Low dark current Sb-based short-wavelength infrared photodetector

The 2 µm wavelength band has significant applications in gravitational wave detection^1–4 and astronomical observation.^5,6 However, in these fields, the signals are weak and background noise interference is high. To meet the requirements for precise measurements, there is a particular need for detectors with low noise to achieve high detection performance. Currently, various material systems cover this spectral range, each with its own advantages. For example, the InGaAs detector, which has good lattice matching with an InP substrate, has an absorption cutoff wavelength of 1.7 µm. Mercury cadmium telluride detectors are mature and exhibit significant performance, but obtaining high-quality large-area materials is challenging, and the manufacturing cost is relatively high.⁷ In contrast, InAs/GaSb/AlSb type-II superlattice (T2SL) structure materials offer several advantages: (1) The epitaxial material possesses large-area uniformity and low cost; (2) it spans a broader range of the infrared spectrum;⁸ and (3) the flexibility in device design can effectively suppress Auger recombination, thereby reducing the dark current of the PD. These advantages position InAs/GaSb superlattice materials as promising candidates for infrared detection and spectral imaging in the visible and the full infrared wavelength range.

Research on InAs/GaSb/AlSb short-wavelength detectors has garnered widespread attention, and there has been rapid progress in the study of the dark current levels of PDs. In 2012, Hoang et al.⁹ conducted a study on InAs/GaSb/AlSb short-wavelength detectors and found that the dark current density was 5.6 × 10⁻⁸ A/cm² at 150 K. After three years, Haddadi et al.¹⁰ investigated high-performance short-wavelength detectors based on InAs/InAs_1−xSb_x/AlAs_1−xSb_x T2SLs. The dark current density was 9.6 × 10⁻⁵ A/cm² at 300 K with a −50 mV bias. In 2019, Dehzangi et al.¹¹ reported on InAs/AlSb/GaSb T2SL detectors, achieving a dark current density of 5.3 × 10⁻⁴ A/cm² at 300 K with an additional −20 mV bias. In 2023, Jiang et al.¹² reported a dark current density of 1.03 × 10⁻³ A/cm² at 300 K with a −50 mV bias. These experimental results fully demonstrate the excellent performance and extensive potential applications of InAs/GaSb/AlSb superlattices in short-wavelength detection.

However, the above studies about InAs/GaSb/AlSb T2SL short-wave infrared PDs have not yet demonstrated sufficiently favorable dark current characteristics for applications in low-noise and high-sensitivity domains. In fact, based on the flexibility of band structure design in type-II superlattice photodetectors (PDs), there is still greater potential for improvement, especially in terms of noise suppression. Therefore, it is necessary to combine both theoretical and experimental aspects to design device structures for noise suppression in photodetectors, fully exploiting the outstanding advantages of type-II superlattice photodetector performance.

In this article, we designed a type-II superlattice photodetector with a dual-barrier structure of InAs/GaSb/AlSb and conducted a detailed analysis of the band characteristics, carrier transport properties, and the advantages of this structure in reducing dark current in the device. Simultaneously, we achieved an ultra-low dark current in experiments, providing substantial evidence of the significant effects of this structure in suppressing diffusion current, recombination current, and electron tunneling dark current.

This detector adopts a PIN-type InAs/GaSb type-II superlattice C_pB_nn structure¹³ with the introduction of a dual barrier region, as illustrated in Fig. 1(a). The detector utilizes a GaSb substrate with a circular mesa design, and the incident light illuminates from above, sequentially passing through the P-type heavily doped region, barrier region, and absorption region. After absorption, a small portion of short-wave infrared light enters the N-type heavily doped Ohmic contact region, with SiO₂ passivation applied to the PD sidewalls. This designed structure contributes to enhancing the device’s performance. In addition, the introduction of a dual barrier region allows more effective control of dark current and the transport of photo-generated carriers, providing robust support for optimizing the detector’s performance. Figure 1(b) represents the schematic diagram of the epitaxial material structure. Figure 1(c) illustrates the HR-XRD (High-Resolution X-ray Diffraction) curve of the epitaxial material of PD in the (004) crystal phase. Multiple satellite peaks of the superlattice are clearly discernible, with the annotated superlattice peak positions for the absorption region, barrier 1 region, and barrier 2 region being 31.36°, 31.61°, and 28.02°, respectively. The first-order satellite peak in the absorption region exhibits a half-width at half maximum (FWHM) of 20 arcsec, indicating a high level of crystalline quality in the material.

