Invited Paper
High Operating Temperature MWIR photon detectors based on Type II InAs/GaSb superlattice
Manijeh Razeghia), Binh-Minh Nguyen, Pierre-Yves Delaunay, Siamak Abdollahi Pour, Edward Kwei-wei Huang, Paritosh Manukar, Simeon Bogdanov, Guanxi Chen. Center for Quantum Devices, Department of Electrical Engineering and Computer Science, Northwestern University, Evanston, Illinois 60208
Abstract Recent efforts have been paid to elevate the operating temperature of Type II InAs/GaSb superlattice Mid Infrared photon detectors. Optimized growth parameters and interface engineering technique enable high quality material with a quantum efficiency above 50%. Intensive study on device architecture and doping profile has resulted in almost one order of magnitude of improvement to the electrical performance and lifted up the 300K-background BLIP operation temperature to 166K. At 77K, the ~4.2 μm cut-off devices exhibit a differential resistance area product in excess of the measurement system limit (106 Ohm.cm2) and a detectivity of 3x1013cm.Hz1/2/W. High quality focal plane arrays were demonstrated with a noise equivalent temperature of 10mK at 77K. Uncooled camera is capable to capture hot objects such as soldering iron. Keywords: Type II superlattice, InAs/GaSb, M-structure, photodetectors, MWIR, focal plane arrays.
INTRODUCTION: TYPE II ANTIMONIDE BASED SUPERLATTICES The Type II InAs/GaSb superlattice was first investigated by Sakaki and Esaki in the 1970s1 and proposed for infrared detection applications by Smilth and Mailhiot in late 1980s2. The superlattice system consists of the materials in the 6.1A family (InAs/GaSb/AlSb) (Figure 1a) which are closely lattice matched to each other and have a type II misalligned band offsets between InAs and GaSb. The conduction band level of InAs is lower than the valance band of GaSb, creating a spatial separation of electrons and holes in InAs/GaSb heterostructure and allowing an effective bandgap varying from 0V to 0.5V in InAs/GaSb superlattices (Figure 1b). Recently, AlSb, the third member of the 6.1A family was incorporated into the conventional InAs/GaSb superlattice to a new superlattice design called M-structure superlattice3. The AlSb layer was inserted in the middle of the GaSb layer, creating potential barriers for both electrons and holes. This potential barrier has been shown to signfificantly enhance the carrier effective mass and provide a more flexible control of the conduction and valence band edges4-6.
a)
Email:
[email protected]
Quantum Sensing and Nanophotonic Devices VII, edited by Manijeh Razeghi, Rengarajan Sudharsanan, Gail J. Brown, Proc. of SPIE Vol. 7608, 76081Q · © 2010 SPIE · CCC code: 0277-786X/10/$18 · doi: 10.1117/12.840422
Proc. of SPIE Vol. 7608 76081Q-1
Figure 1 (a) The energy gap versus the lattice constant of InAs, GaSb, and AlSb compared with other semiconductors. (b) InAs, GaSb, and AlSb energy band lineups. Compared with the state-of-the-art Mecury Cadmium Telluride technology for infrared detection, Type II superlattice, especially M-structure superlattice benefits from a larger effective mass and the flexibility of controlling the energy gap by atomically engineering the superlattice interfaces and layer thicknesses. Using the same techniques, the Auger recombination rate can be significantly reduced7. Despite the potential benefits, Type II superlattice did not really enter the field of infrared detection until the realization of high-quality growth using the state-of-the-art technique of molecular-beam epitaxy (MBE) in the late 1990s. Rapid developments of material quality during the last decade have demonstrated other practical advantages of this material system: covalent-bond stable material, high purity, low defect density, excellent uniformity and high reproducibility. Different photodetector architectures have been utilized to profit from the unique advantages of Type II superlattice, leading to the demonstration of high quality focal plane arrays with comparable performance to the state-of-the-art Mercury Cadmium Telluride technology. In this paper, we will discuss recent efforts in elevating the operating temperature of MWIR Type II superlattice detectors and Focal Plane Arrays (FPA).
