Materials Science and Engineering B xxx (2005) xxx xxx. The improvements of GaN p i n UV sensor on 1 off-axis sapphire substrate

Transcription

1 Materials Science and Engineering B xxx (2005) xxx xxx The improvements of GaN p i n UV sensor on 1 off-axis sapphire substrate Su-Sir Liu a,, Pei-Wen Li a, W.H. Lan b, Wen-jen Lin c a Department of Electrical Engineering, National Central University, Taoyuan, Taiwan, ROC b Department of Electrical Engineering, National University of Kaohsiung, Kaohsiung, Taiwan, ROC c Chung-Shan Institute of Science & Technology, Lung-Tan, Taoyuan 325, Taiwan, ROC Received 11 November 2004; accepted 8 March Abstract A thin gallium nitride (GaN) layer epitaxially grown on misorientation angles of a-plane 1 off-axis sapphire substrate by metal organic chemical vapor deposition (MOCVD) has exhibited excellent film qualities such as enhanced crystallinity, lower defect levels, and less etching pit density. Accordingly, the GaN p i n photodetector fabricated on 1 off-axis sapphire substrate the dark current density decreased from A/cm 2 to A/cm 2 at 3 V, the responsivity increased from 0.06 A/W to A/W at 360 nm with no applied bias; the ultraviolet/visible (UV/vis) rejection ratio increased from to (comparing wavelength nm). A superior device performance could be achieved, as device fabricated on 1 off-axis sapphire substrate Elsevier B.V. All rights reserved. Keywords: GaN; Off-axis; UV sensor; Responsivity; AFM 1. Introduction Gallium nitride (GaN) and its alloys with aluminum and indium have been considered as the most promising materials for semiconductor photonic devices operating in blue and ultraviolet (UV) regions of the spectrum [1 3]. The transparency of high-quality GaN film to photons with wavelengths longer than 360 nm has made it an ideal material for photodetectors capable of retaining near unity quantum efficiency in the UV region of light spectrum while rejecting near infrared and visible light. Therefore, GaN-based photodetectors have been proposed for potentially possible applications such as UV astronomy, flame sensors, missile warning, and space communications [4,5]. It has been revealed that the quality of epitaxially grown GaN film is largely determined by its growth conditions such as substrate nitridation [6,7], buffered layer [8,9], and ramping conditions [10,11]. Nowadays, photodiodes grown on a low-defect lateral epitaxial overgrown (LEO) GaN layer [12], Corresponding author. address: susir.liu@msa.hinet.net (S.-S. Liu) /$ see front matter 2005 Elsevier B.V. All rights reserved. 2 doi: /j.mseb low temperature (LT) AlN interlayer [13], and strain-relief su- 38 perlattice (SLs) interlayer [14] for the defect reduction have 39 been proposed. Among various techniques that have been 40 adopted for GaN-film quality improvement, misorientate of 41 the substrate by reducing threading or edge dislocations has 42 been recognized as a promising technique [15 18]. How- 43 ever, very few devices results have been reported to verify the 44 substrate misorientate effect on device performance. In this 45 work, we experimentally focused on the electrical and phys- 46 ical character discussion as GaN p i n UV sensor grown on 47 c-plane sapphire with misorientation angles of a-plane 1 off- 48 axis substrate for defect reducing; samples were fabricated 49 by metal organic chemical vapor deposition (MOCVD). The 50 corresponding device grown on 0 on-axis has also been fab- 51 ricated for comparisons Experiments 53 The GaN groups of samples utilized in this study were 54 grown on 0 (sample 1) and misorientation angles of a- 55 plane 1 off-axis (sample 2) sapphire substrate, respec- 56

