Critical materials issues in photovoltaics: searching for solutions via theory and simulations
CECAM-HQ-EPFL, Lausanne, Switzerland
Background The possibility to convert sunlight into electrical power is becoming increasingly important as fossil fuels become less available and the environmental, political and physiological cost of our current energy sources becomes more apparent. Although the first solar cell dates back to more than 50 years ago, progress must still be made in solar photovoltaic technology to render sunlight a practical alternative. Although the ultimate improvement on performance will depend on device design and engineering, fundamental scientific issues if addressed properly could guide the choice of the component materials and help optimize their use. Photovoltaics based on crystalline silicon have been predominant since the time of the production of the first (modern) solar cell in 1954. Starting in the 70s, thin film photovoltaics have been investigated especially based on amorphous silicon or chalcopyrites like CuInSe2 (CIS) CuGaSe2 (CGS) and CuIn1-xGaxSe2 (CIGS) . Originally less efficient than silicon, they offered other good characteristics like cost and mass reduction. Also, in contrast to crystalline silicon, chalcopyrites are direct band gap semiconductors, and do not require high purity to function well. Currently, CIGS-based devices are in production. However their efficiency is at most ~20%. Moreover they are expensive. In particular Indium has become rare and needs to be replaced in the perspective of massive production. Also CdS, the typical material forming the buffer layer, will have to be replaced because of the toxic character of cadmium. While great effort is spent on the improvement of these inorganic thin films, novel solutions have appeared on the horizon for the light absorbers, namely quantum dots and organic semiconductors, as well as dye-sensitized (GrÃï¿½ÃÂ¤tzel) cells . In principle, all offer interesting advantages including lower cost and higher flexibility. Moreover the intermediate band (IB) concept - well known in the physics of semiconductors - has been introduced in the engineering of novel solar cells, which provides a mechanism to exceed the limiting efficiency of single gap solar cells. By IB one means an energy band located within the semiconductor band gap; this allows the absorption of two below-band gap energy photons to produce one electron-hole pair and eventually generate enhanced photocurrent without voltage degradation. This scheme can be realized with quantum dots in a host semiconductor or with deep level impurities . It is not a surprise that each choice above implies different fundamental issues to be tackled also by theoreticians in their effort to understand and use their results to help optimize materials design. We restrict our interest to thin film PVs and quantum dots. Theoretical and Computational Approach: State of the Art I. Thin Film Inorganic PVs (CIS et al) Current understanding of the fundamental physics underlying the behaviour of absorber layers of thin film PVs (polycrystalline, multinary materials) relies on the extensive studies pursued over the last 15 years or so by the group of Alex Zunger at the Institute of Solar Energy, Boulder (CO). Examples of issues investigated more recently are
- The role of defects Se-Cu divacancy complex (V-Se-V-Cu) was identified as the source of light- and bias-induced metastabilities in Cu(In,Ga)Se2 (CIGS) based solar cells . The theoretical approach used numerical simulations of equilibrium defect thermodynamics relying on defect formation energies calculated ab initio, and a model describing the transition dynamics after creation of a metastable non-equilibrium state.
- The role of grain boundaries DFT-based calculations led to explain why polycrystalline CuInSe2 solar cells are superior to their crystalline counterpart . GB holes are depressed because an energetic barrier exists for holes from grain interior to grain boundary, which is related to Cu vacancy surface reconstruction. These results were confirmed by more recent experiment on CIGS  and are obviously of critical importance for future design of superior polycrystalline devices.
- The beneficial role of sodium incorporation The presence of small percentages of sodium in CIGS is known to improve conversion efficiencies, which has prompted huge effort in controlling growth and deposition processes (see e.g. ). Based on DFT calculations, the main relevant mechanism of Na in Cu-poor CIS was identified as the replacement of InCu by NaCu, until most of the compensating InCu donors are removed . Moreover NaCu defects or NaInSe2 phases (also predicted to form for high Na concentration) at grain boundaries were proposed to lower the valence-band maximum - owing to the lack of Na d-electron states and thus create a barrier to the GBs for the holes (as above) .
- The possibility of doping Based on DFT calculations, the solubility of transition metal dopants, their site preferences, and electronic and magnetic properties as well were investigated (see e.g. ). Examples of other DFT-based calculations:
- More on the role of sodium: a combined theoretical and experimental (XPS) study of Cu and Na compounds, which are expected to form at the Cu(In,Ga)Se2/In2S3 interface. In CuIn5S8, Na0.5Cu0.5In5S8 and NaIn5S8 compounds variations in the band gaps are confirmed as due to a shift of the top of the valence band with increasing sodium content .
- More on transition metal doping: the effects of cation substitution with Fe in CIS have been investigated on structural and electronic properties of ordered phases also with the aim of making predictions on the influence this replacement might have on photovoltaic conversion efficiencies. Simple analysis of the density-of-states leads to negative conclusions at least for the Fe-rich compounds .
- Intermediate band materials based on CGS: search for suitable candidates. Partial substitution of gallium by transition metal atoms (Ti, V, Cr, Mn) is found to induce the desired additional narrow bands in the band gap of the parent sulphide compound. Ti appears to provide the best characteristics, also concerning stability of the crystal structure .
- DFT-based calculations and beyond. First-principles DFT-based calculations have so far been the leading theoretical approach to the study of the physical properties of the absorber components of real PV devices. It is conceivable that this will continue to be the case for a while, especially in combination with large-scale simulations (e.g. Car-Parrinello molecular dynamics) that allow to investigate directly the formation of defects in inhomogeneous materials, identify probable configurations (structure and composition) of their interfaces, and eventually tackle the intricate problems related to device fabrication (annealing, deposition, doping..). Also, the general current investigation of more accurate exchange-correlation functionals in DFT implementations, notably for the prediction of band gaps, as well as further development and application of TDDFT will have an impact in the study of photovoltaics in general. Especially the combination of the restricted open-shell Kohn-Sham (ROKS) theory and dynamical simulations could be of special interest for the investigation of the excited states of the QDs.
- Materials informatics. The need for improved or novel materials is a common feature of today technologies, and with it the need for combined research in materials screening, growth and characterization. Given the unmanageably large number of a-priori possibilities, structural chemistry and modem computer-aided design can play a unique role in screening potentially important candidates, just as bioinformatics does in the drug discovery process. In particular, virtual screening could complement already existing related experimental effort and/or foster and guide it. As materials informatics, we intend Combinatorial theoretical chemistry, as counterpart of analogous experimental procedures. Combinatorial approaches have been used by experimentalists for the search of thin film PVs based on non-traditional semiconductors, thus producing combinatorial libraries and specific screening methodologies [see e.g. Ref. 17]. - Diagrammatic approaches, which use large databases of empirical, semi-empirical and also quantum parameters on known materials to narrow the search space for potentially beneficial new materials and for the prediction of even more specific information such as stoichiometry, crystal structure and other properties . - The Inverse Band Structure (IBS) method, which is aimed at identifying atomic structures with given desired electronic characteristics. Combined with genetic algorithms for configuration sampling, the IBS scheme has already been applied successfully to determine binary alloys with target physical properties .
- Monte Carlo and classical molecular dynamics simulations. Traditionally these methods are useful for the study of materials processing. For example, they could be used in conjunction with layer-by-layer assembly, which is often employed for exploratory research [see Ref 20]. Special attention will have to be devoted to the selection of the interatomic potentials and possibly more accurate potentials than those currently available will have to be devised.
Claudia Felser (University of Mainz) - Organiser
Tanja Schilling (University of Freiburg) - Organiser
Wanda Andreoni (IBM Research, Zurich Research Laboratory) - Organiser