Materials Science orum Vols. 556-557 (7) pp 965-97 online at http://www.scientific.net (7) Trans Tech Publications Switzerlan Online available since 7/Sep/5 -base Power Converters for High Temperature Applications Leon M. Tolbert a Hui Zhang b Mahu S. Chinthavali c Burak Ozpineci The University of Tennessee Electrical an Computer Engineering Knoxville Tennessee USA Oak Rige National Laboratory National Transportation Research Center Knoxville TN USA a tolbert@utk.eu b hzhang8@utk.eu c chinthavalim@ornl.gov ozpinecib@ornl.gov Keywors: converter inverter high temperature hybri electric vehicle Abstract. As commercial-grae silicon carbie () power electronics evices become available the application of these evices at higher temperatures or frequencies has gaine interest. This paper contains temperature-epenent loss moels for ioes an JETs an simulations for ifferent power converters that are useful for preicting the efficiency of these converters. Aitionally tests to characterize the static an ynamic performance of some available evices are presente to give further insight into the avantages that might be gaine from using evices instea of evices for hybri electric vehicle applications. Introuction Increasing eman for more efficient higher power an higher temperature operation of power converters has le to the evelopment of wie bangap semiconuctor power electronic evices an in particular silicon carbie () power electronics. The high temperature operation capability of these evices increases the power ensity of converters incorporating them because of reuce thermal management an heat sink requirements that yiel reuce weight an cost. To ultimately gain full avantage of power electronics evices requires high temperature packaging techniques gate rives an passive components [-4]. Device an system-level temperature-epenent loss moels have been evelope for transportation applications an utility applications to ai in gauging the impact that these new evices might have in terms of efficiency an heat sink requirements [5-7]. Device an system loss moels for ioes JETs an MOSETs have shown that they will become the evices of choice once issues such as fabrication an packaging of these evices are solve. At present Schottky ioes are the only commercially available evices. These ioes are use in several applications an have prove to increase the system efficiency compare with evice performance when use as antiparallel ioes in brige converters [8-9]. gnificant reuction in weight an size of power converters with an increase in the efficiency is projecte. Some applications such as hybri electric vehicles require that evices be able to hanle extreme environments that inclue a wie range of operating temperatures (from -4ºC to ºC). In the following sections moels an experimental tests for the static an ynamic performance of some commercially available Schottky ioes an experimental samples of JETs an MOSETs in a wie temperature range will be presente. Temperature Depenent Moeling of Schottky Dioes In this section Schottky ioes are characterize by both theoretical analysis an experiments. The moels for static state (forwar conuction) an ynamic state (reverse recovery) will be iscusse. Schottky Dioe Static State. The structure of a power Schottky ioe an its equivalent circuit are shown in ig.. V B is the voltage rop across the Schottky barrier; R D is the resistance of the lightly ope rift region; R S an R C are the resistances of substrate an contact respectively. All rights reserve. No part of contents of this paper may be reprouce or transmitte in any form or by any means without the written permission of the publisher: Trans Tech Publications Lt Switzerlan www.ttp.net. (ID: 6.36.5.3-8/8/87::48)
966 licon Carbie an Relate Materials 6 Metal N- Drift Region N+ Substrate Contact.4.35.3 mulation Experiment CSD5 V B R D R S R C Resistance (ohm).5..5 CSD ig.. (Above) Basic structure an equivalent circuit of Schottky power ioes...5 CSD6 ig.. (Right) On-state resistance of Schottky ioes with respect to temperature (Cree). 73 33 373 43 473 53 Junction Temperature (K) In a Schottky power ioe the thermionic emission process oates in the transport of current across a metal n-type semiconuctor contact []. Uner the forwar bias conition the current flow across the Schottky barrier is given by J qv ( B B )/ kt CT e () where B is the barrier height between the metal an n-type semiconuctor T is the absolute temperature q is the charge of an electron k is Boltzmann s constant C is Richarson s constant which is given by 4 mk q C 3 () h where m is the effective mass of an electron an h is Plank s constant []. or 4H- the theoretical value of Richarson s constant is 46 Acm - K - []. Solving Eq. for V B an neglecting R S an R C (because they are usually small compare to R D for power evices with breakown voltages larger than V) then the total voltage rop across a Schottky power ioe can be expresse as V kt q J CT B ln J R D where R D is given by R D 4V B 3 Ecn. (3) V B is the breakown voltage; E c is the breakown electrical fiel; n is electron mobility which is a function of oping ensity N electrical fiel E an temperature T in the epletion layer. n can be expresse as n E vs Ntot Nref v6k vs ( T) T.8exp 6 (4) T A 3 B T A 3 B N T A. (5) 3 ref N ref
