A molecular beam epitaxy (MBE) system designed for the growth of Hg _l-xCd_xTe alloys is described. The system is equipped with both binary (CdTe) and elemental Hg, Cd, and Te sources and has been used to grow epitaxial layers of CdTe and Hg _1-xCd_xTe with x-value between 0.9 and 0.5 on (111) orientated CdTe substrates. The growth of CdTe on (100) orientated GaAs and 1nP substrates is also reported.

Single-crystal films up to 1 inch diameter of mercury cadmium telluride have been grown on mica substrates using a combination of two techniques. Initially, a film of CdTe was deposited on the substrate in a Hot Wall Epitaxial furnace. In a subsequent step the CdTe film is converted to HgCdTe by an evaporation and diffusion of HgTe at constant tempera-ture. The films were very uniform and had device quality electrical properties. Attempts to produce similar films on 2-inch diameter sapphire substrates resulted in polycrystalline growth.

The refractive index n of the ternary compound Cd_xHg_l-xTe is calculated as a function of frequency and mole fraction x in terms of known experimental parameters. The theoretical result for n is obtained from a quantum mechanical calculation of the dielectric constant of a compound semiconductor, which has been successfully applied to a number of III V and II VI binary and ternary compounds for which experimental data are available. It is given in terms of basic material parameters only, with no adjustable constants. These material parameters consist of band gap energy, effective electron mass and effective heavy hole mass at the band edge, the spin-orbit splitting energy, the lattice constant and the carrier concentration for n-type or p-type materials. If these parameters are known as functions of mole fraction x, the refractive index is completely determined as a function of frequency below the fundamental absorption edge.

A large number of (Hg_.8Cd_.2)Te films, about ten microns thick, were prepared on low-cost Si substrates by r.f..triode-sputtering in a Hg atmosphere. The sensitivity of the film conduction properties to small changes in snutter-deposition parameters (particularly Hg sputtering gas pressure and substrate dc bias and temnerature) and to post deposition annealina parameters was studied by Vanderpauw Hall effect measurements variable tempera-ture and magnetic field. Wavelength dispersive electron-probe microanalysis and optical absorption analysis were used to measure the film composition which under the proper souttering conditions matched the nominal composition value of the pressed-powder target in the 0.2-0.27 x value range. Substrate bias to remove impurities depositing during the sputter-ing process was found to be effective for obtaining n-type films with carrier concentrations as low as 10¹⁵ cm^-3 at 1000^oK. Although the electron mobilities were only about 10% of bulk values, the experimental results indicate that considerable improvement can be obtained by further fine tuning of the deposiiton and annealing parameters.

As in most other aspects of its behavior, the surfaces and interfaces of Hg_l-xCd_xTe are anomalous when compared to the bahavior of other better-understood semiconductors such as Si and the 3-5s. The thrust of this paper will be, first, to outline and, as far as possible, to document these differences. Second, we will make tentative suggestions as to the reason for this difference, indicating the data and reasoning underlying these suggestions. This leads inescapably to the third and very important part of this paper-an attempt to identify the key areas in which we lack critical knowledge and to suggest approaches that may produce such knowledge. Here, as elsewhere in this paper, we attempt to indicate how other work reported at this meeting fit into the overall picture which is beginning to emerge. The fourth section of this paper will give an overview of the passivating overlayers placed on the surface of Hg_l-xCd_xTe and the results obtained from them. It will also attempt to relate these results with the thoughts, questions, and concepts developed earlier in this paper.

Based on the classical theory of epitaxial crystal growth, the misfit dislocations and dangling bond densities of abrupt (111) Hg _{l- x}Cd_xTe heterojunctions have been calculated. It is assumed that the misfit between layers with compositions xi and x2 is accommodated by edge dislocations lying along the <111> directions. This is in agreement with recent experiments on the growth of Hg_l-xCd_xTe epitaxial layers. For the case where (x_{2 -x}1) >0.1 the dangling bond density is on the order of 10¹¹cm^-2. Such large dangling bond densities may produce high interface recombination velocities or band-bending at the interface.

