Search

Dislocations are known to influence the electrical and optical properties of long-wavelength infrared (LWIR) HgCdTe detectors and have been shown to limit the performance of arrays fabricated on heteroepitaxial substrates. To help better understand dislocations in HgCdTe, a new method for preparing HgCdTe diagnostic epitaxial single-crystal samples by chemically removing the supporting CdZnTe substrate has been developed. Using this new sample preparation technique, the behavior of misfit and threading dislocations in HgCdTe epitaxial layers has been investigated by using a defect etch to reveal the dislocations present in the thin HgCdTe films. In most cases etch pits on the surface of the film are spatially correlated with etch pits on the bottom of the HgCdTe film. The small displacements of the related etch pits were used to obtain crystallographic information concerning the paths followed by threading dislocations on allowed slip planes in the HgCdTe crystal. In addition, transmission electron microscopy (TEM) is used to obtain more specific information regarding the Burgers vector of the dislocation. While this new sample preparation technique is useful for studying dislocations in HgCdTe epitaxial layers, it can also be used to study stress from ohmic contacts and passivation layers. The technique can be used for both liquid-phase epitaxy (LPE)- and molecular-beam epitaxy (MBE)-grown HgCdTe on CdZnTe substrates.

HgCdTe material intended for long-wavelength infrared detection is particularly susceptible to damage from stress. As a result, an ideal ohmic contact needs to have good adhesion and low specific contact resistance. The contact should act as a diffusion barrier and induce the least amount of stress in the underlying material. In this paper we present a set of stress measurements from different ohmic contact materials deposited on short- and long-wavelength HgCdTe films grown by liquid-phase epitaxy (LPE). Using a new experimental technique we remove the substrate and measure the stress induced on single- and multilayered HgCdTe cantilevers. To interpret our results, we develop a theoretical model that describes the physics of elastic deformation in HgCdTe layers. Our model is based on classical thin-plate bending theory and explicitly takes into account the realistic boundary conditions that are present in the experimental setup by using a variational approach.

As liquid phase epitaxial (LPE) growth and array fabrication processes have matured to give excellent wafer average performance, the yield limiter for infrared focal plane arrays (IRFPAs), especially large ones, have become outages. In this work, significant progress has been made in identifying the source and eliminating outages from LPE grown Hg1−xCdxTe P-on-n structures. Historically, studies of the sources of outages have employed defect etches to look for dislocations and other crystalline defects, and secondary ion mass spectroscopy (SIMS), imaging SIMS, and sputter initiated resonance ion spectrometry (SIRIS) to look for impurities at critical interfaces. Using these techniques, trends were established, but direct correlation with outages have been observed. In LPE grown materials, where the dislocation densities are always below 5×105 cm−2, and often below 1×105 cm−2 on CdZnTe substrates, dislocations only account for a few outages. In order to understand the source(s) of outages, a failure analysis was performed on several long wavelength IRFPAs. Using a dilute etchant, the metals and then cap layers of some 64×64 pixel IRFPAs which had excellent average performance, but suffered from a high density of pixels with excessive leakage current, were removed. Using a scanning electron microscope with energy dispersive spectroscopy capability, the presence of carbon particles was correlated with excessive leakage current on a 1:1 pixel basis. A series of experiments was then conducted which isolated the source of the particles to the cap layer growth process, which was consequently changed to eliminate them. The process improvements have reduced the particle density to below the measurement limit of the optical measurement technique implemented to monitor the density of particles on witness wafers. These improvements are resulting in IRFPAs with significantly improved operability.

