1. Introduction

Over the last few decades, there have been tremendous advances in the infrared imagers thanks to the rapid progress in detector technologies, with respect to both their performance and fabrication cost. Of all options, antimonide-based type–II superlattices (T2SLs) have emerged as ideal candidates for infrared photodetectors, which are capable of covering the whole infrared range by tuning their energy bandgaps with adopting different superlattice periods. Benefiting from the lattice-matched growth conditions and advances in the molecular beam epitaxy (MBE) technology, this type-II band alignment enables an accurate control of energy band structure and exhibits great flexibility in the design of detector devices for a wide range of applications in different infrared regimes, without degrading the material quality.[1] After long-term development, they are now gradually standing out as strong competitors to the “state-of-the-art” technologies such as mercury cadmium telluride (HgCdTe).[2,3] Recently, there has been prompt increase in the calls for a reduction in the size, weight, and power (SWaP) for mid-/long- wavelength infrared range (MW/LWIR) detection systems, which requires high operating temperature (HOT) devices that can ease the burden of cooling systems. To achieve high temperature operation, the challenge is to deal with the increase of thermal noise at high temperatures caused by the exponential increase of minority carriers’ decay through Shockley-Read-Hall (SRH), radiative, and Auger recombination, especially for the narrow bandgap absorbers falling into MW/LWIR range. [4,5] Therefore, the feasible solution is to increase the minority carrier lifetime which fundamentally determines the performance of infrared detectors regarding to the dark current as well as the quantum efficiency.

On solving this problem, “Ga-free” InAs/InAs_1-xSb_x T2SLs-based material system has been innovatively established, riding on the advantages of significantly longer minority carrier lifetimes, simpler growth, and better defect tolerance than its counterparts InAs/GaSb T2SLs. [6–12] Due to the mismatched growth conditions with InAs, defect-related SRH recombination at the superlattice interfaces would probably reduce the minority carrier lifetime. The locations of the defects to the T2SLs absolute energy states have been proven to have great impact on the SRH recombination process [13], and it seems that defects level in the As-rich InAs/InAsSb T2SLs falls in the conduction band and have less negative effect towards the carrier decay via SRH. [14,15] The hints of the SRH recombination have been found in several studies, especially at the conditions like low temperature and low carrier densities.[16–20] Conventionally, to achieve MWIR HOT photodetector devices, high antimony molar composition (x) strained InAs_1-xSb_x layers are preferred in the superlattice design for the InAs/InAsSb T2SLs-based absorbers, which enlarges the splitting of intersubband caused by the strain effects, leading to a reduction in the Auger recombination rates by limiting the possible transition states for the electrons and holes. [21–23] Additionally, this effect can be further strengthened by a quantum confinement effect with reducing the thickness of InAs_1-xSb_x layers. [24] One breakthrough as an example has been achieved recently by David Z. Ting etal, who have first demonstrated nBn InAs/InAsSb T2SLs barrier infrared detectors/FPAs with a significantly higher operating temperature than the MWIR market-leading InSb FPAs. [21] As an alternative, pBn HOT FPA devices based on InAs/InAsSb T2SLs have not reported yet and the size of pixel still needs to be further scaled down for meeting the demands for high-resolution MWIR imagers.

In this work, we report the demonstration of high-performance pBn MWIR photodetectors involving short-period InAs/InAsSb T2SLs for being used as an efficient absorber to achieve HOT devices. The pBn photodetectors based on this design exhibit quantum efficiency as high as ∼57.6% and dark current density as low as 5.39×10⁻⁵ A/cm² at an operation bias of −400 mV at 150 K. As a proof of concept, we have also fabricated a 15-µm pitch middle-format (640×512) pBn MWIR FPA operating at a high temperature up to 185 K, with a 50% cutoff wavelength of ∼4.8 µm at 150 K, which has achieved the high-resolution imager in this material system. This pBn design will provide new opportunities for achieving small-pitch HOT FPA imagers based on InAs/InAsSb T2SLs for thermoelectric cooling, which would greatly reduce the SWaP for MWIR imaging systems.

