Are IGBTs Still Relevant in the Age of SiC and GaN Silicon?

Why IGBTs Remain Relevant Even as SiC and GaN Gain Traction

It has long been rumored that a shift in the power conversion market is underway. The global market for SiC power devices is predicted to expand rapidly towards the 2030s, and GaN is also broadening its application scope from home appliances to data center power supplies. So, have silicon IGBTs (Insulated Gate Bipolar Transistors) become obsolete?

The answer is no. IGBTs are still overwhelmingly used in systems such as industrial inverters, railways, large UPS, and wind power generation. Factors such as cost, voltage handling capability, proven reliability, and the risks associated with replacement are intricately intertwined. Without dissecting the underlying structure, decisions on "when to switch to SiC" or "where IGBTs are sufficient" can become based on intuition.

Understanding IGBTs: The Synergy of MOSFETs and Bipolar Transistors

An IGBT combines the structures of a MOSFET and a bipolar transistor. It offers the gate-driving method of a MOSFET (voltage-controlled, simplifying circuitry) along with the high current density and low saturation voltage characteristics of a bipolar transistor.

This design philosophy is advantageous in the mid-to-high voltage and mid-to-high current domains. Covering a wide voltage breakdown range from 600V to 6.5kV, IGBTs still offer cost-effectiveness comparable to or better than SiC MOSFETs in low-frequency applications where conduction losses dominate switching losses—that is, in high-power systems operating at frequencies from a few hundred Hz to a few kHz.

However, IGBTs have structural limitations. The "tail current" phenomenon, inherent to bipolar operation where carriers take time to be fully swept out during switching off, leads to increased losses at high-frequency switching. This trade-off forms the basic criterion for application-specific selection.

A "Selection Map" Based on Voltage, Frequency, and Cost

Discussions on "SiC vs. IGBT" often devolve into simple performance comparisons. However, actual selection depends on the operating voltage range, switching frequency, and the overall cost structure of the system.

In terms of voltage, the 600V to 1200V range sees the most intense competition between SiC and IGBTs. Above 1700V, IGBTs remain dominant, and while SiC is entering this space, price differences are still significant. In the high-voltage range of 3.3kV to 6.5kV, IGBTs are nearly unchallenged.

Regarding switching frequency, SiC's superiority becomes clear in the high-frequency range above 20kHz. With lower switching losses, SiC can operate at higher frequencies within the same loss budget. Conversely, for applications around 1 to 10kHz, the low conduction losses and chip cost of IGBTs become advantageous.

This graph illustrates the structure where "SiC and GaN are more advantageous at higher frequencies, while IGBTs maintain competitiveness at lower frequencies." Railway and industrial motor drives are likely to continue being dominated by IGBTs, while EV on-board inverters and high-efficiency chargers can be seen as areas where a transition to SiC is progressing.

Even When SiC Appears to be the "Obvious Choice," Cost Barriers Remain Real

The die cost of SiC is still higher compared to silicon. This is attributed to the manufacturing difficulty and yield rates of SiC wafers. Even with the progression from 4-inch to 6-inch, and further to 8-inch wafer scaling, a cost difference of several times remains.

While SiC adoption is rapidly increasing in EV traction inverters, the background lies in system-level cost recovery through "reduced battery capacity due to lower losses" and "simplification of cooling systems." Even if the device itself is expensive, if the system cost can be recouped, the transition to SiC accelerates. Conversely, in applications where this calculation does not hold true, there remains a rational reason to stick with IGBTs.

Toshiba Device & Storage's triple-gate IGBT is particularly noteworthy in this context. This technology reportedly reduces losses by up to 40.5%, sparking a re-evaluation of "how far Si technology can advance" before considering a switch to SiC. As the cost advantage of SiC further diminishes, evolved IGBTs may continue to remain an option in system design.

The SiC MOSFET "Short-Circuit Withstand Time" Issue: A Design Asymmetry with IGBTs

While SiC possesses excellent characteristics, a widely shared design consideration is the issue of short-circuit withstand time (SCWT).

Short-circuit withstand time determines the "grace period" for the device before protective circuitry activates. A shorter time imposes stricter demands on the response speed of the protection circuit. SiC devices, with their smaller die size and high current density, experience faster temperature rises compared to silicon devices, necessitating a reduction in protection time. This asymmetry becomes a key point for design changes when replacing IGBTs.

