Testing, Characterization, and Modeling of SiC Diodes for Transportation Applications

Transcription

1 Testing, haracterization, an Moeling of ioes for Transportation Applications Burak Ozpineci 1,3 Leon M. Tolbert 1,2 Sye K. slam 1 sislam@utk.eu Fang Z. Peng 2,4 fzpeng@msu.eu 1 ept. of Electrical an omputer Engineering, The University of Tennessee, Knoxville, TN Oak Rige National Laboratory, P.O. Box 29, Oak Rige, TN Oak Rige nstitute for Science an Eucation, Oak Rige, TN ept. of Electrical an omputer Engineering, Michigan State University, East Lansing, M Abstract: The emergence of silicon carbie- (-) base power semiconuctor switches, with their superior features compare with silicon- (-) base switches, has resulte in substantial improvement in the performance of power electronics converter systems. These systems with power evices have the qualities of being more compact, lighter, an more efficient; thus, they are ieal for high-voltage power electronics applications such as a hybri electric vehicle (HEV) traction rive. More research is require to show the impact of evices in power conversion systems. n this stuy, finings of research at Oak Rige National Laboratory (ORNL), incluing evice esign an system moeling stuies, will be iscusse.. NTROUTON Presently, almost all the power electronics converter systems use silicon- (-) base power semiconuctor switches. The performance of these switches is approaching the theoretical limits of the material. Another material, silicon carbie (), with superior properties compare with, is a goo caniate to be use in the next generation of power evices. power evices, with their close-to-ieal characteristics, bring great performance improvement to power converter applications. Some of these avantages compare with -base power evices are as follows: unipolar evices are thinner, an they have lower on-resistances. At low breakown voltages (~ V), these evices have specific on-resistances of 1.12µΩ, aroun 1 times less than those of their counterparts. At higher breakown voltages (~ V), the on-resistance goes up to 29. mω, but it still is 3 times less than that of the comparable evices [1]. With lower R on, power evices have lower conuction losses; therefore, the converters have higher overall efficiency. -base power evices have higher breakown voltages because of their higher electric breakown fiel; for example, Schottky ioes are Prepare by the Oak Rige National Laboratory, Oak Rige, Tennessee 37831, manage by UT-Battelle for the U.S. epartment of Energy uner contract E-A-OR2272. The submitte manuscript has been authore by a contractor of the U.S. Government uner ontract No. E-A-OR2272. Accoringly, the U.S. Government retains a non-exclusive, royalty-free license to publish from the contribution, or allow others to o so, for U.S. Government purposes. commercially available typically at voltages lower than 3 V, but the first commercial Schottky ioes are alreay rate at 6 V. has a higher thermal conuctivity (4.9 W/cm K for an 1. W/cm K for ), an power evices have a lower junction-to-case thermal resistance, R th-jc (.2 K/W for an.6 K/W for ); thus evice temperature increase is slower. evices can operate at high temperatures. evice operation at up to 6 is mentione in the literature [2]. evices, on the other han, can operate at a maximum junction temperature of only 1. is extremely raiation har; that is, raiation oes not egrae the electronic properties of. Forwar an reverse characteristics of power evices vary only slightly with temperature an time; therefore, they are more reliable. -base bipolar evices have excellent reverse recovery characteristics [3]. With less reverse recovery current, the switching losses an EM are reuce, an there is less or no nee for snubbers. Because of low switching losses, -base evices can operate at higher frequencies (>2 khz) not possible with -base evices in power levels of more than a few tens of kilowatts. Although has these avantages compare with, the present isavantages limit its wiesprea use. Some of these isavantages are Low processing yiel because of micropipes. The best wafers available have <1/cm 2, but they are more expensive than the typical wafer with <1/cm 2. High cost $7 for a 6 V, 4 A Schottky ioe (a similar pn ioe is <<$1). Limite availability (only Schottky ioes at relatively low power are commercially available). Nee for high-temperature packaging techniques that have not yet been evelope. At Oak Rige National Laboratory (ORNL), a project team consisting of materials, evice, an systems researchers in cooperation with The University of Tennessee-Knoxville, Auburn University, an Vanerbilt University are builing power MOSFETs. The reason for having a iverse team of people from ifferent isciplines is to evelop better evices by encouraging interaction between the researchers who buil the evices

