Modelling the MOS Device Conductance Using an Extended Tunnelling Model and Subsequent Determination of Interface Traps.
- PDF / 142,910 Bytes
- 6 Pages / 612 x 792 pts (letter) Page_size
- 70 Downloads / 310 Views
E4.25.1
Modelling the MOS Device Conductance Using an Extended Tunnelling Model and Subsequent Determination of Interface Traps. N.Konofaos Computer Engineering and Informatics Dept. & Research Academic Computer Technology Institute, Building B University Campus, Patras, Greece. email: [email protected] ABSTRACT. The contribution of the AC conductance on admittance spectroscopy measurements on Metal-Insulator-Semiconductor (MIS) devices allows the calculation of the interface traps density and the relevant time constant. Two equivalent model approaches can be applied in order to explain the experimental results. One model is the statistical model based on the ShockleyRead-Hall (S-R-H) recombination statistics and the other one is a model based on the quantum tunneling effect. Recent evidence suggest that the tunneling model can be equivalent to the statistical one if a continuum of states is also considered in the modeling, creating an extended tunneling model. In the present paper, a further investigation on the extended model is attempted. Admittance spectroscopy data were collected for various Insulator/Si combinations, such as SrTiO3 and BaTiO3 deposited on Si. Both the S-R-H based and the extended tunneling models were used to analyze the data. The results showed that the extended tunneling can model the conductance of the device successfully and can calculate the interface states density and the traps time constant. INTRODUCTION The performance of Metal-Insulator-Semiconductor (MIS) devices is strongly affected by the presence of traps located at the insulator/semiconductor interface. Various electrical techniques have been proposed in order to identify those states [1-4] relying on the application of an electrical signal across the device. The interface states can be identified by both their capacitive and conductive response to the applied signal. However, extraction of the density of the states from the capacitance cannot derive the traps time constant and usually results in values slightly higher than those provided by the conductance [2]. The conductive response of the traps is directly related to the traps time constant as proposed by Nicollian and Goetzberger and can be used to reveal both the traps density and their time constant [1,2]. Analysing the conductance data though is not an easy task and different models have been proposed, namely a) the discrete, b) the continuum c) the statistical and d) the tunnelling model [1,2,3]. They can all derive the density of interface states (Dit) and the time constant (τ) by connecting the measured parallel conductance versus frequency (Gp vs ω) to Dit and τ. Among the four models, there is not a universally accepted one suitable to fit all the data obtained for various insulator/semiconductor structures. It is well known that the statistical model has been successfully used to fit data obtained from MIS devices where the insulator materials were SiO2 and BaTiO3 [1-4]. However, different models were used in order to analyse the
E4.25.2
experimental dat
Data Loading...
