IGBT, Diode Evolution Impacts Design Of High Power Density Inverter Modules

Fifth generation technology improves the characteristics of fourth generation IGBT and (free-wheeling diode) FWD devices, both switching behavior and electrical losses are improved. Furthermore, an increased operation temperature of the chips up to 175 °C can be used. This article looks at recent developments in each of these areas and examines the impact of increased module power density on overall inverter design by evaluating the thermal management of different designs. This leads to suggested design rules for high power density inverters.

Related Articles

Improved performance characteristics of recent generation IGBT and FWD (free-wheeling diodes) devices support improved module operating specifications. Fig. 1 shows the reduction in static and dynamic losses of the IGBT and FWD, achieved through reduction of the device thickness by almost ten percent. Concurrently, the T_{VJ_OP}was increased by 25K up to 175 °C.

Fig. 1. Switching losses as a function of the static losses for fourth-generation IGBT (in black) and fifth generation device (in red). FWD emitter controlled diode 4 and 5 at TVJ= 150°C.

While reduced die thickness (Fig. 2) improves electrical performance, IGBT short circuit turn-off capability becomes a greater concern as the thermal capability of the die is reduced. Using a copper top layer instead of aluminum on the IGBT compensates for the lower thermal capacity of the die itself. The increased thermal capacity of copper front-side metallization prevents thermal runaway within the established short circuit time of 10µs and enables successful short-circuit turn off.

Fig. 2. Schematic comparison between fourth (left) and fifth generation IGBT (right). Overall thickness is reduced, leading to reduced losses. Change of the metallization from aluminum to copper maintains good short circuit behavior.

For the FWD, if no countermeasure is implemented, the increase of the allowed operating temperature by 25K reduces the surge current capability by five percent. Here also the thick copper layer on the chip helps to handle the higher temperature. Thus the I²t values for the fifth generation emitter controlled diode at 175 °C is on the same high level as a fourth generation diode at 150 °C.

Electrical Performance – Softness

Also linked to die thickness and reduction of the turn-off losses is a higher oscillation tendency under unfavorable conditions. The IGBT oscillation tendency increases at turn-off of higher current, typically at two times the nominal current. Regarding the FWD, this tendency is higher during turn-off of lower current, in the region of 1/10 of the nominal current. The so-called softness can be improved with a higher injection of charge carrier in the die. However, that leads to higher turn-off losses at the IGBT and higher recovery losses at the FWD. The fifth generation IGBT was developed to achieve a very good softness using a minimum of injected charge carriers. As a result, the medium power version of the device can also be used in high power applications. The latest generation FWD also has softness behavior similar to fourth generation device. This ensures the advantages of reduced turn-off losses with clean switching in applications with high output power.

Power Cycling And Thermal Performance

To fully utilize the capability of the fifth generation die, the packaging utilizes .XT assembly technology^{[1, 2]}, an integrated set of interconnection technologies which address limitations for wire bonding as well as chip and substrate soldering. Compared with earlier generation die and assembly with standard interconnects (Fig. 3), fifth generation devices assembled with .XT technology achieve a 10 times higher power cycling capability. Alternatively, a 25 percent higher output power can be driven at same application lifetime as earlier devices. Additionally, the thermal resistance from chip to module base plate R_{TH_JC}is improved by 15%.

Fig. 3. Power cycling target curve of the .XT technology (top);Comparison of the thermal performance of .XT assembly technology with the IGBT4 solder technique (bottom).

First IGBT 5 module – Increased current density

The overall performance improvement for these fifth generation devices leads directly to improved module operating specifications. This is the case for the Infineon PrimePACK™ 2 1200 V module (Fig. 4), which is upgraded from a maximum current of 900 A to 1200 A with no change in operating lifetime.

Fig. 4. Maximum module current with IGBT 4 and 5 in the PrimePACK™ (above right); Comparison of the electrical parameters at different temperatures and currents (table)

Consideration On Different Inverter Designs

These improvements to the module’s operating specifications have an impact on inverter design. The primary area of concern is the inverter thermal management^{[3, 4]}. Increase of the operating temperature of the switching silicon leads to a higher heat sink temperature at given cooling conditions. Considering this, it is very likely that compact air cooled systems will suffer from elevated temperature of components near the power module.

To better understand the influence of different thermal designs, two commercially available 4 kW inverters from drives applications are examined. The focus is the behavior of temperature sensitive components of the control electronics and DC-Link at the higher IGBT operation temperature. Fig. 5 shows the different inverter concepts schematically. While one concept has a dedicated airflow implemented to cool the electronic components, the other inverter has its PCBs cooled by convection. At the same time, the latter is thermally coupled to the heat sink through the power electronic terminals.

Fig. 5. Different inverter design and different power density for drives application.

