Forget silicon. Diamond may be the next semiconductor material

Timing is everything. Technologies, like artificial intelligence, electric vehicles, and quantum computing, that demand unprecedented performance from semiconductors have exposed silicon’s shortcomings. Simultaneously, after decades in development, processes for making single-crystal diamond semiconductors—which can outperform silicon as a computer chip material—are poised for large-scale rollout.

About a dozen companies are developing semiconductor wafers, the basic material on which chips are built, made of synthetic diamond. The challenge for these firms is that their product is now orders of magnitude more expensive to make than its silicon counterpart. But they are inching their way toward market acceptance by reducing costs. If successful, they could usher in a new age of high-performing diamond electronics.

Silicon semiconductors—the workhorse of the electronics industry—are limited by intolerance to high temperatures, inefficient handling of high-voltage current, slow processing speed, and large chip size. These shortcomings have pushed some semiconductor fabricators to build chips on novel but more-expensive wafers made of materials such as gallium arsenide, gallium nitride, and silicon carbide. But semiconductors built on diamond outperform them all on all key metrics, developers say.

The start-up Diamfab is among the companies striving to develop a diamond semiconductor production process that is affordable. And it is making progress. “We did it,” says Gauthier Chicot, CEO and cofounder of the French firm, referring to its opening earlier this year of a pilot plant in Grenoble, France, that creates diamond wafers using microwave plasma chemical vapor deposition (MPCVD). “We are proud to have inaugurated the first European industrial site dedicated to semiconductor diamonds,” he says.

Diamfab is built on 3 decades of technology development at the French National Center for Scientific Research (CNRS) and the University of Grenoble Alpes. The technology was transferred to Diamfab in 2019, when Chicot and his colleagues on the project cofounded the company. “I was lucky because I arrived at CNRS at the moment when the technology was mature enough to start thinking about creating a start-up,” Chicot says.

Most companies making diamond semiconductor wafers use a form of MPCVD. Diamfab begins its process by placing a seed diamond crystal in a reactor, into which it introduces a gas mixture featuring methane—a carbon source—and hydrogen at low pressure. Microwaves then break the molecules apart to form a plasma, which releases carbon atoms that arrange themselves on the surface of the seed, eventually forming a wafer. Diamfab’s customers add transistors, circuitry, and other features to the wafer surface to create computer chips.

Diamond on its own is not a conductor, so Diamfab introduces boron, nitrogen, or phosphorus in addition to the methane. These elements replace some carbon atoms on the diamond surface, creating a defect in the crystal. “This defect will allow the diamond to become a semiconductor,” Chicot says. “Otherwise, pure diamond is insulating.”

Piggybacking on technology for making cosmetic diamonds

Costs for diamond-wafer makers have fallen in recent years thanks in part to a substantial drop in the price of MPCVD reactors, which have become widely available to produce synthetic diamonds for jewelry. Diamfab buys reactors for jewelry-making off the shelf and modifies them, Chicot says.

To further cut processing costs, Diamfab wants to make bigger wafers in each production batch. Other potential cost reduction techniques involve using bigger reactors and adding more diamond seeds per reactor. Variants of chemical vapor deposition, such as laser-plasma CVD, could accelerate diamond growth rates compared with MPCVD.

Diamfab says early data coming out of its pilot facility indicate that the technology will work efficiently at large scale. A handful of other companies are also advancing their own MPCVD processes, and some are already planning or have begun operating their first commercial plants.

Diamond Foundry, which claims to be the first firm to have made a single-crystal diamond wafer, has an aggressive commercialization plan. It opened a commercial plant in Trujillo, Spain, in 2025 and is gradually scaling production. “In 100 years, no-one will remember what Google was but all tech will be based on diamond electronics,” the company predicts on LinkedIn.

Meanwhile, in May, Ookuma Diamond Device completed construction of what it claims is the “world’s first diamond semiconductor factory.” Ookuma is targeting applications that are difficult for conventional semiconductors to handle, such as environments with extreme temperatures or high levels of radiation. It is no coincidence that the company’s plant is in Fukushima, Japan, the site of a nuclear power plant disaster in 2011.

