Skip to main content Accessibility help
×

New atomic layer deposition method yields large crystalline 2D MoS2 thin films

By Prachi Patel February 14, 2018
New atomic layer deposition
Scanning electron microscopy (SEM) image of a 45-nm-thick molybdenum disulfide film deposited using atomic layer deposition (left) and then annealed (right). The inset shows a higher resolution SEM image of the film. Credit: J. Vac. Sci. and Technol.

Sheets of molybdenum disulfide (MoS2) just a few atoms thick hold promise for high-performance, flexible electronics as well as optical applications. But one obstacle the two-dimensional (2D) material faces is the lack of an efficient method to make it in large quantities. Researchers at Argonne National Laboratory have now demonstrated that the atomic layer deposition method could be used to make uniform, crystalline MoS2 thin films as large as a standard 300 mm silicon wafer.

Ultrathin 2D materials, of which graphene is the best known and studied, have electronic properties that surpass silicon’s. But MoS2 has an edge over graphene: it can be made semiconducting. Researchers are now developing applications for the material. But it is still made in the laboratory by peeling flakes off MoS2 crystals.

Some scientists have made thin MoS2 films using chemical vapor deposition (CVD). This involves exposing a hot substrate to a mixture of Mo- and S-containing gases under suitable conditions. But the method does not allow control over the film’s thickness and composition, says chemist Jeffrey Elam. So he and his colleagues at Argonne used atomic layer deposition, a form of chemical deposition in which films are grown layer-by-layer. The method holds the most potential to make uniform monolayer films over large areas, Elam says. It is already used commercially to coat 300 mm silicon substrates with materials like hafnium oxide and titanium nitride.

The researchers chose the novel precursors molybdenum hexafluoride and hydrogen sulfide, which react to make films on a silica substrate at 200°C. X-ray photoelectron spectroscopy measurements confirmed that the deposited films were MoS2. However, Raman spectroscopy and x-ray diffraction measurements failed to identify crystalline MoS2 in the films, and this might be due to unreacted MoFx residues in the films, the researchers say. Annealing the film at 350°C reduced the fluorine concentration and gave high-quality crystalline MoS2.

The researchers were able to make 300-nm-long strips of MoS2 thin films of thicknesses ranging from 12 nm to 60 nm, they report in a recent issue of the Journal of Vacuum Science & Technology A. A single MoS2 layer is around 6.5 Å thick; therefore, a 12 nm film would have about 18 layers. “We’ve demonstrated that our process should be scalable to coating a full-size 300 nm wafer,” Elam says.

While others have made MoS2 films with atomic layer deposition using different precursors, the new precursor combination has an advantage. The reaction between the gases generates heat, Elam says. Tapping this heat to get rid of defects and impurities could reduce the need for annealing, which would help integrate the process into commercial device manufacturing. “When you’re making electronics, there might be materials underneath that won’t last through annealing,” he says. “Ideally you want a process that works at low temperature, say 200°C.”

It would, of course, be desirable to avoid the annealing process altogether, says Tobin Marks, professor of chemistry and materials science and engineering at Northwestern University. But he adds that “the quality of the growth and product characterization are excellent.”

However, a lot more work remains, says Andrey Veovodin, professor of materials science and engineering at the University of North Texas. At 12 nm, the film is too thick to have “the game-changing benefit of 2D confinement,” he says. Plus, the optically measured bandgap that the researchers report is 1.3 eV, while it should be 1.8 eV for a high-quality 2D MoS2 film. “I hope such efforts will continue as ALD has very good potential for 2D large-area growth,” he says. “It’s just too early to claim a victory.”

Read the article in the Journal of Vacuum Science & Technology A.