IEEE ELECTRON DEVICE LETTERS, VOL. 22, NO. 8, AUGUST 2001
373
RESURF AlGaN/GaN HEMT for High Voltage Power Switching Shreepad Karmalkar, Jianyu Deng, Michael S. Shur, Fellow, IEEE, and Remis Gaska
Abstract—A novel HEMT configuration based on the RESURF technique is proposed for very high voltage power switching applications. It employs a p-n junction below the 2-DEG channel and two field plates, one extending from the gate and the other from the drain, to distribute the electric field over the gate to drain separation. 2-D simulations indicate a breakdown voltage 1 KV at on-resistance of 1 m -cm2 (neglecting contact resistances) for the device.
Index Terms—Breakdown voltage, field plate, GaN HEMT, RESURF concept.
m -cm for a HEMT [4], which are much below the ideal. 2D-simulations of a FP-HEMT structure indicate that, even after careful optimization of the FP geometry [8], signifiis possible only by strongly varying cant further increase in the insulator thickness (by a factor 5 or more) along the FP length [9]. However, processing such a variable thickness insulator of controlled dimension may be difficult [10]. As we show in this paper, the RESURF approach [11] combined with FPs over a uniform insulator should allow us to achieve KV.
I. INTRODUCTION
F
IELD EFFECT transistors (FETs) with several hundred ) volts breakdown voltage ( ) and low on-resistance ( will find applications in power switching for factory automation, telecommunications, and motor control [1]. The Baliga for a high figure of merit (BFOM, [1]) gives the ideal ), where , and voltage lateral FET as are the electron concentration, mobility and breakdown field . From this point of view, over the gate to drain separation, AlGaN/GaN HEMTs, where , and , all have high values, cm due to polarization efappear very attractive; cm /V-s due to 2-DEG, and MV/cm fects, due to large bandgap, have been practically attained [2]–[4]. By , the spreading the electric field over higher and higher can be increased up to the channel-substrate junction breakdown voltage, which is very high due to the low substrate doping. exceeding 1 KV As we show, an AlGaN/GaN HEMT with apshould be possible. Ideally, such a device could have proaching 0.1 m -cm according to BFOM. in hetSo far, efforts to spread the field and increase the erostructure FETs have utilized either a larger gate to drain separation [5], [6] and/or the extension of the gate as a field plate portion [7]–[9]. The (FP) over an insulator deposited on the best values practically achieved using the former technique have V at m -cm for a MOSHFET been V at [6], and using the latter technique have been Manuscript received April 17, 2001; revised May 14, 2001. The work at SET, Inc. was supported by Ballistic Missiles Defense/Innovative Science and Technology and managed by the Office of Naval Research (program monitor Dr. John Zolper). The review of this letter was arranged by Editor D. Ueda. S. Karmalkar is with the Electrical Engineering Department, Indian Institute of Technology, Madras 600 036, India, and also with the ECSE Department, Rensselaer Polytechnic Institute, Troy, NY 12180 USA (e-mail:
[email protected]). J. Deng and M. S. Shur are with the ECSE Department, Rensselaer Polytechnic Institute, Troy, NY 12180 USA. R. Gaska is with Sensor Electronic Technology, Inc., Latham, NY 12110 USA. Publisher Item Identifier S 0741-3106(01)06646-0.
II. DEVICE STRUCTURE AND PHYSICS The new structure (Fig. 1) called RESURF-HEMT utilizes, in addition to the gate FP employed earlier, a p-n junction below the heterojunction and a drain FP, for distributing the field over . Both the FPs are deposited on a uniform insulator ensuring process feasibility. The physics of operation is as follows. The p-n junction below the heterojunction depletes the 2-DEG by a vertical electric field, and thus reduces the slope of the horizontal electric field, as per the RESURF concept [11]. Since the junction doping is uniform while the voltage across the junc, increasing toward the drain, the 2-DEG tion varies along depletion solely due to this junction is weak near the gate and strong near the drain. The two FPs compensate this variation to distribute the horizontal electric field more equitably. The gate FP enhances the depletion near the gate due to the negative gate potential, while the drain FP suppresses the depletion near the drain due to the positive drain potential. Unlike the design procedure of existing RESURF devices, e.g., MOSFETs, MESFETs and BJTs [12], where the channel , in charge is optimized for the given p-layer doping and the RESURF HEMT, the p-region doping needs to be adjusted . The for the given 2-DEG channel concentration and RESURF principle is activated when the channel charge is completely depleted by the p-n junction at breakdown [11]. This criterion translates to a p-region doping and thickness of and respectively, based on the order simple one-dimensional junction theory, and assumptions of punch-through conditions and a uniform field in the p-region. KV for In our proposed design (Fig. 1), we targeted cm . III. SIMULATION The device is simulated using Silvaco-ATLAS. The main features of this simulation are as follows (see also [8]). The complex AlGaN layer charge distribution due to the components,
0741–3106/01$10.00 © 2001 IEEE
374
IEEE ELECTRON DEVICE LETTERS, VOL. 22, NO. 8, AUGUST 2001
Fig. 1. RESURF-HEMT structure. Parameter values employed in simulation are also shown; L and t are variable.