2 an the ones who use them. The final power MOSFETs will be use in power electronics converter systems for automotive applications to emonstrate the benefits of base power evices. One of the selecte systems for this project is a traction rive.. EVES A. Schottky an pn junction ioes: The Schottky ioe is the least-complex power evice. Schottky ioes have been fabricate by forming a metal contact with ope samples. Four ifferent samples were prepare; their characteristics are given in Table. All p-type contacts are Al-Ni, an n-type contacts are Au-As-Ni alloy. The -V plots for samples A an in Figs. 1 an 2 o not show any extra resistance ue to the new contact material. This means that Au-As-Ni alloy makes a goo contact for n-. Note that, base on an extensive literature search, Au-As-Ni contacts are a novel approach for evice applications. oping ensities have been calculate from the slope of the -V characteristic plots. The -V plots for samples A an B are shown in Figs. 3 an 4, respectively. The measure values of the activate oping ensities are TABLE HARATERSTS OF THE SAMPLES Sample A B Schottky ioe (p-type epi) pn ioe (p-type epi on n-type substrate) mplante with boron an silicon mplante with boron only sample A, cm 3 ; sample B, cm 3 ; sample, cm 3 ; an sample, cm 3. These values are similar to the applie oping, another sign of a goo contact. B. MOSFETs: The ouble implante metal-oxie semiconuctor (MOS) fiel effect transistor has been frequently use in high-voltage power electronics applications [4, ]. The performance of a MOS evice is limite by the quasisaturation behavior in its characteristics. t is shown that such an effect is ue to carrier velocity saturation because of the high electric fiel, low impurity concentration in the rift layer, an narrow p-boy spacing [6]. A etaile stuy has been mae of the vertical MOS, an an analytical moel has been evelope. Fig. shows the MOS evice, ientifying ifferent regions of operation. An analytical moel is evelope from regional analysis of carrier transport in the channel an rift regions. The current/voltage characteristic in the trioe region is given by Wµ [ ch = Vch 2ox ( VGS VT ) 2L[1 + ( µ 2v L) V ] sat ch (1) ( + ) V where W is the with of the transistor, µ is the effective electron mobility, v sat is the saturation rift velocity, V ch is the channel voltage, V GS is the gate-source voltage, V T is the threshol voltage, ox is the oxie capacitance, o is the boy epletion capacitance, an L is shown in Fig.. The rift region is ivie into three parts: an accumulation region A, a rift region B with a varying cross section area, an a rift region with constant cross section. Voltage across each of these regions is given by ox o ch ] Fig. 1. -V haracteristic of sample A. Fig. 3. -V plot of sample A. Fig. 2. -V haracteristic of sample. Fig. 4. -V plot of sample B.

3 S L G L P S P n + n + Region A L iff Region B W α P W j W t V c /2 V c /2 o Q 1 Q 4 a 1 4 Q 3 i a Q 6 b 3 6 Q i b Q 2 c i c 2 A MOTOR Region V V A B = W j + W E = yy W ( L ( W + W ) j qn µ ) iff = WqN µ n + Fig.. Vertical MOS (Regions A, B, an form the rift region). WqN µ log cotα WqN E ( L + L ) iff P L µ E iff (2) E (3) ( W t W j W LP tanα) V = (4) W ( L + iff LP ) qn µ E where is the rain current, E c is the critical electric fiel, q is the electron charge, an α is the angle shown in Fig.. The total rift region voltage V rift = V A + V B + V, an the voltage across the rain to source V S = V rift + V ch. The current/voltage characteristics are shown for the analytical moel in Fig. 6.. SYSTEM MOELNG The main objective of system moeling is to show some of the system-level benefits of using evices in power converter applications, such as the large reuction in the size, weight, an cost of the power conitioning an/or thermal management systems. The selecte system for this stuy is a hybri electric vehicle (HEV) traction rive. A typical electric traction rive schematic for an HEV is shown in Fig. 7. The schematic consists of a battery, a three-phase inverter (c/ac converter), an an inuction motor. To evaluate any transportation system, its operation while the vehicle is accelerating, ecelerating, stoppe, etc., shoul be consiere. For this reason, it is necessary to simulate the HEV an observe the response of the traction rain current, V GS =2V V GS =18V V GS =1V rain voltage, V S Fig. 6. urrent-voltage characteristics for MOS structure. Fig. 7. Electric traction rive. rive over a riving cycle. Using AVSOR (Avance Vehcle mulator), an HEV moel was simulate in AVSOR over the Feeral Urban riving Scheule (FUS), which is a stanar -secon velocity profile of an average person s vehicle on the way to work [7]. The inverter, in this stuy, uses sinusoial PWM with a switching frequency of 2 khz. The inuction motor is rate at 31 kw, 23 V, 4-pole, 3 rpm. The peak current passing through the switches is 136 A, an the peak voltage across them is the battery voltage V c, which is 3 V in this case. Therefore, the MOSFETs an ioes have to be rate at least 4 V an 2 A. The MOSFETs are not available in this power range, an MOSFETs are not yet available. For comparison purposes in this stuy, we will assume that MOSFETs exist at high power, too. Note that the rive is controlle by constant V/Hz control. The following section will explain the evice loss moeling approach [8]. V. OE MOELNG A. onuction losses: The circuit in Fig. 8 is built to fin the -V characteristics of the ioes. The c voltage supply is varie, an the ioe forwar voltage an current are measure. This test is carrie out at several temperatures of up to 2. The results for both pn an Schottky ioes are given in Fig. 9, in which it can be seen that the forwar voltage of the ioe is higher than that of the ioe. This is expecte because of s wier bangap. Another ifference between these two ioes is their hightemperature behavior. As the temperature increases, the forwar characteristics of the ioe change severely, while those of the ioe stay confine to a narrow region. Note that the pn ioe (negative) an the Schottky ioe (positive) have ifferent polarity temperature coefficients; that is why the slope of the curve at higher currents is increasing in the ioe case an ecreasing in the ioe case with the temperature increase. For the traction rive application, the forwar characteristics of 2 A rate ioes are require. Assume V c UT R oven UT F urrent Probe + V F Fig. 8. -V characterization circuit. -

