Article | Open | Published:

# Dark current reduction in microjunction-based double electron barrier type-II InAs/InAsSb superlattice long-wavelength infrared photodetectors

## Abstract

Microjunction InAs/InAs1−xSbx type-II superlattice-based long-wavelength infrared photodetectors with reduced dark current density were demonstrated. A double electron barrier design was employed to reduce both bulk and surface dark currents. The photodetectors exhibited low surface leakage after passivation with SiO2, allowing the use of very small size features without degradation of the dark current. Fabricating microjunction photodetectors (25 × 25 µm2 diodes with 10 × 10 µm2 microjunctions) in combination with the double electron barrier design results in a dark current density of 6.3 × 10−6 A/cm2 at 77 K. The device has an 8 µm cut-off wavelength at 77 K and exhibits a quantum efficiency of 31% for a 2 µm-thick absorption region, which results in a specific detectivity value of 1.2 × 1012 cm·Hz1/2/W.

## Introduction

After InAs/Ga(In)Sb type-II superlattices (T2SLs) were devised by Sai-Halasz et al.1 and used by Sakaki et al. (ref.2) to make infrared photodetectors, they have been optimized3,4 and have started to challenge mercury cadmium telluride (MCT), which is the state-of-the-art infrared detection technology. Type-II superlattices have been theorized to have multiple advantages over MCT to make infrared photodetectors. MCT compounds are toxic, relatively costly, and can be inhomogeneous across large wafers as opposed to T2SL photodetectors5. Additionally, the valence and conduction energy bands can be separately engineered thanks to the flexibility of superlattice structures. This is done by changing the thicknesses of InAs and Ga(In)Sb layers. AlSb layers can also be introduced to further enhance this tunability6,7, which allows a bandgap range from 1.5 μm to semi-metals8,9.

Nowadays, dark current reduction in T2SL photodetectors is one of the main challenges to increase the specific detectivity (D*), which is the main figure of merit for photodetectors that combines both optical and electrical performance. These photodetectors still suffer from higher dark current density compare to their MCT-based counterparts. The main limiting dark current mechanisms of long-wavelength infrared (LWIR) T2SL photodetectors are the surface leakage and the generation-recombination (G-R) currents3,10,11,12,13,14,15. The G-R current can be described as follows:

$${{\boldsymbol{I}}}_{{\boldsymbol{G}}-{\boldsymbol{R}}}({\boldsymbol{V}})=\frac{{\boldsymbol{A}}\ast {\boldsymbol{W}}({\boldsymbol{V}})\ast {\boldsymbol{q}}\ast {{\boldsymbol{n}}}_{{\boldsymbol{i}}}}{2\ast {{\boldsymbol{\tau }}}_{0}}$$
(1)

$${{\boldsymbol{n}}}_{{\boldsymbol{i}}}=\sqrt{{{\boldsymbol{N}}}_{{\boldsymbol{c}}}\ast {{\boldsymbol{N}}}_{{\boldsymbol{v}}}}\ast {\bf{e}}{\bf{x}}{\bf{p}}(\frac{-{{\boldsymbol{E}}}_{{\boldsymbol{g}}}}{2\ast {{\boldsymbol{k}}}_{{\boldsymbol{B}}}\ast {\boldsymbol{T}}})$$
(2)

where A is the electrical area of the photodetector, W(V) is the depletion region width under a given applied bias voltage (V), n i is the intrinsic carrier concentration, τ 0 is the carrier lifetime, E g is the bandgap energy of the depletion region, and N c and N v are the density of states in the conduction and valence bands, respectively.

The G-R current inversely depends on the carrier lifetime (τ 0). InAs/InAs1−xSbx T2SLs have recently proven to have longer carrier lifetimes than InAs/Ga(In)Sb T2SLs16, because the presence of GaSb related native defects inside the InAs/Ga(In)Sb superlattice creates Shockley-Read-Hall recombination centers. Therefore, gallium-free T2SLs are beneficial for decreasing the dark current density in T2SL-based photodetectors. Recently, high-performance LWIR photodetectors based on InAs/InAs1−xSbx T2SLs have been demonstrated17,18,19,20.

