Investigating group-V doping limits in CdSeTe and potential application of CdSe as tandem top-cells

Abstract

Cadmium selenium tellurium alloys (CdSeXTe(1-X) known as CST) are a photovoltaic specialist's dream: with ideal single-junction and tandem top-cell bandgaps (based on Se stoichiometry) and large absorption coefficient for all stoichiometries (enabling thin-film applications), CST continues to be a promising material for photovoltaic applications. However, CST is not without its problems. Record efficiency CST devices have demonstrated short-circuit current density (JSC) and fill factor (FF) near their theoretical maximum based on measured bandgaps, but continued improvements to device performance has been limited by the open-circuit voltage (VOC), which has been less than 900mV (< 80% of theoretical maximum) for nearly a decade. Advances in absorber doping for p-type conversion have enabled increased carrier densities, moving from roughly 10^14 cm^-3 with group-I doping (copper) to 10^16 cm^-3 using group-V doping (arsenic or phosphorus), but increases in VOC have not been reflected by this fact. This is typically attributed to the so-called "dopant activation" problem, which accounts for the density of acceptor states provided per density of dopant incorporated and tends to be less than 10% in polycrystalline CST. This indicates that roughly 9 out of 10 dopant atoms form defects which may compensate p-type conversion and additionally hinder device performance. Meanwhile, the use of Se alloying to reducing the effective absorber bandgap has afforded increased JSC, but the roles in which Se and group-V dopants play in conjunction with typical device processing is not widely appreciated. In this work, the modern advancements which have allowed for record efficiency CdTe based devices, namely the incorporation of group-V dopants and Se alloying, are examined to address misunderstandings and provide a framework for improving device processing. An investigation into the impact of group-V dopant concentration in CST using optimized device processing conditions reveals that the density of acceptors formed by group-V doping tends to plateau at a point (roughly 1-5 X 10^16 cm^-3) and further incorporation of dopants tends to reduce device performance through increased radiative recombination. Evaluation of a novel process to increase group-V dopant activation, thereby reducing the concentration of nonactive dopant defects, is presented by use of ion implanted oxygen getters. The presence of oxygen in CST devices is inevitable and an oxidated dopant is effectively an inactive dopant. By implanting elements which have a higher affinity for oxidation relative to dopant atoms, the formation of dopant oxides is reduced, and an increased dopant activation is demonstrated. However, this did not improve device performance in practice, indicating that while the methodology of reducing group-V oxides can increase activation, the process of ion implantation itself may introduce additional lattice defects which negate the increased dopant activation. This leads to an examination of the role Se plays in intrinsic CST absorbers independent of group-V doping, revealing an unexpected n-type intrinsic conductivity, which may be a source of defects which compensate the use of group-V dopants. This indicates that work must be done to carefully balance the distribution of Se throughout the absorber bulk, where a concentration gradient, rather than a uniform ternary stoichiometry, is shown to enable the best performance. Finally, pure CdSe absorbers with a large bandgap of roughly 1.7 eV are examined for potential application in tandem PV devices. CdSe absorbers grown at CSU demonstrate the requisite large bandgap and provide insight into limitations based on absorber thickness. This leads to a discussion on CdSe devices with record VOC. To date, published record efficiency CdSe devices have shown >80% of theoretical short circuit current (JSC/JSCSQ) and >60% of theoretical fill factor (FF/FFSQ). However, such record devices have achieved <50% of the theoretical open circuit voltage (VOC/VOCSQ). The development of CdSe devices using novel transport and contact layer structures involving organic semiconductors and transition metal oxides to achieve >60% of VOCSQ (VOC >900 mV) is presented. The limitations of CdSe absorbers are addressed through temperature and intensity dependent photoluminescence measurements, indicating that low charge mobility due to intrinsic trap states in CdSe bulk are the primary limiting factor to further increasing VOC.