Materials Science orum Vols. 556-557 967 Eqs. 3-5 give a comprehensive escription of the static characteristics of Schottky power ioes. The coefficients involve can be foun in many papers an only the breakown voltage of the evices is neee to solve the equations []. Table has a list of the parameters use to moel the forwar characteristic for a V/A 4H- Schottky power ioe. Differentiation of Eq. 3 with respect to J yiels the on-state specific resistance of a power Schottky ioe: R sp V kt R D. (6) J q J Eq. 6 shows the epenence of the ioe s on-state specific resistance to temperature an forwar current. TABLE. PARAMETERS USED IN SIMULATION Property 4H- Breakown electric fiel Ec (kv/cm) Relative ielectric constant. Doping coefficient of.76 Electric fiel coefficient of Coefficient of Coefficient of Coefficient of A 95 B.4 Moreover as the forwar current increases the contribution of the first term in Eq. 6 becomes smaller an can be neglecte. This means that the on-state resistance is nearly constant regarless of forwar current an only changes with temperature at a relatively high current region. Most power Schottky ioes operate in this region. Thus it is reasonable to only consier rift region resistance R D in system moeling. Corresponingly the temperature epenence of the on-state resistance is etere by the change of n with temperature. ig. compares simulation results to experimental results of several Schottky ioes from Cree at ifferent junction temperatures. A 4 Coefficient of B.5 Coefficient of N ref A N ref 7 Maximum saturate velocity v (cm/s) 4.77 7 s Schottky barrier height B (ev).5 Richarson s constant A (Acm - K - ) 46 High Temperature Conuction Tests of Schottky Dioes. Several 6 V 75 A Schottky ioes manufacture by Cree were package by the University of Arkansas for high temperature operation. These package ioes were place with no heat sink in an environmental chamber with an ambient of ºC. The ioes were initially teste in the forwar conucting state at various current levels as shown in ig. 3. ig. 3(b) shows the increase in forwar voltage rop with the temperature increase. The ioe ultimately faile short when the case temperature exceee 55ºC with a forwar current of 5 A. Schottky Dioe Dynamic State. Reverse recovery is the most important ynamic behavior of power Schottky ioes. nce there is no ority carrier injection in power Schottky ioes the epletion layer capacitance eteres their behavior uring reverse recovery. When a reverse biase voltage V R is applie to a power Schottky ioe the with of its epletion layer can be calculate from well-known evice theory: w ( V ) qn R B. (7) 65 6 6 55 5A 5 Temperature (C) 5 45 4 35 3 5 5A A 5A Case temperature A Ambient temperature Voltage (V) 4 3 5A A 5A On-state voltage rop V A 5A 5 4 6 8 4 6 Time () 4 6 8 4 6 (a) (b) ig. 3. (a) Experimental case temperature of a 6 V 75 A Schottky ioe as forwar current is increase. (b) orwar voltage rop of ioe as a function of temperature.
968 licon Carbie an Relate Materials 6 orwor current (A) 7 6 5 5V Qrr=4nC 4 3 7V Qrr=97nC - 35V Qrr=35nC - -3 -.E-6-5.E-7.E+ 5.E-7.E-6 (a) Dioe Switching Loss W.5.5.5 7 6 7 5 5C.5.5 3 3.5 4 4.5 Peak orwar Current A (b) 5C 7C 6C 7C ig. 4. (a) Reverse recovery waveforms of the Schottky ioe. (b) Experimental an peak reverse recovery current values at ifferent temperatures. Then the specific epletion layer capacitance can be represente as C qn. (8) w ( V ) R B As shown by Eq. 8 C is a strong function of V R but is not affecte by the current flowing through it. That is to say the switching loss of a Schottky ioe mainly epens on the reverse voltage. Thus in system moeling it is reasonable to moel the reverse recovery charge of Schottky ioes as a function of their reverse voltage. Specifically the reverse-recovery charge increases approximately linear with V R.5 an the energy loss uring this perio increases linearly with V R.5. Switching Tests of a Schottky Dioe. A Schottky ioe (Cree CSD) was teste at ifferent reverse voltages with the same forwar current. The turn-off characteristics are shown in ig. 4(a). As expecte the reverse-recovery charge increase as the reverse voltage increase an the ratio of reverse-recovery charge at 35 V to that at 5 V is.45 which coincies with the square root of 35V/5V (.449). In aition if the slight changes of an B with temperature are not consiere the reverse recovery behavior of Schottky ioes will be the same at any temperature. This is also consistent with the test results shown in ig. 4(b) where the peak reverse recovery current for a ioe was essentially constant for a temperature range of 7ºC to 5ºC. Thus in system moeling the influence of temperature on the reverse recovery characteristics of Schottky ioes can be neglecte. Power VJET Because a VJET has an on-state resistance similar to that of a Schottky ioe Eq. 3 can also be use to moel its on-state resistance while it is conucting. The switching loss moel for a VJET is iscusse in the following section when analyzing the loss in a power converter. ig. 5(a) shows the I-V curves for a V A JET from e for temperatures from -5ºC to 75ºC. As shown in ig. 5(b) the on-state resistance increases with increasing temperature from.5 ohms at -5ºC to.6 ohms at 75ºC. ig. 5(c) shows the switching losses for the JET as a function of temperature an forwar current. This figure shows the switching losses ecrease by approximately 5% when the JETs were operate at a case temperature of ºC compare to 5ºC.