Equations (1) and (2) also show that the G-R current exponentially decreases with increasing semiconductor bandgap energy inside the depletion region. To reduce G-R current further, semiconductor heterostructures have been used to shift the depletion region from the narrow-bandgap absorption region to a large bandgap barrier (Refs3,11,21). One example of this concept is the nBn device structure, which is able to reduce the G-R current by pushing the depletion region into a large-bandgap electron barrier area22.

In this paper, we start with a similar gallium-free InAs/InAsSb nBn device to Haddadi et al.17. Then we add a second larger-bandgap (Eg > 1 eV) AlAsSb/GaSb superlattice barrier22,23 after the normal MWIR barrier to create a double electron barrier (Fig. 1). It is similar to the structure for the XDBn, which uses two bulk compound barriers24 compared to superlattices in our structure. Using an AlAsSb/GaSb superlattice provides the necessary flexibility to tune the valence band offset while still providing a large energy discontinuity in the conduction band. This additional barrier has a superlattice period of 5/2 monolayers (MLs) of AlAs0.09Sb0.91/GaSb. It maintains the bias dependency below 200 mV. The top n-contact of the original device was then replaced by a 300 nm thick, highly p-doped, GaSb capping layer. We call this a p-type contact double barrier n-type absorption region (CpDBn) structure. The addition of this second barrier pushes part of the depletion region into the larger bandgap region, decreasing the G-R current. The larger bandgap barrier also helps suppress surface band-bending and associated surface leakage.

The G-R current is not only proportional to the width remaining in the narrow bandgap portion of the depletion region, but also to the depletion region area (I G-R A * W(V)). Due to the reduced surface leakage of our large bandgap double electron barrier, it is now possible to reduce the area of the diode. However, reducing the device area also reduces collection of the optical signal. We can create a microjunction device structure by only etching to the level of the depletion region25. The electrical area can be small while the effective optical area is larger. Thus, the G-R current is theoretically decreased by the ratio of the mesa area over the junction area (Fig. 2).

## Results and Discussion

The CpDBn structure was grown with a solid-source molecular beam epitaxy (SSMBE) system on an n-type Te-doped GaSb wafer. The growth was started by a thin GaSb layer followed by a 500 nm thick InAs0.91Sb0.09 etch stop layer. It continued with a 500 nm thick bottom n-contact, then a 2 μm thick absorption region with a superlattice period comprising 30/10 MLs of InAs and InAs0.50Sb0.50, respectively. Next a 500 nm thick MWIR barrier was grown, followed by the newly designed 300 nm thick AlAs0.09Sb0.91/GaSb superlattice barrier, and, finally, a 300 nm thick p-doped GaSb capping layer.

A single element test chip (A) with full-etched photodetector areas ranging from 10 × 10 to 400 × 400 µm2 was processed using the inductive couple plasma (ICP) dry etching technique. Top and bottom contacts were deposited and the diodes were then passivated with SiO2. Windows to the contacts were opened by etching the SiO2 layer with a CF4 plasma in an electron cyclotron resonance reactive ion etching (ECR-RIE) system. Then metal path traces were deposited on the small-size diodes to wire-bond them, since their contact area is too small for direct wire-bonding. The photodetectors were mounted on a 68-pin chip leadless ceramic chip carrier (LCCC) then wire-bonded and cooled down inside a liquid helium cryostat. An unpassivated single element test chip was also processed to measure the optical performance without influence from the SiO2 layer on the optical performance. At 77 K, the single element photodetectors exhibit an 8 μm cut-off wavelength and a peak quantum efficiency (QE) value of ~32% at ~7.3 μm (Fig. 3a) for a 2 µm thick absorption region. Test chip A photodetectors have a dark current density (Fig. 3b) of 1.5 × 10−5 A/cm2 under Vb = −140 mV, the bias voltage in which the quantum efficiency spectrum reaches its saturation point. Test chip A has a 3.1 × 107 Ω·cm sidewall surface resistivity (Fig. 3c). Having such a high surface resistivity enables the use of small dimensions in the microjunction fabrication without significant surface leakage current degrading the device.