The absorption characteristics and carrier transport properties of the material are crucial in the design of photodetectors. Utilizing the 8-band k⋅p theory,¹⁴ we have calculated the key physical quantities of the absorption region, such as the band structure, wave functions, and absorption coefficients. Figure 2(a) illustrates the band structure of a periodical superlattice. The electron wave function $(ψCB12)$ and the heavy hole wave function $(ψHH12)$⁠, respectively, represent the probability distributions of electrons and holes in space. In the superlattice, InAs serves as an electron potential well, confining the electron state to the InAs layer. GaSb acts as a hole potential well, confining the hole state to the GaSb layer. The electron and hole wave functions in the AlSb layer are relatively reduced compared to the GaSb layer, indicating that the AlSb layer acts as a barrier for both types of carriers. Figure 2(b) illustrates the band dispersion of the superlattice. The bandgap E_g between HH₁ and CB₁ is 0.49 eV, corresponding to the difference between the ECBE and the EVBE as shown in Fig. 2(a). The first split-off band is represented as SO₁. Figure 2(c) presents the theoretical calculation of the absorption coefficient in the absorption region. The spectrum lines already marked in the graph indicate the step-like transitions of electrons in the subbands. The absorption coefficient of the PD is calculated to be 984/cm at 2 µm.

The flexibility in band engineering of type-II superlattices allows for a significant reduction in dark current in photodetectors without compromising photocurrent. Figure 3(a) illustrates the band diagram of the C_pB_nn superlattice under a reverse bias of −0.5 V. The number labels correspond to those in Fig. 1(a). In the regions where the valence band and the conduction band form potential barriers for carriers transporting in the opposite directions (electron barriers in the fifth and sixth layers and a hole barrier in the third layer), the carriers generated in the absorption region can freely transport toward the heavily doped N-type and P-type regions under reverse bias, leading to an enhanced carrier extraction efficiency. The inset depicts the electric field distribution in the depletion region, which is predominantly concentrated in the double-barrier region.

This design adopts a structure with separated absorption and depletion regions, significantly reducing the dark current of the photodetector (PD). The bulk dark current consists mainly of four mechanisms, namely (a) diffusion current (DIF), (b) generation–recombination current (GR), (c) band-to-band tunneling current (BTB), and (d) trap-assisted tunneling current (TAT). In Fig. 3(b), an analysis of the main bulk dark current caused by the PD structure is presented. Since a wider bandgap P region with a barrier layer of 6 (E_g = 1.64 eV) is designed in the device structure, and its bandgap is greater than the N region, barrier layer 5, and absorption region 4, the intrinsic carrier concentration in the P region is lower than that in the N region. Therefore, only the contribution of the N region diffusion current to the dark current needs to be considered. The generation–recombination current is caused by dark current due to material defects in the depletion region. The wider the depletion region and the smaller the bandgap, the greater the generation–recombination dark current. Compared to traditional PIN-type detectors, the PD’s depletion region is concentrated in two barrier regions of a thin layer, with a thickness of 250 nm, thereby significantly reducing the generation–recombination dark current. Tunneling dark current is divided into band-to-band tunneling dark current and trap-assisted tunneling dark current. Band-to-band tunneling involves carriers tunneling from the valence band on one side of the junction to the conduction band on the other side of the junction. Trap-assisted tunneling current is caused by minority carriers occupying trap states in the depletion region, tunneling through the junction. Both tunneling mechanisms are related to the electric field strength in the junction region. Adjusting the doping concentration and thickness of the P–N junction region can reduce tunneling dark current. Moreover, compared to the absorption region, the barrier region has a wider bandgap, reducing the generation of tunneling dark current. In addition, the barrier region effectively suppresses the generation of surface leakage current on the device surface.

