高增益GaN基PIN雪崩二极管的制备及p-GaN层载流子浓度的估算

Fabrication of High Gain GaN Based PIN Avalanche Diode and Estimation of p-GaN Layer Carrier Concentration 高增益GaN基PIN雪崩二极管的制备及p-GaN层载流子浓度的估算 1 2 first-author 曹子坤(1993-), 男, 湖北荆州人, 博士研究生, 2016年于北京科技大学获得学士学位, 主要从事氮化镓材料光电器件的研究。E-mail:zikuncao@semi.ac.cn 曹子坤 (1993-), 男, 湖北荆州人, 博士研究生, 2016年于北京科技大学获得学士学位, 主要从事氮化镓材料光电器件的研究。E-mail: zikuncao@semi.ac.cn zikuncao@semi.ac.cn 1 1 1 3 1 1 3 赵德刚(1972-), 男, 湖北钟祥人, 博士, 教授, 博士研究生导师, 2000年于中国科学院半导体研究所获得博士学位, 主要从事宽禁带半导体材料与光电子器件的研究。E-mail:dgzhao@red.semi.ac.cn 赵德刚 (1972-), 男, 湖北钟祥人, 博士, 教授, 博士研究生导师, 2000年于中国科学院半导体研究所获得博士学位, 主要从事宽禁带半导体材料与光电子器件的研究。E-mail: dgzhao@red.semi.ac.cn dgzhao@red.semi.ac.cn State Key Laboratory of Integrated Optoelectronics, Institute of Semiconductors, Chinese Academy of Sciences, Beijing 100083, China State Key Laboratory of Integrated Optoelectronics, Institute of Semiconductors, Chinese Academy of Sciences, Beijing 100083, China 中国科学院半导体研究所 集成光电子学国家重点实验室, 北京 100083 中国科学院半导体研究所 集成光电子学国家重点实验室, 北京 100083 College of Materials Science and Optoelectronic Technology, University of Chinese Academy of Sciences, Beijing 100049, China College of Materials Science and Optoelectronic Technology, University of Chinese Academy of Sciences, Beijing 100049, China 中国科学院大学 材料科学与光电技术学院, 北京 100049 中国科学院大学 材料科学与光电技术学院, 北京 100049 Center of Materials Science and Optoelectronics Engineering, University of Chinese Academy of Sciences, Beijing 100049, China Center of Materials Science and Optoelectronics Engineering, University of Chinese Academy of Sciences, Beijing 100049, China 中国科学院大学 材料科学与光电工程中心, 北京 100049 中国科学院大学 材料科学与光电工程中心, 北京 100049 The fabrication process and test results of the GaN based pin avalanche detector are described in detail. The avalanche in obtained device occurs at a reverse bias of 71 V, and the multiplication factor reaches 5×10 4 . It is found that the p-layer carrier concentration is an important parameter affecting the device performance. Combined with the analysis of electric field intensity distribution, a method to estimate the carrier concentration in p-layer is proposed, and further calculation indicates that the maximum electric field value at avalanche breakdown is 2.6 MV/cm, which is similar to the previously reported values for GaN avalanche detectors. At last, Hall test and Secondary ion mass spectroscopy(SIMS) results are consistent with those estimated by the model calculation. 介绍了GaN基pin雪崩探测器的制作过程和测试结果。制作的器件在71 V反向偏压下发生雪崩，倍增因子达到5×10 4 。我们发现，p层载流子浓度是影响器件性能的重要参数。结合电场强度分布的分析，本文提出了一种估算p层载流子浓度的方法，进一步计算得到刚好雪崩击穿时的最大电场值为2.6 MV/cm，与以往GaN雪崩器件所报道的研究结果相似。最后，霍尔测试和SIMS测量p层载流子浓度的结果与模型计算的估算值吻合。 GaN avalanche detector Poisson equation 氮化镓 雪崩探测器 泊松方程 Introduction 1 As the representative of the third-generation semiconductors, GaN materials are widely used in optical and electrical fields in recent years because of its wide band gap [ 1 1 ] . After the two-step substrate growth technology is mature and the growth conditions of p-GaN are improved [ 2 2 - 3 3 2-3 ] , GaN optoelectronic devices have made great progress. Compared with silicon devices, GaN based avalanche detectors have the advantages of high internal gain and fast response. The forbidden band width of the material is 3.39 eV, which can detect the ultraviolet light. It is widely used in military early warning, medical detection, electronic communication and other fields. In this paper, the fabrication and test results of a pin structure GaN based detector are introduced in detail. The p-layer carrier concentration has a great influence on the device performance. The estimation and measurement of the parameters are helpful to further optimize the device. Therefore, based on the test results of the avalanche detector, this paper proposes a method to estimate the p-layer carrier concentration. At the same time, Hall measurement and SIMS (Secondary ion mass spectroscopy) are used to verify the accuracy of the estimation results. Experiments 1 Materials 2 The GaN samples are epitaxially grown on sapphire substrate by Metalorganic Chemical Vapor Deposition. Because there is a large lattice mismatch between sapphire substrate