Tailoring recipes to specific electronics applications

Developers of diamond semiconductor wafers tout the ability to customize their products for different applications. “In some ways the process of making single-crystal diamond wafers is like cooking,” Chicot says. For example, adding nitrogen to the reactor ensures that a nitrogen atom displaces a carbon atom in the diamond lattice, creating what is known as a nitrogen vacancy center. This center is a quantum system at room temperature, making it suitable for quantum computers, Chicot says. In contrast, silicon semiconductors must be cryogenically frozen to become quantum.

Diamond semiconductors will eliminate the need for expensive and bulky cooling equipment in quantum computers, allowing them to be smaller, Chicot says. Although diamond is more expensive than other semiconductor materials, its superior performance means it is already attractive to quantum-sensing and quantum-computing companies.

“We’re enthusiastic about the potential of synthetic diamond as a next-generation replacement for silicon in semiconductors,” says Shahar Keinan, CEO of the PolarisQB, a start-up that applies quantum technology to drug design.

The use of quantum computing in chemistry, materials science, and drug discovery is maturing rapidly. “As a hardware-agnostic company, we’re excited by how the emergence of synthetic diamond semiconductors could accelerate the future of quantum-driven discovery,” Keinan says. “This breakthrough could unlock major advantages for quantum computing.”

While quantum computing is still largely in development, one nascent market for diamond semiconductors is in quantum sensors used to measure magnetic fields. Such quantum systems have lots of uses, including measuring current flowing through a cable or the magnetic field of the earth, Chicot says. Among early adopters, the Swiss firm Qnami is using diamond quantum chips in scanning magnetometry microscopes and quantum imaging probes for noninvasive imaging at atomic and nanoscale.

EVs and data centers can benefit from diamond semiconductors

The potential market for diamond semiconductors in electric vehicles (EVs) is far bigger. In EVs, an inverter is required to transfer huge voltages from the battery to the motor. At such high voltages, silicon semiconductors are perhaps 90% efficient, so about 10% of the energy in an EV battery is wasted, Chicot says. Tesla was the first company to use silicon carbide semiconductors, which upped efficiency to about 95%. A diamond semiconductor’s energy transfer efficiency could be 99%, he says.

One reason for the greater efficiency in EVs is heat tolerance and dissipation. “Diamond is an even better thermal conductor than copper, for instance,” Chicot says. Consequently, the size and weight of the heat dissipation system can be cut by a factor of three compared with silicon carbide, he adds. “For tomorrow’s electric airplane this will be even more important than for cars.”

Another potential market for diamond semiconductors is managing power delivery for AI data centers, where silicon carbide chips are now used. Power electronics based on diamond semiconductors have a far smaller footprint than their silicon carbide–based counterparts because they can carry between 4 and 50 times as many diodes or transistors on the same surface, Chicot says.

Can Bayram, a professor at the University of Illinois Urbana-Champaign, considers diamond semiconductors necessary for a more reliable and resilient electricity grid. “To meet those electricity demands and modernize the electrical grid, it’s very important that we move away from conventional materials, like silicon, to the new materials that we are seeing being adopted today like silicon carbide and the next generation of semiconductors—ultra-wide bandgap materials—such as aluminum nitride, diamond and related compounds,” he says in a press release after the publication of research in 2023 that demonstrated the performance of diamond semiconductors.

Chicot says Diamfab is focused on such a future. The company expects to keep lowering manufacturing costs so that by 2030 it can start supplying wafers to producers of mass-market power electronics. Other diamond-wafer makers, such as Ookuma, expect to crank up commercial production sooner.

Even if diamond continues to be more expensive than semiconductor materials such as silicon carbide and gallium nitride, Chicot seems confident that diamond’s added value in operation will prevail. “Ultimately, the value and performance that the material will bring to the devices—in power electronics—is huge,” he says.

Alex Scott is a C&EN business reporter covering all chemistry-related issues in Europe.