CHIN. PHYS. LETT. Vol. 26, No. 2 (2009) 028102
GaAs Based InAs/GaSb Superlattice Short Wavelength Infrared Detectors Grown by Molecular Beam Epitaxy * TANG Bao(汤宝), XU Ying-Qiang(徐应强)** , ZHOU Zhi-Qiang(周志强), HAO Rui-Ting(郝瑞亭), WANG Guo-Wei(王国伟), REN Zheng-Wei(任正伟), NIU Zhi-Chuan(牛智川) State Key Laboratory for Superlattices and Microstructures, Institute of Semiconductors, Chinese Academy of Sciences, Beijing 100083
(Received 13 November 2008) InAs/GaSb superlattice (SL) short wavelength infrared photoconduction detectors are grown by molecular beam epitaxy on GaAs(001) semi-insulating substrates. An interfacial misfit mode AlSb quantum dot layer and a thick GaSb layer are grown as buffer layers. The detectors containing a 200-period 2 ML/8 ML InAs/GaSb SL active layer are fabricated with a pixel area of 800×800 𝜇m2 without using passivation or antireflection coatings. Corresponding to the 50% cutoff wavelengths of 2.05 𝜇m at 77 K and 2.25 𝜇 m at 300 K, the peak detectivities of the detectors are 4 × 109 cm·Hz1/2 /W at 77 K and 2 × 108 cm·Hz1/2 /W at 300 K, respectively.
PACS: 81. 05. Ea, 81. 15. Hi, 78. 67. Pt Short wavelength infrared (IR) detectors in the 2.0–3.0-µm range have attracted a great deal of attention due to their potential applications, including spatial remote sensing, infrared imaging, missiles guiding, biology molecules spectroscopy, etc. Compared with HgCdTe and InGaAs of near-infrared materials, the InAs/GaSb SLs are predicted to have much lower Auger recombination rates and smaller tunnelling currents due to their type-II band alignments and large electron effective masses, which are especially important for realizing high-performance interband IR detectors at room temperature. Furthermore, a particular advantage of InAs/GaSb SLs is flexibility in tuning bandgaps, i.e., the operating wavelengths of multibands detectors can be realized by changing the thickness and period of the InAs and GaSb layers. Veryshort-period InAs/GaSb superlattices can be used as near-infrared detectors (from 2–3µm).[1−3] So far, GaSb substrates have been considered as a good choice for growth of lattice-matched InAs/GaSb SLs. However, GaSb substrates are very expensive and ‘undoped’ GaSb has strong absorption in the IR wavelength region due to free-carrier absorption. Therefore, other substrates such as GaAs have been explored for the growth of InAs/GaSb SL devices. In recent years, many works on the growth of GaAs-based mid-infrared[4,5,19] and long wavelength infrared[6,7,20] InAs/GaSb room temperature operating detectors have been reported. However, there have been few reports on short wavelength (between 2.0 µm and 3.0 µm) infrared SL detectors on GaAs substrates. In this Letter, we report GaAs-based short wavelength infrared InAs/GaSb SL detectors grown by molecular beam epitaxy (MBE). The first key issue for the growth of the InAs/GaSb SLs on GaAs substrates
is to produce a high quality buffer layer to solve the problem of nearly 7% lattice mismatch between the GaAs and GaSb, since any roughness leads to inferior lateral transport in the InAs/GaSb superlattices[8] and in-homogeneity of the band gap. In our experiment, we used an AlSb/GaSb layer as the buffer on semi-insulating GaAs (001) substrates, and then grew high-quality InAs/GaSb SLs on top of the buffer layer. Finally, photoconduction detectors were fabricated and measured at room temperature and 77K, respectively. The samples were grown by a VG80H MBE system equipped with low temperature As and Sb cells supplying As4 and Sb4 . In this experiment, we still used GaSb as the buffer, but before the growth of the GaSb buffer layer, an AlSb nucleation layer was deposited.[9] The GaSb/AlSb SL barrier was grown above the AlSb nucleation layer to capture the photoexcited carriers. There were 200 period 2 ML/8 ML InAs/GaSb SLs grown on top of the GaSb buffer layer at a substrate temperature of about 410 ± 10∘ C measured by a standard thermocouple and calibrated with the (1 × 3) to (2 × 5) GaSb surface reconstruction transition.[10] The flux ratio of group V to III was similar to the previously reported values.[11] The intended thickness of InAs and GaSb layers was controlled by the low growth rate of 0.28 and 0.7 ˚ A/s, respectively. As shown in Fig. 1(a), the shape of the AlSb IMF QD is flat and broad, and the QD height and width are 5 and 30 nm, respectively. Balakrishnan et al. proved that the standard Sb-rich growth conditions result in a growth mode characterized by the laterally propagating (90∘ ) interfacial misfit (IMF) dislocations confined to the episubstrate interface when there is a greater quantity of larger adatom (antimony) than
* Supported by the National Natural Science Foundation of China under Grant Nos 60607016 and 60625405, and the National Basic Research Programme of China under Grant No 2007CB936304. ** To whom correspondence should be addressed. Email:
[email protected] c 2009 Chinese Physical Society and IOP Publishing Ltd ○
028102-1
CHIN. PHYS. LETT. Vol. 26, No. 2 (2009) 028102
small adatom (aluminium) on the growth surface.[12] The laterally epitaxy of AlSb can restrain dislocation propagating into the GaSb and SL epilayers. In addition, the AlSb buffer acts as an insulating layer to reduce noise between pixels. The RHEED image (not shown) shows patterns corresponding to relaxed islands that are similar to the result of Balakrishnan. [0 ]
01
(b)
(a)
As In b aS G
AlSb IMF-QD on GaAs
1 nm
Fig. 1. (a) Dark-field HRTEM images showing an AlSb IMF QD on the GaAs substrate. (b) Dark-field HRTEM images showing the InAs (dark) and GaSb (light) layers in the SL structures of 2 ML/8 ML for the widths of the InAs and GaSb, respectively.
