IM Engineer Study Notes
← Back to list
Hardware & Semiconductor
#전력반도체#SiC#GaN#와이드밴드갭#IGBT
Last updated · 2026-10-19

Power Semiconductors and Compound Power Devices (SiC/GaN)

1. Overview

A power semiconductor is the general term for semiconductor devices that switch or rectify high voltage and large current in order to convert (AC↔DC), control (adjust voltage, current, frequency), and distribute electrical energy. Whereas logic and memory semiconductors, which handle computation and storage, deal with "information," power semiconductors are fundamentally different in that they deal with "energy."

The fundamental reason power semiconductors rose to prominence as an independent industrial and academic field is that an intermediate stage of "power conversion" inevitably sits between the generation and consumption of energy. A battery delivers direct current (DC) but an electric motor runs on alternating current (AC); the low-voltage DC produced by a solar panel must be turned into high-voltage AC to connect to the grid; and a laptop adapter steps 220V AC down to a few volts of DC. A power semiconductor acts as the switch at every point where such conversion occurs, and the conversion efficiency directly governs energy loss, heat generation, and the size, weight, and lifetime of the product. In other words, the power semiconductor is "the heart valve of every device that uses electricity."

Traditionally this role has been served by silicon (Si)-based devices (diodes, thyristors, MOSFETs, IGBTs). However, as demand surged for handling power at higher voltages, faster, and with less loss—as in electric vehicles (EVs), renewable energy, data centers, and fast charging—silicon ran into the limits of its material properties. What emerged to break through this limit was the wide band-gap (WBG) compound semiconductor, namely silicon carbide (SiC) and gallium nitride (GaN). This document first organizes the classification and operating principles of power semiconductors, then explores in depth—from the material, circuit, and system perspectives—why WBG devices are replacing silicon.

In terms of scale, power semiconductors are overshadowed by memory and logic yet constitute a far from small market, and riding the vast currents of electrification and carbon neutrality they are one of the fastest-growing semiconductor fields. This is because the electrification of a single vehicle demands hundreds of additional power semiconductors, while data centers, renewables, and industrial automation pull up demand simultaneously. What is interesting is that this field does not follow the logic of process-node competition. Whereas logic semiconductors raise performance through line-width scaling down to a few nm, power semiconductors compete through innovation in materials (SiC/GaN), device structure, and packaging even at relatively large processes of hundreds of nm to a few μm. Thus, even without a cutting-edge fine-process fab, a latecomer that secures material, design, and packaging capability has room to enter, making it a field that nations strategically cultivate.

The core characteristics of power semiconductors can be summarized in three points. First, high withstand voltage and large current: they must endure voltages of hundreds of V to several kV and currents of tens to thousands of A. Second, switching characteristics: they must repeat conduction (on) and blocking (off) quickly while keeping the switching loss of the transition small. Third, thermal management: since loss ultimately turns into heat, heat-dissipation design and the junction temperature limit constrain performance. Because these three characteristics trade off against one another, power semiconductor design comes down to a system-optimization problem of "which device to use, for which application, and how to cool it."

2. Classification of Power Semiconductors and the Power Conversion Scheme

Power semiconductors are classified hierarchically according to controllability and the voltage/frequency region. The simplest diode is a passive rectifying device that only determines the current direction; the thyristor (SCR) controls only turn-on; and the MOSFET/IGBT are active switches that control both turn-on and turn-off via the gate. The structure diagram below shows the full classification of power semiconductors and their application areas.

graph TD
    A["Power Semiconductor"] --> B["Rectifying device (Diode)"]
    A --> C["Phase-control device (Thyristor/SCR)"]
    A --> D["Fully controllable switch"]
    D --> E["Power MOSFET (low-voltage, high-frequency)"]
    D --> F["IGBT (high-voltage, large-current)"]
    D --> G["WBG device (SiC/GaN)"]
    G --> G1["SiC MOSFET (600V - kV class)"]
    G --> G2["GaN HEMT (100V - 650V)"]
    B --> H["Power supply, rectifier circuits"]
    E --> I["Adapters, DC-DC, data-center power"]
    F --> J["EV inverters, rail, grid, industrial motors"]
    G1 --> K["EV traction inverter, fast charging, PV inverter"]
    G2 --> L["Fast chargers, 48V server power, AI power"]

As the structure diagram shows, the greater the "controllability" of a power semiconductor, the more sophisticated the applications it can handle. A diode sets only the current direction and a thyristor only the turn-on instant, but MOSFET/IGBT/WBG devices control both the turn-on and turn-off instants, freely shaping voltage, current, and frequency. It is precisely this full-control capability that is the foundation of modern power conversion—which produces waveforms as in inverters and converters—and WBG devices are both a continuation of and a leap beyond existing devices in that they exercise this capability "faster and at higher voltage."

