GaSb superlattice

Passivation of type II InAs/GaSb superlattice photodetectors Andrew Hood, Yajun Wei, Aaron Gin, and Manijeh Razeghi 1

)

Northwestern University, Center for Quantum Devices, ECE Department, Evanston, IL 60208 Meimei Z. Tidrow Missile Defense Agency, 7100 Defense Pentagon, Washington, DC 20301 Vaidya Nathan Air Force Research Laboratory, AFRL/VSSS, Kirtland AFB, NM 87117

ABSTRACT Leakage currents limit the operation of high performance type II InAs/GaSb superlattice photodiode technology. Surface leakage current becomes a dominant limiting factor, especially at the scale of a focal plane array pixel (< 25 µm) and must be addressed. A reduction of the surface state density, unpinning the Fermi level at the surface, and appropriate termination of the semiconductor crystal are all aims of effective passivation. Recent work in the passivation of type II InAs\GaSb superlattice photodetectors with aqueous sulfur-based solutions has resulted in increased R0A products and reduced dark current densities by reducing the surface trap density. Additionally, photoluminescence of similarly passivated type II InAs/GaSb superlattice and InAs GaSb bulk material will be discussed. Keywords: InAs, GaSb, superlattices, infrared, type II, passivation, ammonium sulfide, sulfur, photoluminescence 1.

INTRODUCTION

A great number of novel devices based on the 6.1 Å family of III-V materials have been recently demonstrated or proposed. The direct bandgap nature of many Sb-based materials and their alloys provides for interesting optoelectronic devices, throughout a wide range of wavelengths, for near to very long wavelength infrared optical applications. GaSb, as well as other III-V materials, exhibits a very chemically active surface. Sb-based devices have long since been troubled with rapid oxidation of their surfaces in air and from the presence of high surface-leakage currents. Relatively thick oxides, on the order of several nanometers, form when GaSb is exposed to air and this layer can degrade overall device performance by introducing surface states and conductive leakage pathways. 1,2

3,4

Unlike silicon, which has a natural, stable passivation of SiO , GaSb and other III-Vs must be passivated by another means to reduce the surface state density and unpin the Fermi level at the surface. A number of chemical and thin film treatments have been suggested and successfully applied for the passivation of various III-V materials , including InAs and GaSb. Most of the demonstrated wet chemical methods proposed in the literature involved chalcogen-based coating of the material surface and device sidewalls. We have recently shown that an aqueous ammonium sulfide treatment of type II InAs/GaSb superlattice photodiodes decreased the trap density and decreased reverse-bias dark current density by two orders of magnitude. Additionally, current modeling showed the dominant current component change from trapassisted tunneling to generation-recombination in the unpassivated and passivated cases, respectively. In the following 2

5,6

7

1)

316

Electronic mail: [email protected] Quantum Sensing and Nanophotonic Devices II, edited by Manijeh Razeghi, Gail J. Brown, Proceedings of SPIE Vol. 5732 (SPIE, Bellingham, WA, 2005) 0277-786X/05/$15 · doi: 10.1117/12.597140

we will discuss additional work we have done in characterizing the effect of ammonium sulfide surface treatments of bulk InAs, GaSb as well as type II InAs/GaSb superlattice photodiodes. In typical photovoltaic detectors, surface effects are not readily observed in the current-voltage (IV) characteristics of large dimension devices; 100s of microns in diameter, for example. But as device sizes are shrunk in order to increase density and resolution such as in focal plane array pixels (< 30 µm × 30 µm), the ratio of the diode perimeter to the bulk’s cross-sectional area increases and surface effects become more pronounced. The relationship of the device, bulk, and surface resistance-area products, as a function of relative device dimension, is given by the parallel resistor equation of 1

RA 0

 1 = + ( )   R0 A bulk

1

r

surface

0,

  

p  ,  A 

(1)

is the dynamic surface resistance per unit length, p is the perimeter of where (R0A) is the bulk R A product, r the diode, and A is the cross-sectional area of the detector junction. Passivation aims to increase the second term on the right hand side of Equation 1 so that the overall device impedance is dictated only by bulk material parameters. bulk

0

0,surface

8

2.

