RESOURCES

Industrializing Diamond Semiconductors: What Still Separates the Material from the Device

Diamond's bottleneck has never been physics — it is a reproducible wafer and a usable n-type layer. Drawing on peer-reviewed literature and the industry moves of 2026, this piece maps where diamond's "ultimate semiconductor" advantage is real, where n-type doping and dislocation density still block device-grade material, which segments already generate revenue (thermal management, high-power optics, quantum sensing), and what size and capacity actually constrain. The takeaway for investors and buyers: carat output is not electronic-grade capability.

diamond semiconductor2026-07 · 9 min

01The Conclusion First: The Money Is in the Material Layer, Not the Device Layer

Read the 2026 disclosures together and the picture is unambiguous: over the next three to five years, most revenue in diamond semiconductors will not come from power devices. It will come from two businesses that need no PN junction at all — thermal management (heat spreaders and GaN-on-diamond bonded substrates) and high-power optical windows — with quantum sensing close behind. The reason is not that the physics is weak. It is that of the four prerequisites in the value chain, only one is improving at a visible rate: large-area, low-dislocation single crystal (improving), controllable n-type doping (still a research problem), mature dielectric and interface processing, and reliable ohmic contacts. Power devices need all four at once; heat spreaders and windows need the first plus purity. For investors, this reframes diamond from a long-dated option on a device breakthrough into a business that already books orders — just at the material and packaging layer.

02How Large the Physical Advantage Really Is — and Why Orders of Magnitude in FOM Do Not Equal a Product

On paper, diamond has no rival. The ultra-wide-bandgap review by Xu et al. in Oxford Open Materials Science lists room-temperature values of 5.5 eV bandgap, 13 MV/cm critical field, 7,300 cm²/V·s electron mobility, 5,300 cm²/V·s hole mobility, and 2,290 W/(m·K) thermal conductivity — the highest thermal conductivity in the UWBG family. The 2024 Materials review by Zhao et al. uses slightly different conventions (5.47 eV, a theoretical 20 MV/cm field, 22 W/cm·K) but reaches the same verdict. Because the Baliga figure of merit scales with ε·μ·E_c³, the cube of the critical field lets diamond dominate any comparison: in its 2023 announcement, Diamond Foundry cited the textbook figure directly, putting diamond at 17,200× silicon and 60× silicon carbide. The operative phrase is "textbook." Those numbers assume intrinsic material, ideal interfaces, and fully ionized dopants. The same Materials review compiles the measured records — 874.6 MW/cm² BFOM, 10 kV breakdown, 60 kA/cm² current density. Respectable laboratory results, but the gap between them and 17,200× is precisely what the next two sections are about.

03Wall One: The 0.59 eV Problem in n-Type Doping

Diamond's most famous defect is the absence of a good donor. The same review gives activation energies of 0.37 eV for boron (p-type) and 0.59 eV for phosphorus (n-type). At 0.59 eV, the overwhelming majority of phosphorus donors simply do not ionize at room temperature: the epilayer stays highly resistive, and electron mobility is further suppressed by compensating centers and doping-induced defects. This is not a matter of insufficient process effort — it follows from diamond's small lattice constant and the small Bohr radius of its impurities. Even boron's 0.37 eV causes heavy carrier freeze-out at room temperature, forcing industry to buy conduction with high doping concentrations, which in turn risks a collapse from band conduction into hopping conduction. The consequence is written directly into the device family tree: the Materials review states plainly that the lack of n-type diamond limits bipolar device development, leaving PIN diodes and IGBT-class structures largely conceptual in diamond. In 2024, a NIMS team demonstrated the world's first n-channel diamond MOSFET — a lightly phosphorus-doped epilayer over a heavily doped contact layer — with field-effect mobility of roughly 150 cm²/V·s at 300°C. It is the most substantive step on this path in a decade, but note the temperature: elevated heat helps donors ionize, which is exactly why room-temperature operation remains the hard case. The industry's pragmatic detour is the hydrogen-terminated two-dimensional hole gas and surface transfer doping, at the cost of thermal stability and reliability.

