In polycrystalline thin-film cadmium telluride (CdTe) solar cells, atomic defects (dopants: copper (Cu), arsenic (As); and selenium (Se) alloy) have significantly enhanced hole density and minority carrier lifetime. Density functional theory (DFT) has predicted the atomic configurations of relevant defects and their electronic structures. Yet, experimental evidence of the defects, especially their spatial distribution across the absorber, is still lacking. Herein, since it can probe local atomic structure of elements of interest with trace-elemental sensitivity, nanoprobe X-ray absorption near edge structure (XANES) spectroscopy was used to elucidate atomic structures of Cu, As, and Se. After XANES spectra were measured from CdTe devices, the atomic information was extracted from the measured spectra by fitting them with reference spectra, which were simulated from 1) point defects and grain boundaries (GBs) predicted by DFT; 2) secondary phases which could form under processing conditions. XANES analysis of various device architectures revealed structural inhomogeneities across the absorbers from point defects to secondary phases. The majority of the Cu dopant atoms form secondary phases with surrounding atoms even inside the absorbers, explaining the low dopant activation. When entering the target lattice site (Cd), Cu forms a complex with chlorine (Cl) and becomes a donor defect, compensating hole density. Compared to Cu, As dopant tends to enter the target site (Te) more frequently, explaining higher hole density in As-doped CdTe. Notably, As on the Te site forms neutral charged complexes with Cl. Although they are not as detrimental as the Cu-Cl complex, the As-Cl complexes may be responsible for low dopant activation and compensation observed in As-doped CdTe devices. Complementary to the DFT prediction, this work provided the distribution of Se local structures across the absorber, specifically the variation of Se-Cd bond lengths in differently performing areas. Under environmental stressors (heat and light), it showed atomic reconfiguration of Se and Cl at GBs, and Se diffusion into the bulk, co-occurring with device degradation. This framework was also extended to study defect evolution in other thin-film solar cells (CIGS and emerging perovskite). XANES analysis has shed light on atomic defects governing solar cell performance and stability, which are crucial in pushing the efficiency toward the theoretical efficiency limit.

Thin film solar cells are based on polycrystalline materials that contain a high concentration of intrinsic and extrinsic defects. Improving the device efficiency in such systems relies on understanding the nature of defects - whether they are positive, negative, or neutral in their influence - and their sources in order to engineer optimized absorbers. Oftentimes, these are studied individually, as characterization techniques are limited in their ability to directly relate material properties in individual layers to their impact on the actual device performance. Expanding the tools available for increased understanding of materials and devices has been critical for reducing the translation time of laboratory-scale research to changes in commercial module manufacturing lines. The use of synchrotron X-ray fluorescence (XRF) paired with X-ray beam induced current and voltage (XBIC, XBIV respectively) has proven to be an effective technique for understanding the impact of material composition and inhomogeneity on solar cell device functioning. The combination of large penetration depth, small spot size, and high flux allows for the measurement of entire solar cell stacks with high spatial resolution and chemical sensitivity. In this work, I combine correlative XRF/XBIC/XBIV with other characterization approaches across varying length scales, such as micro-Raman spectroscopy and photoluminescence, to understand how composition influences device performance in thin films. The work described here is broken into three sections. Firstly, understanding the influence of KF post-deposition treatment (PDT) and the use of Ag-alloying to reduce defect density in the Ga-free material system, CuInSe2 (CIS). Next, applying a similar characterization workflow to industrially relevant Ga-containing Cu(In1-xGax)Se2 (CIGS) modules with Ag and KF-PDT. The influence of light soaking and dark heat exposure on the modules are also studied in detail. Results show that Ag used with KF-PDT in CIS causes undesirable cation ordering at the CdS interface and affects the device through increased potential fluctuations. The results also demonstrate the importance of tuning the concentration of KF-PDT used when intended to be used in Ag-alloyed devices. Commercially-processed modules with optimized Ag and KF concentrations are shown to have the device performance instead be dominated by variations in the CIGS composition itself. In particular, changes in Cu and Se concentrations are found to be most influential on the device response to accelerated stressors such as dark heat exposure and light soaking. In the final chapter, simulations of nano-scale XBIC and XBIV are done to contribute to the understanding of these measurements.