SAMPLE PREPARATION AND PASSIVATION TREATMENT

2.1. Material growth Type II InAs/GaSb superlattice samples for passivation experiments were grown using an Intevac Mod Gen II solid source molecular beam epitaxy (MBE) system with cracker and SUMO cells for group III and group V sources, respectively. Atomic force microscopy and high resolution x-ray diffraction were used to characterize the grown superlattice material. Atomic steps were routinely observed and root-mean square surface roughness was typically 1.2-1.6 Å over 20 µm × 20 µm areas. The superlattice to substrate lattice mismatch was controlled to be within 0.1%.

2.2. Device processing In the case of passivated diode characterization the following structures were grown and processed as follows. The diode structures were grown on p-type GaSb(001) substrate with a ~1.5 µm thick p-type GaSb buffer layer. The InAs/GaSb superlattice structure consisted of a 0.5 µm p-type region (NA ~1×1018 cm-3), a 2.0 µm non-intentionally doped i-region, a 0.5 µm n-type region (ND ~1×1018 cm-3), and a 100 Å InAs n+ capping layer. Top contacts of Ti/Pt/Au were evaporated and diode mesas were isolated using standard UV photolithography techniques and a citric acid-based wet chemical etch. The mesa diodes discussed here had dimensions of 400 µm × 400 µm. After etching, samples were rinsed with isopropanol, then given an ammonium sulfide treatment, and subsequently transferred to a nitrogen purged environment for short term storage before taking measurements.

2.3. Passivation treatment The passivation treatment consisted of a prepared ammonium sulfide solution and was applied by immersing the sample in this solution. An effect noticed by ourselves and others is that the ammonium sulfide solution degrades over time and this change in composition can impact device results and passivation effectiveness.9 Careful storage and use of the ammonium sulfide must be followed to reduce the change in composition of the solution. When we first received the ammonium sulfide from our supplier, successful passivation was achieved with a simple dilution of the stock solution, but over time, reproducibility was sporadic. To circumvent this problem, we dissolved 2.5g of elemental sulfur per 100 ml of heated ammonium sulfide, prior to dilution, to increase the free sulfur content (S) and possibly the degree, x, of polymerization of the catenated sulfur (Sx). This provided solutions that resulted in more reliable and repeatable

Proc. of SPIE Vol. 5732

317

passivation procedures. This solution, diluted 1:5 with de-ionized (DI) water, was used as our standard method of passivation, unless otherwise noted.

3. PASSIVATION CHARACTERIZATION 3.1. Optical and scanning electron microscopy (SEM) After the passivation treatment samples were studied under optical microscope. With typical passivation times of 5 minutes and under, no change in the surface or mesa image was observed. From time to time there were clumps of dust-like particles on the surface that we attributed to free sulfur that was not completely dissolved in the solution. We did not observe any change in the efficacy of successful passivation with the presence of these clumps. Investigation by scanning electron microscope allowed for qualitative analysis of the mesa sidewalls. SEM images of passivated and unpassivated mesa sidewalls are shown in Figure 1. Figure 1(a) shows an unpassivated mesa sidewall of a type II InAs/GaSb superlattice photodiode with dimensions of approximately 25 µm × 25 µm. Residues of etch by-products are seen on the unpassivated mesa sidewalls. Figure 1(b) shows a passivated mesa, of the same size as that in (a), which has been passivated for 17 minutes in an ammonium sulfide solution heated to 60 °C. The calculated etch rate of the GaSb buffer layer is 125 nm per minute under the conditions specified, an order of magnitude faster than the etch rate reported by others at room temperature and with a more concentrated solution.10 The ammonium sulfide solution appears to have preferentially etched the GaSb buffer layer compared to the superlattice. We attribute the reduced etch rate of the superlattice from the 1:20 ratio of etch rates between InAs and GaSb.11 It goes without saying that caution should be exercised in the passivation of small dimension GaSb-based devices so as to avoid over-etching, as the etch rate of the GaSb is sensitive to both concentration and temperature. It is partly for this reason that we keep our passivation treatment to within 1 to 5 minutes. With respect to the surface quality and cleanliness of the mesa sidewalls, the ammonium sulfide treatment has removed the etch residue and there are no noticeable, regular decorations of the sidewall surface.

(b) (a) Figure 1 - Scanning electron microscope (SEM) micrographs of (a) unpassivated and (b) passivated type II InAs/GaSb superlattice photodiodes. Etching of the GaSb buffer layer and the superlattice is observed in the passivated sample. The passivated sample was immersed in a 60 °C ammonium solution for 17 minutes.