2 2 S.-S. Liu et al. / Materials Science and Engineering B xxx (2005) xxx xxx force microscopic (AFM) and a surface profiler (Dektak3) 82 were carried out to the surface morphology analysis.the 83 interface state capacitance with frequency was analyzed 84 by using HP 4284 parameter analyzer company with light 85 shield model Hg-402L mercury probe. Hall measurement 86 ( K) was also used to characterize the mobility and 87 carrier concentration. The I V characteristics of the p i n 88 photodetectors were analyzed using an HP 4156 parameter 89 analyzer. In addition, the spectral response of the p i n UV 90 detector was studied using a monochrometer with uniform W Xenon illumination. A standard Si-based UV enhanced 92 photo-detector was also used for purpose of calibration Results and discussion Fig. 1. (a) GaN p i n UV sensor structure grown on 0 or 1 off-axis sapphire substrate. (b) LT-PL intensity analysis as GaN grown on 0 (sample 1) and 1 off-axis (sample 2) sapphire substrate. tively, by MOCVD system, trimethylgallium (TMGa) and ammonia (NH 3 ) as gallium and nitrogen sources, SiH 4 as n-type and Cp 2 Mg as p-type doping source, respectively. During deposition process, hydrogen gas was used as the carrier gas. Prior to the GaN growth, a low temperature amorphous-phase GaN buffer layer (520 C, layer thickness = 30 nm) was first growth and followed by H 2 annealing (1120 C). The device structure from bottom to top consists of a 1 m GaN buffer layer, 2 m Si lightlydoped GaN layer (n = cm 3 ),a1 m heavilydoped n-gan layer (n = cm 3 ),a1 m un-doped GaN layer, and terminated by a 200 nm Mg-doped p-gan layer (p = cm 3 ) shown as Fig. 1(a); mesa etching was performed by ICP-RIE using Cl 2 and Ar as the etching sources for device isolation and contact patterning. The final device pattern consists of two circular rings for p- and n-contact electrodes as: Ni/Au (20 nm/150 nm) and Ti/Al/Ti/Au (20 nm/100 nm/20 nm/150 nm), respectively. Finally, a N 2 furnace annealing process at 600 C was performed for 10 min. Double-crystal-high-resolution X-ray diffractometer (XRD) for FWHM analysis, photoluminescence (PL) for spectrum analysis, a scanning electron microscope (SEM) to characterize the etching pit density (EPD in H 3 PO 4 at 280 C for 5 min) of GaN film were employed. Atomic Fig. 1(b) shows the low temperature (17 K) PL intensity 95 of GaN grown to a thickness of 2 m on0 (sample 1) and 96 misorientation angles of a-plane 1 off-axis (sample 2) sap- 97 phire substrate, respectively. It was observed that at lower 98 energy (inset, Fig. 1(b), ev ranges), the PL inten- 99 sity of sample 2 is quite flat than sample 1 which shows that 100 sample 2 has less defects in GaN bulk. Moreover, at higher 101 energy ( ev) measurement, it was observed that near 102 GaN bandedge ( 3.48 ev), the PL intensity of GaN grown 103 on sample 2 is larger and presents a better film quality than 104 sample 1. Misoriented slightly of a-plane 1 off-axis, possibly 105 due to reduction of undesirable impurities and/or the reduc- 106 tion of nitrogen vacancies. Low-temperature PL reveals an 107 unidentified peak, possibly related to an impurity or the sub- 108 strate, which has great intensity on 0 on-axis sample than off-sample at lower energy level. Regardless of the opti- 110 cal quality of the on-axis samples, the off-axis samples have 111 dim yellow band emission, bright band edge emission; these 112 factors lead to the possible conclusion that slight misorienta- 113 tion decreases the incorporation of unwanted impurities, such 114 as silicon, carbon, or oxygen, and/or reduces the density of 115 nitrogen vacancies [17]. 116 To correlate with the electrical data, Fig. 2 shows the XRD 117 (0 0 4) rocking curves of GaN grown to a thickness of 2 mon (sample 1) and misorientation angles of a-plane 1 off-axis 119 (sample 2) sapphire substrate, respectively. It was observed 120 that sample 2 with a pure GaN free exciton peak intensity was 121 observed and presented a narrow FWHM (240 s 1 ) shown as 122 Fig. 2(b) which presented a good film quality than sample FWHM (266 s 1 )infig. 