Understanding the radiative and non-radiative properties of semiconductor materials is a prerequisite for optimizing the performance of existing light emitters and detectors and for developing new device architectures based on novel materials. Due to the ever increasing complexity of novel semiconductor systems and their relative technological immaturity, it is essential to have design tools and simulation strategies that include the details of the microscopic physics and their dependence on the macroscopic (continuum) variables in the macroscopic device models. Towards this end, we have developed a robust full-band structure based approach that can be used to study the intrinsic material radiative and non-radiative properties and evaluate the same characteristics of low-dimensional device structures. A parallel effort is being carried out to model the effect of substrate driven stress/strain and material quality (dislocations and defects) on microscopic quantities such as non-radiative recombination rate. Using this modeling approach, we have extensively studied the radiative and non-radiative properties of both elemental (Si and Ge) and compound semiconductors (HgCdTe, InGaAs, InAsSb and InGaN). In this work we outline the details of the modelling approach, specifically the challenges and advantages related to the use of the full-band description of the material electronic structure. We will present a detailed comparison of the radiative and Auger recombination rates as a function of temperature and doping for HgCdTe and InAsSb that are two important materials for infrared detectors and emitters. Furthermore we will discuss the role of non-radiatiave Auger recombination processes in explaining the performance of light emitter diodes. Finally we will present the extension of the model to low dimensional structures employed in a number of light emitter and detector structures.
This paper presents one-dimensional numerical simulations and analytical modeling of InAs nBn detectors having n-type barrier layers with donor concentrations ranging from 1.8×1015 to 2.5×1016 cm-3. We consider only “ideal” defect-free nBn detectors, in which dark current is due only to the fundamental mechanisms of Auger-1 and radiative recombination. We employ a simplified nBn geometry, with the absorber layer (AL) and contact layer (CL) having the same donor concentration and comparable thicknesses, to reveal more clearly the underlying device physics and operation of this novel infrared detector. Our simulations lead to a new model for the ideal nBn with an n-type barrier layer (BL) that consists of two ideal backto- back photodiodes connected by a voltage-dependent series resistance representing hole conduction within the BL. Increasing the BL donor concentration lowers exponentially the hole concentration in the BL, thereby exponentially increasing the BL series resistance. Reductions in dark current and photocurrent due to the valence band barrier in the n-type BL only become appreciable when the BL series resistance becomes comparable to or exceeds the sum of the diffusion current resistances of the AL and CL. This new model elucidates the overwhelming importance of the electrical type and doping concentration of the BL to the operation of the nBn detector.
Due to the ever increasing complexity of novel semiconductor systems, it is essential to possess design tools and simulation strategies that include in the macroscopic device models the details of the microscopic physics and their dependence on the macroscopic (continuum) variables. Towards this end, we have developed robust multi-scale modeling capabilities that begin with modeling the intrinsic semiconductor properties. The models are fully capable of incorporating effects of substrate driven stress/strain and the material quality (dislocations and defects) on microscopic quantities such as the local transport coefficients and non-radiative recombination rate. Using this modeling approach we have extensively studied UV APD detectors and infrared focal plane arrays. Particular emphasis was placed on HgCdTe and InAsSb arrays incorporating photon trapping structures as well as two-color HgCdTe detectors arrays.
Numerical simulations play an important role in the development of large-format infrared detector array tech- nologies, especially when considering devices whose sizes are comparable to the wavelength of the radiation they are detecting. Computational models can be used to predict the optical and electrical response of such devices and evaluate designs prior to fabrication. We have developed a simulation framework which solves Maxwell’s equations to determine the electromagnetic properties of a detector and subsequently uses a drift-diffusion ap- proach to asses the electrical response. We apply these techniques to gauge the effects of cathode placement on the inter- and intra-pixel attributes of compositionally graded and constant Hg1−xCdxTe mid-wavelength infrared detectors. In particular, the quantum efficiency, nearest-neighbor crosstalk, and modulation transfer function are evaluated for several device architectures.
This paper presents one-dimensional numerical simulations and analytical modeling of ideal (only diffusion current and only Auger-1 and radiative recombination) InAs nBn detectors having n-type barrier layers, with donor concentrations ranging from 1.8×1015 to 2.5×1016 cm-3. We examine quantitatively the three space charge regions in the nBn detector with an n-type barrier layer (BL), and determine criteria for combinations of bias voltage and BL donor concentration that allow operation of the nBn with no depletion region in the narrow-gap absorber layer (AL) or contact layer (CL). We determine the quantitative characteristics of the valence band barrier that is present for an n-type BL. From solution of Poisson’s equation in the uniformly doped BL, we derive analytical expressions for the valence band barrier heights versus bias voltage for holes in both the AL and the CL. These expressions show that the VB barrier height varies linearly with the BL donor concentration and as the square of the BL width. Using these expressions, we constructed a phenomenological equation for the dark current density versus bias voltage which agrees reasonably well with the shape of the J(V) curves from numerical simulations. Our simulations suggest that the nBn detector should be able to be operated at or near zero-bias voltage.