Hooking up wiring to 2D semiconductors

Two sheets of molybdenum disulfide show why it's so hard to find any place to put impurities of any sort inside what's essentially a single molecule.

As chip manufacturers are getting close to the limits of their ability to scale features down, materials scientists are working hard to provide the raw materials that would let us build circuitry from individual molecules, such as carbon nanotubes. One class of molecules that may find a home in future chips is sheets of material that are a single atom thin.

Although graphene, a sheet of carbon atoms, tends to attract the most attention, there are actually a variety of atomically thin materials. And, while graphene is not normally semiconducting, a number of the alternatives are. One of these alternatives, molybdenum disulphide (MoS2), has already been used to create functional electronics. Unfortunately, the performance of these circuits has been erratic. Now, a collaboration of researchers at Rutgers University and Los Alamos National Lab has figured out why: hooking up wires to an atomically thin material is really hard. Fortunately, they've also figured out a solution.

Although MoS2 appears to have what it takes to make great circuitry, early attempts at using it have been inconsistent. As the authors of the new paper note, the mobility values (a measure of how quickly electrons move through the circuit) reported for these circuits can vary by as much as a factor of 400. The problem, the authors suspected, comes from wiring up the circuits. Although it's easy to deposit metal on top of an atomically thin material like MoS2, it's another thing entirely to make sure electrons can easily hop across that junction.

In regular circuitry, a mismatch between the wiring and a semiconductor is usually handled by adding a small amount of an impurity to the semiconductor. This process, known as doping, brings the semiconductor's properties a bit closer to those of the wiring, making for a match that lets electrons flow with minimal hassle. The problem comes about when you attempt to add impurities to a material that's only an atom or two thickit simply doesn't work out.

But doping isn't the only way to change the properties of an atomically thin semiconductor. Since these are essentially single molecules, it's possible to change their properties by altering the molecular structure. As it turns out, MoS2 can come in two forms, or phases, that are distinguished by slightly different arrangements of the bonds extending from the metal. One of these (the 2H phase) is the semiconductor that circuitry depends on. But the other, called the 1T phase, behaves more like a metal.

The secret to the new work is that the authors have figured out a chemical process that can convert between the two. Exposure to an organic-lithium compound will cause local conversion of the semiconducting form to the metallic one. As long as the researchers can control what parts of the MoS2 get exposed to the chemical, they can create a pattern of metallic material on the semiconductor. After that, it's simply a matter of hooking up the wiring exclusively to the metallic parts.

This allows electrons to readily jump from the wiring and into the MoS2 sheet before they come in contact with the semiconducting portion of the circuitry. The authors built a series of field-effect transistors using this approach and show that all measures of their performance are significantly better than devices that rely only on the semiconducting form of MoS2. In addition, the performance was consistent; they made 25 devices of each type and showed their behavior was all very similar.

That's the good news. There are a number of caveats that the authors helpfully point out, the most significant of which is that they're not especially good at limiting the chemical treatment to only those areas where they want to hook up wiring. The other issue is that the metallic1T phase is what's called "metastable"it can convert back to the semiconductor under the right conditions. And, as of yet, we don't know what those conditions are or whether they're likely to pop up during normal operations of a chip.

Nevertheless, the authors do a nice job of identifying a problem and showing that it can be solved. And, more generally, the work does a nice job of showing why it can take a while to go from a material breakthrough to an actual product. While MoS2 may have all the properties of a wonder-semiconductor, putting it into an actual product involves sweating a tremendous number of details (including attaching wiring to it). It's the less glamorous detail work that follows discoveries that actually allows us to do something with the discovery.

See the article here:
Hooking up wiring to 2D semiconductors