INTRODUCTIONThe II-VI narrow-gap Hg 1Ϫx Cd x Te single crystals are frequently investigated because of their applicability for the preparation of long-and middle-range infrared detectors. The quality of a final device strongly depends on the quality of each partial technological step. Dry etching (DE) is a widely used technological operation for structuring and removing a material in the semiconductor industry. In the case of p-type (HgCd)Te, Ar ion milling or reactive ion etching in Ar or H 2 /CH 4 plasma result in the formation of the n-type layer under the etched surface over relatively large depths (Ϸ100 m) at room temperature within short times (Ϸ10 min). 1-5 Very good quality p-n junctions were produced by DE, and high-performance IR photovoltaic detectors were fabricated. 6,7 Up to now, however, the determination of optimal etching conditions for formation of shallow p-n junctions with low surface damage, predefined thickness, and minimal lateral extension of the converted layer remains a technological problem.The model mostly used to explain the p-n conversion is based on an assumption that Hg interstitials (Hg I ) are liberated during DE by collisions of etching atoms with the surface, diffusing fast into the sample, 8 and recombining with Hg vacancies (V Hg ) or extrinsic acceptors (As, Au). 9-11 Consequently, residual uncompensated donors prevail and convert the sample into the n-type.The purpose of this paper is to make precise the preceding model in a more detailed form. We analyze the character and the properties of the surface source of Hg I especially and its influence on the broadening of the n-type layer. Coupled one-dimensional and two-dimensional equations describing diffusion of Hg I and their recombination with V Hg are solved to characterize the propagation of the p-n junction in the case of nonshielded and partly shielded surfaces. A seeming discrepancy between various experiments, which produce different time dependencies of the p-n junction depth (from square root to linear), being continuously reported in the literature, 1,3,12,13 is explained by the effect of the formation of the surface source of Hg I at the start of the etching and possibly by the density gradient of the acceptors. The ratio depth/lateral extension of the p-n junction under the shielding mask is calculated and analyzed as well.
MODELThe model of the type conversion of p-(HgCd)Te, assuming ultrafast diffusion of Hg I liberated at the etched surface and its recombination with V Hg , was accepted in the past as the most promising model to explain all reported experimental results. It appeared to be fruitful also in the case of extrinsicallyThe extended model describing the formation and propagation of a converted n-type layer in p-(HgCd)Te during dry etching (DE) based on the ultrafast diffusion of Hg interstitials and their recombination with Hg vacancies is presented. A couple of one-and two-dimensional equations are solved numerically to characterize the kinetics of the p-n junction. The time dependence of the p-n ju...