This article presents a systematic study on the electronic transport mechanisms of CuGaSe2-based thin film solar cells. A variety of samples with different types of stoichiometry deviations, substrates and buffer layers is investigated. We propose two transport models, namely tunneling enhanced volume recombination and tunneling enhanced interface recombination, which allow to explain the observed features for all devices under consideration. The doping level of the absorber layer turns out to be the most decisive parameter for the electronic loss mechanism. The doping is influenced by the type of stoichiometry deviation as well as by the Na content of the substrate. High doping levels result in tunnel assisted recombination. The best solar cells display the lowest tunneling rates. For these devices treatments of the absorber surface by air-annealing and/or the deposition temperature of the CdS buffer layer are decisive for the final device performance. We use the investigation of the open-circuit voltage relaxation to verify the assumptions on the dominant loss mechanism in the different devices.

No abstract

CuGaSe 2 /CdS/ZnO heterostructures with different CuGaSe2 stoichiometry deviations, glass substrates with different Na content and varying CdS buffer deposition procedures are analyzed with admittance spectroscopy, deep level transient spectroscopy, and capacitance–voltage measurements. Cu-rich CuGaSe2 exhibits two acceptor-like bulk traps with activation energies of 240 and 375 meV. The density of both defect states is reduced by air annealing at 200 °C. Ga-rich CuGaSe2 material displays a tail-like energetic distribution of acceptor defects. The maximum of this distribution is at an energy of 250 meV. Defect densities and doping concentrations of Ga-rich material are considerably lower than in Cu-rich material. The different defect and doping densities found in the present investigation fully explain the efficiency gain which has recently been made by changing the material stoichiometry, the glass substrate and the CdS-deposition method for CuGaSe2-based thin film solar cells.

The microstructure of the CdS/CuGaSe2 interface region in Cu-rich CuGaSe2-based polycrystalline thin film solar cells with KCN-treated absorber layers are characterized. Two recipes for the chemical bath deposition (CBD) of CdS with different bath temperatures (60 and 80 °C) are compared. Coherent Cu–Se precipitates are observed in both cases in the grains of the absorber layer. This precipitation cannot be avoided and seems to be a principal limitation for the performance of Cu-rich CuGaSe2-based thin film solar cells. There is a significant difference between both recipes concerning the interaction with the absorber layer surface. For bath temperatures of 80 °C the interaction is much stronger and Cu–S inclusions are found in the buffer layer. These may be responsible for shunts across the pn junction. Owing to the reduced interaction of the CdS deposited at 60 °C there are no Cu–S inclusions. For the 80 °C recipe the CdS/CuGaSe2 interface region consists of a continuous transition zone with low defect density, whereas for the 60 °C recipe the interface is sharper, but the CdS layer contains a high density of stacking faults. The structure of the CdS layer depends also on the bath temperature and the growth orientation of the CuGaSe2 grains. CdS(80 °C) crystallizes predominantly in the zincblende structure and contains less linear and planar defects than CdS(60 °C) which tends to incorporate hexagonal regions in the cubic matrix. Strains due to lattice mismatch as well as mixture between wurtzite and zincblende structures were revealed in high resolution transmission electron microscopy (HRTEM) images of the CdS(60 °C) layer. For CdS(80 °C) the strain is relaxed by twinning and small-angle grain boundaries which were imaged by HRTEM. A suitable CdS buffer layer for Cu-rich absorber layers could not be obtained by CBD because of either the low crystal quality [CdS(60 °C)] or the formation of Cu–S inclusions [CdS(80 °C)]. The enhanced interaction with the Ga-rich absorber layer and improved quality of CdS(80 °C) results in an improved device performance of Ga-rich CuGaSe2-based solar cells.

This paper summarizes recent material and device results obtained at the Institute of Phy sical Electronics at Stuttgart University (lP£). Properties of the mat erial system Cu(ln, Ga)(S, Se h were analy sed and wherever possible a correlation bet ween the material properties and the device characteristics is made. Different high-vacuum techniques of absorber preparation are presented and compared. The f ormation of different alloy s of the family Cu(ln, Ga)(S, S eh is possible for the co-evaporation and at least one of the sequential evaporation techniques. The model f or Cu-rich grow th of Cul nSe2 kno wn from the co-evaporation process can also be used for the Cu-rich growth in the sequential evaporation processes. The surf ace composition of slightly (In, Gas-rich bulk compositions is always determined to be the defect chalcopyrite Cu(/n, GahSe5. Solar cells prepared with different processes and therefore different morphologie s yielded similar device performance. An exponential decay of the density of states from the valence and conduction bands was obtained. Improved cell performance is achieved using absorber layers with higher carrier concentrations. The carrier concentration can be increased by using Na-containing substrates or by utilizing a new Cd-free buffer layer. Device efficiencies in the range of 15% were achieved using the Cd-free buffer layer.1 -5 and 12% for CulnS2.6 But, not only for single cells on a laboratory scale, the material systems passed the status of only being a promising candidate for solar cell applications. Modules with more than 40 W have been realized." Submodules of 50 em? revealed efficiencies of more than 10%7 and the upscaling process to full-size modules is ongoing. To overcome the actual limitations of these cells it is very important to understand the correlation between crystallographical, electrical and optical properties and the achieved solar cell data. This paper tries to illustrate some of these connections.