Application of multivariate analysis techniques to safeguards of the electrochemical treatment of used nuclear fuel

Several countries have shown interest in developing the electrochemical treatment of used nuclear fuel (UNF), commonly termed pyroprocessing. From a proliferation perspective, an advantage of pyroprocessing is its inability to isolate pure plutonium. However, because plutonium is present in the process, it needs to be effectively safeguarded to protect against diversion of plutonium-containing materials that could be post-processed elsewhere. The most complicated unit to safeguard in the process is the electrorefiner where UNF is chemically separated into several material entities. Molten LiCl-KCl serves as the electrolyte into which actinides (including uranium and plutonium), rare earths, and other active metals from the fuel are partitioned after being oxidized to chloride salts. Various voltammetric methods are being developed to measure the concentration of actinides in the molten salt in near-real-time. However, these methods have mostly been applied to molten salt mixtures containing a single actinide. Unfortunately, the presence of multiple actinides will create interferences in the electrochemical responses which could make traditional analysis of voltammetric data inaccurate for determining individual species concentrations. It is proposed to use multivariate techniques to more accurately predict concentrations of multiple actinides from voltammetric data. Two techniques, principal component analysis and partial least squares, are demonstrated on experimental and simulated data for molten salt mixtures containing uranium and plutonium. These techniques consistently yielded more accurate predictions of uranium and plutonium concentrations than simply using peak height. Possible methods of employing multivariate techniques in safeguarding an electrorefiner are also demonstrated.