Synthesis, Characterization, and Stability of Two Americium Vanadates, AmVO3 and AmVO4

In search for chemically stable americium compounds with high power densities for radioisotope sources for space applications, AmVO3 and AmVO4 were prepared by a solid-state reaction. We present here their crystal structure at room temperature solved by powder X-ray diffraction combined with Rietveld refinement. Their thermal and self-irradiation stabilities have been studied. The oxidation states of americium were confirmed by the Am M5 edge high-resolution X-ray absorption near-edge structure (HR-XANES) technique. Such ceramics are investigated as potential power sources for space applications like radioisotope thermoelectric generators, and they have to endure extreme conditions including vacuum, high or low temperatures, and internal irradiation. Thus, their stability under self-irradiation and heat treatment in inert and oxidizing atmospheres was tested and discussed relative to other compounds with a high content of americium.


INTRODUCTION
The americium isotope 241 Am is formed during storage of plutonium via β − -decay of 241 Pu with a half-life of 14.33 years. Due to 241 Am accumulation in existing stocks of civil-separated plutonium in Europe, and its relatively high specific power of 0.114 W/g, 241 Am has been proposed for use in radioisotope power systems (RPSs) 1 and is under consideration by the European Space Agency (ESA) as an energy source for future European space missions. 2 The requirements for a stable solid form are very diverse, ranging from storage on earth and operation in space to safe performance in case of accidents and post-accident scenarios. Our group studied several ceramic forms containing significant specific Am amounts, 3−5 the present study being in line with the past efforts.
The synthesis and crystal structure of AmVO 3 were reported for the first time by Keller and co-workers. 6,7 In their work, AmVO 4 was prepared by dissolving AmO 2 and V 2 O 3 as hydroxides, mixed, dried, and reacted at 1250°C under an oxidizing atmosphere. The produced AmVO 4 was subsequently reduced to obtain AmVO 3 . The authors proposed the crystal structure for the two compounds (perovskite-type for AmVO 3 and zircon-like for AmVO 4 ) and gave the lattice parameters.
According to the Inorganic Crystal Structure Database (ICSD), there are no compounds of americium and vanadium with a fully characterized crystal structure. We present here the complete crystal structure refinement of the two vanadates by Rietveld analysis at room temperature. Spectroscopic and microscopic studies were carried out to check the purity of the samples. The high-resolution X-ray absorption near-edge structure (HR-XANES) technique was applied at the Am M 5 edge to characterize the Am oxidation states. The stability of AmVO 3 and AmVO 4 under self-irradiation and heat treatment in an inert and oxidizing atmosphere was tested and discussed relative to other americium-containing compounds.

EXPERIMENTAL SECTION
Caution: 241 Am is a highly radioactive isotope (t 1/2 = 432.8 years, specific activity of 126.8 GBq/g). All work presented in this paper were carried out in glove boxes in radiological laboratories licensed for handling actinides. When appropriate, shielding and remote handling tools were used to protect the workers in these experiments.  4 were assessed under an argon/hydrogen and oxygen atmosphere, respectively (to preserve the oxidation state of the respective compounds) using a Netzsch STA 449C thermogravimetric analysis instrument. The temperature was controlled by a Pt-PtRh (10%) thermocouple. The measurements were conducted on pellet fragments (11−78 mg) up to 1500°C in alumina crucibles, and the applied heating and cooling rates were 10°C/min.

XRD.
Room temperature XRD analyses were performed on about 10 mg of powdered material loaded in a bicomponent epoxy resin using a Bruker D8 Advance diffractometer (Cu Kα radiation, 40 kV, 40 mA) mounted in a Bragg−Brentano configuration. The diffractometer was equipped with a curved Ge(1,1,1) Kα 1 monochromator, a ceramic copper tube, and a LinxEye position sensitive detector. The XRD patterns were recorded using a step size of 0.01°across the 10°≤ 2θ ≤ 120°angular range. Structural analysis was performed by the Rietveld method using the Jana2006 software. 8 Since all the crystallographic parameters could not be independently refined, some structural constrains were applied. For the atomic displacement parameter, isotropic displacement parameters U iso were used, and the following equation has been considered: 9 The V−O distances were constrained to 2 Å for the perovskite structure AmVO 3 10 and to 1.71 Å for the zircon structure AmVO 4 . 11 2.4. Raman Spectroscopy. Raman spectroscopy measurements were performed on fragment of pellets (2−3 mg) at room temperature and ambient pressure on a polycrystalline specimen using a Horiba Jobin-Yvon T64000 spectrometer. For technical reasons, two lasers at different wavelengths were used; a 647 nm Kr+ laser and a 660 nm solid state laser excitation source. A 50× long focal objective was used to irradiate the sample and collect the backscattered light. Great care was taken to avoid sample damage or laserinduced heating. Measurements were performed with few tenths of a milliwatt incident power.