A microjunction test chip (B) with various sizes of microjunction was also processed from the same wafer using the same process and passivation, with an additional ICP dry etching step to create the microjunctions. Two parameters in the design of the microjunction must be considered: The ratio of the mesa area over the microjunction area and the distance between the microjunction and the outside collection area. The distance is important because the main mechanism for the lateral collection is diffusion; the diffusion length will limit the lateral collection of the photo-carriers or may require extra bias voltage to collect carriers via a drift mechanism. However, applying additional bias voltage will increase the dark current. Therefore, a compromise will have to be made. With a 25 × 25 µm2 mesa, 5 × 5 µm2, 10 × 10 µm2 and 15 × 15 µm2 microjunction photodetectors were processed to study which device design has the best overall performance.

As test chip B is passivated with SiO2 like test chip A, the optical performance of these microjunction photodiodes is compared with test chip A to study the bias dependence (Fig. 4) and relative blackbody integrated photo-response. All sizes of microjunction diodes reach full quantum efficiency, however, the bias at which this occurs (bias dependence) increases when the microjunction area becomes smaller (Fig. 4a). The photodetector needs a higher bias voltage to have lateral collection of all photo-carriers.

The electrical performance of the different microjunction photodiodes was measured with a cold shield. As a result, 25 × 25 µm2 mesa/10 × 10 µm2 microjunction diodes have the best overall performance with a dark current density of 6.3 × 10−6 A/cm2 (Fig. 5a) at Vb = −180 mV quantum efficiency saturation bias (Fig. 4b). Thus, these diodes’ dark current density was decreased 2.5 times compared to test chip A’s electrical performance. The dark current of the microjunction photodiode is lower at smaller bias voltages. However, the surface leakage is higher at larger bias voltages because the junction size is smaller. The special detectivity is 1.2 × 1012 cm·Hz1/2/W for microjunction test chip B (Fig. 5b) and is 1.6 higher than test chip A.

The 25 × 25 µm2 mesa/15 × 15 µm2 microjunction diodes showed almost as good performance with 6.5 × 10−6 A/cm2 at Vb = −160 mV, however, the 25 × 25 µm2 mesa/5 × 5 µm2 microjunction diodes have 1.3 × 10−5 A/cm2 at Vb = −250 mV because the lateral collection requires a significantly larger applied bias voltage.

## Conclusion

In summary, we reported the use of a new p-type contact double barrier n-type absorption region (CpDBn) photodetector structure to improve passivation quality and suppress the G-R current. The double barrier consists of a large bandgap (>1 eV) AlAsSb/GaSb barrier and an undoped MWIR barrier to reduce the G-R current by shifting a portion of the depletion region into this second barrier. The large bandgap barrier suppresses the surface leakage current and the G-R current. This surface leakage suppression allows the use of a microjunction that further improves the electrical performance by reducing the bulk dark current originating from the junction area, especially the G-R current. A study of the bias dependence of optical performance of 5 × 5 µm2, 10 × 10 µm2 and 15 × 15 µm2 area microjunction photodiodes for a 25 × 25 µm2 mesa area combined with electrical measurement. The 25 × 25 µm2 mesa/10 × 10 µm2 microjunction area photodiode structure exhibits a dark current reduction of 2.5 times at the quantum efficiency saturation bias. The structure increases the special detectivity value by 1.6 times to 1.2×1012 cm·Hz1/2/W.

## Methods

### Growth

To grow the structure, we used a molecular beam epitaxy (MBE) system that is equipped with group III SUMO cells and group V valved crackers. Silicon and beryllium were introduced in the superlattice as n-type and p-type dopants, respectively. The structure was grown on tellurium (Te)-doped (001) GaSb n-type substrate.

### Fabrication

The processing of single element test chips can be separated in five steps: Mesa isolation, metal contacts deposition, passivation, window opening, and path trace metal deposition. After removing residue from the surface, a standard lithography defines mesa shapes and sizes. The sizes vary from 10 × 10 µm2 to 400 × 400 µm2. The etching is performed first with an inductive coupled plasma reactive ion etching (ICP-RIE) system with a BCl3/Ar gas mixture. After the dry etching, no wet etching is used as it has been proven that dry etching only creates less surface leakage for passivated LWIR samples. A solvent-based cleaning procedure in an ultrasonic bath is then performed to clean the sample thoroughly. Then a second lithography is done to define the top and bottom contacts, which are deposited in an electron beam evaporator and consist of a layer of titanium followed by a layer of gold. This metallization was chosen to create ohmic contacts. Lift-off is then performed and the same cleaning procedure is used to clean the diodes sidewalls. A 1.5 µm thick SiO2 layer is deposited on the sample using plasma enhanced chemical vapor deposition (PECVD). Then another lithography is performed to define a window to the contact. An electron cyclotron resonance reactive ion etching (ECR-RIE) system is used with a CF4 plasma to etch the SiO2 layer. As some of the contact sizes are very small, direct wire bonding is not possible and an additional step is necessary: A path trace metal is deposited to bridge the top contact to an adjacent larger pad for wire bonding.

