Effect of graphite and graphene oxide on thorium carbide microstructural and thermal properties

Thorium carbide to be tested as target material for the production of 225Ac with the ISOL method, was produced via carbothermal reduction of ThO2 nanoparticles by graphite and graphene oxide, respectively. The use of graphene oxide (GO) as carbon source resulted in a reduced reactivity compared to graphite, confirmed by the presence of unreacted ThO2 mainly in the core of the samples. The reacted ThO2 or ThC2–GO showed a faster reactivity in air, mainly observed as ThC2 amorphization. The specific surface area of the ThC2–GO samples was almost doubled compared to ThC2–graphite samples. The effect of these microstructural features was analysed in terms of thermal diffusivity and calculated thermal conductivity that were both reduced in ThC2–GO samples, however the difference with ThC2–graphite samples decreased at increasing temperature. The present study shows that the use of unreduced GO inhibits the solid-state reaction between ThO2 and C; on the other hand, the high reactivity of the ThC2 so produced is expected to be beneficial for the 225Ac production with the ISOL method, affording a high release efficiency. It is expected that the use of reduced GO could represent a good solution for highly efficient ThC2 targets.

Since its appearance in clinical research literature in the late 1990's, Targeted Alpha Therapy (TAT) has been considered an extremely promising approach to treat cancer 1 . The benefits of using alpha-emitting radioisotopes to selectively induce localized damage to cancer cells come from peculiar properties of alpha particles with respect to beta radiation. Due to their much larger mass, alpha particles result in a higher linear energy transfer, i.e., they deposit higher amounts of energy over shorter ranges.
Among the potentially employable alpha-emitting radioisotope for this purpose, 225 Ac has recently attracted much interest due to the combination of a suitable 10-day half-life, the emission of four alpha particles in its decay chain and the generation of 213 Bi, another isotope considered for TAT applications 2 . The development of 225 Ac-based radiopharmaceuticals presents several challenges, mostly linked to the availability of sources and/or production techniques of this particular isotope and to the purity of the obtained products. At present, 225 Ac for clinical research is primarily produced from the decay of 229 Th stocks located in the USA, Russia and Germany, in turn obtained from stockpiles of 233 U 3 . Other explored approaches are the use of 226 Ra targets bombarded by protons or neutrons 4 and the direct irradiation of 232 Th with high-energy protons 5 . None of these techniques is free from the necessity of developing several stages of chemical purification to eliminate co-produced and undesired isotopes of actinium and other elements.
With the aim of developing new and more efficient production methods, ISOL (Isotope Separation On-Line) facilities are being considered as sources to get high purity 225 Ac for research purposes 2 . In the framework of such existing or future facilities, traditionally devoted to the production of radioactive ion beams for nuclear physics, several research programs on medical isotopes production are now ongoing. Examples include MEDICIS at CERN 6 , ISAC at TRIUMF 7 , ISOLPHARM at INFN 8 and ISOL@MYRRHA at SKC-CEN 9 . The presence of a mass separation stage in these infrastructures ensures high purity, due to the absence of same-element contaminants, which cannot be removed with chemical separation methods. www.nature.com/scientificreports/ The research reported here is aimed at the development of a nanostructured thorium carbide-based target for 225 Ac production with the ISOL method. Historically, thorium carbide targets have played a rather marginal role in ISOL facilities 10,11 compared to the commonly used uranium carbide ones, due to lower yields of radionuclides relevant for nuclear physics and more difficult synthesis and handling. The need for optimized targets at the nanoscale level has lately arisen in the ISOL target community; stable porous nanostructures with high specific surface area (SSA) have been associated with increased performance consistency during irradiation, guaranteeing high yields to experimental users 12,13 .
Our synthesis of high-SSA thorium carbides, was inspired by recent work carried out at JRC Karlsruhe on thorium oxide nanopowders 14,15 . The idea behind this work was to let nano-ThO 2 react with two different carbon sources (graphite and graphene oxide) at high temperature in an inert atmosphere, to obtain ThC 2 -carbon nanocomposites and to verify the retention of high SSA in the final material. Composites materials demonstrated 16 better thermal properties and thus are of interest for our application. In our previous collaborative work, graphene had the effect of improving thermal properties of uranium carbide-carbon composites 17 . In the present study, graphene oxide was used, with the aim of exploiting its low-temperature reduction 18 and the related gas release during thermal treatment, thought to be beneficial for the creation of interconnected micro-and meso-porosity, and consequently for obtaining a high SSA.

