A team of researchers from Korea and Japan have had a breakthrough with a semiconductor material that they claim could be a candidate to replace silicon in future electronics. In the 7 August issue of Science they report the creation of a transistor where the channel consists of layers of a two-dimensional material molybdenum ditelluride (MoTe2).
Everyone’s on the lookout for a replacement for silicon, because, great as it is, it has two drawbacks: its electronic properties degrade when the silicon layer is thinned to just one or a few atoms thick, and its indirect bandgap make it difficult to use silicon in optoelectronics, explains Heejun Yang, a physicist at Sungkyunkwan University, Suwon, Korea, who led the research.
One class of candidate materials that researchers have been examining are transition metal dichacogenides, which include molybdenum disulfide and tungsten diselenide. But making good electrical connections between these materials and metal contacts has been a problem.
Molybdenum ditelluride, synthesized for the first time in the 1960s was little investigated because it could not be obtained in a pure form that could be used in electronics. Last year things changed when Yang and his colleagues at Sungkyunkwan University obtained a very pure form of molybdenum ditelluride, and made a surprising discovery. "I was studying MoTe2 and we were using a technique called Raman spectroscopy. Each time we measured we obtained results that were different. I got quite disappointed because I really didn’t understand why. As we have a solid, it should have produced the same Raman spectrum for each measurement," says Yang.
A laser can both thin out a sample of molybdenum ditelluride, and change the top layer from semiconducting (2H) to metallic (1T’).Illustration: Sungkyunkwan University
However, the scientists quickly realized that they were onto something new. In his spectroscopy experiments, Yang had control over the laser’s wavelength and power. "What we found is that the laser increases the temperature up to 400 degrees Celsius locally, and that because tellurium atoms can be easily sublimated, you have a thinning effect of the MoTe2 sample," says Yang. “I realized that the different results were caused by the resulting changing thickness of the samples, and the presence of different structural phases."
This observation gave them a way to reduce the number of layers in the material. (But because the substrate, silicon, that carried the sample absorbed too much heat, they could not create a single monolayer.)
The researchers found that laser irradiation, besides thinning the sample, also resulted in two structural phases of the material: One phase, called 2H (for “hexagonal”), is semiconducting, and the other phase, 1T', is metallic. What’s more, the researchers found that parts of a 2H top layer of a MoTe2 sample could be changed by a laser pulse into a 1T' metallic layer.
The prototype transistor had a channel of semiconducting MoTe2 about 2 micrometers long. At both ends of the channel, the semiconducting top layer was converted to a metallic phase by irradiation with a laser. "Once our material changes into a metallic state, we can deposit gold or aluminum junctions, which is very good for a high speed device," says Yang. So, in fact, the electrical source and drain junctions for the transistor were not outside of the channel material, but were part of it—a novelty in 2-D transistor design. "This is the first realization of a metal-semiconductor junction" in the same 2-D material, says Yang. This so-called homojunction between the two phases is very efficient and allows a charges to move swiftly—10 to 50 times higher than silicon, says Yang.
Another advantage is that the MoTe2 semiconducting material has practically the same bandgap as silicon—1 electron volt a versus silicon’s 1.1 eV. That similarity might help ease the replacement of silicon transistors by MoTe2 transistors in the design of logic circuits.
A prototype transistor with a single layer channel is next for Yang. But MoTe2 doesn’t seem to yield monolayers easily. Unlike single atomic-layer graphene, which can be produced easily by exfoliating graphite, they will have to look for another way. "The layer to layer interaction is weak for graphene. Unfortunately, in MoTe2 the layer by layer interaction is quite strong. Students have tried “scotching,” but we couldn't get monolayers," says Yang.
However, among 2-D semiconductors, MoTe2 might be worth the extra effort. "The low bandgap compared to that of other two-dimensional semiconductors is a promising point," he says.