The most basic device, the diode, is a passive device that uses the rectifying action of a PN junction to let current flow in only one direction, converting AC to DC. Lacking a control terminal, it cannot actively regulate switching, but its recovery characteristic (reverse recovery) has a large effect on overall circuit loss. In particular, the reverse recovery current that flows momentarily when a diode transitions to the reverse direction in high-speed switching circuits is a chief culprit of loss and noise, so the SiC Schottky barrier diode (SBD), which has almost none of this current, is rapidly replacing the Si fast diode. The thyristor (SCR) is a semi-controllable device whose turn-on instant alone can be controlled by the gate; once it conducts, it does not turn itself off until the current approaches zero, so it is still used in regions of very high power and low switching frequency such as ultra-high-voltage transmission (HVDC) and large-capacity rectification.

Among silicon devices, the MOSFET is advantageous for voltage drive and high-speed switching, but at high voltage its on-resistance (R_DS(on)) grows sharply, making it suitable for low-voltage, high-frequency regions below a few hundred volts. Because of the so-called "silicon limit"—where on-resistance grows in proportion to the 2 to 2.5 power of the withstand voltage—using a MOSFET at high voltage makes heat generation unbearable. By contrast, the IGBT (Insulated Gate Bipolar Transistor) combines the gate-drive convenience of a MOSFET with the low conduction loss of a bipolar device, and has dominated the high-voltage, large-current region of 600V to 6.5kV (EV propulsion, rail, wind, industrial motor drives) for more than two decades. However, the IGBT has a structural weakness: the tail current generated as minority carriers are swept out at turn-off makes switching loss large, so it is hard to raise the frequency above tens of kHz. It is precisely this dual constraint of silicon—"MOSFETs cannot be used at high voltage, and IGBTs cannot raise their frequency"—that is the niche WBG devices have penetrated.

To understand the power-semiconductor industry, one must also note the stage-by-stage particularities of the value chain. SiC/GaN power semiconductors tightly interlock across every stage: (1) single-crystal ingot/wafer (material), (2) epitaxy growth, (3) device design/front-end (fab), (4) module packaging/back-end, and (5) gate driver/control. In SiC in particular, the material (wafer) accounts for about half the cost, giving it a unique structure in which, unlike memory and logic, the competitiveness of the material stage governs the entire business viability. For this reason, material companies such as Wolfspeed and Coherent (II-VI) hold industry leadership, and device makers compete to sign long-term wafer supply contracts with them.

Power conversion is divided into four types according to the form of the input/output power, and a power semiconductor switch is central to each conversion. Rectification (AC→DC) is handled by diodes/bridges, inversion (DC→AC) by MOSFET/IGBT/SiC, conversion (DC→DC) by MOSFET/GaN, and the cycloconverter (AC→AC) by thyristors. For example, in an EV the traction inverter that turns the battery's DC into three-phase AC for the motor directly governs driving efficiency, and which device is used there changes the single-charge range by a few percent. The table below compares the characteristics of the main devices.

Category Si MOSFET Si IGBT SiC MOSFET GaN HEMT
Voltage region ~900V 600V-6.5kV 650V-kV class 100V-650V
Switching speed Fast Slow (tail current) Very fast Fastest
Conduction loss Large at high voltage Small Small Small
Frequency Hundreds of kHz ~20kHz Hundreds of kHz MHz class
Typical applications Adapters, DC-DC EV, industrial motors EV inverter, PV Fast charging, server power

The key to read from the table is not simple superiority but a division of regions. MOSFET/GaN suit low-voltage, high-frequency, while IGBT/SiC suit high-voltage, large-current, and no single one dominates every region. This means power semiconductor design is not a matter of choosing "the best device" but of choosing "the device that best fits this application's voltage, current, frequency, and cost constraints." But why does this division of regions arise? The table alone does not explain "why the difference occurs." The key lies in the material property called band-gap energy. The next chapter traces how this property leads to device performance, down to the circuit and system stages.