Results
An overview of the properties of graphite and graphene oxide-derived thorium carbide pellets is given in Table 1. In the following, these materials will be referred to as ThCx-graphite and ThCx-GO, respectively. The results of elemental analysis (carbon and oxygen content) on both the starting materials and the final pellets are shown in Table 2. Both materials showed higher weight losses with respect to those calculated based on stoichiometry (17 wt%), as reported in "Methods" section. This is generally attributed to phenomena such as water adsorption on starting reagents, binder decomposition and oxygen contamination in the carbon sources, none of which are considered in the stoichiometry. On the other hand, if one considers the actual C/O ratio found in ThCx-graphite and ThCx-GO by elemental analysis, the weight loss calculated with the corrected stoichiometry is around 20 ± 1 wt%, this agrees with the experimental values of 19.6 ± 0.1 wt% for ThCx-graphite but is higher than 17.7 ± 1.6 wt% for ThCx-GO. The remaining graphite (1.4 wt%) or GO (3.3 wt%) is then transformed to the C being present in the sample. The lower weight loss observed for ThCx-GO indicates incomplete conversion, which could be an effect of increased surfaces or lower reactivity, as shown in the following. Despite showing different volumetric shrinkages and weight losses, the two materials ended up having similar densities and, thus, total porosities. The total porosity was obtained from a calculation which considers the designed stoichiometry of the final materials (ThC 2 + 2C), and therefore does not take into account any discrepancy in the composition of the final products. The data reported in Table 2 are close to those reported by the supplier's certificate of analysis. Graphite has a higher C content than graphene oxide, in which O is present together with H either as water or in different organic groups. In the case of ThCx-graphite and ThCx-GO, again considering the final anticipated stoichiometry ThC 2 + 2C, Th represents about 83 wt% of the final mass. The three ThCx-GO samples used for the elemental analysis showed a difference in the carbon content indicating a clear non-homogeneity in the sample, resulting in a higher error with respect to ThCx-graphite. Figure 1 shows the evolution of the CO release during the separate thermal treatments of ThCx-graphite and ThCx-GO samples. Since two batches of roughly the same mass of reagents were loaded into the furnace in each treatment (6 pellets with a total mass of about 3 g), the evolution and total quantity of emitted CO should be similar. While this is true for the low-temperature CO release associated with binder decomposition 19 , a remarkable difference can be seen at high temperature. Above 1700 K indeed a higher gas release was observed in the ThCx-graphite treatment. Part of it could be attributed to the oxidation of graphite components of the www.nature.com/scientificreports/ furnace, since the thermal treatment of the ThCx-graphite was carried out before that of the ThCx-GO and after a period of inactivity. However, a further heat treatment of the samples reproduced the characteristic gas release and suggested that the outgassing of the internal furnace wall is minor. This outgassing usually happens at temperatures well below 1700 K, so the difference in CO release can be better ascribed to incomplete carburization of ThCx-GO. Figure 2 reports X-ray diffraction (XRD) patterns of the starting reagents, whereas those of ThCx-graphite and ThCx-GO are shown in Fig. 3. The ThO 2 nanoparticles obtained by hydrothermal decomposition of the thorium oxalate have a size of about 7 (± 1) nm routinely verified by powder XRD and calculated from the line broadening of 5 isolated peaks in the diffractogram. They have been analysed by TEM, as reported in 14,15 . Figure 3 clearly shows that ThCx-graphite is solely composed of ThC 2 as the crystalline phase whereas ThCx-GO contains a mixture of two crystalline phases ThC 2 and ThO 2 . The analyses of the XRD patterns of both samples including the Le Bail fitting confirm the cubic structure for ThO 2 (a = 5.5988 (2) Å, Fm-3m_space group 225) and the monoclinic structure for ThC 2 (a = 6.694 (2) Å, b = 4.2405 (12) Å, c = 6.746 (3) Å, β = 103.84 (2)°, C2/cspace group 15) found for the ThCx-GO whereas for the ThCx-graphite only the monoclinic structure is found for ThC 2 (a = 6.6983 (6) Å, b = 4.2369 (4) Å, c = 6.7478 (6) Å, β = 103.836 (3)°, C2/c-space group 15). This could arise either from an incomplete reaction, as suggested by the aforementioned CO release during the treatment and the lower mass loss during the thermal treatment, or with a lower probability from oxidation during the handling and transport of the samples to the characterization devices.