2(a), respectively. The contem- 124 porary studies also show that high-quality GaN films can be 125 grown on sapphire substrate if the misorientation angle is 126 chosen appropriately [15 18]. 127 Fig. 3 shows the average dark current density 128 (1 mm 1 mm active region) of p i n UV sensors grown 129 on 0 (sample 1) and misorientation angles of a-plane off-axis (sample 2), respectively. Dark current is usually 131 attributed to the hopping of charged carriers via localized 132 defect-related states (traps) in the depletion region. Experi- 133 mental results indicate that the dark current density of sample 134

3 S.-S. Liu et al. / Materials Science and Engineering B xxx (2005) xxx xxx 3 Fig. 2. XRD (0 0 4) analysis as GaN grown on (a) 0 sapphire (sample 1), FWHM = 266 s 1 and (b) 1 off-axis sapphire (sample 2), FWHM = 240 s 1. Fig. 3. Dark current analysis as GaN p i n UV sensor grown on 0 (sample 1) and 1 off-axis (sample 2) sapphire substrate, respectively. Fig. 4. (a) Hall measurement of carrier concentration vs. temperature (K 1 ) which shows the activation energy (E a ) as GaN grown on 0 (sample 1) and 1 off-axis (sample 2) sapphire substrate, respectively. (b) Interface state capacitance vs. frequency as GaN grown on 0 (sample 1) and 1 off-axis (sample 2) sapphire substrate, respectively. Fig. 5. Spectrum responsivity analysis as GaN p i n UV sensor grown on 0 (sample 1) and 1 off-axis (sample 2) sapphire substrate, respectively.

4 4 S.-S. Liu et al. / Materials Science and Engineering B xxx (2005) xxx xxx is A/cm 2 which is about 2 orders of magnitude lower than sample 1 ( A/cm 2 ) at reverse bias of 3 V. Although the dark currents do fluctuate, however, more than 90 C of the p i n diodes retain their reverse currents between 10 pa/cm 2 and 20 pa/cm 2 at 0 V to 3V, 139 which are compatible with the results reported in GaN-based 140 p i n diodes[19]. The turn-on voltages of samples 2 and were 1.6 V and 2 V; the corresponding ideality factors were Fig. 6. (a and b) The etching pit density (EPD) on top p-gan film as devices grown on 0 and 1 off-axis of sapphire substrate, respectively; also (c and d) the AFM surface roughness morphology of p-gan film as devices grown on 0 and 1 off-axis of sapphire substrate, respectively.

5 S.-S. Liu et al. / Materials Science and Engineering B xxx (2005) xxx xxx obtained by fitting the low-level injection region of dark current (Fig. 3) are 2.4 and 3.4, respectively. The lower turn-on voltage and lower ideality factor could be achieved on sample 2 which represents a lower defect density of the film. The corresponding electrical characteristics such as lower dark current in sample 2 can be attributed to the improvement of GaN film quality. The crystal defects or impurities in the material may easily create energy levels within the bandgap and form recombination centers, which degrade the performance of UV photo detector. The improved analysis of defect level was characterized by Hall measurement ( K) and interface state capacitance analysis, respectively. Fig. 4(a) shows the Hall measurement of carrier concentration with temperature (K 1 ) as GaN grown to a thickness of 2 m with the same silane (SiH 4 ) flow on 0 (sample 1) and misorientation angles of a-plane 1 off-axis (sample 2) sapphire substrate, respectively. The slope of curve was calculated and defined as activation energy (E a ). It is observed that sample 1 presented two slope (m 1 = 19.3 mev, m 2 = 30.5 mev) of activation energy (E a ); the higher activation energy E a = 30.5 mev could be recognized as deep-level behavior. On the other hand, sample 2 has