Schottky junctions on semi-insulating LEC gallium arsenide for X- and /spl gamma/-ray spectrometers operated at and below room temperature

Abstract

This work deals with the study of a Schottky junction used as an X- and /spl gamma/-ray detector in a spectrometer operated in the temperature range from -30/spl deg/C to +22/spl deg/C. The device (7 mm/sup 2/ active area and 100 /spl mu/m thickness), fabricated on liquid encapsulated Czochralski (LEG) semi-insulating Gallium Arsenide, is designed with a noninjecting ohmic contact which allows biasing voltages up to 550 V. At room temperature (22/spl deg/C) the energy resolution is found to be relatively poor (15.5-keV full-width at half-maximum (FWHM) at 59.5 keV) due to the large junction reverse current, whose density (7-37 nA/mm/sup 2/ at V/sub bias/=100-500 V) is within the typical values for Schottky junctions on SI LEC GaAs. By cooling of the detector to -30/spl deg/C, the noise of the reverse current is drastically lowered, thus achieving electronic noise levels around 160-180 rms electrons (1.6-1.8 keV FWHM), At 500-V bias, the /sup 241/Am spectrum has been resolved down to an energy of 4 keV with charge collection efficiency of cce=97% and a resolution of about 2-keV FWHM for the Np L lines and 2.4-keV FWHM for the 59.5-keV /spl gamma/ photons. The linearity of the detector has been measured to be better than /spl plusmn/0.6% within the explored energy range (14-59 keV). From the experimental spectra, it has been analyzed how either the electronic noise or the trapping of the signal charge contribute to the energy resolution of the spectrometer. The result is that despite the high measured cce. The trapping gives a contribution higher than 1.5 keV FWHM for the 59.5-keV spectral line. A comparison between the experimental results and Monte Carlo simulations, based on the Hecht model of charge trapping in detectors, is shown to give a satisfactory justification of the observed phenomena. A total mean drift length of carriers has been experimentally derived, finding an exponential dependence upon the bias voltage applied to the detector.