X-ray topography provides a very sensitive map of lattice mismatch between a HgCdTe LPE epitaxial layer and its (111) CdZnTe substrate. A well-defined Crosshatch pattern in the three «110» directions indicates a positive room-temperature lattice mismatch. For conditions of near-perfect lattice matching (±0.003% mismatch), the Crosshatch pattern disappears, presumably because there are few or no misfit dislocations present near the interface, and a region free of topographic contrast is observed. The crosshatch-free region occurs for a small positive room-temperature mismatch (about 0.02%); this is attributed to differences in the lattice matching condition at room temperature and the growth temperature. For negative mismatches, where the film is in tension, a mosaic pattern, rather than a crystallographically oriented Crosshatch, is observed in the topograph. Rocking curve full width at half maximum of the epitaxial layer is minimized in the crosshatch-free zone at a value nearly equal to that of the substrate. Etch pit density of the HgCdTe layer shows a strong minimum for perfect room temperature lattice matching, with values as low as 1 x 104 cm−2. For nearly lattice matched layers, Crosshatch is present throughout the thickness of the epitaxial layer except for a narrow graded-composition region near the substrate interface. Crosshatch contrast appears to result from long-range strain fields associated with a misfit dislocation network near the substrate interface. Spatial variations in topographic features and mismatch across relatively small lateral distances are caused by variations in substrate alloy composition. For truly lattice-matched substrates, better control over the substrate lattice parameter is required.

Results of large-area (up to 1000 cm2/run) Cd1-xZnxTe heteroepitaxy on both GaAs and GaAs/Si substrates by metalorganic chemical vapor deposition (MOCVD) are presented. Cd1-xZnxTe (x = 0-0.1) films exhibited specular surface morphology, 1% thickness uniformity (standard deviation), and compositional uniformity (Δx) of ±0.002 over 100 mm diam substrates. For selected substrate orientations and deposition conditions, the only planar defects exhibited by (lll)B Cd1-xZnxTe/GaAs/Si films were lamella twins parallel to the CdTe/GaAs interface; these do not propagate through either the Cd1-xZnxTe layer or subsequently deposited liquid phase epitaxy (LPE) HgCdTe layer(s). Background Ga and As-impurity levels for Cd1-xZnxTe on GaAs/Si substrates were below the secondary ion mass spectroscopy detection limit. Preliminary results of HgCdTe liquid phase epitaxy using a Te-rich melt on Si-based substrates resulted in x-ray rocking curve linewidths as narrow as 72 arc-sec and etch-pit densities in the range 1 to 3 x 106 cm2.

We report current-voltage data for back-illuminated mesa photodiode test structures fabricated by arsenic-diffusion into n-type LPE HgCdTe films. Arsenic diffusion was carried out in a sealed quartz ampoule containing a source of both Hg and As. The arsenic-diffused p-on-n photodiodes were characterized at 70 K and 80 K. The cutoff wavelength was about 11 μm at 80 K. The data for 400 μm diameter photodiodes fabricated by the arsenic diffusion process are very similar to those from a conventional two-layer LPE P-on-n process for material with approximately the same cutoff wavelength. We outline process and doping level changes that should improve detector performance.

We report results for 64 X 64 simultaneous MW/LW dual-band HgCdTe Focal Plane Arrays (FPAs). The MW and LW average cutoff wavelengths at 78 K are in the 4.27 - 4.35 micrometer and 10.1 - 10.5 micrometer ranges respectively. The unit cell size is 75 X 75 micrometer2. These staring dual-band FPAs exhibit high average quantum efficiencies (MW: 79%; LW:67%), high median detectivities (MW: 4.8 X 1011 cm- (root)Hz/W; LW: 7.1 X 1010 cm-(root)Hz/W), low median NE(Delta) Ts (MW: 20 mK; LW: 7.5 mK for TSCENE equals 295 K and f/2.9), large dynamic ranges (MW: 77 dB; LW: 75 dB), and 87% stare efficiencies for both the MW and LW spectral bands. The dual-band HgCdTe detector array is fabricated from a four- layer P-n-N-P film grown in situ by MOVPE. The dual-band silicon CMOS input circuit utilizes a unique floating-direct- injection approach to achieve separate and simultaneous integration of both bands within each unit cell. There are two Compact Signal Averager circuits in each unit cell, to average subframes within every frame for each spectral band, allowing full stare efficiency in both spectral bands, as well as variable band-independent transimpedance gains. These data confirm that all key features of our P-n-N-P dual-band HgCdTe detector and our dual-band input circuit function as designed.

