Decoupling of Component Diffusion in a Glass-Forming ${\mathrm{Zr}}_{46.75}{\mathrm{Ti}}_{8.25}{\mathrm{Cu}}_{7.5}{\mathrm{Ni}}_{10}{\mathrm{Be}}_{27.5}$ Melt Far above the Liquidus Temperature

We report ${}^{95}\mathrm{Zr}$ and ${}^{57}\mathrm{Co}$ radiotracer diffusivities and viscosity data in the equilibrium liquid state of a bulk metallic glass forming ${\mathrm{Zr}}_{46.75}{\mathrm{Ti}}_{8.25}{\mathrm{Cu}}_{7.5}{\mathrm{Ni}}_{10}{\mathrm{Be}}_{27.5}$ melt (Vitreloy 4) far above the liquidus temperature $T_l$ that are not affected by convection, as evidenced via quasielastic neutron scattering. Zr diffusion is strongly decoupled from diffusion of the smaller components by more than a factor of 4 at $T_l$, although it obeys the Stokes-Einstein equation. The results suggest that, in the present Zr-based metallic glass forming systems, diffusion and viscous flow start to develop solidlike, i.e., energy-landscape-controlled, features already in the stable liquid state more than 300 K above the mode coupling temperature $T_c$.

Figure 1

${}^{95}\mathrm{Zr}$ and ${}^{57}\mathrm{Co}$ penetration profiles. The activity of ${}^{95}\mathrm{Zr}$ and ${}^{57}\mathrm{Co}$ is plotted on a semilog scale vs the square of the penetration depth divided by the annealing time. Full and half-full symbols display ${}^{57}\mathrm{Co}$ and ${}^{95}\mathrm{Zr}$ diffusivities, respectively. Data points with open symbols close to the surface are affected by surface hold up and were not taken into account. Note that the profiles in the central part were obtained from simultaneous diffusion of both tracers to rule out systematic errors.

Figure 2

${}^{95}\mathrm{Zr}$ and ${}^{57}\mathrm{Co}$ penetration profiles from simultaneous diffusion of both tracers with different sequence of the tracer layers and otherwise identical conditions. (a) ${}^{57}\mathrm{Co}$ layer on top of ${}^{95}\mathrm{Zr}$ layer and (b) ${}^{95}\mathrm{Zr}$ layer on top of ${}^{57}\mathrm{Co}$ layer. Note the good agreement which allows hold-up effects caused by the bottom layer on diffusion of the top layer to be excluded.

Figure 3

(a) Arrhenius plot of diffusion in ${\mathrm{Zr}}_{46,75}{\mathrm{Ti}}_{8,25}{\mathrm{Cu}}_{7,5}{\mathrm{Ni}}_{10}{\mathrm{Be}}_{27,5}$ (Vitreloy 4) in the stable and in the supercooled liquid state. The present ${}^{95}\mathrm{Zr}$ and ${}^{57}\mathrm{Co}$ diffusivities, including data points obtained with the long capillary technique (green triangles), are shown together with diffusivities from QNS [29] and tracer diffusivities from the literature (Knorr et al. [23], Rehmet et al. [24], Ehmler et al. [25], and Zumkley et al. [26]). The QNS data essentially represent diffusivities of Cu and Ni [29, 30]. Also shown are diffusivities converted via the Stokes-Einstein equation from the present viscosity data measured using electrostatic levitation. The radius of 145 pm was taken, which represents Zr atoms; however, taking radii of the smaller components would not lead to any significant changes on the present log scale. Dashed lines mark the quasieutectic melting temperature $T_m$ and the critical temperature $T_c$ of the mode coupling theory determined form quasielastic neutron scattering. The caloric glass transition temperature $T_g$ (not shown) is 580 K (from DSC at $20\mathrm{K}/\min$). (b) Magnified view of (a) showing only the range of the stable liquid state.