Available online at www.sciencedirect.com Procedia Chemistry (2012) 791 – 797 ATALANTE 2012 International Conference on Nuclear Chemistry for Sustainable Fuel Cycles Assessment of the complete core of the reference pyrochemical process, developed by the CEA E Mendesa*, O Conocarb, A Laplacea, N Douyèreb, M Miguirditchiana a b CEA Marcoule, Nuclear Enregy Division, Radiochemistry & Processes Department, SCPS/LEPS, F-30207 Bagnols-sur-Cèze, France CEA Marcoule, Nuclear Enregy Division, Radiochemistry & Processes Department, SEAT/GEDM, F-30207 Bagnols-sur-Cèze, France Abstract Pyrochemical separation technology using high-temperature molten salts presents a potential interest for an overall separation strategy Pyrochemical R&D program, launched at CEA, aims at demonstrating the feasibility of a grouped separation of actinides with sufficient decontamination The process is based on two liquid/liquid extraction steps: a selective extraction in liquid aluminium from molten fluoride and the actinide back-extraction in molten chloride The present work focuses on a complete study of this core of process The reductive extraction has been assessed as function of the salt composition wile the back-extraction step focused on the behaviour of the actinides 2012Elsevier The Authors Publishedand/or by Elsevier B.V under responsibility of the Chairman of the ATALANTE 2012 ©©2012 B.V Selection peer-review SelectionCommittee and/or peer-review under responsibility of the Chairman of the ATALANTE 2012 Program Program Keywords: Separation process, Pyrochemistry, liquid/liquid extraction Introduction Among the various techniques of reprocessing, pyrochemical separation technology using high-temperature molten salts and metal media presents a potential alternative to hydrometallurgical technology (PUREX process and its analogues), especially in case of the treatment of minor actinides-rich materials [1] The pyrochemical * Corresponding author Tel.: +33-466796311; fax: +33-466796567 E-mail address: eric.mendes@cea.fr 1876-6196 © 2012 Elsevier B.V Selection and/or peer-review under responsibility of the Chairman of the ATALANTE 2012 Program Committee doi:10.1016/j.proche.2012.10.120 792 E Mendes et al / Procedia Chemistry (2012) 791 – 797 R&D program, launched at the CEA Marcoule in the late 90s, aims to demonstrate the feasibility of an innovative grouped separation of the actinides with sufficient decontamination of fission products Previous experimental studies showed that fluoride medium is very attractive in terms of confinement: direct fluoride immobilization into glass matrix up to 15 wt.% has been successfully proved at lab-scale [2] The physical and chemical properties of the product are compatible with an industrial-scale melting process such as La Hague vitrification process These results drove us to assess the potentialities of fluoride melts and to develop a reference process for actinides separation from the fission products The core of this process is based on two different liquid/liquid extraction steps The first step consists in the selective actinides reductive extraction Several studies [3 – 6] highlighted the advantages (in terms of selectivity and salt reprocessing) of performing such extraction step by contacting a fluoride salt, containing the dissolved spent fuel, with a liquid aluminium phase The aluminium acts both as a reductive reagent and a solvent, regarding the actinides The occurring reaction is described by the equilibrium (1): AnF3(salt phase) + Al(metal) ļ An(metal) + AlF3(salt phase) (1) The efficiency of the extraction is given by the distribution ratio DAn = [An]metal / [AnF3]salt (concentrations in g/g) and the selectivity is given by the separation factor SFM/M’ = DM / DM’ The second step of the process is the actinide back-extraction from the Al matrix Several techniques have been inventoried and surveyed and liquid-liquid oxidative extraction was selected for further investigation The process consists in contacting the liquid Al containing the actinides with a pure molten chloride salt containing the oxidizing agent AlCl3 The process reaction is described by the equation (2): An(metallic solvent) + AlCl3(salt) ļ Al(metallic solvent) + AnCl3(salt) (2) The two steps of the core of process have previously been assessed separately The selective recovery of actinides by liquid/liquid reductive extraction was successfully performed [7, 8] The oxidative liquid/liquid backextraction survey led to the determination of both optimal set-up and experimental conditions and shown very promising results regarding the back-extraction of U (the most difficult element to be back-extracted) [9] The present work now focuses on a complete study of the core of process The reductive extraction step has been studied as function of the fluoride salt composition while