No abstract

The dependence of the surface structure, composition, and electronic properties of three low index SnO 2 surfaces on the annealing temperature in vacuum has been investigated experimentally by low energy He + ion scattering spectroscopy ͑LEIS͒, low energy electron diffraction ͑LEED͒, scanning tunneling microscopy ͑STM͒, and angle resolved valence band photoemission ͑ARUPS͒ using synchrotron radiation. Transitions from stoichiometric to reduced surface phases have been observed at 440-520 K, 610-660 K, and 560-660 K for the SnO 2 ͑110͒, ͑100͒, and ͑101͒ surfaces, respectively. Density functional theory has been employed to assess the oxidation state and stability of different surface structures and compositions at various oxygen chemical potentials. The reduction of the SnO 2 surfaces is facilitated by the dual valency of Sn, and for all three surfaces a transition from Sn͑IV͒ to Sn͑II͒ is observed. For the ͑100͒ and ͑101͒ surfaces, theory supports the experimental observations that the phase transitions are accomplished by removal of bridging oxygen atoms from a stoichiometric SnO 2 surface, leaving a SnO surface layer with a 1 ϫ 1 periodicity. For the ͑110͒ surface the lowest energy surface under reducing conditions was predicted for a model with a SnO surface layer with all bridging oxygen and every second row of in-plane oxygen atoms removed. Ab initio atomistic thermodynamic calculations predict the phase transition conditions for the ͑101͒ surface, but there are significant differences with the experimentally observed transition temperatures for the ͑110͒ and ͑100͒ surfaces. This discrepancy between experiment and thermodynamic equilibrium calculations is likely because of a dominant role of kinetic processes in the experiment. The reduction of surface Sn atoms from a Sn͑IV͒ to a Sn͑II͒ valence state results in filling of the Sn-5s states and, consequently, the formation of Sn derived surface states for all three investigated surfaces. The dispersion of the surface states for the reduced ͑101͒ surface was determined and found to be in good agreement with the DFT results. For the ͑110͒ surface, the 4 ϫ 1 reconstruction that forms after sputter and annealing cycles was also investigated. For this surface, states that span almost the entire band gap were observed. Resonant photoemission spectroscopy identified all the surface states on the reduced SnO 2 surfaces as Sn derived.

Recombination is critically limiting in CdTe devices such as solar cells and detectors, with much of it occurring at or near the surface. In this work, we explore different routes to passivate p-type CdTe surfaces without any intentional extrinsic passivation layers. To provide deeper insight into the passivation routes, we uniquely correlate a set of characterization methods: surface analysis and time-resolved spectroscopy. We study two model systems: nominally undoped single crystals and large-grain polycrystalline films. We examine several strategies to reduce surface recombination velocity. First, we study the effects of removing surface contaminants while maintaining a near-stoichiometric surface. Then we examine stoichiometric thermally reconstructed surfaces. We also investigate the effects of shifting the surface stoichiometry by both “subtractive” (wet chemical etches) and “additive” (ampoule anneals and epitaxial growth) means. We consistently find for a variety of methods that a highly ordered stoichiometric to Cd-rich surface shows a significant reduction in surface recombination, whereas a Te-rich surface has high recombination and propose a mechanism to explain this. While as-received single crystals and as-deposited polycrystalline films have surface recombination velocities in the range of 105–106 cm/s, we find that several routes can reduce surface recombination velocities to <2.5 × 104 cm/s.

Efficient p-type doping in CdTe has remained a critical challenge for decades, limiting the performance of CdTe-based semiconductor devices. Arsenic is a promising p-type dopant; however, reproducible doping with high concentration is difficult and carrier lifetime is low. We systematically studied defect structures in As-doped CdTe using high-purity single crystal wafers to investigate the mechanisms that limit p-type doping. Two As-doped CdTe with varying acceptor density and two undoped CdTe were grown in Cd-rich and Te-rich environments. The defect structures were investigated by thermoelectric-effect spectroscopy (TEES), and first-principles calculations were used for identifying and assigning the experimentally observed defects. Measurements revealed activation of As is very low in both As-doped samples with very short lifetimes indicating strong compensation and the presence of significant carrier trapping defects. Defect studies suggest two acceptors and one donor level were introduced by As doping with activation energies at ~88 meV, ~293 meV and ~377 meV. In particular, the peak shown at ~162 K in the TEES spectra is very prominent in both As-doped samples, indicating a signature of AX-center donors. The AX-centers are believed to be responsible for most of the compensation because of their low formation energy and very prominent peak intensity in TEES spectra.

One of the main limiting factors in CdTe solar cells is its low p-type dopability and, consequently, low open-circuit voltage (VOC). We have systematically studied P and As doping in CdTe with first-principles calculations in order to understand how to increase the hole density. We find that both P and As p-type doping are self-compensated by the formation of AX centers. More importantly, we find that although high-temperature growth is beneficial to obtain high hole density, rapid cooling is necessary to sustain the hole density and to lower the Fermi level close to the valence band maximum (VBM) at room temperature. Thermodynamic simulations suggest that by cooling CdTe from a high growth temperature to room temperature under Te-poor conditions and choosing an optimal dopant concentration of about 1018/cm3, P and As doping can reach a hole density above 1017/cm3 at room temperature and lower the Fermi level to within ∼0.1 eV above the VBM. These results suggest a promising pathway to improve the VOC and efficiency of CdTe solar cells.