IGBT reliability solved: deliberately cap the current below latch-up
Toshiba engineer Akio Nakagawa discovered that the solution to IGBT latch-up burnout wasn't raising the safe current limit—it was deliberately keeping the device's saturation current below the latch-up threshold, so it would saturate before it could fail.
The argument · tap a timestamp to hear it
Thyristors lock up until forced off, justifying search for better switches
Before IGBT, the thyristor (silicon controlled rectifier) suffered from a fundamental flaw: once the gate triggered conduction from anode to cathode, the device self-locked and couldn't turn off until the anode current dropped below zero or a specific threshold. This worked fine for AC applications where the line voltage naturally zeros each half-cycle, but direct-current circuits required bulky, power-hungry external circuits to force the current down and reset the device. This structural limitation spurred the search for new switching devices—first junction transistors, then field-effect transistors, and eventually the IGBT itself.
IGBT succeeded by combining MOSFET's speed with BJT's current capacity
Baliga had spent years watching the semiconductor industry's endless tug-of-war between power BJTs and power MOSFETs. Bipolar transistors could handle high voltages but consumed power proportional to output current and were slow to switch. MOSFETs switched fast and efficiently but couldn't sustain high voltages. His insight was radical: combine them. The IGBT's structure reflected this hybrid ambition—it kept the MOSFET's voltage-controlled gate but added BJT-style emitter and collector, so conduction began with MOSFET inversion-layer current and then triggered minority-carrier injection to dramatically reduce on-state resistance and handle large currents.
Organizational pressure, not pure science, accelerated IGBT to market
In the early 1980s, a GE vice president responsible for motor-speed drives for air conditioning and heat pumps issued an ultimatum: the research lab must develop a BJT replacement for his applications, or he would tell Jack Welch that these research facilities were unnecessary. Baliga seized the moment to revive his hybrid MOSFET-BJT concept, now framing it not as a narrow device but as a platform applicable across GE's consumer appliances, lighting, and other product lines. The vice president was convinced and escalated to Welch. Within months, the project was funded and accelerated. The first wafer run occurred in August 1981; the technology was published in December 1982; and the commercial IGT product launched in June 1983.
RCA's simultaneous invention created permanent ambiguity about IGBT's parentage
Unknown to GE, RCA was following a parallel path. Carl Wheatley and Hans Becke, seeking a new-generation power semiconductor, pursued the concept of a power MOSFET with some bipolar character to achieve higher current density. They filed a patent in 1980 and commercialized a device they called COMFET in 1983—the same year GE launched the IGT. The market was suddenly awash in competing names for what was essentially the same innovation, muddying the historical record and creating a patent thicket that would tangle invention attribution for decades.
IGBT's internal parasitic transistor is a structural latch-up trap
The IGBT's superior current-handling came from a clever hybrid structure, but it carried a hidden structural defect. Inside the device sat an unintended parasitic NPN transistor, quiescent under normal operation but capable of catastrophic activation. When current became large enough, hole flow from the P-type collector through the N-type drift region and into the P-type body could induce sufficient voltage drop to accidentally trigger this parasitic NPN, which then formed a self-sustaining current loop with the intended PNP transistor. Once locked in this state, switching off the gate had no effect—thermal runaway and device failure followed. Baliga's ‘Deep P+’ masking technique reduced but did not eliminate this hazard. System designers had to constantly monitor operating margins to avoid exceeding safe-area limits.
Current saturation, not higher limits, was the real latch-up solution
The standard engineering response to latch-up had always been to design devices with higher latch-up current thresholds—raise the bar higher. But Toshiba's Akio Nakagawa proposed the inverse strategy: instead of raising the latch-up threshold, lower the device's saturation current to stay below it. By deliberately capping the device's saturation characteristics, he ensured the IGBT would reach current saturation before it could trigger the parasitic thyristor. Under normal operation, the device would simply stop accepting additional gate drive—saturation would occur first, latch-up never. Toshiba began selling fully latch-up-proof IGBTs in 1985, and this design philosophy rapidly displaced earlier power BJT and MOSFET designs in mid and high-voltage converter markets.
IGBT's invention credit remains disputed, analogous to integrated circuits
The question of IGBT invention remains contested. Baliga was earliest to publish the conceptual design (1979), first to bring a device to market (1982 publication, 1983 launch), and most prolific in subsequent patents (over 20) and scholarship (an 800-page monograph). For these reasons, much of the semiconductor industry credits him as the primary inventor or co-inventor. However, Baliga's early devices still exhibited thyristor-like behavior under high current—they were not truly latch-up-free across the full operating range. RCA's patent application, filed around the same time, explicitly claimed ‘latch-up-proof operation across the full working range’—the feature that Nakagawa actually realized. The analogy to integrated circuits is apt: Kilby receives credit for inventing the IC concept; Noyce and Hoerni for the planar process that made ICs practical. Here, Baliga pioneered the IGBT concept; Nakagawa solved its most critical practical limitation.
Silicon IGBT still dominates mid-power applications despite SiC and GaN competition
The past decade has seen the rise of wide-bandgap semiconductors—silicon carbide and gallium nitride—promising superior switching speed and efficiency in power applications. These newer materials continue to gain market share in high-power and high-frequency domains. Yet the humble silicon IGBT has not retreated. It remains the dominant switching device in mid-power applications, particularly for motor control and power conversion. The real legacy of IGBT is not dominance forever but rather the ability to do work that would otherwise be impossible: air conditioners that dial down fan speed after reaching set temperature, consuming a fraction of the energy; wind turbines that rectify variable-frequency generator output into storable DC current; electric vehicles that convert battery DC into the variable-frequency AC that drives the motor. These applications would not exist in their current form without the IGBT.
In their own words · checked verbatim
It's more like as if the two got mixed up inside the teleportation pods from that movie, the Fly. … It's semiconductor body horror.
He told them bluntly, if Research doesn’t come up with something then he will tell Welch that the labs were unnecessary.
I thought, what I’ll do is I’ll make the power MOSFET and I’ll have to make sure that it does have some bipolar action as well, and this will give me a lot of high current-density capability
Carl F. Wheatley12:28
This parasitic NPN BJT reacts with the normal PNP BJT to create a self-sustaining loop like that reactor at the end of the Spider-Man 2 movie.
Nakagawa's insight - which looks obvious in hindsight but is also why it is so brilliant - was to lower the device's saturation current to below its latch-up current limit.
In the end, it all comes down to what you think is the IGBT we are talking about.
Figures
| First-generation power MOSFET power rating | less than 1 watt | 5:15 |
| GE's first IGBT wafer run | August 1981 | 11:24 |
| IGBT commercial launch | June 1983 | 12:28 |
| Patents filed by Baliga for IGBT improvements | over 20 | 19:34 |
| Pages in Baliga's IGBT monograph | over 800 | 19:34 |
Glossary
- IGBT
- Insulated-gate bipolar transistor—a power semiconductor combining MOSFET-style gate control with BJT-style high-current conduction.
- latch-up
- A runaway conduction state where parasitic thyristor action prevents the gate from turning off the device, requiring external current limiting to stop.
- conductivity modulation
- Simultaneous injection of minority and majority carriers to dramatically reduce resistance and enable higher current conduction.
- Deep P+
- A heavily boron-doped region added to limit hole flow and reduce parasitic thyristor action in IGBT.
- parasitic NPN BJT
- An unintended NPN transistor structure within IGBT that can be accidentally triggered, causing latch-up and device failure.
How to listen
Power electronics engineers selecting semiconductor devices, or anyone interested in semiconductor invention history and patent disputes.