and GaN material, a thin buffer layer should be grown on sapphire substrate first. Then u-GaN, n-GaN, i-GaN and p-GaN are grown in order. u-GaN is the nucleation layer grown for reducing the lattice mismatch of subsequent growth. The i-GaN layer is unintentionally doped [ 4 4 ] , but because of impurities, it shows weak n-type. The thickness of p, i and n layers is 0.2, 0.1, 1.8 μm respectively. The carrier concentration in i-layer is 3×10 16 cm -3 and that in n-layer is 5×10 18 cm -3 by the previous tests. The carrier concentration in p-layer can be obtained by the subsequent calculation and tests. Device Fabrication 2 In order to remove the surface contamination, the samples were cleaned with acetone and isopropanol, respectively, in a 75 ℃ water bath. Then, the inorganic impurities and the oxide layer on the surface are removed with a mixed solution of HCl and hydrogen peroxide. Both p-type and n-type contacts were then patterned and deposited using a standard lift-off process. A mesa structure was formed by Inductively coupled plasma(ICP)etching [ 5 5 ] .The metal film is deposited by magnetron sputtering. The passivation layer is grown to reduce the leakage current through the mesa sidewalls, preventing the devices from undergoing premature breakdown under reverse bias [ 6 6 ] . The photosensitive area of the device is about 300 μm×300 μm. The transparent electrode is Ni/Au [ 7 7 ] , with thickness of 15/50 nm respectively. The thickening stage adopts Ti/Al/Ti/Au [ 8 8 - 9 9 8-9 ] with thickness of 15/250/50/250 nm. After the process treatment, the overall structure of the device is as shown in Fig. 1 Fig. 1 . Results and Discussion 1 In order to evaluate the defects of the material, the rocking curves of X-ray diffraction(XRD) are measured by Omega scanning. As shown in Fig. 2 Fig. 2 , the full-widths at half-maximum(FWHMs)at the GaN(002) reflection and(102)reflection are 212.04 arcsec and 219.6 arcsec, respectively. The calculated results show that the screw dislocation density is 9.03×10 7 cm -2 and the edge dislocation density of GaN samples is 2.56×10 8 cm -2 . The above data indicate that the defect densities in the material are relatively small [ 10 10 ] . The GaN material is also tested by atomic force microscopy (AFM). The result is shown in Fig. 3 Fig. 3 . The scanning area is 5 μm×5 μm. The total arithmetic average roughness(RA)and root mean square roughness(RQ) of the samples were 0.2 nm and 0.3 nm respectively. The surface morphology is flat as a whole, and the obvious step flow is also observed exhibiting excellent surface properties. The I - V test device is Keithley 2400, and the current error is about 100 pA. For protecting the instrument, the maximum scanning voltage range is set to 75 V, and the voltage step is set to 0.1 V. First, I - V curve I dark is measured in the dark. Afterwards, using xenon lamp with 75 W power as light source, the detector area is illuminated by 360 nm band light through a scanning grating monochromator, and the I - V curve I photo is measured. By using the equation of avalanche multiplication, the gain factor of avalanche multiplication M ph can be determined as [ 11 11 - 12 12 11-12 ] http://html.publish.founderss.cn/rc-pub/api/common/picture?pictureId=3015863&type= http://html.publish.founderss.cn/rc-pub/api/common/picture?pictureId=3015863&type=small http://html.publish.founderss.cn/rc-pub/api/common/picture?pictureId=3015863&type=middle where I ph0 and I dark are the photocurrent and dark current respectively. The molecular terms of Eq.1 correspond to the current when avalanche multiplication occurs, while the denominator terms of Eq.1 correspond to the current when avalanche multiplication does not occur. Since the photocurrent of the photodiode does not significantly increase after 10 V, the unmultiplied currents are sampled at 10 V for calculating the avalance gain. The typical results are shown in Fig. 4 Fig. 4 . It can be seen that under the reverse bias voltage of no more than 11 V, the current increases very slowly with the voltage, which is less than 1 nA. When the voltage reaches 71 V, avalanche