A single 2D crystal becomes a transistor, in the lab

For more than two decades, researchers have hoped that atomically thin semiconductors could take over from silicon as devices shrink. The obstacle was rarely the semiconductor itself. It was everything around it. At the Institute for Basic Science (IBS), a team led by Jo Moon-Ho, director of the Center for Van der Waals Quantum Solids, has now built a transistor starting from a single layered crystal, converting selected layers into the insulator and contacts the device needs. The work is published in the journal Science.

A transistor needs three different materials: a semiconductor to carry current, an insulating layer to control it, and metal contacts through which current enters and leaves. The surface of a 2D semiconductor offers few sites where other materials can attach, so a uniform insulator is hard to deposit, and depositing metal directly can damage the thin layer or create barriers that reduce performance. Peeling and stacking separate 2D materials one by one works in the lab but is hard to scale over large areas.

The IBS team used the crystal itself as the only starting material. They began with tungsten diselenide (WSe₂), whose electrical behavior changes when some of the atoms around the tungsten layer are replaced. The upper layers were oxidized into an insulating tungsten oxide, which sits directly above the channel and serves as the gate dielectric, with an effective thickness below 1 nanometer. Other regions were converted into tungsten ditelluride (WTe₂), a semimetal used for the contacts, while unconverted regions kept their semiconducting properties. The reaction can be stopped after a chosen number of layers, at roughly 0.7 nanometer per WSe₂ monolayer.

In the test devices, a change of only 61.3 millivolts in gate voltage increased the current tenfold, close to the theoretical room-temperature limit of about 60 millivolts per decade, and the on/off current ratio exceeded 10 million. The WTe₂ contacts reduced the barrier for charge injection to about one-third of the barrier seen with direct platinum contacts. The team also built n-type MoS₂ transistors using the same oxide, then stacked p-type and n-type transistors into a complementary inverter, the basic building block of digital logic, and showed that the voltage inverted as expected.

For chip designers, the promise is a route to what Jo calls ultralow-power, ultrascaled and vertically integrated semiconductor devices. The first gain would fall to researchers working on 2D electronics, because the method avoids the peeling and stacking step that the team describes as poorly suited to mass production. The study gives no timeline, no production process and no large-area test. Its results come from individual devices built in the lab, and turning them into manufactured circuits remains a separate step.