2.5. Scanning Electron Microscopy. Images of the samples (fragment of pellets of 1−2 mg) were obtained in a Philips/FEI XL40 SEM operated at 25 kV, equipped with a SAMx energy-dispersive Xray analysis system (EDX). This microscope (high-voltage unit, column, chamber, and turbomolecular pump) was placed inside a glovebox, while the components that are not getting in contact with the active materials (primary vacuum system, the water-cooling circuit, and the acquisition electronic) were outside. 12 2.6. XANES and HR-XANES. For AmVO 3 and AmVO 4 compounds, the Am M 5 HR-XANES spectroscopy technique was performed at the ACT station of the beamline for catalysis and actinide research (hereafter CAT-ACT beamline) of the KIT Light Source, Karlsruhe, Germany. 13 Spectra acquisitions were done utilizing a Johann type X-ray emission spectrometer. The incident beam was monochromatized by a Si(111) double-crystal monochromator (DCM), focused to 500 × 500 μm, and subsequently narrowed down by slits onto the sample to a spot size of about 200 × 200 μm. The X-ray emission spectrometer consists of four Si(220) crystals with a 1 m bending radius and a single diode VITUS silicon drift detector (Ketek, Germany), which together with the sample were arranged in a Rowland circle geometry. 13 AmO 2 reference sample was used to calibrate Am M 5 HR-XANES spectra. The main absorption maximum was set to 3890 eV for AmO 2 . 14 The maxima of the WLs are located at AmCl 6 3− (3888.4 eV), AmFe 2 (3887.5 eV) reported elsewhere. 15,16 Both Am M 5 edge data cited above were measured in conventional fluorescence mode and were similar to AmVO 3 and AmVO 4 (3888.5 eV) recorded in this work. The slight divergence of the absorption maximum of AmO 2 in the current work (3890 eV) from the data reported by Epifano et al. 17 (about 3891.5 eV) is a result of different experimental energy resolutions (for details cf. Discussion in the Supporting Information and Figure S5). The multiposition sample cell (containing <1 mg powdered material embedded in bicomponent glue) was placed into a double containment, where the inner compartment was sealed by 8 μm and the outer compartment by 13 μm Kapton foil. The HR-XANES spectra were measured with a step size of 0.1 eV from −10 to +25 eV from the white line (WL) of the respective edge and 0.5 eV in all other parts of the spectra. At least two spectra were averaged for each sample. The sample, crystals, and detector were enclosed in a box filled with helium to minimize intensity losses due to scattering and absorption of photons in air. Constant helium flow was maintained to keep the oxygen level below 0.1%.
Additionally, Am L 3 edge and V K-edge XANES measurements were performed at the INE-Beamline 18 of the KIT Light Source, Karlsruhe, Germany. The radiation protection measures were kept identical to those used at the ACT station. 19 Two Ge(422) and two Si(111) crystals were mounted in the double-crystal monochromator (DCM) for Am L 3 and V K edge measurements, respectively. The beam was focused on a ∼0.5 mm × 0.5 mm spot on the sample. Zr or V metal foils were used for energy calibration for the Am L 3 edge or V K edge XANES, respectively. V 2 O 3 (Sigma-Aldrich, 99.99%) and V 2 O 5 (Merck, 99.95%) powders were mixed with cellulose and pressed into the pellets and used as references for V K edge XANES. The XANES spectra were measured in fluorescence mode with step sizes of 0.25 and 0.8 eV from −10 to +25 eV from the white line (WL) of the respective edges for the V K edge and Am L 3 edge and 4 eV steps in the post-edge area of the spectra. At least two spectra were averaged for each sample. Measurements were performed in air, and no radiation damage was observed during the measurements.