namely the polarization dipole, the insulator-donor layer interface and the ionized unintentional doping, is effectively modeled as a positive charge sheet, , along the heterojunction; decides the 2-DEG concentration. The is extracted directly from the simulated breakdown – curve, as the voltage at the intersection of the extrapolated rapidly rising and almost saturated segments of the – curve. The – curve is simu, lated for gate voltage, , equal to the threshold voltage, which is obtained by linear extrapolation of the – curve axis, for small (e.g., 0.05 V). The impact ionizato the , with measured values of tion is modeled as /cm and V/cm for GaN [3]. is not sensitive to the exact values of other The simulated material parameters (see [8]). IV. RESULTS AND CONCLUSION A simulated V is obtained [curves , Fig. 2] using the two FPs and the p-n junction, for p-region thickness m, m and other parameter values given for this device is 1.07 m -cm in Fig. 1. The simulated (neglecting contact resistances), which is similar to that of fabricated devices [4], [6] without the p-n junction. For a standard HEMT (without the FPs and the junction) having the same and 0.5 m thick n-substrate, simulations yield V m -cm , implying that the improved design and by 10 times, causing just 14% rise in the enhances the . Since simulations show that FPs do not affect , this rise can be traced to the slight depletion of the 2-DEG by the p-n junction, whose negative p-depletion charge attracts some of the and terminating on the 2-DEG. field lines emanating from The contributions of the p-n junction and drain FP are clearly brought out in Fig. 2. When the p-n junction is removed (curves , Fig. 2), falls drastically to 370 V, i.e., by a factor of 3, and the current at the threshold increases by an order of magV is lower than the maximum achievnitude. (This able value of 630 V reported for the FP structure [8], since the values of and in Fig. 1 are not the optimum [8]). The junction removal eliminates the field under the drain FP [curve , Fig. 2(b)], and thus, any effect of this FP. On the other hand, in
Fig. 2. Simulated characteristics showing the effects of the p-n junction and the gate FP. (a) I –V curves; (b) magnitude of the horizontal field distribution along the 2-DEG channel. Legend—A: PN junction and both FPs, B : FPs only, C : PN junction and gate FP. V V , which is 2.45 V for curves A and C , and 2.85 V for curve B .
0
=
0
the presence of the p-n junction, removal of the drain FP lowers to 625 V (curves , Fig. 2), i.e., by 40%. In the presthe ence of the p-n junction, the drain FP introduces an additional peak in the field distribution, and at the same time, reduces the sharp peak at the drain end, making the distribution more even [compare curves and , Fig. 2(b)]. Note that the use of the p-n junction alone (without any field plates) provides only a limited advantage. In this case, the simulated device threshold current has been found to be as high as 15 mA/mm at a drain voltage of 300 V. It is thus concluded that the p-n junction should be used in conjunction with both the FPs for maximum benefit. as a function of with and It is informative to plot the without the p-n junction (Fig. 3) to illustrate the role of the p-n junction. The plot shows an approximately linear rise followed by saturation as expected from the HEMT physics. Also given in the figure, is the ideal behavior based on assumptions of uniform , 1-D vertical field in the channel substrate juncfield over tion and breakdown field of 3 MV/cm. Clearly, for the device with FPs and the junction, the 2-D simulated saturation value
KARMALKAR et al.: RESURF AlGaN/GaN HEMT FOR HIGH VOLTAGE POWER SWITCHING
375
from the grounded substrate case (Fig. 3), confirming that the novel RESURF HEMT should work with both grounded as well as floating substrates. In a later publication, we shall discuss optimization of the device structure, and provide more details of the effects of a floating substrate on dc and switching characteristics. In conclusion, we proposed a novel very high voltage AlGaN/GaN HEMT structure with two FPs based on the RESURF technique, useful in power switching applications. REFERENCES
Fig. 3. Breakdown voltage as a function of the gate to drain separation. The V , which is 2.45 ideal curve assumes a breakdown field of 3 MV/cm. V V for devices with the p-n junction (floating or grounded substrate) and 2.85 V for devices without the p-n junction (i.e., field plates only).