3. Material Properties and Device Operation of Wide Band-Gap (WBG) Compound Semiconductors

Whereas the band gap of silicon is about 1.12eV, 4H-SiC is about 3.26eV and GaN about 3.4eV, 2.9 to 3 times wider. A wider band gap means more energy is needed for an electron to jump to the conduction band, and from this three decisive advantages cascade. The first is a rise in the breakdown field. While silicon's breakdown field is about 0.3MV/cm, SiC/GaN is about 3MV/cm, roughly 10 times higher. To obtain the same withstand voltage, the drift region can be made 1/10 as thin, which leads directly to a sharp reduction in on-resistance. It is for exactly this reason that a SiC MOSFET achieves low loss with a far smaller area than a Si MOSFET of the same withstand voltage.

Where GaN particularly shines is electron mobility. The two-dimensional electron gas (2DEG) that forms naturally at the AlGaN/GaN heterojunction interface has little impurity scattering, so electrons move very fast; thanks to this, the GaN HEMT has small parasitic capacitance and extremely fast turn-on/turn-off, enabling MHz-class switching. When switching is fast, the same power can be processed at a higher frequency, and when the frequency rises, the size of the inductors, capacitors, and transformers that temporarily store energy shrinks inversely. The answer to "why is a GaN charger small" lies in exactly this property, and it is the key principle that drives up power density (W/cm³).

The second is high-temperature operation. A wide band gap lowers the probability that electrons are thermally excited, so the device retains its semiconductor characteristics above 200°C—beyond the 150-175°C where silicon shows its limits. This reduces the heat-dissipation burden and simplifies the cooling system, returning as a real benefit in EVs by reducing the size and weight of coolant loops and heat sinks. The third is thermal conductivity; SiC in particular, at about 4.9W/cm·K, is more than three times that of silicon (about 1.5W/cm·K), so it can draw out generated heat quickly. The detailed architecture diagram below shows the causal chain by which WBG material properties propagate into device-, circuit-, and system-level gains.

flowchart LR
    subgraph PHY["Material"]
      P1["Wide band gap (3.3eV)"]
      P2["High breakdown field (~3MV/cm)"]
      P3["High electron mobility (GaN)"]
    end
    subgraph DEV["Device"]
      D1["Thin drift layer -> low Rds(on)"]
      D2["Small parasitic capacitance -> fast switching"]
      D3["High-temp operation (over 200C)"]
    end
    subgraph SYS["System"]
      S1["Lower switching loss -> high efficiency"]
      S2["Higher frequency -> smaller passives"]
      S3["Less cooling burden -> lighter weight"]
    end
    P2 --> D1 --> S1
    P3 --> D2 --> S2
    P1 --> D3 --> S3

The measure that bundles these individual properties into a single metric to gauge a device's potential is Baliga's Figure of Merit (BFOM). BFOM is proportional to permittivity, electron mobility, and the cube of the breakdown field; because the breakdown field enters as a cube, SiC/GaN, with a field 10 times higher, are evaluated as having, in theory, hundreds of times the performance potential of Si. Of course, real devices fall far short of this theoretical value owing to process maturity, defects, and packaging limits, but this figure is precisely the fundamental basis explaining "why the industry moves to WBG while bearing enormous investment." In other words, the WBG transition is not a fad but a direction dictated by material properties, and the remaining task is how economically that potential can be realized.

The point of the causal diagram above is that WBG's advantage is not the superiority of any single metric but a cascade effect amplified and propagated from material → device → system. A 10-fold breakdown field leads to reduced on-resistance, small parasitic capacitance to higher frequency, and a wide band gap to high-temperature tolerance; these three converge at the system stage into efficiency, miniaturization, and weight reduction. Therefore, when evaluating a WBG device one must look not at a single parameter but at how much this cascade is realized in the final product—and what blocks that realization is the cost, reliability, and design difficulty examined later.

SiC and GaN belong to the same WBG family but differ in device structure and strengths. The SiC MOSFET has a vertical structure like silicon, with current flowing through the substrate, making it advantageous for high voltage and large current. 1200V is the representative rating, and its main arenas are regions handling high power such as EV traction inverters, solar PCS, and fast chargers. By contrast, the GaN HEMT (High Electron Mobility Transistor) is a lateral device that uses the two-dimensional electron gas (2DEG) formed at the heterojunction (AlGaN/GaN) interface as its channel; its electron mobility is very high, enabling MHz-class ultra-fast switching, but its withstand voltage is limited to 650V or below. Thus GaN shines in medium-to-low voltage, ultra-high frequency regions such as fast-charging adapters, DC-DC converters, and 48V server power.