2.7. α Self-Irradiation. Due to the high alpha activity of americium, the doses accumulated by the AmVO 3 and AmVO 4 at the time of different measurements must be considered. Daily doses of about 7.8 × 10 15 and 7.4 × 10 15 α/g were built up for AmVO 3 and AmVO 4 , respectively. Only TGA and XRD techniques could be applied on freshly prepared material. The calculated doses of the materials at the time of the different measurements are summarized in Table 1. Between measurements, the samples were stored under the atmosphere of the glovebox where the samples were produced (nitrogen with up to 10,000 ppm oxygen).

Materials Characterization: Fresh Material.
During the synthesis of AmVO 4 by oxidation of an AmVO 3 specimen, the DTA measurement ( Figure S1) performed in oxygen indicated a single exothermic event reaching a maximum energy release at 450°C with an associated weight gain of 4.5  4 showing a zircon-like I4 1 /amd structure ( Figure 1 and Table 2). Residual NpO 2 can be detected in quantifiable amounts, which indicates that this daughter element of 241 Am does not fully integrate the two vanadate compounds and segregates (at least partially) into an oxide phase. The Np L 3 edge XANES data are shown in Figure  S4, supporting the above statement.
The crystal structures suggest that americium stays in the oxidation state III in the materials whatever the redox condition, while vanadium can have the oxidation state III or V, which results in ABO 3 perovskite or ABO 4 zircon structures, respectively. One can notice that during the oxidation process, the density of the material strongly decreases from 9.565 g/ cm 3 for the close-packed AmVO 3 to 6.914 g/cm 3 for the oxyanionic AmVO 4 , which indicates a volume increase of 45% during oxidation. The formation of a zircon-like structure for AmVO 4 is in accordance with the works of Keller et al. 6,7 or of Goubard et al. 20 showing slightly lower but similar lattice parameters ( Table 2).
TGA analyses, under air and argon/hydrogen for AmVO 4 and AmVO 3 , respectively, revealed no significant weight loss for both materials, indicating a good thermal stability of the compounds at high temperature. However, XRD analysis of AmVO 3 after TGA measurement showed the presence of a new unidentified phase, AmVO 3 remaining as the main compound, suggesting the beginning of a degradation of the material at this temperature. In contrast, XRD measurement of AmVO 4 after TGA showed that the material remains fully unchanged, indicating an excellent thermal stability of this material under oxidative conditions. The evolution of the unit-cell parameters/volumes of the MVO 3 and MVO 4 compounds ( Figure 2) confirms that the Am-vanadates belong to the corresponding crystallographic families. Only the perovskite phases crystalizing in the orthorhombic Pbnm space group are considered in the MVO 3 part of Figure 2. The atomic coordinates and displacement parameters for AmVO 3 and AmVO 4 compounds are presented in Table 3.
3.2. Materials Characterization: α Self-Irradiated Material. Due to the impact of the COVID-19 pandemic,

Inorganic Chemistry
pubs.acs.org/IC Article SEM, XANES, and Raman characterization could only be performed about 1−2 years after synthesis. Due to the high alpha activity of 241 Am, the effect of self-irradiation must be considered in this condition, the main effect being amorphization of the crystal structure. The amorphization of the two vanadates was followed through XRD analyses, confirming that SEM, XANES, and Raman were performed on amorphous material. However, the results further described below seem to show only a limited impact of this amorphization on the microstructure (SEM) and electronic structure (XANES) of the two vanadates, while Raman spectroscopy showed limited impact of self-irradiation for AmVO 4 and a chemical transformation due to Raman laser heating for AmVO 3 .
The XRD results reveal that the volume of the close-packed AmVO 3 perovskite structure increases by about 4.2% after a dose of 1.03 × 10 18 α/g, while that of the oxyanionic AmVO 4 zircon structure shows a contraction of about 1.8% after a similar dose (Table 4). Therefore, the increased lattice disorder under α self-irradiation in this low-density structure results in a more compact arrangement. One can clearly see the swelling in AmVO 3 and the contraction in AmVO 4 in Figure 3, with a shift of the diffraction peaks into lower and higher 2Θ values, respectively. After a dose of about 1.65 × 10 18 α/g (7 months), the main diffraction peak of AmVO 3 is still visible, which indicates that the amorphization process is well advanced but not completed, while at that dose, the amorphization of AmVO 4 seems fully completed.