To investigate this phenomenon of possible ThC x oxidation in more detail, XRD patterns were collected at different times with samples embedded in oil and measurements under air on both the pellet surfaces and the crushed powders, as shown in Fig. 4. Time t 0 corresponds to the starting of the XRD measurement. The results unambiguously show that:  is thus not possible to assign its appearance to the partial hydrolysis or oxidation of carbide resulting from exposure to moist air. Therefore, it must be residual (unreacted) in accordance with reduced weight loss as reported in Table 1, which indicates incomplete conversion.
In order to shed more light on the ThCx-GO composition, a carburized sample was studied in more detail (Fig. 5) by polishing it until half thickness was reached and taking an XRD spectrum of the inside surface, to be then compared to the outside surface. The results agree with those of Fig. 4: oxygen is found inside the pellet and not on the surface, providing additional evidence that the oxide-carbide conversion was incomplete. In fact, the inside surface seems to be composed only of ThO 2 , even though the presence of ThC 2 cannot be excluded completely from the diffractogram (Fig. 5), since the ThC 2 XRD peaks are less intense than the ones for ThO 2 due to the lower symmetry (monoclinic for ThC 2 vs cubic for ThO 2 ).
Raman spectra of graphite, GO, ThCx-graphite and ThCx-GO samples have been collected and are reported in the Supplementary Data S1 section. The main features of graphitic carbon-based materials, G and D bands at 1500 and 1350 cm −120 , are observed. The most intense peak for the graphite is the G band at 1500 cm −1 , while  www.nature.com/scientificreports/ for GO, D-band and G-band have equal intensity. The G peak is ascribed to the bond stretching of all pairs of sp 2 atoms in both the rings and chains, while the D band is ascribed to defects in the sample and its intensity is related to the degree of disorder. Considering these assignments and comparing the samples after carburization, ThCx-GO shows the highest ratio of intensity I D /I G thus pointing to the highest amount of defects 19 . SEM images taken on the surface of samples of the two materials are reported in Fig. 6. Both show a microstructure composed of a dispersion of residual carbon (dark grains) in a partially sintered thorium carbide matrix. In ThCx-GO, carbon is clearly visible, contrary to what was found in UC x -graphene in another study 17 , but it is equally well dispersed. It must be remarked that in this case graphene oxide was used instead of pure graphene. ThCx-graphite microstructure resembles that of previously obtained carbides using the very same graphite 17,19 , whose size is in the order of tens of μm.  www.nature.com/scientificreports/ TEM images and the corresponding ring diffraction patterns are reported in Figs. 7 and 8 for ThCx-GO and ThCx-graphite, respectively. As in SEM, not many structural differences can be observed between the two materials. No evidence of a discrepancy in carbon distribution around thorium carbide grains is found, contrary to what reported for uranium carbide 17 . The analysis of ring diffraction patterns suggests, however, that the crystalline part of the ThCx-GO sample is polyphasic, with the onset of a cubic phase (ThO 2 ) and a monoclinic one (ThC 2 ), whereas ThCx-graphite is found to be monophasic. N 2 physisorption analyses have been performed on the starting materials, ThO 2 nanoparticles, graphene oxide and graphite, and on the composite pellets after the carburization reaction. While the analysis on GO and graphite was carried out under standard conditions using a Micromeritics ASAP-2020 surface area and porosimetry system, measurements on ThO 2 powders and on samples after thermal treatments were carried out with a Micromeritics Gemini surface area analyser, stored inside a glove box under inert atmosphere and equipped with a LN Dewar with reduced volume (0.6 l). Therefore, under these conditions only a limited number of data points in the adsorption isotherm were collected and the pore size analysis through the BJH model was not feasible with reasonable accuracy.