318


3.2. Photoluminescence of passivated InAs, GaSb, and InAs/GaSb superlattice To assess the effectiveness of our passivation technique on the suppression of non-radiative surface states, we performed photoluminescence (PL) experiments on passivated materials. The idea behind these experiments was to systematically show the improvement of PL signal from type II InAs/GaSb material and its constituent compound semiconductor materials when processed with the ammonium sulfide surface treatment. In the case of bulk InAs and GaSb, substrates are readily available. The type II InAs/GaSb superlattice material is grown at our facilities, but PL of the superlattice sidewall surface proved to be challenging. Our approach is summarized in the next section. The tested materials included InAs(100) substrate material, GaSb(001) substrate material, and MBE-grown type II InAs/GaSb superlattice material. PL measurements were conducted in a closed-cycle helium cryostat. The excitation source was an Ar+ ion laser emitting at 514 nm (green), chopped at a 50% duty cycle, and directed upon the sample through a sapphire window. The laser spot was approximately 200µm in diameter. The photoluminescence was collected with a 20° FOV sapphire lens and directed at the slit of a grating monochromator. The photoluminescence signal was recorded by a lock-in amplifier in series with a preamplifier and a liquid nitrogen cooled InSb detector for wavelengths greater than 1.75 µm or thermoelectrically cooled InGaAs detector for wavelengths between 800 nm and 1750 nm. A 2-14 µm bandpass filter was used, with the InSb detector and a KRS-5 window was used with the InGaAs detector, to suppress harmonic peaks from the laser beam. Fluent laser power was minimized to reduce the laser penetration depth and maximize the observation of surface effects.

3.2.1. Photoluminescence of passivated GaSb-p(001) substrate Two small samples were cleaved from a p-type GaSb(001) wafer. The GaSb wafer was supplied in a Fluoroware tray and sealed by two outer bags within an inert atmosphere. The wafer had been removed periodically to cleave samples from but otherwise it was stored in a nitrogen purged box. One of the cleaved samples was treated for 5 minutes in the aforementioned ammonium sulfide solution. After the passivation treatment, both samples were left in air for one day. Figure 2 shows the relative photoluminescence of the passivated and unpassivated GaSb samples at 18 K and fluent laser power of 50 mW. The unpassivated signal has been smoothed to reduce noise and to allow for a more clear comparison. Three major peaks at 756, 777, and 794 meV are observed and are attributed to the commonly known B, A, and BE4 peaks. The main peak, A, at 777 meV involves the conduction band to native acceptor transition whereas the BE4 peak is the decay of a bound exciton to the same acceptor level. The peak at B is actually a broad unresolved feature of the peaks known as, in decreasing energy, B, A-LO, U, and B-LO, in the literature.12


319

)u .a (e sn op se R ec ne cs en i m ul ot oh P

GaSb Substrate - (NH ) S

4 2 x

100

10

passiv ated

GaSb Substrate - unpassiv ated - 1 day in air

A

BE4

B, A-LO, U, B-LO

1

0.1

0.01

1E-3

18 K 50 mW Fluent Power

1E-4 730

740

750

760

770

780

790

Energy (meV)

800

810

Figure 2 - An increase in the photoluminescence signal by over a factor of 30 was observed in passivated GaSb samples, compared to the unpassivated sample. The PL response, normalized to the A peak of the unpassivated GaSb sample, as well as the factors of increase are given in Table 1. Overall, the PL response of the passivated sample was strongly enhanced compared to the unpassivated sample, in some cases by a factor of up to 45. It is concluded that the surface recombination velocity has been reduced by the ammonium sulfide surface treatment. As reported by Salesse, , the increase seen in the BE4 to A ratio in the passivated sample suggests an increase in the bound exciton lifetime which is strongly influenced by the surface stateinduced electric field. It is possible that by satisfying the dangling bonds or other surface states with the ammonium sulfide treatment, the surface electric field is reduced along with the surface recombination velocity. et al

Transition A (777 meV) BE4 (793 meV) B (756 meV) BE4/A Unpassivated GaSb 1.0 0.14 0.10 0.14 Passivated GaSb 30.7 6.11 2.29 0.20 Factor of increase 30.7 43.64 22.9 42.9 % Table 1 – Photoluminescence response, normalized to the A peak of unpassivated GaSb.

