The critical charge density of 4H-SiC thyristors

IEEE TRANSACTIONS ON ELECTRON DEVICES, VOL. 45, NO. 1, JANUARY 1998 307 The Critical Charge Density of 4H-SiC Thyristors Michael E. Levinshtein, John W. Palmour, Member, IEEE, Sergey L. Rumyanetsev, and Ranbir Singh Abstract— The critical charge density which determines the maximum voltage ramp (dV =dt) of a thyristor, the minimum value of the gate control current, and the parameters of current filamentation has been determined in SiC thyristors both theoretically and experimentally. For 4H-SiC thyristors blocking 300 to 400 V, the critical charge density has been found to be 15 03 and 1014 cm03 at a forward voltage of 5 and 100 V, 22 10 cm respectively for an operating temperature of 560 K. The critical current density j0 below which the turn-on state is localized has also been estimated theoretically and experimentally. While theoretical calculations predict its value to be 2 2 102 A/cm2 ; experimental results show a range of 3 2 102 to 7:6 2 102 A/cm2 : I. INTRODUCTION T HYRISTORS made with silicon carbide (SiC) are expected to show great performance advantages over those made with Si or GaAs because of SiC’s higher breakdown electric field, carrier saturation velocities [1] and thermal conductivity than either Si or GaAs. A high breakdown electric field allows the design of SiC thyristors with thinner and higher doped base layers than identically rated Si or GaAs thyristors. Such devices may be expected to show fast switching and low residual voltage drop at very high current densities. A high thermal conductivity should allow higher current density operation of SiC thyristor. The first p-n-p-n SiC thyristors were demonstrated in late 80’s [2], [3]. 6H-SiC [4] and 4HSiC [1], [5] thyristors have been demonstrated recently. A large bandgap of SiC is also expected to result in a much higher operating temperature and a higher radiation hardness than Si and GaAs thyristors. SiC thyristors have been shown to operate successfully at 800 K [6], [7]. Recent 4H-SiC thyristors produced by Cree Research, Inc., have shown a high forward blocking voltage of 400 V and a residual voltage drop of only 4 V at forward current density of A/cm The main parameters for the thyristor turn-on process are: the time constant of the current rise time , the critical charge density , the residual voltage during turn-on process , and the spread velocity of the “on” state Recently and dependencies have been investigated for 4H-SiC thyristors [8]. This paper addresses the need to evaluate the critical charge density of these thyristors. Manuscript received June 7, 1996; revised April 16, 1997. The review of this paper was arranged by Editor T. P. Chow. This work was sponsored in part by the Office of Naval Research under the Manufacturable University Research Initiative (Contract N00014-95-1-1302) and NASA Lewis Research Center (Contract NAS3-26927). M. E. Levinshtein and S. L. Rumyanetsev are with the Ioffe Institute of Russian Academy of Science, St. Petersburg, Russia. J. W. Palmour and R. Singh are with Cree Research, Inc., Durham NC 27703 USA. Publisher Item Identifier S 0018-9383(98)00288-3. II. THE CRITICAL CHARGE DENSITY CONCEPT The concept of the critical charge density was put forward in [9] and developed in [10], [11]. The critical charge density of of a thyristor determines the maximum voltage ramp the thyristor [11], the minimum value of the gate controlcurrent density [12], the spread velocity of the “on” state [13], the holding current, and the parameters of current filamentation in gate-controlled thyristors [14]. This concept unifies the numerous methods known to turn-on a thyristor. These methods can be divided into two groups: pulse methods and dc methods. The pulse methods include turn-on in , short gate pulse, short light a thyristor induced via high pulses etc. The term “short pulse” implies that the duration is much less than the minority of the actuating pulse carrier lifetime in the thinner and higher doped base of the thyristor. The minimum critical charge , required to turn on the thyristor is an intrinsic parameter to describe the “sensitivity” of the thyristor. It is noteworthy that the current flowing through the thyristor cannot be used as a measure of this sensitivity. This is because a thyristor would not turn on in the limiting case of -function current pulse when the maximum current achieves an infinite value independent of the charge generated in the thyristor. The dc methods of thyristor turn-on include very slow (quasistatic) increasing