Theoretical calculations^15–17 can quantitatively characterize the influence of different dark current mechanisms on the bulk dark current of the PD. Alchaar¹⁸ analyzed that the primary influencing factor of current is diffusion current. As shown in Fig. 3(c), the theoretical calculation curve of the dark current of the PD without dual-barrier at 300 K, the total dark current closely aligns with the diffusion current curve. At a voltage of −0.5 V, the calculated diffusion current density, generation–recombination current, and band-to-band tunneling current are 1.53 × 10⁻⁶, 8 × 10⁻¹⁰, and 9.4 × 10⁻¹³ A/cm², respectively. Figure 3(d) shows the theoretical calculation of dark current density for the dual-barrier structure detector. At a bias voltage of −0.5 V, the total dark current, diffusion current, generation–recombination current, and interband tunneling current values are, respectively, 2 × 10⁻¹⁰, 1.35 × 10⁻¹⁰, 6 × 10⁻¹¹, and 5.4 × 10⁻¹³ A/cm². Due to the significant decrease in diffusion current of the detector with a dual-barrier structure, the total dark current is four orders of magnitude lower than that of the PD without a dual-barrier structure depicted in Fig. 3(c).

Figure 4(a) presents the dark current density curves of a mesa detector with a diameter of 500 µm at different temperatures as a function of bias voltage. As the temperature decreases from 300 to 180 K, the dark current density gradually decreases, reaching saturation at a bias voltage greater than −0.2 V. At a bias of −0.5 V, the dark current density at 300 and 180 K is 2.5 × 10⁻⁶ and 2.05 × 10⁻⁹ A/cm², respectively.

Figure 4(b) shows the relationship between the dark current density of the device and the reciprocal temperature (1/T) at a bias of −0.5 V within the temperature range of 120 to 300 K. The straight lines are fitted by diffusion dark current and generation–recombination dark current with the reciprocal of temperature, and the intersection of the pink and yellow lines occurs at 240 K. Between 243 and 300 K, the main dark current of the device is diffusion current, while below 243 K, the main contribution is from the generation–recombination dark current. In the temperature range of 240–300 K, the dark current exhibits an Arrhenius-type behavior^9,19–21 with an activation energy of ∼0.45 eV, consistent with the bandgap of the absorption region. As analyzed in Fig. 3(b), the substantial increase in the operating temperature without increasing dark current is attributed to the suppression of generation–recombination current. At lower temperatures, generation–recombination current dominates, and the intrinsic region is unintentionally doped, reducing Shockley–Read–Hall (SHR) recombination. The double-barrier region is sufficiently thick at 250 nm, reducing tunneling dark current.

As shown in Fig. 5(a), the PD with a diameter of 500 µm, under a bias voltage of −0.5 V, exhibits its spectral responsivity measured through a Fourier-transform infrared spectrometer. As the test temperature increases from 180 to 300 K, the 100% detection cutoff wavelength of the PD expands from 2.3 to 2.4 µm. The experimentally measured 100% cutoff wavelength aligns with the theoretically calculated absorption coefficient spectrum as shown in Fig. 2(c). During the test, a step-like absorption between SO₁ and CB₁ is observed in the wavelength range of 1.82–2 µm, consistent with Fig. 2(c). At a temperature of 300 K, the responsivity at 2 µm wavelength is 0.083 A/W. According to the absorption coefficient spectrum in Fig. 2(c), the theoretically calculated responsivity and quantum efficiency of the PD with an absorber thickness of 1 µm at a wavelength of 2 µm are 0.138 A/W and 8.5%, respectively. The experimental test results show a much lower value of 5% compared to the theoretical prediction. Strategies can be improved, such as reducing surface reflection of incident light, increasing the thickness of the absorption region, and designing more optimized absorption coefficient of absorption region to enhance quantum efficiency.

