| Photovoltaic cells having the structures CuInSe2-CdO and CuInSe 2-CdS-CdO have been fabricated on a cut or cleaved p-type Bridgman-grown CuInSe2 ingot. The CdO window layer in both cases was deposited by dc reactive sputtering. In the CuInSe2-CdS-CdO structures, the CdS interlayer was deposited by a solution growth method and these cells, incorporating the CdS interlayer, yielded a conversion efficiency of about 6.9%. By comparison, the cell with the structure CuInSe2-CdO, where the lattice mismatch between the CuInSe2 and the CdO is some 23%, showed a surprising efficiency of 6.8%. This result, indicating that the defect centers at the CuInSe2-CdO interface arising from the large lattice mismatch, play a smaller role than expected, suggests that the seat of photovoltaic action may lie away from the substrate-window layer interface as a buried homojunction within the CuInSe2. While the conversion efficiencies Of the two kinds of cells were similar, there were slight differences in their other characteristics whereby the device without the CdS interlayer had a higher illuminated open circuit voltage, lower short circuit current density, higher forward ideality factor above 2 and higher internal series resistance. The acceptor concentration of the stoichiometrically-prepared CuInSe 2 was estimated to be about 1017 cm-3 in the bulk of the wafers but almost an order of magnitude smaller than this near the surface after annealing in argon. Using the photocurrent-capacitance method, minority electron diffusion lengths in the CuInSe2 were estimated to be about 3 mum for the devices made with cleaved substrates and about half this value for the cells made with abrasively-polished substrates. |