Another decisive difference between the two devices lies in the substrate and cost structure. SiC makes devices on a SiC single-crystal substrate, but because single-crystal growth is extremely slow (on the order of a few mm per day) and defect control is difficult, the substrate itself is expensive. GaN, by contrast, can use the GaN-on-Si approach of growing a GaN thin film on a cheap silicon substrate, giving it great potential to lower cost by leveraging existing silicon fab (up to 200/300mm) infrastructure. However, GaN-on-Si faces the challenge of managing stress and defects caused by differences in lattice constant and thermal expansion coefficient. In this way SiC has the profile "an expensive substrate in exchange for high voltage and high reliability," and GaN "limited withstand voltage in exchange for low cost and ultra-high frequency"—opposite cost-performance profiles—and it is more accurate to see the two materials not as competitors but in a complementary relationship, dividing different regions along the voltage/frequency axes.

A practically important point is that the GaN HEMT is inherently normally-on (depletion). If current flows when the gate voltage is zero, it can lead to a short-circuit accident at power-up, so for power use it must be made normally-off (enhancement). To achieve this, a p-GaN gate structure or a cascode scheme that connects a low-voltage Si MOSFET in series with a high-voltage GaN is used. In this way, as WBG devices realize their material advantages in actual products, the difficulty of gate-drive and protection circuits rises alongside, and this becomes one axis of the trade-off discussed later. For example, the SiC MOSFET has a narrower gate-drive voltage margin than silicon, so its oxide is easily damaged under overvoltage, and it requires a dedicated driver that turns it off firmly with a negative (-) gate voltage. GaN switches so fast that it responds sensitively even to small parasitics, so it demands a design that places the driver right next to the device and shrinks the loop area to the extreme. In the end, the designer must recognize from the outset that WBG performance is realized not in the numbers on the device datasheet but in the surrounding circuit design capability that draws those numbers out.

4. Application Cases and Practical Implications versus Si

The effect of the WBG transition gains persuasive power when confirmed not by abstract material properties but by concrete product figures. The common pattern is a value chain running "efficiency improvement → reduced loss and heat → shrunken cooling and passives → compact, light, high-density," overlaid with the per-application arithmetic that "the economic value of efficiency is proportional to voltage, current, and operating time." The four cases below show how this pattern is realized as concrete gains in different applications. The most symbolic case is the EV traction inverter. The Tesla Model 3 (mass-produced in 2017) was the first mass-produced EV to adopt a SiC MOSFET (supplied by STMicroelectronics, 24 devices per module) in its traction inverter, reducing inverter loss versus the IGBT to extend range by a few percent on the same battery and to make the inverter compact and light. Thereafter Lucid Air used six of Wolfspeed's XM3 half-bridge SiC modules per inverter, and SiC traction inverters have become the de facto standard for premium EVs. In EVs, where a 1% difference in power loss ties directly to range and battery cost, device choice is no longer a component-level matter but one of product competitiveness.

The second case is the fast-charging adapter. A 65W laptop adapter used to be a brick-sized Si-based unit, but using a GaN HEMT raises the switching frequency to the MHz class, greatly reducing passives such as transformers, inductors, and capacitors. As a result, a GaN charger of the same output cuts volume and weight to less than half while being even more efficient. This is the direct cause of the consumer-felt value of a "small charger," and a product-level demonstration of the Chapter 3 causal chain that passive miniaturization stems from higher frequency.

The third case is data-center and AI power. As the power density of GPU racks for generative AI surges, power-supply efficiency has become a key variable governing electricity bills, cooling costs, and carbon emissions. The industry is reorganizing server power around 48V, reviewing 800V HVDC distribution, and aggressively adopting GaN/SiC for conversion in these segments. Navitas, Innoscience, and others have introduced 48V conversion and 800V HVDC solutions for AI data centers, while onsemi is developing 650V devices on a 200mm GaN-on-Si platform with GlobalFoundries—competition is heating up. Because a 1%p efficiency improvement translates in large data centers into savings of hundreds of millions of won in annual electricity costs, the economic incentive to adopt WBG is clear.