Materials Science orum Vols. 556-557 969 (a) (b) (c) ig. 5. (a) Experimental I-V curve for a V A JET for temperatures from -5ºC to 75ºC. (b) On-state resistance of JET versus temperature. (c) Switching loss for JET for temperature range from 5ºC to ºC. ull-brige Converter Power Loss Moel A common converter foun in traction applications an utility applications is a three-phase full-brige inverter [7 ]. Converter loss moels have three parts namely single evice loss moels converter power loss moels an thermal moels as shown in ig. 6. ngle evice moels are the basis of the converter moels which escribe evice characteristics in both conuction an switching perios uner ifferent temperatures. The converter power loss moels use an averaging technique to estimate the system power loss uner a specific control strategy. Their inputs are evice characteristics given by single evice moels an the system operation variables (inverter c input voltage ac output current moulation inex an power factor) calculate by the moel. Their outputs are the power losses of switches an ioes. These power losses are fe into the thermal moels to get real-time junction temperatures of evices. At the same time the temperatures are fe back to the single evice moels in orer to upate evice characteristics. A thermal equivalent circuit moel of the converter that inclues the evice packaging an heat sink has been evelope to etere what temperature the evices in the converter will be for given ambient conitions an converter loss profile. ig. 7 shows the simulation results comparing a -base inverter with a -base inverter for a battery charging an ischarging application. ull etails of the moel can be foun in [9]. The power loss savings of the Schottky ioe ue to its inherent lower on-state resistance an better reverse recovery characteristics reuce the temperature rise of the system ramatically when using the same heatsink an ambient temperature as the system. This resulte in lower losses in the -base system an an efficiency greater than 99% for the -base system compare to just more than 96% for the -base system as shown in ig. 7(). ngle evice moel: On-state resistance switching characteristics Device tests Parameters Battery moel: SOC-V oc Internal Resistances Inverter system power loss moel: Averaging technique Control Strategy Temperature Loop Thermal moel: Equivalent circuit ig. 6. Converter loss moel for battery charging application.
97 licon Carbie an Relate Materials 6 Summary This paper presente some moels that can ai in etering the expecte system benefits of using evices in various applications. As power electronics evices become more reaily available more research into the packaging an application of these evices will be neee to fully exploit their inherent high-temperature capabilities. Issues that will have to be aresse inclue high temperature gate rives passive components an packaging which inclues the substrate encapsulation ie attach an connection. ss (W) Power Lo ss (W) 8 6 4 8 6 4 4 6 8 4 35 3 Power Loss of Switches ig. 7. (a) JET an IGBT evice power losses uring ischarge Acknowlegment (single evice) (b) ioe power losses (c) converter system power losses () converter efficiency. The authors woul like to thank Anant Agarwal an Jim Richmon of Cree for proviing the Schottky ioes. We also woul like to thank re Barlow of the University of Iaho for the packaging of these evices while he was at the University of Arkansas. References Total Lo 5 5 5 (a) Total Power Loss of Converter IGBTs JETs 4 6 8 (c) Power Loss (W) Efficiency (%) 5 45 4 35 3 5 5 5 Power Loss of Dioes 4 6 8 99.5 99 98.5 98 97.5 97 96.5 96 (b) Converter Efficiency 95.5 4 6 8 [] D. C. Hopkins D. W. Kellerman R. A. Wunerlich C. Basaran J. Gomez: IEEE Applie Power Electronics Conference (6) p. 87. [] D. Katsis B. Geil T. Griffin G. Koebke S. Kaplan G. Ovrebo S. Bayne: IEEE Inustry Application Society Annual Meeting (5) p. 399. [3] T. unaki J. C. Bala J. Junghans A. S. Kashyap. D. Barlow H. A. Mantooth T. Kimoto T. Hikihara: IEEE Power Electronics Specialists Conference (5) p. 3. [4] M. Chinthavali B. Ozpineci L. M. Tolbert: IEEE Applie Power Electronics Conference (5) p. 3. [5] H. Zhang L. M. Tolbert B. Ozpineci: IEEE Workshop on Computers in Power Electronics (6) p. 99. [6] L. M. Tolbert B. Ozpineci S. K. Islam. Z. Peng: SAE Transactions Journal of Passenger Cars - Electronic an Electrical Systems (3) p. 765. [7] H. Zhang L. M. Tolbert B. Ozpineci M. Chinthavali: IEEE Inustry Applications Society Annual Meeting (6). [8] B. Ozpineci M. Chinthavali A. Kashyap L. M. Tolbert A. Mantooth: IEEE Applie Power Electronics Conference (6) p. 448. [9] H. Zhang L. M. Tolbert B. Ozpineci M. Chinthavali: IEEE Inustry Applications Society Annual Meeting (5) p. 63. [] B. J. Baliga: Moern Power Devices 987. []. Roccaforte S. Libertino. Giannazzo C. Bongiorno. L. Via an V. Raineri: Journal of Applie Physics vol. 97 (5). [] M. Roschke. Schwierz: IEEE Trans. on Electron Devices vol. 48 () p. 44. ()