3.2.2. Photoluminescence of passivated InAs(100) substrate Two samples were cleaved from an InAs(100) wafer and treated in the same manner with which the GaSb wafer, in the previous section, was passivated and compared. The passivated and unpassivated relative PL spectra of the InAs samples are shown in Figure 3. While not as resolved as the GaSb spectrum, a clear enhancement of the PL signal response was observed in the case of the passivated sample indicating, once again, suppression of surface states. The passivated signal was enhanced by as much as ~60 times compared to the unpassivated photoluminescence signal.

320


). u. a( es no ps e R ec ne cs en i m ul ot oh P

InAs Substrate - unpassivated - 1 day in air InAs Substrate - (NH4)2Sx passivated

100

10

1

18 K 100 mW Fluent Power

0.1

360

370

380

390

400

410

420

Energy (meV)

430

440

450

Figure 3 –PL response from unpassivated and ammonium sulfide passivated InAs substrate. 3.2.3. Photoluminescence of passivated MBE-grown type II InAs/GaSb superlattice material The improvement of the photoluminescence signal in ammonium sulfide surface treated InAs and GaSb suggests similar enhancement of the InAs/GaSb superlattice photoluminescence response. As explained in section 2.1, the grown samples are finished with an InAs n+ capping layer and this makes direct photoluminescence measurements of a superlattice surface difficult. Even with the capping layer removed, the periodic InAs/GaSb structure is not exposed in the same manner in which a mesa sidewall would be. Since we are concerned with the status of surface states on the mesa sidewalls, we would like to observe the characteristics of these sidewalls by photoluminescence. This is not a trivial task since the exposed superlattice is typically only 3 µm high and nearly normal to the sample surface. In order to make the superlattice available for excitation by the laser, a type II InAs/GaSb sample was mounted onto a Teflon holder and slowly lowered, by a controlled DC motor, into a solution of our standard citric-acid based etchant. During this lowering and etching process, the etchant was agitated by a magnetic stirring bar to ensure that the reactants were circulated and an even etch rate was maintained. The sample was then removed from the etchant, rinsed with DI water for 2 minutes, and then passivated. The end result was a graded profile of the superlattice over a distance of approximately 1 mm. For this experiment the grown superlattice structure was only 400 nm thick and had a designed cutoff wavelength of around 3.7 µm (335 meV). For thicker structures we were able to obtain longer graded profiles with lengths of up to ~1 cm. Figure 4 shows a portion of a gradually etched superlattice. Individual InAs and GaSb layers of the superlattice can be clearly seen. The overall length of the etched superlattice region in Figure 4 was ~7 mm.


321

Figure 4 - An SEM micrograph of gradually etched type II InAs/GaSb superlattice. The GaSb buffer layer is the solid grey region on the right. The type II InAs/GaSb sample was mounted, as the InAs and GaSb samples had been before, in the liquid He cryostat and photoluminescence measurements were conducted. Reference spectra above and below the graded region were recorded, representing the presence of the full superlattice and no superlattice, respectively. Figure 5(a) is a schematic representation of the gradually etched sample.

InAs capping layer A superlattice B GaSb C buffer layer

~1 mm

). .ua (7 Unpassivated graded superlattice - pos. B es Passivated graded superlattice - pos. B ~0.5 mm up from passivated graded area - pos. A no6 ~0.5 mm down from passivated graded area - pos. C ps e5 R ec 4 ne cs 3 en 14 K i 2 50 mW Fluent Power m ul 1 ot oh0 P 310 315 320 325 330 335 Energy (meV)

