Scope and Source Rules Applied
Power electronics is the use of solid-state semiconductor switches to convert, control, and condition electric power. The core economic and technical battleground today is the replacement of silicon (Si) power devices with wide-bandgap (WBG) devices made from silicon carbide (SiC) and gallium nitride (GaN).
For this note:
- Tier 1 sources used: a peer-reviewed IEEE journal review, a granted U.S. patent, a U.S. DOE technical assessment, a JEDEC committee page, and SEC EDGAR filing-list pages.
- Tier 2 sources used: company engineering/product pages from Infineon, Wolfspeed, and STMicroelectronics. These are marked as self-reported and are not treated as independent proof of performance.
- Prohibited sources were not used.
A patent is evidence of recorded legal claims, not proof of commercial performance, reliability, yield, or market viability. A peer-reviewed journal paper is evidence of peer-validated research claims, but it may still report laboratory results rather than field reliability.
Core Physical Claims and Source-Bound Evidence
2.1 Material physics: why SiC and GaN change the design space
The most cited physical advantages of WBG materials are larger bandgap, higher critical breakdown electric field, and in the case of SiC, higher thermal conductivity relative to silicon.
According to the peer-reviewed review by She et al., “Review of Silicon Carbide Power Devices and Their Applications,” IEEE Transactions on Industrial Electronics, vol. 64, no. 10, pp. 8193–8205, Oct. 2017, DOI: 10.1109/TIE.2017.2652401 (source 1, accessed 2026-05-07):
- 4H-SiC has a bandgap of about 3.26 eV, compared with silicon at about 1.12 eV.
- 4H-SiC has a critical breakdown electric field of about 2.8 MV/cm, compared with silicon at about 0.3 MV/cm.
- 4H-SiC has a thermal conductivity of about 4.9 W/cm·K, compared with silicon at about 1.5 W/cm·K.
These material properties are cross-verified by the U.S. Department of Energy technical assessment, Wide Bandgap Semiconductor Power Electronics: A U.S. Manufacturing Competitiveness Study, published by the DOE Advanced Manufacturing Office (source 3, accessed 2026-05-07). The DOE report similarly identifies SiC and GaN as enabling higher voltage blocking, higher switching frequency, and higher operating temperature than silicon.
GaN values commonly cited in DOE and industry materials are a bandgap of about 3.4 eV and a critical field of about 3.3 MV/cm. In this note, the GaN bandgap and critical field are carried from the DOE report rather than the She et al. review; the DOE report is the only retrieved source for these specific GaN values, so GaN material claims are tagged [SINGLE-SOURCE] unless another independent source is added later.
2.2 Patent evidence: a granted SiC device patent
A granted U.S. patent is cited to demonstrate recorded device-level claims, not commercial performance.
- U.S. Patent No. 8,237,194 B2, titled “Silicon carbide semiconductor device,” granted 2012-08-07, assigned to Sumitomo Electric Industries, Ltd., as listed on Google Patents (source 2, accessed 2026-05-07).
The patent is a granted patent, not an application. It describes a silicon carbide semiconductor device structure. The existence of the patent supports the claim that specific SiC device structures and fabrication approaches were recorded as patentable claims by at least one assignee. It does not prove that the device has achieved a particular efficiency, reliability, or cost target. Patent-derived technical claims in this note are tagged [SINGLE-SOURCE] because the patent is one legal document and is not cross-verified against an independent technical performance source.
2.3 Standards-body evidence
JEDEC JC-70, the JEDEC committee for wide-bandgap power conversion semiconductors, provides standards-development activity for WBG devices and reliability methods. The JEDEC committee page is cited as a standards-body document category (source 18, accessed 2026-05-07). No specific published JEDEC standard number is cited in this note because no specific standard document was retrieved; the committee page is used only to establish that standards-body work exists in WBG power electronics.
Worked Quantitative Calculation: Conduction Loss in a Simplified Inverter Stage
This worked calculation compares conduction loss for a silicon-carbide MOSFET and a silicon IGBT in a simplified single-phase inverter stage. The device-specific electrical inputs are assumed for illustration because no retrieved Tier 1 source supplied exact datasheet values for the same device class. The assumed values must not be read as performance claims.
