Search

We propose simple, analytical, approximations of carrier density for HgCdTe semiconductors that have non-parabolic energy bands and are highly degenerate. The proposed expressions are in the form similar to the classical Boltzmann's approximation without any adjustable parameter. These relations can be applied to HgCdTe for the case where electron densities are very high and the material is highly degenerate, e.g. highly accumulated surface due to passivant induced negative fixed charge density in n-HgCdTe photoconductor device. The expressions are valid in the LWIR and MWIR bands, and can be tuned further to apply to other bands.

We present the results of calculations for surface shunt resistance due to the passivant fixed charge density (Qss) in n-HgCdTe photoconductive (PC) detectors. To the best of our knowledge, this is the first detailed calculation involving the actual majority carrier profile at the accumulated surface. The effect of surface field (or potential) due to heavily accumulated density of majority carriers on surface mobility (μs) has been investigated in detail by employing the Schrieffer and Goldstein models using random diffuse scattering. Additionally, the effect of lateral field (applied to these devices) on surface mobility is included by invoking the model of Yoo et al. The above effects were not taken into account in previous simplified models. For narrow-band, n-type HgCdTe the effects of carrier degeneracy and band non-parabolicity cannot be neglected. In this work, a one-dimensional model including these effects has been developed to evaluate the detector resistance and responsivity. A proper two-layer (bulk and surface) responsivity model is developed. The results are compared with the widely-used approximate one-layer model of Reine and with the step model proposed by Bhan and Gopal. It is shown that, for the Reine model, the agreement with the present model depends on the value of μs chosen. The trend of the step model agrees with the Reine model, but both models show disagreement with the present one-layer and two-layer profile models for Qss ≈ (1010–1012) cm−2.

The dependence of the zero-bias resistance-area (RA) product and the quantum efficiency (η) of variable-area diode arrays is numerically calculated by solving the diffusion equation in a cylindrical, three-dimensional geometry in the thick base approximation. The calculation is done for long-wavelength IR HgCdTe n+-on-p diffusion-limited photodiodes at 77 K. The inverse resistance-area product 1/(RA) and the square root of the quantum efficiency, η1/2, are plotted against the perimeter-to-area (P/A) ratio. The 1/RA results are fitted to a quadratic dependence on the P/A ratio. The dependence of the 1/RA on the minority carrier diffusion length, the junction depth and the surface recombination velocity (SRV) is evaluated. An empirical expression is proposed that largely accounts for the dependence of the coefficients of the quadratic on these parameters and is more general than those used in previous studies. The results are also in reasonable agreement with the results of Briggs, expressed in terms of the parameter f3D, that are valid for zero junction depth and zero SRV. The slope of the quantum efficiency versus P/A plot, which is approximately a straight line, is related to an effective length Lopt, that also depends on the diffusion length, junction depth, SRV and the absorption coefficient α. The parameter Lopt varies as α1/2.

In this paper, the varying composition (x) of the Hg1-xCdxTe epilayer is related to a new parameter namely the shift in the forward bias voltage at a constant current (Vcon) of the photovoltaic (PV) diodes in an array environment. The dependence of Vcon on the composition x is investigated both theoretically and experimentally. It is shown that Vcon is almost linearly proportional to x and hence can be utilized as a calibration standard for mapping x or non-uniformity in x. Here, this is done by fabricating nine 16×16 photovoltaic diode arrays on liquid-phase epitaxy-grown mercury cadmium telluride, Hg1-xCdxTe, of nominal composition x = 0.227. The composition x or cut-off wavelengths (λc) determined from Fourier transform infrared transmission for three specific PV diodes having different compositions (x) are correlated with the shift in Vcon of current-voltage (I-V) characteristics of these diodes. Following this, Vcon versus x is plotted which is used as a calibration curve for mapping non-uniformity in x in terms of Vcon for a given technology. Furthermore, it is proposed that, in addition to the usual information, the I-V characteristics can be used to monitor any process-induced changes in the composition of the starting material.