The fourth case is the solar/ESS power conversion system (PCS). The inverter that turns the DC produced by solar panels into grid AC, and the bidirectional charge/discharge converter of an energy storage system (ESS), operate all day, so even a 0.5%p efficiency difference is cumulatively reflected in generation revenue. A SiC-based string inverter reduces switching loss to achieve around 99% EU efficiency and, through higher frequency, cuts reactors and filters to lower installation area and weight. At plant scale, 0.5%p of efficiency translates into substantial additional generation over decades of operation, so demand for SiC PCS is steadily rising alongside the expansion of renewables. This case shows another axis of power semiconductor economics—"the value of efficiency is proportional to operating time."

That said, WBG does not immediately replace silicon in every region. The core reason is cost and maturity. SiC wafers are several times more expensive than silicon because single-crystal growth is difficult and defect (micropipe, etc.) control is tricky, and GaN has high circuit/layout design difficulty due to noise (EMI) from high-speed switching and the sensitivity of its gate drive. Therefore, in low-voltage, low-cost, high-volume regions (general appliances, low-output power supplies) silicon is still economical, while WBG penetrates first in EVs, renewables, fast charging, and AI power, where the value of high efficiency and high density exceeds the cost. Device choice is a matter not only of performance superiority but of a total-cost-of-ownership (TCO) judgment: "can the added cost be recovered through the value of efficiency and miniaturization?"

5. Deep Dive: Latest Trends, Supply Chain, and Standards

WBG power semiconductors have passed the stage of material-property debate and entered a phase of industrial competition over "who can mass-produce more cheaply, more stably, and more reliably." The variables that govern this phase are wafer diameter scaling, packaging/module innovation, standards/reliability systems, and supply-chain leadership. Below, these axes are examined in turn.

The biggest inflection point in the power-semiconductor industry is the transition of SiC wafers to 200mm (8 inch). Moving from the existing 150mm (6 inch) to 200mm increases the diameter by about 1.7 times, greatly increasing the chips per wafer, and this is seen as the key to structurally lowering SiC device cost. Wolfspeed brought the world's first 200mm-dedicated SiC fab online in Marcy, New York, USA, and formalized commercial shipment of 200mm materials in September 2025; Infineon, STMicroelectronics, and China's Sanan have also moved to secure 200mm lines. However, the 8-inch transition hinges on yield stabilization, and some analyses hold that early mass-production yields fall short of expectations—showing that "material superiority is not immediately supply stability."

In the GaN camp, higher voltage and data-center applications lead the trend. In 2024, Infineon announced the 650V G5 CoolGaN family on 8-inch wafers, targeting consumer, data-center, industrial, and solar uses, and market research firms (Yole, TrendForce, etc.) forecast the GaN power-device market to grow to about USD 3 billion by 2030, roughly four times (around 40% CAGR) the 2025 level (forecasts vary by firm and should be taken as indicative). Demand from AI and robotics is cited as a new driver of this growth.

As important as the device is packaging and module technology. No matter how fast and low-loss a WBG device is, if the package's parasitic inductance is large, excessive voltage spikes and ringing occur during high-speed switching, destroying the device or eroding its performance. Accordingly, low-inductance modules with reduced wire bonding, double-sided cooling that extracts heat from both sides of the chip, sintered silver joints that withstand high temperatures, and ceramic substrates (AMB) are advancing together. In other words, power-semiconductor competitiveness in the WBG era is decided not by the "chip" alone but by integrated design capability over "chip + package + module," another core capability domestic firms must catch up on.

The standards and reliability aspect also deserves attention. Because malfunction of a power semiconductor ties directly to safety issues such as fire and electric shock, automotive devices are subject to AEC-Q101 (discrete semiconductor reliability) and modules/systems to the functional-safety standard ISO 26262 (automotive). WBG devices have a relatively short history, so long-term reliability data is still accumulating, and WBG-specific reliability issues such as gate-oxide degradation in SiC MOSFETs and the rise of dynamic on-resistance (dynamic R_DS(on)) in GaN are being actively researched. Domestically, too, power semiconductors are designated a core system-semiconductor item, driving localization of materials, devices, and modules, but the high dependence on overseas sources for SiC single-crystal/epi materials and design capability leaves supply-chain self-reliance as a task.