Microjunction test chips undergo a process similar to the passivated single element test chips. However, the microjunction test chip has an extra etching step: The first etching step defines the diode microjunction and the second etching defines the mesa. The microjunction sidewalls are covered by photoresist during the mesa etching and are therefore protected. Then metal contact deposition, passivation, and window opening are performed in the same way as for the passivated single element test chip. As focal plane arrays typically have pixel areas around 25 × 25 µm2, diodes were fabricated with this mesa area and microjunction areas of 5 × 5 µm2, 10 × 10 µm2 and 15 × 15 µm2. Since the top contact sizes are very small, direct wire bonding is not possible. The same path trace metal deposition is used to allow wire bonding.

### Device testing

The processed test chips are mounted onto a 68-pin leadless ceramic chip carrier (LCCC) for electrical and optical characterization. The connections from the top and bottom contacts of the sample to the bond pads of the LCC were made with a gold wire bonder. After bonding, the samples were loaded into liquid helium cryostat that has the capability to measure from 4 K to 300 K. The test chips are connected to a switching matrix that allows connection to a semiconductor parameter analyzer to measure the electrical properties, a fourier transform infrared spectroscopy (FTIR) system to measure the optical response, or a lock-in amplifier to measure the calibrated blackbody integrated response.

Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

## References

1. 1.

Sai‐Halasz, G. A., Tsu, R. & Esaki, L. A new semiconductor superlattice. Applied Physics Letters 30, 651–653 (1977).

2. 2.

Sakaki, H., Chang, L. L., Sai-Halasz, G. A., Chang, C. A. & Esaki, L. Two-dimensional electronic structure in InAs-GaSb superlattices. Solid State Communications 26, 589–592 (1978).

3. 3.

Nguyen, B.-M., Hoffman, D., Delaunay, P.-Y. & Razeghi, M. Dark current suppression in type II InAs∕GaSb superlattice long wavelength infrared photodiodes with M-structure barrier. Applied Physics Letters 91, 163511 (2007).

4. 4.

Nguyen, B.-M., Hoffman, D., Huang, E. K.-W., Delaunay, P.-Y. & Razeghi, M. Background limited long wavelength infrared type-II InAs/GaSb superlattice photodiodes operating at 110 K. Applied Physics Letters 93, 123502 (2008).

5. 5.

Nguyen, B. M., Chen, G., Hoang, M. A. & Razeghi, M. Growth and Characterization of Long-Wavelength Infrared Type-II Superlattice Photodiodes on a 3-in GaSb Wafer. IEEE Journal of Quantum Electronics 47, 686–690 (2011).

6. 6.

Nitta, J., Munekata, H., Razeghi, M. & Nguyen, B.-M. Proceedings of the 14th International conference on Narrow Gap Semiconductors and SystemsBand gap tunability of Type II Antimonide-based superlattices. Physics Procedia 3, 1207–1212 (2010).

7. 7.

Hoang, A. M., Chen, G., Haddadi, A. & Razeghi, M. Demonstration of high performance bias-selectable dual-band short-/mid-wavelength infrared photodetectors based on type-II InAs/GaSb/AlSb superlattices. Applied Physics Letters 102, 011108 (2013).

8. 8.

Mohseni, H. et al. Very long wavelength infrared type-II detectors operating at 80 K. Applied Physics Letters 77, 1572–1574 (2000).

9. 9.

Hoang, A. M., Chen, G., Haddadi, A., Abdollahi Pour, S. & Razeghi, M. Demonstration of shortwavelength infrared photodiodes based on type-II InAs/GaSb/AlSb superlattices. Applied Physics Letters 100, 211101 (2012).

