60). Finally, a post-edge resonance peak (feature C in Figure and Table S2) was observed for all compounds, with maxima between 7830.8 and 7831.8 eV.

Figure 9

XANES data collected for x = 0.10, x = 0.30, and x = 0.60, alongside Dy2O3, Dy2TiO5, and Dy2Ti2O7 reference compounds, with features A, B, and C labeled. Insets on the right show enlargements of features B and C.

There were several qualitative trends noted in the XANES spectra of the reference compounds and Dy-titanates. Primarily, it was clear that the edge position (7792.5 ± 0.3 eV) and energy position of feature A were very similar for all reference compounds and Dy-titanates, indicating that all samples contained Dy in the same oxidation state. As the reference compounds all contained Dy uniformly as Dy3+, it was therefore considered that Dy entered the zirconolite-pyrochlore solid solution entirely as Dy3+. Further confidence in Dy oxidation state assignment was informed from bond valence sum analyses, the results of which are summarized in Table . The speciation of Dy3+ is encouraging, as this remains comparable to Am3+; previously, the synthesis of Am2Ti2O7 by calcination in air between 1200 and 1300 °C has been shown to result in the complete reduction of Am4+ to Am3+.[45] Feature B presented a variation that may be dependent on the coordination of Dy3+ cations. This is evidenced by the reference compounds (Dy2TiO5 and Dy2O3) having different energy maxima (∼7812.4 eV) compared to the Dy-titanates x = 0.10, 0.30, 0.60, and 1.00 (Dy2Ti2O7) (∼7815.9 eV) (Figure and Table S2). This shift in maxima position could be attributed to an increase in O coordination of the Dy3+ atoms, as the feature maximum for the 6-fold Dy2O3 is lower than that of the 8-fold Dy2Ti2O7; however, a more comprehensive systematic analysis of L1- and L2-edges would be needed to confirm this trend. Moreover, the intensity of feature B present in Dy-titanates was also observed to vary as a function of Dy concentration and thus changing structure type. This can clearly be seen when comparing feature B intensity between zirconolite-2M (i.e., x = 0.10) in which Dy3+ was targeted equimolar across both the 8- and 7-fold sites, and pyrochlore (i.e., x = 0.60) in which Dy3+ cations occupy only one 8-fold coordinated site. This is qualitative evidence that suggests that there is an agreement between the targeted self-balancing charge substitution scheme for zirconolite and Dy3+ being split between two crystallographically distinct sites at a lower concentration. Qualitative trends were also observed for feature C, in which a decrease in maxima (7830.8 eV) intensity was observed in correlation with increased Dy doping of the Dy-titanates. Additionally, a shift in maxima position of ∼0.9 eV is seen between the Dy-titanate samples and the reference compounds (Dy2O3 and Dy2TiO5) possibly indicating a change in Dy coordination environment. A similar coordination-related energy shift in this feature has also been noted in the Dy and Sm L3-edges for other complex materials and has been proposed as the result of increased average Ln–O bond distance.[43,46] This qualitative trend is broadly consistent with EXAFS analyses (discussed below) whereby a slight decrease in the average Dy–O bonds was observed with increasing Dy concentration (i.e., when Dy was modeled as the absorbing atom in zirconolite-2M, 4M, and pyrochlore).

Table 3

Bond Valence Sums for Dy in Selected Compositions

composition	bond valence sum (Dy···O x8)
x = 0.10 (zirconolite-2M)	3.125
x = 0.30 (zirconolite-4M)	2.98
x = 0.60 (pyrochlore)	3.021
x = 1.00 (pyrochlore)	3.016

