Silicon has powered computer chips for decades, but engineers are increasingly running into the material’s physical limits. To keep making electronics smaller and more powerful, researchers are investigating ways to combine silicon with new ultrathin materials.
One promising group of materials is known as transition metal dichalcogenides (TMD). Among the leading candidates is molybdenum disulfide, a material just three atoms thick, consisting of a layer of molybdenum between two layers of sulfur.
Removing a Single Atomic Layer
For future transistors that combine silicon and TMD materials, manufacturers may need to selectively remove atoms from only the upper sulfur layer while leaving the underlying layers untouched.
One common way to remove surface atoms involves plasma, the energetic state of matter found in the Sun and other stars. Plasma research has also been a major focus at the U.S. Department of Energy’s (DOE) Princeton Plasma Physics Laboratory (PPPL) for the past 75 years.
Under carefully controlled conditions, particles within a plasma can strike the surface of a TMD material and knock atoms loose. The challenge is achieving enough energy to remove sulfur atoms from the top layer without harming the molybdenum layer beneath. Because the difference between success and damage is so small, developing a reliable process has proven difficult.
Using computer simulations, researchers found that treating molybdenum disulfide with oxygen or fluorine before plasma exposure can make the process much more controlled. Their findings were published in the Journal of Physical Chemistry Letters.
Oxygen and Fluorine Expand the Safety Margin
The simulations revealed that pretreatment dramatically lowers the energy required to remove sulfur atoms.
On an untreated surface, dislodging a sulfur atom requires about 30 electron volts. That threshold falls to roughly 10 electron volts when fluorine is added and about 14 electron volts when oxygen is used.
This difference is important because plasma ions do not all carry identical amounts of energy. Some have more energy than others. On an untreated surface, the range between removing sulfur atoms and damaging the molybdenum layer below is so narrow that some ions are likely to cause unwanted damage.
Lowering the sulfur removal threshold to 10 or 14 electron volts creates a wider operating window. As a result, manufacturers would have more flexibility to remove the top sulfur layer cleanly while preserving the rest of the material.
Letting Chemistry Do the Work
Rather than relying entirely on physical impacts to break atoms free, the researchers found a way to use chemistry to assist the process.
When an incoming ion strikes an oxygen-treated surface, two oxygen atoms can combine with a nearby sulfur atom to form sulfur dioxide, a stable gas that can naturally leave the surface. Fluorine behaves in a similar way, creating sulfur-fluorine compounds that are easier to remove.
“We are not directly breaking the bonds,” said Yury Polyachenko, a graduate student in chemistry at Princeton University who also worked at PPPL during the summer of 2025 and is the study’s lead author. “We are forming some intermediate products, such as sulfur dioxide. This intermediate product is much easier to break off.”
Expanding the Approach to Other Materials
The researchers plan to continue studying the technique to better understand its effects.
“The next step is figuring out how much damage the process causes, not just whether it causes damage,” Polyachenko said. “After that, we want to see whether the same approach works for related materials — swapping molybdenum for tungsten, or sulfur for selenium — to find out how broadly this idea can be applied.”
The research team also included Igor Kaganovich and Shoaib Khalid of PPPL, along with PPPL alumnus Yuri Barsukov.
The work was supported by DOE, the Office of Science, Fusion Energy Sciences and Basic Energy Sciences, as part of the Extreme Lithography & Materials Innovation Center, a Microelectronics Science Research Center, under contract number DEAC02-09CH11466.
The simulations were performed at the National Energy Research Scientific Computing Center (NERSC), a DOE Office of Science user facility at Lawrence Berkeley National Laboratory, operated under contract number DE-AC02-05CH11231. Additional computing resources included the Stellar, Della and Tiger clusters at Princeton University and NERSC award BES-ERCAP36136.