Luminescence measurements inform a strategy for unlocking the full potential of CdTe-based photovoltaics
Jundt, Pascal M., author
Sites, James R., advisor
Sampath, Walajabad S., committee member
Yost, Dylan C., committee member
Gelfand, Martin P., committee member
Kuciauskas, Darius, committee member
Cadmium telluride (CdTe) photovoltaics are characterized by simplicity and speed of fabrication with low usage of materials, all of which translate into low cost. These significant advantages have earned CdTe the second-highest adoption rate of all photovoltaic technologies. However, conversion efficiencies, while functional, are significantly lower than the theoretical limit for this material. This discrepancy is almost entirely a discrepancy in voltage, and the so-called "voltage deficit" of CdTe has stubbornly persisted for decades. While many strategies are being pursued to attempt to reduce the voltage deficit, this issue is fundamentally one of excessive nonradiative recombination due to defects within the absorber material, as will be demonstrated in this dissertation. Recombination is evaluated primarily by luminescence measurements, and as such this class of measurements is particularly relevant to the challenges faced by CdTe research today. The rate of recombination is parameterized by the carrier lifetime, and time-resolved photoluminescence (TRPL) is the most common method of determining this parameter in CdTe. Historically, accurate determination of bulk lifetime was as simple as extracting the time constant of the slowest component of a TRPL decay. However, significant gains in material passivation and doping over the last few years have both decreased the relative influence of trap-assisted recombination and increased the influence of p-n junction fields on TRPL measurements. Consequently, when measurements are performed on complete cells, extracting the tail time constant from a TRPL decay no longer necessarily gives an accurate representation of the bulk material lifetime, and the result is distorted by field effect contributions. This fact is not necessarily well-known by the CdTe community, and extraction of the tail time constant is still the most common way to report lifetimes, even in measurements on complete state-of-the-art cells. This dissertation demonstrates the skewing effects of junction fields, and identifies under which conditions they manifest and how. To probe field effects, external electrical bias was incorporated during TRPL measurements, which allows fairly precise manipulation of fields. Biased TRPL measurements were performed on a variety of samples, and a model was developed to substantiate and better explain the results. It was found that the same characteristics which enable good performance (high lifetime, doping, and mobility) are the same which add complexity to TRPL interpretation. It was also found that field effects can be effectively suppressed by significant forward bias, leading to far more accurate determination of bulk lifetime. TRPL and external radiative efficiency (ERE) luminescence measurement results have indicated very low rates of nonradiative recombination and associated very high lifetime for some CdTe-based materials deposited at Colorado State University, particularly the cadmium selenium telluride (CdSeTe) alloy. While these attributes should allow voltages approaching 1 V and efficiencies on the order of 25%, when incorporated into "traditional" cell architectures these materials typically achieve middling performance at best, and often no performance at all. To unlock the great potential indicated by luminescence measurements, a different cell architecture is proposed which aims to accommodate these materials and take advantage of their characteristics. In an n-i-p configuration, an intrinsic absorber material is sandwiched between two carrier-selective contacts, at least one of which must be transparent. This design eliminates the requirement that the absorber be doped, which penalizes lifetime. Based on the findings of modeling reported here, undoped CdSeTe appears to be an ideal intrinsic layer material. The currently-utilized SnO2:F/MgZnO front contact appears to be excellent as the n-type electron-selective layer. The one missing component is the p-type hole-selective layer; modeling in this dissertation describes in detail what attributes are required of this material. Most important is band alignment with CdSeTe, which should produce a valence band offset as close to zero as possible, and a conduction band offset which forms a sufficiently high electron barrier. Sufficient p-type doping is also quite important. Based on these criteria, ZnTe was identified as a suitable candidate material, and several cells were fabricated with this architecture. While preliminary cells achieved relatively poor performance compared with traditional designs, J-V curves were surprisingly well-behaved, and the almost immediate achievement of functioning cells using an entirely new approach is promising. Luminescence characterization of these structures identified several areas for improvement, namely the use of a p-type dopant other than copper and the replacement of ZnTe with another material with similar band structure but more compatible lattice constant.
Includes bibliographical references.