The SE micrographs presented in Figure 4 were recorded about 2 years after the synthesis. Even if the specimens had a relatively high level of self-irradiation at the time of the measurements, it can be observed that the microstructures are typical for solid state reactions, confirming the self-homogenization of the reaction mixture during heat treatment.
To assess the oxidation state of both americium and vanadium in the AmVO 3 and AmVO 4 , XANES spectroscopy characterization was performed. Am M 5 edge HR-XANES, Am L 3 edge XANES, and V K edge XANES spectra were collected. The HR-XANES data at the Am M 5 edge reveals that Am atoms have an Am III oxidation state for both AmVO 3 and AmVO 4 compounds. The energy position of the main absorption resonance (white line, WL) maximum in the AmVO 3 and AmVO 4 spectra ( Figure 5, blue and red solid lines) is located at about 3888.5 eV that is characteristic of the main absorption intensity of Am III , consistent with previously reported data. 14,15 Note that a very limited number of Am M 5 data are published to date. To be able to compare to published spectroscopic data on compounds containing Am III , Am L 3 edge XANES experiments were also performed prior to collecting the V K-edge data. The AmVO 3 , AmVO 4 , and AmO 2 XANES spectra were obtained following background subtraction by fitting a linear polynomial to the pre-edge region of the absorption spectrum and normalized at the maximum (WL) intensity for Am M 5 ( Figure 5) and at the high energy range for Am L 3 and V K edge data (Figures 6 and  7). The maxima of the WLs located at Am III VO 3 (18,521 eV eV), Am III VO 4 (18,521 eV), and Am IV O 2 (18,525 eV) are  22 The difference in the tail feature around 18,533 eV as well as smearing of the shape-resonance located at about 18,556 eV for AmVO 3 as compared to AmVO 4 suggests a different level of amorphization. Therefore, even though at this level of selfirradiation, both AmVO 3 and AmVO 4 are lacking a long-range order as observed by XRD, the degree of amorphization is somewhat different as revealed by Raman data (see below) and suggests that amorphization affects less the local structure in AmVO 4 as compared to AmVO 3 .
To directly compare the electronic structure of vanadium, the background-subtracted and normalized V K-edge XANES spectra for AmVO 3 and AmVO 4 (blue and red solid lines, respectively) as well as reference V 2 O 3 and V 2 O 5 compounds (orange dotted and black dashed lines, respectively) are shown in Figure 7. As expected, for AmVO 3 , vanadium has a V III oxidation state since all spectral features (A−D) match the spectral features of the V 2 O 3 reference spectrum. The reduced intensity of the pre-edge feature A in AmVO 3 compared to V 2 O 3 may indicate changes in the electronic structure of vanadium as a result of AmVO 3 amorphization due to selfirradiation effects as well as the influence of Am-presence in the resulting AmVO 3 after reaction of AmO 2 and V 2 O 5 at 1250°C , which differs from the pure V 2 O 5 structure. The V K-edge XANES spectrum of AmVO 4 exhibits a strong pre-edge feature (A) characteristic to V V that is well known to be due to a formally forbidden (very weak) 1s → 3d electronic transition, which is dipole-allowed when the local O h symmetry is distorted. 23−25 Features B and B′ seem to be smeared out for AmVO 4 , indicating a change of the VO 5 square-pyramidal polyhedral configuration in V 2 O 5 to tetrahedral in AmVO 4 . The findings are consistent with previously reported data by   Benzi et al. 26 for V 2 O 5 and palenzonaite (natural V V ) compounds. In that study, feature C′ in the tetrahedral geometry is located at about 5510 eV, matching the position of the shape-resonance C′ observed for AmVO 4 .
In conclusion, all the oxidation states probed through XANES and HR-XANES measurement are in perfect agreement with the expectations based on the initial crystallographic structure of the two materials. The effect of self-irradiation and amorphization has therefore no impact on their oxidation states.