The origins of carrier mobility (μe) were thoroughly investigated in hydrogenated indium oxide (IO:H) and zinc-tin oxide (ZTO) transparent conducting oxide (TCO) thin films. A carrier transport model was developed for IO:H which studied the effects of ionized impurity scattering, polar optical phonon scattering, and grain boundary scattering. Ionized impurity scattering dominated at temperatures below ~240 K. A reduction in scattering charge Z from +2 to +1 as atomic %H increased from ~3 atomic %H to ~5 atomic %H allowed μe to attain >100 cm^2/Vs at ~5 atomic %H.

In highly hydrogenated IO:H, ne significantly decreased as temperature increased from 5 K to 140 K. To probe this unusual behavior, samples were illuminated, then ne, surface work function (WF), and spatially resolved microscopic current mapping were measured and tracked. Large increases in ne and corresponding decreases in WF were observed---these both exhibited slow reversions toward pre-illumination values over 6-12 days. A hydrogen-related defect was proposed as source of the photoexcitation, while a lattice defect diffusion mechanism causes the extended decay. Both arise from an under-coordination of the In.

An enhancement of μe was observed with increasing amorphous fraction in IO:H. An increase in population of corner- and edge-sharing polyhedra consisting of metal cations and oxygen anions is thought to be the origin. This indicates some measure of medium-range order in the amorphous structure, and gives rise to a general principle dictating μe in TCOs---even amorphous TCOs. Testing this principle resulted in observing an enhancement of μe up to 35 cm^2/Vs in amorphous ZTO (a-ZTO), one of the highest reported a-ZTO μe values (at ne > 10^19 cm^-3) to date. These results highlight the role of local distortions and cation coordination in determining the microscopic origins of carrier generation and transport. In addition, the strong likelihood of under-coordination of one cation species leading to high carrier concentrations is proposed. This diverges from the historical indictment of oxygen vacancies controlling carrier population in crystalline oxides, which by definition cannot occur in amorphous systems, and provides a framework to discuss key structural descriptors in these disordered phase materials.

Hg1-xCdxTe (MCT) has historically been the primary material used for infrared detectors. Recently, alternative substrates for MCT growth such as Si, as well as alternative infrared materials such as Hg1-xCdxSe, have been explored. This dissertation involves characterization of Hg-based infrared materials for third generation infrared detectors using a wide range of transmission electron microscopy (TEM) techniques.

A microstructural study on HgCdTe/CdTe heterostructures grown by MBE on Si (211) substrates showed a thin ZnTe layer grown between CdTe and Si to mediate the large lattice mismatch of 19.5%. Observations showed large dislocation densities at the CdTe/ZnTe/Si (211) interfaces, which dropped off rapidly away from the interface. Growth of a thin HgTe buffer layer between HgCdTe and CdTe layers seemed to improve the HgCdTe layer quality by blocking some defects.

A second study investigated the correlation of etch pits and dislocations in as-grown and thermal-cycle-annealed (TCA) HgCdTe (211) films. For as-grown samples, pits with triangular and fish-eye shapes were associated with Frank partial and perfect dislocations, respectively. Skew pits were determined to have a more complex nature. TCA reduced the etch-pit density by 72%. Although TCA processing eliminated the fish-eye pits, dislocations reappeared in shorter segments in the TCA samples. Large pits were observed in both as-grown and TCA samples, but the nature of any defects associated with these pits in the as-grown samples is unclear.

Microstructural studies of HgCdSe revealed large dislocation density at ZnTe/Si(211) interfaces, which dropped off markedly with ZnTe thickness. Atomic-resolution STEM images showed that the large lattice mismatch at the ZnTe/Si interface was accommodated through {111}-type stacking faults. A detailed analysis showed that the stacking faults were inclined at angles of 19.5 and 90 degrees at both ZnTe/Si and HgCdSe/ZnTe interfaces. These stacking faults were associated with Shockley and Frank partial dislocations, respectively. Initial attempts to delineate individual dislocations by chemical etching revealed that while the etchants successfully attacked defective areas, many defects in close proximity to the pits were unaffected.