Improved device technology for epitaxial Hg1-xCdxTe infrared photoconductor arrays

The performance of Hg1-xCdxTe infrared photoconductors is strongly dependent on the semiconductor surface conditions and, in particular, the degree to which the surface contributes to recombination of photogenerated excess carriers. Although published photoconductor fabrication processes based on bulk Hg1-xCdxTe address this issue by fully passivating both major surfaces (i.e. front and back) with anodically grown native oxide, passivation of the sidewalls is neglected. In this paper it is shown both theoretically and experimentally that leaving the sidewalls unpassivated can result in approximately a factor of two reduction in responsivity for long-wavelength infrared (LWIR) detectors used in high-resolution thermal imaging systems. Detector arrays are typically fabricated on x=0.23 Hg1-xCdxTe representing a cut-off wavelength of 9.4 mu m and use individual element sizes of approximately 50*50 mu m2. We describe in detail for the first time a device technology which enables the fabrication of Hg1-xCdxTe photoconductor arrays such that the entire surface of the semiconductor is effectively passivated, including the sidewalls. Of particular interest is the fact that this improved device technology is compatible with present-day Hg1-xCdxTe epitaxial growth processes. This is in contrast to current photoconductor technology which is primarily based on bulk Hg1-xCdxTe. Experimental results are presented which compare device performance of LWIR detectors fabricated using the improved photoconductor technology with current published photoconductor technology. These results clearly indicate that detectors fabricated on liquid phase epitaxially (LPE) grown x=0.23 Hg1-xCdxTe material using the improved photoconductor device technology achieve much higher responsivities and detectivities. Furthermore, it is shown that only a fully passivated device structure is capable of exploiting any future improvements in bulk minority carrier lifetime as it approaches the Auger recombination limit.