Isotherms collected for the original GO and graphite are reported in Fig. 9. GO displays a remarkably high SSA (around 250 m 2 /g) with a presence of micropores, evidenced by a steep increase in N 2 adsorption at low relative pressures (< 10 −2 ), and narrowly distributed slit-like mesopores, as indicated by the hysteresis type H3, according to IUPAC 21 . As for the pore size distribution, in the micropores range the distribution obtained using  www.nature.com/scientificreports/ Horwath-Kawazoe model shows a maximum at 0.6 nm, whereas in the mesopores range a clear peak at around 3.8 nm is observed by applying the BJH model to desorption (pore size distribution graphs not reported). Table 3 reports specific surface areas of the starting materials and the final carbides, evaluated using BET and single point adsorption methods.
Thermal diffusivity α data (in m 2 s −1 ) measured up to 1550 K by the laser flash technique developed at JRC Karlsruhe are reported in Fig. 10. They show a higher diffusivity in the ThCx-graphite sample than in that of ThCx-GO. The difference between both samples decreases with temperature, from 28% at 500 K to 8% at 1550 K. Figure 11 shows calculations of the thermal conductivity λ obtained from the thermal diffusivity α and the density ρ (in kg m −3 ) of samples obtained after heat treatment and corrected for thermal dilatation using data from 22 , with specific heat C p of ThC 2 and ThO 2 extracted from 23,24 respectively. Based on XRD, ThCx-graphite is composed of only ThC 2 within the uncertainty of the measurement, while ThCx-GO contains ThC 2 and ThO 2 . Since the exact composition and the ThC 2 /ThO 2 ratio of ThCx-GO is not retrievable by Rietveld analysis due to its low XRD signals and its tendency to rapidly amorphize, only rough estimates can be made. We considered three different compositions for ThCx-GO, each associated with a percentage of ThC 2 in the sample (100% ThC 2 , 50% ThC 2 -50% ThO 2 and 10% ThC 2 -90% ThO 2 ). At low temperature, the specific heat of ThO 2 23 is slightly higher than that of ThC 2 24 , whereas at high temperature the specific heat of ThC 2 is much higher than that of ThO 2 , resulting in diverging data for the three ThCx-GO compositions. In all cases however, the ThCx-graphite conductivity is higher than that of ThCx-GO, confirming the diffusivity trends. The difference in thermal conductivity between ThCx-graphite and ThCx-GO tends to decrease with temperature. Literature about thermal conductivity data for the monoclinic α-ThC 2 is very scarce but, as pointed out by Manara et al. 23 , an estimated value is 24 W/(m K) obtained from electrical resistivity measurements for fully dense ThC 2 at 298 K, whereas experimental data for a 72% ρ th (theroretical density) sample yield 24 W/(m K) at 443 K and 20.5 W/(m K) at 627 K. On the other hand, Pillai and Malakkal 25,26 reported much lower values for the thermal conductivity of 94% ρ th and 95% ρ th ThC 2 samples in the 300-1200 K temperature range. In particular, around 600 K the ThO 2 thermal conductivity is found to be around 7 25 or 8 26 W/(m K), less than half the value reported for a 72% ρ th ThC 2 23 .

Discussion
Differences in reactivity between ThCx-graphite and ThCx-GO. In the prepared ThCx-GO samples, ThO 2 is observed, in the inner of the pellets; additionally during the carboreduction less weight loss and a lower CO release are observed. All these points together can be attributed to the inefficient reduction of GO in argon atmosphere and incomplete transformation. Indeed, the literature reports that the thermal reduction of GO becomes efficient at 2023 K under high vacuum or reducing atmosphere (H 2 or Ar-H 2 ). The oxygen atoms in GO are present as hydroxyl, epoxy and carboxyl groups, of which the most thermally stable are the latter two 27 . In this study, the carbothermal reduction process was run under Argon atmosphere and up to 1923 K, so a lower efficiency seems reasonable. The 21 wt% of graphite or GO in the starting mixtures with nano-ThO 2 are sufficient to receive complete reduction to ThC 2 plus the formation of carbon. The oxygen content in the starting GO (5.5 wt%) drives the reaction stoichiometry to ThO 2 so that during the carboreduction it seems reasonable that a lower amount of carbon is formed. Since the carbothermal reduction of ThO 2 is a solid-state diffusion process a lower amount of one component could reduce the contact and therefor affect the conversion negatively. Additionally the GO shows a tenfold higher SSA compared to graphite (Table 3); this indicates the presence of a less dense structure with lower contacts, which is reflected as well in the observed lower thermal diffusivity and conductivity of the ThCx-GO compared to ThCx-graphite. A less dense structure could be another reason why the surface controlled transformation of the nano-ThO 2 is slower in the carbothermal reduction with GO than with graphite.