of the anode voltage or gate current; varying temperature or light intensity at a constant anode and gate voltages, etc. In such cases, the minimum current is a suitable parameter to describe the sensitivity of the thyristor. Depending upon the turn-on method, this could be either the gate current or the anode current. The charge generated in the thyristor cannot be used as an intrinsic parameter because in the limiting case of very slow change in the actuating current, the charge tends to infinity. It has been shown in [10], [11] that the concept of the critical charge density can be used for dc turn on methods also, but in this case, the critical charge represents the excess charge of the minority carriers and not the charge generated by instantaneous changes in operating conditions. For example, in the case of turn-on of the thyristor by gate current at constant anode voltage, the more is gate current, the more is the charge of minority carriers in the bases of the thyristor. which is required to turnMinimal minority carrier charge on the thyristor represents a suitable parameter to describe the sensitivity of the thyristor. and the critical charge It has been shown [10], [11] that are of the same order of magnitude for a thyristor. Thus, the goal of the critical charge density concept is to provide a unified criterion of thyristor turn-on sensitivity. References [10], [11] establish the relation between these quantities, the 0018–9383/98$10.00  1998 IEEE 308 IEEE TRANSACTIONS ON ELECTRON DEVICES, VOL. 45, NO. 1, JANUARY 1998 on the minority carriers distribution along dependence of the thyristor bases, their dependence on pulse duration etc. The critical charge density depends upon the area where the thyristor is turned on. It may be the whole area of the thyristor, if the thyristor is turned on with the effect, or a very small portion if the thyristor is turned on with a small beam of light (see, for example, [15]). Hence, critical charge per unit activated area is a more intrinsic parameter. But the most intrinsic parameter is the critical charge per unit volume of the thyristor. It is well known, for example, that the turn-on of the thyristor by gate current results in an inhomogenous distribution of excess minority charge along the thin higher doped base. The density of excess charge is maximum near the gate electrode and decays monotonically along the base [16]. Hence, if the gate current increases slowly, the thyristor is turned-on at the point near the gate electrode where the minority carrier charge density is maximum and is equal to the critical charge density. With the critical charge known, the volume critical charge density is given by (1) where is the thyristor area, is the thickness of the thin base. Expression (1) contains only the thickness of the thin base and not the sum of the thicknesses of the thyristor bases (the thick and the thin base). This is because for equal charge introduced to each of the bases, the efficiency of the minority charge is much more in the thin base of the thyristor [10], [11]. The volume critical charge density is often expressed as (a) (2) where is the electron charge. In such a form, the value of critical charge density can be compared conveniently with the base doping density and with the minority carrier concentration at the emitter p-n junction. For silicon thyristors, the critical charge density falls in the range of 10 cm for power high-voltage thyristors to 10 cm for special thyristors designed with very low values and very low gate control current values [17], [18]. The same range of values is typical for GaAs thyristors [19]. To our knowledge, the critical charge density of SiC thyristors has not been determined yet. In this study, 4H-SiC thyristors manufactured at Cree Research, Inc., were used. The design of the devices is the same as reported previously in [8]. Schematic diagrams for the device (top view and the cross sections of the thyristor) are shown in Fig. 1(a)–(c). The operating area of the device was cm , the thickness of the p-base was m, the thickness of the n-base was m The doping level of the n-base was cm The maximum forward breakover voltage was 300 to 400 V. To determine the critical charge density the well-known “ technique” was used. A forward dc voltage was applied to the thyristor. Then, an additional pulse of amplitude (rise time ns) was applied. The 50 pulse generator TR 0306 with the rise time of 0.3 ns was used to form the pulse. High-frequency 50 circuit with the low (b) (c) Fig. 1. Schematic diagram of