SUPERLATTICE DESIGNS Type II InAs/GaSb superlattices are formed by alternating the two constituent InAs and GaSb layers. These two materials do not share any common atom; therefore, the interfaces between them invoke the presence of other materials such as InSb, GaAs or mixed ternaries/quaternaries between them. InSb is ~7% compressively strained on GaSb substrates while GaAs suffers from a tensile strain with a lattice mismatch of ~-7%. Due to such large lattice mismatches, an interface layer as thin as one atomic monolayer (ML) can critically change the average lattice parameter and the crystalline stress of the structure. Naturally, one would think of combining 1 interface layer of InSb and the other interface of GaAs to balance the strain, however, this could lead to a considerable degradation of the optical efficiency of the material. Shown in Figure 2 is the band alignment of two periods of a Type II InAs/GaSb superlattice design with two types of interfaces: 1 ML of InSb or 1 ML of GaAs. In Type II superlattices, the optical transition due to the interband recombination of an electron/hole pair occurs mostly at the interfaces where the localized electrons in the InAs and holes in the GaSb layer overlap the most. In the case of InSb interface, the layer enlarges the Type II broken-gap alignment, thus allowing for a stronger tunneling of carriers and a heavier overlap between the
Proc. of SPIE Vol. 7608 76081Q-2
wavefunctions. In contrast, the GaAs interface creates a potential barrier for both electrons and holes, expulsing the carriers away form the interface and thus, reducing the overlap. The optical process in superlattice with GaAs interfaces is therefore less probable than those with InSb interfaces. For this, we only consider and realize superlattices with InSb-like interfaces.
Figure 2- Band alignment of superlattice with InSb and GaAs interfaces For theoretical modeling, we utilized the Empirical Tight Binding Method8, which can describe the effect of stress in each individual layers. Figure 3 shows the cut-off wavelength of different superlattice configurations. Each dot represents one superlattice configuration with fixed InAs and GaSb layer thicknesses. The superlattice designs are selected to have a lattice mismatch within 4000ppm from the GaSb substrate, based on empirical experience assuring for high quality material growth. It can be seen clearly that shorter wavelength superlattices tends to have larger lattice mismatch than longer wavelength designs. It is due the requirement of narrower InAs electrons’ well for higher conduction band and for larger energy gap. In InAs/GaSb superlattices, the InAs layer is -0.6% lattice mismatch to the GaSb substrate. Thinner tensile InAs layer will make the structure more compressive with larger lattice mismatch.
Cut-off wavelength (μm)
14 12 10 8 6 4 2 0 -2000
-1000
0
1000
2000
3000
4000
Lattice mismatch to GaSb ( ppm)
Figure 3- Cut-off wavelength distribution of superlattices as a function of lattice mismatch to the GaSb substrate. The dots represent the possible superlattice configurations by varying the InAs and GaSb layer thicknesses while keeping the two interfaces purely InSb. One of the main challenges for MWIR detection is thus the selection of superlattice design with correct cut-off wavelength and minimal compressive strain to assure high material quality. To solve the problem, a new interface engineering method was invented. By using a small proportion of Gallium in the InSb interfaces, the average lattice constant of the interfaces is reduced, bringing the average lattice constant
Proc. of SPIE Vol. 7608 76081Q-3
of superlattice closer to that of the GaSb substrate. The compressive strain of the structure will be suppressed, allowing for higher material quality. A clear advantage of using GaxIn1-xSb type interfaces is that it is highly controllable and highly repeatable because of the non-volatility of the Ga and In species on the sample surface at a growth temperature close to 400 °C. The calculation of mixed interface superlattices is performed in the Empirical Tight Binding frame work8, with variation of the Gallium molar fraction x1 and x2 at the two GaxIn1-xSb interfaces. As an example for superlattice consisting of [(InAs)7-Gax1In1-x1Sb-(GaSb)11-Gax2In1x2Sb]N, Figure 4a shows the calculated map of cutoff wavelength in the full range for x1 and x2 ( from 0 to 1). The cutoff wavelength can range from 3.68 μm to 4.20 μm, which corresponds to an energy variation of ~40 meV. The calculated map of mismatch (absolute values) is shown in Figure 4b. It is clear that there exists a zero mismatch composition line. With almost zero lattice mismatch, superlattice of very high crystalline quality can be grown with significant thickness in the microns range.
