The methodology currently applied in France and Europe to develop new highly selective hydrophilic or lipophilic compounds for partitioning LLRN is more or less the same as elsewhere in the world:

(i) It starts with the design of a new structure of complexing molecule that could suit the chemical properties of the target LLRN. Radiochemists can search, through comprehensive literature reviews, for existing natural or tailored molecules known for their ability to complex or extract mimicking nonradioactive elements. However, although significant progress has been made in recent years on the use of computational tools to develop macrocyclic ligands for selective metal binding (Hay et al., 2004, 2005), radiochemists can hardly rely on computational modelling approaches, such as those adopted by pharmacologists to design effective drugs, to develop potentially interesting ligands to selectively extract minor actinides from PUREX raffinates, probably because the intrinsic complexity of the chemical systems implemented in their partitioning processes deters chemists from identifying and understanding the physicochemical phenomena underlying solvent extraction, both qualitatively and quantitatively.

Whatever the objectives when describing a chemical system (its energy, the behaviour of its molecular orbital electrons, the main source of its chemical reactivity, its mechanics, or the geometry of its studied complexes), this system can be conceptualized and modelled in different referentials. Nevertheless, these referentials consist of parameters and basic data only achievable after establishing chemical models, which in turn have to be confronted with experimental physi­cal data. Unfortunately, these experimental data are not easy to obtain, since preparing and analyzing radioactive samples in glove boxes is often difficult and limited to specific chemical environments.

Furthermore, the conditions used are usually far from those encoun­tered when reprocessing spent nuclear fuel at industrial scale, where concentrated solutions are implemented. In concentrated solutions, the chemical reactivity of the radionuclides is actually considerably modified as compared to that in diluted solutions. This difficulty, which is currently overcome by using approximate interpolated and/or extrapolated functions, makes the quantitative prediction of the behaviour of a solute in a complex medium almost impossible without experimental data. However, recent coupling of two mathematical approaches proved to be successful in predicting the thermodynamic properties of concentrated solutions: Molecular Dynamics on the one hand, which allows microscopic parameters to be calculated (such as the diameter of the hydrated cations in infinitely dilute solutions or the parameters describing the changes in the size of the hydrated cations with the concentration), and the Binding Mean Spherical Approximation theory on the other hand, which allows osmotic coef­ficients and thus activity coefficients (yb which accounts for the solute chemical reactivity in concentrated solutions) to be estimated via a microscopic representation of the solutions (Ruas et al., 2006).

Statistical approaches aiming at establishing empirical mathemati­cal relationships between sets of ligand structures and their chemical properties (such as complexation or extraction of target LLRN) are sometimes used to help design optimized ligand structures (Ionova et al., 2001). For instance, a quantitative structure-activity relationship can correlate chemical structures with well defined chemical reactivity parameters (Drew et al., 2004a, Varnek et al., 2007). On a more fun­damental level, quantum chemistry and molecular dynamics calcula­tions, such as density functional theory, have been used concomitantly with structural and spectroscopic characterization to enrich basic knowledge of actinide complex formation and/or extraction (Boehme and Wipff, 1999a, b, Karmazin et al., 2002, Baaden et al., 2003, Coupez et al., 2003). Beyond the difficulties of characterizing the radioactive systems investigated, this modelling approach seeks to define inde­pendent physical criteria that can be experimentally observed at different scales (molecular, microscopic as well as macroscopic) and correlated with modelling calculations (Guillaumont et al., 2006, Foreman et al., 2006, Petit et al., 2007, Gaunt et al., 2008).

(ii) Small quantities (sometimes a few hundred milligrams) of the desired compound are synthesized and chemically characterized to validate its structure and determine its purity utilizing suitable analytical tools such as nuclear magnetic resonance (NMR), mass spectrometry or elementary analysis.

(iii) The complexing and/or extracting properties of the synthesized com­pound are then assessed by general screening tests in micro-tubes, first on surrogate solutions, then on solutions spiked with the target LLRN and containing some of (or all) the competing elements present in the genuine nuclear waste.

(iv) If the ligand presents promising complexing and/or extracting proper­ties toward the target LLRN, the chemical reactions involved may be studied more deeply from a thermodynamic and kinetic standpoint, but this requires the synthesis of larger quantities of compounds. These investigations are carried out both at molecular and supra — molecular scales implementing analytical methods adapted to the nuclear environment to probe radioactive complexes. Analytical tech­niques used to investigate complexes at molecular scale include NMR (Wietzke et al., 1998, Lefranqois et al., 1999, Dozol and Berthon, 2007), UV-Visible spectrophotometry (Miguirditchian et al., 2006), X-ray crystallography (Wietzke et al., 1998, Berthet et al., 2005, Baaden et al., 2003, Coupez et al., 2003, Foreman et al., 2006, Gaunt et al., 2008), X-ray absorption spectroscopy (Hudson et al., 1995, Denning et al., 2002, Den Auwer et al., 2004, Gannaz et al., 2006), time-resolved laser — induced fluorescence spectroscopy (Colette et al., 2004, Pathak et al., 2009), microcalorimetry (Miguirditchian et al., 2005), gas chromatog­raphy and electrospray ionization mass spectrometry (Lamouroux et al., 2006, Leclerc et al., 2008, Antonio et al., 2008). Examples of analytical techniques used to investigate complexes at supra-molecu — lar scale are vapour pressure osmometry, small-angle neutron scatter­ing, and small-angle X-ray scattering (Hudson et al., 1995, Erlinger et al., 1999, Berthon et al., 2007).