2. Experiments

The pBn device was grown on a 2-inch Te-doped n-type (10¹⁷ cm⁻³) GaSb substrate using a solid source molecular beam epitaxy (SSMBE) reactor. The growth process started with a 171 nm GaSb buffer layer to smooth out the surface, then a 500 nm heavily n-doped InAs/InAs_0.66Sb_0.34 type-II superlattices (T2SLs)-based bottom contact layer (10¹⁸ cm⁻³) was grown, which was followed by a 3.0 µm-thick undoped absorption region (10¹⁵ cm⁻³) consisted of same period InAs/InAs_0.66Sb_0.34 T2SLs, a 120 nm slightly p-type doped AlAsSb-based electron barrier, and a 300 nm heavily-doped p-GaSb as top contact layer (10¹⁸ cm⁻³) (see, Fig. 1(a)). Silicon (Si) and Beryllium (Be) were used as the n- and p-type dopants, respectively. In particular, the MWIR superlattice absorption region design consists of an ultra-short period of InAs/InAs_0.66Sb_0.34, respectively, per period with a 5.2-µm nominal cut-off wavelength at 150 K. High-resolution X-ray diffraction peaks from the as-grown wafer in Fig. 1(b) clearly identify the epitaxial layers corresponded to GaSb substrate, InAs/InAsSb T2SLs absorber, and AlAsSb barrier, respectively. The thickness of InAs/InAsSb in each superlattice period is determined to be 30.2 Å and 9.2 Å, with a total thickness of ∼3.9 nm per-period, which is well consistent with the epi-design. Figure 1(c) shows the cross-sectional transmission electronic microscopic (TEM) image of InAs/InAsSb strained-layer T2SLs absorber. The bright and clear spots in the bright-field transmission electron diffraction pattern of the InAs/InAsSb strained-layer T2SLs absorber suggest high quality as-grown materials, as presented in the inset of Fig. 1(c). The as-grown wafer for the MWIR photodetectors and FPA processing exhibited excellent surface morphology as measured by atomic force microscopy (AFM), with a root mean square (RMS) roughness of 3.0 Å over an area of 10×10 µm².

 

Fig. 1. (a) Schematic diagram depicting the pBn device design and (b) High-resolution X-ray diffraction of the as-grown wafer. (c) Cross-sectional transmission electronic microscopic image of short-period InAs/InAsSb strained-layer T2SLs absorber. Inset shows the bright field transmission electron diffraction pattern of the InAs/InAsSb strained-layer T2SLs absorber.

Download Full Size | PDF

A 640 × 512 middle-format focal plane array (FPA) with the pixel size of 12 µm was designed to be integrated with a 15 µm-pitch read-out integrated circuit (ROIC). As a reference, an array of circular single-element photodetectors with diameters of 380 µm was processed together with the FPA device. Detailed descriptions about the FPA fabrication process can be found in our previous study. [25] As the processing was over, the FPA was mounted on an 84-pin leadless ceramic chip carrier, wire-bonded and loaded in a dewar camera system for imaging test. Example images were taken by the FPA using an f/2.0 lens, and integration time was adjusted throughout all operation temperatures as increasing from 77 K to 185 K for the ability to accommodate the higher frame rates and for the dynamic range of the camera. In detail, integration times for taking the IR pictures in this work were 1.12 ms, 0.51 ms, 0.32 ms, and 0.2 ms at 77 K, 120 K, 150 K, and 185 K, respectively. The electrical and optical characterization for the pBn single-element photodetectors was carried out with the front-side illumination configuration. Dark-current characteristics was measured using a circular single-element photodetector with a diameter of 380 µm covered by a copper cold shield at different temperatures. For the optical test, a calibrated 500 ^°C blackbody was used as the IR light source and Fourier transform infrared (FTIR) spectrometer was used to obtain the absorption spectra. The optical response was measured from a 380 µm photodetector with an optical aperture of 210 µm. Saturated quantum efficiency (QE) as a function of applied bias voltage was evaluated based on the output from the lock-in amplifier when the device was illuminated by the calibrated blackbody.

3. Results and discussion

Fig. 2(a) presents the dark current density versus applied bias voltage characteristics of a circular single-element photodetector with a diameter of 380 µm as temperature increasing from 90 K to 300 K. Under −400 mV applied bias, the photodetector at 150 K exhibits a dark current density of 5.39×10⁻⁵ A/cm², whereas at T = 200 K, the dark current density is 2.73×10⁻³ A/cm². Figure 2(b) presents the Arrhenius plot of the dark current density versus inverse temperature (1000/T) from 100 K to 300 K at an operation bias of −400 mV. A clear transition in the dark current mechanisms can be identified with increasing temperatures from G-R to diffusion. The violet and red dashed lines represent the diffusion and G-R limits, respectively. By using the formula J_diff ∼ T³exp(-E_a/K_bT), the associated activation energy (E_a) can be calculated as ∼206 meV, which is fairly close to the bandgap of as-designed InAs/InAs_0.66Sb_0.34 strained-layer T2SLs obtained from 100% cut-off wavelength at 150 K (∼238 meV). Here, J_diff, K_b, and T represent the dark current caused by diffusion, Boltzmann constant and the operating temperature, respectively. In addition, the dominant dark current for the pBn photodetector is caused by diffusion at operating temperatures above 170 K, while G-R current dominates below this temperature, whereby the cross-over temperature (T₀), defined as the transition point from G-R current to diffusion current, can be estimated as T₀= 170 K. It’s noteworthy that, in the pBn devices, although the depletion area electrical field mostly falls in the wide bandgap barrier regions, the G-R current seems evitable. [26] This G-R current can hopefully be further suppressed by adopting a n-doped barrier design in the pBn device structure, confirmed by theoretical predictions [27], which suggests room of improvements in the device design.

 

Fig. 2. (a)Typical temperature-dependent dark current density characteristics and (b) Arrhenius plot of the dark current density under an applied bias of −400 mV for a circular 380 µm single element photodetector.