340

(a) (b) Figure 5 – (a) A schematic representation of the 3.7 µm cutoff wavelength type II InAs/GaSb superlattice sample gradually etched over a distance of 1 mm. The superlattice thickness was 400 nm. (b) Photoluminescence of gradually etched type II InAs/GaSb superlattice. Locations A, B, and C represent the laser focus positions on the sample for spectral measurements of the superlattice photoluminescence peak. When focused upon the graded portion for both the passivated and unpassivated samples, the 322


signal response was maximized, within the graded region, at 3.7 µm before commencing the scan. This was done in order to minimize the possibility that the enhancement seen in the passivated spectrum was due to greater superlattice thickness and thus higher absorption. But given the high absorption coefficient of GaSb in the visible region, luminescence is mainly produced within 10s of nanometers from the surface, and thus luminescence across most of the graded region should be the same. Additionally, multiple spectra were taken on the graded portion of the superlattice, varying the spot of incidence along the lateral direction, and the same relationship was observed; an increase in the superlattice photoluminescence peak of ~50% was realized for the ammonium sulfide passivated sample shown in Figure 5(b). While not as sizeable an increase in the PL signal as was seen with the passivated GaSb and InAs substrates, the type II InAs/GaSb sample underwent additional processing and the elongation and exposure of the many InAs/GaSb interfaces most likely manifested a large number of surface states that could not be completely satisfied by the passivation treatment. Nonetheless, the improvement in the PL signal is strong motivation for ammonium sulfide as a passivant of type II InAs/GaSb superlattices.

3.3. Electrical characterization of ammonium sulfide-passivated type II InAs/GaSb superlattice photodiodes An MBE-grown Type II InAs/GaSb superlattice structure with a designed cutoff wavelength of ~8 µm was fabricated into 400 µm × 400 µm mesa diodes as described in section 2.2. Some diodes were passivated while the others were untreated. Samples were indium bonded to a leadless chip carrier. Top n-contacts were bonded to the chip carrier pads and the substrate was used as a common p-contact. The chip carrier was placed into a custom-designed coldfinger and socket inside an open-cycle liquid nitrogen cryostat with a coldshield covering the detectors. The cryostat was pumped to a vacuum of ~2.0 × 10 Torr and then cooled to 80 K. -5

Current-voltage (IV) measurements were taken with an HP 4155A parameter analyzer and dark current modeling was conducted as described elsewhere. Measured and calculated dark current densities are shown in Figure 6 (a) and (b) for the unpassivated and passivated diodes, respectively. 7

100

100

Unpassi vated - N

10

2 )

~ 3.0 x 10 cm 14

d trap

-3

1

m /c A ( yti sn e D tn err u C

Passi vated - N

10

2 )

0.01

-3

1

m /c A ( yti sn e D tn er ru C

0.1

~ 1.0 x 10 cm 11

d trap

0.1

0.01

1E-3

1E-3

1E-4

Calculated total curren t Calculated trap -assisted tunneling current

1E-5

Calculated d iffusion current Calculated G-R current

80K

1E-7 -2.0

-1.5

Calculated total current Calculated trap -assisted tun nelin g current

1E-5

Calculated tunneling current

1E-6

M easured

1E-4

Measured

-1.0

-0.5

Voltage Bias (V)

0.0

Calculated tunnelin g current Calculated diffusion current Calculated G-R current

1E-6

80K

1E-7 -2.0

-1.5

-1.0

-0.5

0.0

Vol tage Bi as (V)

Figure 6 - Measured and modeled dark current densities for unpassivated (a) and passivated (b) type II InAs/GaSb photodiodes. Surface leakage current is represented by a trap-assisted tunneling current. Current modeling included diffusion, generation-recombination (G-R), direct tunneling, and trap-assisted tunneling current components. Fitting parameters included carrier lifetimes, trap density, and trap level. The fitted trap densities were 3.0 × 10 cm and 1.0 × 10 cm for the unpassivated and passivated detectors, respectively, representing an over 2 order of magnitude decrease in the reverse-bias dark current density. 14

-3

11

-3


323

One of the most important characteristics of an effective passivation technique is for it to remain stable and effective over time. Figure 7 shows the IV characteristics of a passivated photodiode, measured at 77 K, immediately after the ammonium sulfide treatment and after 3 months exposure to the ambient environment. The nearly identical currentvoltage curves indicate that the passivation treatment did not degrade even upon exposure to atmosphere for up to three months. Such behavior has been repeatedly observed in ammonium sulfide passivated type InAs/GaSb photodiodes in our labs. ) 2 m c/ A ( yti sn e D tn er ru C kr a D

3

10

2

10

3 months exposure to ambient environment Immediately after etching and passivation

1

10

0

10

-1

10

-2

10

-3

10

-4

10

-5

10

-6

10

-7

77 K

10

-3.0

-2.5

-2.0

-1.5

-1.0

-0.5

0.0

Bias Voltage ( V )