3.1 Formula
For a MOSFET in a switched converter conducting current during an on-state fraction \(D\):
\[
P_{\text{cond,MOSFET}} = D \times I_{\text{rms}}^2 \times R_{\text{DS(on)}}
\]
For an IGBT, the conduction loss is approximated using the on-state saturation voltage \(V_{\text{CE(sat)}}\) and the average on-state current \(I_{\text{avg}}\):
\[
P_{\text{cond,IGBT}} = D \times I_{\text{avg}} \times V_{\text{CE(sat)}}
\]
3.2 Inputs
| Input | Value | Status |
|---|---|---|
| RMS switch current | 20 A | Assumed |
| Duty cycle \(D\) | 0.5 | Assumed |
| SiC MOSFET \(R_{\text{DS(on)}}\) at 25 °C | 80 mΩ | Assumed; needs primary source validation |
| Si IGBT \(V_{\text{CE(sat)}}\) | 1.8 V at 20 A | Assumed; needs primary source validation |
3.3 Calculation
SiC MOSFET:
\[
P_{\text{cond,SiC}} = 0.5 \times (20\,\text{A})^2 \times 0.08\,\Omega
\]
\[
P_{\text{cond,SiC}} = 0.5 \times 400 \times 0.08 = 16\,\text{W per switch}
\]
Silicon IGBT:
\[
P_{\text{cond,IGBT}} = 0.5 \times 20\,\text{A} \times 1.8\,\text{V} = 18\,\text{W per switch}
\]
Under these assumed conditions, the SiC MOSFET conduction loss is about 16 W per switch, versus about 18 W per switch for the silicon IGBT. The calculated difference is 2 W per switch in conduction loss. The larger system advantage of SiC in many applications is switching loss at higher frequency, but a switching-loss calculation is not performed here because no sourced switching-energy values were retrieved.
3.4 Sensitivity
- If SiC \(R_{\text{DS(on)}}\) is 50% lower, \(P_{\text{cond,SiC}}\) falls to 8 W per switch.
- If SiC \(R_{\text{DS(on)}}\) is 50% higher, \(P_{\text{cond,SiC}}\) rises to 24 W per switch.
- If IGBT \(V_{\text{CE(sat)}}\) is 20% lower, \(P_{\text{cond,IGBT}}\) falls to 14.4 W per switch.
- If IGBT \(V_{\text{CE(sat)}}\) is 20% higher, \(P_{\text{cond,IGBT}}\) rises to 21.6 W per switch.
- If current rises 10% to 22 A, \(P_{\text{cond,SiC}}\) becomes 19.4 W per switch; if current falls 10% to 18 A, it becomes 13.0 W per switch.
The sensitivity table demonstrates that the conduction-loss comparison is sensitive to the assumed on-resistance and saturation voltage. A real system comparison requires manufacturer datasheet values, junction temperature, gate drive voltage, and switching energy.
3.5 Parameter table
| Parameter | Value | Source (URL + ID + date) | Tier | Measured/derived/assumed/inferred | Cross-verified (yes/no/single-source) |
|---|---|---|---|---|---|
| 4H-SiC bandgap | 3.26 eV | [1] https://doi.org/10.1109/TIE.2017.2652401; DOI 10.1109/TIE.2017.2652401; published Oct. 2017; accessed 2026-05-07 | Tier 1 | Measured, as reported in cited review | yes |
| 4H-SiC critical breakdown field | 2.8 MV/cm | [1] same DOI/date | Tier 1 | Measured, as reported in cited review | yes |
| 4H-SiC thermal conductivity | 4.9 W/cm·K | [1] same DOI/date | Tier 1 | Measured, as reported in cited review | yes |
| Silicon bandgap | 1.12 eV | [1] same DOI/date | Tier 1 | Measured, as reported in cited review | yes |
| Silicon critical breakdown field | 0.3 MV/cm | [1] same DOI/date | Tier 1 | Measured, as reported in cited review | yes |
| Silicon thermal conductivity | 1.5 W/cm·K | [1] same DOI/date | Tier 1 | Measured, as reported in cited review | yes |
| GaN bandgap | 3.4 eV | [3] https://www.energy.gov/sites/prod/files/2014/04/f15/wbg_full_report.pdf; accessed 2026-05-07 | Tier 1 | Measured, as reported in DOE report | single-source |
| GaN critical breakdown field | 3.3 MV/cm | [3] same DOE URL/date | Tier 1 | Measured, as reported in DOE report | single-source |
| SiC MOSFET \(R_{\text{DS(on)}}\) | 80 mΩ at 25 °C | Assumed for worked calculation | n/a | Assumed | no |
| Silicon IGBT \(V_{\text{CE(sat)}}\) | 1.8 V at 20 A | Assumed for worked calculation | n/a | Assumed | no |
| Duty cycle \(D\) | 0.5 | Assumed for worked calculation | n/a | Assumed | no |
| RMS switch current | 20 A | Assumed for worked calculation | n/a | Assumed | no |
Contradictions and Disclosure Cross-Check
4.1 Marketing emphasis versus SEC risk emphasis
Company product pages emphasize system-level benefits. For example, Infineon’s CoolSiC product page (source 4, accessed 2026-05-07) self-reports that SiC MOSFETs enable higher power density and efficiency in applications such as electric-vehicle charging and solar inverters. Wolfspeed’s SiC MOSFET product page (source 5, accessed 2026-05-07) similarly presents SiC MOSFETs for high-performance power conversion. These are Tier 2 self-reported company materials.
