/en/research/projects/3-01/subprojects/6_semic/index.html
3-01 Dynamics at Surfaces and Structuring
Project coordinator(s): A. Rosenfeld, M. Weinelt

Subproject: Carrier dynamics at semiconductor surfaces

We perform time-, energy-, and angle-resolved two-photon photoemission (2PPE) spectroscopy to study electron dynamics at semiconductor and metal surfaces.

2PPE gives access to the occupied and unoccupied surface and valence electronic states. As light source we use ultrashort laser pulses. They are provided by a 200 kHz Ti:Sa regenerative amplifier (RegA 9050, Coherent) and a homebuilt oscillator, which generates 800 nm pulses of 38 fs duration. Subsequent parametric amplification in two Optical Parametric Amplifiers (OPA 9400 and OPA 9800, Coherent), pulse compression by prism pairs and second harmonic generation provide pump and probe pulses of 30 to 80 fs duration with photon energies between 0.8 eV and 5.4 eV. An ultra-high vacuum chamber (UHV, pressure < 3 x 10-11 mbar) is used to prepare and sustain clean, well-ordered surfaces over several hours (Fig.1).

Fig.1: The UHV endstation consists of preparation and analyzer chamber (upper and lower part) separated by a gate valve. The hemispherical electron analyzer can be seen in the lower left part, the laser system in the background.

The hemispherical energy analyzer PHOIBOS 100 (SPECS GmbH, Berlin) equipped with a 2D-CCD imaging-detector allows for measuring the intensity of photoemitted electrons within a sizeable range of emission angles and kinetic energies at the same time. By introducing a time delay between the pump and probe pulses, allowqs us to follow the temporal evolution of the carrier population in surface as well as bulk states. An important aspect of this experimental setup is that all information about different electronic states is gathered within one image. There we can analyze the lifetimes of different states very precisely with a relative temporal resolution of only a few femtoseconds. As an example we show a spectrum of the Cu(111) surface in Fig. 2. The 2PPE pump-probe measurements of the n=0 Shockley surface state, and the n=1 and n=2 image-potential states in Fig. 3 highlight the potential of the set-up resolving ultrafast dynamics at surfaces.

Fig.2: Left: Schematic of the Cu(111) projected bulk-band structure including the occupied Shockley surface state (SS) and the first two image-potential states n=1 and n=2. Right: Energy- and angle-resolved 2PPE-intensity of the Cu(111) surface at zero pump-probe delay.

Fig.3: Time-resolved measurements of the regions marked in Fig. 2 by colored squares. The red curve gives the cross-correlation of pump and probe pulses. The shift of the blue and green traces correspond to the lifetimes of n=1 and n=2 image-potential states, respectively.

A very recent experiment was performed on the Si(100) surface. Tuning the photon energy of the pump pulse between 4.5 and 5.4 eV unoccupied states up to the Si(100) vacuum level are populated, e.g. the image-potential states n=1 and n=2 in Fig. 4.

Fig. 4: Surface projected bulk-bandstructure close to the Γ – point of the surface Brillouin zone, in- cluding the surface states Dup and Ddown and image-potential states n=1 and n=2 of Si(100).

Our measurements corroborate the existence of the Rydberg-like series of image-potential states on the semiconducting silicon surface. Figure 5 shows the first two image-potential states at zero time delay (Td=0 fs) between pump and probe pulses. For Td > 100 fs n=3 and n=4 states are resolvable, due to the longer lifetimes of the higher image-potential states.

Fig. 5: 2PPE-intensity as a function of the kinetic energy of the photoemitted electrons. For zero delay time between pump and probe pulses n=1 and n=2 image-potential states can be seen. For increasing delay Td the n=3 and n=4 image-potential states are observed.

As we tune the pump-pulse photon-energy across the Dup to n = 1 and Dup to n = 2 transitions, significant variation of both the 2PPE peak positions and intensities are seen (Fig. 6). Below resonance we observe the Dup initial state with the kinetic energy following the pump-pulse photon-energy. Above resonance the Dup intensity is significantly reduced and the peak position reflects the respective image-potential-state resonances. Around the resonance the lowering in energy of the expected peak position and the intensity variations can be explained by a Fano-like resonance, caused by interference between transitions from Dup to the discrete image-potential-state resonances and the bulk states of the silicon conduction band.

Fig.6a: False color representation of the Si(100) 2PPE intensity as a function of pump-pulse photon-energy. The photon energy of the probe pulse ionizing the system at zero delay time is kept constant. b: Kinetic energy of the intensity maxima in a. For the Dup to n=1 and n=2 transitions a clear shift of the maximum to photon energies below those expected from the kinetic energies of the n=1 and n=2 intermediate states is observed. This shift and the negligible intensity after resonance indicate a Fano-like interference. c: Measured intensity along the inital-state energy (red solid line in b.) and simulated Fano-profile for q=-3.4.

In future experiments we plan to compare electron dynamics at silicon and germanium surfaces.

 

[main page of project]
M. Weinelt
weinelt@mbi-berlin.de
Page last modified: 2011/04/06
Technical responsibility:
www-adm@mbi-berlin.de