4 ioe Forwar urrent, A Arrows point at the increasing temperature ioe Forwar Voltage, V Fig. 9. Experimental -V characteristics of the an ioes in an operating temperature range of 27 to 2. that the 2 A ioe is equivalent to twenty 1 A ioes connecte in parallel. Then, in the -V characteristics of the resulting ioe, s an R s shoul be change accoringly. The ioe equation is q( V R ) nkt = s 1 s e. () s is irectly an R s is inversely proportional to the area of the evice. For the 2 A ioe, the area is increase 2 times; therefore, s increases 2 times an R s ecreases 2 times. The resulting ioe equation is q( V ) R s nkt = 2 2 s e 1. (6) f a line is rawn along the linear high-current portion of the -V curves of (6) extening to the x-axis, the intercept on the x-axis is V an the slope of this line is 1/R. After these parameters are obtaine for each ioe at ifferent operation temperatures, as shown in Fig. 1, the conuction losses can be calculate using (7) [8]: P = R M cosφ + V M cosφ, con, 8 3π 2π 8 (7) where M is the moulation inex an φ is the power factor. The conuction losses for a ioe an for a ioe are plotte in Figs. 11 an 12, respectively. These plots show that for low temperatures (T < ), the conuction loss of the ioe is less than that of the ioe, an vice versa for higher temperatures (T > ). This is because the series resistance of the Schottky ioe is increasing while that of the pn ioe is ecreasing. This increase seems to be a isavantage in the Schottky ioe case; however, note that the ioe cannot withstan temperatures of over 1. B. Switching losses: The most important part of the ioe switching loss is the reverse recovery loss. The rest of the losses are negligible. Reverse recovery losses, in this paper, will be calculate experimentally. For this purpose, the chopper circuit in Fig. 13 was built. The main switch Q is turne on an off at 1 khz with a uty ratio of 7%. When Q is on, the ioe is off, an the current is force by the c supply to buil up through the loa an Q. When Q turns off, the loa current starts flowing through an the loa. n this moe, the c supply is out of the loop; therefore, the current starts ecreasing. After a while, Q is turne on again an then turns off. The typical ioe turn-off waveforms are given in Fig. 14. These experimental waveforms show that the ioe switching losses are almost three times more than those of the ioe. The peak reverse recovery current, R, an the switching loss, P sw, of the ioes are measure at ifferent operating temperatures with varying loa currents. The results are plotte in Figs. 1 an 16. n Fig. 1, the R of the ioe is higher than that of the ioe at any operating temperature. As the temperature increases, the ifference increases because the R of the ioe increases with temperature, but that of the ioe stays constant. Note that the ioe faile when operating at 1 an 4. A, V, V R, Ω Toven, (a) Toven, (b) Fig. 1. Variation of (a) V an (b) R with temperature. ioe onuction Loss, W ncreasing Temperature 2, 7, 12, 17, 22 M= ioe Forwar urrent, A Fig. 11. onuction losses for a 2A ioe. ioe onuction Loss, W ncreasing Temperature 2, 7, 12, 17, 22 M= ioe Forwar urrent, A Fig. 12. onuction losses for a 2A ioe.