04Wall Two: Size, Dislocations, and the Mosaic-versus-Heteroepitaxy Fork

The size reality is more conservative than the marketing. Xu et al. state that commercial single-crystal diamond tops out at roughly 10 mm. The Materials review catalogues the best results across three parallel routes: 300 mm has been reported for polycrystalline diamond substrates, 92 mm for heteroepitaxy, while the largest homoepitaxial "wafer" is a two-inch mosaic assembled from 24 individual crystals — and the grain boundaries in a mosaic tile disqualify it for vertical power devices, confining it to thermal and some optical uses. Heteroepitaxy is the other route: Qu et al. (arXiv:2404.08446, 2024) grew two-inch free-standing diamond on an Ir/YSZ/Si(001) stack using laser-patterned templates, with (400) and (311) rocking-curve FWHM of 313.5 and 359.3 arcseconds and an etch-pit dislocation density around 2.2 × 10⁷ cm⁻². That last number is the key to reading the whole industry timeline: how large a wafer can be and how good it can be are two independent progress bars, and device yield tracks the second. On 16 June 2026, Element Six and Orbray announced a reproducible process for three-inch wafer-scale single-crystal diamond, with four-inch in development and two-inch thermal-bonding wafers being prepared for volume production at Element Six's CVD facility in Gresham, Oregon. Diamond Foundry had already produced the first 100 mm single-crystal diamond wafer by heteroepitaxy on 6 November 2023, naming further defect-density reduction as its next objective. Diamond Foundry names further defect-density reduction as its next objective, while Element Six/Orbray pursue larger diameters (4-inch in development) alongside quality and volume — defect reduction matters at least as much as scaling up, which is itself evidence that diameter is not the only bottleneck.

05Three Segments That Already Ship — All of Them Route Around Doping

Identifying which applications already generate cash is more useful than guessing the year a diamond MOSFET arrives — and these three segments share one trait: they exploit diamond's thermal conductivity, optical transparency, or spin properties, and depend on neither doping nor junctions. First, thermal management for RF and compute. Akash Systems pursues GaN-on-diamond: on 13 November 2024 it announced a non-binding preliminary memorandum of terms with the U.S. Department of Commerce under the CHIPS Act, comprising $18.2 million in direct funding plus $50 million in combined federal and California tax credits, over $68 million in total; the company states its diamond cooling reduces GPU hot-spot temperature by 10–20°C. Bonding a 2,290 W/(m·K) material to the back of a GaN active layer physically hands junction-temperature budget back to the designer — value that waits on no doping breakthrough. Second, high-power optical windows, which demand high conductivity, low absorption, and thermal-shock resistance simultaneously; diamond is one of very few options for multi-kilowatt laser and millimeter-wave windows, and the requirement is purity and low stress, not electrical activation. Third, quantum sensing: nitrogen-vacancy centers operate at room temperature and demand ultra-low nitrogen background and low strain, entirely orthogonal to device-grade doping. Notably, the joint 2026 statement from Element Six and Orbray lists 6G wireless, power and RF electronics, sensing, thermal management, and quantum technologies together as targets — the ordering itself signals where the industry's priorities sit.

06Timescales: Three Steps, Not One

What follows is a read on industry trends, not a timeline commitment by any supplier. Diamond's industrialization is far clearer when split into three independent steps. Step one is a reproducible large-format plate: from Element Six/Orbray's reproducible three-inch process to four-inch development, progress here is observable and reasonably forecastable, because it is mainly plasma and thermal-field engineering. Step two is device-grade crystal quality: pushing the ~10⁷ cm⁻² dislocation densities in the open literature down to a level that sustains high-voltage device yield requires long iteration on stress management, nucleation layers, and thick-layer growth — historically this step never obeys linear extrapolation. Step three is n-type and interfaces: from the NIMS result of 150 cm²/V·s at 300°C to a room-temperature, enhancement-mode n-channel device with a reliable gate dielectric, this is widely expected to be the longest leg. The reasonable inference: if step one and thermal-bonding processes reach volume first, diamond enters the supply chain at scale as a passive component — heat spreader, bonded substrate, window — while large-scale substitution by diamond power devices becomes possible only once steps two and three converge together. Any claim that collapses all three steps into a single date should be discounted.