breakdown occurs, and the current increases exponentially. Limited by the maximum current of the instrument, the maximum gain measured is 5×10 4 , which will be higher in practice. It is assumed that the layers of GaN are uniform, and the interface charge of each layer is ignored. We use the catastrophe junction approximation and assume that there is only one kind of conductive impurity in each GaN layer. The charge distribution of pin device is roughly as shown in Fig. 5 Fig. 5 . The electric field in semiconductor is affected by the charge distribution, which can be expressed as Eq.2 by Poisson equation: http://html.publish.founderss.cn/rc-pub/api/common/picture?pictureId=3015868&type= http://html.publish.founderss.cn/rc-pub/api/common/picture?pictureId=3015868&type=small http://html.publish.founderss.cn/rc-pub/api/common/picture?pictureId=3015868&type=middle where φ is the potential, E is the electric field intensity, x is the distance, ρ is the charge density which can be approximately considered as the elementary charge e multiplying donor or acceptor concentration, K s is the dielectric constant of GaN, and ε 0 is the permittivity of free space. According to Poisson equation and carrier continuity equation [ 13 13 - 16 16 13-16 ] , the distribution of electric field intensity of the device can be drawn as Fig. 6 Fig. 6 . It can be seen that with the increase of the reverse bias voltage, the electric field gradually diffuses to the p and n regions. Because of the high carrier concentration in n-layer and relatively low diffusion in p-layer, p-layer will be depleted earlier than n-layer. The contact type between metal and semiconductor strongly depends on the difference of their work functions. In order to form Ohmic contact between p-type material and metal, it is generally required that the work function of metal should be larger than that of p-type semiconductor material. For p-GaN materials, the work function is about 7.5 eV [ 17 17 ] . But the metal largest work function Pt is only 5.65 eV. In fact, the p-layer GaN will contact with the metal to form a Schottky contact with a very small potential, and then the Ohmic contact effect will be produced by using the tunneling effect under the high reference impurity. Fig. 7(a) Fig. 7(a) shows the conduction band diagram of pin device and metal under zero bias or low bias. At this time, the electrons in the p-layer pass through the i region to the n region, which belongs to minority carrier conduction. As shown in Fig. 7(b) Fig. 7(b) , when the p-layer is just exhausted, the metal electrons can cross the barrier of p and i layers to the n-layer. It can be considered that the metal and the pin device are connected to form a Schottky node, and the potential height will be higher than the general Schottky node. Because the Schottky reverse current is generated by the majority carrier, while the pin junction current is generated by the minority carrier, the reverse current of Schottky junction is much larger than that of pin junction. The current generated by the majority carrier will be much greater, as shown in Fig. 7(c) Fig. 7(c) . With the reverse voltage becoming larger, the Schottky potential is eliminated and the current becomes larger and larger. So, we can infer that when the I - V curve direction bias reaches 11 V, the current suddenly increases(shown in Fig. 4 Fig. 4 ), because the p-layer is just completely depleted as Fig. 7(b) Fig. 7(b) . According to the principle of integration, the area enclosed by the E - X image is approximately equal to the sum of the voltages, i.e . the magnitude of the applied reverse bias voltage V R1 and the built-in potential difference V bi . The V vi is about 3.2 V and the V R1 is about 11 V when the p-layer is just exhausted. If the concentration of i and n-layer and the thickness of p and i layer are known, using the area of E - X image as the sum of reverse bias voltage and built-in potential difference, the slope of p-layer line K p in E - X image is calculated as shown in Eq.3: http://html.publish.founderss.cn/rc-pub/api/common/picture?pictureId=3015873&type= http://html.publish.founderss.cn/rc-pub/api/common/picture?pictureId=3015873&type=small