Solid-SWP Elecrronic.~Vol. 38, No. 5, PP. 989-995, 1995 zyxwvutsrqpo 0038-1101(94)00216-9 Copyright Q 1995 Elsevier Science zyxwvutsrqpo Ltd Printed in G re a t Brita in. All rig hts re se rve d 0038-I lOi/ -$9.50 + 0.00 Pergamon A METHOD FOR THE DETERMINATION OF BARRIER HEIGHTS FROM THE CAPACITANCE-VOLTAGE CHARACTERISTICS OF A SCHOTTKY JUNCTION CONTAINING BULK DEEP TRAPS A. COLA’, L. VASANELLI’ and P. MURET’ ‘C.N.R., Istituto Materiali per I’Elettronica, Via Arnesano 73 100, Lecce, Italy and ‘C.N.R.S., Laboratoire d’Etudes et des Propriktbs Electroniques des Solides. P.B. 166, 38042 Grenoble Cedex. France zyxwvutsrqponmlkjih (Received 6 June 1994: in revised ,form 29 September 1994) Abstract-In Schottky barrier diodes, appreciable concentrations of deep levels in the space charge region influence high frequency capacitance measurements. The barrier height, calculated by the usual analysis of l/C’ vs V plot, exhibits large steps at temperatures corresponding to the relaxation of the traps occurring during the voltage sweeps. In this work we develop a model which takes into account the partial ionization of the deep levels. to determine the correct parameters of the Schottky barrier by temperature dependent capacitance+voltage measurements. The influence of two kinds of bulk traps, already detected by DLTS. is clearly evidenced in the l/C’ vs V intercept and taken into account for the calculation of the true barrier height and trap concentrations. The correlation coefficient of the linear regression is very significantly improved if this model is used. The analysis is applied to Ti/GaAs Schottky diode prepared by ion beam sputtering. 1. INTRODUCTION It has been previously shown that special analyses are required for the interpretation of the Schottky barrier capacitance measurements on semiconductors con- taining deep traps[l-61. The linearity of the l/C* vs V plots is strongly affected by the presence of deep levels, invaliding the usual analysis which neglects the contribution of traps to calculate the Schottky barrier height and the shallow dopant concentration[6-91. This is true even in high frequency capacitance measurements, where the deep levels can not be in equilibrium with the test signal. In current GaAs technology, appreciable concentration of native traps, as the well-known EL2 level, are expected. Moreover, during the technological processes for the preparation of the device (metallization, etching, etc.), defects can be created near the metal- semiconductor interface[ 1 O-1 31. Lateral barrier non- uniformities also give rise to non-idealities[l4], but in such case, a greater effect on the current than on the capacitance is expected. In these cases the ideal Schottky barrier properties are altered, thus making more complicate the determination of the barrier height, which is the most important parameter in Schottky diodes. These difficulties become evident when one attempts to compare the results obtained by different techniques, such as current-voltage, capacitance-voltage and internal photo-emission[ 15.161. In this paper, we analyze the influence of appreci- able concentrations (i.e. same order of magnitude as the shallow level) of deep levels uniformly dis- tributed in the space charge region, on the calculation of the barrier height from temperature dependent capacitance-voltage measurements. This study has been carried on Ti/GaAs Schottky barriers, where previous DLTS measurements evidenced the presence of two deep levels[l7]. We will discuss the results of this analysis of C-V data measured over a wide range of temperature. starting from a model based on the abrupt space charge region approximation in the case of high trap concentration, taking into account the relaxation of the two deep traps during the voltage biasing. The model can be applied to a semiconductor containing one or two traps. A suitable measurement methodology has been developed, in order to com- pare meaningfully the outcome of the model with the experimental data. As a result, we have obtained the correct dopant and deep levels concentrations and the temperature dependence of the barrier height. Finally, the barrier height values have been compared with those calculated from current-forward voltage measurements. 2. EXPERIMENTAL The capacitance measurements were performed using an HP-4192A LCR-meter, at the signal fre- quency 1 MHz, having verified that at this frequency the measurements were not affected by the series resistance or the deep levels. The current measure- ments were performed using an HP-414OB PA-meter. The diode was placed in a liquid nitrogen cryostat; 989