Figure 5(b) shows the response characteristics of a PD with a diameter of 500 μ m at an incident wavelength of 1750 nm under varying reverse bias voltages. The responsivity increases with increasing reverse bias voltage. Under the bias voltage range of 0–0.125 V, as the reverse current of the P–N junction has not reached reverse saturation, the current is mainly composed of minority carriers (electrons and holes) in the P and N regions diffusing toward the depletion region. The minority carrier concentration is strongly influenced by temperature, with higher temperatures leading to higher intrinsic carrier concentrations. Therefore, at reverse biases lower than 0.125 V, the responsivity of the PD increases with increasing temperature. As the reverse voltage continues to increase, exceeding 0.125 V, the reverse current caused by minority carriers reaches saturation. At this point, the gradient of the minority carrier concentration no longer changes with bias voltage, and the dark current of the PD is mainly influenced by diffusion current and generation–recombination dark current. Therefore, as the temperature increases, the dark current increases, and the extraction efficiency of photo-generated carriers decreases, leading to a reduction in responsivity. Under a reverse bias of −0.5 V, the responsivity at 1750 nm wavelength is 0.12 A/W@220 K and 0.09 A/W@300 K. The short-wave infrared device has already achieved the desired responsivity, and future work could further optimize the layer ratio of the superlattice to increase the absorption coefficient or enhance the device’s absorption length to achieve the desired responsivity.

The detectivity is an important parameter describing the detector. Antimony-based superlattice infrared detectors typically exhibit high sensitivity, capable of detecting weak light signals. This allows them to deliver excellent performance even under low light intensity conditions. Considering thermal noise and shot noise, the detectivity (D∗) of the PD can be calculated using the relationship $D∗=R/((4KT)/(RA))+2qJ$⁠, where R is the device responsivity, RA is the differential resistance-area product, K is the Boltzmann constant, T is the temperature of the device, q is electron charge, and J is the dark current. In Fig. 5(c), at −0.5 V bias and a wavelength of 2 µm, the detectivity of a 500 µm diameter PD is presented. The trend of detectivity corresponds to that of dark current, decreasing with decreasing temperature. At 300 K and 180 K, the detector’s detectivity is 1.48 × 10¹⁰ and 6.61 × 10¹¹ cm Hz^1/2/W, respectively. The low dark current contributes to the device achieving higher detectivity. Table I provides a comparison of the performance of the present work with other SWIR photodetector technologies.

In summary, we introduced a dual-barrier C_pB_nn structure InSb/GaSb/AlSb T2SL PD, employing a design structure with separated absorption and depletion regions. The design is aimed at effectively suppressing the generation of dark current. Each layer of the detector’s material is grown using molecular beam epitaxy equipment, resulting in high-quality single-crystal materials. Simultaneously, a thorough theoretical analysis of the device structure’s dark current suppression is conducted. Based on the 8-band k p theory and utilizing the nextnano software for calculations, the band structures of the barrier and absorption regions are obtained. The energy band diagram of the device at a bias voltage of −0.5 V, the absorption coefficient in the absorption region, and simulation results of the dispersion relationship in the absorption region are computed. In addition, parameters related to carrier transport are derived. In terms of experimental data, the dark current surface density of the detector at −0.5 V bias voltage is found to be 2.5 × 10⁻⁶ and 2.05 × 10⁻⁹ A/cm² at 300 and 180 K, respectively. Within the temperature range of 243–300 K, diffusive dark current is identified as the main factor, while below 243 K, composite dark current becomes predominant. Through a detailed analysis of experimental data and theoretical calculations, the mechanisms behind the effective suppression of composite, tunneling, and surface dark currents in the dual-barrier structure are explored. For a detector with a diameter of 500 µm, at a −0.5 V bias voltage, the responsivities at 2 µm wavelength are determined to be 0.083 A/W@300 K. The 100% cutoff wavelengths of the detector at 300 K are consistent with theoretical calculations, measuring 2.4 µm. At temperatures of 300 and 180 K, the detector’s detectivities are found to be 1.48 × 10¹⁰ and 6.61 × 10¹¹ cm Hz^1/2/W, respectively.