Sulphur passivation of gallium antimonide surfaces

Sulphur

passivation

of gallium

antimonide

surfaces

P. S. Dutta, K. S. Sangunni, and H. L. Bhat Department of Physics, Indian Institute of Science, Bangalore-

012, India

Vikram Kumar Solid State Physics Laboratory, Lucknow Roaci, Delhi-110 0.54, India

(Received 15 April 1994; accepted for publication.20 July 1994) Improvement in optical and electrical properties were observed after sulphur passivation of gallium antimonide surface. Enhancement of photoluminescence intensity up to 60 times, reduction in surface state density by two orders of magnitude, and reverse leakage currents by a factor of 20-30 were obtained as a result of surface passivation. While the reduction of surface recombination is attained, the surface is not unpinned.

In recent years, GaSb has become an important III-V compound semiconductor for optoelectronic device applications.’ Previous studies reveal that devices fabricated on GaSb exhibit large leakage currents.’ The problems of high surface state densities, surface Fermi level pinning and a residual native oxide layer on the surface, has hindered the development of devices based on GaSb.’ Hence, suitable surface passivation is an important step for the further development of GaSb device technologies. Previous experience with GaAs, AlGaAs, and InP wherein various dry and wet chemical treatments have been used to improve the surface-related characteristics justifies this point.“-” Of the various reagents used for surface treatments including HaS gas, Na,S, WOH, and (NH&S, the ammonia sulphides with excess sulphur [(NH&& (x= l-3)] were found to be the most effective. However, very little work has been done to investigate the passivating effect of sulphur on GaSb.7 In this letter, we present our results on the effect of sulphur passivation on the optical and electrical properties of the GaSb surface. GaSb single crystal samples (Te doped II type, and undoped p type) were used in this study. The crystals were grown in our laboratory by the vertical Bridgman technique.’ Surface passivation of GaSb substrates was carried out using ammonium sulphide [(NH,&&.]. Photoluminescence spectroscopy was used to characterize the optical properties before and after passivation. The PL measurements were carried out using a MIDAC Fourier transform photoluminescence spectrometer with a resolution of 1 meV. The room-temperature PL measurements were performed with the samples placed in a sealed He-tilled chamber. For low temperature measurements, the samples were freely suspended in liquid He at 4.2 K. An A? laser operating at 5145 A was used as the excitation source with excitation levels of 3.5 and 0.166 mW/cm” at 300 and 4.2 K respectively. A liquid-nitrogen cooled Ge photodetector was used for detection. The influence of various etchants before (NH4j2Sx treatment were studied first. The wafers were etched using different etchants, soaked in (NH&$ solution kept at 50 “C, and then dried in Nz gas. The strength of the room-temperature PL signal is shown in Fig. 1 as a function of surface treatment. Since the PL observed was a typical band-to-band transition, we indicate relative intensities only. In order to Appi. Phys. Let-t. 65 (13), 26 September 1994

find an optimum condition which could yield the best PL intensity, the sulphur treatments were carried out for various times (as indicated in Fig. 1). The variations in PL intensity after (NH&S, treatments for different times are believed to result from the gradation in passivated surface film quality. Reduction in PL intensity due to overpassivation is attributed to scattering by the polycrystalline surface film reconstructed with sulphur. Furthermore, the effect of rinsing the residual sulphur film with de-ionized water was also evaluated. As can be seen from Fig. 1, improvement in PL efficiency was observed even after thoroughly rinsing off the bulk of the film in de-ionized water, presumably due to the presence of a passivating submonolayer surface phase. The effect of etchants on the PL yield is evident from Fig. 1. The highest PL intensity was observed for the samples etched in HCl:H*O followed by sulphur (S) treatment. Reduction of the PL intensity for this sample was very minimal even after a few days indicating the stable nature of the passivated surface, This is because HC1:H20 usually removes the surface oxide layer without affecting the surface. Other etchants sometimes etch the surfaces nonstoichiometrically thereby increasing the surface state density. The low temperature PL spectra for the untreated and the treated samples are shown in Fig. 2. The PL spectrum of untreated GaSb showed residual acceptor (777 meV) and acceptor bound exciton (792 meV), whereas for S-treated GaSb, free exciton (810 meV), and acceptor bound excitons (797 meV, 801 meV) with enhanced intensity (nearly 60 times) were observed. This indicates that the dangling bonds at the surface are terminated by sulphur, which leads to the reduction of recombination centers of excitons. The highly resolved spectral features seen at low temperature in the passivated sample is likely to be due to the reduction in nonradiative surface recombination centers. This is also reflected in our enhanced room-temperature PL intensities. Capacitance-voltage (C-V) measurements on Schottky diodes fabricated by thermal evaporation of various metals on the untreated and the treated samples indicate the barrier heights to be independent of metal work function. Hence, the surface Fermi level pinning seems to exist even after sulphur passivation. From current-voltage (I-V) measurements at room temperature (Fig. 3), the ideality factor for the forward 1-V characteristics (in the low bias regime) was found to be in the range of 1.4-1.6 for the untreated samples and l.l-