Fitting of the EXAFS region provided insight into the local structure of Dy3+ including the immediate coordination environment and structure over a range of up to ∼4.5 Å from the central Dy atom (Figure ; Table ). Analysis of the Dy2Ti2O7 reference compound (i.e., x = 1.00) produced a good fit (R-factor = 0.0155) that consisted of 2 O atoms at 2.19(2) Å, 6 O atoms at 2.45(2) Å, 6 Ti atoms at 3.57(1) Å, 6 Dy atoms at 3.58(1) Å, 12 O atoms at 3.92(5) Å, 6 O atoms at 4.51 Å, and an O–Ti multiple scattering pathway at 4.79 Å with a degeneracy of 24. This model is in excellent agreement with the expected Dy2Ti2O7 structure as previously determined by single-crystal X-ray diffraction.[34] Fitting of the Dy2Ti2O7 reference compound informed the fitting parameters of the x = 0.60 compound, as from XRD and TEM analyses, this was confirmed to also adopt the cubic pyrochlore structure, despite partial occupancy of Ca and Zr on the A site. An excellent fit (R-factor = 0.0076) to the data was produced with a similar model to the Dy2Ti2O7 standard, albeit with a lower coordination of Dy3+ atoms and fewer long-range order shells fitted. The best-fit model consisted of 2 O atoms at 2.21(2) Å, 6 O atoms at 2.43(2) Å, 6 Ti atoms at 3.53(1) Å, 3.6 Dy atoms at 3.59(2) Å, and 12 O atoms at 3.93(5) Å. The number of second shell Dy···Dy paths was reduced from 6 to 3.6 (x = 1.00 and 0.60, respectively), which is in line with the reduced concentration of Dy in the x = 0.60 sample relative to the x = 1.00 (Dy2Ti2O7) compound. As the coordination shell only contains 3.6 Dy, this leaves a remaining degeneracy of 2.4 that is made up of Ca and Zr; however, attempts to fit these atoms into the EXAFS fits proved unsuccessful, with the addition resulting in significantly worse fits (see Supporting Information for further details). This may be due to the limited data obtained of the samples (k-range could only be fit to 12), and therefore, a much higher k-range may be needed to deconvolute the Ti, Dy, Ca, and Zr that all manifest at a similar point in the EXAFS spectrum. Indeed, such limitations have been observed previously in different systems with different elements (U in iron (oxyhydr)oxides); however, the need for a high k-range, and the addition of molecular dynamics modeling, to deconvolute many overlapping shells in the EXAFS has been clearly shown.[47,48]

Figure 10

Dy L3 edge XAS spectra for Dy2Ti2O7 (i.e., x = 1.0), x = 0.6, x = 0.3, and x = 0.1 samples (where x refers to the structure Ca1–Zr1–Dy2Ti2O7). Left: k3-weighted EXAFS. Right: Fourier transform of k3-weighted EXAFS, using a Hanning window function. Black lines are data, and red lines are the best modeled fits for the data.

Table 4

Fitting Parameters for EXAFS Data Presented in Figure a

		path
sample	parameters	O1	O2	Ti1	Ti2	Zr1	Dy1	O3	O4	O Ti MS
	N	2	6	6			6	12	6	24
Dy2Ti2O7 (x = 1.0)	σ2(10–3) (Å2)	3(1)	13(1)	13(2)			7(1)	15(7)	5(4)	9b
E0 = 1.4(14)	R(Å)	2.19(2)	2.45(2)	3.57(1)			3.58(1)	3.92(5)	4.51(3)	4.79b
R-factor = 0.0155	α(%)	100.0	100.0	100.0			100.0	100.0	99.7
	N	2	6	6			3.6	12
x = 0.6	σ2(10–3) (Å2)	7(2)	15(2)	15(2)			7(1)	19(7)
E0 = 1.6(13)	R(Å)	2.21(2)	2.43(2)	3.53(1)			3.59(2)	3.93(5)
R-factor = 0.0076	α(%)	99.9	100.0	100.0			99.4	100.0
	N	5	3	6			0.6	8
x = 0.3	σ2(10–3) (Å2)	12(4)c	12(4)c	13(1)			4(3)	18(4)
E0 = 2.6(6)	R(Å)	2.29(5)	2.41(2)	3.49(1)			3.59(4)	4.05(4)
R-factor = 0.0028	α(%)	100.0	100.0	100.0			94.6	100.0
	N	4	4	2	4	1		12
x = 0.1	σ2(10–3) (Å2)	5(1)c	5(1)c	9(7)d	9(7)d	7(3)		19(3)
E0 = 3.2(5)	R(Å)	2.28(1)	2.43(1)	3.33(3)	3.52(2)	3.58(10)		4.13(2)
R-factor = 0.0040	σ(%)	100.0	100.0	94.0	99.4	100.0		99.9