(v) As the compound will sooner or later be degraded whilst being imple­mented to partition the target LLRN, through acidic hydrolysis and a/у radiolyses of the spent nuclear fuel dissolution solution, under­standing the ligand degradation pathways will help improve its chemi­cal resistance (by modifying its structure or changing the formulations of the separation system) and minimize the effect of its degradation products on the process efficiency (by developing specific solvent washing methods).

(vi) In parallel, systematic parametric tests to acquire partitioning data in test tubes (such as the solvent loading capacity or the variation of the distribution ratios of the target LLRN with a given parameter, as for instance the extractant concentration, acidity, pH, or ionic strength) will allow the formulations of the different aqueous and organic solu­tions to be optimized and the mass action laws to be determined. Thermodynamic models will consequently be developed to interpo­late the extraction isotherms.

(vii) In some cases a counter-current separation flowsheet is calculated utilizing process simulation codes, taking into account both the phase transfer kinetics and the hydraulic characteristics of the contactors (e. g., mixing efficiency, droplet size). The implementation (after scaling up the synthesis of the compound) of this counter-current flowsheet in laboratory scale contactors (mixer-settlers, centrifuges, small-scale pulsed or rotating columns), first on surrogate feeds to test the hydro­dynamic behaviour of the partitioning system, and then on genuine nuclear waste feeds (arising from the dissolution of spent nuclear fuel in concentrated nitric acid) validates or invalidates the process separa­tion performance. Analysis of the counter-current pilot test and com­parison of the observed experimental results with modelling calculations improve the accuracy of the process simulation codes, which can subsequently be employed to extrapolate the design of the workshops in the industrial reprocessing plant.

This methodology is rather long and complicated: it requires a range of skills and involves many researchers as the work progresses. The design of a highly selective hydrophilic or lipophilic compound is therefore intrinsi­cally combined with the development of the partitioning process in which the compound is used alone or in a synergistic mixture. The optimizations of both the selective compound itself and of the related partitioning process are therefore almost inseparable.

In fact, only a very few molecules are developed up to the counter-current demonstration test and most often the compounds investigated do not reach the third step, thus implying continuous iteration between steps (i) and (iii). It is commonly accepted, although difficult to satisfy, that highly efficient and selective compounds must fulfil many criteria, among which are:

• simplicity of preparation: excessive synthesis costs to scale up the pro­duction of a highly selective complexant or extractant might be a disin­centive for an industrial application;

• high complexation or extraction efficiency and a high selectivity toward the target element(s) to ensure high decontamination factors;

• reversibility of the target element complexation or extraction to ensure a high recovery yield;

• high resistance to chemical and radiochemical attack, generating only manageable degradation compounds;

• no formation of precipitate, the occurrence of which could trap some radionuclides and thus decrease their recovery yields, not to mention a criticality event;

• minimum secondary waste generation (burnable compounds consisting of carbon, hydrogen, oxygen, and nitrogen atoms are expected to release only gases);

• fast mass transfer kinetics to ease the implementation of the partitioning process in short time contactors;

• no occurrence of stable emulsions, which block counter-current process implementation.

With regard to the extractant loading capacity (that is to say the amount of metallic cation it can take up from the aqueous feed without inducing third phase formation or organic phase splitting), the major factor to con­sider is the balance between its hydrophilic character and its lipophilicity. The hydrophilic character of a ligand is enhanced by the polar nature of its chemical functions bearing the electron-donor atom(s), whereas the lipophilicity of a ligand is favoured by the length of the hydrophobic carbon chains grafted onto its skeleton. However, due to the presence of these two differing parts in its structure (the polar head interacting with the metallic cation(s) at the water/oil interface and the long hydrocarbon tail(s) facilitat­ing its dissolution and that of its metallic complex(es) in organic diluents), the compound usually presents surface active properties which may cause aggregation in the organic phase. Although aggregation could improve the extraction efficiency through the formation of reverse micelles (Chiarizia et al., 1999, Yaita et al., 2004, Jensen et al., 2002, 2007, Testard et al., 2008), it may also lead to the formation of a stable emulsion in certain experimen­tal conditions (Nave et al., 2004) and consequently to hydrodynamic prob­lems when running the process.

With regard to the selectivity and the complexation/extraction efficiency of a given ligand, considered as a base in Pearson’s ‘Hard and soft acids and bases’ theory (Pearson, 1963), the main factors governing the thermo­dynamics of its chemical reactivity are (i) the nature (‘soft’ or ‘hard’) and (ii) the number of its electron-donor atom(s), which interact with the elec­tron deficient metallic cation (considered as an acid in Pearson’s theory), as well as (iii) its structure (i. e., the three-dimensional geometric orientation of its chemical function(s) bearing the electron-donor atom(s)). The better the fit between the ligand donating function(s) and the metallic cation free orbitals, the stronger the coordination interactions. Multiple denticity, chelation, cyclization, and pre-organization of the chemical functions are complementary but nevertheless significant parameters that induce energy- stabilizing effects on the thermodynamics of the chemical reactivity and selectivity of a given ligand. For instance, the higher the degree of coordina­tion of a neutral extractant to the targeted metallic cation, the stronger the metal-ligand bonding interactions, because of the increased entropy varia­tion term (AS > 0) due to amplification of the system disorder, mainly resulting from dehydration of the metallic cation (Musikas, 1986, Nash, 1993).