07For Investors and Buyers: Carat Output Is Not Electronic-Grade Capability

A scissor pattern is opening up on the supply side. Gem-grade lab-grown diamond is commoditizing fast: FY2025-26 figures released by India's Gem & Jewellery Export Promotion Council (GJEPC) on 15 April 2026 show polished lab-grown diamond exports of $1.13 billion, down 10.55% year over year (−6.52% in rupee terms) while volumes rose — the textbook shape of falling prices on rising units. Over the same period, gold jewellery studded with lab-grown diamonds grew 31.30% to $1,425.51 million, showing value migrating from loose stones to finished goods. Capacity is highly concentrated: China Daily reported in June 2024 that China accounts for roughly 95% of global lab-grown diamond output, with Henan contributing about 80% of the national total and a single CVD plant reaching an annual capacity of about 700,000 carats. Put those two datasets side by side and the point becomes clear: carats are a gem-grade metric, nearly uncorrelated with electronic-grade capability. Electronic-grade procurement and investment should look at a different scorecard — nitrogen and boron background (Type IIa or not), dislocation and stress distribution, thickness uniformity within and across plates, batch-to-batch consistency of plasma and chamber pressure, single-run yield rather than annual tonnage, and whether the supplier owns the reactor technology itself. That last item matters unusually much in this cycle: when yield is governed by growth-chamber reproducibility, a company that builds its own equipment has a shorter process-iteration loop and lower marginal cost of capacity — precisely the two most expensive things about crossing from gem-grade to electronic-grade.

08Where ENTASK Sits on This Path

ENTASK (Xi'an Entask Semiconductor Technology Co., Ltd., founded September 2020 in Xi'an, registered capital RMB 20 million) occupies a vertically integrated position on this path: it builds MPCVD equipment and produces diamond material. The Gen-4.5 and Gen-5 systems currently offered use the in-house E-MG010K microwave generator (2450 MHz, 10 kW/15 kW): the 10 kW configuration provides a 92 mm growth-area diameter with 12 mm crystal growth height, and the 15 kW configuration a 120 mm growth area, paired with the InsightAction and WiseAction intelligent control systems. Published stability metrics are MTBF above 10,000 hours, more than 1,500 hours of continuous full-load operation without failure, chamber-pressure stability of ±0.005 kPa, single-run yield of no less than 99%, monthly output of 200–300 ct per reactor, and a single-run record of 856 ct (January 2026). Mapped onto the scorecard above, three items are verifiable here: chamber-pressure and batch consistency, single-run yield, and in-house reactor development. On the material side there are three lines: rough (rectangular accounts for 90%; standard 20×15×9.5, 21×17×9.7 and 24×18×10 mm, non-standard 30×30×15 and 35×35×12 mm; D–E color, VVS–VS, Type IIa, IGI grading reports available); single-crystal optical windows from 20×20 to 45×45 mm at 0.5–1.2 mm custom thickness; and heat spreaders made exclusively as single crystal, never polycrystalline, at 1800–2500 W/m·K (supplier-stated range; measurement temperature and method per the accompanying report — note Element Six specifies >1900 W/m·K for single-crystal MCC and >2000 W/m·K for its top polycrystalline grade at 300 K, and values roll off with temperature) and 20×20 to 30×30 mm. The company holds ISO 9001:2015 certification and two published patent applications (published by CNIPA; not yet granted) — CN119269411A (360° stress analyzer for transparent crystals) and CN120178750A (industrial communication system for semiconductor equipment) — and supplies mainland China as well as India, the Middle East, Europe and the Americas. In this article's framing, that capability set maps onto the segments that have already landed: the single-crystal purity and consistency demanded by thermal management and optical windows, not the n-type layer demanded by power devices. (The ENTASK specifications and performance figures below are supplied by the company and have not been independently verified by a third-party test house; buyers should inspect the raw records on site.)

REFERENCES