Figure 4- (a) The calculated map of cutoff wavelength for superlattice of [(InAs)7-Gax1In1Sb-(GaSb) The cutoff wavelength varies between 3.68-4.20 μm, corresponding to an 11-Gax2In1-x2Sb]N. x1 energy difference of ~40 meV. (b)The calculated map of the absolute values for the lattice mismatch in between the superlattice of [(InAs)7-Gax1In1-x1Sb-(GaSb)11-Gax2In1-x2Sb]N and the GaSb (001) substrate. Within a narrow stripe like region, zero lattice mismatch can be obtained.
MATERIAL GROWTH AND CHARACTERIZATIONS Optimal superlattice design was grown using an Intevac Modular Gen II molecular beam epitaxy system equipped with As and Sb valved cracker sources, on p-type epi-ready GaSb (001) substrates. Growths were performed mostly on quarter and whole 2” wafers for both large size detectors and focal plane array processing. The lattice mismatch to the GaSb substrate was controlled below 0.1%. The surface morphology of the samples was studied with a Digital Instruments Nanoscope IIIa atomic force microscope (AFM). The root mean square (RMS) roughness for typical superlattice growths has been demonstrated below 1.1 Å over an area of 10 μm×10 μm (Figure 5a). Small FWHM peaks and high ordered diffraction confirm the expected layer thickness of each individual layer and the interface sharpness. The uniformity of the material growth across a 2” wafer is shown in (Figure 5b). Photoluminescence was performed on 4 cleaves portions from a 2” wafer and showed identical spectrums at all locations (Figure 6). The integrated intensity slightly decreases at the hazy area at edge of portion C (near the wafer holder’s clamp) but this haziness is not a typical problem in superlattice growth.
Proc. of SPIE Vol. 7608 76081Q-4
Lattice mismatch: ~1200ppm
100000
Superlattice Peak FWHM~25 arcsec
Intensity ( a.u.)
10000
GaSb buffer Peak FWHM~16 arcsec (1)
(-1)
(2)
(-2) 1000 (-4)
(-3)
100
60 A
60 A (3)
60 A
60 A
(4)
10
28
30
32
Omega( degree)
Figure 5- Standard structural characterization of as grown samples: Left) 10 × 10 μm2 AFM scan shows clear, long atomic steps, Right) High Resolution X-ray Diffraction shows higher order diffraction patterns with narrow FWHM.
Figure 6- Material uniformity across a 2” wafer: left) 4 strips A,B,C,D were cleaved from a 2” wafer for PL measurement. Portion C has a hazy area due to the clamp of the wafer holder but this is not a typical problem of MBE growth of Type II superlattice. Right: PL signal at different locations of the four cleaved portions. The dips at ~4.2μm are due to the absorption of CO2 in the lab environment.