the 4H-SiC thyristor structure under investigation. (a) Top view. (b) and (c) Cross section views of the thyristor. LEVINSHTEIN et al.: CRITICAL CHARGE DENSITY OF 4H-SiC THYRISTORS 309 (a) (a) (b) (b) Fig. 2. (a) The dependencies of the critical charge density ncr versus forward voltage V0 : 1) – 4H-SiC thyristor, 560 K; 2 – GaAs thyristor, 300 K. (b) The dependence of the central (collector) p-n junction capacitance Cjc versus forward voltage drop V0 for the 4H-SiC thyristor, 560 K. 1.2 nH) was used to apply the parasitic inductance value pulse to the anode of the thyristor. The minimum value of sufficient to turn on the thyristor was registered. The magnitude of the charge appearing in both bases of the thyristor is equal to (3) is the capacitance of the central (collector) p-n where junction of the thyristor [see Fig. 1 (b), (c)]. The critical charge density was calculated according to (1). The dependence was measured at a frequency of 1 kHz in the range of forward voltage with an accuracy of 10 pF. III. RESULTS AND DISCUSSION A. Critical Charge Density at High Temperatures K and forward voltage V, the At switching in 4H-SiC thyristors is qualitatively analogous to the effect in Si and GaAs thyristors. The curve (1) in Fig. 2(a) shows the dependence of the critical charge density versus forward voltage Curve (2) shows the dependence for GaAs thyristor with approximately the same breakover voltage [20]. The critical charge value for the SiC thyristor is rather small pC at V and pC for V. However, due to a small -base thickness m and very small thyristor cm , the critical charge density is area sufficiently high: cm at V and cm at V. 1 Fig. 3. (a) The dependencies of the pulse amplitude V turned-on the 4H-SiC thyristor by the dV =dt effect versus forward voltage V0 at 300 K. The pulse duration (ns): 1 – 20, 2 – 130, 3 – 400. (b) The same dependencies for commercial Si thyristor KU-103, 300 K. The pulse duration (ns): 10 – 20, 20 – 130, 30 – 400. B. The Effect at Room Temperature K the effect has been found to be At values for a SiC anomalous within the whole range of the thyristor. One can see this easily by comparing the dependencies for SiC [Fig. 3 (a)] and Si [Fig. 3 (b)] thyristors. The usual effect is characterized by an “infinitely long” current pulse. That is, the exponential increase of the pulse as anode current begins before the end of the given by (3). It is possible to turn-on the thyristor by a shorter pulse duration, but that requires a correspondingly larger pulse It can be seen from the inset of Fig. 3(b) that for a Si thyristor, a smaller pulse duration requires a larger magnitude in the whole range of dc voltage value This dependencies is typical for all Si and GaAs kind of thyristors at any temperatures. One can see in Fig. 3(a) that the opposite is true for SiC V, a smaller gate pulse duration requires thyristor. At magnitude to turn on the thyristor. a smaller To appreciate better this very unusual situation, let us V, a pulse is imagine for example that at is applied to the thyristor. The magnitude of the pulse ns. However, large enough to turn-on the thyristor at if the pulse is not interrupted after 20 ns, the thyristor is not turned on. Hence, the applied pulse impedes the turn on V, does the dependence process. Only at take the usual form [compare with Fig 3(b)]. There is no clear physical picture of this phenomenon at present. Investigations show that two effects take part for this phenomenon. K and , the capacitance of First, at is sufficiently larger than the the collector p-n junction (Fig. 1). A high capacitance of the emitter junction 310 IEEE TRANSACTIONS ON ELECTRON DEVICES, VOL. 45, NO. 1, JANUARY 1998 Fig. 4. Capacitance–voltage dependencies measured between gate and cathode electrodes. 1–3 forward voltage, T(K): 1 – 300, 2 – 390, 3 – 560. 4–5 reverse voltage, T(K): 4 – 300, 5 – 390. 