Electro-Optical Characteristics of P^sup +^n In^sub 0.53^Ga^sub 0.47^As Hetero-Junction Photodiodes in Large Format Dense Focal Plane Arrays

• Published in: Journal of electronic materials Vol. 44; no. 8; p. 2813
• Main Authors: Dewames, R, Littleton, R, Witte, K, Wichman, A, Bellotti, E, Pellegrino, J
• Format: Journal Article
• Language: English
• Published: Warrendale Springer Nature B.V 01.08.2015
• Subjects: Diodes; Infrared imaging systems; Mathematical models; Simulation
• ISSN: 0361-5235; 1543-186X
• DOI: 10.1007/s11664-015-3706-8

This paper is concerned with focal plane array (FPA) data and use of analytical and three-dimensional numerical simulation methods to determine the physical effects and processes limiting performance. For shallow homojunction P^sup +^n designs the temperature dependence of dark current for T < 300 K depends on the intrinsic carrier concentration of the In^sub 0.53^Ga^sub 0.47^As material, implying that the dominant dark currents are generation and recombination (G-R) currents originating in the depletion regions of the double layer planar heterostructure (DLPH) photodiode. In the analytical model differences from bulk G-R behavior are modeled with a G-R like perimeter-dependent shunt current conjectured to originate at the InP/InGaAs interface. In this description the fitting property is the effective conductivity, σ ^sub eff^(T), in mho cm^sup -1^. Variation in the data suggests σ ^sub eff^ (300 K) values of 1.2 × 10^sup -11^-4.6 × 10^sup -11^ mho cm^sup -1^). Substrate removal extends the quantum efficiency (QE) spectral band into the visible region. However, dead-layer effects limit the QE to 10% at a wavelength of 0.5 [mu]m. For starlight-no moon illumination conditions, the signal-to-noise ratio is estimated to be 50 at an operating temperature of 300 K. A major result of the 3D numerical simulation of the device is the prediction of a perimeter G-R current not associated with the properties of the metallurgical interface. Another is the prediction that for a junction positioned in the larger band gap InP cap layer the QE is bias-dependent and that a relatively large reverse bias ≥0.9 V is needed for the QE to saturate to the shallow homojunction value. At this higher bias the dark current is larger than the shallow homojunction value. The 3D numerical model and the analytical model agree in predicting and explaining the measured radiatively limited diffusion current originating at the n-side of the junction. The calculations of the area-dependent G-R current for the condition studied are also in agreement. Unique advantages of the 3D numerical simulation are the ability to mimic real device structures, achieve deeper understanding of the real physical effects associated with the various methods of junction formation, and predict how device designs will function.