Self-Diffusion in Polycrystalline Nickel

Abstract

The self-diffusion of nickel has been studied in polycrystalline sanlples by a sectiorling technique. There is evidence of grain boundary diffusion below temperatures of 1150" C. The results obtained between 1150' and 1400" C are representative of volume diffusion and are represented by the expression D = 3.36 e-69.8001RT.A conlparison is made with the results of other self-diffusion studies using Zener's hypothesis. INTRODUCTIONThe most precise measurements made on the rate of volume self-diffusion in metals have been carried out with single crystal specimens. Measurements made with polycrystalline specimens have been in good agreement with those made with single crystal specimens only when the diffusion study was confined to temperatures close to the melting point of the metal. None of the reported investigations (1, 2, 3) made of the self-diffusion of nickel in polycrystalli~~e samples have satisfied this criterion. Therefore, it was thought worth while to report some worlc on the rate of nickel self-diffusion in the temperature range 1050' to 1400' C. E X P E R I M E N T A L Diffusion coefficients were obtained by the standard sectioning method (4) and by a modification of Gruzin's integral residue method (5). In some cases both techniques were en~ployed to analyze the same specimen, which permitted a direct coinparison of the results obtained with each.The nickel tracer employed contained two long-lived isotopes, Ni63 decaying with the emission of a 67-kev p-particle and Ni5Vecaying by K-electron capture with the emission of a Co K, X-ray (6). A Coj8, gamma-emitting impurity present in the tracer solution was completely removed by an ion exchange procedure (7). In the 27r internal source proportional counter employed, the ratio of detected disintegrations of Nim to Ni59 was in excess of 5x103 for almost all sources counted. Thus, with one exception which will be discussed later, it was assumed that the tracer consisted of a single active species, Ni63.The diffusion specimens consisted of polycrystalline nickel cylinders about 0.60 in. in diameter with the active tracer electroplated on one face. The total analyzed impurity

DIFFUSION OF Ni63 IN IRON, COBALT, NICKEL, AND TWO IRON–NICKEL ALLOYS

“…A comparison of the present results for self-diffusion of nickel with those of a previous investigation (2) shows that the two studies are in close agreement as is shown in Table I1 where the values are tabulated. Both investigations gave higher values for the activation energy than that reported by Hoffman, Pikus, and Ward (6).…”

“…The area under the curve of C(x, t ) versus x was integrated to determine a value for Co. Agreement between the duplicate specimens was within 1%. The diffusion coefficient calculated from equation [2] for the mean value of Co was in excelleilt agreement with that calculated from the sectioning data in the ilorillal inanner using equation [I]. This agreement indicates that good results should be obtaiilable with this surface activity technique.…”

“…Therefore, while the tracer layer was thin enough to satisfy the boundary For personal use only. conditions of equation [2], it appears to have been thick enough to cause a concentration effect.…”

“…I, and are typical of those obtainecl a t all temperatures. For surface activities below 4500 c.p.m., the curves are linear conforrniilg to equation [2] and extrapolate to pass through the origin. However, for the diffusion of nicltel into iron, cobalt, and the two iron-nickel alloys the plots are curved in the initial region, the curvature correspollding to a decrease in the diffusion coefficieilts with tirne.…”

“…However, for the diffusion of nicltel into iron, cobalt, and the two iron-nickel alloys the plots are curved in the initial region, the curvature correspollding to a decrease in the diffusion coefficieilts with tirne. This curvature represents either a concentration dependence of the diffusion coefficient or a failure t o satisfy the boundary conditions of equation [2] by using a tracer layer of finite thickness. This latter supposition was not substantiated, as a check experiment showed that a plot of C(0, t ) versus l/dt is linear for nicltel self-diffusion even for diffusion times short enough to promote much less diffusion than that obtained in those speciinens where curvature was observed.…”