Design 1 has the heat sink completely inside the housing, while in design 2 the heat sink is a part of the housing and larger in volume by circa 70 percent.

The benefits of the 25K higher maximum operation temperature are examined in two tests: one with an increase of the output current by 20 percent and the other with a decrease of the cooling efforts by 25 percent. The temperature of various electronic components is monitored with thermocouples and the results are shown in Fig. 6. Both inverters are tested under load and overload conditions up to T_vjof 175 °C.

Fig. 6. Temperature behavior of the electronic components by increasing the IGBT operation temperature up to 175 °C in two different inverter designs. Dotted lines represent the maximum allowed temperature for the respective components.

The measured temperatures inside of Inverter 1 are lower in all operation points in comparison to Inverter 2. Due to the fact that Inverter 1 has the higher power density, this is a good indication of better thermal design compared to Inverter 2. The temperature spread over the different measurement points also is larger for the first design in comparison to the second design. Especially at T_{vj_op}= 175°C the captured points of all measured components are distributed across just 12K within Inverter 2. At the same operation point, the temperature spread in Inverter 1 is 49K. The Inverter 1 design, with a small heatsink close to the air output side, leads to relatively “cold” areas inside the housing and the IGBT junction temperature does not influence all areas of the inverter evenly. With Inverter 2, as the large heatsink warms up, convection and conduction along the module terminals heats and stresses the various electronic components uniformly. Using the first design, the development engineer has the possibility to place the thermal sensitive components in the “cold” areas of the inverter, while this is not possible with Design 2.

It is not surprising that increasing the output power leads to an increase of the inverter temperature. This is not only due to the higher IGBT operation temperature, but also due to the higher current through DC-Link capacitors and PCBs. The results show that, compared with Design 2, the PCB and the IGBT driver in Design 1 experience a higher temperature elevation when the current increase causes a rise in T_{VJ_OP}from 150 °C to 175 °C. This can be explained by the stronger coupling of these components with the heatsink in the more compact design.

Nevertheless, the margin to the maximum allowed component temperature in Design 1 is higher than in Design 2.

Regarding the DC-Link capacitors, the temperature rise is similar in both designs. Design 2 requires 105°C capacitors against 85 °C in Design 1, a clear disadvantage for Design 2 due to the passive cooling of this part of the inverter.

Test 2, reduction of the cooling efforts, is equivalent to an increase of the heatsink thermal resistance R_{th_ha}. In this test, Design 2 is seen to have serious problems. Already at a temperature of T_vj= 115°C most of the components are above the maximum permissible temperature. The increased thermal resistance leads to an increase of the heatsink temperature, the heat remains in the inverter housing and is dissipated towards the control part of the inverter through convection and conduction via the module terminals. Design 1 manages this situation better. All measured parts operated inside the allowed temperature up to a Tvj_op of 175 °C of the IGBT.

The distance to the maximum allowed operation temperature of the electronic components during the operation of the IGBT at 175 °C and the increase compared to operation at T_vj= 150 °C is shown in Fig. 7.

Fig. 7. Distance to the maximum allowed operation temperature of the electronic components during the operation of the IGBT at different junction temperatures.

A core element when increasing the power density in inverters is the lifetime consideration of the components. Using the same component technology at higher temperature automatically leads to a decrease of the lifetime. The inverter design with “cold” areas and forced convection on electronic level helps to handle this challenge. It is shown that despite the dimensioning of DC-link capacitors, which has to be adjusted to higher current anyway, an inverter design implementing the designated PCB cooling may draw benefits from increased junction temperatures straight away. In contrast, considerable redesign efforts have to be considered if the electronics part is thermally coupled to the main heat sink through power terminals of the module.

Compared to the fourth generation IGBT and FWD devices, both switching behavior and electrical losses are improved with the fifth generation technology. Furthermore, an increased operation temperature of the chips up to 175 °C can be used. These, together with .XT interconnection technology, lead to a higher power density and longer lifetime. With improved module specifications, inverter thermal design must be evaluated. Care must be taken regarding the temperature increase seen at all system components to guarantee that the frequency inverter`s lifetime remains constant. A smart inverter design is the best way to handle an IGBT with T_VJequal to 175°C without the need of expensive components designed to handle the higher temperature.

References

[1] A. Ciliox et al: New Module generation for higher lifetime, PCIM , Nuremberg, Germany, 2010

[2] K. Guth et al: New assembly and interconnects beyond sintering methods, PCIM, Nuremberg, Germany, 2010

[3] K. Vogel et al: IGBT with higher operation temperature - Power density, lifetime and impact on inverter design, PCIM 2011, Nuremberg, Germany

[4] K. Vogel et al: IGBT inverter with increased power density by use of a high-temperature-capable and low-inductance design, PCIM 2012, Nuremberg, Germany