only one slope and lowering of activation energy (m 3 = 16.5 mev) was observed. Meanwhile, the mobility with temperature was carried out on both samples shown in inset of Fig. 4(a). It is observed that sample 2 has higher mobility than sample 1. The lower activation energy and higher mobility on sample 2 also show that the improvement of GaN film quality on 1 off-axis sapphire substrate. Fig. 4(b) shows as the C V measurement on samples 1 and 2, respectively. It was observed that at lower frequency (100 Hz), the related interfacial states capacitance, C P produced by defect extended into the surface of the bulk were measured about F and F on samples 1 and 2, respectively. The local defects or interface states hopping in the detector with bias, presented a higher dark current. This effect was studied extensively in amorphous semiconductors and was found to be the dominant mechanism of the dark current flow. Fig. 5 shows the spectrum response of visible-blind UV photodiodes at zero bias; the average responsivity exhibits significantly larger during UV spectral response ranges during nm visible-blind window as devices grown on 0 (sample 1) and misorientation angles of a-plane 1 offaxis (sample 2) sapphire substrate, respectively. Two devices roll off abruptly at wavelengths greater than 360 nm since they are transparent in this region. The decrease of responsivity at shorter wavelengths on samples 1 and 2, can be explained by the absorption in the p-layer. Sample 1 has lower responsivity as wavelength below 360 nm than sample 2 indicating that higher surface recombination since the carriers are created close to the surface and must diffuse to the junction in order to be collected. The response tail for photon energies below the band gap (wavelength > 360 nm) can be explained by the mechanisms of central photoionization in various defects such as dislocations during grain boundaries, or interface. These defect centers could generate energy lev- 198 els in the semiconductor band gap and they can be ionized by 199 photons with energies below the band gap. Meanwhile, high- 200 energy photons could also ionize these centers and provoke 201 band-to-band transition in the bulk, and a reduction of UV/vis 202 rejection ratio is observed. Experiment shows that the maxi- 203 mum responsivity on sample 2 is A/W at 360 nm with 204 no applied bias, which corresponds to an internal quantum 205 efficiency of 43 C is due to the A- and B-exciton absorp- 206 tion [20]. The corresponding responsivity of sample 1 at the 207 same condition is much lower ( 0.06 A/W). Moreover, the 208 rejection ratio of sample 2 is about (comparing 209 wavelength 360 nm and 450 nm) which is 1 order of magni- 210 tude higher than sample 1 ( ) at the same wave- 211 length, respectively. Through PL spectrum, XRD FWHM, 212 Hall measurement, and interface state capacitance analysis 213 discussed in Figs. 1, 2 and 4, we confirm that the improved 214 quality of GaN layers on sample 2 could be obtained, which 215 induced a higher responsivity and rejection ratio on sample was observed. 217 Fig. 6(a and b) shows the SEM photograph of etching 218 pit density (EPD) on top p-gan film as devices grown on (sample 1) and misorientation angles of a-plane 1 off- 220 axis (sample 2) sapphire substrate, respectively. The EPD 221 of sample 2 is cm 2, which is about 2 order lower 222 than sample 1 ( cm 2 ), the possible reason is that 223 slight misorientation decreases the incorporation of unwanted 224 impurities (such as silicon, carbon, or oxygen) and/or reduces 225 the density of nitrogen vacancies [17]. The improved grain 226 alignment and lower EPD may be responsible for the superior 227 electrical properties in sample 2. Through defect reduction, 228 the efficiency of carrier collection can be increased and a 229 higher response or rejection ratio on sample 2 is thus formed. 