This paper reports the deposition of (100) GaAs and (111)B CdZnTe layers on silicon substrates up to 4-inch diameter to produce substrates suitable for liquid phase epitaxy (LPE) of high-quality HgCdTe layers. Metalorganic chemical vapor deposition is used for both GaAs and CdZnTe in a reactor capable of deposition onto eighteen 3-inch or ten 4-inch wafers per run. An encapsulation scheme is described that prevents contamination of a Te melt by Si or GaAs during LPE growth. Excellent uniformity of thickness and Zn concentration are achieved in the MOCVD films. The CdZnTe films show only lamellar twins close to the GaAs interface; no twins capable of propagating into the HgCdTe layer are formed. These substrates have been used for the growth of pure HgCdTe films having a dislocation density that is only a factor of 2 to 4 higher than that measured in similar films grown on bulk CdTe substrates.

In the system both end members and the phases and have been reported to be useful transparent conductors. The present report concerns amorphous sputtered films having compositions ranging from 20 to 80 mole percent in . We have determined the properties of both as‐deposited and annealed films as functions of both film composition and sputtering conditions. From these results, the optical and electrical properties of the films have been correlated with film structure and composition. The optimum conditions for obtaining highly conductive and transparent films are described. Films having a transmission of 80%–90% in the visible and a sheet resistance of less than 5 Ω/sq can be deposited under these conditions.

The electrical properties of 2000–3000‐A‐thick films of hafnium, hafnium nitrides, and hafnium dioxide have been studied. The properties of these films can be related to film structure and composition. Conditions were established under which hafnium films can be prepared by dc sputtering in pure argon with density and resistivity approaching that of pure bulk hafnium. Additions of low concentrations of nitrogen to the argon sputtering atmosphere result in the deposition of films consisting of a solid solution of nitrogen in hafnium. At intermediate nitrogen concentrations films of HfN, a compound with metallic properties, are obtained. Higher nitrogen concentrations in the sputtering atmosphere result in films of nonstoichiometric HfN containing excess nitrogen and finally of films of a higher nitride of hafnium having semiconducting properties. Sputtering in argon‐oxygen mixtures above a critical oxygen concentration results in the deposition of HfO2 films with insulating properties.

Hall coefficient and conductivity measurements were made on single crystals of CdTe at temperatures up to 950°C while controlling the partial pressure of either cadmium or tellurium. Measurements made with a cadmium reservoir showed the material to ben-type due to the presence of a doubly ionized native donor, in agreement with the results of Whelan and Shaw. The apparent enthalpy of formation of the donor and its concentration along the cadmium-rich solidus were obtained. Measurements made with a tellurium reservoir close to tellurium saturation are in disagreement with the usual model, which attributes formation ofp-type material under these conditions to the presence, of a native acceptor defect. The results are consistent with a hole concentration originating from an excess of acceptor impurities rather than from a native acceptor.

The total vapor pressure, , over has been determined between 980° and 1169°K by using a sum manometric technique. The lower end of an evacuated, inverted, fused silica U‐tube was immersed in a large reservoir of melt inside a transparent furnace. The value of and the density of were obtained at each temperature by visually observing the height to which the melt rose in the U‐tube as a function of the external argon pressure applied to the reservoir surface. In separate experiments a sealed, evacuated, fused silica optical cell was maintained at 1000°C while the temperature of a sidearm reservoir containing was held at a lower temperature. By measuring the optical density as a function of wavelength between 1990 and 5500Aa, it was possible to determine the partial pressure of , , and the partial optical density of as functions of the temperature between 943° and 1200°K. If it is assumed, consistent with all the evidence, that Te2 and are the principal vapor species, the results of both sets of experiments are represented by According to these expressions neither vaporizes nor effuses congruently since is never more than 15% of . The density of is 7.4 g/cm3 at 973°K and 7.0 g/cm3 at 1170°K.