With exceptional atmospheric transparency，the mid-wave infrared （MWIR） region plays a vital role in various critical imaging applications. In comparison to the short-wave infrared （SWIR），MWIR offers a stronger signal and improved night vision performance. Furthermore，MWIR boasts superior refractive limit resolution compared to long-wave infrared （LWIR） and benefits from better sensitivity due to the more substantial variation of the blackbody radiation curve with temperature.

Recent advances in InAsSb-based MWIR photodetectors have demonstrated high performance at elevated operating temperatures，outperforming conventional InSb detectors that require operation at temperatures in the liquid nitrogen range^［1-7］. The primary source of dark current in InSb detectors is the generation of carriers in the depletion region due to Shockley Read-Hall （SRH） Generation-Recombination （G-R） centers^［8-9］.

To address these limitations，high-quality InAsSb-based XBn structure infrared detectors can be grown on GaSb substrates. These detectors incorporate a wide energy gap barrier with a low valence band offset，effectively blocking the electronic component of current in the n-type absorber while allowing unimpeded minority hole transport through the barrier layer^［6］. Consequently，the reduced depletion region in the narrow gap absorber of the barrier heterostructure suppresses the thermal generation of carriers，a significant source of dark current in the narrow gap absorber. In contrast，the thermal effects on the wide bandgap barrier are negligible，when in the depleted state.

While there is a wealth of theoretical literature on the doping of barrier layers in XBn structures^［10-13］，few experimental studies have been reported^{［14，15］}. In our study，we focus on high-quality PBn unipolar barrier InAsSb materials grown on GaSb substrates. We investigate the light response and dark current characteristics of PBn InAsSb for varying p-type doping concentrations in both the contact layer （CL） and barrier layer （BL）. Additionally，we perform a detailed analysis of the valence band and tunneling characteristics.

We grew high-quality lattice-matched InAsSb materials using molecular beam epitaxy on GaSb substrates. The growth process involved several key steps：initially，the GaSb substrate underwent degassing at 350 °C for 30 minutes in a pre-degassing chamber to eliminate adsorbed surface gases. Subsequently，the GaSb substrate was transferred to the growth chamber，where deoxidation occurred at 520 °C under a Sb atmosphere. Following deoxidation，a 250 nm GaSb buffer layer was grown at 500 °C to smoothen the surface. Subsequently，a 3 μm InAs_0.905Sb_0.095 absorption layer （AL） was grown at 440 °C，followed by a 150 nm AlAs_0.92Sb_0.08 barrier layer.

It's important to note that GaSb exhibits a thermal expansion coefficient （TEC） of approximately 7.75×10^-6°C^-1，which is about 1.5 times greater than InAs （4.52 × 10^-6 °C^-1） and InSb （5.37 × 10^-6 °C^-1） ^［16］. This TEC difference results in a Sb composition of around 9% on GaSb at room temperature for lattice-matched InAsSb. To mitigate strain during the growth process and attain high crystal quality，the Sb composition of the AL was set at 9.5%. Finally，a 300 nm Be-doped p-type GaSb electrode layer was grown. Figure 1 illustrates the layer structure of the InAsSb films.

To prevent the space charge region from extending too deeply into the AL，which could increase the G-R current，we maintained the AL and the first half of the BL as unintentionally doped. Adjustments to the p-type doping concentration were made in the second half of the BL and the CL，and this approach led to the fabrication of InAsSb unit photodetectors. Table 1 provides an overview of the p-type doping concentrations for the three PBn structure samples.

The properties of InAsSb material were assessed using Fourier-transform infrared （FTIR） spectroscopy and Nomarski microscopy. Photodetector processing involved inductively coupled plasma （ICP） dry etching，SiO₂ surface passivation，and Ti/Au electrode deposition. The photo-response and temperature-variable dark current characteristics of the photodetectors were examined using a probe system，B1500A semiconductor parameter test system，and a temperature-variable metal Dewar. To calculate the energy band structure of PBn InAsSb photodetectors，we utilized Silvaco software.

The Sb composition in both InAsSb and AlAsSb was precisely tuned to 9.5% and 8%，respectively. At 300 K，the cutoff wavelength of InAsSb was determined to be 4.5 μm，as measured through FTIR transmission spectroscopy （as shown in Fig. 1（b））. The surface morphology of InAsSb material was examined using a Nomarski microscope，revealing a remarkably low defect density of less than 300 cm^-2. The HRXRD analysis yielded a FWHM of 27 arcsecs for InAsSb. The combination of a high uniform HRTEM，a low FWHM，and a smooth surface （see Fig. 2） indicates the high quality of the material.

We fabricated unit cell photodetectors with a size of 200 μm×200 μm，and we conducted measurements of photocurrent and dark current （J_dark） in the temperature range from 77 K to 300 K. Furthermore，we investigated the Activation Energy （E_a） using Arrhenius equation under varying temperature conditions^［17］.

