Dual-Gate p-gan Gate High Electron Mobility Transistors for Steep Subthreshold Slope. Jong-Ho Bae and Jong-Ho Lee

Transcription

1 Copyright 2016 American cientific Publishers All rights reserved Printed in the United tates of America Article Journal of Nanoscience and Nanotechnology Vol. 16, 1 5, ual-ate p- ate High Electron Mobility Transistors for teep ubthreshold lope Jong-Ho Bae and Jong-Ho Lee epartment of Electrical and Computer Engineering and the Inter-University emiconductor Research Center (IRC), eoul National University, eoul , Korea A steep subthreshold slope characteristic is achieved through p- gate HEMT with dual-gate structure. Obtained subthreshold slope is less than 120 V/dec. Based on the measured and simulated data obtained from single-gate device, breakdown of parasitic floating-base bipolar transistor and floating gate charged with holes are responsible to increase abruptly in drain current. In the dual-gate device, on-current degrades with high temperature but subthreshold slope is not changed. To observe the switching speed of dual-gate device and transient response of drain current are measured. According to the transient responses of drain current, switching speed of the dual-gate device is about 10 5 sec. Keywords: Al/, ual-ate, ate Current, HEMTs, Impact Ionization, p-, ubthreshold lope, Threshold Voltage, Transient. 1. INTROUCTION The Al/ high electron mobility transistor (HEMT) has been widely investigated for high-frequency and high-power switching applications due to its wide band gap, high breakdown voltage, high mobility and 2-dimensional electron gas (2-E). 1 The 2-E formed by Al/ hetero-junction has high mobility and high carrier concentration. ince the 2-E is formed at Al/ interface all along the channel, field effect transistor (FET) with 2-E shows normally-on characteristics. However, for high-power applications, normallyoff characteristic with high threshold voltage (V T above 3 V is desirable. 2 Recently, p- gate HEMT on i substrate with a high V T ( 3 V) and a low gate current (I was reported To understand the behavior of the device, I of such a p- gate HEMT was analyzed including trapping behavior and hole charging in the p- layer in our previous work. 12 In this paper, we investigated the mechanism of sudden increase of the I and drain current (I observed in p- gate HEMT based on conduction mechanism of I described in our previous work. 12 We also verified the mechanism by using TCA simulation including impact ionization model. By adapting dual-gate structure Author to whom correspondence should be addressed. with floating gate formed between gate and source considering the mechanism observed in simulation and measurement data, V T 4 V and subthreshold slope less than 120 V/dec are achieved. ince the device is working with hole charging in the channel and floating gate, transient response is analyzed to observe the switching speed of the device. The switching speed of the device is 10 s. 2. EXPERIMENTAL ETAIL 2.1. evice tructure The p- gate HEMTs in this work were fabricated at amsung. The devices have Ti/Au gate formed on p- (100 nm)/al (15 nm)/ hetero-structure grown on 8-inch i (111) substrate as depicted in Figure 1. To reduce punch-through, the layer is doped with carbon 1 m below the channel. The p- gate layer is doped with an Mg ( cm 3 and hole concentration obtained by hall measurement is cm 3.The p- layer depletes 2-E under the gate and the device works as normally-off characteristic. The metal/p-/ Al/ structure is a kind of floating base n p n heterojunction bipolar transistor, because the metal/p- is a chottky junction and p-/al/ forms a heterojunction p n diode. The dual-gate device was fabricated with single-gate HEMT device on same wafer, which J. Nanosci. Nanotechnol. 2016, Vol. 16, No. xx /2016/16/001/005 doi: /jnn

