The Origin of Slow States at the Interface of a-Si:H and Silicon Nitride
- PDF / 361,780 Bytes
- 6 Pages / 420.48 x 639 pts Page_size
- 52 Downloads / 198 Views
THE ORIGIN OF SLOW STATES AT THE INTERFACE OF a-Si:H AND SILICON NITRIDE R. A. STREET and C. C. TSAI Xerox Palo Alto Research Center Palo Alto, California 94304 U.S.A. ABSTRACT Transient photoconductivity is used to investigate the origin of slow states near the interface of a-Si:H and silicon nitride. A graded composition of the nitride layer near the interface greatly increases the density of slow states. We deduce that slow states are bulk nitride traps and that the magnitude of charge storage is largely determined by the composition dependence of the localization radius of electrons within these traps. The kinetics of charge storage and release are found to be very different and are interpreted in terms of an activation step at the interface. INTRODUCTION The interface between hydrogenated amorphous silicon (a-Si:H) and silicon nitride contains excess localized electronic states. In general there is charge transfer within the states near the interface, causing band bending and charge accumulation [1,2]. For the a-Si:H/nitride interface, the band bending is such that electrons accumulate in the a-Si:H. The charge density is found to depend on the order of deposition, being much larger for nitride films grown on a-Si:H (top nitride) than vice versa [1]. When the interface is biassed, charge will either be removed or added to the interface states. The states can be characterized as fast or slow, depending on the time required for the charge state to change. Such a distinction is not very precise, since it refers to the specific time of the measurement, and in practice the response times vary over a very wide range. Previous measurements have shown that some electrons are removed fom the interface in a time of - 10-7 sec at 300 K [1]. On the other hand, slow relaxation of transistor characteristics with a time constant of hours is also observed and attributed to charge release [3,4]. In this paper, we are mostly concerned with the origin of the slow states with time constants in excess of about 1 sec. There are two mechanisms that can account for a slow release time of carriers. One mechanism is that a large activation energy is required to excite a carrier from a localized state, and the second is that the carrier must tunnel from a localized state within the nitride to the semiconductor. The combined effects give a release rate of PR of PR = coo exp (-2 R/Ro)exp(-E/kT)
(1)
where R0 is the tunnelling radius, R is the distance to the interface, and E is the relevant activation energy. Eq. 1 applies to a single tunnelling event. A release mechanism that depends on-multiple hopping of carriers to the interface will have a different release rate, although with qualitatively the same dependence on R/R0 and E,except that R is the average distance between hops. It was originally proposed that slow states were electrons trapped deep in the nitride [3,4], so that the long time constant originated largely from the tunnelling term. Recently, Hepburn et al. have proposed instead that the charge is at the interface, and the
Data Loading...
