HWNEWS
Researchers say a simple chemical tweak fixes the contact problem that has held back one half of the atom thick transistor family most likely to follow silicon.
Silicon has been the channel material in the smallest transistors on the market for most of the chip era. The smallest transistors in research have used semiconductors just one atom thick. The atom-thick branch has always been missing one reliable half. A four-lab collaboration has now named the specific physical reason, and demonstrated a specific chemical fix.
In a paper published 01 October 2026 in Nature Communications, researchers from Huazhong University of Science and Technology, Hong Kong Polytechnic University, UC Santa Barbara, and the National University of Singapore report a high-performance p-type transistor built from a single layer of tungsten diselenide, a 2D material. The fix is chemical, not exotic. An oxygen treatment suppresses the atomic defects that have blocked the p-type half of the 2D family for years, and the contact region ends up heavily doped with positive carriers. The measured contact resistance falls from 1,869 to 104 ohm-micrometers, an 18× drop, and a 16-nanometer-channel device runs in a transport regime that is usually reserved for textbook diagrams.
The work is the kind of advance that matters because it is falsifiable, not because it is final. It is a single paper, from a single materials system, at a single channel length. The numbers are real, and the physics is named. Whether the result survives at wafer scale and inside a real complementary logic stack is a different question, and a longer one.
Every transistor has two ends: a channel where the current flows, and metal contacts that feed current in and out. The contact is the smaller structure. It is also the one that breaks first. When a metal is laid down on a 2D semiconductor like tungsten diselenide, the contact region's electronic energy gets pinned at a value the metal imposes, a phenomenon called metal-induced gap states, or MIGS. The result is a large energy barrier at the contact, and a large resistance that does not get better just by making the channel shorter.
The conventional workaround on the silicon side is heavy doping at the contact, deliberately introducing charged atoms that bend the energy bands in the semiconductor and let current tunnel through. Tungsten diselenide, a layered material that can be peeled or grown down to a single molecular layer, has resisted that trick. The two most common defects in grown WSe2 are selenium vacancies, missing selenium atoms in the lattice, and those vacancies trap the dopants before they can do their job.
The new paper uses oxygen as the dopant instead. Oxygen fills the selenium vacancies, the channel is left with fewer electron traps, and the contact region ends up heavily doped with positive carriers. The MIGS pinning still happens, but now the heavy doping in the contact is strong enough to bend the energy bands past the pinning and admit current.
The result is a p-type field-effect transistor with a 16-nanometer channel. Hole mobility, the speed at which the positive carriers move under an applied field, rises from 55 to 218.3 square centimeters per volt-second. The paper describes that figure as approaching the theoretical limit for monolayer tungsten diselenide.
Contact resistance drops from 1,869 to 104 ohm-micrometers. Saturation current, the maximum current the device can push per micrometer of channel width, reaches 1,635 microamps per micrometer at a drain-source voltage of -1.2 V. Most of the carriers, 81 percent, cross the 16-nanometer channel without scattering, a regime called ballistic transport, at room temperature. Ballistic transport at room temperature is what textbooks point at when they describe the upper limit of how fast a transistor can switch. Reaching 81 percent of it is a real materials result, not a marketing line.
As relayed by Semiengineering, the paper is also a stress test on a specific bet. The semiconductor industry has been told for years that sub-1-nanometer complementary metal-oxide-semiconductor, the dominant chip architecture, will need new channel materials once silicon stops scaling. Tungsten diselenide and the broader family of 2D semiconductors are the most credible candidates. The catch has always been that the n-type half, the devices carrying negative electrons, was relatively easy to build, and the p-type half was not. Without both halves in the same materials family, the path to a 2D complementary logic stack does not open.
This paper does not open it yet. The advance is bounded in three concrete ways.
The result is a research demonstration on a single channel length, 16 nanometers, with one doping chemistry, oxygen, on one materials system, monolayer tungsten diselenide. Independent groups have not yet reproduced a contact resistance of 104 ohm-micrometers in this materials system, and the paper does not include wafer-scale data, yield numbers, or any signal of a process that can be ported from a cleanroom at HUST to a foundry line at a major chipmaker.
The complementary n-type half is still its own problem. A 2D complementary metal-oxide-semiconductor stack needs p-type and n-type devices with comparable performance, on the same substrate, with compatible processing. The new paper does not address the n-type side at all.
And the integration question is open. Even if monolayer tungsten diselenide can be grown uniformly across a 300-millimeter wafer, the rest of the manufacturing flow, the gate dielectric, the interconnect, the packaging, is currently built for silicon. None of that is the paper's job to solve, and the authors do not claim to.
The next credible milestones are concrete: wafer-scale monolayer WSe2 growth with controlled defect density, an n-type 2D device at comparable performance, and a working complementary logic cell that uses both. Until those land, the result is a named mechanism, a real number, and a four-lab collaboration that has narrowed the field. That is the right size of progress for one paper.