1. Introduction
The mid-wavelength infrared (MWIR) photodetector operate in the wavelength range from 3 to 5 µm and hold significant promise in modern infrared devices due to the wide range of applications, such as industrial gas monitoring, environmental components analysis, and flame detection[1, 2]. Early studies of MWIR photodetectors primarily focused on HgCdTe and InSb materials. HgCdTe offers many advantages such as the tunable bandgap, high electron mobility, and the well-established industrial manufacturing foundation. However, the instability of its bulk, surface, and interface properties leads to poor uniformity and low production yield[3]. InSb exhibits excellent performance, high stability, and relatively low cost. However the response area is limited by the bandgap of the InSb material, which restricts its applications[4]. Additionally, the rapid progress in antimonide-based infrared material in recent years have demonstrated outstanding performance in MWIR detection area.
The concept of the InAs/GaSb T2SLs was firstly introduced by Sai-Halasz et al. in 1977[5] and was used in the infrared detection area by Smith and Mailhoit in 1987[6]. To create a hole barrier in the valence band, reduce the dark current, and suppress both the diffusion and tunneling effects in the depletion region, Nguyen developed an M-type barrier composed of AlSb/GaSb/InAs/GaSb in 2007[7]. By tailoring the InAs and GaSb components, the bandgap of the superlattice can be precisely tuned within the range of 0.0413 to 0.413 eV, which means the response wavelength from 3 to 30 μm. InAs/GaSb T2SLs are considered as a promising candidate, whose performance can be comparable to that of HgCdTe[8, 9].
Generally, T2SLs photodetectors achieve mesa isolation through etching process, and the doping of contact layers are precisely controlled by MBE growth. The etching process includes plasma and chemical etching. During the etching process, the sidewalls are forcibly broken, resulting in numerous dangling bonds that cause surface energy band bending and the surface leakage channels[10]. This introduces dark current associated with the sidewall surface condition, which negatively impacts the device performance[11−14]. Meanwhile, as the mesa decreases, the proportion of sidewall related dark current increases, severely limited the performance of smaller mesa devices[15]. Therefore, the common etching and passivation processes to increase the performance of the infrared photodetectors have inherent limitations. Etching introduces the sidewall roughness, while the subsequent surface passivation adds further complexity, restricts the development of focal plane arrays (FPAs)[16]. To address these issues, a planar junction InAs/GaSb T2SLs based on ZnS diffusion is developed. This approach eliminates the sidewall formation through etching, as mesa isolation achieves by the diffusion region selection. Compared to the traditional mesa etching process, the planar junction photodiode reduces the manufacturing costs. By minimizing the sidewall surface of the device, the surface leakage current can be effectively suppressed by minimizing the sidewall surface area. Common fabrication techniques of planar junction devices included ion implantation and diffusion. However, ion implantation incurs both high costs and damage in lattice structure of T2SLs. In contrast, diffusion is a simpler process and causes less damage due to the absence of high-energy particle bombardment, making it more suitable for selective regional doping in T2SLs[17]. Planar junction infrared detectors based on materials such as HgCdTe, InAs, and InSb have been successfully demonstrated[18−21]. However, reports on planar photodetectors that utilize InAs/GaSb T2SLs structures are remain limited[22, 23].
This paper presents a fabrication method for MWIR photodetectors featuring a p-type channel by Zn diffusion. The n-type contact layer of InAs/GaSb/AlSb/GaSb T2SLs, the barrier layer of InAs/GaSb/AlSb/GaSb T2SLs, the absorption region of InAs/GaSb T2SLs, followed by the top contact layer of intrinsic InAs/GaSb T2SLs were grown by MBE system. Finally, the selective area thermal diffusion was performed using a rapid thermal annealing system.
