Bulk growth of cadmium mercury telluride (CMT) using the Bridgman/accelerated crucible rotation technique (ACRT)

Abstract

Cadmium mercury telluride (CMT, Cd x Hg 1− x Te) is still the pre-eminent infrared material, despite the difficulties associated with its production and subsequent processing. By varying the x value the material can be made to cover all the important infrared (IR) ranges of interest (i.e. 1–3, 3–5 and 8–14 μm). This paper will focus on the technological aspects of the production of large bulk crystals, their assessment and their use in photoconductive IR detectors. Bridgman growth takes place in simple two-zone furnaces with the pure elements contained in thick-walled high-purity silica ampoules. The thick ampoule walls are needed to contain the high (up to ∼70 atmos) mercury vapour pressures within the ampoules. We have found it necessary to purify both the mercury and the tellurium on site before use to obtain the required electrical properties. There is marked segregation of the matrix elements in standard Bridgman growth that is both a disadvantage and an advantage. Its disadvantage is that the yield of material in terms of composition for the most commonly required regions is low. The advantage is that all regions of interest can be produced in the same crystal, to some extent. A further advantage is that segregation of impurities also occurs and this leads to low background donor levels in Bridgman material. The other main disadvantages of the Bridgman technique are that the material is non-uniform in composition in the radial direction, as well as in the growth direction, resulting in low yields, and there are numerous grain and sub-grain boundaries. We have therefore developed an improved process based on the addition of the accelerated crucible rotation technique (ACRT). Here, ampoules are subjected to periodic acceleration/deceleration in their rotation, rather than constant rotation as in the Bridgman process. The major effect of this is to stir the melt during growth and produce flatter solid/liquid interfaces. This, in turn, improves the radial and axial compositional uniformity of the material, normally by a factor of at least ten-fold, and reduces somewhat the grain and sub-grain densities. The only drawback is that the long-wavelength material is now p-type as-grown and must be annealed in mercury vapour to convert it to n-type for use in photoconductive detectors, although the 1–3 and 3–5-μm material is n-type as-grown. An additional marked advantage of ACRT is that the improved radial compositional uniformity enables larger diameter material to be produced (up from 13 to 20 mm in our case). Assessment of the material includes wavelength mapping of both radially cut slices and axially cut planks; the latter gives useful information on the shape and change in the solid/liquid interface as growth proceeds. Hall-effect measurements are used routinely to characterise the material electrically. Images taken with an IR camera, and a blackbody source, reveal features in slices, e.g. cracks, inclusions of second phase and swirl patterns; the origin of the latter is as yet unknown. Some chemical analysis data from laser scan mass spectrometry (LSMS) for earlier 13-mm-diameter material and secondary ion mass spectrometry (SIMS) for later 20-mm-diameter material is also presented, which demonstrates the high-purity levels reached in this material. We are currently routinely growing 20-mm-diameter, 200-mm-long crystals of ∼0.5-kg weight with good uniformity of composition and good electrical properties for first-generation photoconductive IR detector programmes that require low carrier concentration n-type material. Compositions produced range from x∼0.7 (∼1.3 μm) through to x∼0.0 (i.e. HgTe), the latter in the tail ends of the crystals.