the back-extraction step focused on the behaviour of the extracted actinides and possible lanthanides The study led to an overview of the feasibility of the core of the reference process developed by the CEA Experimental Al, Cu, LiF, AlF3, NdF3, LiCl, CaCl2 and AlCl3 were purchased from Sigma-Aldrich with > 99.99% purity The preparation of AnF3 is described elsewhere [7] Al–Cu alloy (78–22 mol%) was prepared by dissolving the suitable quantity of copper in liquid aluminium at 800° C This operation was performed under argon sparging in a stainless steel reactor using a graphite crucible 2.1 Reductive liquid/liquid extraction The reductive liquid/liquid extraction step has been assessed as a function of the fluoroacidity of the salt (i.e AlF3 content) on UF4, PuF3, AmF3 and NdF3, mixtures as representative as possible of the reprocessing conditions of ADS and transmutation targets Three compositions of LiF-AlF3 salts have been used, corresponding (from the less to the most fluoroacidic salt) to: LiF-AlF3 Eutectic (E1) (85-15% mol.), LiF-AlF3 Cryolithic (C) (75-25% mol.) and LiF-AlF3 Eutectic (E2) (65-35% mol.) The amounts of salt containing the fluoride mixtures and metallic phases are summarized in table For convenient reasons these experiments were named Run E1, C and E2 (depending on the molar composition of the 793 E Mendes et al / Procedia Chemistry (2012) 791 – 797 fluoride salt) The experiments were carried out using the device (high-temperature liquid–liquid contactor: HTLLC) described in [7] The two liquid phases were contacted at 830°C during 4h under argon atmosphere and a sampling of the two phases was performed in order to determine the distribution ratio DM for each elements and the separation factors SFM/M’ The set-up was cooled down to room temperature and the phases were mechanically separated Since the distribution ratio may vary with the temperature, it was necessary to sample each phase after cooling down (sample of the metallic phase was done by remelting the ingot) Table Fluoride salt and metal quantities used for the reductive liquid-liquid extraction step Run E1 Run C Run E2 Metal (g) 20.0 20.6 20.8 Salt (g) 19.7 18.9 20.0 Containing (as tri or tetra fluorides) : U (mg) Pu (mg) Am (mg) Nd (mg) 190 26 100 52 180 26 96 50 174.5 25.5 96 50 When necessary, i.e when an important coextracted fraction of Nd was observed, an additional scrubbing step was performed by contacting the obtained metallic phase with a pure LiF-AlF3 (cryolithic) salt, in order to remove one part of the Nd The cryolithic composition was chosen to have a good compromise between the amount of Nd and An removed 2.2 Oxidative liquid/liquid back-extraction The metallic ingots coming from the reductive extraction step were recovered to perform the oxidative liquid/liquid back-extraction LiCl-CaCl2 (30-70 mol%)-AlCl3 (introduced as NaAlCl4 such as AlCl3/An ratio equals 7, Actinide amount being given by the previous step analysis) mixtures were prepared This salt composition was previously identified as the best one to perform the back-extraction [9] Due to the high volatility of AlCl3 the back-extraction experiments were carried out using a tightened inconel container ensuring a perfect tightness of the system The back-extraction experiments were carried out by contacting the metallic phase (15 to 18g depending on the recovered amount from previous step) with ~20g of the chloride salt mixture inside the crucible at 700°C, during 4h, under Ar atmosphere The temperature of the system was then cooled down to room temperature The two phases were mechanically separated and samples of the two phases were performed (metallic sample was done by remelting the ingot) 2.3 Sample preparation and analysis The samples coming from the different steps of the study were dissolved: Fluoride salt samples: in HNO3 (3 M) and Al(NO3)3 (1 M) at 90–100° C [10] Metal samples: in HNO3 (4 M) and HF (0.7 M) Chloride salt samples: in HNO3 (0.5 M) The elements present in both phases were quantified by: Į, Ȗ spectroscopy (for Pu and Am), X-ray fluorescence (for U) and ICP-QMS (for Nd) x x x 794 E Mendes et al / Procedia Chemistry (2012) 791 – 797 Results and discussion 3.1 Reductive extraction step Fig depicts the variation of the distribution ratio DM of the studied elements versus the AlF3 molar ratio All distribution ratio decrease while the fluoroacidity (AlF3 content) of the salt increases These results are in good agreement with those already determined by Conocar et al [7] Considering the following reaction: AnF3(salt phase) + Al(metal) ļ An(metal) + AlF3(salt phase) The distribution ratio DAn can be expressed as a function of the equilibrium constant of the reaction and the AlF3activity: D An K J AnF J An ( Al ) a AlF where ȖAnF3 and ȖAn(Al) are the activity coefficients of An in the salt and in the