The chemistry of the anodic oxide of Hg_l-xCd_xTe is reviewed. The oxide is generally believed to contain little Hg and primarily consist of CdTeO₃ with smaller amounts of HgTe0₃, CdTe20₅, HgTe20₅, or Te0₂. Consistent with the weakening Hg-Te bond strength for lower x value material, a severely defective interface between the oxide and the substrate is reported for Hg0_.8Cd_0.2Te, but not for Hg_0.7Cd_0.3Te. The detailed nature of these defects, however, remains open to debate. Models include a Hg-depleted layer in the semiconductor and an inhomogeneous oxide layer near the interface.

The band structure as well as electrical and optical properties of Hg_l-xMn_xTe are remarkably similar to those of Hg _l-xCd_xTe, making Hg_l-xMn_xTe a competitive candidate for applications in infrared detectors. In this paper we examine the properties of Hg_1-xMn_xTe which are relevant to this application. Hg_l-xMn_xTe is formed by substitution of tie magnetic Mn ions for Hg in the HgTe lattice; although Mn is not a group II element, Hg_l-xMn_xTe crystallizes in the zincblende structure, forming good quality crystals up to x 0.3S. Its energy gap and related band parameters vary with x, but at a rate about twice as fast as in Hg_l-xCd_xTe. The electrical properties of Hg_1-xMn_xTe are again similar to those of Hg_l-xCd_xTe, including electronic mobilities and doping characteristics. As-grown Hg_l-x Mn_xTe is p-type due to a natural tendency to form Hg-vacancies, which act as acceptors and whose concentration can be reduced by appropriate annealing. Because Mn is a magnetic ion, Hg _l- differs from Hg_l-xCd_xTe in its magnetic properties, as well as in the behavior of its electrical and optical properties in the presence of a magnetic field. For example, for x > 0.17 Hg_1-xMn_xTe exhibits a transition to the spin glass phase at low temperatures. Furthermore, the presence of Mn ions leads to an exchange interaction between the localized magnetic moments and the band electrons, which in turn affects the band parameters and leads to new and spectacular effects in the transport and optical properties. It is to be emphasized that magnetic properties associated with the Mn ions, such as electron paramagnetic resonance, magnetic susceptibility, etc., also offer a unique handle by which the distribution of Mn over the lattice (e.g., its tendency to form clusters) can be studied.

The present performance status of silicon IR detectors, including spectral response, detectivity and operational temperature for use in the 2-2.5, 3-5 and 8-14 pm principal windows of the atmosphere, are summarized and discussed. In general, all of the detectors are background radiation noise limited below some temperature which is a function of the background flux density. This temperature is tabulated for the detectors for a background provided by a 30° field of view of the 300K background. These temperatures are in general lower than required for equivalent intrinsic detectors, but the special properties of silicon when they are mated to CCDs, particularly for the longer wavelengths, including the electrical impedance and the character of the noise may offset the temperature advantage of the intrinsics when used in imaging arrays.

Infrared systems applications often require low noise detectors to achieve background limited performance. In this paper, the dark current (noise) mechanisms of InSb MIS detectors will be discussed from a theoretical viewpoint and a model of the dark current phenomena will be presented. Device measurements will be presented to qualify the model. The total well capacity of an MIS detector is determined by the insulator capacitance and the height of the applied voltage pulse that switches the device into deep depletion (Q = C_oxV_I. As V_I is increased beyond some threshold value, the dark current is seen to increase more rapidly, leading to excess detector noise. This "breakdown" is attributed to carrier tunneling from states at the interface into the conduction band. This is due to the higher fields associated with increasing the voltage pulse height. A model that qualitatively describes this phenomenon is presented along with experimental data for InSb devices. InSb MIS infrared detectors integrate minority carriers that are generated in the potential well beneath a transparent metal gate (see Figure 1). This integrated charge is read out by collapsing the potential well, which causes the collected carriers to recombine in (or be injected into) the substrate, thereby creating a current in an external circuit. The well is then reestablished, initiating charge integration. This is the principle of the Charge Injection Device (CID).1 The collected charge includes both photogenerated carriers (which arise from incident radiation) and dark current. This paper is the result of a study of the physical mechanisms that contribute to the generation of dark carriers in CIDs which will ultimately limit the detector's performance.

