Linear x-ray detector array made on bulk CdZnTe for 30~100 keV energy

Journal of Electronic Materials, Jun 1997 by Yoo, S S, Jennings, G, Montano, P A

A high energy x-ray generator (Phillips MG-225) with a tungsten rotating anode target was used to measure the detector efficiency of a large area detector. Continuous x-ray energy was obtained by changing the diffraction angle of a (311) orientation single crystal Si monochromator. The measured x-ray photon energy varied between 30 to 100 keV. The x-ray beam size was small (2 x 2 mm^sup 2^) enough to avoid the radiation onto the edge of the detectors. A 5 cm thick NaI detector of 100% efficiency at these energies was used as a reference.15 For the x-ray energy scan measurements, the x-ray generator tube power was set to 120 kV and 1 mA and pulses were counted by a CAMAC based counter. However, with the high tube voltage, the third harmonic x-ray energy photons appeared in the low monochromatic x-ray energies.

The response of the linear array pixel elements were measured using a Co^sup 57^ source. Negative bias was applied to the bottom contact, so that the collected charges are mostly due to electrons. The radiation source was located 2 cm away from the linear array. For counting efficiency measurements, the radiation source was sealed with 5 mm thick lead blocks with a 1 mm diameter collimator.

RESULTS AND DISCUSSION

X-ray energy scan measurement was performed using a CdZnTe detector with a 1 x 1 cm^sup 2^ area, and 1.7 mm thickness and compared with measurement made using a NaI detector as shown in Fig. 2. Two peaks observed at 58.9 and 67.2 keV are K^sub a^ and K^sub beta^ x-rays of the tungsten rotating anode target from both detectors. A slight increase of the count rate in the low energy is observed for the NaI detector. This is due to the increased number of third harmonic energy photons by diffraction from the Si crystal at energies below 40 keV. Because of the low absorption rate at the third harmonic x-ray photon energies, the CdZnTe detectors do not show such an increase in counts. It is also noted that, for the CdZnTe detector, relatively low count rate is observed in the low energy range. This is due to the absorption of the low energy x-ray photons by the conductive rubber contact placed in the radiation side of the CdZnTe detector, which contains a significant amount of silver. However, the attenuation within the rubber contact seems to be negligible as the energy increases. Thus, by ignoring the attenuation in the high energy, the overall CdZnTe detector counting efficiency was determined by taking the ratio between the counts of the CdZnTe and the NaI detector. We obtained a count ratio as high as 90% at 60 keV from the energy scan measurements. However, the ratio slowly decreases with further increase in x-ray photon energy, and reaches 75% at 100 keV. The calculated x-ray photon absorption efficiency for the 1.7 mm thick CdZnTe detector is 100% at 60 keV and decreases to 80% at 100 keV. This result illustrates that the counting efficiency of the CdZnTe detector reaches over 90% compared with the theoretical expectation.

Dark current-voltage measurements at 300K were carried out before and after the fabrication process, and the results are shown in Fig. 3. As a reference, current-voltage measurement was made for a large area detector before the fabrication of the linear array detectors. The measured dark current was as low as 30 nA at 100 V bias, and the corresponding resistivity was calculated to be 3.3 x10^sup 10^ Wcm. Three linear arrays of different thickness were separately fabricated from the same sample. One of the CdZnTe samples was chemically etched in a standard 5% Br^sub 2-^ methanol solution for 5 min. The others were mechanically thinned using 0.5 (mu)m size alumina polishing powder, then followed by chemical etching in the same solution, but for shorter time. The estimated thickness of the etched layers is 20~30 (mu)m. The effect of the damage on the surface by less etching is clearly seen from the current-voltage measurement of both arrays of less chemical etching time. The calculated resistivity was nearly one order of magnitude less than the first one. It seems that the surface damaged region created during the grind polishing is deeper than the etched layer thickness. However, the array made by chemical etching shows nearly the same resistivity as that of the large area detector.