Inelastic electron-phonon scattering and time-resolved laser-induced phase transformation in aluminum.

In a beautiful experiment Williamson, Mourou, and Li' have measured directly the time it takes for the solid-liquid transition of a free-standing aluminum film. This was done by a direct measurement of the disappearence of electron diffraction patterns after the application of a short laser pulse. The results are interpreted' as due to an expansion of a critical nucleus of the melt throughout the solid sample thus leading to complete melting. One may doubt the validity of this model for melting through nucleation and, in addition, the model depends on parameters which are not independently measured. It is the purpose of this comment to propose a different explanation of the results of Williamson, Mourou, and Li; this alternative explanation relies on the inelastic electron-phonon scattering in the system. In order to determine whether Fig. 3 of Ref. 1 could be related to the time for the homogeneous melting of the sample, we have performed a molecular-dynamics simulation of the melting of a Lennard-Jones crystal. In this simulation a set of 4000 particles with periodic boundaries interact via Lennard-Jones potentials at constant pressure and some prescribed starting temperature T, . This system, at T, and zero pressure, forms a stable fcc crystal at values of T, below —0.8 in the usual molecular-dynamics reduced units. An fcc crystal was equilibrated at zero pressure and T, =G.75, and at a time, conveniently referred to as t = 0, the temperature of the system was raised from T, to T =0.8, i.e., slightly above the melting temperature. The equations of motion of all particles were solved simultaneously and followed in time, as is usual in molecular-dynamics simulations. Figure 1 shows the intensity of the (100) fcc peak as a function of time. Notice that after t —8 psec (for argon) this intensity drops essentially to zero, indicating a melting of the Lennard-Jones crystal. After the melting process is over the diffusion constant was found to be D =4.8X10 cm /sec, characteristic of self-diffusion in liquids; before heating, i.e., before t =0, the solid in equilibrium showed mean square atomic displacement (r ) = 0.76 A with no diffusion. Another simulation with T, =0.7 (i.e., the system at a temperature further down in the solid r'ange) gave (r2): —0.48 A, t = 27 psec, and D = 4.7X 10 s cm /sec when the temperature was raised to T =0.8. An earlier calculation, 3 with constant volume (not constant pressure), gave essentially the same results. Since these times are orders of magnitude shorter than the measurements of Ref. 1, it is clear that homogeneous melting cannot explain the results of Ref. 1. Changes in the potentials are not expected to change this qualitative 1000-