The amplitude reduction factor (S02) for all samples was 0.95; N is the degeneracy; σ2 is the Debye–Waller factor; R is the interatomic distance; α is the result of the F-test indicating the confidence that adding the path improves the fit (>67% is equal to 1σ and >95 % is equal to 2σ in terms of standard deviation).

Indicates that the path was parameterized using O1 and Ti1 parameters.

Indicates that the parameters were linked.

Indicates that the parameters were linked. The general formula relating to “x” is Ca1–Zr1–,Dy2Ti2O7.

Fitting of the EXAFS for the x = 0.1 sample to a zirconolite-2M model produced a good fit (R-factor = 0.0040) with the model consisting of 4 O atoms at 2.28(1) Å, 4 O atoms at 2.43(1) Å, 2 Ti atoms at 3.33(3) Å, 4 Ti atoms at 3.52(2) Å, 1 Zr at 3.58(10) Å, and 12 O at 4.13 Å. This is in agreement with the expected zirconolite-2M structure. The x = 0.3 sample proved the most challenging to fit, due to the composition adopting the complex zirconolite-4M structure. The zirconolite-4M structure is composed of both pyrochlore and zirconolite-2M structural units, resulting in a system of mixed coordination environments for the Dy3+ ions. Consequently, a fit that represented a mixture of the zirconolite-2M and pyrochlore structure was attempted, using a pyrochlore CIF file as the basis. A good fit was obtained (R-factor = 0.0028) which consisted of 5 O atoms at 2.29(5) Å, 3 O atoms at 2.41(2) Å, 6 Ti atoms at 3.49(1) Å, 0.6 Dy atoms at 3.59(4) Å, and 8 O atoms at 4.05(4) Å. The reduced occupancy of the distal O shell (8 instead of 12) is likely representative of the lower degree of long-range order in the structure. The degeneracy of the Dy shell is concordant with expected occupancy of the zirconolite-2M structure. As with the previous samples and as discussed above, the Ca and Zr atoms could not be fit due to the limited k-range obtained.

While EXAFS fitting suggests that the Dy3+ cations are 8-fold coordinated by O atoms in all compositions, this coordination environment is only expected to be the sole environment for the Dy2Ti2O7 pyrochlore reference compound. Upon doping the zirconolite-2M structure with Dy3+, it is expected that the Dy atoms will occupy the Ca (8-fold) and Zr (7-fold) sites, with a possible preference for the Ca site on the basis of compatible ionic radii (Ca2+ = 1.12 Å and Dy3+ = 1.027 Å in 8-fold coordination).[49] In the x = 0.1 sample, an even split of 4 O atoms at 2.28(1) Å (Zr site) and 4 O atoms 2.43(1) Å (Ca site) may suggest that Dy is equally distributed across the Ca (8-fold) and Zr (7-fold) sites. Indeed, altering the total occupancy of the first O shell to 7.5 (expected in a system with a 50:50 split of 7- and 8-fold sites) by setting the degeneracy of the aforementioned O atoms to 3.75 each produces an equally valid fit, albeit with a slightly worsened R-factor of 0.0073, with a similar effect being observed in the x = 0.3 composition (increased R-factor of 0.0032). This suggests that Dy3+ may be doped in equal amounts across the Ca and Zr sites, in good agreement with the targeted self-charge balancing substitution regime, and moreover, the qualitative observations of features B and C in the XANES region.

