Fundamental nature of deep defects in hydrogenated amorphous silicon revealed by electron spin resonance and modulated photocurrent measurements

• Main Author: Kwon, Daewon
• Format: Dissertation
• Language: English
• Published: ProQuest Dissertations & Theses 01-01-1997
• Subjects: Condensation; Condensed matter physics; Materials science
• ISBN: 9780591489392; 0591489392

Modulated photocurrent (MPC) measurements in intrinsic hydrogenated amorphous silicon (a-Si:H) have revealed a dominant deep band of electron traps with a thermal emission energy near 0.6 eV. The MPC signal results from the capture and re-emission of photo-generated electrons. Therefore, the MPC band might result either from $\rm D\sp0/D\sp+$ or from $\rm D\sp-/D\sp0$ transitions, where D represents the dominant deep defect in a-Si:H which has three charge states ($\rm D\sp+,\ D\sp0,\ D\sp-)$. Proponents of the recently popular "defect pool model" have assigned a $\rm D\sp0/D\sp+$ transition to this band. In our study, the defect band has been identified by a direct comparison between the magnitudes of defect densities revealed from the MPC spectra with the spin densities measured by electron spin resonance (ESR). Measurements were carried out both for an intrinsic and for a lightly n-type doped a-Si:H samples for a variety of metastable states. In the various metastable states the Fermi level, E$\sb{\rm F}$, varies from less than 0.5 eV to 0.73 eV below the conduction band mobility edge, E$\sb{\rm C}$. The comparison between MPC and ESR clearly identifies that the MPC band arises from the $\rm D\sp-/D\sp0$ transition of the D defect. In intrinsic a-Si:H a thermal energy shift of the MPC spectra peak by more than 0.1 eV was also observed in response to applied light bias. However, the magnitude of the defect band decreased by only 25%. The ESR measurements under the same light bias conditions resulted in a decrease of defect spin density by less than 5%. Both the standard defect model for a-Si:H as well as the defect pool model fail to explain these results. However, within a defect relaxation picture, the network has an ability to re-adjust the electronic energy of the defect in order to maintain a minimum total system energy. Therefore, our results are most readily explained by such a defect relaxation model.