2 ual-ate p- ate HEMTs for teep ubthreshold lope Al p- Nucleation layer/buffer i-wafer Al F p- Nucleation layer/buffer i-wafer Figure 1. chematic cross-sectional view of a single-gate and dual-gate p- gate HEMTs on silicon substrate. has a floating gate (metal/p- on Al layer) between the gate and source as depicted in Figure Measurement and imulation The C I V and transient response of the devices were obtained using a Agilent B1500A and WFMU module. To verify the physical model based on measured data, we performed a TCA simulation. Basic parameters including electron and hole concentrations, and the generation rate due to impact ionization are simulated with ILVACO, Atlas TCA tool. 13 The device structure used in device simulation has the same physical dimension as that of measured single-gate device. Physical models used in the simulation includes RH, Auger, field dependent mobility, incomplete ionization model and elberherr s impact ionization model with band-to-band generation model. 13 The doping concentration of acceptor (N a in p- is cm 3 and hole concentration in neutral region of the p- is cm 3, which means the incomplete ionization model is working well with a reported activation energy (E a of Mg in (0.17 ev) REULT AN ICUION 3.1. C I V of ingle-ate p- HEMT Figure 2 shows transfer curves of a single-gate p- gate HEMT measured and simulated at room temperature ev. 1 ev. 2 im. V = 1 V W = 100 µm Figure 2. I and I versus V at V = 1 V. The solid and dashed lines represent measured and simulated I V, respectively. In the red circled region, I and I abruptly increase. I (A) Bae and Lee Normally-off behavior is observed with a V T of 2 Vand I is smaller than I in order of 3. ome of the singlegate devices show steep increase of I and I in C I V sweep. In the red-circled region in Figure 2, a steep increase in I and I is observed at V between3and4v.as described in our previous work, 12 When the V is around 3.5 V, floating base BJT is turned on by holes which come from gate chottky junction. We can consider that floating base BJT is turned-on by positive feedback caused by impact ionization in the device which shows steep increase in I and I imulation Results of ingle-ate p- HEMT To verify the physical model for the sudden increase in both current, we performed a TCA simulation. Band diagrams as a parameter of V (1, 3 and 5 V) are showninfigure3.asv increases, conduction and valence bands move down and p- depletion region becomes wide. Also, both conduction and valence band become steep with high electric field in depletion region. In Figure 3, impact generation rate increases as V increases. The impact generation rate obviously increases at high V in reverse biased chottky junction. As more holes are generated in the reverse biased chottky depletion region, holes are accumulated at p-/al heterojunction interface. As a result, because of the accumulated holes at the p-/al interface, the p n barrier formed by p-/al/ heterojunction becomes small and Energy Band (ev) Impact en. Rate ate V = 1~6 V V = 1, 3, 5 V p- Al Figure 3. imulated parameters through gate vertical stack (metal/p- /Al/) as a parameter of V : energy band diagram (1, 3, 5 V) and impact generation rate (1 6 V). 2 J. Nanosci. Nanotechnol. 16, 1 5, 2016

3 Bae and Lee more electrons move from the channel to p- gate across the barrier as shown in Figure 4. After that, the injected electrons generate more e h pairs in the p- depletion region and generated holes lower the barrier further, which in turn incurs more electron injection from the channel. This process is accelerated by a positive feedback, resulting in a sort of breakdown of a floating base BJT. As a result, I increase steeply. The physical model including impact ionization and breakdown of a floating base BJT explains well the measured I described in Figure 2. Moreover, as more holes are generated in p- layer at V of 4 V, generated holes are injected into channel up to cm 3. As a result, energy band moves down with injected holes and more electrons are accumulated in the channel as shown in Figures 4 and. Therefore more electrons easily flow from the source to the drain, resulting in the increase of I. It has been reported that the hole injection into the channel makes I increase and second g m peak. 15 In our case, the amount of