10. 10.

Hood, A. et al. Near bulk-limited R0A of long-wavelength infrared type-II InAs∕GaSb superlattice photodiodes with polyimide surface passivation. Applied Physics Letters 90, 233513 (2007).

11. 11.

Vurgaftman, I. et al. Graded band gap for dark-current suppression in long-wave infrared W-structured type-II superlattice photodiodes. Applied Physics Letters 89, 121114 (2006).

12. 12.

Wei, Y. et al. 309–315.

13. 13.

Bogdanov, S., Nguyen, B.-M., Hoang, A. M. & Razeghi, M. Surface leakage current reduction in long wavelength infrared type-II InAs/GaSb superlattice photodiodes. Applied Physics Letters 98, 183501 (2011).

14. 14.

Mou, S. et al. Surface channel current in InAs∕GaSb type-II superlattice photodiodes. Journal of Applied Physics 102, 066103 (2007).

15. 15.

Chen, G. et al. Surface leakage investigation via gated type-II InAs/GaSb long-wavelength infrared photodetectors. Applied Physics Letters 101, 213501 (2012).

16. 16.

Steenbergen, E. H. et al. Significantly improved minority carrier lifetime observed in a long-wavelength infrared III-V type-II superlattice comprised of InAs/InAsSb. Applied Physics Letters 99, 251110 (2011).

17. 17.

Haddadi, A., Chen, G., Chevallier, R., Hoang, A. M. & Razeghi, M. InAs/InAs1−xSbx type-II superlattices for high performance long wavelength infrared detection. Applied Physics Letters 105, 121104 (2014).

18. 18.

Haddadi, A. et al. High-performance short-wavelength infrared photodetectors based on type-II InAs/InAs1−xSbx/AlAs1−xSbx superlattices. Applied Physics Letters 107, 141104 (2015).

19. 19.

Hoang, A. M., Chen, G., Chevallier, R., Haddadi, A. & Razeghi, M. High performance photodiodes based on InAs/InAsSb type-II superlattices for very long wavelength infrared detection. Applied Physics Letters 104, 251105 (2014).

20. 20.

Kim, H. S. et al. Long-wave infrared nBn photodetectors based on InAs/InAsSb type-II superlattices. Applied Physics Letters 101, 161114 (2012).

21. 21.

Callewaert, F., Hoang, A. M. & Razeghi, M. Generation-recombination and trap-assisted tunneling in long wavelength infrared minority electron unipolar photodetectors based on InAs/GaSb superlattice. Applied Physics Letters 104, 053508 (2014).

22. 22.

Maimon, S. & Wicks, G. W. nBn detector, an infrared detector with reduced dark current and higher operating temperature. Applied Physics Letters 89, 151109 (2006).

23. 23.

Martyniuk, P. & Rogalski, A. Performance limits of the mid-wave InAsSb/AlAsSb nBn HOT infrared detector. Optical and Quantum Electronics 46, 581–591 (2014).

24. 24.

De Lyon, T. J., Rajavel, R. D. & Sharifi, H. (Google Patents, 2014).

25. 25.

Aifer, E. H. et al. Shallow-Etch Mesa Isolation of Graded-Bandgap “W”-Structured Type II Superlattice Photodiodes. Journal of Electronic Materials 39, 1070–1079 (2010).

## Acknowledgements

The authors would like to acknowledge the Walter P. Murphy fellowship, DARPA, the Army Research Laboratory (ARL), the Air Force Research Laboratory, and NASA for their support, interest, and encouragement, as well as their colleagues for their help and advice.

## Author information

### Affiliations

1. #### Center for Quantum Devices, Department of Electrical Engineering and Computer Science, Northwestern University, Evanston, 2220 Campus Drive, RM 4051, Evanston, IL, 60208-0893, USA

• Romain Chevallier
•  & Manijeh Razeghi

### Contributions

Romain Chevallier conceived and designed this project, performed the devices fabrication and measurement and wrote the paper. Abbas Haddadi grew the material with SSMBE and performed structural characterization. Manijeh Razeghi supervised the project. All authors analyzed the data, revised and modified the manuscript.

### Competing Interests

The authors declare that they have no competing interests.

### Corresponding author

Correspondence to Manijeh Razeghi.