F. Siebke

Solution of the ZnO/p contact problem in a-Si:H solar cells

Abstract This paper addresses the problem of preparing a good p-layer contact with zinc oxide as TCO. Our approach was to deposit pin cells with different p-layer recipes on ZnO coated SnO2:F and on uncoated SnOi2:F in one run, in order to obtain a direct comparison of the interface properties of the two TCO materials under the condition of as equal as possible surface morphology. The pin cells prepared on the ZnO surface exhibit a lower fill factor (FF). Our experiments demonstrate that the hydrogen interaction with the ZnO surface plays the most decisive role for the ZnO/p contact. We explain the observed effects using a band diagram of the ZnO/p interface and show that the accumulation layer at the ZnO surface — caused by atomic hydrogen in the plasma — is responsible for the low FF in pin cells. Based on this model the contact problem is solved by introducing a μc-n-Si intralayer between ZnO and p-layer resulting in an identical high FF on both ZnO and SnO2 substrates.

Interfaces in a-Si:H Solar cell structures

Abstract The performance of amorphous silicon based solar cells depends on the tailored properties of the various layer materials making up the cell structure as well as on the properties and on the design of the interface regions between the layers. The electronic properties related to the various interfaces are markedly influenced by the Fermi level position within these regions, and by structural properties and chemical compositions resulting from the preparation conditions. Results are presented for the p/i and the TCO/p interfaces and discussed with respect to device performance. Further examples of interface effects are described which are related to chemical reactions and hydrogen diffusion in the course of sample preparation.

Correlation between structure and optoelectronic properties of undoped microcrystalline silicon

The correlation between crystallinity and optoelectronic properties of undoped microcrystalline silicon is investigated. The use of the constant photocurrent method for measurement of microcrystalline silicon is discussed. This method measures the true absorption coefficient only if the crystallinity is a major fraction of a sample. If the crystalline volume fraction is less, it underestimates the absorption coefficient at smaller photon energies. At these energies, carriers are mainly photogenerated in the crystalline phase. Carriers generated in isolated grains contribute less to the photocurrent than carriers generated in grains which are embedded in percolation paths. The shape of constant photocurrent method spectra of microcrystalline silicon with a poor crystallinity differs from spectra of material with better crystallinity.

Material Basis of Highly Stable a-Si:H Solar Cells

We achieved a stabilized efficiency of 9.2 % after only 8 % relative degradation for an a-Si:H/a-Si:H stacked cell with the top-cell i-layer prepared at 140 °C using a high hydrogen dilution of the silane process gas. From a comprehensive characterization of p-i-n cells and the corresponding i-layer material prepared at 140 °C and 190 °C substrate temperature with different hydrogen dilutions, we conclude that the performance of these pin cells strongly correlates with the material properties of the corresponding i-layers. High fill factors after light soaking are reflected in a good microstructure, high photo-conductivity, and relatively low defect density. Whereas the initial V oc is limited by interface recombination, volume recombination dominates the forward-dark current after light soaking. The stabilized Voc as well as the short-circuit current densities correlate with the optical bandgap of the i-layer.

Improved analysis of the constant photocurrent method

Abstract A numerical model has been developed to simulate constant photocurrent method spectra. It takes into account the position of the Fermi energy and the full set of optical transitions between localized and extended states under subbandgap illumination, capture, emission and recombination processes. The comparison of simulated and measured spectra yields information about the density of localized gap states in amorphous silicon, that is the valence-band tail, the integrated defect density, the defect distribution in energy and the charge state of the defect states. For n- and p-type hydrogenated amorphous silicon (a-Si:H) we achieved good agreement between simulation and experimental data. In the annealed state the defect absorption is dominated by a single defect peak which can be attributed to D − states in n-type material and D + states in p-type a-Si: H. In undoped a-Si: H we observed more charged than neutral defect states, confirming the predictions of the defect-pool model. Furthermore, the s...

Surface and bulk density of states of a-Si:H determined by CPM and total-yield photoelectron spectroscopy

The density of states (DOS) of a-Si:H and the influence of light soaking were studied with the constant photocurrent method (CPM) and total-yield photoelectron spectroscopy for films prepared at substrate temperatures between 100°C and 400°C. While by the first method information about the bulk is obtained the other yields the DOS in the surface-near region. At the surface we observed a relative to the bulk enhanced density of states as well as an enhanced influence of light soaking. In addition changes of the DOS of fresh films occur upon annealing at 190°C which can be significantly larger than the light induced effects.

Defect distributions in a-SixGe1−x:H

Gap states in a-SiGe:H alloys were examined by numerical simulations of sub-bandgap absorption spectra measured by the constant photocurrent method and photothermal deflection spectroscopy. In contrast to simple deconvolution methods our analysis uses occupation statistics and takes into account the condition of charge neutrality. The simulations yield information on the energy distribution and the charge state of the defects. The results reveal the coexistence of charged and neutral defects. The defect distributions are similar to those found in amorphous hydrogenated silicon. In the investigated range of compositions charged states dominate the defect density. Taking the position of the defect states as a reference level both band edges shift towards the defect states with decreasing band gap. In contrast to electron spin resonance measurements, no evidence for a distinction between Si-related and Ge-related defect states can be found in sub-bandgap absorption spectra.

More insights from CPM and PDS. Charged and neutral defects in a-Si:H

CPM and PDS spectra of annealed and degraded a-Si:H are analyzed. Numerical simulations of CPM and PDS data using occupation statistics yield information on the energy distribution and the charge state of the defects. The simulations reveal the coexistence of charged and neutral defects resembling the predictions of the defect-pool model. Charged states dominate the defect densities of annealed and degraded a-Si:H. In the case of spatial homogeneous defect densities, different sensitivities of CPM and PDS on charged and neutral defects cause different defect absorptions detected by both methods. Spatially inhomogeneous defect densities caused, e.g. by voids or columnar growth are detected by combining CPM and PDS since PDS detects the total defect density while CPM favors regions with low defect densities.

CPM and PDS -- A critical interpretation of experimental results

CPM and PDS spectra of a-Si:H yield identical shape of the Urbach tail, while the defect absorption measured by PDS differs significantly from CPM. In this work an analysis of CPM and PDS spectra of annealed and degraded films is presented. Numerical simulations of CPM and PDS data, taking into account optical transitions, capture and emission processes as well as the Fermi level, yield information on the energy distribution and the charge state of the defects. The simulations reveal the coexistence of defects in the D{sup {minus}}, D{sup +} and D{sup 0} states. The defect distribution is dominated by charged states as predicted by the defect-pool model. Good agreement between measured and simulated PDS and CPM spectra can be obtained in the case of a homogeneous defect density. It is shown that differences between CPM and PDS are due to different sensitivities of the techniques to charged and neutral defect states. Microscopic inhomogeneities may cause significant additional differences.

Stable and metastable defect distributions in undoped and doped a-Si:H obtained from analysis of the constant photocurrent method

The defect distributions of amorphous silicon with various doping levels are determined by comparison of measured and simulated constant photocurrent method spectra for the annealed as well as for the degraded state. In both cases the defect distributions are dominated by charged states. The dopant induced changes in the density of localized states are discussed in the context of the defect-pool model.