Quantum efficiency as a device-physics interpretation tool for thin-film solar cells

Abstract

Thin-film solar cells made from CdTe and CuIn1–xGaxSe2 p-type absorbers are promising candidates for generating pollution-free electricity. These devices have band gaps which are well-suited for absorption of sunlight. Most importantly, the materials used in these devices can be deposited in a variety of industry-friendly ways, so that the cost associated with manufacturing is generally lower than for competing technologies such as crystalline silicon. The challenge faced by the thin-film photovoltaics (PV) community is to improve the electrical properties of devices, without straying from low-cost techniques that are amenable to industry. This dissertation will focus on the use of quantum-efficiency (QE) measurements to deduce the device physics of thin-film devices, in the hope of improving electrical properties—and efficiencies—of PV materials. While photons are the source of the energy required for creating electron-hole pairs which produce PV power, photons can also have secondary effects, such as modifying the band structure of the solar cell. Under illumination, photoconductivity in the commonly-used CdS window layer can result in band profiles which differ from those present in the dark. By investigating the effect of bias light during QE measurements, one can determine whether photoconductivity plays a role in device operation, in which case QE measurements performed under non-standard conditions may be in error. QE data is presented here which was taken under a variety of light-bias conditions, from dark to 0.40 sun intensity. These results suggest that in most cases, 0.10 sun of white-light bias incident on the CdS layer is sufficient to achieve accurate results. Effects are observed in the short-wavelength region and the near-bandgap region if no conduction-band offset is present. If a large conduction-band offset results in an electron energy barrier, effects are observed throughout the entire spectrum. QE results can be described by simple models which are based on carrier collection by drift and diffusion, and photon absorption. These models are sensitive to input parameters such as carrier mobility and lifetime, which are difficult to measure directly. One model, referred to as the "Simple-drift model", accurately predicts QE results in the thin-film devices studied. By comparing calculated QE curves with experimental results, it was determined that minority-carrier (electron) lifetimes in CdTe are the order of 0.01-0.1 ns. Lifetime determinations for CdTe devices with varying amounts of copper in the back-contact show that increasing amounts of copper correlates with shorter electron lifetimes. This supports the hypothesis that copper serves as a recombination center in CdTe. The spatial uniformity of QE results has been investigated with the LBIC apparatus, and several individual experiments are described which investigate the effect of device processing conditions on cell uniformity. Virtually every electrical variation that can occur in a solar cell can occur in a nonuniform fashion, and can often be detected with the LBIC apparatus. LBIC was used to directly show that the presence of copper reduces photocurrent, likely due to Cu serving as a recombination center in CdTe. Additionally, the use of a highly-resistive TCO layer is demonstrated to improve cell uniformity, due to mitigating shunt-paths and weak diodes. Optical effects are shown to be a cause of nonuniformity in non-thin-film devices, but are not usually dominant in thin-film devices. Forward biased LBIC measurements indicate that conduction-band barriers which cause voltage-dependent collection in some CIGS devices are spatially nonuniform. Devices with transparent back contacts have been suggested by others as a novel device structure for thin-film PV devices. Light-beam-induced current (LBIC) results on ITO back-contact devices show that with front illumination, a significant portion of long-wavelength photons are reflected back into the device from the ITO/air interface, improving long-wavelength QE. CdTe devices which received different back contact surface treatments were studied; including those that received a strong bromine/dichrol/hydrazine (BDH) etch and those that received a weak bromine in methanol etch. While front-side LBIC showed similar uniformity, back-side results showed improved uniformity in BDH-etched devices, attributed to better (more ohmic) back contacts in these devices. In thin-absorber devices, the uniformity trend would likely extend to front-side measurements as well.