SFB 616


Pictures SFB616

 Project B2:
The Dynamics of Surface Atoms after Electronic Stimulation

 The subject of project B2 is to study the dynamical behaviour of clean and adsorbate covered surfaces subsequent to short, intense laser pulse excitation. The laser energy initially deposited in the electron system is transferred to the phonon system resulting in a transient, local heating of the topmost atomic layers. In this project we will study the impact of this heating on the surface structure with electron diffraction by means of a pump and probe experiment (see Fig. 1).

Fig 1: Principle of the time-resolved RHEED experiment.

 A femtosecond laser pulse is split in two parts. One part serves as a pump pulse which is directed onto the sample in order to excite the surface. The second part is frequency tripled yielding laser pulses at 4.55 eV. These pulses are directed onto the backside of the photocathode of a pulsed electron gun. Via photoemission short electron pulses are created which are accelerated up to energies of 15 keV. Using an electrostatic lense the electrons are focussed onto the sample. The scattered electrons are detected in a multichannelplate-detector (MCP-detector). The resulting diffraction patterns are recorded with a cooled CCD-camera and the images are stored for further analysis.

The electron pulses probe the surface at fixed times t 0 . By changing the path length for the pumping laser light in the delay line the moment of excitation can be varied. This enables the setting of a delay D t between pump and probe pulse: for D t > 0 the pump pulse arrives before the probe pulse and for D t < 0 the excitation occurs after the electron pulses probed the surface. Diffraction images recorded at different fixed delays reveal a direct measure of the dynamic changes of the surface structure.

Fig 2: Dependence of the (0,0)-spot intensity on the Pump-pulse traveltime for Bi grown on Si(001). The inset shows a RHEED-diffraction pattern of the Bi(0001)/Si(001)-sample.

 The temporal resolution of the experiment is determined by the electron pulse width and the velocity mismatch. The initially short electron pulse broadens due to space charge effects and the vacuum dispersion originating from the finite energy width of the produced photoelectrons. In order to keep these effects small we use high energetic electrons. This has the advantage that the time where the above mentioned effects can act on the pulse are strongly reduced. At these high energies, however, the electrons are only surface sensitive at glancing angles of incidence. In such a geometry the electrons probe the sample over its total width. The pumping laser pulse excites the whole sample at the same time due to the perpendicular incidence on the sample. As the electrons probe the surface at different times the resulting diffraction pattern is an averaged image of the structural changes over the travel time needed to probe the total width of the sample. This effect is called the velocity mismatch because the velocities of laser and electron pulse are very different. For 10 keV the velocity mismatch accounts to 70 ps and is the limiting factor in this experiment. Using thinner samples or tilted laser pulse fronts this contribution to the time resolution can be reduced below 10 ps leaving the electron pulse width as determinant of the time resolution.

Fig. 2 shows the first result obtained with the current setup where a thin Bi-film deposited on a Si(001) substrate was heated with a short laser pulse.

 In the graph the dependence of the peak intensity of the (0,0)-spot is shown as a function of the pump-pulse traveltime. At short times the the intensity remains constant. In this temporal region the pump pulse follows the probe pulse. For the 5.5 keV electrons a steep intensity decrease is found at a pump-pulse traveltime of 6.25 ns indicating the temporal overlap of pump and probe pulses. The sample is heated by the laser pulse resulting in an increased vibration of the surface atoms. As a consequence the diffracted intensity is reduced due to the Debye-Waller-effect. At shorter traveltimes (to the right in Fig. 2) the intensity recovers which is explained by the cooling of the surface. At an electron energy of 7 keV the same general behaviour is observed. However, the dip in the intensity curve is shifted to smaller times. To ensure temporal overlap the laser pulses have to arrive earlier at the sample for the higher energetic electrons. The observed shift corresponds to the different kinetic energies of the two experiments and is in accordance with the expected shift (cf. Fig. 1).

In order to get a better understanding of the transient surface temperature, the observed intensity behaviour in Fig. 2 has to be converted into a temperature scale. This is done by measuring the stationary Debye-Waller-effect of the Bi/Si(001)-system. The result of this measurement is shown in Fig. 3a) where the intensity of the (0,0)-spot in dependence of the sample temperature is displayed.

Fig 3.

 Fig. 3: a) shows the stationary Debye-Waller-Effekt of a Bi(0001)/Si(001) sample. The exponential fit is in accordance with the Debye-Waller behaviour yielding a surface Debye-temperature of 48 K. The squares in b) show the timeresolved data from Fig. 2 converted to a temperature scale using the temperature calibration of 3a). The decay time of the exponential fit in b) was determined to be 640 ps. The blue curve is the result of the convolution of the temperature dependence from the heat-diffusion-model with a rectangular function of width 80 ps. The convolution accounts for the velocity mismatch. Using this intensity-temperature calibration, the data in Fig. 2 for the 7 keV electrons are converted to a surface temperature. The time dependent surface temperature is shown as squares in Fig. 3b). In fig. 3b) the time-axis was shifted in a way that the observed temperature rise was set to zero delay. For negative delays the surface temperature stays constant at 80 K the base temperature at which the experiment was conducted. At a delay of 0 ps the temperature rises linearly and reaches its maximum of about 170 K after 80 ps. The observed linear increase can be explained with the timeresolution of the experiment because a much faster increase is expected. As discussed before the timeresolution is determined by the velocity mismatch which is 80 ps for 7 keV electrons at 4° angle of incidence and a samplewidth of 4mm. After the maximum temperature is reached the surface cools down exponentially with a time constant of 640 ps.

 For comparison with the data, the one dimensional heat diffusion model was solved with literature values for bismuth and the parameters of the experiment. In this model the surface region is heated by a short laser pulse on a length scale of the lights adsorption depth. Subsequently the surface cools down by heat diffusion into the bulk. The resulting temporal profile of the surface temperature was convoluted with a rectangular function of width 80 ps to account for the velocity mismatch and is shown in Fig. 3b) as blue line. The observed temperature increase is very well reproduced confirming that the resolution at 0ps delay is limited by the velocity mismatch. In contrast, the heat diffusion model can not describe the temperature decrease at higher delays. This implies that this model is not applicable to the thin film structure of our experiment. The boundary between 6 nm thin Bi-film and the underlying Si-substrate plays is the dominating factor on the heat conduction (thermal boundary conductance).


Internationaler Workshop 2008

Workshop 2008

B2 Möllenbeck et al.
PDF (0.7 MB)

DPG 2008

DPG 2008
B2 Hanisch et al.
 PDF (1.5 MB)

Begehung 2005

Begehung 2005
B2 Janzen et al.
 PDF (1.6 MB)