We have performed electronic structure calculations of the interaction of potassium and oxygen with graphite (GR), individually and as coadsorbates. We use up to three graphite planes to represent the graphite surface, but we show that the main physics is correctly described by a single graphite layer. At low coverage the potassium–graphite bond is largely ionic, and the variation of the K–GR bond energy with the lateral position of the K atom in the graphite unit cell is very small. We study the interaction between atomic oxygen and graphite. We find that O binds strongest at the bridge site, but the barrier for diffusion is rather small. The frequency for the perpendicular O–graphite vibrational mode is remarkably low (53 meV), reflecting the relative slow variation of the O–graphite interaction energy with the separation $z$ between the O atom and the graphite surface. We consider the adsorption of ${\mathrm{O}}_{\mathrm{2}}$ on a clean graphite surface and on a graphite surface with a low concentration of potassium. On the clean surface the ${\mathrm{O}}_{\mathrm{2}}$ –graphite interaction is found to be repulsive (the weak attractive van der Waals interaction is not included in our theoretical method), in accordance with the extremely low sticking coefficient observed for ${\mathrm{O}}_{\mathrm{2}}$ on clean graphite. When potassium is adsorbed on the graphite surface, ${\mathrm{O}}_{\mathrm{2}}$ chemisorbs at the potassium sites which is consistent with the large sticking coefficient observed for ${\mathrm{O}}_{\mathrm{2}}$ on a potassium covered surface. The energy barrier towards dissociation of ${\mathrm{O}}_{\mathrm{2}}$ on the clean graphite surface is estimated to be similar to that of gas phase ${\mathrm{O}}_{\mathrm{2}}.$ For ${\mathrm{O}}_{\mathrm{2}}$ on K/graphite we find that ${\mathrm{O}}_{\mathrm{2}}$ chemisorbs “side-on” K, and that the barrier for dissociation is much smaller than in the gas phase or on the clean graphite surface.

Chalcopyrite semiconductors are of considerable interest for application as absorber layers in thin-film photovoltaic cells. When growing films of these compounds, however, they are often found to contain CuAu-like domains, a metastable phase of chalcopyrite. It has been reported that for CuInS₂, the presence of the CuAu-like phase improves the short circuit current of the chalcopyrite-based photovoltaic cell. We investigate the thermodynamic stability of both phases for a selected list of I–III–VI₂ materials using a first-principles density functional theory approach. For the CuIn–VI₂ compounds, the difference in formation energy between the chalcopyrite and CuAu-like phase is found to be close to 2 meV per atom, indicating a high likelihood of the presence of CuAu-like domains. Next, we calculate the spectroscopic limited maximum efficiency (SLME) of the CuAu-like phase and compare the results with those of the corresponding chalcopyrite phase. We identify several candidates with a high efficiency, such as CuAu-like CuInS₂, for which we obtain an SLME of 29% at a thickness of 500 nm. We observe that the SLME can have values above the Shockley–Queisser (SQ) limit, and show that this can occur because the SQ limit assumes the absorptivity to be a step function, thus overestimating the radiative recombination in the detailed balance approach. This means that it is possible to find higher theoretical efficiencies within this framework simply by calculating the J–V characteristic with an absorption spectrum. Finally, we expand our SLME analysis to indirect band gap absorbers by studying silicon, and find that the SLME quickly overestimates the reverse saturation current of indirect band gap materials, drastically lowering their calculated efficiency.