These two points together make us come to the conclusion that the ThO 2 present in the ThCx-GO is unreacted nano-ThO 2 . In order to examine this further experiments are planned.
On the other hand, the reduced reactivity of the ThO 2 -GO mixture has the positive effect of preserving part of its initial large surface area, with a clear effect on the final material SSA and the total pore volume. A more voluminous structure of the final ThC 2 product seems to be advantageous with respect for the irradiation and isotope yield later and makes further experiments even more attractive.  www.nature.com/scientificreports/ Differences in sample behavior in air. Literature about the amorphization of thorium carbides is limited to one work 28 , which can be included in the more general research on hydrolysis of ThC 2 29,30 . The hydrolysis of ThC 2 in air at room temperature was found to be accompanied by the production of several different hydrocarbon gases and to result in the formation of ThO 2 and an amorphous phase, which was not fully identified but thought to be hydrated ThO 2 . The effect of exposing UC x -graphite to moist air at 30 °C for 48 h was investigated in a previous work 31 . Despite a visible (but not quantifiable) carbon depletion and oxygen enrichment on the surface, the morphology of carbide grains remained unaltered, testifying to the great stability of the carbidegraphite system to external conditions. The ThCx-GO phase evolution during XRD investigations (Fig. 4) seems to be in agreement with this phenomenon, whereas ThCx-graphite was found to be stable throughout the entire measurements campaign; ThCx-GO is more reactive towards air than ThCx-graphite. This different reactivity against air could be explained as well by a more voluminous structure of ThCx-GO compared to ThCx-graphite facilitating its reaction with air.
Based on the findings of this work we present the synthesis of ThC 2 (ThCx-graphite) as product of the carboreduction of nano-ThO 2 with graphite as suitable starting materials for the production of radioisotopes. The carboreduction of nano-ThO 2 with graphene oxide forms ThCx-GO, a product of incomplete carboreduction and a mixture of ThO 2 , carbon and ThC 2 . However, properties like the lower thermal diffusivity and conductivity or the higher SSA of the ThCx-GO compared to ThCx-graphite support a voluminous structure being attractive for the recovery of radioisotopes after irradiation. This makes the nano-ThO 2 carboreduction with GO becoming worth for further investigations towards the development of target materials for irradiation experiments dedicated to isotope production for medicinal applications.

Methods
Samples preparation. The process going from sample preparation to pressing of the pellet took place in glove box equipped with an MBraun gas purification system. The samples were kept under nitrogen atmosphere with a content of H 2 O < 1 ppm and of O 2 < 50 ppm. ThO 2 nanoparticle sample was prepared following a method described in the literature 15  The process was repeated three times. Afterwards, working under argon, the autoclave was filled with about 5.5 g oxalate powder and 3 ml of water and the hydrothermal reaction was performed for 18 h at 523 K. The resulting ThO 2 nanocrystal's were washed with water, ethanol and acetone, in order to gradually decrease the polarity of the solution. Each washing step was followed by centrifugation and the supernatant was removed, the residue was dried by evaporating the acetone from the last washing. Both graphite and graphene oxide were used as carbon sources for the synthesis of ThC 2 . Graphite powders of size < 45 μm and graphene oxide powder (15-20 layers, 4-10% edge oxidized) were purchased from Sigma-Aldrich. The reactants were first individually mixed in a planetary ball-mill (2 h at 3000 rpm), with jar and balls made of ZrO 2 , leading to a powder size < 5-10 μm. The powders were then mixed in an agate mortar using the mass percentages reported in Table 1, corresponding to the planned reaction.