6 forward voltage for Si thyristor KU-103, T = 300 K. ratio does not occur commonly in Si and GaAs thyristors. It will be shown that at , the turn on effect changes qualitatively. It can even be process by said that the term “ effect” is invalid at Secondly, the spectrum of the deep levels near junction is essentially different from the spectrum of the deep levels junction. This follows directly from the temperature near dependence of the effect. In Fig. 4 several curves displaying the dependence of the capacitance As a result, value increases with an increase in anode bias (curves 1–3 in Fig. 4). A large enough bias results in the condition A further increase in the bias leads to a decrease in the total capacitance A similar relation has been observed for SiC thyristors at forward bias (curves 1–3). At reverse bias, dependencies have the usual form (compare curves 4 and 5 with curve 6). One can see from Fig. 4 that and junctions are characterized not only by the different capacitance values at , but also by the different temperature dependencies of the capacitance. This directly proves a different spectrum of the deep levels near the junctions. The pulse applied is divided between , , and junctions. Because the voltage drop on can be neglected. Having at as it is usually realized for Si and GaAs thyristors, the pulse is almost completely applied to the central junction, provided to V. Majority carriers are displaced into the thyristor bases by the expanded space charge region of the central junction. These excess majority carriers initiate the injection of minority carriers from emitter p-n junctions. In this way, the usual mechanism is realized. , the essential As can be see from Fig. 3, at different mechanism of the turn on is realized in 4H-SiC thyristor provided K. C. Spread of Turn-On State in 4H-SiC Thyristors (4) versus bias voltage are shown. The voltage bias was applied between the gate and cathode contacts. For curves 1–3, the polarity of the voltage corresponds to the reverse bias on the junction and to the forward bias on the junction (forward bias on the thyristor). Dashed lines 4, 5 represent the dependencies for another polarity of voltage between the gate and cathode contact. Dotted curve 6 in Fig. 4 represents the dependence for commercial Si thyristor KU-103 (forward bias). In Si and GaAs thyristors, the doping level of the thick base is usually the same near both junctions. In this case, at As forward bias is applied, decreases and increases. Even at to is the built in potential), the condition is fulfilled and value is practically equal to (4). For Si thyristors the condition is valid if to V; and for GaAs thyristors—if to V. However, if at , the other situation is realized. A major part of the bias applied falls on the junction The capacitance increases with an increase in forward voltage as value is practically constant. At (5) It is well known that for every thyristor structure, there exists a critical current density at which the turned-on state does not spread and occupies only part of the structure. Under this condition, the critical density of minority carrier charge in the thin base of the thyristor is equal to [13]. For GaAs and Si thyristors, the relationship between and has been investigated in [19]. The value of can be estimated using the simplest model [13], [21]. It is assumed that in the turned-on state the minority carrier concentration is constant along the thyristor bases. For this case (6) For thin n-base of the thyristor, one can assume that the common-base current gain is equal to 0.8. Using the usual expression for the common base circuit configuration gain, it is easy to obtain the hole lifetime in the thin base of the thyristor: s The diffusion coefficient cm s has been taken for this calculation. To estimate the electron lifetime in the wide thyristor base , one can assume the usual relation between wide base width and electron diffusion length Using the electron diffusion coefficient cm s a of s was derived. to be cm (see Fig. 2), we Taking a value of obtain a value of A/cm The thyristor structures investigated in this paper are not designed specifically to study LEVINSHTEIN et al.: CRITICAL CHARGE DENSITY OF 4H-SiC THYRISTORS 311 In Fig. 5(b), the boundaries of the on-state region are shown for the same thyristor structure at mA. Knowing the anode current value and the area of the onstate , one can find the value. The experimental values of fall in the range of and A/cm Taking into account the qualitative character of the theoretical model an agreement between theoretical and experimental estimations are quite satisfactory. IV. CONCLUSION (a) has been The value of the critical charge density investigated in SiC thyristors for the first time. The effect has been used to investigate the voltage and temperature dependencies of the critical charge density. At high temperatures K the “ effect” characteristics are analogous qualitatively to the effect in Si and GaAs thyristors. At 560 K, the value is equal to approximately cm at a forward voltage of 5 V and decreases monotonically with an increase in anode voltage. At an anode voltage of 100 V, is approximately 10 cm At K, the “ effect” is anomalous within the whole range