2. Method
The InAs/GaSb superlattice epitaxial layers were grown on double-side polished n-type GaSb (100) substrates using an MBE system. The growth parameter optimized were optimized during the growth process to minimize the strain, thereby preventing the defects and lattice mismatches. This approach effectively mitigates the degradation of the performance and ensures the stability of the fabrication process. The substrate and epitaxial structure of the material are shown in Fig. 1(a). A buffer layer with the thickness of 1 μm was deposited at the interface between the substrate and the n-type contact layer to improve the surface quality. To prevent the generation of epitaxial defects, a lattice-matched InAs/GaSb/AlSb/GaSb M-type barrier superlattice with a periodicity of 10/1/5/1 ML was grown. This structure effectively guided and collected photogenerated carriers. Previous results indicate that the M-type barrier structure significantly reduces dark current[24]. Moreover, the n-contact layer with the same composition as the M-barrier layer and high doping concentrations reduced the lattice mismatch during epitaxial growth. The superlattice in the absorption region consisted of 9.25/7 ML of InAs/GaSb materials, while the contact layer was an undoped 9.25/7 ML InAs/GaSb superlattice, serving as a diffusion-selective region. The band structure was shown in Fig. 1(b).
As shown in Fig. 2, after the material growth, a 200 nm-thick silicon dioxide (SiO2) layer was uniformly deposited on the entire epitaxial wafer using plasma-enhanced chemical vapor deposition (PECVD). The diffusion regions are defined by photolithography in the SiO2 layer. These exposed regions were etched by using the reactive-ion etching (RIE) system. A 200 nm-thick ZnS layer, serving as the Zn thermal diffusion source, was then deposited by using CVD. To suppress the volatilization during the diffusion process, a 200 nm-thick SiO2 layer was deposited. The Zn-diffusion process was conducted in a rapid thermal processing (RTP) reactor. After the diffusion, the residual SiO₂ and ZnS materials were removed by using Hydrofluoric acid and Hydrochloric acid. The Ti/Pt/Au (50/50/300 nm) electrodes were fabricated by magnetron sputtering at the electrode positions. The spectral response measurements were performed by using a Fourier transform infrared (FTIR) spectrograph equipped with a temperature controller. The current−voltage (I−V) measurements were conducted by using a semiconductor device analyzer. The responsivity was evaluated with a blackbody radiation source settled at 773 K, combing with the photocurrent measured by using a lock-in amplifier.
3. Results and discussion
The diffusion profile was analyzed by using the secondary ion mass spectrometry (SIMS)[25]. As the diffusion temperature increased from 300 to 450 °C, the results revealed that the diffusion depth expanded from 45 to 60 nm, as illustrated in Fig. 3(a). The diffusion depth demonstrated the significant sensitivity with the temperatures below 400 °C. The dark current was measured by using I−V measurement. Fig. 3(b) shows the dark current density of the device with different diffusion temperatures at 0 V bias. With the increasing diffusion temperature, the dark current decreased within the increase of diffusion temperature from 300 to 400 °C due to the enhanced diffusion depth. However, when the temperature reached 450 °C, the dark current increased sharply, attributed by the high defect density inside the MWIR superlattice induced by the elevated temperatures. The results suggest that excessively high diffusion temperatures failed to further improve the diffusion depth and caused more internal damage and lattice defects.
As shown in Fig. 3(b), the lowest dark current is observed at a diffusion temperature of 400 °C. Accordingly, the dark current characteristics of a single-element photodetector with a diameter of 300 μm were measured within the temperature range from 77 to 280 K under this diffusion condition. Fig. 4(a) presents the I−V curves at different temperatures. At 77 K, with a bias of 0 V, the dark current density is 8.67 × 10−5 A/cm2. Fig. 4(b) shows the Arrhenius plot, where the dark current density is inversely proportional to temperature under a diffusion temperature of 400 °C and a bias of 0 V. In the temperature range from 77 to 160 K, the extracted activation energy is Ea = 127 meV, calculated by the following equation:
Jdiff=T3exp(−EakbT), | (1) |
where kb represents the Boltzmann constant, T is the temperature. The fitted activation energy is slightly greater than half the bandgap value (Eg = 218 meV) of the 9.25/7 ML InAs/GaSb superlattice absorber, derived from the absorption edge of the infrared response spectrum. This indicates that the generation-recombination current dominates, with a minor contribution from the diffusion dark current.
The spectrum response was characterized by using the Fourier transform infrared (FTIR) system. The blackbody infrared radiation was used as the calibration source to determine the quantum efficiency (QE), and the specific detectivity (D*) was calculated afterwards.