injected holes increases rapidly due to positive feedback related to the hole generation, I also increases steeply C I V of ual-ate p- HEMT Figure 5 shows the measured I V curves of three dual-gate p- gate HEMTs and simulated I V curves. The curves show similar V T and I. Obtained subthreshold slope from one of the device is 120 V/dec, as plotting in inset of the Figure 5. In the dual-gate device, I is controlled by V and potential of the floating gate (V F. ince the single- and dual-gate devices are fabricated on same wafer and V T of the single-gate device shown in Figure 2 is 2 V,which ual-ate p- ate HEMTs for teep ubthreshold lope 250 µv step < 120 µv/dec V = (V) Meas. im. V = 1 V 3 devices Figure 5. I and I versus V at V = 1 V. The open and closed symbols represent log I and log I, respectively. The I and I increase abruptly around 4 V of V. means V T of floating gate is also 2 V.WhenaV is less than 4 V, V F is smaller than V T since floating gate is output node of a kind of voltage divider. The voltage is divided by two resistances: one is consist of gate chottky junction, Al/AN heterojunction, and 2-E region from the gate to floating gate and the other is 2-E region from floating gate to the source. Because a resistance from the source to floating gate is smaller than total resistances of others, V F should be close to V. Therefore V F is less than V T and the channel under the floating gate still remains as depleted although V is larger than V T. As a result, the dual-gate device is off and only leakage current flows through depletion region. At a V larger than 4 V, impact ionization occurs in the p- depletion region as explained above. The generation rate of e h pairs by impact ionization increases, and then a lot of holes are I (A) e - conc. (cm 3 ) F e - conc. (cm 3 ) h + conc. (cm 3 ) F h + conc. (cm 3 ) Figure 4. Contours of e and h + distribution as a parameter of V (1 to 6 V) in a single-gate p- HEMT. The e and h + concentrations under the gate increase abruptly at V larger than 3 V. Figure 6. Contours of e and h + distributions as a parameter of V (1 to 6 V) in a dual-gate p- HEMT. The e and h + concentrations under the floating gate increase abruptly at V larger than 4 V. J. Nanosci. Nanotechnol. 16, 1 5,

4 ual-ate p- ate HEMTs for teep ubthreshold lope Bae and Lee ºC 30 ºC 60 ºC 100 ºC Figure 7. I V curves of a dual-gate p- gate HEMT as a parameter of T (10, 30, 60, 100 C). injected into the Al and channel layer. Because V F is smaller than V, holes easily move to the floating gate and the channel under the floating gate. Resultant electron concentration under the floating gate increases abruptly and channel under the floating gate turns on as shown in Figure 6. This process is caused by impact ionization and a positive feedback, resulting in steep subthreshold slope. Figure 7 shows temperature (T dependence of the subthreshold slope in dual-gate device. As T increases, I slightly decreases due to optical phonon scattering Also, the hysteresis observed in drain leakage current (around a V of 2 V) decreases, which seems that the injected holes in channel and floating gate are removed quickly through thermal emission and trap-related conduction. However, there is no dependence on subthreshold slope, which means that the I increase does not depend on T, which is different from conventional field effect devices Transient Response I To investigate switching speed, transient responses of the I with V step is measured as plotted in Figure 8. Ramp-up time of the step bias is 10 7 sec and base time before measurement is 1 sec. When the V is changed from 0 to 6 and 8 V (from off to on state), the device turned on (I 1 ma) before measurement and it takes 10 s for the device to fully turn on (equivalently I 3 ma). imilarly, when the V is changed from 6 and 8 to 0 V, the device turns off within 10 s. The time to turn off depends on the V which is considered as the amount of holes in the channel and floating gate. ince the leakage level of our high-frequency I t measurement system is about 1 A in this current measure range. 