Scientists at MIT, along with researchers from IBM, the University of California at Los Angeles, and Kyungpook National University in South Korea, have found a way to produce graphene with fewer wrinkles, and to iron out the wrinkles that do appear. The team reports that the techniques successfully produce wafer-scale, "single-domain" graphene - single layers of graphene that are uniform in both atomic arrangement and electronic performance.
After fabricating and then flattening out the graphene, the researchers tested its electrical conductivity. They found each wafer exhibited uniform performance, meaning that electrons flowed freely across each wafer, at similar speeds, even across previously wrinkled regions.
This is an important step as graphene will need to be single-domain in order to become a widespread semiconductor material for industry, since making millions of devices on it will require the performance of the devices to be the same in every location.
A normal CVD process often produces relatively large, macroscropic wrinkles in graphene, due to the properties of the underlying copper substrate and the process of pulling the graphene out from the acid. The alignment of carbon atoms is not uniform across the graphene, creating a "polycrystalline" state in which graphene resembles an uneven, patchwork terrain, preventing electrons from flowing at uniform rates. Back in 2013, while working at IBM, members of the team developed a method to fabricate wafers of single-crystalline graphene, in which the orientation of carbon atoms is exactly the same throughout a wafer - using a silicon carbide wafer with an atomically smooth surface rather than using CVD. The material did have, however, tiny step-like wrinkles on the order of several nanometers. The researchers then used a thin sheet of nickel to peel off the topmost graphene from the silicon carbide wafer, in a process called layer-resolved graphene transfer.
In their recent work, the team discovered that the layer-resolved graphene transfer irons out the steps and tiny wrinkles in silicon carbide-fabricated graphene. Prior to transferring the layer of graphene onto a silicon wafer, the team oxidized the silicon, creating a layer of silicon dioxide that naturally exhibits electrostatic charges. When the researchers then deposited the graphene, the silicon dioxide effectively pulled graphene's carbon atoms down onto the wafer, flattening out its steps and wrinkles. Unfortunately, the team says this ironing method would not work on CVD-fabricated graphene, as the wrinkles generated through CVD are much larger, on the order of several microns.
To test whether the flattened, single-crystalline graphene wafers were single-domain, the researchers fabricated tiny transistors on multiple sites on each wafer, including across previously wrinkled regions. "We measured electron mobility throughout the wafers, and their performance was comparable," the team says. "What's more, this mobility in ironed graphene is two times faster. So now we really have single-domain graphene, and its electrical quality is much higher [than graphene-attached silicon carbide]."
While there are still challenges to overcome before graphene could be used in commercial electronics, the group's results give researchers a preliminary path to reliably manufacture pristine, single-domain, wrinkle-free graphene at wafer scale.