In Fig. 3（a），the dark current of sample A is presented，measured in the temperature range from 77 K to 287 K. In this sample，the CL is p-type doped at a concentration of 5×10¹⁷ cm^-3，and the BL is composed of a 75 nm layer （p-type doped ranging from 1×10¹⁴ cm^-3 to 3×10¹⁷ cm^-3） followed by a 75 nm UID layer. Notably，the turn-on voltage for this sample is approximately -0.2 V，and a 'second plateau' with a turn-on voltage of around -0.5 V is observed. This effect is attributed to the higher p-type doping in the CL （5×10¹⁷ cm^-3），which surpasses the BL doping and introduces a hole trap between the CL and the BL.

Figure 3（b） shows the dark current of sample B，which was measured within the temperature range from 77 K to 277 K. In this case，the CL is p-type doped at a concentration of 1.5×10¹⁷ cm^-3，lower than the CL in sample A （5×10¹⁷ cm^-3） and matching the BL's doping level （3×10¹⁷ cm^-3）. By reducing the p-type doping in the CL，the 'second plateau' turn-on voltage disappears.

Figure 3（c） displays the dark current characteristics of sample C，measured over the same temperature range. In this sample，the CL is p-type doped at 1.5×10¹⁷ cm^-3，while the BL remains unintentionally doped. Notably，the dark current of sample C is one order of magnitude lower than samples A and B at positive bias，and the turn-on voltage of -2 V significantly surpasses the values for samples A and B （-1.4 V and -0.4 V，respectively）. At -0.5 V and 150 K，the dark current for sample C is approximately 5×10^-6 A/cm².

Analyzing the photodetector's dark current I-V test results，the conclusions are as follows：

1） Higher p-type doping in the BL results in easier tuning under high reverse bias，as demonstrated by the dark current characteristics of samples B and C.

2） Sample A exhibits a higher p-type doping concentration in the CL compared to the BL，introducing a hole trap that forms a valence band barrier，consequently raising the turn-on and tuning voltage.

3） In the case of sample C，the UID BL effectively blocks majority carriers，both electrons from the AL and holes from the CL. This suggests that the UID BL serves as a complementary barrier.

The Arrhenius fitting results of dark current in Fig. 4 （a）-（c） indicate that E_a falls within the temperatures spanning from 250 K to 150 K and the applied reverse bias increases from 0 V to 3 V. The activation energy for both diffusion current and G-R current can be reasonably approximated as：

When the calculated E_a matches the energy band gap （E_G），the dark current is dominated by the diffusion current. Conversely，when E_a corresponds to half of the energy band gap （E_G/2），the dark current is predominantly governed by the G-R current. At a temperature of 150 K，the activation energy is measured approximately 200 meV，and the proportion of G-R current within the dark current becomes more prominent.

At reverse bias voltages lower than 2.0 V，the E_a values for all three samples surpass 250 meV at 250 K. However，as the applied reverse bias voltage exceeds 2.0 V，the E_a values exhibit varying trends. Samples A and C experience a slight decrease in E_a，while sample B's E_a significantly drops to 100 mV. Because the p-type doping of the CL at 1.5×10¹⁷ cm^-3 is lower than the doping level of the BL at 3×10¹⁷ cm^-3 in sample B，this lower p-type doping facilitates easier tuning at the interface between the CL and the BL under high reverse bias.

In contrast，in sample A，where the p-type doping of CL （5×10¹⁷ cm^-3） exceeds that of BL （3×10¹⁷ cm^-3），the majority electron current and minority hole carriers may be blocked，resulting in a drop in E_a as bias decreases under low reverse bias voltage at 150 K.

Through an analysis of the dark current and E_a characteristics，it becomes apparent that there may exist two distinct space charge regions at the interface of the PBn structure：a） between the BL and the AL； b） between the BL and the CL.

By adjusting the p-type doping concentrations in both the BL and the CL，the tuning current and the occurrence of a 'hole trap' at the valence band can be modified. The dark current characteristics can be effectively modeled using the back-to-back diode equations^［18］：

The equation β = q/nkT incorporates fundamental factors，where q is the electron charge，k stands for the Boltzmann constant，T is the temperature，and n denotes the ideality factor. In our model，V_on1 and V_on2 represent the operating voltages under negative and positive bias，respectively. This model encompasses resistance from two space charge regions （R_SCR1 and R_SCR2） and an ideality factor （β = q/nkT），which can be adjusted to account for diffusion or G-R （Generation-Recombination） currents.

We define the positive polarity as a configuration where a negative bias is applied to the n-side of diode 1 and a positive bias is applied to the p-side of diode 2. When a negative bias of V = V₁is applied，it results in a current density J= J₁，with J_A1 representing the plateau current and V_on1 being the open turn-on voltage. Conversely，when a positive bias of V = V₂ is applied，it leads to a current density J = J₂，with J_A2representing the plateau current and V_on2 signifying the open turn-on voltage.

This model is adept at elucidating the presence of two space charge regions at the interface of PBn structure，notably between the BL and the AL，and between the CL and the BL.

