Molybdenum disulfide, a two dimensional semiconductor that’s just 3 atoms thick, has had a big year. In October, a group of researchers made a 1-nanometer transistor from the material, showing that even if silicon transistors stop shrinking, the new material might provide a path forward. In December, at the IEEE International Electron Devices Meeting in San Francisco, researchers presented work they say shows that molybdenum disulfide not only makes for superlative single transistors, but can be made into complex circuits using realistic manufacturing methods.
At the meeting, a group from Stanford showed that transistors made from large sheets of MoS2 can be used to make transistors with 10-nanometer-long, gate having electronic properties that approach the material’s theoretical limits. The devices displayed traits close to ballistic conduction, a state of very low electrical resistance that allows the unimpeded flow of charge over relatively long distances—a phenomenon that should lead to speedy circuits. Separately, a team from MIT demonstrated complex circuit elements made from MoS2 transistors.
Most of the work on molybdenum disulfide so far has been what Stanford electrical engineer Eric Pop calls “Powerpoint devices.” These one-off devices, made by hand in the lab, have terrific performance that looks great in a slide. This step is an important one, says Pop, but the 2D material is now maturing.
The Stanford lab’s transistors are not as small as the record-breaking ones demonstrated in October. What’s significant, says group leader Pop, is that these latest transistors maintained similar performance even though they were made using more industrial-type techniques. Instead of using Scotch tape to peel off a layer of molybdenum disulfide from a rock of the material, then carefully placing it down and crafting one transistor at a time, Pop’s grad student started by growing a large sheet of the material on a wafer of silicon.
In a transistor, a gate electrode switches the semiconductor channel between conducting and insulating states. In the Stanford device, the tricky part was coming up with an easy way to make a small gate atop the molybdenum disulfide without harming it, says Pop. That is, until his student, Christopher English, realized they could harness the power of rust. English chose a somewhat unusual material, aluminum, to serve as the gate electrode. He deposited a 20-nanometer finger of aluminum on the molybdenum, then allowed it to oxidize and shrink down to a smaller size. The gate ends up being about 10 nanometers.
At these relatively small dimensions, the molybdenum disulfide transistors approach their ultimate electrical limit, a state called ballistic conduction. When a device is small enough (or at low enough temperature), electrons will travel through the conducting medium without scattering because of collisions with the atoms that make up the material. Transistors operating ballistically should switch very fast and enable high-performance processors. Pop estimates that about 1 in 5 electrons moves though the rusty transistors ballistically. By further improving the quality of the material (or making the transistors smaller), he expects that ratio to improve. The important thing, he says, is the way they achieved this: using methods that could translate to larger scales. “We have all the ingredients we need to scale this up,” says Pop.
Zippy nanoscale transistors are great on their own, but they’re useful only if you can build them into circuits. Researchers from MIT demonstrated just that by constructing working registers and latches. They managed the feat, says electrical engineer Dina El-Damak, by creating computer-aided design software tailored to MoS2. This sort of software is common in the silicon world and enables designers to come up with new circuits relatively easily. (El-Damak worked on the molybdenum disulfide project at MIT and is now a professor at the University of Southern California in Los Angeles.)
Since molybdenum disulfide is so new, not many circuit designers have worked with the material. So far, most work has been done by trial and error, one device at a time. The MIT group can create an informed circuit design, using their computer models to simulate the best and worst cases, based on the material’s known properties and the performance of previous devices, says El-Damak. Then the group fabricates the design that seems most likely to work, tests its performance, and feeds the results back into the program. “By doing this, we have more confidence in scaling up this technology,” she says.
Both Pop and El-Damak say molybdenum disulfide is unlikely to be a direct replacement for silicon. The material will either be used to build complementary systems on top of silicon chips, or it will be used on its own in flexible, transparent electronics. It’s also possible that some other 2D semiconductor will end up being a better option. Molybdenum disulfide is a few steps ahead because researchers have worked with it more than, say, tungsten selenide, and know how to grow the material over large areas.
The Stanford and MIT research demonstrates important progress in this field, says Deji Akinwande, an electrical engineer at the University of Texas at Austin who co-chaired the IEDM session on 1D and 2D devices. People who work in industry are always asking when these materials will be made into useful circuits, and now it’s happening, he says. “Industry is starting to take this more seriously, now that it’s no longer just the grad student in the basement working on it,” he says.