for-arsenic replacement due to a combination of anion segregation and background incorporation across the InAs-on-GaSb interface,[13] together with spatially graded indium-for-gallium replacement due to cation segregation across the GaSb-on-InAs interface.[14,15] HRXRD profiles covering the symmetric 004 reflection range are shown in Fig. 2 for a 200 period InAs (2 ML)/GaSb (8 ML) SL. The measured profiles show diffraction peaks originating from the GaAs substrate, the AlSb nucleation layer, the GaSb/AlSb SLs and the strain-relaxed GaSb buffer layer. In addition, well-resolved SL diffraction peaks up to the second order are observed, indicated by 0, ±1, etc. The periodic width of the SL is 32.45 ˚ A according to the structure of InAs (2 ML)/InSb(0.3 ML)/GaSb(8 ML). Thus, InSb thickness exceeds the quantity enough to compensate for the strain. The lattice mismatch to the GaSb buffer is +7.8 × 10−3 . Compared to longperiod SLs, the same deviation of InSb IF thickness for short period SLs will produce larger relative error. Thus controlling the InSb IF thickness of short-period SLs accurately is an important and challenging work.
Fig. 3. Absorption spectra of the InAs(2 ML)/GaSb (8 ML) SLs at 300 K. Fig. 2. HRXRD scan of a 200-period InAs (2 ML)/GaSb (8 ML) SLs.
Figure 1(b) shows the dark-field image of the 30 ˚ A SLs. The individual layers within the SLs are revealed in this image, showing little roughness in the interfaces and small fluctuations in thickness. For our MBE growth using As4 as the group-V element, since the surface mobility and activity of As4 adatom is lower than As2 , a higher V/III flux ratio has to be applied to maintain the surface smoothness during the InAs layer growth. Then, a relatively higher As partial pressure leads to a partial replacement of Sb by As at the interfaces when growing GaSb layers and incorporates the As atoms into the GaSb layer, which is considered to be the dominant reason of the interfaces disorder as far as our uncracked As source system was concerned. Automatically, the detectivity may drastically decrease due to such a high level of interfacial disorder. Other possible reasons are spatially graded antimony-
Figure 3 displays the absorption spectra of InAs (2 ML)/GaSb(8 ML) SLs measured at 300 K. The cutoff wavelengths are 500 meV. There is a step around 750 meV, which should be attributed to the absorption of GaSb layer. This SL system involving midwaves and long wavelengths has been modelled using a variety of computational approaches,[16,17] including the popular envelope-function approximation (EFA) method. The band gaps of InAs/GaSb SLs calculated by different researchers have great differences.[18] Interfaces and the spin-orbit band have a larger effect on the band gaps of short period SLs than those of long-period SLs. Thus, to predict the band gap of short-wave SL detectors becomes more difficult. This makes modified computational approaches indispensable. The photoconductor detectors with a pixel area of 800 × 800 µm2 were fabricated by standard lithography and subsequent etching by tartaric acid solution. A 2000 ˚ A Ti/Au film was sputtered on the SLs to form
028102-2
CHIN. PHYS. LETT. Vol. 26, No. 2 (2009) 028102
ohmic contacts. For electrical measurements, the samples were indium bonded to a copper heatsink and attached to the cold finger of a liquid nitrogen cryostat. The spectral response of the devices was measured with a fourier transform infrared (FTIR) system. The detectivity of the short-wavelength detector based on InAs(2 ML)/GaSb(8 ML) SL was tested with 1.2 mA current under a blackbody temperature of 900 K.
buffer layers. The peak detectivity of the photoconduction detectors at 50% cutoff wavelength 2.25 µm is 2 × 108 cm·Hz1/2 /W at 300 K and 4 × 109 cm·Hz1/2 /W at 77 K. We can use cracked As source to reduce the V/III flux ratio for producing a better interface and greatly enhance the detectivity when fabricating the detector devices in structure of PIN together with introducing the passivation and antireflection layer. This InAs/GaSb superlattices detector system surely have a great potential and will have a much better performance after optimizing the growth condition and device structure.
