Fabrication and passivation of GaSb photodiodes

Transcription

1 Journal of Crystal Growth 310 (2008) Fabrication and passivation of GaSb photodiodes Sujatha Sridaran, Ashonita Chavan, Partha S. Dutta Rensselaer Polytechnic Institute, Troy, NY 12180, USA Available online 3 December 2007 Abstract A novel Zn diffusion technique in n-gasb substrate from a low temperature chemical bath deposited ZnS layer has been developed to obtain high breakdown voltages. Junctions formed by this technique have breakdown voltages of 18:5 V, low reverse leakage current (0.01 0:03 A=cm 2 at 3 V), excellent reverse current saturation and ideality factor of 1:3. The high breakdown voltages obtained are due to the co-doping of zinc and sulfur from the ZnS film. Sulfur forms shallow and deep levels that compensate the p-doping of zinc. The non-linear relation of the inverse of the zero-bias resistance area product (1=R 0 A) versus perimeter to area ratio ðp=aþ in these diodes indicates surface leakage is the dominant leakage mechanism. CdS has been used to passivate the mesa photodiodes. After passivation, the 1=R 0 A product reduces from 0.3 to 0:02 O 1 cm 2 for a 150 mm diameter device. The 1=R 0 A product is also independent of the diode dimension confirming effective passivation. ZnS surface passivation on the mesa walls is not effective and is found to increase the leakage current. r 2007 Elsevier B.V. All rights reserved. PACS: Ea; Fx; z; Vv; Dw Keywords: A1. Diffusion; A1. Doping; A1. Semiconducting III V materials; A3. Chemical bath deposition; B1. Gallium antimonide; B1. Zinc sulfide; B3. Photodiodes 1. Introduction Gallium antimonide (GaSb) based semiconductor materials are very attractive for high speed electronic and optoelectronic applications in the near to mid-infrared wavelength region [1]. There has been continuous research in the materials and device aspect of GaSb, though not as rigorously as other III V compounds such as GaAs, InP or GaN. The slow maturity in the antimonide based device technology can be attributed to the high leakage currents and low breakdown voltages of GaSb devices. To obtain high breakdown voltages low doping concentrations are essential near the metallurgical junction to support a high electric field. Undoped as-grown GaSb is always p-type in nature with a residual acceptor concentration of cm 3 [1]. Though Pino et al. [2] have reported net donor concentration of 1: cm 3 at 300 K in GaSb, bulk growth of tellurium (Te) compensated GaSb to obtain Corresponding author. address: duttap@rpi.edu (P.S. Dutta). high resistivity samples still remains a technical challenge. Lower doping levels ( cm 3 ) can be achieved using growth from non-stoichiometric melts by liquid phase epitaxy at low temperatures or by compensation using Te doped GaSb in the growth solution [3,4]. Formation of p n junction by Zn diffusion from vapor source always results in highly doped p-region [5,6]. Heinz [7] has demonstrated a low doped p-region ( cm 3 ) by Zn diffusion from spin-on solid source. In this work, we have demonstrated high breakdown voltages by co-doping Zn and S from a low temperature chemical bath deposited ZnS layer. The ZnS layer protects the GaSb surface during the high temperature diffusion step from the evaporating Sb, in addition to being a source of Zn for diffusion. CdS and ZnS layers have also been employed as a passivation layer on these devices, and its impact on the surface leakage has been evaluated. The simple co-doping technique developed in this work, circumvents the need to incorporate complex epi-layer or bulk growth in the fabrication process to obtain high breakdown voltage GaSb devices /$ - see front matter r 2007 Elsevier B.V. All rights reserved. doi: /j.jcrysgro

