Improving the Accuracy in Determining the Insulator Capacitance in Metal–Insulator–Semiconductor Structures

Abstract

The results of the voltage–capacitance spectroscopy of interface states in metal–insulator–semiconductor (MIS) structures are critical functions of the accuracy in determining the insulator capacitance C _i, which is typically no higher than a few fractions of a percent. This substantially limits the energy range of the observed spectrum of the interface states (ΔE ≅ 0.5 eV for Si-based MIS structures) and the sensitivity to the density of the interface states at the spectrum edges (ΔN _ss ≅ 1 × 10¹⁰ cm^–2 eV^–1). We propose a method for minimizing these errors that is based on a sequential variation of the initial estimate C _i ≅ C _{i 0} → C _ij, j = 0, 1, 2, ... and the identification of singular points in the dependences $\overline {\delta V_g^{ac,in} }$ and $\overline \delta _{ac,in}$ on C _ij, where $\overline {\delta V_g^{ac,in} }$ are the mean arithmetic values of the voltage difference between the experimental and ideal voltage–capacitance characteristic and $\overline \delta _{ac,in}$ are the rms deviations of the voltage $\overline {\delta V_g^{ac,in} }$ values taken in the high-accumulation (ac) and inversion (in) regions from $\delta V_g^{ac,in}$ values. The highest (≅10^–4%) accuracy in determining C _i is achieved in the regions of the equidistant experimental and ideal voltage–capacitance characteristic. This method, combined with the technique of Ψ _s ^′/Ψ _s diagrams, ensures an extension of ΔE to ≅0.9 eV at ΔN _ss ≅ 1 × 10¹⁰ cm^–2 eV^–1 and the possibility of determining the sign and density of the fixed charge in the gate insulator.