This Diamond Transistor Is Still Raw, But Its Future Looks Bright

Researchers in Japan have developed a diamond FET with high hole mobility

3 min read
Left, a diagram labelled diamond, graphite and hBN. Right, a grey microscope image with 6 orange points facing each other

A diamond substrate [gray], whose surface has been hydrogen-terminated, is laminated with a cleaved gate insulator [pink] of single-crystalline hexagonal boron nitride (h-BN). The graphite gate electrode, source electrode, and drain electrode are shown on top in black. Right: Optical microscopy image of the fabricated FET.

NIMS/Nature Electronics

Diamonds may be forever, wrote Ian Fleming, author of the James Bond novels. But a cynical engineer might add, “Yes, forever on the periphery of being practical as a material for semiconductors.” For despite the material’s upsides—a bandgap wider than competing silicon carbide and gallium nitride (GaN), excellent heat conduction, and the ability to operate at much higher temperatures and voltages than silicon—diamond’s downsides cloud much of the material’s sparkle.

Cost is an obvious obstacle. Compared to silicon, silicon carbide can be 30 to 40 times as expensive, and GaN between 650 to 1,300 times as expensive. Synthetically produced diamond materialfor semiconductor research has a price tag about 10,000 times that of silicon. Another issue is the small size of diamond wafers, with the largest commercially available size smaller than 10 square millimeters. Doping the material using ion implantation is difficult, and the material’s charge-carrier activation becomes less efficient at room temperature.

“Because of these disadvantages, use of diamond for semiconductor devices has been a challenge,” says Takahide Yamaguchi, principal researcher at Japan’s National Institute for Materials Science (NIMS) in Tsukuba (50 kilometers northeast of Tokyo) and the corresponding author of a paper published in December on a diamond field-effect transistor (FET) in Nature Electronics. “And though we haven’t overcome them, we have demonstrated promising results with the material that could facilitate the development of diamond devices for applications like low-loss power conversion and high-speed communications.”

Yamaguchi’s “promising results” refers to the development by a team of NIMS researchers, including himself, of a diamond FET with high hole mobility. This reduces conduction loss and boosts operational speed. In addition, the transistor exhibits normally-off behavior: The cessation of electric current flow through the device when the gate voltage is switched off. This makes it particularly suitable for fail-safe power electronic applications, says Yamaguchi.

Key to the researchers’ success is that they have been able to remove electron acceptors (impurities) from a diamond’s hydrogen-terminated surface. Hydrogen-termination covers the surface of the diamond with hydrogen atoms that bond to the outer carbon atoms. When the surface is exposed to air, it becomes electrically conducting because surface transfer doping is induced by adsorbed airborne acceptors.

“A number of R&D projects around the world have used hydrogen-terminated surface and surface transfer doping to create diamond FETs,” says Yamaguchi. “But all these devices have exhibited very low mobility of between just 1 to 10 percent of diamond’s original hole mobility, as well as normally-on behavior in many cases.”

The problem, he explains, is that surface transfer doping requires acceptor states on the diamond’s surface, but ionized acceptors cause carrier scattering that decreases hole mobility. Surface transfer doping also makes the design and fabrication of diamond FETs different from those ofstandard devices. Such issues have prevented device and circuit engineers from seriously considering implementing diamond FETs.

To overcome this obstacle, the NIMS team use single-crystalline hexagonal boron nitride (h-BN) for the gate insulator, instead of an oxide such as alumina, the go-to choice. Also, they have devised a new fabrication method that prevents the device being exposed to air. First, the researchers hydrogenate the diamond surface in a chemical vapor deposition chamber using hydrogen plasma. The substrate is then transferred in a vacuum suitcase to a glove box filled with argon gas, where it is laminated with a cleaved h-BN thin crystal.

The result is an FET “with hole mobility five times that of conventional FETs using oxide gate insulators, and more than twenty times that of GaN and SiC p-channel FETs,” says Yamaguchi. He adds that FETs with high hole mobility operate with lower resistance, which reduces conduction loss. “So a twenty-times increase in channel mobility means a one-twentieth reduction loss in the channel,” he explains.

That’s the good news. In its present form, however, the device is not ready for practical use. For example, it requires the addition of a drift layer so it can withstand high voltages. But adding such a layer increases conduction loss.

“Nevertheless, while the device isn’t ready yet for real applications, our research shows that acceptors are not necessary for inducing conductivity in hydrogen-terminated diamond, as was previously believed,” says Yamaguchi. “In fact, the reduction of the acceptor density even improves mobility and device performance.”

In addition, he says their results indicate that the development of diamond FETs is possible using standard designs. With further research, then, he believes they can improve device performance and devise a practical fabrication method for mass production. And, with Ian Fleming in mind, he adds that the announcement last October in Japan of the development of a mass production method for 2-inch diamond wafers suggests, “It won’t take forever for the remaining challenges to be solved.”

The Conversation (0)

Two Startups Are Bringing Fiber to the Processor

Avicena’s blue microLEDs are the dark horse in a race with Ayar Labs’ laser-based system

5 min read
Diffuse blue light shines from a patterned surface through a ring. A blue cable leads away from it.

Avicena’s microLED chiplets could one day link all the CPUs in a computer cluster together.

Avicena

If a CPU in Seoul sends a byte of data to a processor in Prague, the information covers most of the distance as light, zipping along with no resistance. But put both those processors on the same motherboard, and they’ll need to communicate over energy-sapping copper, which slow the communication speeds possible within computers. Two Silicon Valley startups, Avicena and Ayar Labs, are doing something about that longstanding limit. If they succeed in their attempts to finally bring optical fiber all the way to the processor, it might not just accelerate computing—it might also remake it.

Both companies are developing fiber-connected chiplets, small chips meant to share a high-bandwidth connection with CPUs and other data-hungry silicon in a shared package. They are each ramping up production in 2023, though it may be a couple of years before we see a computer on the market with either product.

Keep Reading ↓Show less