In this paper a model for calculating the quantum efficiency and the specific detectivity of both and photodiodes, considering both front- and backside illuminated configurations, is presented. The diode is analysed by assuming a step model for the concentration profile. The effects of the surface recombination velocities at different surfaces/interfaces of the diode, the minority carrier lifetimes and the carrier concentrations are considered. It is pointed out that the parameter to be optimized for a photodiode is and not the zero-bias resistance - area product , although the latter combines with the quantum efficiency to determine the behaviour of . Based on our calculations, it is observed that the values of different parameters that maximize are different from those that are optimal for high quantum efficiency. For sufficiently high , dominates ; otherwise, both factors affect . The calculation of done here refers to the long-wavelength IR region where often is low, with the result that the injection efficiency into the CCD multiplexer is decreased, which has been taken into account.

Voltage - temperature characteristics of photovoltaic diodes fabricated on HgCdTe epitaxial layers have been studied. The diodes are operated with forward bias in a constant-current mode. Plots of forward voltage versus temperature (T) clearly reveal well resolved peaks suggestive of a trap-assisted tunnelling (TAT) mechanism. The nonlinearity of versus T plots in terms of these peaks correlates with reverse leakage currents and resistance - area products whereas the ideality factor correlates with TAT. Two such peaks are observed in our samples near 120 and 90 K. The origin of these peaks is explained by a multi-step tunnelling recombination model.

The results of a quasi-two-dimensional model for calculating passivant-induced surface leakage currents due to band-to-band tunnelling in mercury cadmium telluride (MCT) photovoltaic (PV) diodes are presented. The object is to assess the effect of a fixed surface state charge density due to a passivant on the zero-bias resistance - area product for the technologically important case when surface state charges accumulate the MCT surface. Calculations are carried out to estimate the tolerable value of beyond which the of the MCT diode degrades. To the best of our knowledge, this is the first time that such a detailed calculation involving the acceptor concentration profile near the surface has been reported for long-wavelength IR (LWIR) MCT photodiodes. This calculation has been done numerically, and hence the depletion width (pinched near the surface), the electric field and the band-to-band tunnelling are calculated as a function of depth, layer by layer, from the passivant - semiconductor interface. The currents - diffusion, generation - recombination, band-to-band and trap-assisted tunnelling - have been calculated in each layer, and the zero-bias resistance - area product is determined for each mechanism. Hence, the resultant is calculated. The results are compared with the earlier step model of Bhan and Gopal (Semicond. Sci. Technol. 9 (1994) 289), which assumed a surface layer of constant concentration . The present model indicates that for photodiodes with a cut-off wavelength of and an acceptor concentration , operating at 77 K, a would degrade significantly. This value is insensitive to composition (in the LWIR). For an diode, the tolerable value of is found to depend on the concentration in the layer. Further, the tolerable value of for an diode is much higher than for an diode, for donor concentrations .

The effect of the layer in a structure made in long-wavelength (x = 0.229) epilayers on the zero-bias resistance - area product is theoretically analysed, and compared with a structure. A step profile is assumed for the concentration in the and layers, fabricated by ion implantation and annealing. Various current components (diffusion currents in the , and p layers, depletion layer and surface generation - recombination (g - r) currents, band-to-band and trap-assisted tunnelling) are taken into account. The effect of the surface recombination velocities, and the concentrations (p-side trap, layer, and p layer) are discussed in detail. The results of this model clearly indicate the conditions under which a diode can perform better than a diode. In broad terms, a structure is more suitable for p-type epilayers with lower trap concentrations and higher lifetimes, whereas a structure is advantageous, in terms of increasing , for higher trap concentrations and lower lifetimes. Another advantage of the structure is that reduced tunnelling implies that a higher acceptor concentration can be accepted for device fabrication.

The authors report the first model of Zener tunnelling capacitance and its effect on the C-V characteristics of Hg1-xCdxTe. For narrow band Hg1-xCdxTe, with x=0.2, even at moderate doping densities, C-V curves exhibit low-frequency behaviour at measurement frequencies as high as 1 MHz (both for p-type and n-type MCT). The model predicts the reported trends in C-V characteristics as a function of composition, doping density and measurement frequency.