High-temperature Properties of Schottky Diodes Made of Silicon Carbide Krzysztof Górecki, Damian Bisewski, Janusz Zarębski Department of Marine Electronics, Gdynia Maritime University, Poland, Ryszard Kisiel, Marcin Myśliwiec Warsaw University of Technology, Institute of Microelectronics and Optoelectronics, Poland

Outline Introduction Tested diode Measurement method Investigations results Conclusions

Introduction (1) One of high-temperature semiconductor material is silicon carbide (SiC). Devices made of this material theoretically can operate at much higher temperatures than silicon devices. However, the value of the maximum operating temperature given by manufacturers of SiC semiconductor devices does not exceed 250°C, and the typical values of this temperature belong to the range from 150 to 175oC. Obtaining higher values of the operating temperature is limited by the package construction. The package should be resistant to high temperature generated by the semiconductor structure as a result of a self-heating phenomenon The ability of removing heat generated in the electronic device can be characterized by thermal resistance Rth or transient thermal impedance Zth(t).

Introduction (2) Values of these parameters depend on the construction of the cooling system of the considered device, and a very essential component of this system is the device case. Warsaw University of Technology undertook investigations that led to the construction of a package for SiC Schottky diodes that would be resistant to high temperature. The measurements of thermal and electrical properties of the packaged SiC Schottky diodes including selected constructions of the cases at different cooling conditions were performed at Gdynia Maritime University. In this paper electrical and thermal properties of the considered diode are investigated at different cooling conditions. The measurements were performed at very wide range of change of the junction temperature of this diode.

Tested diode Chip structures of the investigated SiC diode were developed and manufactured in laboratories of the Institute of Electron Technology. Diode chips were assembled in the laboratories of Warsaw University of Technology. The tested structure is mounted on a ceramic DBC substrate (Al2O3) coated on both sides by a 200 mm copper layer. The top side of the substrate is covered by a 3 mm galvanically applied nickel layer and a 1.5 mm gold layer. The SiC structure is assembled into package using the silver micropowder sintering technique. The SiC structure top contact is connected to the package pads by four 50 mm diameter gold wires. Leads of the diodes have the form of copper strips 2 mm wide and 30 mm long. These leads are coated by a thin layer of silver. The results of measurements of the tested diode characteristics at temperatures up to 350oC do not show a loss of performance of the diodes.

Measurement method Transient thermal impedance is determined using the indirect electrical method. In this method the cooling curve of the tested diode is measured. The voltage drop on the forward biased diode at the constant current IM is used as a thermally-sensitive parameter. The measurements are performed in three steps (calibration, heating and cooling) using the measurement set-up. In the measuring set-up the source IM delivers the measuring current of the investigated diode (DUT) located inside the thermostat. The voltage source VEE with the resistor RE sets the heating current IH measured by the ammeter. The switch S1 is controlled by the PC - the position of this switch depends on the measurement step. The diodes protect the current source IM against short-circuit when the switch is closed. The values of the voltage vD of the DUT are recorded using the 16-bit A/D converter The maximum sampling rate of the converter is 500 kS/s.

Investigations results The characteristics of the investigated diode are measured. The investigations were performed at different values of dissipated power and at different cooling conditions. The considered diode is mounted using thermally-conductive glue to liquid heat-sinks, connected to forced water-cooling system or it operates without any heat-sink. This heat-sink is made of copper and brass. Using the measurement method transient thermal impedance waveforms of the considered diode are measured.

Investigations results (2) The results of measurements Z(t) for the diode: mounted on the heat-sink connected to the water-cooling system for the diode mounted on the heat-sink without water-cooling system It is visible that the time indispensable to obtain the thermally steady-state depends on cooling conditions. This time is the shortest and equal to about 10 s for the diode mounded on the heat-sink connected to the water cooling system, whereas it is the longest and equal to about 3000 s for this diode mounded on the heat-sink not connected to the water cooling system.

Investigations results (3) The dependences of thermal resistance on the dissipated power of the considered diode operating at different cooling conditions.

Investigations results (4) For the diode operating without any heat-sink thermal resistance is a decreasing function of dissipated power. The observed dependence proves that the dominant role in heat removal is played by convection and radiation, whose intensity increases with an increase of the device temperature resulting from an increase of the dissipated power. In the considered range of changes of the dissipated power, thermal resistance of the diode decreases by even 30%. In turn, thermal resistance of the diode situated on the heat-sink with water cooling is even ten times smaller than for the diode operating without any heat-sink. At these cooling conditions, an increase of the dependence Rth(p) can be observed. It proves that the dominant mechanism of heat removal is heat conduction. Thermal resistance of the diode situated on the heat-sink without water cooling has the values within the range from 16 to 18 K/W.

Investigations results (5) The current-voltage characteristics of the considered diode operating at different cooling conditions. As a results of self-heating phenomena the characteristics move right when the cooling conditions worsen. This means that series resistance of the considered diode strongly depends on temperature. An increase in the value of this resistance causes the quasi-horizontal shape of the characteristics at high values of the diode current. The maximum allowable value of this current increases even three times when the cooling conditions change from the best to the worst.

Investigations results (6) The dependence of the internal temperature of the investigated diode on the dissipated power. The maximum value of the internal temperature of the diode situated on the heat-sink is higher than 300oC, and for the diode without any heat-sink this temperature increases even to 500oC. In spite of the fact that the internal temperature has such a high value, the investigated device operates properly. This proves, that the considered diode is a high temperature device.

Conclusions In the paper, the results of investigations of the silicon carbide Schottky diode operating at high values of the device internal temperature are presented. The presented results of measurements prove that the considered structure of this diode together with the mounting base elaborated at Warsaw University of Technology can operate at the junction temperature equal to even 500oC. Such a high value of the device internal temperature is much higher than the allowable value of the maximum temperature in commercially offered semiconductor devices made of silicon carbide. Using the mounting manner of the considered semiconductor device could make it possible to produce real high temperature semiconductor devices. The open problem is enclosuring of the considered device, which makes it possible to use the mounted device in high temperature at the high reliability level.