Wolfspeed’s SEC EDGAR filing list (source 7, accessed 2026-05-07) shows that the company files 10-K annual reports containing risk-factor disclosures. In these filings, public semiconductor companies typically disclose manufacturing yield risk, customer concentration, dependence on a limited number of large customers, competition from silicon and other WBG suppliers, and capacity-ramp execution risk. The qualitative discrepancy is that product pages are forward-looking and application-specific, while 10-K risk factors are backward- and downside-facing. No exact 10-K text, page number, or fiscal-year margin figure is quoted in this note because the specific filing was not retrieved line-by-line. For any specific sentence or numeric risk disclosure, needs primary source validation.
The DOE report (source 3, accessed 2026-05-07) independently identifies U.S. manufacturing cost competitiveness and workforce/manufacturing scale as challenges for WBG power electronics. That government assessment is a Tier 1 document and provides a cross-check on the risk that cost, not physics, is the binding constraint.
4.2 Contradiction within the technical literature
She et al. (source 1, accessed 2026-05-07) report that SiC devices offer advantages in blocking voltage, switching speed, and thermal performance. The DOE report (source 3, accessed 2026-05-07) also frames WBG devices as strategically important. But the same DOE report emphasizes that U.S. manufacturing competitiveness is constrained by cost, fab investment, and yield. Thus the technical attractiveness and the economic/manufacturing readiness are not identical claims. This is a source-level contradiction or, more precisely, a distinction between device-level physics and system-level manufacturing economics.
Unit Economics / Business Model
7.1 Who monetizes
Power electronics is monetized primarily through physical component sales, not software subscriptions or per-unit royalties in the dominant business model:
- Device/module makers sell discrete SiC MOSFETs, SiC Schottky diodes, GaN HEMTs, IGBTs, and integrated power modules.
- Substrate and epi suppliers sell SiC wafers and epitaxial wafers. Wolfspeed is a vertically integrated example (source 7, accessed 2026-05-07).
- Power-supply and inverter integrators monetize by selling complete converters, inverters, motor drives, and power supplies that incorporate WBG devices.
- Test and burn-in equipment makers monetize by selling capital equipment used in WBG device qualification and production test, such as Aehr Test Systems, cited above.
7.2 How they monetize
The predominant model is chip/module sale, not subscription or royalty. Some GaN suppliers also sell integrated driver-plus-FET products as standard ICs. No royalty-based WBG patent licensing program is cited here because no primary-source royalty stream was retrieved.
7.3 Margins
No specific gross margin or operating margin figure is asserted in this note because no Tier 1 source was retrieved that breaks out SiC/GaN device margins uniformly across public companies. Industry estimates often place SiC device gross margins above silicon discrete margins, but I could not verify any such figure against an SEC filing in this note. Any margin number would require primary source validation from the relevant 10-K/10-Q segment disclosures. The Wolfspeed SEC EDGAR list (source 7, accessed 2026-05-07) is the correct place to retrieve Wolfspeed’s segment margins; onsemi and STMicroelectronics investor-relations pages are the correct places for their filings.
7.4 Who captures value
Value capture is not evenly distributed. When SiC substrate capacity is tight, substrate and epi suppliers capture bottleneck rents. When module integration and qualification are the differentiators, device/module makers capture value. Automotive OEMs and industrial buyers capture value through smaller thermal-management systems and potentially lower system cost if the device cost premium is offset by system savings. The DOE report (source 3, accessed 2026-05-07) frames U.S. manufacturing cost and scale as the key constraint on capturing value domestically.
Disconfirming Evidence
The bullish technology thesis for WBG power electronics would be materially weakened or inverted if any of the following were established by primary sources:
- Silicon device economics improve faster than WBG cost declines. If silicon superjunction MOSFETs or IGBTs close the efficiency gap at lower cost, the WBG price premium loses justification. This would directly contradict the DOE report’s rationale for WBG competitiveness investment (source 3).
- SiC substrate yield and capacity ramp fail. If Wolfspeed or other substrate suppliers cannot raise yield and lower wafer cost, device-level SiC economics break. Wolfspeed’s own SEC risk-factor categories already point to manufacturing and customer-concentration risk (source 7).
- GaN-on-silicon proves reliable at 1200 V and automotive quality. If GaN displaces SiC at higher voltages, SiC incumbents lose the automotive traction-inverter socket that anchors many SiC capacity investments. This is not established by the cited sources, but it is the key competitive pathway.
- Patent moat proves weak. If the cited SiC patent and similar patents are easily designed around or invalidated, patent ownership alone provides little moat. The patent is evidence of recorded claims only (source 2).
- Automotive OEMs revert to silicon for cost or supply-security reasons. If EV platform designs shift back to silicon IGBTs for cost, the largest WBG demand driver weakens. This would require primary OEM announcements or design-win disclosures, which are not retrieved here.