4. CONCLUION We have characterized the mechanism of steep increase in I and I of p- gate HEMT. In device simulation, the breakdown of the parasitic floating base bipolar junction transistor is responsible for the I increase, since there is a positive feedback through hole generation due to the impact ionization in reverse biased chottky junction (gate/p-) and barrier lowing due to hole accumulation in p- layer. By the positive feedback, the drain current also can be increased very rapidly. In a dual-gate device which has the floating gate formed between the gate and the source, the steep increase of drain current at a threshold voltage of 4 V was observed. The subthreshold slope of the drain current was below 120 V/dec. The channel under the floating gate turns on quickly by the hole generated from positive feedback process in main gate. Characteristics in the dual-gate device have no appreciable change with temperature. To see the switching speed of the dual-gate device, I transient response were measured. ince the injected holes in floating gate and dominantly controls the I, switching speed of the dualgate device is about 10 s, which is considered as hole charging/ discharging time. V step (V), V = 1 V 0 to 6 6 to 0 0 to 8 8 to 0 Acknowledgment: This work was supported by the Center for Integrated mart ensors funded by the Ministry of cience, ICT and Future Planning as lobal Frontier Project (CI-2012M3A6A ). Time (sec) Figure 8. Transient responses of I as a parameter of V switching. The V steps from 0 V to 6 V (or 8 V) (turn-on) and from 6 V (or 8 V) to 0 V (turn-off). It takes about 10 s for dual-gate device to get steady state. References and Notes 1. Power - Technologies for Power Electronic Applications: Industry and Market tatus and Forecasts Yole evelopment (2012), pp M. A. Khan, Q. Chen, C. J. un, J. W. Yang, M. Blasingame, M.. hur, and H. Park, Appl. Phys. Lett. 68, 514 (1996). 3. Y. Uemoto, M. Hikita, H. Ueno, H. Matsuo, H. Ishida, M. Yanagihara, T. Ueda, T. Tanaka, and. Ueda, IEEE Trans. Electron evices 54, 3393 (2007). 4. M. Meneghini, C. de anti, T. Ueda, T. Tanaka,. Ueda, E. Zanoni, and. Meneghesso, IEEE Electron evice Lett. 33, 375 (2012). 4 J. Nanosci. Nanotechnol. 16, 1 5, 2016

5 Bae and Lee 5. O. Hilt, A. Knauer, F. Brunner, E. Bahat-Treidel, and J. Wurfl, Proc. IP (2010), pp O. Hilt, F. Brunner, E. Cho, A. Knauer, E. Bahat-Treidel, and J. Wurfl, Proc. IP (2011), pp X. Hu,. imin, J. Yang, M. Asif Khan, R. aska, and M.. hur, Electron. Lett. 36, 753 (2000). 8. T. Mizutani, M. Ito,. Kishimoto, and F. Nakamura, J. Appl. Phys. 77, 657 (1995). 9. I. Hwang, H. Choi, J. Lee, H.. Choi, J. Kim, J. Ha, C.-Y. Um,.-K. Hwang, J. Oh, J.-Y. Kim, J. K. hin, Y. Park, U.-I. Chung, I.-K. Yoo, and K. Kim, Proc. IP (2012), pp I. Hwang, J. Kim, H.. Choi, H. Choi, J. Lee, K. Y. Kim, J.-B. Park, J. C. Lee, J. Ha, J. Oh, J. hin, and U.-I. Chung, IEEE Electron evice Lett. 34, 202 (2013). 11. I. Hwang, J. Oh, H.. Choi, J. Kim, H. Choi, J. Kim,. Chong, J. hin, and U.-I. Chung, IEEE Electron evice Lett. 34, 605 (2013). ual-ate p- ate HEMTs for teep ubthreshold lope 12. J.-H. Bae,. Hwang, J. hin, H.-I. Kwon, C. H. Park, H. Choi, J.-B. Park, J. Kim, J. Ha, K. Park, J. Oh, J. hin, U.-I. Chung, K.-. eo, and J.-H. Lee, Int. El. evices Meet. (2013), pp Atlas User s Manual, ilvaco Inc., anta Clara, CA (2012). 14. H. Nagai, Q.. Zhu, Y. Kawaguchi, K. Hiramatsu, and N. awaki, Appl. Phys. Lett. 73, 2024 (1998). 15. Y. Uemoto, M. Hikita, H. Ueno, H. Matsuo, H. Ishida, M. Yanagihara, T. Ueda, T. Tanaka, and. Ueda, IEEE Trans. Electron evices 54, 3393 (2007). 16. L. Hsu and W. Walukiewicz, Phy. Rev. B 56, 1520 (1997). 17. B. L. elmont, M. hur, and M. troscio, J. Appl. Phys. 77, 657 (1995). 18. J. F. Zhang, Y. Hao, J. C. Zhang, and J. Y. Ni, ci. China er. F, Inf. ci. 51, 780 (2008). 19. K. Lee, M.. hur, T. J. rummond, and H. Morkoç, J. Appl. Phys. 54, 6432 (1983) Zanato,. okden, N. Balkan, B. K. Ridley, and J. chaff, emicond. ci. Technol. 19, 427 (2004). Received: 20 March Accepted: 20 April J. Nanosci. Nanotechnol. 16, 1 5,