http://html.publish.founderss.cn/rc-pub/api/common/picture?pictureId=3015873&type=middle where K p , K i , K n are the corresponding slopes of E - X images of p, i and n layers respectively, X i , X n are the thickness of i layer and n layer respectively. It can be calculated that K p is 3.6×10 14 V/m 2 . Using the Poisson equation of Eq.2, the carrier concentration in the p layer can be calculated to be 1.8× 10 17 cm -3 according to the value of K p . In order to further verify the accuracy of estimation methods and results. We analysis the avalanche when the reverse voltage just reaches V R2 71 V. When avalanche occurs, it needs high enough electric field strength inside the device to drive the carrier to collide with the lattice. However, the maximum electric field strength of the device with the same material is generally considered to be basically the same when avalanche happens. If the p-layer carrier concentration is supposed as the result of calculation above, namely 1.8×10 17 cm -3 , the area enclosed by the E - X image is equal to the reverse voltage V R2 71 V plus the built-in voltage 3.2 V. According to the above assumptions, using the area of E - X image as the sum of reverse bias voltage V R2 and built-in potential difference V bi , the maximum electric field intensity can be calculated as 2.6 MV/cm by Eq.4 which is basically consistent with the reported value of previous studies [ 18 18 - 26 26 18-26 ] : http://html.publish.founderss.cn/rc-pub/api/common/picture?pictureId=3015874&type= http://html.publish.founderss.cn/rc-pub/api/common/picture?pictureId=3015874&type=small http://html.publish.founderss.cn/rc-pub/api/common/picture?pictureId=3015874&type=middle Experimental methods are also used to analyze the p-layer carrier concentration [ 20 20 , 27 27 ] . p-GaN: Mg+ layers are grown with the same growth conditions for tests. SIMS can be applied, and the concentration of Mg atoms shown in Fig. 8 Fig. 8 is about 2×10 19 cm -3 . According to previous studies [ 2 2 ] , 1%-2% of all Mg atoms are ionized at room temperature [ 28 28 ] . At the same time, the sample was also used for Hall test at 300 K. The average carrier concentration in the p-layer is about 3.15×10 17 cm -3 . These experimental results are very close to the previous estimate of 1.8×10 17 cm -3 . Conclusion 1 In this work, the test results of the studied avalanche detector show that the dark current of the device is less than 1 nA when the bias voltage is less than 11 V. The gain of APD is 5×10 4 . The overall performance is quite good. The method of estimating carrier concentration in the p-layer is in line with the test results of Hall and SIMS measurements. The proposed method can be used to measure the carrier concentration in p-layer in the avalanche diode for optimizing the device design. References PANDA D K, LENKA T R. Modeling and simulation of enhancement mode p-GaN Gate AlGaN/GaN HEMT for RF circuit switch applications[J]. J. Semicond ., 2017, 38(6):064002-1-6. KOZODOY P, XING H L, DENBAARS S P, et al .. Heavy doping effects in Mg-doped GaN[J]. J. Appl. Phys ., 2000, 87(4):1832-1835. RAZEGHI M, ROGALSKI A. Semiconductor ultraviolet detectors[J]. J. Appl. Phys ., 1996, 79(10):7433-7473. ZHAO D M, ZHAO D G. Analysis of the growth of GaN epitaxy on silicon[J]. J. Semicond ., 2018, 39(3):033006. LI X L, MA P, JI X L, et al .. Implementation of slow and smooth etching of GaN by inductively coupled plasma[J]. J. Semicond ., 2018, 39(11):113002-1-6. CAO X A, LU H, LEBOEUF S F, et al .. Growth and characterization of GaN pin rectifiers on free-standing GaN[J]. Appl. Phys. Lett ., 2005, 87(5):053503-1-3. CHENG Z J, SAN H S, FENG Z H, et al .. High open-circuit voltage betavoltaic cell based on GaN pin homojunction[J]. Electron. Lett ., 2011, 47(12):720-722. WANG X D, HU W D, CHEN X S, et al .. Dependence of dark current and photoresponse characteristics on polarization charge density for GaN-based avalanche photodiodes[J]. J. Phys. D :Appl. Phys ., 2011, 44(40):405102-1-11. LI T, BECK A L, COLLINS C, et al .. Improved ultraviolet quantum efficiency using a semitransparent recessed window AlGaN/GaN heterojunction p-i-n photodiode[J]. Appl. Phys. Lett ., 1999, 75(16):2421-2423. ZHANG S, ZHAO D G, JIANG D S, et al .. Investigation of responsivity decreasing with rising bias voltage in a GaN Schottky barrier photodetector[J]. Semicond. Sci. Technol ., 2008, 23(10):105015-1-6. WANG X D, HU W D, PAN M, et al .. Study of gain and photoresponse characteristics for back-illuminated separate absorption and multiplication GaN avalanche photodiodes[J]. J. Appl. Phys ., 2014, 115(1):013103-1-8. ZHANG Y, YOO D, LIMB J B, et al .. GaN ultraviolet avalanche photodiodes fabricated on free-standing bulk GaN substrates[J]. Phys. Status Solidi C, 2008, 5(6):2290-2292. MCCLINTOCK R, YASAN A, MINDER K, et al .. Avalanche multiplication in AlGaN based solar-blind photodetectors[J]. Appl. Phys. Lett ., 2005, 87(24):241123-1-3. MORESCO M, BERTAZZI F, BELLOTTI E. Theory of high field carrier transport and impact ionization in wurtzite GaN. Part Ⅱ: application to avalanche photodetectors[J]. J. Appl. Phys ., 2009, 106(6): 063719-1-9. XIE F, LU H, CHEN D J, et al .. Metal-semiconductor-metal ultraviolet avalanche photodiodes fabricated on bulk GaN substrate[J]. IEEE Electron Device Lett ., 2011, 32(9):1260-1262. WANG F X. Theoretical study of back-illuminated separated absorption and multiplication AlGaN APDs with different structural parameters[J]. Opt. Quantum Electron ., 2017, 49(6):216. TREXLER J T, PEARTON S J, HOLLOWAY P H, et al .. Comparison of Ni/Au, Pd/Au, and Cr/Au metallizations for Ohmic contacts to p-GaN[J]. MRS Proc. , 1996, 449:1091. BUYANOVA I A, WAGNER M, CHEN W M, et al .. Photoluminescence of GaN:effect of electron irradiation[J]. Appl. Phys. Lett ., 1998, 73(20):2968-2970. KOLNIÍK J, OGUZMAN İ H, BRENNAN K F, et al .. Monte Carlo calculation of electron initiated impact ionization in bulk zinc-blende and wurtzite GaN[J]. J. Appl. Phys ., 1997, 81(2):726-733. LIMB J B, YOO D, RYOU J H, et al .. GaN ultraviolet avalanche photodiodes with optical gain greater than 1000 grown on GaN substrates by metal-organic chemical vapor deposition[J]. Appl. Phys. Lett ., 2006, 89(1):011112-1-3. VASHAEI Z, CICEK E, BAYRAM C, et al .. GaN avalanche photodiodes grown on m -plane freestanding GaN substrate[J]. Appl. Phys. Lett ., 2010, 96(20):201908-1-3. PAU J L, BAYRAM C, MCCLINTOCK R, et al .. Back-illuminated separate absorption and multiplication GaN avalanche photodiodes[J]. Appl. Phys. Lett ., 2008, 92(10):101120-1-3. BAYRAM C, PAU J L, MCCLINTOCK R, et al .. Performance enhancement of GaN ultraviolet avalanche photodiodes with p-type δ-doping[J]. Appl. Phys. Lett ., 2008, 92(24):241103-1-3. LIN J C, SU Y K, CHANG S J, et al .. High responsivity of GaN p-i-n photodiode by using low-temperature interlayer[J]. Appl. Phys. Lett ., 2007, 91(17):173502-1-3. MCCLINTOCK R, PAU J L, MINDER K, et al .. Hole-initiated multiplication in back-illuminated GaN avalanche photodiodes[J]. Appl. Phys. Lett ., 2007, 90(14):141112-1-3. HUANG Y, CHEN D J, LU H, et al .. Back-illuminated separate absorption and multiplication AlGaN solar-blind avalanche photodiodes[J]. Appl. Phys. Lett ., 2012, 101(25):253516-1-4. CHOI S, KIM H J, ZHANG Y, et al .. Geiger-mode operation of GaN avalanche photodiodes grown on GaN substrates[J]. IEEE Photonics Technol. Lett ., 2009, 21(20):1526-1528. SMORCHKOVA I P, HAUS E, HEYING B, et al .. Mg doping of gan layers grown by plasma-assisted molecular-beam epitaxy[J]. Appl. Phys. Lett ., 2000, 76(6):718-720. 10.3788/fgxb20204106.0707 TN36 fgxb-41-6-707 Supported by National Key R＆D Program of China(2018YFB0406903，2016YFB0401801，2016YFB0401803); National Natural Science Foundation of China(61674138，61674139，61604145，61574135，61574134，61904172); Science Challenge Project(TZ2016003) 国家重点研发计划（2018YFB0406903，2016YFB0401801，2016YFB0401803）；国家自然科学基金（61674138，61674139，61604145，61574135，61574134，61904172）；科学挑战计划（TZ2016003）资助项目 2020-03-18 2020-04-15 2020-09-15 版权所有©《发光学报》编辑部2020 Copyright ©2020 Chinese Journal of Luminescence.All rights reserved. 2020 Chinese Journal of Luminescence 6 41