Narrowing to the domestic ecosystem, a structure is forming in which foundry firms build SiC/GaN-dedicated process lines, material/device specialists invest in SiC wafers/epi and device design, and automakers, battery, and energy firms as demand sources pull localization forward. However, SiC single-crystal materials and high-reliability module design are still at the catch-up stage, so a self-reliant ecosystem is completed only when the government's power-semiconductor cultivation projects, joint industry-academia-research R&D, and demonstration/reliability-certification infrastructure proceed in parallel. Without vertically linked capability from material to module, device design alone can hardly secure a sustainable edge in global competition.

Another major axis on the application side is rail, the grid, and industrial motors. Replacing a rail vehicle's propulsion inverter with SiC raises regenerative-braking efficiency and shrinks transformers and cooling equipment, achieving vehicle lightening and energy savings at once, and in fact several high-speed and metro rail systems are adopting SiC inverters. In transmission and distribution, IGBTs and thyristors still dominate ultra-high-voltage, large-capacity regions such as HVDC, FACTS, and power STATCOMs, but as kV-class SiC devices mature they can dramatically reduce converter-station loss and scale, so long-term replacement is expected. In short, WBG is widening its penetration from consumer devices into the deep layers of industrial and energy infrastructure.

6. Considerations and Implications

From a professional engineer's perspective, power semiconductors expand beyond "which device is superior" to "how to embed power-semiconductor technology into the organization's product, energy, and supply-chain strategy." The considerations below present axes of decision-making at the multiple layers of device, circuit, system, supply chain, and policy.

First, device choice should be approached as TCO-based system optimization. WBG has a high device unit price, but its high efficiency and high frequency can shrink passives, cooling systems, and batteries to lower total system cost and life-cycle cost. Therefore one should compare Si/SiC/GaN from a TCO perspective that integrates efficiency, size, heat dissipation, and reliability—not the "device BOM unit price"—and design on the premise of a division of labor in which silicon serves low-voltage, high-volume regions and WBG serves high-efficiency, high-density regions.

Second, the secondary problems that high-speed switching creates (EMI, gate drive, layout) must be addressed together from the start of design. WBG's fast voltage/current slew rates (dv/dt, di/dt) are both the source of efficiency and a cause of noise, ringing, and malfunction. Because performance gains are canceled if the gate driver, parasitic-inductance-minimizing layout, and EMI filter are separated from device selection and bolted on later, device, drive, packaging, and layout must be co-designed as one.

Third, the reliability and functional-safety verification system must be realigned to WBG characteristics. Applying silicon-based reliability models as-is can miss WBG-specific degradation such as SiC gate-oxide degradation or GaN dynamic resistance. In safety-critical fields such as automotive, energy, and medical, compliance with standards like AEC-Q101 and ISO 26262 must proceed alongside degradation-mechanism-based verification through long-term accelerated-life testing and field-data accumulation.

Fourth, supply-chain risk and a localization strategy must be considered together. Because SiC single-crystal material and 200mm-transition capability are concentrated in a few firms, building a procurement strategy by looking only at device performance exposes one to supply disruption and price volatility risks. Securing multiple supply sources, long-term supply contracts, localization investment at the material/epi/packaging stages, and design capability through industry-academia-research cooperation is the path that simultaneously secures technological competitiveness and supply stability.

Fifth, one must recognize the strategic value from the perspective of the energy transition and carbon neutrality. Improving power-semiconductor efficiency contributes directly to society-wide energy efficiency and carbon emissions through EV adoption, renewable-energy integration, and data-center power savings. Viewed as a foundational technology of national energy and industrial competitiveness beyond a mere component of an individual product, it requires mid-to-long-term investment in R&D, talent, and the ecosystem.

Sixth, a phased, selective adoption strategy aligned with per-application roadmaps is realistic. Rather than rushing a full transition to WBG, a hybrid strategy that first applies it to traction inverters, fast charging, and AI power—where the efficiency/density value is greatest—while keeping silicon in low-cost, low-voltage regions is reasonable. By gradually accumulating design assets, verification data, and supply sources and widening the application scope in line with the device price-decline curve, one can manage the initial cost burden and reliability risk while not missing the right moment for the technology transition.

References

In one line: A power semiconductor is a device that converts and controls electrical energy; SiC/GaN with their wide band gaps surpass the material limits of silicon (MOSFET/IGBT) through high-efficiency, high-frequency, high-temperature operation and are rapidly penetrating EVs, renewables, fast charging, and AI power—but TCO-based system optimization that accounts for cost, reliability, and supply-chain maturity is the key to success.