Schottky barrier infrared detectors have made significant advances in the past several years. These have occurred on many fronts, and have resulted in focal planes which can image in the staring mode with more than 8000 active elements. Initial experimental results of the early 1970's were done with Palladium Silicide detectors. Images made from this material have a cutoff wavelength of about 3.5 micrometers and can adequately cover the near infrared bands with moderate quantum efficiency. In the mid to late 1970's, it was found that PtSi could be formed into infrared sensing arrays with a cutoff wavelength of about 4.7 micrometers. Unfortunately, the early arrays did not cover the region from 4.7 to 5.2 micrometers and suffered from rather low quantum efficiencies. Improvements in processing have overcome both of these initial difficiencies. In fact, PtSi arrays now have cutoff wavelengths of nearly 6.0 micrometers and moderate quantum efficiencies from 3.0 to 5.0 micrometers. They have been measured to be background limited at f/3 and have fixed pattern fingerprints of less than 0.3%. The minimum resolvable temperature at f/3 is less than 0.3°C. Our latest experiments have concentrated in IrSi detectors. These devices have cutoff wavelengths beyond 9.0 micrometers, and have excellent quantum yields in the 3.0 to 5.0 micro-meter region.

Two-dimensional arrays of a novel infrared detector, the Charge Imaging Matrix, (CIM), based on HgCdTe charge transfer devices have been demonstrated on p-type HgCdTe of 0.13 eV and 0.25 eV bandgap at 78K. The purpose was to demonstrate a detector for scanned infrared focal plane applications. A principle feature of a CIM is that it can be operated at high readout rates with off-chip TDI (time-delay-integration) using silicon circuitry and thus avoiding the excessive well capacity requirements of CCDs at long wave-lengths. A CIM pixel has three principle elements: (1) a metal-insulator-semiconductor (MIS) detector with semitransparent metal gate for infrared detection separated from (2) a sense diode by (3) an opaque MIS transfer gate for charge transfer control. CIMs can be operated at high readout rates because their operation does not require charge injection as does CID operation. Instead, the charge is extracted from the sense diode during the preset operation allowing data rates to exceed one megahertz. Arrays with up to 9 x 8 pixels on 76.2 um centers were fabricated on material with hole concentrations of 1-5 x 10¹⁵ cm^-3 at 78K using processes similar to those used to fabri-cate HgCdTe CCDs and CIDs [1]. Diodes were formed by boron ion implantation. CIM devices with 4-5 pm and 9-10 pm cutoff wavelengths were tested at 78K for operability, spectral response, inter-pixel crosstalk, and detectivity. Their design, fabrication, and electro-optical tests will be discussed.

This paper is a summary of work done in the development of monolithic lead salt-silicon infrared focal plane technology. The photoconductive detector materials, PbS and PbSe are chemically deposited onto premetallized silicon MOSFET integrated circuit wafers. A variety of structures based on an implanted PMOS process were fabricated and evaluated. Operational results of an eight-element PbS array multiplexed on-chip are presented along with radiometric measurements on other integrated PbS-silicon MOSFET structures. PbS imagery is shown using one element of a 20-element array integrated on-chip.

Multispectral response photovoltaic lead-salt infrared (IR) detectors have been fabricated with a high degree of reliability. These detectors are made with metallic lead junction contacts on multilayer P-type PbS, PbS_0.5Se_0.5' and PbSe epitaxial films which have been chlorine-treated by baking in an air oven together with PbC1,. Detectors with nearly BLIP operation at liquid nitrogen temperature have been fabricated. Further improvement on the optical performance of these detectors can be achieved by refinements in the fabrication process.

The different types of state-of-the-art hybrid structures for photovoltaic focal plane arrays are subject to inherent fabrication problems. The island technology used for front-side illumination presents problems caused by the thinning and bonding of the photodetector material and by the forming of the connections. Devices for backside illumination are difficult to produce and very much dependent on the quality of the CdTe substrate. A new device which consists of a symmetrical front-and-back structure permits detection to take place on one face and electrical connection on the other. Its fabrication draws on classical HgCdTe array fabrication processes. This new symmetrical two-face structure yields diodes whose characteristics and manufacturing yields are entirely comparable to those obtained for conventional HgCdTe arrays.