metallic phase, respectively This equation clearly shows that DAn decreases with an increase of aAlF3, AlF3 molar fraction Thus, a decrease of the D values was expected from run E1 to run E2 Fig clearly shows the selectivity between An and Nd The estimation of the separation factors SFAn/Nd led to values higher than 250 in every cases (i.e elements, and salt bath) This observation is also in good agreement with the literature [7] 1000 Kdcoefficient massique Distribution DM 100 10 U Pu Am Nd 0.1 0.01 10 15 20 25 30 35 40 % molaire AlF3) Composition saline LiF-AlF3 ((AlF LiF-AlF3 salt composition mol%) Fig Trends of the distribution ratio versus the AlF3 content in the salt Table (first column) summarizes the efficiency of the reductive extraction step for each element These efficiencies were estimated from the analysis of samples made after cooling down the reactor and were calculated regarding the initial amount of the elements introduced in the salt phase Table shows that the amount of U in the metallic phase is lower than the one expected regarding the DU values (the extraction efficiency should be > 99%) This difference could be due to partial precipitation of uranium during the experiment Such observation had already been observed during preliminary studies of the E Mendes et al / Procedia Chemistry (2012) 791 – 797 uranium reductive extraction The extraction efficiencies of Pu and Am seem to be in close agreement with the D values calculated previously As expected, a non neglectable quantity of Nd was coextracted together with the actinides during run E1, i.e 27% of Nd in metallic phase This coextraction is mainly due to the fluorobasicity of the LiF-AlF3 E1 composition and further justified the implementation of an additional scrubbing step The implementation was latter a good opportunity to assess the behaviour of a core of process including a scrubbing step: The efficiency of the scrubbing step, calculated regarding the initial amount of each element present in the metal phase before scrubbing, shows an important decontamination of the metallic phase, i.e ~80% of the remaining Nd was removed with a minimum loss of actinides 3.2 Oxidative back-extraction step Table (second column) summarizes the extraction efficiencies of U, Pu, Am and Nd after contacting the three metal ingots prepared by reductive liquid/liquid extraction, with CaCl2-LiCl + AlCl3 The percentage of elements back-extracted during the oxidative back-extraction step is calculated regarding the initial amount of the elements present in the metallic phase (coming from the previous step) Table shows that the back-extraction efficiencies of the studied elements remain almost constant for the three experiments This behaviour was expected since the experimental conditions of this step are the same for the three runs; only the initial amount of An and Nd in the metallic phase slightly differs from one experiment to the other Pu and Am are quantitatively back-extracted (>99%) in a single stage while the back-extraction efficiency of U varies from 87 to ~95% The observations made for the behaviour of U, Pu and Am confirm that U is the most difficult actinide to be back-extracted as mentioned in [9] The Nd back-extraction is also important, i.e >90% for run E1 and C and >85% for run E2 No selectivity between An and Ln was expected in these experiments which demonstrate the importance, for the whole process, to reach a high An/Ln separation at the reductive extraction step As a conclusion, run E1, Run C and run E2 successfully demonstrated the feasibility of a back-extraction of the actinides initially present in the Al-Cu matrix These runs also confirmed the optimal experimental conditions, i.e CaCl2-LiCl (70-30 mol %) + AlCl3 (such as AlCl3/An = 7), identified in the previous study [9] 3.3 Global discussion on the feasibility of the core of process Table (third column) summarizes the global efficiency of the complete core of process for each studied element It is representative of the efficiency of a single batch of the core of process, i.e one extraction + one back-extraction step The global efficiency is estimated by the final amount of each element in the chloride salt, regarding the initial amount of the elements introduced in the fluoride salt Table shows very promising results regarding the recovery of Pu and Am: 96 to 98 % of these elements could be recovered in the chloride salt after run E1 and run C in a single batch The results obtained after run E2 seem to indicate that no quantitative recovery of these elements can be achieved in a single batch using a starting LiF-AlF3 E2 mixture This observation is mainly due to the efficiency of the reductive extraction step Concerning the behaviour of uranium, it seems that no quantitative recovery can be achieved in a single batch regardless of