Molecular Beam Epitaxy Growth of High-Quality HgCdTe LWIR Layers on Polished and Repolished CdZnTe Substrates

Journal of Electronic Materials, Jun 2005 by Singh, R, Velicu, S, Crocco, J, Chang, Y, Et al

We report here molecular beam epitaxy (MBE) mercury cadmium telluride (HgCdTe) layers grown on polished and repolished substrates that showed state-of-the-art optical, structural, and electrical characteristics. Many polishing machines currently available do not take into account the soft semiconductor materials, CdZnTe (CZT) being one. Therefore, a polishing jig was custom designed and engineered to take in account certain physical parameters (pressure, substrate rotational frequency, drip rate of solution onto the polishing pad, and polishing pad rotational velocity). The control over these parameters increased the quality, uniformity, and the reproducibility of each polish. EPIR also investigated several bromine containing solutions used for polishing CZT. The concentration of bromine, as well as the mechanical parameters, was varied in order to determine the optimal conditions for polishing CZT.

Key words: HgCdTe, molecular beam epitaxy (MBE), CdZnTe, long wavelength infrared (LWIR), polished substrates

INTRODUCTION

Systems based on mercury cadmium telluride (HgCdTe) grown by molecular beam epitaxy (MBE) provide the highest performance thermal imaging detectors. These systems are made by depositing an epitaxial layer of HgCdTe on cadmium zinc telluride (CdZnTe or CZT) substrates by MBE. They dominate the market in military systems, in which they are critical components of unmanned aerial vehicles, precise targeting systems, and track and kill missile defense systems. They also are being used to identify and track targets at night, during low visibility weather conditions, and through obscurants such as dust or smoke. Currently, devices based on HgCdTe must be cooled using liquid nitrogen. The ability to provide devices capable of operating efficiently at higher temperatures is of particular interest for defense applications.

There also are mmmany potential nondefense applications, including among others sensors for chemical and biological toxin detection for homeland security and pollution monitoring, sensors for the early detection of structural fatigue and other failures at nuclear power plants, sensors for distant detection of gas leaks, imagers for cameras and the auto industry, and, perhaps most important, infrared (IR) and x-ray imaging for medical and dental applications. Improved IR medical imaging would greatly aid in surgery, the early detection of breast cancer and neonatal and infant disorders, location and treatment causes of pain, detection of vascular disorders, diagnosis of Raynaud's disease, the monitoring of effects of drug therapy, the measurement of allergies, etc.

Due to its nature, MBE demands much higher quality substrates than other techniques, especially for the growth of HgCdTe. Substrates epi-ready for the MBE growth of II-VI semiconductors, and especially for HgCdTe, are difficult to manufacture and polish. This difficulty is a major factor contributing to the high cost and hence limited deployment of systems based on MBE-grown HgCdTe. The ability to polish CZT substrates to an epi-ready state and to repolish MBE-grown HgCdTe/CdZnTe wafers that do not meet MBE specifications would be of critical importance both for reducing Defense Department costs and for making HgCdTe-based detectors viable for nondefense applications. Substrates that have been damaged, used for MBE calibration, etched for EPD measurements, etc., could be restored to pristine condition and reused for MBE growth at a cost far lower than that of purchasing new epi-ready substrates. Also, the CZT substrates on which HgCdTe detectors are grown are not fabricated at all in the United States. Currently, the technology for producing such high-quality surfaces on CZT substrates is available only at a single vendor in Japan, putting the U.S. military in a vulnerable position.

We report here on the repolishing of used or damaged CZT substrates. The softness of CdZnTe renders its polishing to an epi-ready state nontrivial. We have investigated combinations of different basic polishing techniques (hydroplane,1·2 noncontact, and pressure-dependent polishing) and measured the resultant substrate surface quality using several advanced characterization methods, including investigations of the quality of HgCdTe layers grown by MBE on the polished substrates.

It is generally accepted that in order for a substrate to be considered epi-ready, it must have a very low surface roughness, i.e.,

The final polishing procedure adopted consists of a three-step process: mechanical polishing, chemical polishing, and CMP. The mechanism for mechanical polishing involves the "scratching" away of material from the surface via abrasive particles in a slurry. Mechanical polishing eliminates any macroscopic roughness and creates a very flat, uniform reflective surface from which to start. However, the substrate surface is left rough after this step. Furthermore, it is generally accepted that the imparted subsurface damage after this step can reach as far as 3 times as deep as the diameter of the abrasive particle used. Therefore, mechanical polishing cannot serve as a final step.