2 S. Sridaran et al. / Journal of Crystal Growth 310 (2008) Experimental details The experimental details of the chemical bath deposition of ZnS, sealed tube diffusion followed by the device fabrication is as follows [8]. The substrates used in this work were single crystal Te-doped n-gasb with carrier concentrations cm 3. Prior to all processing, the samples were degreased with hot xylene, acetone and methanol (XAM cleaning) followed by a 1 min etch in warm hydrochloric acid (HCl) to remove the native oxide. To form the p n homojunction photodiode Zn diffusion was carried out from a low temperature chemical bath deposited ZnS layer of thickness 1000 A. The cleaned samples were immersed vertically in the chemical bath containing M zinc chloride (ZnCl 2 ), 0.02 M ammonium chloride (NH 4 Cl), M thiourea (N 2 H 4 CS) in 50 ml of deionized water. The ph and the temperature of the bath were maintained at 10.3 and 85 1C, respectively. The bath was stirred continuously at 200 rpm using a magnetic stirrer. Deposition was carried out for 45 min to obtain thickness of 1000 A. The film thickness was measured using a Dektak 8 surface profiler after etching regions of the ZnS film. Simultaneous depositions were carried on glass slides and the optical energy gap measured using a Varian Cary 500 UV-Vis-NIR spectrometer was found to be 3:6 ev. The ZnS deposited samples were sealed in an evacuated quartz tube ð10 6 torrþ and diffusion was performed at 500 1C for 10 h. The secondary ion mass spectrometry (SIMS) analysis of the Zn and S profiles after the diffusion is shown in Fig. 1. It is important to note that 50 nm of the SIMS profile from the surface is arbitrary and not considered for evaluation. The sheet resistance of the p-region measured using the four point probe was found to be 198 O=sq. Hall measurements were performed on the p-region by forming ohmic contacts with indium (In). Extreme caution was exercised during contact formation to ensure that the p n junction does not get shorted. Hall measurements result in a carrier concentration of cm 3 and carrier mobility of cm 2 =V s. After the diffusion, the ZnS layer on the GaSb surface was removed by rinsing in warm HCl for 2 min. ZnS diffuses on both the front and the back sides of the wafer. Therefore, to enable ohmic contact formation to the n-gasb substrate, the p-region from the backside is removed by etching for 30 s in chromic oxide solution (Cr 2 O 3 :H 2 O:HF as 32 g:120 ml:10 ml) after protecting the front side with Apiezon-W wax. Ohmic contacts on n-gasb were provided by the electron beam (E-beam) evaporation of 200 A of tin and 1000 A of gold followed by a rapid thermal annealing (RTA) at 350 1C for 15 s. Ohmic contacts on p-gasb were provided by E-beam evaporation of 200 A of titanium and 1000 A of gold after patterning and lift-off. The front side was further patterned and mesa etched using sodium potassium tartrate, HCl and H 2 O 2 (12 g:33 ml:9 ml diluted in 500 ml of deionized water) [9] for 10 min to define diodes of diameter varying from 150 to 500 mm. The schematic of the fabricated device is shown in Fig Results and discussions I V characterization of diodes was carried out at room temperature using HP 4140B I V meter. Standard pin probes were used to make connection to the top metal. The bottom metal was in contact with the metallic chuck, which was grounded. The large metal chuck also acted as a heat sink to keep the temperature of the device constant during the measurements. Dark I V characteristics were obtained when the probe station was enclosed in a dark chamber to prevent any ambient incident radiation. A representative plot of the dark I V characteristics of the ZnS diffused mesa photodiodes with varying diode area is shown in Fig. 3. Diodes varying in diameter from 150 to 500 mm were tested. In the forward bias regime, all diodes exhibit a turn-on voltage of 0:3 V, typical of GaSb devices [3]. The measured reverse saturation current density varies from 4: to 7: A=cm 2 and the ideality factor varies from 1.30 to 1.35 across diodes with diameters between 150 and 500 mm. In the reverse bias region, all diodes exhibit high breakdown voltage and low leakage current as seen in Metal Contact (200 Å Ti and 1000 Å Au) ZnS diffused p- region Te doped n-gasb Metal Contact (200 Å Sn and 1000 Å Au) Fig. 1. Secondary ion mass spectrometry analysis of Zn and S profile diffused from ZnS thin film in n-gasb. Fig. 2. Schematic of the cross section of mesa p n homojunction photodiode.