230 Fig. 6 (c and d) shows the AFM roughness analysis of p- 231 GaN film on samples 1 and 2, respectively, the mean rough- 232 ness of sample 2 (R a = nm), which is higher than sam- 233 ple1(r a = nm). The reasons why such rougher surface 234 morphologies on 1 off-axis sample are not clear at present, 235 but it is considered that the appropriate atomic steps of the 236 substrate formed by the slight misorientation can relieve the 237 lattice mismatch between sapphire and GaN, which is one of 238 the reason why for the change rougher in surface morphology 239 [21] Conclusions 241 In conclusion, GaN p i n UV sensor grown on 0 and 242 misorientation angles of a-plane 1 off-axis sapphire sub- 243 strate have been fabricated and characterized, respectively. 244 The GaN p i n UV sensor grown on 1 off-axis sapphire 245 substrate has exhibited lower dark current, higher response 246 sensitivity, and higher rejection ratio compared to that grown 247 on 0 on-axis sapphire substrate, these results which are in 248 accordance with the state-of-the-art of dislocation reduction 249 technique as GaN p i n UV detector fabricated by using LT 250

6 6 S.-S. Liu et al. / Materials Science and Engineering B xxx (2005) xxx xxx inter layer, superlattice (SLs), or epitaxial lateral overgrowth (ELOG) technique. Through the electrical and physical characteristic analysis by LT-PL, XRD, Hall measurement, interface state capacitance, and EPD analysis in our experiments, the truly improvements of data on 1 off-axis were obtained. Meanwhile, the effects of GaN p i n UV detectors fabricated on different misorientation angles are under developed in our laboratory, and will be characterized and discussed in the future. References [1] M.N. Yoder, IEEE Trans. Electron. Devices 43 (1996) [2] S.N. Mohammad, A.A. Salvador, H. Morkoc, Proc. IEEE 83 (1995) [3] H. Morkoc, S. Strite, G.B. Gao, M.E. Lin, B. Sverdlov, M. Burns, Appl. Phys. Lett. 76 (1994) [4] M. Razeghi, A. Rogalski, Appl. Phys. Lett. 79 (1996) [5] E. Monroy, F. Calle, J.L. Pau, E. Munoz, F. Omnes, B. Beaumont, P. Gibart, J. Cryst. Growth 230 (2001) 537. [6] K. Uchida, A. Watanabc, F. Yano, M. Kouguchi, T. Tanaka, S. Minagawa, Appl. Phys. Lett. 79 (1996) [7] D. Kum, D. Byun, J. Electron. Mater. 23 (1997) [8] H. Amano, N. Sawaki, I. Akasaki, Y. Toyoda, Appl. Phys. Lett. 48 (1986) 353. [9] S. Nakamura, Jpn. Appl. Phys. Lett. 30 (1991) [10] C.F. Lin, G.C. Chi, M.S. Feng, J.S. Tsang, J.M. Hong, Appl. Phys. 274 Lett. 68 (1996) [11] L. Sugiura, K. Itaya, J. Nishio, H. Fujimoto, Y. Kokubo, Appl. Phys. 276 Lett. 82 (1997) [12] G. Parish, S. Keller, P. Kozodoy, J.P. Ibbetson, H. Marchand, P.T. 278 Fini, S.B. Fleischer, S.P. DenBaars, U.K. Mishra, E.J. Tarsa, Appl. 279 Phys. Lett. 75 (1999) [13] J.C. Carrano, T. Li, P.A. Grudowski, C.J. Eiting, D. Lambert, J.D. 281 Schaub, R.D. Dupuis, J.C. Campbell, Electron. Lett. 34 (1999) [14] C.Q. Chen, J.P. Zhang, M.E. Gaevski, H.M. Wang, W.H. Sun, R.S.Q. 284 Fareed, J.W. Yang, M. Asif Khan, Appl. Phys. Lett. 81 (2002) [15] M. Fatemin, A.E. Wickenden, D.D. Koleske, M.E. Twigg, J.A. Fre- 287 itas Jr., R.L. Henry, R.L. Henry, R.J. Goman, Appl. Phys. Lett (1998) [16] T. Someya, K. Hoshino, Y. Arakawa, Appl. Phys. Lett. 79 (2001) [17] P.A. Grudowski, A.L. Holmes, C.J. Eiting, R.D. Dupuis, Appl. Phys. 292 Lett. 69 (1996) [18] T. Yuasa, Y. Ueta, Y. Tsuda, A. Ogawa, M. Taneya, K. Takao, Jpn. 294 Appl. Phys. Lett. 38 (1999) L703 L [19] J.C. Carrano, T. Li, P.A. Grudowski, C.J. Eiting, D. Lambert, J.D. 296 Schaub, R.D. Dupuis, J.C. Campbell, Electron. Lett. 34 (1999) [20] W. Liu, M.F. Li, S.J. Chua, Y.H. Zhang, K. Uchida, Appl. Phys. 298 Lett. 71 (1997) [21] J.M. Lee, Mater. Res. Soc. Symp. Proc. 67 (1986)