Proc. of SPIE Vol. 7608 76081Q-5
DEVICE FABRICATION AND OPTIMIZATIONS Type II InAs/GaSb superlattice devices are typically grown with an N-on-P polarity as shown in Figure 7a. The devices consist of two 0.5μm thick p and n-contact and an intentionally undoped i-region. Photodiodes with linear sizes varying from 100-400μm are fabricated using standard photolithography techniques, and a combination of dry etching with Electron Cyclotron Resonance (ECR) and wet etching using a citric acid based solution. Top and bottom contacts are deposited using an electron beam metal evaporator. The background concentration of the intrinsic region, measured via Capacitance-Voltage technique9, could be as low as 5x1014cm-3. This low background doping level is a good indication for high material quality, which allows for a long carrier diffusion length and high optical response. Device with 4μm thick i-region exhibits a quantum efficiency in excess of 50%. Higher quantum efficiency can be obtained by simply growing thicker device as the diffusion length is expected to be much longer than the device’s length10. However, the low background concentration means higher minority carrier concentration, which leads to a strong diffusion dark current, especially at high temperature. Figure 6b shows R0A vs. 1/T plot which reveals the dominant dark current mechanisms of a MWIR detector at different temperature regimes. Near cryogenic temperature, the device impedance decreases slowly with the increase of temperature. The R0A vs. 1/T plot shows an Arrhenius characteristic with slope less than half of the bandgap. This is due to either the generation-recombination current or the surface leakage, which is dictated by a performance discrepancy between different sized diodes. In the temperature range from 120-200K, the R0A follows a linear line with slope close to the energy gap of the material, meaning that the dark current is dominated by the diffusion regime. In order to increase the operating temperature of MWIR detectors, it is important to improve the performance of the device at this temperature range; thus, efforts must be spent on reducing the diffusion current. T (K) 9
10
400 300 250
200
8 7
10
6
10
5
3
10
2
10
1
10
0
10
-2
10
-3
10
-4
D
10
-1
if f
10
E
GaSb-Sub
Ti/Pt/Au
r d o ge ka li ea l R G ce r fa su te mi
4
10
us = 0 io n . 2 lim 17 e V i te d
2
n i p
10
2
a
Ti/Pt/Au R0A (Ω cm )
400μm
100
Target for operation temperature
10
400μm
150
4
6
8
10
12
14
-1
1000/T (K ) Figure 7- a) Schematic diagram of a Type II superlattice photodiode. B) Quantum efficiency spectrum of a MWIR photodiode at different measured temperatures.
The diffusion current is a fundamental mechanism in a diode structure. The relation between applied voltage V and the diffusion current density Jdiff is given by:
⎛ ⎛ eV J diff = J 0 ⎜ exp ⎜ ⎝ kT ⎝
⎞ ⎞ ⎟ − 1⎟ ⎠ ⎠
Proc. of SPIE Vol. 7608 76081Q-6
Equation 1
Where VB is the applied voltage bias; q is the electron charge; k is the Boltzmann constant and T is the absolute temperature. The first exponential part is the contribution of the majority carrier diffusion and the constant J0 is the drift part. At reverse bias, the later dominates and creates a saturated overall diffusion current asymptotically approaching J0. This current can be calculated from
x ⎞ ⎛ D x D J 0 = eni2 ⎜ h tanh n + e tanh p ⎟ Lh N A Le Le ⎠ ⎝ N D Lh
Equation 2
with:
n = N c N v exp− 2 i
D=
Eg kT ;
N c ,v
⎛ kTme, h ⎞ = 2⎜ 2 ⎟ ⎝ 2π h ⎠
kT μ e ; L = Dτ
3/ 2
Equation 3
Equation 4
The current density J0 depends on the bandgap Eg and temperature T via the intrinsic carrier concentration ni; on the material quality via the diffusion length Le,h and diffusivity De,h or diffusion lifetime (τe,h) and on the device design via the doping concentrations NA,D and the device thicknesses xn,p. From Equation 2, we can see that at a fixed temperature and for a given material, there are two way to reduce the diffusion current: • •
Reducing the device thicknesses: xn or xp Increasing the doping level NA,D.
Both methods have been applied for LWIR Type II superlattice detectors, and showed their effectiveness11, 12. However, the first method is not practical because the improvement of the electrical performance will impair the optical performance. A thinner device means less absorption length; the infrared radiation will not be absorbed and converted into electrical signal. We took the second option and applied to our MWIR devices. Similar to the study in Ref 12, the active region of the device was intentionally doped with Berylium to decrease the concentration of minority electrons, which are responsible for the diffusion current in a p-N photodiode. As shown in Figure 8a, by doping the active region p-type, the differential resistance of the device could be improved by one order of magnitude. The R0A of the doped sample reached the measurement system-limit of 106Ohm.cm2 but it should be much higher according to the interpolation of the diffusion-type I-V at zero bias.