3 1592 S. Sridaran et al. / Journal of Crystal Growth 310 (2008) Fig. 3. Dark I V characteristics of ZnS diffused mesa GaSb photodiodes of diameter (i) 150 mm, (ii) 200 mm, (iii) 300 mm and (iv) 500 mm. Current Density (A/cm 2 ) (i) (ii) (iii) (iv) 150 μm 200 μm 300 μm 500 μm Voltage (V) Zn and the S diffuse into the GaSb substrate. Typically, Zn doping of GaSb results in a heavily p-type region. Sulfur doping of GaSb exhibits both shallow and deep levels which are DX like in nature [10]. The presence of these levels compensate the heavy doping of the p-region, resulting in a p n þ junction with high breakdown voltages. This is the first report of high breakdown voltages obtained in GaSb by co-doping the substrate with Zn and S from a ZnS film. The properties of the ZnS diffused p n junction diodes are tabulated in Table 1. With an incident optical radiation, the diodes exhibit a photocurrent. With increasing light intensity, the photocurrent also increases as measured on a 300 mm diameter device (Fig. 5). The diodes show excellent reverse current saturation even with an incident radiation. At a reverse bias of 5 V, and arbitrary incident radiation and the 300 mm devices exhibit photocurrent as high as 300 ma. At higher reverse bias, the leakage current increases and the diodes breakdown at voltages lower than under dark conditions. The light I V characteristics of the diodes were obtained with no dynamic temperature control. The metal chuck acted as a good heat sink. However it is possible to have some leakage current contributions from heat generated due to illumination that could lead to the Table 1 Properties of ZnS diffused p n homojunction diodes Diode diameter ðmmþ Reverse saturation Ideality Breakdown current density (A/cm 2 ) factor voltage (V) 150 4: : : : Zero-bias resistance area product (O cm 2 ) Fig. 4. Dark J V characteristics of ZnS diffused mesa GaSb photodiodes of diameter (i) 150 mm, (ii) 200 mm, (iii) 300 mm and (iv) 500 mm. Fig. 4. At a reverse bias of 3 V, the reverse leakage current density varies between 0.01 and 0:05 A=cm 2. The reverse leakage current density is also found to increase with increasing reverse bias until the device reaches breakdown. Breakdown voltages measured at 2 ma reverse current ranges between 18:5 V for 150 mm diameter devices to 16:5 V for 500 mm diameter devices. All diodes exhibit excellent current saturation and sharp breakdown. The breakdown voltage increases as the device area decreases. The observed breakdown voltages are among the highest reported in literature. Similar breakdown voltages have been observed by Heinz [7] using a spin-on Zn source. The high breakdown voltages obtained on the ZnS diffused GaSb diodes are due to the co-doping of GaSb by Zn and S. The SIMS analysis (Fig. 1) shows that both the Fig. 5. I V characteristics of 300 mm diameter ZnS diffused p n junction device with increasing light intensity.

4 S. Sridaran et al. / Journal of Crystal Growth 310 (2008) lowering of the breakdown voltages (during illumination). Additionally with increasing reverse bias, the width of the depletion region increases. The junction depth from the surface is 0:15 mm. At higher reverse biases, the depletion region reaches the surface, where the surface recombination velocity is high. With the doping profile as obtained by the SIMS analysis, at a reverse bias voltage of 4 V, the entire p-region is depleted. Therefore, with increasing incident intensity a large number of photogenerated carriers are lost by surface recombination-generation, resulting in high reverse leakage current and low breakdown voltages. To evaluate the effect of Zn diffusion from the ZnS layer, Zn diffusion using the leaky box technique was also carried out at 500 1C for 5 h. All the remaining processing steps were performed simultaneously. The diodes have similar turn-on voltages (0:3 V) and extremely low leakage currents at low reverse bias (0 1 V). However, the leakage current increases at higher reverse biases and the device breaks down between 2 and 3 V reverse bias. Both the diodes show an increase in the reverse saturation current density with decreasing device dimension (increasing perimeter to area ratio), which is indicative of surface leakage in mesa etched diodes. However, at a reverse bias of 3 V, the diodes fabricated on Zn diffused from ZnS film have leakage current two orders of magnitude lower than the diodes fabricated on Zn diffused from the vapor phase. This is due to the fact that devices formed by vapor phase diffusion, have a very high surface Zn concentration and the junction formed is a p þ n þ junction. As a result the device has a low breakdown voltage and high leakage current at low reverse bias. But the junction formed using the ZnS layer is a p2n þ junction, and can support high electric field before reaching the breakdown voltage (17 V). Therefore, these diodes exhibit a lower leakage current at a reverse bias of 3 V. A plot comparing the breakdown voltages of the two sets of diodes is shown