IGBT advancements enhance solar, industrial applications

IGBTs are bipolar power switches that are designed in terms of both their physical structures and material doping levels to meet the needs of many different applications. There are a variety of physical structures employed by different manufacturers that achieve specific performance characteristics. After these common structure elements are implemented, such as a trench-style gate, the key performance trade-offs within a technology family occur due to process steps and differing doping concentrations that impact the switching and conduction losses. Performance improvements from one product generation to another are a function of changes to both the physical structure and advanced manufacturing techniques.

The TRENCHSTOP5 IGBT technology platform illustrates this performance evolution. It advances the existing TRENCHSTOP families by implementing a very fine pitch stripe-cell style gate structure along with advanced thin-wafer technology. The stripe-cell style gate increases the MOS channel width and reduces gate charge Qg, resulting in lower Vcesat and Eon.

The reverse blocking capability of an IGBT depends on the thickness and doping concentration of the n-doped layer. One fabrication method is to build an epitaxial layer on a Si substrate wafer. This Si wafer is then ground down to the desired thickness to address both the required blocking capability as well as the lowest conduction and switching losses. TRENCHSTOP technology utilizes a different approach. It omits the epi-layer and directly dopes the Si substrate wafer. This reduces the thickness of the n-drift region of the technology as well as the number of steps needed to manufacture the devices. Since IGBT devices are bipolar they have an excess carrier and hole doping concentrations to achieve a low on-state voltage, both on-state and switching losses are dependent on the total amount of carriers in the on state condition. Consequently, thinner Si for IGBTs results in lower conduction losses and switching losses. (Aigner) The TRENCHSTOP 5 technology has a thickness of Fig. 1 .

Fig. 1: IGBT device structures for TRENCHSTOP and TRENCHSTOP5 technology platforms. Note the vertical stripe cell gate structure and the thinner n- basis (substrate) or so called drift region.

This combination of features reduces both switching losses and conduction losses compared to previous generations. The technology platform is applied today in the HighSpeed5 IGBT product family, which includes two versions to address different requirements for SMPS design.

H5: Allows plug and play replacement of existing MOSFETs and previous generation (HighSpeed3) devices without any design change. The optimized Field Stop design allows very soft voltage rise during hard commutation turn-off at low R_g(off) with very high dI/Dt.

F5: Is a fast version providing the highest efficiency. With these devices, it is recommended to use a split R_g configuration to control voltage overshoots at turn-off of high load current. They are a best fit for optimized PCB design with low stray inductance in the commutation loop.

Performance characteristics of H5 and F5 vs H3
Using a 40-A device for comparison of the HighSpeed 5 version H5 to the HighSpeed 3 devices, conduction losses are reduced from 1.95 to 1.75 V. Concurrently, switching losses have improved from an Eon of 0.61 to 0.3 mJ and the Eoff is reduced from 0.29 to 0.16 mJ. In both phases of the switching cycle, losses are reduced by half of their previous values. This places the IGBTs in the performance range of a super junction MOSFET device, well beyond the capabilities of other available IGBTs. The static and dynamic characteristics of the H5 and F5 variants of the HighSpeed 5 families are listed in Fig. 2 .

Fig. 2: Static and Dynamic Characteristics of new versions (H5/F5) compared to the existing High Speed 3 technology.
IGBTs as plug-and-play for MOSFETs?
With this dramatic shift toward an idealized power switch, is the TRENCHSTOP5 technology ready to replace conventional MOSFETs? Not in every application, since there is a trade-off between the reduction in switching and conduction losses. With increased cell density and the required optimized carrier profile, the device loses short-circuit capability and consequently is restricted from applications that require this functionality. This restriction, however, still leaves open sockets within such market segments as inductive heating, welding, UPS, and solar inverters.

Efficiency improvements in resonant topologies
The H5 is attractive for ZVS topologies, like phase shift full bridge or LLC, for three reasons: the low Q_g , the minimal Q_oss , and the available duo-pack configuration with anti-parallel fast diode. A low Q_g reduces driving losses, improves the light-load efficiency, and enables a reduction in the delay time by keeping the same ZVS window from cycle to cycle. This in effect increases the actual duty cycle and shortens the conduction time of the freewheeling diode. A low Q_oss value allows transitioning to ZVS in accordance with low load currents. It also contributes to light-load efficiency improvement and provides turn on and turn-off loss optimization over the entire power range. The anti-parallel fast diode in the duo-pack configuration provides higher ruggedness, especially in ZVS topologies during critical working conditions like startup, overload, or short circuit, and burst-mode operation, where a key parameter to improve reliability is the low t_rr of the diode.

Solar inverters improve system cost and efficiency
The new device series address solar inverters by offering designers higher power density, lower system costs, and efficiencies up to and beyond 98%, in the case of H4 or H6 inverter topologies. Thermal factors such as heat sinking, ventilation, board layout and derating objectives have a serious impact in these designs. Ultimately the power losses for the IGBTs are key factors in the equation. 

Consider the power loss improvement of the High Speed H5 compared to the previous best-in-class device. Figure 3 shows the difference in total power loss per IGBT vs. switching frequency, using a simple condition of a 20A square wave with 50% duty cycle and junction temperature 100°C.

Fig. 3: Comparison of power losses versus switching frequency for different generations of High Speed IGBTs.

At a typical 40 W, for a standard TO-247 device, the maximum operating frequency of the previous generation IGBT with max junction temperature of 100°C is around 28 kHz. In comparison, the H5 device can be driven up to 50 kHz. This increase in operating frequency allows designers to use smaller and lighter-weight magnetics and reduce electrolytic capacitors in some cases. So while there is an investment in design efforts to optimize layout for higher frequency operation, the benefits are seen in the bill of materials. This is important for applications like solar and UPS, where the passive components are a high percentage of the total costs.

In a final comparison, the new IGBT version was compared to the previous device in a 3-kW solar inverter with input of 350 Vdc, output of 230 Vac, and operating frequency of 20 kHz. Figure 4 shows an efficiency improvement of 0.4% for the inverter. Considering an expected a life span of more than 20 years for some solar installations, this relatively small improvement has major impact on operational costs and consequently, the ROI of a solar installation.

Fig. 4: Efficiency curves for a 3-kW solar inverter with 350-Vdc input and 230-Vac output, operating frequency of 20 kHz.  High Speed H5-enabled an efficiency improvement of 0.4% for this inverter

References :
Aigner, K. “Thin Chips for Power Applications.” Burghartz, Joachim. Ultra-thin Chip Technology and Applications. New York: Springer, 2011. 328-330.
Chiola, Davide. “High Speed 5 IGBT achieves Platinum Efficiency Standard in Commercial SMPS applications.” PCIM. China, 2012.
Griebl-Kimmer. “TRENCHSTOP 5: A New Application Specific IGBT Series.” PCIM Europe. Nuremberg, Germany: PCIM, 2012.