Icosahedral Order, Frustration, and the Glass Transition: Evidence from Time-Dependent Nucleation and Supercooled Liquid Structure Studies

One explanation for the glass transition is a geometrical frustration owing to the development of non-space-filling short-range order (icosahedral, tetrahedral). However, experimental demonstrations of this are lacking. Here, the first quantitative measurements of the time-dependent nucleation rate in a ${\mathrm{Zr}}_{59}{\mathrm{Ti}}_3{\mathrm{Cu}}_{20}{\mathrm{Ni}}_8{\mathrm{Al}}_{10}$ bulk metallic glass are combined with the first measurements of the evolution of the supercooled liquid structure to near the glass transition temperature to provide strong support for an icosahedral-order-based frustration model for the glass transition in Zr-based glasses.

Figure 1

(a) Number of quasicrystal grains per unit volume $N_v$ as a function of annealing time at the nucleation temperature $T_N$. The growth temperature was $430°\mathrm{C}$. The slope of $N_v$ versus $t$ equals $I^s$ at long annealing times. The induction time $\theta$ is obtained by extrapolating $N_v$ in the steady-state regime to the time axis (illustrated by the dashed line for the $395°\mathrm{C}$ data). (b) Fit to scaling relation predicted from classical theory of nucleation [Eq. (1)].

Figure 2

Plot of the log of the product of the experimentally measured steady-state nucleation rate and the induction time as a function of the inverse driving free energy and temperature for $i$ phase nucleation in a ${\mathrm{Zr}}_{59}{\mathrm{Ti}}_3{\mathrm{Cu}}_{20}{\mathrm{Ni}}_8{\mathrm{Al}}_{10}$ metallic glass. The straight line is a best fit to the data points. The vertical error bars reflect the experimental uncertainty in measurements of the induction time and steady-state nucleation rates. The horizontal ones reflect the uncertainty in the quasicrystal metastable liquidus temperature (see text) as well as that in the specific heat measurements of the supercooled liquid near the glass transition.

Figure 3

The $g(r)$ computed from the experimental $S(q)$ (circles) and the RMC fits (solid lines); the data were sparsely plotted to show the quality of the fits.

Figure 4

(a) A normalized BOO analysis of ${\mathrm{Zr}}_{59}{\mathrm{Ti}}_3{\mathrm{Cu}}_{20}{\mathrm{Ni}}_8{\mathrm{Al}}_{10}$ alloy at various temperatures. The arrow indicates the direction of decreasing temperature. (b) The intensity of the 1551 Honeycutt-Andersen index (characteristic of ISRO) as a function of liquid temperature. The dashed line is a polynomial fit.