5 i L 6 V c oven i =UT while the ioe survive that temperature an faile at a higher 2 an 4 A. To obtain a 2 A ioe moel for the traction rive, assume as before that 2 of the ioes teste here are connecte in parallel to form a 2 A ioe. The switching losses multiplie by 2 gives the switching losses of the 2 A an ioes (Fig. 16). V. RESULTS urrent Probe + Voltage v solator - The HEV traction rive is simulate over the FUS cycle using the moel evelope in the previous sections. The loss profiles of a ioe an a MOSFET in the rive are shown in Fig. 17. ioe losses are lower than ioe losses mostly because the ioe has lower reverse recovery losses. On the other han, MOSFET losses are lower because the switching losses are similar but MOSFET conuction losses are lower. The reason for lower conuction losses is the lower specific on-resistance [1] (R on,sp () = Ω-cm 2, R on,sp (4H-) = Ω- cm 2 ). Total energy loss (six ioes an six MOSFETs) is 92W s for the inverter an 338W s for the inverter over the FUS cycle. The corresponing efficiency (Fig.18) of the inverter is 8 8%, while that of the inverter is 9 9%. This is a 1 percentage points increase in the average efficiency. As a result, the battery in the HEV with the inverter will nee less charging than the one with the inverter. The loss profiles in Fig. 17 are fe to the thermal moels of the evices. The resulting junction temperature profiles Q i UT Fig. 13. Reverse recovery loss measurement circuit. L 1 R 1 Peak Reverse Recovery urrent, A , 61, 17, 11, 2, Peak Forwar urrent, A are shown in Fig. 19. Natural air-coole heatsinks are use to limit the junction temperature to 1 for an 17 for. The latter temperature limit is foun on the atasheet of the nfineon Schottky ioe use in this stuy [9]. The resulting heatsink volumes an masses for each evice an each inverter are given in Table. Using evices instea of their counterparts in an HEV traction rive reuces the size an weight of the heatsink to one-thir. Note that a heatsink usually occupies one-thir TABLE HEATSNK MASS AN VOLUME FOR EAH EVE AN NVERTER Volume (cm 3 ) Mass (g) ioes ioes MOSFETs MOSFETs inverter inverter Fig. 1. Peak reverse recovery values with respect to the forwar current at ifferent operating temperatures. ioe Switching Loss, W , 61, 17, 11, 2, Peak Forwar urrent, A Fig. 16. ioe switching loss of 2A ioe at ifferent operating temperatures. Fig. 14. Typical reverse recovery waveforms of the pn an Schottky ioe (2 A/iv.). ioe losses, W MOSFET losses, W Time, s Fig. 17. Total loss profile for a ioe (top) an a MOSFET (bottom).

6 Total inverter losses, W Efficiency () Efficiency () Time, s the volume of the converter an weighs more than the electronics. Theoretically, evices can work at higher temperatures. f new packaging techniques are evelope so that these higher temperatures coul be use as the junction temperature limits, then the amount of cooling require woul be less, an more weight an volume savings woul be possible. V. ONLUSONS Even the first evices show the superiority of compare with. One of the challenges for builing power evices is the contact material. n this paper, a new contact material, Au-As-Ni alloy, is introuce. The -V characteristics of the Schottky ioes built with this contact material show that it makes soli an reliable contact with. System stuies show that power electronics systems using power evices are on average 1 percentage points more efficient because of the low losses of the power evices. Moreover, with their high-temperature operation capability, they have less stringent cooling requirements. n the HEV traction rive stuie here, using power evices saves 1392 cm 3 of space an 3.7 kg of weight. The weight reuction an efficiency increase result in an increase in the fuel economy of the vehicle an a longer lifetime for the battery. Fig. 18. Total losses an the efficiency of the inverter over the FUS cycle. ioe Junction Temperature, MOSFET Junction Temperature, Time, s Fig. 19. Junction temperature profiles of the ioes an MOSFETs in the three-phase inverter. Note that technology is still in its infancy. More stuies like the one etaile in this paper are require to forecast the impact of power evices. When this technology matures, for power evices in the meium-tohigh power range, the future will be. REFERENES [1] M. Bhatnagar an B. J. Baliga, omparison of 6H-, 3-, an for power evices, EEE Trans. on Electron evices, 4 (3), pp. 64 6, March [2] K. Shenai, R. S. Scott, an B. J. Baliga, Optimum semiconuctors for high power electronics, EEE Transactions on Electron evices, 43 (9), pp , Sept [3] A. Elasser, M. Kheraluwala, M. Ghezzo, R. Steigerwal, N. Krishnamurthy, J. Kretchmer, an T. P. how, A comparative evaluation of new silicon carbie ioes an state-of-the-art silicon ioes for power electronic applications, EEE AS Annual Meeting onference Proceeings, pp , [4] B. J. Baliga, Moern Power evices, Wiley, New York, []. A. Grant an J. Gower, Power MOSFETs: Theory an Application, Wiley, New York, [6] M. arwish, Stuy of the quasi-saturation effect in VMOS transistors, EEE Transactions on Electron evices, vol. E-33 (11), pp , [7] [8] B. Ozpineci, L. M. Tolbert, S. K. slam, an M. Hasanuzzaman, Effects of silicon carbie () power evices on PWM inverter losses, The Annual onference of the EEE nustrial Electronics Society (EON'1), pp , 21. [9]