of the forward voltage values: At V, the smaller duration of the “ pulse,” the smaller is the pulse amplitude of the turned on the thyristor. The possible reasons of such a condition have been considered. Based on the obtained value, the critical current density has been estimated using a simple model is the anode current density below which the anode current is localized). Calculations give A/cm Experimental values for are between to A/cm REFERENCES (b) Fig. 5. On-state in 4H-SiC thyristor at different values of anode current Ia : (a) Ia = 3:8 mA. (b) Ia = 6:6 mA. the spread of the on-state [18], [19], [22]. However, one can value experimentally. estimate the is high enough 15 mA), If the anode current value the on-state occupies all areas of the structure. This can be judged from the recombination radiation [19], [22] which for SiC is in a visible wavelength. The radiation can be observed easily through an optical microscope. It is worth noting that only a portion of the electrode shines. Light scattering into the substrate produces a luminous “halo” around the “on” parts of the structure. This halo allows one to judge qualitatively on the localization of the on-state. In some thyristors, one can observe that a decreasing anode current localizes the on-state to a part of the structure (see Fig. 5). In Fig. 5(a), the boundaries of the luminous halo are shown of 3.8 mA. Since it is difficult at the anode current to observe the on-state distribution under the opaque anode electrode, it was assumed that the boundary of the on-state under the electrode is a straight line [dashed line in Fig. 5(a)]. [1] P. P. Joshi, “Monte Carlo calculations of the temperature- and fielddependent electron transport parameters for 4H-SiC,” J. Appl. Phys., vol. 78, p. 5518, 1995. [2] V. A. Dmitriev, S. N. Vainshtein, M. E. Levinshtein and V. E. Chelnokov, “Silicon carbide dynistor,” Sov. Tech. Phys. Lett., vol. 13, p. 6, 1987. [3] , “First SiC dynistor,” Electron. Lett., vol. 24, p. 1032, 1988. [4] J. A. Edmund, J. W. Palmour, and C. H. Carter, Jr., “Junction devices in 6H-SiC,” in Proc. Int. Semicond. Dev. Res. Symp., p. 487 Dec. 4-6, 1991. [5] J. W. Palmour, J. A. Edmund, H. S. Kong, and C. H. Carter, Jr., “Vertical power devices in silicon carbide,” in Silicon Carbide and Rel. Mater. Inst. Phys. Conf. Series, 1994, vol. 137, p. 499. [6] J. W. Palmour, S. T. Allen, and D. G. Waltz, “4H-SiC power switching devices,” in Silicon Carbide and Rel. Mater. Inst. Phys. Conf. Series, 1995, vol. 142, p. 813. [7] A. N. Andreev, A. M. Strel’chuk, N. S. Savkina, F. M. Snegov, and V. E. Chelnokov, “Study of SiC-6H dynistor structure,” Semiconductors, vol. 26, p. 561, 1995. [8] M. E. Levinshtein, J. W. Palmour, S. L. Rumyanetsev, and R. Singh, “Turn-on process of 4H-SiC thyristors,” IEEE Trans. Electron Devices, vol 44, p. 1177, July 1997. [9] K. Weiser, R. S. Levit, M. I. Nathan, G. Burns, and J. Woodall, “Indium Phosphide laser characteristics,” Trans. AMIE, 1964, p. 230. [10] A. I. Uvarov, “Critical turn-on charge of a thyristor,” in Physics of Electron-Hole Junctions and Semiconductor Devices, S. M. Ryvkin and Yu V. Shmartsev, Eds. New York, NY: Consultants Bureau, 1971, pp. 170–179. [11] , “Conditions for turning on a thyristor by short gate current pulses,” in Physics of p-n Junctions and Semiconductor Devices, S. M. Ryvkin and Yu. V. Shmartsev, Eds. New York, NY: Consultants Bureau, 1971, pp. 216–223. [12] M. I. Dyakonov and M. E. Levinshtein, “Theory of propagation of the turned-on state in a thyristor in the presence of a gate current,” Sov. Phys. Semicond., vol. 12, p. 992, 1978. 312 [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] IEEE TRANSACTIONS ON ELECTRON DEVICES, VOL. 45, NO. 1, JANUARY 1998 , “Theory of propagation of the turned-on state in a thyristor,” Sov. Phys. Semicond., vol. 12, p. 426, 1978. , “Parameters of a current filament in a gate-current-controlled thyristor and its turn-off gain,” Sov. Phys. Semicond., vol. 14, p. 283, 1980. V. A. K. Temple, “Controlled turn-on thyristor,” IEEE Trans. Electron Devices, vol. ED-30, p. 816, 1983. J. R. Hauser, “The effects of distributed base potential on emitter current injection density and effective base resistance for stripe transistor geometry,” IEEE Trans. Electron Devices, vol ED-11, p. 238, 1964. M. Suzuki, N. Sowaki, K. Iwata, and T. Nishinaga, “Current distribution at the lateral spreading of electron-hole plasma in the thyristor,” IEEE Trans. Electron Devices, vol ED-29, p. 1222, 1982. H. Dodson and R. L. Longini, “Probed determination of