0003-6951/94/65(13)/l 695/3/$6.00

Q 1994 American Institute of Physics

1695

Downloaded 14 Dec 2004 to 128.113.30.116. Redistribution subject to AIP license or copyright, see http://apl.aip.org/apl/copyright.jsp

I reased

I Etched

I r I , I IIll I , 10 20 30 40 Etched 10 20 30 40 24 Etc , min I min L hours I later (NHq125, (NH412 Sx treatment, treatment, dried D. I. rinsed, dried

PIG. 1. Relative PL intensities at 300 K for three samples as a function of surface treatment. (A): sample etched using HC1:H20 (l:l), (0): sample etched using Brz methanol, (0): sample etched in HCl%NO, (3O:l) or CH3COOH:HN03:HF (40:18:2).

1.2 for the passivated ones. The ideality factor between 1 and 2 indicates that electron diffusion current and surface recombination current are comparable in magnitude. Hence, the decrease in the ideality factor after passivation is mainly due to the reduction in surface recombination current. The reverse current in Schottky diodes decreased by a factor of lo-20 (see Fig. 3). This may be due to the formation of a

@x20 0 xl

@

~

792

lw\ l-47 c

770

I

I

I

I

780

790

800

a10

Photon Energy ImeV) FIG. 2. PL spectra of 1 untreated and 2 S treated p-GaSb at 4.2 K.

1696

Appl. Phys. Lelt., Vol. 65, No. 13, 26 September 1994

Voltage (Volts)

FIG. 3. 1-V characteristics of untreated (A) and S .tieated (e) n-GaSb samples at 300 K. (forward characteristics 1 and reverse characteristics

2). very thin dielectric layer between the semiconductor and the metal. S treatment after diode fabrication was found to reduce the reverse current further by a factor of 2-5. This can be due to the fact that S treatment changes the fixed charge in the oxide and thus diminishes the surface channel conductivity. DLTS measurements were carried out on MIS structures to study the effect of passivation on the surface state densities. The oxide layers of the MIS structures were formed by anodic oxidation of GaSb using either a ,solution of KOH in methanol9 or an aqueous solution of H,P04 with a pH of 2.5.l’ Mere exposure to ambient air for a long time also gives rise to an insulating layer. However, the MIS structures fabricated using this technique were found to be leaky. To separate the contribution due to the bulk deep level in GaSb, we also performed DLTS measurements on the Schottky diodes of GaSb. The results are shown in Fig. 4. The Schottky samples primarily show the bulk-dominated features. Hence the DLTS spectra of the Schottky diodes exhibited only a minor difference between the surface treated and the un: treated samples. DLTS signals observed from the MIS struo ture originate at the semiconductor-insulator interface states. As can be seen, the MIS structures fabricated after S treatment exhibit a drastic reduction in surface trap densities (note that the peak at 150 K is due to bulk deep level in GaSb). Using the conventional conductance-frequency measurements,” the surface trap densities were found to be in the range of 1X lOlo-5X 10” cm-’ eV-l and 3X1O’2-8X1O12 cme2 eV-’ for the treated and the untreated samples, respectively. The iowest density of state was obtained for the sample etched using HCI:H,O (1:l) and treated by sulphur. The surface state density was found to increase for various sample treatments in this order: etched in HCI:H,O and S treated, etched in Br, methanol (0.03% Brd and S treated, etched in HCI:HNO, (3O:l) and S treated, and Dutta et al.