Figure 7 - A passivated type II InAs/GaSb photodetector measured immediately after the passivation treatment and after 3 months of exposure to lab air. 4. CONCLUSION We have developed an effective, reliable, and stable ammonium sulfide-based passivation technique for type II InAs/GaSb photodiodes. Improved photoluminescence from type II InAs/GaSb superlattice as well as its constituent materials, InAs and GaSb, were observed when treated with this passivation technique. IV measurements demonstrated a reduction of the reverse-bias dark current density for passivated diodes and diodes maintained their IV behavior for a period of time up to 3 months in air after being passivated. Effective passivation and reduction of the surface state density will be necessary for the realization of high-resolution, high-performance, midwave and longwave infrared focal plane arrays based on the type II InAs/GaSb material system, and ammonium sulfide-based passivation techniques provide such properties.

ACKNOWLEDGEMENTS The authors would like to acknowledge the support of Lt. Col. Todd Steiner of the Air Force Office of Scientific Research (AFOSR), and Dr. Gail Brown and Dr. Paul Levan of the Air Force Research Laboratory (AFRL). This project is jointly funded by the MDA, AFOSR, and AFRL under contract #F49620-01-1-0087. P.S. Dutta, H.L. Bhat, V. Kumar, “The physics and technology of gallium antimonide: An emerging optoelectronic material”, J. Appl. Phys., 81(9), 5821-5870, 1997. 2 M. Razeghi, “Overview of antimonide based III-V semiconductor epitaxial layers and their applications at the center for quantum devices”, Eur. Phys. J. AP, 23, 149-205, 2003. 3 Y. Wei, A. Gin, M. Razeghi, G.J. Brown, “Advanced InAs/GaSb superlattice photovoltaic detectors for very long wavelength infrared applications”, Appl. Phys. Lett., 80(18), 3262-3264, 2002. 4 Y.Wei, A. Gin, M. Razeghi, G.J. Brown, “Type II InAs/GaSb superlattice photovoltaic detectors with cutoff wavelengths approaching 32 µm”, Appl. Phys. Lett., 81(19), 3675-3677, 2002. 1

324


D.Y. Petrovykh, M.J. Yang, L.J. Whitman, “Chemical and electronic properties of sulfur-passivated InAs surfaces”, Surf. Sci., 523, 231-240, 2003. 6 B.Wu, G. Xia, Z. Li, J. Zhou, “Sulphur passivation of the InGaAsSb/GaSb photodiodes”, Appl. Phys. Lett., 80(7), 1303-1305, 2002. 7 A. Gin, Y. Wei, A. Hood, A. Bajowala, V. Yazdanpanah, M. Razeghi, M. Tidrow, “Ammonium sulfide passivation of Type-II InAs/GaSb superlattice photodiodes”, Appl. Phys. Lett., 84(12), 2037-2039, 2004. 8 L. Burkle, F. Fuchs, R. Kiefer, W. Pletschen, R.E. Sah, J. Schmitz, “Electrical Characterization of InAs/(GaIn)Sb infrared superlattice photodiodes for the 8 to 12 µm range”, Mat. Res. Soc. Symp. Proc., 607, 77-82, 2000. 9 R. Iyer, D.L. Lile, “Role of polysulfides in the passivation of the InP surface”, Appl. Phys. Lett., 59(4), 437-439, 1991. 10 E. Papis, A. Piotrowska, E. Kaminska, K. Golaszewska, W. Jung, J. Katcki, A Kudla, M. Piskorski, T.T. Piotrowski, J. Adamczewska, “Sulfide treatment of GaSb surface: influence on the LPE growth of InGaAsSb/AlGaAsSb heterostructures”, Vacuum, 57, 171-178, 2000. 11 D.Y. Petrovykh, M.J. Yang, L.J. Whitman, “Chemical and electronic properties of sulfur-passivated InAs surfaces”, Surf. Sci., 523, 231-240, 2003. 12 W.J. Jiang, Y.M. Sun, M.C. Wu, “Electrical and photoluminescent properties of high-quality GaSb and AlGaSb layers grown from Sb-rich solutions by liquid-phase epitaxy”, J. Appl. Phys., 77(4), 1725-1728, 1994. 5


325