The Raman spectrum of the aged AmVO 4 (Figure 8) suggests that the VO 4 tetrahedra in the zircon-like structure are intact despite the amorphization of AmVO 4 at this level of selfirradiation found by XRD. The five clearly identifiable internal modes of the [V−O 4 ] tetrahedron, ν 1 (A 1g ), ν 3 (B 1g ,E g ), ν 4 (B 1g ), ν 2 (A 1g ), and ν 2 (B 2g ), were found at very similar wavenumbers as for the lanthanide LnVO 4 zircon compounds, 27 as shown in Table 5, though somewhat lower than the orthovanadate of neodymium, whose ionic radius is closest to trivalent Am III . The two clearly identifiable low-frequency external modes, which reflect the motion between the [V−O 4 ] tetrahedron and the Am III ion, were observed, of which the lowest is the T(B 1g ) and the highest is assigned to the R(E g ). The spectrum also shows a very weak mode at the position of one of the T(E g ) modes. The T(B 1g ) mode is substantially lower than that found for the lanthanide LnVO 4 compounds. 27 This can be explained by the observation by Moura et al. 28 that the T(B 1g ) mode in the Tb(V 1−x P x )O 4 solid solution shows a different broadening, indicating that it is related primarily to the Tb motion on the [Tb−O 8 ] sublattice. The strong shift observed here thus supports that it is due to the mass effect between the actinide and lanthanide series.
The shifts of the modes in AmVO 4 compared to the LnVO 4 series are very similar to AmPO 4 5 and LnPO 4 . 29 However, the strong broadening of the modes after aging that was observed for AmPO 4 did not occur. This may be explained by   analogues. 29 Surprisingly, the Raman spectrum of the aged AmVO 3 sample was identical to that of AmVO 4 , suggesting that oxidation took place. We exclude that the oxidation is related to the damage build-up since it is not observed in the XRD and XANES results at similar dose. We therefore conclude that it is triggered by heating the laser during the Raman measurement.
3.3. Closer Look at the α Self-Irradiation Mechanism in AmVO 4 . Our results for AmVO 4 amorphization derived from XRD analysis differ significantly from the observations in the study of Goubard et al., 20 who studied the radiation damage build-up in AmVO 4 during a much longer period of 5 years (14 × 10 18 α/g). They reported that the structure remained at least partially crystalline during this period and became predominantly amorphous after 5 years, which is about 10 times slower than in our work. Moreover, Goubard et al. 20 did not observe a volume contraction under α self-irradiation like in the present study, but a moderate lattice expansion with a volume increase to a maximum of about ΔV/V 0 = 1.2%. At a comparable dose to our study (1 × 10 18 α/g), they found ΔV/ V 0 = 0.15%.
The reasons for the discrepancies with the work of Goubard et al. 20 remain unclear, and one can only hypothesize the following: • A different content of neptunium (daughter element of americium) in the initial material. Note that Goubard et al. 20 did not give details about the purity of the used Am. The exact quantity of Np that could be incorporated in AmVO 4 is not known, but we observed NpO 2 as the separate phase in this work. Moreover, uptake of Np IV in the structure would require charge compensation, for example, by V III , which was not observed by XANES in our work. • The higher synthesis temperature in the current work (1000°C vs 600°C), which could have resulted in a more crystalline material and thus less initial disorder. As a result, a potential time span of contraction could have been absent or short in the study by Goubard et al. 20 It should be noted that their first data point at 0.7 × 10 18 α/g is anomalous in their ΔV/V 0 vs dose curve and indicates a minimal expansion (ΔV/V 0 = 0.03%) after approximately 90 days. • The difference in production scale (hundreds of milligrams here versus 1 mg by Goubard et al. 20 ) and more probable methods (synthesis temperature and duration) could have resulted in substantially different crystallite sizes, powder morphologies, and densities, thus potentially affecting the impact of the alpha decay through annihilation of defects, in particular via diffusion to sinks (grain boundaries). 33 • Radiogenic helium can interact with defects (vacancies), but the effect of this will depend on the microstructure and crystallinity. The larger the grain size, the larger the helium fraction retained in the lattice, affecting the defect recombination kinetics. Similarly, the initial disorder (crystallinity) will affect the helium retention. Unfortunately, information on the microstructure of the material synthesized by Goubard et al. 20 is missing.   The first three hypotheses could explain the slight difference of the initial lattice parameters (see Table 2), but not the different kinetics. Although the damage kinetics could be affected by size and geometry of samples analyzed in the XRD instruments, the discrepancy is huge. The last two hypotheses could explain the different recombination kinetics, but remain speculative, in the absence of comparative microstructural data.