turn-on spread of large area thyristors,” IEEE Trans. Electron Devices, vol. ED-13, p. 478, 1966. S. N. Vainshtein, Yu. V. Zhilyaev, and M. E. Levinshtein, “Propagation of turned-on state in gallium arsenide thyristors,” Sov. Phys. Semicond., vol. 21, p. 77, 1987. S. N. Vainshtein, I. I. Diakonu, Yu. V. Zhilyaev, and M. E. Levinshtein, “Fundamental switching parameters of gallium arsenide thyristors,” Sov. Tech. Phys. Lett., vol. 28, p. 359, 1983. R. L. Longini and J. Melngailis, “Gated turn-on of four layer switch,” IEEE Trans. Electron Devices, vol. ED-10, p. 178, 1963. H. Yamasaki, “Experimental observation of the lateral plasma propagation in a thyristor,” IEEE Trans. Electron Devices, vol. ED-22, p. 65, 1975. Michael E. Levinshtein received the M.S.S.E. degree from Leningrad Electrotechnical Institute, Leningrad, USSR, in 1963, the Ph.D. degree in physics from A.F. Ioffe Institute of Physics and Technology, Leningrad, in 1970, and the Doctor of Science (Habilitation) degree in physics from A.F. Ioffe Institute of Physics and Technology in 1980. Since 1967, he has been with A.F. Ioffe Institute of Physics and Technology. His research has included hot electrons, Gunn effect, power and superpower Si, SiC, and GaAs devices, low-frequency and 1/f noise in semiconductors and semiconductor devices. He has published five books and is a co-editor of two collections of the best Russian papers in semiconductor physics and technology: Best of Soviet Semiconductor Physics and Technology, (1987–1988), (American Institute of Physics, 1991), and Best of Soviet Semiconductor Physics and Technology, (1989–1990), (Singapore: World Scientific, 1995). He is a Principal Scientist of Ioffe Institute of Russian Academy of Science, Professor of St. Petersburg Technical State University, and a Visiting Professor of the University of Virginia, Charlottesville. John W. Palmour (M’95) received the B.S. and Ph.D. degrees from North Carolina University, Raleigh, in 1982 and 1988, respectively. His major was in materials science and engineering with a minor in electrical engineering. His doctoral research concentrated on processing techniques and transistor development in SiC devices, and he demonstrated an SiC MOSFET operating at 650  C: After graduating, he became a co-founder of Cree Research, Inc., Durham NC, where he has been Senior Scientist since its formation, concentrating on device processing techniques and transistor development in SiC. He has been responsible for the development of high-voltage, high-temperature 4H-SiC power MOSFET’s and thyristors, as well as high-frequency MESFET’s and planar n-channel and p-channel 6H-SiC MOSFET’s. He has coauthored over 100 publications in various conference proceedings and refereed journals and is an inventor on 13 issued U.S. patents and seven foreign patents concerning semiconducting SiC. He also serves on the Board of Directors for Cree Research, Inc. Sergey L. Rumyanetsev received the M.S.E.E. degree from Leningrad Electrotechnical Institute, Leningrad, USSR, in 1977, the Ph.D. degree in physics from Leningrad Polytechnical Institute in 1986, and the Doctor of Science (Habilitation) degree from A.F. Ioffe Institute of Physics and Technology, Lenningrad, in 1996. From 1977 to 1980, he was with A.F. Ioffe Institute of Physics and Technology. From 1977 to 1989, he was with Industrial and Scientific Corporation “Svetlana.” Since 1989, he has been a Senior Scientist of A.F. Ioffe Institute of Physics and Technology. His research has included computer simulations and experimental investigations of microwave devices, experimental and theoretical investigations of low frequency noise in semiconductors and semiconductor devices. His current research interest include low frequency noise, wide bandgap semiconductors, and conducting polymers. He published a number of papers and he is a coeditor of one (Singapore: World Scientific, 1996). Ranbir Singh received the B.Tech degree from the Indian Institute of Technology, New Delhi, in 1990, and the M.S. and Ph.D. degrees from North Carolina State University, Raleigh, in 1992 and 1997, respectively, all in electrical engineering. His graduate experience included exposure to a wide variety of both bipolar and MOS families of devices, with his specialty in characterizing the cryogenic operation of Si power devices. In August 1995, he joined Cree Research, Inc., Durham NC, where he conducts research on SiC power devices. His interests include development of SiC power MOSFET’s, thyristors, Schottky diodes, and novel power devices in SiC.