The optical properties were characterized at 77 K. The devices exhibited the 50% cut-off wavelength of 5.26 μm, which aligns well with the theoretical design. The D* of photodetectors fabricated at various diffusion temperatures was calculated using the equation provided below:
D∗=ληhc(2qJ+4kbTRA)−12. | (2) |
In which η means the quantum efficiency, λ means the wavelength, h is the Planck constant, c is the speed of light, q is the electric charge, J is the current density, kb is the Boltzmann constant, T is the measurement temperature, and RA is the differential resistance area.
As shown in Fig. 5, the diffusion temperature of 400 °C yields the best performance. The infrared photodetector demonstrates a dark current density of 8.67 × 10−5 A/cm2 at 77 K. Under zero bias, the QE reaches to 42.3%, and the maximum detectivity is 3.9 × 1010 cm·Hz1/2/W. These results indicate that this device has good performance and meaningful for the focal plane arrays.
In addition, Fig. 5 presents the QE spectrum and the specific detectivity of photodiodes fabricated at 300, 350, and 450 °C under the measurement temperature of 77 K. The maximum quantum efficiencies were 24.2%, 9.5%, and 23.4%, while the maximum detectivities were 2.06×1010, 4.26×109, and 5.86×109 cm·Hz1/2/W. The device fabricated at the diffusion temperature of 400 °C formed a p-type channel of appropriate depth in the contact layer, demonstrated excellent performance over a range of diffusion temperatures. At 450 °C, more defects were generated in the SLs. At 300 and 350 °C, the low diffusion temperature prevented the diffusion of zinc ions into the absorption layer, leading to a lower detectivity.
Finally, the 640 × 512 focal plane arrays were fabricated by using the same diffusion process at a diffusion temperature of 400 °C. The focal plane arrays were bonded to the readout integrated circuit (ROIC) by using a flip-chip bonding process. The substrate was thinned to 500 nm through the chemical mechanical polishing to complete the device fabrication process[26]. At 77 K, the focal plane arrays exhibited the average detectivity of 1.12 × 109 cm·Hz1/2/W and the NETD of 539 mK. The average detectivity of the FPAs is derived from the differential measurement method by using the area source blackbody, which is different from the detection method for the peak detectivity of a single device, whose numerical conversion is based on the blackbody temperature and spectral response calculation[27]. Comparing with the results in the literature, there are extremely few studies on the FPAs of MWIR InAs/GaSb T2SLs planar junctions. The NETD of our detector is an order of magnitude higher than that of the InAs/GaSb mid-wave high-temperature mesa FPAs reported previously[28, 29]. The FPAs with a cut-off wavelength of 8 μm reported in 2014 had a NETD of 1200 mK at 81 K operating temperature, which was decreased to 450 mK with the addition of a shallow etching process optimization. However, the NETD of the mesa-type FPAs of the same device was 40 mK[30]. From the above results, the planar junction process still needs to be further investigated to achieve the desired reduction in device noise.
4. Conclusion
In summary, by using the CVD-grown ZnS as the diffusion layer, this study successfully fabricated the InAs/GaSb MWIR planar photodetectors, whose 50% cut-off wavelength is 5.26 μm at 77 K. The device exhibits optimal performance at a diffusion temperature of 400 °C, achieving a maximum quantum efficiency of 42.3% and a peak detectivity of 3.9 × 1010 A/cm2 under an applied bias of 0 V at 77 K. Analysis indicated that with the absence of sidewalls, the primary contributors to the dark current are generation−recombination and diffusion processes. The dark current density at 0 V bias was measured to be 8.67 × 10−5 A/cm2. Furthermore, a 640 × 512 focal plane array was fabricated, yielding an average detectivity of 1.12 × 109 cm·Hz1/2/W and a NETD of 539 mK.
Acknowledgments
This work was supported by the National Key Technologies R&D Program of China (Grant Nos. 2024YFA1208904, 2019YFA0705203), Major Program of the National Natural Science Foundation of China (Grant Nos. 62004189, 61274013), the Strategic Priority Research Program of the Chinese Academy of Sciences (Grant No. XDB0460000), and the Research Foundation for Advanced Talents of the Chinese Academy of Sciences (Grant No. E27RBB03).