Microchip's 700V/1200V SiC MOSFETs list a typical short-circuit withstand time of 3μs under specific conditions. Compared to the typical 10μs of general Si IGBTs, this difference directly impacts protection circuit design.

DESAT (Drain-Source voltage detection) is a protection method that monitors the on-state drain-source voltage to detect overcurrent. To accommodate the shorter withstand time of SiC, optimization of blanking time settings and thresholds is required. This also affects gate driver selection and design.

There is a trade-off between short-circuit withstand time and on-resistance (Ron). Mitsubishi Electric addresses this issue by introducing a p-type protective layer in their trench-type SiC-MOSFETs. ROHM's 4th generation SiC MOSFETs also aim to achieve both low RonA and high short-circuit withstand capability through their unique structure. The differing approaches by each manufacturer are aspects worth examining in product selection, going beyond datasheet figures to understand the underlying structural technology.

Where Does GaN Fit In? Organizing the Segmentation Among the Three

When discussing IGBTs and SiC, the question of where GaN fits in arises. GaN (Gallium Nitride) is predominantly a lateral device, with its current voltage limit generally around 650V. While research into vertical GaN is progressing, mass production is considered some time away.

GaN's current primary battleground is in high-frequency applications below 650V. Adoption is rapidly expanding in areas such as AC/DC converters for data centers, smartphone chargers, and on-board chargers (OBCs) for vehicles. In this domain, it can achieve higher operating frequencies than SiC MOSFETs, enabling further miniaturization and higher efficiency.

This segmentation is not fixed; as SiC's high voltage capability increases and costs decrease, it will encroach on IGBT territory. However, the pace of this encroachment varies by application. Currently, there are limited signs of IGBTs for 3.3kV–6.5kV railway applications being replaced by SiC in the near future, whereas replacement is steadily progressing in EV on-board inverters.

Decision-Making Criteria at the Intersection of Procurement, Design, and Business

If the "IGBT vs. SiC" discussion is solely a technical comparison, it overlooks perspectives from procurement and business. Supply chain structure also becomes a factor in decision-making.

The concentration of SiC wafer suppliers remains high, making procurement risk non-negligible. On the other hand, IGBTs are built upon mature manufacturing technology, with multiple suppliers competing, offering advantages in price negotiation power and supply stability.

onsemi offers a complete SiC portfolio, including SiC MOSFETs, SiC diodes, and SiC modules, from 650V to 1700V. Alongside Infineon's CoolSiC, these are increasingly incorporated into business plans as key SiC vendors. While the participation of major players like these is improving the stability of SiC supply, a comprehensive risk assessment of the entire supply chain, including wafer procurement, is being conducted in parallel with technology selection.

From a design perspective, there's an understanding that replacing IGBTs with SiC is not a "pin-compatible swap." Changes to gate drivers, redesign of short-circuit protection parameters, and re-evaluation of EMI characteristics—when these accumulate, the cost of replacement extends beyond component differences to development man-hours and certification fees. For updating existing systems, the simple calculation that "switching to SiC will improve efficiency" does not always hold true.

While these figures vary depending on system scale and organizational structure, they serve as a reminder of the decision-making criterion to "not compare solely based on component costs."

What to Confirm Next

From the preceding analysis, it is evident that IGBTs are not an "obsolete technology" but rather a technology that "continues to be an optimal choice under specific conditions." The challenge lies in determining whether these "specific conditions" align with one's own system.

First, it is essential to confirm where your application falls within the matrix of operating voltage and frequency. Next, conduct a calculation to see if the system-level cost structure—including not only component costs but also cooling, filters, board area, and certification costs—can truly be improved by adopting SiC. Then, if SiC is chosen, the suitability of short-circuit withstand time and gate driver design will be the initial technical investigation themes.

Information on the supply side, such as SiC wafer procurement risks and trends towards 8-inch wafer transition, along with each company's manufacturing capabilities, can be grasped more holistically if design and procurement review these aspects together. Related developments, such as the joint development of 8-inch SiC substrates by Mitsubishi Electric and Coherent, and the SiC module supply agreement between STMicroelectronics and Ampere for EVs, can also be referenced as examples for measuring the maturity of the SiC supply chain.

Whether continuing to use IGBTs or transitioning to SiC, the basis for such a decision should be articulated along the axes of voltage, frequency, cost, and risk, rather than simply following trends. This distinction is what differentiates the quality of technology selection.