=
0 0
approaches the ideal saturation value, which is the channel-sub. However, the average horizontal field over given strate by the slope of the linear portion of the 2-D simulated curve is 1.5 MV/cm, which is a factor of 2 below the ideal. Additional field spreading techniques are required to further improve this value. It is interesting to see the effect of setting the p-substrate, shown grounded in Fig. 1, floating, since a device with a floating p-substrate may be easier to fabricate. Simulation of the charge conditions within the device in this case shows V, a portion of the channel-substrate junction that, for near the source gets mildly forward biased. This is to force the current, which is injected into the substrate by the reverse biased portion of channel-substrate junction, back into the channel near the source. The forward bias reduces the effective device threshold voltage somewhat, due to the body effect. However, the breakdown voltage is not significantly different
[1] B. J. Baliga, “Trends in power semiconductor devices,” IEEE Trans. Electron Devices, vol. 43, pp. 1717–1731, Oct. 1996. [2] S. T. Sheppard, K. Doverspike, W. L. Pribble, S. T. Allen, J. W. Palmour, L. T. Kehias, and T. J. Jenkins, “High-power microwave GaN/AlGaN HEMT’s on semi-insulating silicon carbide substrates,” IEEE Electron Device Lett., vol. 20, pp. 161–163, Apr. 1999. [3] K. Kunihiro, K. Kasahara, Y. Takahashi, and Y. Ohno, “Experimental evaluation of impact ionization coefficients in GaN,” IEEE Electron Device Lett., vol. 20, pp. 608–610, Dec. 1999. [4] N.-Q. Zhang, S. Keller, G. Parish, S. Heikman, S. P. Denbaars, and U. K. Mishra, “High breakdown GaN HEMT with overlapping gate structure,” IEEE Electron Device Lett., vol. 21, pp. 421–423, Sept. 2000. [5] R. Gaska, Q. Chen, J. Yang, A. Osinsky, M. A. Khan, and M. S. Shur, “AlGaN/GaN heterostructure FET’s with offset gate design,” Electron. Lett., vol. 33, no. 14, pp. 1255–1257, July 1997. [6] G. Simin, X. Hu, N. Ilinskaya, A. Kumar, A. Koudymov, J. Zhang, M. Asif Khan, R. Gaska, and M. S. Shur, “A 7.5 kW/mm current switch using AlGaN/GaN metal–oxide–semiconductor heterostructure field effect transistors on SiC substrates,” Electron Lett., vol. 36, pp. 2043–2044, Nov. 2000. [7] J. Li, S. J. Cai, G. Z. Pan, Y. L. Chen, C. P. Wen, and K. L. Wang, “High breakdown voltage GaN HFET with field plate,” Electron Lett., vol. 37, pp. 196–197, Feb. 2001. [8] S. Karmalkar and U. K. Mishra, “Enhancement of breakdown voltage in AlGaN/GaN high electron mobility transistors using a field plate,” IEEE Trans. Electron Devices, vol. 48, pp. 1515–1521, Aug. 2001. [9] S. Karmalkar and U. K. Mishra, “Very high voltage AlGaN/GaN HEMT’s using a field plate deposited on a stepped insulator,” Solid-State Electron, to be published. [10] B. J. Baliga, Modern Power Devices. New York: Wiley, 1987. [11] J. A. Appels and H. M. J. Vaes, “High voltage thin layer devices,” in IEDM Tech. Dig., 1979, pp. 238–240. [12] A. W. Ludikhuize, “A review of RESURF technology,” in ISPSD’2000, pp. 11–18.