Proc. of SPIE Vol. 7608 76081Q-7
T (K)
T (K)
100
240 200
p-doped Active region Undoped Active region 6
7
8
9
10 -1
1000/T (K )
11
12
13
D* at 4.0 μm (cm Hz
1/2
/ W)
150
2 R0A (Ω.cm )
250200 7 10 6 10 5 10 4 10 3 10 2 10 1 10 0 10 -1 10 -2 10 4 5
10
13
10
12
160
120
80
TBLIP=166K D*BLIP: 300 K 2π FOV
10
11
10
10
10
TBLIP = 136 K *
D BLIP 300K 2π FOV p-doped Active region undoped Active region
9
4
5
6
7
8
9
10 -1
11
12
13
1000 / T (K ) a) b) Figure 8-Comparison of (a)-differential resistance and (b) specific detectivity between intrinsically-undoped devices and intentionally p-doped devices. The overall performance of the device is shown in Figure 8b,. The detectivity of undoped devices attains 1.5x1013 cm.Hz1/2/W at 77K but decreases rapidly with increased temperature due to the decrease of R0A and the increase of the dark current. The background limited performance (BLIP) with a 300Kbackground is achieved at temperature below 136K. By intentionally p-doped the active region, the detectivity at 77K reaches 3x1013 cm.Hz1/2/W and the BLIP temperature rises to 166K.
FOCAL PLANE ARRAYS FABRICATION AND TESTING The Focal plane array fabrication involves significantly more steps compared with single element detectors processing (Figure 9). The FPA mesas are first formed using UV lithography and citric acid based wet etching or BCl3 based dry etching techniques. Each FPA die has a format of 320×256 consisting of 25μm×25μm square mesas with a pitch of 30 μm. Ti/Pt/Au is used as both top and bottom Ohmic contacts. Passivation step is then made to form a protection layer. Depending on different passivation techniques, different methods are applied for contact window opening. 6μm thick indium bumps are deposited to hybridize the array to the ROIC. Due to the oxidization nature of indium in ambient atmosphere environment, the FPA samples were diced before the indium evaporation process. The ROIC and the array are hybridized at room temperature. The gap between the ROIC and the FPA can be underfilled using low viscosity epoxy before the GaSb substrate is mechanically thinned down to 50 μm. The substrate can then be completely removed using wet etching. The FPA hybrid is finally mounted on a ceramic leadless chip carrier (CLCC) with vacuum grease and then gold wire bonded for testing.
Proc. of SPIE Vol. 7608 76081Q-8
P+ - Region GaSb (0.2 µm) P+ - Region InAs/GaSb (MWIR) (0.5 µm)
Active (π - region) InAs/GaSb (6.5 µm)
M- Barrier (undoped) InAs/GaSb/AlSb/GaSb (0.5 µm)
N+ - Region InAs/GaSb/AlSb/GaSb (MWIR) (0.5 µm)
N+
- Contact Layer InAsSb Buffer (1.5 µm)
Type-II wafer grown by MBE
Array pixel by ECR dry and wet etching
Metallization for ohmic contact on the detector
FPA ROIC
Packaging on a leadless chip carrier
FPA ROIC
ROIC FPA
QDIP
Test for imaging
Passivation-In bump deposition on FPA and ROIC
QDIP ROIC
ROIC QDIP
Flip chip bonding
Substrate Thinning
1
Figure 9-FPA fabrication procedure.
T = 81K
T=140K
T = 100K
T = 81K
T = 120K T = RT
T=200K
T = 300K
Figure 10-Pictures realized with the MWIR FPA at 81K up to Room Temperature. The 320x256 array with a 30 µm pitch was tested from 77 to 300K, at a frame rate of 32.64 Hz and an integration time of 22.98 ms. The Noise Equivalent Temperature Difference (NEDT) measured at 81K presented a peak at 10 mK with F#=2.3, which is equivalent to the state-of-the-art HgCdTe and QWIP technologies. Figure 10 shows a representative IR imaging of a MWIR Type II superlattice FPA. The camera is capable to image a human being at up to 150K, and can capture a hot soldering iron at uncooled condition.