the initial fluoride salt composition This could mainly be due to a possible precipitation of part of the uranium during the reductive extraction step Table also highlights the advantages of implementing run E1 with a scrubbing step: 92% back-extraction efficiency was estimated for Nd during E1 If no scrubbing step had been implemented, a final amount of Nd in chloride salt of 25% (of the initial concentration) could have been expected By addition of the scrubbing, only 1.8% of the initial Nd was measured in the chloride salt after run E1 795 796 E Mendes et al / Procedia Chemistry (2012) 791 – 797 Table Extraction, back-extraction and total recovery efficiency of An and Nd Efficiency (%) Run Extraction Back-extraction Scrubbing E1 Global efficiency +/- U 82.4 2.3 87.4 72.6 2.7 Pu 98.8 3.3 99.1 95.9 0.4 Am 99.1 2.8 99.9 96.4 0.4 Nd 27.2 78.1 92.2 1.8 - +/C U 85.3 94.7 81.7 1.8 Pu 97.7 99.0 97.7 0.3 Am 98.5 99.8 98.5 0.2 Nd 10.8 92.4 10.8 - U 64.0 94.2 +/E2 59.3 4.1 Pu 84.0 98.9 83.4 1.7 Am 84.0 99.98 84.0 1.6 Nd 4.4 86.2 3.8 - Conclusion The present work focused on a complete assessment of the core of the reference pyrochemical process developed by the CEA for nuclear spent fuels reprocessing The reductive extraction step was investigated as function of the initial fluoroacidity of the LiF-AlF3 mixture The back-extraction step focused on the behaviour of the extracted actinides and possible lanthanides Moreover the core of process was also studied regarding the implementation of an additional scrubbing step The reductive extraction step exhibits an expected behaviour, already observed [7], as a function of the AlF3 content in the salt phase: the extraction efficiency decreases with the fluorobasicity of the salt The backextraction experiments confirmed preliminary results [9], i.e U is the most difficult actinide to be back-extracted Since Ln are as efficiently back-exctracted as the An, no selectivity can be expected at this step A maximum of Ln should therefore be separated at the extraction step This observation justified the implementation of a scrubbing step after the extraction, using LiF-AlF3 E1 mixture because of the non neglectable amount of Nd coextracted together with the An Finally, the global efficiency of the core of process gave very promising results although it seems not possible to reach the complete recovery of uranium in a single batch, mainly because of a partial precipitation during the reductive extraction step Pu and Am recovery is almost quantitative by using the more fluorobasic salt mixture as initial fluoride salt phase The next studies, already started, will now focus on the head-end steps of the developed pyrochemical process E Mendes et al / Procedia Chemistry (2012) 791 – 797 Acknowledgements Part of this work was carried out with the European Commission financial support within the framework of the ACSEPT Collaborative Project (FP7-EURATOM no 211267) References [1] H.P Nawada, K Fukuda, J Phys Chem Solids 66 (2005) 647 [2] A Grandjean, in: Proceedings of Mater Res Soc Symp., vol 848, Symposium FF, Paper FF9-32, 2005 [3] J Lacquement, S Bourg, H Boussier, O Conocar, C Hamel, A Laplace, C Maillard, L Donnet and J Duhamet in: Proceedings of Int Conf Atalante 2004, Nỵmes, France, June 21–25, 2004 [4] J Finne, G Picard, S Sanchez, E Walle, O Conocar, J Lacquement , J.-M Boursier, D Noel, Journal of Nuclear Materials 344 (2005) 165–168 [5] O Conocar, N Douyere and J Lacquement, Journal of Alloys and Compounds, 389, 29-33, 2005 [6] J Lacquement, S Bourg, H Boussier, O Conocar, A Laplace, M Lecomte, B Boullis, J Duhamet, A Grandjean, P Brossard, D Warin, in: Proceedings of Int Conf Global 2005, Tsukuba, Japan, Paper No 153, 2005 [7] O Conocar, N Douyere and J Lacquement, Journal of Nuclear Materials, 344, 136-141, 2005 [8] Jérôme Lacquement, Hubert Boussier, Annabelle Laplace, Olivier Conocar, Agnès Grandjean, Journal of Fluorine Chemistry 130 (2009) 18–21 [9] E.Mendes, O Conocar, A Laplace, N Douyère, J Lacquement, M Miguirditchian, Procceding of Molten Salts Chemistry and Technology, MS-9, Trondheim, 2011 [10] O.J Wick, Plutonium Handbook: A Guide to the Technology, vol I and II, American Nuclear Society, La Grange Park, IL, 1980 797 ... 3.8 - Conclusion The present work focused on a complete assessment of the core of the reference pyrochemical process developed by the CEA for nuclear spent fuels reprocessing The reductive extraction... identified in the previous study [9] 3.3 Global discussion on the feasibility of the core of process Table (third column) summarizes the global efficiency of the complete core of process for each... representative of the efficiency of a single batch of the core of process, i.e one extraction + one back-extraction step The global efficiency is estimated by the final amount of each element in the chloride