Ix view of advances in the theory of metalfmetal-ion exchange and of applications of modern techniques of electrochemical measurement to solid electrodes, it is timely to re-emphasise the necessity for control of the metallurgical condition of the metal . Jacquet,' electroplating copper on to cast copper from acid sulphate solutions, observed that oxide particles acted as nuclei . Vaughan and Pick" extended these observations using the electron microscope to follow the early stages of nucleation on a variety of copper substrates. They showed that particular care must be taken in the study of classical nucleation to exclude possible impurity effects, especially where impurity is segregated. Bond et al.$ showed that segregation occurred in the solidification of commercial high purity aluminium, resulting in precipitates about 100 nm in size and leading to pitting during etching with 0 .5 M NaCl-3 % H202. To avoid pitting, either exceptionally highly refined metal was needed, or special solidification conditions were required .
The passage of a molten zone along the length of a metal rod is an established method of purification (zone-refining) that gives, under suitable conditions, a single crystal. The slow freezing of molten metal in the production of single crystals by most methods also promotes the redistribution of impurity . A treatment of the subject has been given by Pfann.* In the present work a rod of nickel was refined in a floatingzone electron-beam melter .5 The starting material was 99 . 965 per cent nickel, the principal impurities being iron (180 at-ppm), silicon (75 at-ppm) and sulphur (50 at-ppm). The amount of oxygen present before melting was less than 4 ppm . A section cut from the centre of the resulting single crystal carried the impurities iron (180 at-ppm), silicon (4 at-ppm) and sulphur (1 at-ppm). All other impurities were reduced to a very low level . Zone refining was most effective in reducing the amount of residual impurities except for iron, which in these concentrations is freely soluble in both solid and liquid nickel. Iron was shown to have no effect in the electroplating experiment to be described .
Cellular solidification substructure (a well-known phenomenon), was found in the section cut from the rod . Associated with the substructure was a very fine precipitate phase occasionally forming small clusters at the tripoints of the hexagonal cells (Fig . ) . This enhanced nodal segregation has been described elsewhere :7 the resulting precipitate particles were large enough for microprobe analysis (1 ,ram) and were shown to be NiO . A very considerable segregation concentration effect had occurred. In regions of the section where the impurity content was high, cell walls were decorated with precipitate and electrolytic etching produced corresponding * Manuscript received 7 July 1970.