Proc. of SPIE Vol. 7608 76081Q-9
CONCLUSION In summary, we have presented the development progress of Type-II InAs/GaSb MWIR photodetectors fabricated at the Center for Quantum Devices. By optimizing the material growth and device structure, device performance has been improved by one order of magnitude, allowing for higher operating temperature. The devices obtained Background limited (BLIP) operation at 166K. This temperature can be achieved by a Thermo Electric Cooler instead of liquid nitrogen cooling, which can potentially reduce the cost and size of the system significantly. MWIR Focal Plane Arrays have been demonstrated with a capability to image a human being up to 150K and can image hot objects when operated at room temperature.
REFERENCES [1] Sai-Halasz, G.A., Tsu, R., and Esaki, L., "A new semiconductor superlattice", Applied Physics Letters, 30, (12), pp. 651 (1977) [2] Smith, D.L., and Mailhiot, C., "Proposal for strained type II superlattice infrared detectors", Journal of Applied Physics, 62, (6), pp. 2545 (1987) [3] Nguyen, B.M., Razeghi, M., Nathan, V., and Brown, G.J., "Type-II M structure photodiodes: an alternative material design for mid-wave to long wavelength infrared regimes," in Quantum Sensing and Nanophotonic Devices IV, 1 ed. vol. 6479 San Jose, CA, USA: SPIE, 2007, pp. 64790S. [4] Nguyen, B.-M., Hoffman, D., Delaunay, P.-Y., and Razeghi, M., "Dark current suppression in type II InAs/GaSb superlattice long wavelength infrared photodiodes with M-structure barrier", Applied Physics Letters, 91, (16), pp. 163511 (2007) [5] Hoffman, D., Nguyen, B.-M., Huang, E.K.-w., Delaunay, P.-Y., Razeghi, M., Tidrow, M.Z., and Pellegrino, J., "The effect of doping the M-barrier in very long-wave type-II InAs/GaSb heterodiodes", Applied Physics Letters, 93, (3), pp. 031107 (2008) [6] Nguyen, B.-M., Hoffman, D., Delaunay, P.-Y., Huang, E.K.-W., Razeghi, M., and Pellegrino, J., "Band edge tunability of M-structure for heterojunction design in Sb based type II superlattice photodiodes", Applied Physics Letters, 93, (16), pp. 163502 (2008) [7] Mohseni, H., Litvinov, V.I., and Razeghi, M., "Interface-induced suppression of the Auger recombination in type-II InAs/GaSb superlattices", Physical Review B (Condensed Matter and Materials Physics), 58, (23), pp. 15378 (1998) [8] Wei, Y., and Razeghi, M., "Modeling of type-II InAs/GaSb superlattices using an empirical tight-binding method and interface engineering", Physical Review B (Condensed Matter and Materials Physics), 69, (8), pp. 085316 (2004) [9] Hood, A., Hoffman, D., Wei, Y., Fuchs, F., and Razeghi, M., "Capacitance-voltage investigation of high-purity InAs/GaSb superlattice photodiodes", Applied Physics Letters, 88, (5), pp. 052112 (2006) [10] Nguyen, B.-M., Hoffman, D., Wei, Y., Delaunay, P.-Y., Hood, A., and Razeghi, M., "Very high quantum efficiency in type-II InAs/GaSb superlattice photodiode with cutoff of 12 mu m", Applied Physics Letters, 90, (23), pp. 231108 (2007) [11] Hood, A., Hoffman, D., Nguyen, B.-M., Delaunay, P.-Y., Michel, E., and Razeghi, M., "High differential resistance type-II InAs/GaSb superlattice photodiodes for the long-wavelength infrared", Applied Physics Letters, 89, (9), pp. 093506 (2006) [12] Hoffman, D., Nguyen, B.-M., Delaunay, P.-Y., Hood, A., Razeghi, M., and Pellegrino, J., "Beryllium compensation doping of InAs/GaSb infrared superlattice photodiodes", Applied Physics Letters, 91, (14), pp. 143507 (2007)
Proc. of SPIE Vol. 7608 76081Q-10
