Amorphous/crystalline-silicon heterojunction solar cells

Heterojunctions between hydrogenated amorphous silicon, a-Si:H, and crystalline silicon, c-Si, can be made at low process temperatures of typically 150-200 ^oC and therefore offer a low-temperature and low-cost alternative to the traditional silicon wafer technology. The so far best results have been obtained by Sanyo Corporation. Using textured n-type silicon wafers Sanyo reached an efficiency of 20.7 % for a laboratory cell where for both the emitter and the back contact a-Si:H technology was employed. The focus of our work is to develop this low-temperature technology for application in case of temperature sensitive silicon absorbers such as poly-Si films but also multicrystalline or band-grown Si-material. Since these absorbers are in general p-type we so far have worked with heterojunctions on p-type Si wafers only. The best result which we obtained in close collaboration with the FU Hagen in the frame of a networking project (BMBF) is illustrated in Fig.1.

The structure is TCO/a-Si:H(n⁺)/c-Si(p) with back surface field (BSF). The transparent conducting oxide, TCO, was indium tin oxide (ITO) but similar results have also been obtained using ZnO:Al as window material. This solar cell with an area of 1 cm² has a certified efficiency of 16.23 %. This is to our knowledge the best result so far published for this inverse structure on a p-type flat Si wafer. The most critical part in this cell structure is the preparation and optimization of the a-Si:H emitter. In order to develop a deeper understanding of the physics of this device we carefully analysed bulk and interface properties of such amorphous layers and studyied the cell parameters as a function of the thickness d. The result is that the optimum emitter thickness is close to d = 5 nm. At higher values of d the optical absorption in the emitter becomes a limiting factor and at lower values there is a detrimental influence of the TCO work function and of emitter shorts on the a-Si:H/c-Si heterojunction. It turned out that the properties of the above solar cells were not limited by the quality of the heterointerface. This results from the fact that the active heterointerface is perfectly passivated by an optimized wafer precleaning and an a-Si:H(n) layer deposition process. For this purpose special analytical tools including such for getting information on an atomic depth scale were developed. The amorphous films were deposited by plasma enhanced chemical vapor deposition (PECVD) from gas mixtures of silane SiH₄ and phosphine PH₃ at a substrate temperature of 170 ^oC. UV-excited constant final state yield spectroscopy (CFSYS) is used to study the density of states distribution in such films. Comparison with results from photoyield measurements proved that this method has advantages as compared with UPS and gives the most reliable results.

Fig. 2 compares the gap state distributions and Fermi level positions of undoped and phosphorus doped films. Doping clearly leads to a shift of the Fermi level towards the conduction band from E_C – E = 0.5 eV to 0.25 eV. This is connected with a pronounced increase of the concentration of deep gap states and a flattening out of the band tail at the valence band. Qualitatively these observations are similar as in case of thick amorphous films. However, there are pronounced quantitative differences in particular in case of undoped a-Si:H which for thick films would have much lower N(E) values.

Fig. 3 shows that the doping effect saturates at a gas phase doping level of about 10⁴ ppm. For the performance of the solar cell the Fermi level position in the thin film is a crucial parameter. The Fermi level position determines the minimal emitter layer thickness which allows to screen the field penetration from the TCO-a-Si:H(n) contact and thus the reachable band bending in the absorber. The Fermi level position here amounts to about E_C-E_F = 0.25 eV. Of course this value strongly depends on the deposition and posttreatment conditions. This is the first investigation of the behavior of ultra thin a-Si:H emitter films on c-Si substrates by CFSYS.

The dependence of the cell parameters on the thickness d of the emitter layer is presented in Fig. 4. These results reflect the opposing influence of two effects which leads to optimum performance at a value of around 5 nm. The short circuit current j_SC decreases continuously with increasing d due to the increase of optical absorption in the a-Si:H(n⁺) layer and their extreme low contribution to the photocurrent if the a-Si:H emitter layer becomes thicker (dead layer). The second opposing influence consists in the decrease of the open circuit voltage V_OC. Two explanations may be given to this behavior. The break down of V_OC may be the result of imperfections (shunts) in the ultra-thin a-Si:H film. However, measurements of surface photovoltage (SPV) suggest as an alternative explanation that the band bending in the Si-wafer is reduced at very low thicknesses. The TCO-layer may form a Schottky contact with the a-Si:H(n⁺) the field of which penetrates into the c-Si space charge region thus reducing V_OC. Such results suggest that the TCO is not just acting as a window material but also influences the electronic structure of the heterojunction. Therefore the complete contact system TCO/a-Si:H/c-Si has to be analysed and considered in a process of optimisation.

Fig. 5 shows that in fact the internal quantum efficiency in the short wavelength region is strongly enhanced when the thickness descreases from 30 nm to 5 nm. The blue response in the case of a 5 nm thick emitter layer is limited by the TCO absorption edge. This results make evident that the whole TCO/a-Si:H/c-Si system has to be analyzed and considered for an optimization of the solar cell device. A detailed analysis of the temperature dependence of V_OC and the dark saturation current of the hetero solarcell presented in Fig. 1 and Fig. 5 shows that the cell efficiency is not limited by the a-Si:H(n)/c-Si(p) interface, but by the quality of the absorber and the rear contact. Therefore further improvements are possible by an optimisation of the back contact and the absorber. In addition the use of light trapping structures will increase J_SC and will not reduce V_OC as long as the interface of the heterojunction is well passivated.