Photoconductivity of transparent perovskite semiconductor BaSnO3 and SrTiO3 epitaxial thin films

Abstract

The perovskite semiconductor BaSnO3 (BSO) are attracting much attention recently due to high electron mobility and excellent oxygen stability. Since the existence of defect such as threading dislocation in BSO, the mobility of epitaxial thin film of BSO is suppressed and smaller than that of the single crystal of BSO. Since the photoconductive behavior changes according to the defect state in material, I have studied the photoconductivity in epitaxial thin film of BSO in order to investigate the defect states in the film. Because of the lack of reference in photoconductivity in BSO, I have chosen SrTiO3 (STO) for the comparative study due to its same perovskite structure with BSO and almost identical band gap size with that of BSO. I made the epitaxial thin films of BSO and STO using pulsed laser deposition technique. Although STO and BSO have the same perovskite structure and almost the same bandgap size, their photoconductive behavior is very different when UV light is illuminated. The photoconductivity in BSO thin film has much longer time component than that of STO. Due to the long time component, the magnitude of the BSO increased slowly so that it is more than 25 times higher than that of STO after 3 hours of illuminations and decreased very slowly when the illumination is removed. The different time constant of the photoconductivity in BSO and STO epitaxial film is due to the different properties of the defect states in BSO and STO films. The defect states in BSO film seems to consist of deep acceptor level so that the electrons trapped at the deep acceptor level excite to the conduction by illumination band, making hole traps which deter the recombination process. On the other hand, the defect states in STO film seems to have few acceptor state or very shallow acceptor levels if any. So there are few trapped electrons at the shallow level, making trapped holes at the shallow level by illumination, which are released easily by thermal excitation of electron from the valence band so that they can hardly contribute to a long time component of the photoconductivity. In the spectral responses the photocurrent in BSO starts to increase from sub-band gap energy while the photocurrent in STO starts to increase from band gap energy. This shows existence of deep level in BSO and non-existence of deep level in STO. The spectral responses also show that the band to band transition is the main mechanism of the photoconductivity in both BSO and STO film. I also studied the photoconductivity in BSO after annealing in O2 at 750 C to investigate the defect states after the annealing process. I annealed the sample in O2 at 750 C for 1 hour and cooled it slowly to the room temperature. I repeated this procedure for several times. After that, the fast increasing component of the photoconductivity increased greatly and the magnitude of the photoconductivity increased about 10 times. This probably due to increased mobility by increased crystallinity after annealing process. However, the decay curve has much slower component so that the current does not returned to its original dark current level even after very long time about 40 hours. Rather, it seems to stay at certain value about 1 pA. This long time component is probably due to the increase in the concentration of trapped electrons at the deep levels close to mid-gap as the crystal quality becomes better and the defect states, especially the deep levels close to the valence-band edge is reduced, even though the overall electron density is reduced by reduced oxygen vacancy. Furthermore, I have studied photoconductivity in BLSO 0.011% to investigate the electron doping effect on the defect states in BSO. The sample was subjected to the same annealing process in O2 gas to minimize the effect of oxygen vacancy on the photoconductivity in the sample. The magnitude of the photoconductivity increased very much and was about 400 times higher than that of undoped sample. The decay curve drops slowly but after 40 hours it seems to reach a current level about 2 pA and to remain at this level as well. The decay curve can be fitted by stretched exponential function with constant current term. The most probable time component of the probability function in the stretched exponential function is P F p3 %. However, since the constant current component has infinite time constant, there should be an explanation on it. As the electron density increases by La doping, the electrons are trapped at the deep acceptor levels so that the energy of the occupied deep acceptor level distributed broadly from the energy close to the valence-band edge to the energy close to the mid-gap energy. The increased electron density makes more deep acceptor level to be occupied so that the photoconductivity increased accordingly. The occupied deep level close to mid-gap also increases, so that the constant current component is created.