By Jamie Oberdick
Researchers from the National Science Foundation-sponsored Two-Dimensional Crystal Consortium (2DCC-MIP) - Materials Innovation Platform may have come up with a solution for a bottleneck that has confounded researchers trying to develop high-quality 2D semiconductors for next generation electronics such as Internet of Things (IoT) and artificial intelligence.
Tungsten diselenide is a semiconductor that holds promise as a material for nanosheet transistors, which are electronic devices that use atomically-thin layers of material to control and manipulate electrical current flow. Such research has gained importance since last year’s passing of the CHIPS and Science Act, which is designed to boost America’s efforts to onshore the production and development of semiconductor technology.
While it would make for a highly effective semiconductor, synthesizing a single-layer sheet of tungsten diselenide that is three atoms thick over sapphire wafer areas as large as 8 to 12 inches in diameter has proven to be a challenge.
This is due to a defect in the material known as “mirror twins.” Mirror twin boundaries form from opposite-oriented tungsten diselenide crystals on the sapphire wafers and are so named because the arrangement of atoms on one side of the boundary is the mirror opposite of the other side of the boundary. This defect scatters electrons as they move through the 2D layer, which in turn reduces the performance of the nanosheet transistor.
In the study published in Nature Nanotechnology, the researchers turned to metal organic chemical vapor deposition (MOCVD), a technology that is used to deposit ultra-thin, single crystal layers onto a substrate, in this case a sapphire wafer. 2DCC-MIP researchers pioneered the use of this technique for the synthesis of wafer-scale transition metal dichalcogenides like tungsten diselenide.
“To achieve single-layer sheets with a high degree of crystalline perfection, we used sapphire wafers as a template to align the tungsten diselenide crystals as they deposit by MOCVD on the wafer surface,” said Joan Redwing, director of the 2DCC-MIP, distinguished professor of materials science and engineering and electrical engineering and lead author of the study. “However, the tungsten diselenide crystals can align in opposite directions on the sapphire substrate. As the oppositely oriented crystals grow larger in size, they ultimately meet up with one another on the sapphire surface to form the mirror twin boundary.”
To solve this issue and get most of the tungsten diselenide crystals to align with the sapphire crystals, the researchers took advantage of “steps” on the sapphire surface. The sapphire single crystal that makes up the water is highly perfect in physics terms; however, it is not perfectly flat at the atomic level. There are steps on the surface that are a mere atom or two tall with flat areas between each step. The researchers made an interesting and significant discovery about these steps.
The step on the sapphire crystal surface is where the tungsten diselenide crystals tended, but not always, to attach and the crystal alignment when attached to the steps tended to be in all one direction.
“If the crystals can all be aligned in the same direction, then mirror twin defects in the layer will be reduced or even eliminated,” Redwing said.
The researchers found that by controlling the MOCVD process conditions, most of the crystals could be made to attach to the sapphire at the steps. And during the experiments, they made a bonus discovery: If the crystals attach at the top of the step, they align in one crystallographic direction, if they attach at the bottom, they align in the opposite direction.
“In experimental work performed by Dr. Haoyue Zhu, a postdoctoral scholar in the 2DCC-MIP and Dr. Tanushree Choudhury, an assistant research professor in the 2DCC-MIP, we found that it was possible to get the majority of the crystals to attach at either the top or the bottom edge of the steps,” Redwing said. “This would provide a way to significantly reduce the number of mirror twin boundaries in the layers.”
This in turn led to researchers in the 2DCC-MIP Theory/Simulation/Data facility lead by Dr. Nadire Nayir, a postdoctoral scholar in Prof. Adri van Duin’s group, to develop a theoretical model of the atomic structure of sapphire surface to explain why the tungsten diselenide attached to the top or bottom edge of the steps. If the surface of the sapphire was covered with selenium atoms, then they would attach to the bottom edge of the steps, if the sapphire is only partially covered so that the bottom edge of the step lacks selenium atom, then the crystals attached to the top.
To confirm this theory, the Penn State 2DCC-MIP researchers worked with Krystal York, a graduate student in Prof. Steven Durbin’s group at Western Michigan University, who contributed to the study while a visiting scholar as part of the 2DCC-MIP Resident Scholar Visitor Program (RSVP). York learned how to grow tungsten diselenide thin films via MOCVD while using 2DCC-MIP facilities for her Ph.D. thesis research. Her experiments helped confirm that indeed, this method worked.
“While carrying out these experiments, Krystal observed that the direction of tungsten diselenide domains on sapphire switched when she varied the pressure in the MOCVD reactor,” Redwing said. “This experimental observation provided verification of the theoretical model that was developed to explain the attachment location of tungsten diselenide crystals on steps on the sapphire wafer.”
This study could have a significant impact on semiconductor research, as it could enable other researchers working on the synthesis of 2D semiconductors to reduce mirror twin domains. In fact, Redwing notes, wafer-scale tungsten diselenide samples on sapphire produced using this novel MOCVD process are available to researchers outside of Penn State via the 2DCC-MIP user program.
This work is important, according to Redwing, because semiconductor chips must advance to meet the demands of a new world of electronics that is coming.
"Applications such as artificial intelligence and the internet-of-things will require further performance improvements as well as ways to reduce the energy consumption of electronics,” Redwing said. “High-quality 2D semiconductors based on tungsten diselenide and related materials are important materials that will play a role in next generation electronics.”
Along with Redwing, Nayir, van Duin, York, Durbin, Choudhury and Zhu, other authors of the paper include from Penn State Anushka Bansal, Benjamin Huet, Kunyan Zhang, Saiphaneendra Bachu, Thomas V. McKnight, Nicholas Trainor, Aaryan Oberoi, Ke Wang, Saptarshi Das, Shengxi Huang, Nasim Alem and Vincent H. Crespi. From the Oak Ridge National Laboratory, other authors include Alexander A. Puretzky and from Western Michigan University, Robert A. Makin.
Support for this work was provided by the National Science Foundation through the 2DCC-MIP-MIP.