Discussion within Context of Existing Literature

Several systematic CaZrTi2O7–REE3+Ti2O7 solid solutions have been reported in the wider literature, the published phase fields from which are summarized in Table and Figure , with descriptions below. A comparison of the ionic radii of lanthanides and minor actinides is provided in Table S1. The radius ratio necessary to stabilize the cubic pyrochlore structure in A2B2O7 ceramics is clearly demonstrated by varying the size of the A3+ site, in agreement with the model proposed by McCauley.[50] The bracketed areas displayed along the x-axis in Figure represent the dominant phase present at a given substitution value.

Table 5

Summary of Reported Phase Fields for CaZrTi2O7–REE3+Ti2O7 in Order of Increasing Ionic Radius (8-Fold Coordination), Where REE3+ = Y, Dy, Gd, Sm, Nd, and Cea

		phase assemblage
solid solution	ionic radius (Å)	x = 0.10	x = 0.20	x = 0.30	x = 0.40	x = 0.50	x = 0.60	x = 0.70	x = 0.80	x = 0.90	x = 1.00
Ca1–xZr1–xY2xTi2O7	1.019	2Mb	2M + 4M	2M + 4M	4Mb	4M + Py	4M + Py	Pyb	Pyb	Pyb	Pyb
Ca1–xZr1–xDy2xTi2O7	1.027	2Mb	2M + 4M	2M + 4M	4M + Py	4M + Py	Pyb	Pyb	Pyb	Pyb	Pyb
Ca1–xZr1–xGd2xTi2O7	1.053	2Mb	2M + 4M + Pe		2M + 4M + Pe + Py	4M + Py	Pyb		Pyb		Pyb
Ca1–xZr1–xSm2xTi2O7	1.079	2M + Pe	2M + 4M + Pe	4M + Pe	4M + Py	4M + Py	Pyb	Pyb	Pyb	Pyb	Pyb
Ca1–xZr1–xNd2xTi2O7	1.109	2M + Pe	2M + 4M + Pe	4M + Pe	4M + Pe + Py	Py + Pe	Py + Pe	Py + Pe	Py + Pe + NT	NTb	NTb
Ca1–xZr1–xCe2xTi2O7+δ	1.143	2M + 4M + Pe	4Mb	4Mb	Py + Pe

Key: 2M = zirconolite-2M; 4M = zirconolite-4M; Py = cubic pyrochlore; Pe = perovskite; NT = Nd2Ti2O7 (monoclinic double-layered perovskite, space group P21).

Indicates phase purity.

Figure 11

Visual summary of phase fields reported in CaZrTi2O7–REE3+Ti2O7 solid solutions with nominal formulation Ca1–Zr1–REE2Ti2O7 (Key: 2M = zirconolite-2M; 4M = zirconolite-4M; Py = cubic pyrochlore; NT = Nd2Ti2O7 (monoclinic double-layered perovskite, space group P21)).

Gd: Zhang et al. synthesized the Ca1–Zr1–Gd2Ti2O7 solid solution (0.0 ≤ x ≤ 1.0) by oxide synthesis, with bulk pellets sintered at 1400 °C for 48 h.[25] The zirconolite-2M phase was present over the compositional range 0.0 ≤ x ≤ 0.4, while the 4M polytype was stabilized between 0.2 ≤ x ≤ 0.5. The cubic pyrochlore phase was formed in compositions for which x ≥ 0.4, and present as a single phase above x = 0.6. These phase fields are near-identical to those reported in the present work, which is unsurprising given the relative ionic radii of Gd3+ and Dy3+ cations (1.053 and 1.027 Å, respectively).

