Abstract

There has been a great deal of interest in the investigation of the primary steps of phase transitions at semiconductor surfaces excited by an ultrashort laser pulse. Raman experiments [1] revealed that energy from the electronic system is transferred to the phonon system (lattice) on a time scale of about 1-2ps. Direct observation of the sample surface after excitation showed the onset of melting to be within a few or less than 1ps [2]. Second harmonic generation in reflection turned out to be a powerful tool to detect symmetry changes of the sample surface [3-5]. The general observation is that the drop of the second harmonic efficiency occurs within the first one hundred fs after excitation. In contrast the reflectivity of the fundamental changes on a time scale of several hundred fs up to one ps depending on the excitation density. While the SHG probes only about a layer of lOnm the reflectivity of the fundamental is determined by a layer thickness of several hundred nm. Therefore one possible explanation for the different time constants rests on propagation effects of the melt front, another assumes an intermediate state reflecting a changed lattice symmetry which is reached before a considerable part of energy is transferred to the phonon system (cold melting) [3]. We supplemented pump-probe and time resolved SHG in reflection by transient grating measurements. The latter are extremly sensitive against propagation effects because they change the grating structure and thus the diffraction efficiency. Corresponding measurements are shown in Fig.l. It is intresting to note that observable first-order diffraction occured only at excitation densities above the melting threshold. From Fig.1 it is evident that the diffraction dynamics is similar to that of the fundamental reflection. A theoretical model explains this as a consequence of the fact that propagation effects can be neglected within the first few ps. An upper limit for the velocity of the melt front velocity can be estimated and is in the order of 500 m/s. Since the diffraction rise time of the UV-probe (310nm) is considerably longer than the decay of the second harmonic efficiency the existence of an intermediate state or phase prior the melting becomes even more likely.

© 1992 The Author(s)