In the past two years, IRCCD staring sensor technology has advanced to where, except for resolution, performance is comparable with scanning sensor technology. In this paper, we discuss the signal processing requirements for Schottky IRCCD sensors and describe a real time digital Schottky IRCCD image processor. This processor conditions the raw Schottky IRCCD data to standard composite video output and provides first order correction of sensor nonuniformity, optical shading and background suppression. The processor design is based on simple low cost circuitry which can be expanded to meet the support requirements for future Schottky IRCCD focal planes which have pixel counts approaching standard visible television resolu-tion. Examples of real-time processed thermal images will be shown.

Short wavelength (λ_c = 2.5 μm) 32 x 32 HgCdTe focal plane arrays have been fabricated for use in an Airborne Imaging Spectrometer (AIS) developed by the Jet Propulsion Labora-tory for NASA. An Imaging Spectrometer provides simultaneous imaging of several spectral bands for applications in the sensing and monitoring of earth resources. The detector material is HgCdTe grown on CdTe substrates using liquid phase epitaxy. Planar processing is used to make photovoltaic detectors on 68 um centers. The detector array is mated to a silicon charge coupled device multiplexer to make hybrid focal plane arrays. Results show high performance detectors with a mean R_oA = 9.6 x 10⁷ Ω --cm² and IleakAge (-100 mV) = 0.037 pA at 120K and near zero background. The yield and uniformity are high. The ratio of the standard deviation of the dc responsivity to the mean is 3% for 98.5% of the pixels. The D1.0 = 1.3 x 10¹² cm - âœ"fiz/W at a background of 10¹³ ph/cm2-s and 120K which is close to the background limited (BLIP) D* of 1.9 x 10¹² cm- âœ"Hz/W.

A theoretical analysis has been done to estimate background clutter leakage noises due to vibrational motion in an AC-coupled mosaic sensor. The motion is separated into uniform drift-dominant and jitter-dominant cases. For the former case, a mathematical expression is obtained to give mean square clutter leakage in a summation form which includes the effects of background scene spatial variations, optical transfer function, footprint size, drift velocity and a smear function. The summation expression is simplified to obtain closed forms for equally spaced fill and spill samples and also for samplings at the end of a frame tine. In the jitter-dominant case, the leakage estimate is shown to depend upon both vibrational frequency and the statistics of the jitter motion.

This paper describes a high resolution CCD multiplexer for focal plane imaging systems. The multiplexer incorporates quadrilinear readout registers to achieve two times the resolution of conventional bilinear structure while using the same design rules. Complete parallel charge transfer are ensured by a novel buried channel poly gate isolation scheme. A monolithic silicon photodiode array of 8 Am pitch, 3533 elements was designed with the multi-plexer. Video preprocessing circuits of high speed four to one channel stitching, compensated sample and hold and bad pixel deletion were integrated on chip for improved performance. The modulation transfer functions due to the geometry and the transfer inefficiency are discussed. The theoretically calculated total MTF agrees with the experimental result. At Nyquist frequency of 62.5 c/mm the total MTF is better than 0.6 in the absence of the diffusion MTF degradation. The noise spectrum of the CCD and the output amplifier are presented. The RMS noise of the CCD in dark is approximately 0.35 my over 1 MHz bandwidth. The CCD noise increases with light input attributed primarily to the shot noise. The low noise nature of the multiplexer makes it ideal for the high resolution low light level detection applications.

Design constraints for optimal resolution of pushbroom scanned linear arrays are studied. Resolution and detector signal-to-noise ratio (SNR) trade-offs are analyzed with respect to detector geometry, quantum integration timing, flight dynamics, and optical point spread function. The point spread function model includes the effects of optical aberrations, diffraction limiting effects and platform vibration. Optimal resolution in the scan direction is defined for discretely sampled point sources. The system model maximizes detector responsive area in time delay and integrate (TDI) arrays under an optimal resolution constraint.