A solution of 2 wt% of phenolic resin in acetone (15 wt%) was added dropwise as binder. The powders were then further mixed in the ball mill (3 h at 3000 rpm). The pressing conditions and the heat treatment conditions were similar to those reported in 17 . At least five pellets of 13 mm diameter and about 1 mm thickness were prepared for each batch (ThO 2 -Graphite and ThO 2 -GO). The heat treatment of the samples was done by batch with a Degussa Type VSL annealing furnace under Argon flow (5 N) at a flow rate estimated 3-5 l/min. Before flushing with Ar passing through a moisture trap, we first proceed with a primary vacuum of 10 -3 mbar of the furnace chamber. During the heat treatment, a slight Ar overpressure in the furnace is present. The furnace is located in a glove box under N 2 whose oxygen trace is below 0.5%. Leakage into the furnace before Ar flushing cannot be excluded. However, it is limited to the trace of oxygen present in the glove box whose content is below 0.5%. The samples were maintained on a graphite crucible standing on alumina disk deposited on molybdenum grids at 1963 K for 24 h, heating rate 2 K/min. To follow the process of the carbothermal reaction, monitoring of the CO evolution during the heat treatment, we used a Residual Gas Analyser (Siemens Ultramat 6). With an analytical balance Mettler Toledo Model SAG204, we determined the weight loss after thermal treatment of the samples.
Samples chemical and physical characterization. ELTRA CS-800 instrument was used to analyse the carbon and oxygen contents on powdered samples. The method goes through direct combustion of samples using the infrared absorption detection technique 17 . The bulk density ρ bulk was measured via the weight over volume ratio; the total porosity of the samples was calculated by: P tot = 1 − ρ bulk /ρ th , where ρ th is the theoretical density of the final sample, calculated by the mixture rule taking into account the volumetric fractions of ThC 2 and free carbon present in the final samples (ThC 2 + 2C) and their theoretical solid densities, 9.14 g/cm 3 for ThC 2 23 , between 1.9 and 2.3 g/cm 3 for carbon 19 , averaging the obtained results. The shrinkage percentage was determined considering only the average thickness of the pellet measured with an absolute Digimatic Indicator ID-S (Mitutoyo), before and after heat treatment. Scanning electron microscopy was performed on a Philips XL 40 using a tungsten filament (200 V-30 keV).The Transmission Electron Microscope (TEM) study was performed using an FEI Tecnai G2 model, equipped with a GATAN Tridiem camera and a GATAN Imaging Filter. The field emission gun was operated at 200 kV during the study. The TEM was adapted for the examination of highly active or irradiated nuclear materials thanks to a flange that was inserted in the octagon and which hosts the objective lenses and a glove box mounted on this flange around the CompuStage rotation holder 32 .
(1) ThO 2 + 6C → ThC 2 + 2C + 2CO. www.nature.com/scientificreports/ The sample was prepared by crushing, as explained in previous work 33 ; the so-prepared sample grids were then introduced into a Plasma cleaner machine in order to eliminate any organic residues, and brought to the TEM for the analysis making use of a La Calhène DPTE system. Samples embedded in oil were characterized by powder X-ray diffraction on a Rigaku MiniFlex 600 diffractometer in θ-2θ configuration using Cu K α1-α2 radiation. The patterns were measured from 5° to 120° in 2Θ angle with a scan/step of 0.02° scan/step and a scan speed of 0.29°/min. The software Topas version 5 was used to refine the XRD data.

Functional properties characterization.
A shielded "laser-flash" device designed and constructed at JRC Karlsruhe 34,35 was used for the measurement of the thermal diffusivity α of the samples. The detailed description has been provided in previous study on UC 2 samples 17 . The precision of measurements is better than 5% and is mainly determined by the variation in sample thickness. The experiments were carried out starting at about 500 K with the aim of measuring α up to the maximum temperature of about 1550 K and then back to 500 K in order to observe the effect of possible recovery effects or sample composition changes (oxidation) that might have taken place during laboratory thermal annealing.

Data availability
The datasets generated and/or analysed during the current study are available from the corresponding author on reasonable request.