Our observations for zircon-type AmVO 4 are different from our results for the monazite-type AmPO 4 . 5 The zircon structure is closely related to the monazite LnMO 4 structure, both made of isolated tetrahedra of V or P, connected by trivalent metal ions in eight or nine coordinations, respectively. Under the influence of temperature and/or pressure, a transition can take place. 30,31 For AmPO 4 , a significant expansion of the unit cell was observed (ΔV/V 0 = 1.8% at 0.6 × 10 18 α/g), whereas the AmVO 4 cell contracts with irradiation time. However, the dose at which the transition to a fully amorphous phase takes place is close in both studies, slightly higher for AmVO 4 compared to AmPO 4 , which is in line with the observation of Meldrum et al. 29 that the critical amorphization temperature is slightly higher for zircon-type LnPO 4 compounds as well.
These contrasting observations are not easy to combine. Volume contraction due to α self-irradiation is known for Am 2 Zr 2 O 7 pyrochlore 34 and was attributed to the stability of the [Zr−O 6 ] octahedra and their rotation in response to the increase in disorder around americium. Although Am 2 Zr 2 O 7 has a close-packed pyrochlore structure, which is very different from the open oxyanionic structures of zircon and monazite, this explanation may help understand. The volume in the LnVO 4 series is discontinuous, with the zircon (xenotime) structure (eight-coordinated M III ) being larger than the monazite (nine coordination), 35 similar to the LnPO 4 series. 36 So, in case the radiation damage in zircon-type AmVO 4 affects predominantly the americium coordination sphere leading to a higher coordination with a concomitant rotation of the [V− O 4 ] tetrahedra, the volume will decrease, as we observed here. In AmPO 4 , the Am ions are already nine-fold coordinated, resulting in prompt expansion of the lattice with increasing dose. Of course, also differences in the radiation resistance of the [V−O 4 ] and [P−O 4 ] tetrahedral entities may play a role.

CONCLUSIONS
Two americium vanadates were produced and characterized in terms of crystal structure, cation oxidation states, and chemical, thermal, and radiation stability. The XRD and XANES measurements demonstrate that americium stays in the oxidation state III in the material whatever the redox condition, while vanadium shows the oxidation state III in the perovskite-like AmVO 3 or V in the zircon-like AmVO 4 .
After an accumulation of a dose of 10 18 α/g, the closepacked AmVO 3 shows a volume expansion of 4.2% while the low density oxyanionic AmVO 4 contracts with 1.8% in volume (in contrast with other literature reports 20 ). With a fast amorphization behavior, large volume variation, and interchangeable structure as a function of the presence of the oxygen atmosphere and temperature, americium vanadates do not appear to be good candidates as Am forms for radioisotope power systems. Moreover, 237 Np (the decay product of 241 Am) tends to segregate as a fluorite fcc secondary phase, creating interfaces and inducing stress in the material. Comparing the existing data on Am-containing ceramics with fluorite, 3 pyrochlore, 34 zircon, monazite, 5 and perovskite 4 structures, the (Am,U)O 2 2,3 fluorite solid solution seems currently to be the most suitable form for space applications. ■ ASSOCIATED CONTENT
Thermogravimetric behavior of AmVO 3 and AmVO 4 ; overview of the AmVO 3 and AmVO 4 specimens measured by SEM and EDX; XRD diffraction pattern of AmVO 3 after TGA measurement; Np L 3 XANES spectra of the AmVO 3 and AmVO 4 ; discussion on the difference in the experimental energy resolution leading to an energy shift of the main maxima of the An M 4,5 edge HR-XANES spectra; discussion on the neptunium dioxide residual phase from XRD data (PDF) istration, validation, and writing (review and editing); J.-Y.C. and D.B.: Investigation; D.F.: Project administration, resources, and writing (review and editing); R.J.M.K.: Project administration, resources, supervision, and writing (review and editing); K.P.: Conceptualization, investigation, methodology, supervision, and writing (original draft; review and editing).