in Fig. 6. The breakdown voltage of the junction formed by zinc sulfide diffusion is 6 times higher than the diodes formed on junctions by vapor phase Zn diffusion. Finally, the 1=R 0 A as a function of the perimeter to area ratio ðp=aþ is plotted in Fig. 7. From the model proposed by Gopal [11], the non-linear behavior of 1=R 0 A versus P=A as seen in Fig. 7 is indicative of the surface limited performance of the diodes depending upon the diffusion length of the minority carriers. The R 0 A increases from 21 to 115 O=cm 2 as the diode diameter increases from 150 to 500 mm for the diodes formed on layers from the ZnS diffusion and from 38 to 226 O=cm 2 for diodes of the same area and formed by the leaky box diffusion technique. To evaluate the impact of ZnS and CdS passivation on the surface limited performance of the reverse leakage, the metal was protected with photoresist and a low temperature deposition of ZnS or CdS was performed. The effect of CdS as a passivation layer on GaSb devices is well known from prior work [12,13,8]. When CdS is used as the Breakdown voltage (V) Zn diffusion from ZnS thin film Zn diffusion from vapor phase Zn Perimeter to area ratio (cm -1 ) Fig. 6. Comparison on breakdown voltages of p n junction devices fabricated by Zn diffusion from ZnS thin film and vapor phase Zn diffusion. 1/R 0 A(Ω -1 cm -2 ) Zn diffusion from ZnS thin film Zn diffusion from vapor phase Zn Perimeter to area ratio (cm -1 ) Fig. 7. 1=R 0 A as a function of P=A for p n junction devices fabricated by Zn diffusion from ZnS thin film and vapor phase Zn diffusion. passivation layer, reduction in the reverse leakage current up to a order of magnitude was observed in several devices. Additionally, the 1=R 0 A product reduced from 0.3 to 0:02 O 1 cm 2 for a 150 mm. The 1=R 0 A product is also independent of the diode dimension, thereby indicating that the leakage due to the surface component has been significantly reduced and the leakage current is purely diffusion limited. The J V characteristics of the unpassivated, CdS passivated and ZnS passivated diodes are shown in Fig. 8. It is seen that the diodes passivated by ZnS exhibit higher leakage than even the unpassivated diodes. At a reverse bias of 3 V, the leakage current of the ZnS passivated diodes is 2 times higher than the unpassivated diodes and an order of magnitude higher than the CdS passivated diodes. The higher leakage of the ZnS

5 1594 S. Sridaran et al. / Journal of Crystal Growth 310 (2008) leakage and the 1=R 0 A product is independent of diode dimensions. Therefore, Zn diffusion from a ZnS film followed by CdS passivation are effective means to overcome the high reverse leakage currents and low breakdown voltages typical of vapor phase diffused p n junctions. Additionally, the simple co-doping technique developed in this work, circumvents the need to incorporate complex epi-layer or bulk growth in the fabrication process to obtain high breakdown voltage GaSb devices. Acknowledgements The authors would like to thank National Science Foundation (NSF-STI ) for funding this research and Dr. Ishwara B. Bhat for his technical assistance. Fig. 8. Dark J V characteristics of 500 mm diameter diodes before and after passivation with ZnS and CdS. passivation can be attributed to the high affinity of Zn for oxygen. The oxide formed on the surface deteriorates the performance of the diodes. 4. Conclusions In conclusion, we have developed a new zinc diffusion technique from a low temperature chemical bath deposited ZnS layer. The fabricated diodes exhibit breakdown voltages 6 times higher than the vapor phase Zn diffused diodes. S from the ZnS film forms deep and shallow levels that compensates the heavy doping of Zn thereby resulting in higher breakdown voltage devices. The diodes have low reverse leakage current (0.01 0:03 A=cm 2 at 3 V) and excellent reverse current saturation. CdS passivated diodes exhibit an order of magnitude reduction in the reverse References [1] P.S. Dutta, H.L. Bhat, V. Kumar, J. Appl. Phys. 81 (9) (1997) [2] R. Pino, Y. Ko, P.S. Dutta, J. Electron. Mater. 33 (9) (2004) [3] Y.-M. Sun, J.-M. Wang, M.-C. Wu, Jpn. J. Appl. Phys. 34 (12A) (1995) L1579. [4] F. Capasso, M.B. Panish, S. Sumski, IEEE J. Quant. Electron. QE-17 (2) (1981) 273. [5] G. Rajagopalan, N.S. Reddy, H. Ehsani, I.B. Bhat, P.S. Dutta, R.J. Gutmann, G. Nichols, O. Sulima, J. Electron. Mater. 32 (11) (2003) [6] O.V. Sulima, A.W. Bett, Solar Energy Mater. Solar Cells 66 (2001) 533. [7] C. Heinz, Electron. Lett. 22 (5) (1986) 276. [8] S. Sridaran, Ph.D. Thesis, Rensselaer Polytechnic Institute, [9] V. Bhagwat, J.P. Langer, I.B. Bhat, P.S. Dutta, T. Refaat, M.N. Abedin, J. Electrochem. Soc. 151 (5) (2004) A728. [10] P.S. Dutta, K.S.R.K. Rao, K.S. Sangunni, H.L. Bhat, Appl. Phys. Lett. 65 (11) (1994) [11] V. Gopal, Semiconductor Sci. Technol. 9 (1994) [12] A. Chavan, A. Chandola, S. Sridaran, P.S. Dutta, J. Appl. Phys. 100 (6) (2006) [13] S. Sridaran, A. Chavan, P.S. Dutta, Appl. Phys. Lett. 89 (14) (2006)