VUV Spectroscopy of Atoms, Molecules and Surfaces

88 Chapter 5. Femtosecond VUV core-level spectroscopy ...
approach it is implicitly assumed that the electrons can be designated a temperature
which may not be justified within the first ps after the presence of
the desorbing laser pulse [46]. Since the electron temperature depends on the
pulse duration and the absorbed fluence but not the wavelength, the wavelength
dependence of the desorption yield observed by the group of Mazur
has been attributed to the influence of non-thermalized substrate electrons
[35].
Attempts have been made to deduce some of the time scales involved
in the electron-mediated desorption proces by means of pump-probe measurements.
Most of these have applied a two-pulse correlation scheme where
the desorption yield is monitored as a function of pump-probe delay [39,
47, 51, 63, 64]. These measurements have revealed a peak with a ∼1 ps
width (FWHM) on top of a broad plateau with a decay time of ∼100 ps.
The ∼100 ps decay time has been attributed to equilibriation of the laserheated
region with the ambient substrate [51] but the assignment of the ∼1ps
spike is less clear. Bonn et al. have performed two-pulse correlation measurements
for CO and CO2, generated from the oxidation of CO, on Ru(0001)
yielding correlation times of 20 ps and 3 ps, respectively [47]. These were
taken as evidence for phonon- and electron-mediated processes, respectively,
with reference to the much longer cooling time of the lattice compared with
the electron gas. The short-lived component was attributed to the electronmediated
mechanism on account of a comparison with the sub-picosecond
electron-phonon coupling time which determines the cooling time of the hotelectron
gas [39, 47]. This was supported by the fact that the energy barrier
to CO desorption is significantly lower than that for CO2 formation, the latter
being given by the activation energy for breaking of the Ru–O bond of
the dissociatively adsorbed O atoms [47].
It remains unclear what the correlation time really means [35], i.e. with
which physical proces it should be identified. An upper limit to the actual
desorption time, i.e. the time required for the breakage of the adsorbatesubstrate
bond, has been set to 325 fs from a measurement of the change
in second-harmonic signal of the probe as a function of pump-probe delay
by Prybyla et al. [41]. The second-harmonic signal depends on the degree of
adsorbate coverage and thus allows a more unambiguous interpretation of the
observed time scale. An even more direct interpretation will be allowed by
the technique of fs VUV core-level spectroscopy where the size of a substrate
core-level shift depends on its chemical surroundings, in this case the degree
of adsorbate coverage.