For the photocurrent of Sample C，the simulation parameters include：n=20，R_SCR1 = 1×10⁹ Ω·cm，R_SCR2= 8×10¹⁰ Ω·cm，J_A1=4×10^-8 A/cm²， J_A2 = 1×10^-11 A/cm²，V_on1 =-0.5 V，V_on2=-0.05 V，and T=77 K.

Figure 5（b） displays the fitting of the photocurrent for Sample C，where the simulation results （indicated by the purple dashed line） closely align with the experimental results （depicted by the blue solid line）. These simulation results further substantiate the existence of two space charge regions located on both sides of the barrier layer.

In Fig. 5（a），the Arrhenius fitting results of the dark current for sample C are presented under various reverse biases. At an applied bias of -0.2 V，diffusion current overwhelmingly dominates，which is approximately four times higher than the Rule 07 reference （indicated by the purple dashed line）. However，under -0.5 V and -1.0 V biases，the contribution of the G-R current increases. This surge in G-R current under higher applied bias voltages may be attributed to the expansion of space charge regions.

Figure 5（b） shows the photocurrent characteristics of the three samples at 77 K. The intriguing 'second-plateau' turn-on voltage is observed in the BL p-type doped samples A and B. Notably，this turn-on voltage is exceptionally high in sample A，where the p-type doping concentration of the CL exceeds that of the BL. However，sample B shares the same 'second-plateau' turn-on voltage as sample C. This suggests that the higher p-type doping in the CL compared to the BL introduces an additional barrier in the valence band，effectively blocking the photocurrent.

The photocurrent of the sample with the UID barrier layer is aptly simulated by the back-to-back diode model，signifying the presence of two space charge regions located on both sides of the barrier layer.

When comparing the photocurrent behaviors of samples A，B，and C，it becomes evident that the p-type doping in the barrier layer introduces the distinctive 'second plateau' turn-on voltage. Silvaco simulations of the energy band structure （as shown in Fig. 5（c）） further confirm that p-type doping creates a hole trap within the valence band. At low reverse bias voltages，this hole trap restricts both the photocurrent and dark current. As the reverse bias voltage increases，the confined carriers within the hole trap can be effectively extracted.

In samples with an undoped barrier layer，the valence band remains relatively flat，and no 'second-plateau' turn-on voltage is observed. In contrast，samples with p-type doping in the barrier layer exhibit very low turn-on voltages，some even reaching positive biases. Both experimental results and theoretical calculations strongly suggest that p-type doping in the barrier layer introduces an additional barrier within the valence band.

This phenomenon can be attributed to two key factors：1） P-type doping in the BL tilts the valence band，reducing the valence band offset between the AL and the BL to zero or even to a positive bias； 2） In the PBn structure，the role of the AlAsSb barrier layer in governing the dark current and photocurrent of the photodetectors predominantly depends on the interface band alignment between the BL and the AL and the high impedance of the wide bandgap AlAsSb layer.

The introduction of p-type doping at the interface between the BL and the AL effectively eliminates the barrier effect of this interface on carriers. P-type doping in the BL facilitates easy tunneling under positive bias voltage，reducing the junction impedance. This diminished barrier effect on the conduction band electrons weakens its influence on most carriers. Consequently，it allows for the transverse tunneling of electrons from the AL to the electrode layer at a lower positive bias voltage.

Conversely，samples that remain unintentionally doped in the BL do not exhibit tunneling characteristics even at a positive bias voltage of 3 V. Under these conditions，the majority of electron carriers in the AL are effectively blocked.

This study presents a comprehensive examination of PBn InAsSb photodetectors with a p-type doped BL，incorporating both simulation and experimental results. It is noteworthy that the p-type doping concentration demonstrates minimal influence on the dark current （J_dark）. However，it significantly impacts several other key parameters，including activation energy （E_a），turn-on voltage，and tuning current.

The introduction of p-type doping creates a hole trap in the valence band of AlAsSb，leading to the emergence of the 'second plateau' turn-on voltage. This hole trap subsequently blocks both photocurrent and dark current under low reverse bias voltages. As the reverse bias voltage increases，the previously blocked carriers within the hole trap become extractable. The p-type doped BL is instrumental in facilitating tunneling at lower bias voltages.

Notably，the 'second-plateau' reverse turn-on voltage and the positive tunneling characteristics vanish in the case of the UID BL sample. In such samples，the presence of two space charge regions on either side of the UID BL plays a pivotal role in blocking the majority carriers in both the absorption and contact layers.

Through the analysis of p-type doping in InAsSb photodetectors，we gain valuable insights into transport properties and the energy band structure within PBn designs. This knowledge is instrumental for optimizing structure designs and enhancing the performance of high operating temperature （HOT） photodetectors.

The MWIR InAsSb application market is promising for a wide range of applications，because the HOT photodetectors can achieve SWaP³ （size smaller，wight lighter，performance higher，power lower and price cheaper）. IRnova，SCD，Raytheon Vision Systems and Leonardo UK are developing MWIR InAsSb products for the next generation infrared equipment.