The paper presents the method and results of low-frequency noise measurements of modern mid-wavelength infrared photodetectors. A type-II InAs/GaSb superlattice based detector with nBn barrier architecture is compared with a high operating temperature (HOT) heterojunction HgCdTe detector. All experiments were made in the range 1 Hz - 10 kHz at various temperatures by using a transimpedance detection system, which is examined in detail. The power spectral density of the nBn’s dark current noise includes Lorentzians with different time constants while the HgCdTe photodiode has more uniform 1/f - shaped spectra. For small bias, the low-frequency noise power spectra of both devices were found to scale linearly with bias voltage squared and were connected with the fluctuations of the leakage resistance. Leakage resistance noise defines the lower noise limit of a photodetector. Other dark current components give raise to the increase of low-frequency noise above this limit. For the same voltage biasing devices, the absolute noise power densities at 1 Hz in nBn are 1 to 2 orders of magnitude lower than in a MCT HgCdTe detector. In spite of this, low-frequency performance of the HgCdTe detector at ~ 230K is still better than that of InAs/GaSb superlattice nBn detector.

Low frequency noise spectroscopy is a valuable tool for studying narrow gap semiconductor materials and devices. In the paper, the method of traps investigation with low frequency noise spectroscopy was presented, together with quantitative analysis of the results obtained for high-operating-temperature, fast response, Hg _{1− x} Cd _x Te mid-wavelength infrared detectors. The hole trap levels in these devices are Cd-content independent and take the value E _T ≅ 140 meV or E _T ≅ 40 meV. The level at E _T ≅ 140 meV was found for almost all tested detectors. The level around E _T = 40 meV was found in two samples. Apart from the hole traps, three electron traps were also found. Their energy levels follow the trend lines E _T = 0.35 E _g ( x ) and E _T = 0.75 E _g ( x ). All trap energies are consistent with the results reported in the literature for Hg _{1− x} Cd _x Te devices.

1/f noise-dark current correlation of p-i-n InAs/ GaSb T2SL detectors was analyzed. Experiments include current and noise measurements versus voltage and temperature. Modeling was used to decompose dark current into its components and correlate them with the observed 1/f noise. As a result, the total 1/f noise at arbitrary bias and temperature can be expressed as a sum α _sh (I _sh ) ² + α _G-R (I _G-R ) ² + α _tun (I _tun ) ¹ . 1/f noise from diffusion current was not observed, probably due to a very small noise coefficient, which is less than α _diff <; 10 ^-11 . The generation-recombination (G-R) current has relatively small contribution to the total 1/f noise in the inspected samples, because its noise coefficient is α _G-R = 2 × 10 ^-8 . The largest contribution to 1/f noise comes from the shunt current. The shunt current 1/f noise coefficient α _sh is temperature independent and proportional to the shunt resistance, α _sh ~ R _sh . Its value can be so high that shunt-originated 1/f noise is observed at high temperature region, in which current is limited by the G-R and diffusion components. At high reverse bias, tunneling currents introduce additional contribution to 1/f noise, which exhibits linear dependence on the tunneling current.

In this report, we present results of an experimental investigation of a near mid-gap trap energy level in InAs _{10 ML} /GaSb _{10 ML} type-II superlattices. Using thermal analysis of dark current, Fourier transform photoluminescence and low-frequency noise spectroscopy, we have examined several wafers and diodes with similar period design and the same macroscopic construction. All characterization techniques gave nearly the same value of about 140 meV independent of substrate type. Additionally, photoluminescence spectra show that the transition related to the trap centre is temperature independent. The presented methodology for thermal analysis of dark current characteristics should be useful to easily estimate the position of deep energy levels in superlattice photodiodes.

The empirical 1/f noise model for p\(^{+}\)-p-n infrared detector made of type-II InAs/GaSb superlattice material is presented. It is shown that 1/f noise magnitude can be accurately estimated if dark current contributions are determined and noise coefficients are known. It is found that the shunt, the bulk generation–recombination, and the trap-assisted tunneling currents contribute to the total 1/f noise. No 1/f noise connected with the diffusion and the band-to-band tunneling currents is observed.