Downloaded 14 Dec 2004 to 128.113.30.116. Redistribution subject to AIP license or copyright, see http://apl.aip.org/apl/copyright.jsp

(al

100

158

150

200

250

300

Temperature (K) FIG. 4. DLTS spectra of (a) GaSb Schottky diode before S treatment 1 and after S treatment 2 . (b) DLTS spectra of GaSb MIS structure. 1 etched in Bra methanol (0.03% Br& 2 etched in Brz methanol (&03% Br2) and S treated, 3 etched in CH, COOHHNO, :HF (40:18:2), 4 etched in CHsCOOH:HNOa:HF (40:18:2) and S treated, 5 etched in HCl:HZO (1:l) and S treated.

etched in CHaCOOH:HNOa:HF (40:18:2) and S treated. From our optical and electrical studies, it can be inferred that the reduced surface recombination rate (as evidenced by the increase in PL intensity) is mainly due to the reduction in surface state density (as electrically measured). There can be an additional contribution due to the increase in band bending brought in by the surface treatments.” Downward band bending near the surface gives rise to a field which tends to repel electrons that may approach the surface from the bulk regions. Since the sites for recombination are presumed to be near the surface, recombination can be effectively decreased by inhibiting carrier transport to the surface. For technological applications, it is essential to produce a passivating surface which not only reduces the density of surface states but also remain chemically stable under atmo-

spheric conditions for long periods. In our case, the enhanced gain in PL intensities after passivation persisted for a few months. The leakage current in the Schottky diodes prepared using the unpassivated samples increased by approximately one order of magnitude after one year and the MIS prepared on the unpassivated samples were found to be leaky after a few months. However, the Schottky diodes and the MIS structures fabricated after passivation showed no sign of degradation even after one year. To check the chemical stability of the passivating tllm, heat treatment at 200-250 “C in Nz atmosphere for lo-30 min was performed. It was found that the passivating effect remained even after such a treatment. In fact, sulphur forms many stable binary compounds with Ga and Sb. The beneficial effect of the treatment should be quite general and similar improvements in performance could be obtained on other GaSb-based devices. Moreover, the passivating film thus formed is chemically robust and does not hinder the long-term device performance. In conclusion, we have investigated the effects of (NH&& surface treatment on optical and electrical properties of GaSb. Enhancement in PL intensity, reduction both in surface state density and diode leakage current were observed as a result of passivation. However, the surface pinning still persists. Under optimized conditions, one can form a fully passivating thin interfacial sulphide layer. On such an interface it is possible to fabricate devices with improved characteristics. One of the authors (PSD) would like to thank CSIR (India) for the award of senior research fellowship (SRF). This work is supported by the UGC-COSIST programme. ‘A. G. Mimes and A. Y Polyakov, Solid State Electron. 36, 803 (1993). ‘B. J. Skromme, C. J. Sandroff, E. Yablonovitch, and T. Gmitter, Appl. Phys. L&t. 51, 2022 (1987). ‘R. N. Nottenburg, C. J. Sandroff, D. A. Humphrey, T. H. Hollenbeck, and R. Bhat, Appl. Phys. Lett. 52, 218 (1988). 4L. Koenders, U. Blomachez, and W. Month, J. Vat. Sci. Technol. B 6, 1416 (1988). ‘I. L. Chuang, M. S. Carpenter, M. R. Melloch, M. S. Lundstrom. E. Yablonovitch, and T. Y. Gmitter, Appl. Phys. Lett. 57, 2113 (1990). “R. Iyer, R. R. Chang, A. Dubey, and D. L. Lile, J. Vat. Sci Technol. B 6, 1174 (1988). 7A. Y. Polyakov, M. Stam, A. G. Milnes, and T. E. Schlesinger, Mater Sci. Eng. B 12, 337 (1992). sP. S. Dutta, H. L. Bhat, K. S. Sangunm, and Vi&ram Kumar, Proceedings of the Conference on Emerging Optoelectronic Technologies, Bangalore, India (Tata McGraw-Hill, New Delhi, 1992), p. 287. “C. W. Fischer, N. Leslie, and A. Etchells, J. Vat. Sci. Technol. 13, 59 (1976). r”D E Aspnes, B. Schwartz, A. A Studna, L. Derick, and L. A. Koszi, J. A,,, Phys. 48, 3510 (1977). ” E. H. Nicollian and J. R. Brews, h4OS (Metal Oxide Semiconductor) Physics and Technology (Wiley, New York, 1982). ‘aR. S. Besser and C. R Helms, Appt. Phys. Lett. 52, 1707 (1988).

Appl. Phys. Lett., Vol. 65, No. 13, 26 September 1994 Dutta et a/. 1697 Downloaded 14 Dec 2004 to 128.113.30.116. Redistribution subject to AIP license or copyright, see http://apl.aip.org/apl/copyright.jsp