References
Fig. 4. Responsivity spectra of the InAs 2 ML/GaSb 8ML superlattice at 77 and 300 K. Inset: schematic of the InAs/GaSb SL photoconductive detector.
Figure 4 shows the photoresponse spectra of detector at 77 K and room temperature. The cutoff wavelength (defined as the point where the photoresponse drops to 50% of the maximum value) is 2.05 µm at 77 K and 2.25 µm at 300 K, respectively. The blackbody peak detectivity is 2 × 108 cm·Hz1/2 /W at room temperature and 4 × 109 cm·Hz1/2 /W at 77 K. This appears lower compared with other systems (such as MCT) due to two reasons. Firstly, there is no passivation or anti-reflection coating, which is clearly helpful for increasing the detectivity by reducing the leak currents and enhancing the quantum efficiency. Secondly, 0.65 µm is not sufficiently thick to absorb all the light in terms of absorption layer. There is also a GaSb absorption step near 1.7 µm in the photo response spectra. GaAs based short wavelength infrared 2 MlL/8 ML InAs/GaSb superlattices photoconduction detectors grown by molecular beam epitaxy were fabricated for the first time. An interfacial misfit mode AlSb quantum dot layer could restrain dislocations propagating into the GaSb and then the InAs/GaSb superlattices epilayer. Short-period InAs/GaSb superlattices were then grown on high quality GaSb/AlSb
[1] Johnson J L, Samoska L A., Gossard A C., Merz J, Jack M D, Chapman G R, Baumgratz B A, Kosai K and Johnson S M 1996 J. Appl. Phys. 80 1116 [2] Fuchs F, Weimar U, Pletschen W, Schmitz J, Ahlswede E, Walther M, Wagner J and Koidl P 1997 Appl. Phys. Lett. 71 3251 [3] Rehm R, Walther M, Schmitz J, Fleißner J, Fuchs F, Ziegler J and Cabanski W 2005 SPIE 5957 595701 [4] Zhang X B, Ryou J H, Dupuis R D, Petschke A, Mou S, Chuang S L, Xu C and Hsieh K C 2006 Appl. Phys. Lett. 88 072104 [5] Rodriguez J B, Christol P, Ouvrard A, Chevrier F, Grech P and Joullie A 2005 Electron. Lett. 41 362 [6] Mohseni H, Tahraoui A, Wojkowski J, Razeghi M, Brown G J, Mitchel W C and Park Y S 2000 Appl. Phys. Lett. 77 1572 [7] Mohseni H, Wojkowski J, Razeghi M, Brown G and Mitchel W 1999 IEEE J. Quantum Electron. 35 1041 [8] Blank H, Thomas M, Wong K and Kroemer H 1996 Appl. Phys. Lett. 69 2080 [9] Hao R T, Xu Y Q, Zhou Z Q, Ren Z W, Ni H Q, He Z H and Niu Z C 2007 J. Phys. D: Appl. Phys. 40 1080 [10] Bracker A S, Yang M J, Bennett B R, Culbertson J C and Moore W J 2000 J. Cryst. Growth 220 384 [11] Hao R T, Xu Y Q, Zhou Z Q, Ren Z W, Ni H Q, He Z H and Niu Z C 2007 J. Phys. D: Appl. Phys. 40 6690 [12] Balakrishnan G, Tatebaysshi J, Khoshakhlagh A, Huang S H, Jallipalli A, Daswon L R and Huffaker D L 2006 Appl. Phys. Lett. 89 161104 [13] Steinshnider J, Harper J, Weimer M, Lin C H, Pei S S and Chow D H 2000 Phys. Rev. Lett. 85 4562 [14] Steinshnider J, Weimer M, Kaspi R and Turner G W 2000 Phys. Rev. Lett. 85 2953 [15] Renard C, Marcadet X, Massies J, Pr´ evot I, Bisaro R and Galtier P 2003 J. Cryst. Growth 259 69 [16] Dente G C and Tilton M L 1999 J. Appl. Phys. 86 1420 [17] Hall K C, Gundogdu K, Altunkaya E, Lau W H, Flatte M E, Boggess T, Zinck J J, arvos-Carter W B and Skeith L 2003 Phys. Rev. B 68 115311 [18] Szmulowicz F, Haugan H, and Brown G J 2004 Phys. Rev. B 69 155321 [19] Zhen H L, Li N, Xiong D Y, Zhou X C, Lu W and Liu H C 2005 Chin. Phys. Lett. 22 1806
028102-3
