Institute Silicon Photovoltaics

Constant Final State Yield Spektroskopie

for the characterization of amorphous silicon

Basics

In contrast to insulators and crystalline semiconductors, amorphous silicon has electronic states not only in the bands, but also in the band gap. They result from strain in the amorphous network and from rupture of Si-Si bonds. At densities below 10²⁰  eV^-1cm^-3 (as compared to >10²¹ eV^-1cm^-3 in the valence band) they are below the detection limit of conventional UPS spectroscopy. Using Constant-Final-State-Yield spectroscopy (CFSYS) [1-3], these densities of states are still detectable. In CFSYS, the energy E_phot of the photons impinging on the sample is he variable, the kine­tic energy E_kin where the analyser detects the electrons is the fixed quantity (in conventional UPS, this is the other way round). E_kin is chosen such, that the electrons are detected that are just able to overcome the work function barrier Φ_Sample from the sample into the vacuum.

Measurement setup

To vary E_phot, a Xe high pressure lamp with double monochromator is used which gives a sufficient photon flux (light intensity) up to ~ 7 eV (near UV). Thus, the energy interval that can be covered with CFSYS ranges from the Fermi edge (E_bind=0) to (E_bind=−7eV−F_Probe) at about −3,5 eV. Thus, only states in the band gap (a-Si:H: E_gap~1,7 eV) and in the valence band close to the band edge are accessible to the measurement.

Advantage

The advantage of this PES variant lies in the high information depth, i.e. the maximum depth in which a photoelectron can be created that can still exit into the vacuum and be detected by the energy analyzer. At photoelectron energies of 4-7 eV the information depth, which is approximately identical to the mean free path of the electrons between inelastic scattering events λ_mpf, is ~5-10 nm ². In contrast, for excitation with E_phot=21,2 eV (He I), l_impf ~0,5 nm ². Thus, it is possible to measure an average density of states in the first ~10 nm of a sample. This is ideally suited fort he investigation of ~5-10nm thin a‑Si:H emitter, BSF and passivation layers such as the ones used in a‑Si:H/c‑Si heterojunction solar cells. The density of states N_OCC(E) can be measured down to 10¹⁵ eV^-1cm^-3. For very thin a‑Si:H layer thicknesses, the PES spectrum contains additional contributions from the valence band of the underlying c‑Si substrate. Therefore, it is possible to determine the band offset at the a‑Si:H/c‑Si heterojunction from such spectra with a minimum of assumptions [5,6].

 ² The reason is, that the low energy of the photoelectrons is not sufficient to excite phonons in the sample material. Therefore, the main inelastic scattering process cannot take place.

Example of a CFSYS measurement of differently doped, ~10nm thin a-Si:H films on crystalline silicon (c-Si) substrates [6]: E_V marks the valence band edge of the a-Si:H. With increasing n-type doping of the a-Si:H, the defect density in the band gap increases, and the Fermi level E_F (vertical arrows) moves towards the conduction band edge E_C. Electronic states above E_F are not visible in photoelectron spectroscopy, since they are not occupied by electrons.

 References