The Influence of Oxide on the Electrochemical Processes in K2NbFT-NaCI-KCI Melts

Transient electrochemical techniques showed that in NaCI-KC1 melts the reduction of K2NbF7 occurs through a two-step reaction Nb(V) --> Nb(IV) --> Nb. When oxide ions were introduced, cyclic voltammetry indicated that the waves corresponding to reduction ol the complex NbF~progressively disappeared. Oxo complexes, such as NbOF~ n 3) , are reduced at a potential close to that of the Nb(IV) fluoro complexes. Niobium metal deposition was perturbed by the formation of niobium suboxides like NbO or Nb4Q. For molar ratios of oxide to Nb greater than one, a black layer of niobium oxide, NbO2, appeared at the electrode surface. The present study confirms the high oxygen affinity of niobium and shows that a careful purification of the electrolyte and feeding materials is required for producing niobium with a low oxygen content. Thermal reduction is the major processing route to produce metallic niobium. However, it is generally necessary to refine the metal considerably to satisfy industrial demands. This has motivated research on alternative processes that involve high temperature electrolysis in molten salts such as alkali metal fluorides or chlorides. The preferred electrolytes are the alkali chlorides which are cheaper and less toxic than fluorides. It has been shown that the mechanism of the electrodeposition process depended on temperature and the nature of the electrolyte. 1-4 At temperatures lower than 650~ the formation of niobium subhalides perturbs the deposition. ~'6 In chloride melts, such as LiCI-KCI, NaCI-KCI, NaCi-AiCl3, the oxidation states Nb(V), Nb(IV), and Nb(III) are stable. 5-8 A recent study ~ has shown that the presence of small amounts of fluoride ions modify the equilibrium conditions. In NaCIKCI, at 720~ for molar ratios of F/Nb, greater than two, the Nb(III) species are not stable anymore and direct reduction of Nb(IV) to metallic niobium is observed. This result explains the differences in the electrochemical behavior between niobium solutions prepared from K2NbF7 and from NbCIs. The presence of oxide ions has a considerable influence on the deposition process. These ions are a common impurity of the melts, and ignoring their presence may lead to erroneous interpretations. Moreover, there have been a few attempts to recover niobium from oxides dissolved in fused salts. It has been shown that niobium oxofluoro complexes in alkali fluoride melts reduced to niobium metal. The presence of a certain amount of oxide actually improved the current efficiency. 9 However, with NaCI-KCI as solvent, K2NbF7 reduced to NbO when oxide ions were present in the melt. I~ This result is in agreement with recent experiments, I~ which indicate that niobium is the only refractory metal which reacts with its oxides giving rise to suboxides such as NbO and Nb4Os. The aim of the present work is to study the mechanism of the electroreduction of K2NbF7 in NaCI-KCI melts as a function of the oxide ion content. Experimental The electrochemical cell consisted of an outer Hastelloy | (Cabot Corporation) shell (a cylinder closed at the bottom end). The glassy carbon vessel containing the molten salt was placed inside this envelope. ~ The experiments were * Electrochemical Society Active Member. Unit6 de Recherche Associ~e au CNRS No. 430. carried out under on argon atmosphere. The NaCI-KCI mixture was purified according to a procedure involving high vacuum drying, chlorine bubbling, argon flushing, electrolysis, and finally filtering in a separate silica cell. 13 Anhydrous potassium heptafluoroniobate (Alpha Products) was purified by recrystalllzation from an aqueous hydrogen fluoride solution. Sodium oxide was obtained from Aldrich (98%) and was used without further purification. A few experiments were carried out using barium oxide or niobium pentoxide from Merck. The reference electrode was a Ni2+/Ni electrode (i mole percent (m/o) NiCl2 in NaCI-KCI); this electrode was standardized by the use of an internal reference system. 14 The counterelectrode was a graphite rod (No. 208 Le Carbone Lorraine). Platinum or tungsten (Matthey reagent, diameter 1 ram) wires were used as working electrodes. The generation of the potential sweep was performed with a Tacussel GSTP4 programmer and a Tacussel PRT 20/10