Sm: Similar phase fields were reported by Jafar et al. in the Ca1–Zr1–Sm2Ti2O7 (0.0 ≤ x ≤ 1.0) system, fabricated by solid-state reaction at 1300 °C (24 h).[28] Zirconolite-2M was found to coexist alongside a minor perovskite phase when targeting x = 0.10, with mixtures of zirconolite-2M, 4M, and/or perovskite formed in the compositional range 0.20 ≤ x ≤ 0.35. Cubic pyrochlore and zirconolite-4M were reported for x = 0.40 and 0.50, with single-phase pyrochlore observed beyond x ≥ 0.60.

Y: Jafar et al. also synthesized corresponding Ca1–Zr1–Y2Ti2O7 compositions with a sintering temperature of 1300 °C.[26] No perovskite phases were identified at any interval, with single phase 2M and 4M formed at x = 0.10 and 0.40, respectively, with the intermediate compositions composed of a mixture of polytypes. At x = 0.60, zirconolite-4M was found to coexist with pyrochlore; however, cubic pyrochlore was stabilized as a single phase beyond x ≥ 0.70.

Nd: The phase fields observed when doping Nd3+ into the structure were observed to deviate from the conventional 2M → 4M → Py phase evolution, as zirconolite-2M, 4M, or cubic pyrochlore was not formed as a single phase at any compositional interval.[51] Rather, co-mixtures of 2M, 4M, and/or perovskite were present between 0.10 ≤ x ≤ 0.40, progressing to mixtures of cubic pyrochlore and perovskite between 0.50 ≤ x ≤ 0.70. As the end-member Nd2Ti2O7 structure is not capable of adopting the cubic pyrochlore structure on the basis of ionic radii (Nd3+ = 1.109 Å), a monoclinic double-layered perovskite structure was preferentially formed beyond x = 0.90.

Ce: Meng et al. synthesized the corresponding Ce solid solution; however, as Ce is known to adopt both Ce3+ and Ce4+ oxidation states in zirconolite, the targeted composition allowed for nonideal oxygen stoichiometry, and thus denoted as Ca1–Zr1–Ce2Ti2O7+δ.[4] Moreover, Ce 3d XPS data confirmed that Ce was consistently distributed as ∼50% Ce3+, inferring that the excess positive charge could be compensated by cation vacancies. Unfortunately, the solid solution was not progressed beyond x = 0.40, but, given the relatively large radius of Ce3+ in comparison to other REE3+ cations, the formation of single-phase zirconolite-4M was detected at a lower concentration, corresponding to both x = 0.20 and 0.30. When targeting x = 0.40, a co-mixture of pyrochlore and perovskite was formed.

Conclusions

Zirconolite-structured materials are a candidate wasteform for the immobilization of Pu and other highly radioactive minor actinide species that may be derived from future advanced reprocessing cycles for spent nuclear fuel. To this end, the novel Ca1–Zr1–Dy2Ti2O7 solid solution was fabricated by a conventional solid-state route, with Dy3+ deployed as an inactive surrogate cation to replicate the partitioning behavior of minor actinides such as Am3+. XRD, SEM, TEM-ED, and XAS techniques were used to characterize a series of distinct phase transformations, with Dy3+ cations fully immobilized in the zirconolite-2M phase at a concentration corresponding to x = 0.10, followed by progressive mixtures of zirconolite-2M, 4M, and/or pyrochlore in the compositional interval 0.20 ≤ x ≤ 0.50. Increasing the nominal Dy3+ concentration beyond x = 0.60 resulted in the formation of single-phase pyrochlore, successfully forming the end-member Dy2Ti2O7. Analyses of the Dy L3 XANES and EXAFS regions confirm uniform Dy3+ speciation, consistent with previously observed Am3+ under similar processing conditions, and determine that the coordination environment of Dy cations was consistent with occupation in zirconolite-2M, zirconolite-4M, and pyrochlore-structures when targeting x = 0.10, 0.30, and 0.60, respectively. Given the exceptional radiation stability and chemical durability of zirconolite and pyrochlore solid solutions, it is expected that trivalent minor actinide species could be successfully accommodated in solid solution at any compositional interval in the CaZrTi2O7–Dy2Ti2O7 system.