Download Full Size | PDF

Figure 3 shows the saturated quantum efficiency spectrum for the processed pBn MWIR photodetector at 150 K and −400 mV. The dip emerging at 4.2 µm in the spectrum corresponds to the CO₂ absorption in the air. [28] The 50% cut-off wavelength is around 4.8 µm at 150 K, and the saturated peak quantum efficiency reaches as high as 57.6% at 4.4 µm. It slightly increases to 62.2% at 185 K, due to the absorber energy bandgap narrowing, as shown in the inset of Fig. 3 (right). A spectral peak responsivity (@-400 mV) emerging at 4.4 µm reach 1.89 and 2.04 A/W at 150 K and 185 K, respectively. Particularly, we should point out that, with the short-period of InAs/InAsSb strained-layer T2SLs, cut-off in the optical response seems to be faster as compared to its counterparts, as it may be caused by the wave-function of the minority carriers (holes) over-laps more with that of electrons. [21,26] Moreover, as shown in the inset of Fig. 3 (left), quantum efficiency at around 4.4 µm exhibits a saturation behavior as a function of applied bias voltage in a wide range of temperatures, which indicates efficient blocking of carriers by the barrier. The turn-on voltage for this pBn device is about −300 mV, defined as the bias where the 90% of peak quantum efficiency has reached, which is slightly higher than nBn devices, suggesting band alignment for the AlAsSb barrier and the absorbers needs to be optimized.

 

Fig. 3. Quantum efficiency spectra of the photodetector device at 150 K and reverse bias of −400 mV. Inset shows the temperature dependent quantum efficiency at 4.4 µm at different reverse bias.

Download Full Size | PDF

After performing electrical and optical characterization, the shot-noise limited specific detectivity (D*) at peak responsivity at 4.4 µm was calculated according to the equation depicted as follows:

 

Fig. 4. Specific detectivity (D*) spectra of a photodetector device at 150 K and −400 mV applied bias in front-side illumination configuration without any anti-reflection coating. Inset shows specific D* as a function of temperature at 4.4 µm. Dashed line shows the f/2 and 2π field-of-view (FOV) background-limited performance (BLIP) D* at 3-5 µm.

Download Full Size | PDF

Figures 5(a)–5(d) present example images taken with an pBn MWIR FPA as the temperature varying from 77 K up to 185 K. The images maintain a good quality until temperatures raised beyond 150 K, without exhibiting any obvious signs of degradation. Figure 5(e) shows temperature-dependent NETD and the pixel operability for an InAs/InAsSb T2SLs-based MWIR FPA measured with a black body at a temperature of 30 °C and an f/2 optics. At 77 K and 120 K, the FPA device achieves noise equivalent temperature difference (NETD) values as low as 15.1 mK and 27.8 mK, respectively. At 150 K, the FPA device exhibits NETD of ∼39.5 mK. As the temperature further goes up to 185 K, the NETD rapidly degrades to 84.9 mK, attributed to the increase in the detector pixel’s dark current density. As discussed before, this effect can be mitigated by a suppression of G-R dark current with an optimal doping of AlAsSb-based barrier. Meanwhile, by using a standard two-point non-uniformity correction and standard criteria for the identification of defective pixels, the pixel operability at 150 K reaches 99.6% and slightly declines to 99.4% at 185 K. It is noteworthy that the pBn FPA performance in the aspect of NETD versus temperature is not as stable as the nBn device, as the higher G-R current is one of the contributing factors. Another fact is that FPA pixel size after the mesa definition in this study is only about 11 µm, which is much smaller (3-4 times smaller), leading to a severer surface leakage that hopefully solved by better surface passivation [21].

 

Fig. 5. (a-d) Example images taken with InAs/InAsSb strained-layer T2SLs MWIR FPA operating from 77 K to 185 K. (e) Noise equivalent temperature difference (NETD) and FPA pixel operativity measured from 77 K to 185 K. An f/2.0 optics is used and the integration time is adjusted at all data points to avoid capacitor saturation.

Download Full Size | PDF

4. Conclusion

In this work, 15 µm-pitch 640 × 512 middle-format pBn mid-wavelength infrared (MWIR) high operating temperature (HOT) focal plane array based on InAs/InAsSb strained-layer type-II superlattices has been successfully demonstrated with an operation temperature up to 185 K, by suppressing Auger recombination with a design of short superlattice period. With a high operability of ∼99.6%, the FPA device exhibits ∼15.1 mK and ∼27.8 mK NETD at 77 K and 120 K, accordingly. As temperature further increased to 150 K and 185 K, the FPA NETD values decline to 39.5 mK and ∼84.9 mK, with a specific detectivity of 4.43 × 10¹¹ and 1.33 × 10¹¹ Jones, respectively. This work presents a new thought in the device structure and absorber design based on antimonide Ga-free T2SLs for building up HOT MWIR FPA imagers that can be massively produced, which paves the way for fast, low-cost, compact IR detection systems for widespread imaging applications.