Assembling functional nanoscale devices atom-by-atom has taken a big step forward with the demonstration of scanning transmission electron microscopy (STEM) e-beam manipulation of single Si atoms on a graphene lattice, including directed motion of Si through the lattice to a target defect and incorporation into the graphene lattice in a selected location. This proof of principle by scientists at Oak Ridge National Laboratory (ORNL) could be useful in the fabrication of atomic scale devices needed in small quantities—like qubits and vacancy centers—and for studying chemical reactions at the atomic level.

While moving atoms using a scanning tunneling microscope (STM) was introduced in the early 1980s, in recent years teams of investigators have shown that STEM, with its atomically focused electron beam tip capable of being positioned with picometer precision, could make subtle changes in a material’s structure on the molecular and atomic levels. The ORNL team harnessed this power of STEM to produce a vacancy in graphene, controllably move an Si adatom across the surface toward the vacancy, and then insert the Si into the graphene lattice to produce a thermodynamically stable system. 

“The key to controlling this process lies in being able to vary the dose of the electron beam over a range of energies,” says Sergei Kalinin, director of the ORNL Institute for Functional Imaging of Materials. “This allows us to do something very special. We can visualize the sample with low [electron] dose imaging, switch to a local high dose to change the sample and make the Si atoms move, and then visualize the sample again at low dose.”

Silicon is almost always found as a contaminant on the surface of graphene—the source being the glassware or solvents used in the graphene synthesis process. To introduce a single Si substitutional atom at a specific lattice site, a 100 keV STEM beam was placed on the desired lattice site for 1–2 seconds to create a defect. Then, by scanning the beam over a 1–2 nm area near the defect to sputter away contaminants, the defect healed by incorporating an Si atom, as shown by subsequent imaging at 60 keV. Silicon dimers were created in a similar manner, demonstrating the ability to build devices comprising multiple atoms, which will be needed for most practical applications of this technology.

Once the Si atom was substituted in the lattice, the researchers demonstrated the ability to move it to a desired alternate location in the lattice to show further control of the process. In this case, a 60 keV beam was scanned over the substituted Si defect and its neighboring C atoms in the direction of the target lattice location. Si movement occurs by temporary removal of a neighboring C atom, followed by the filling of the vacancy by the Si atom, which also pulls the ejected C atom into the Si atom’s former space on the lattice. Repeated scans of the beam in the direction of the target site produced somewhat random motion of the Si atom through the lattice. Eventually, though, the Si atom arrived at the targeted lattice site. The researchers write that scanning over the neighboring C atoms appears to “encourage” the Si atom to move toward the target.

“This is an amazing example of turning bad to good—beam damage into controlled defect generation and doping,” says Stephen Pennycook of the National University of Singapore, who was not involved in this research.

Future research is aimed at determining which position of the beam is most conducive to making the atom move to the desired lattice location to remove some of this random motion. “Is the ideal beam position 1 angstrom away from the atom or 1.3 angstroms from the atom, or some other distance?” asks Ondrej Dyck of ORNL, the first author of the article. 

To date, all experiments have been done manually, but that will soon change. “What we want to do now is to use elements of automated experiments and machine learning to teach the microscope to look for the optimal position, and to learn which position produces the most efficient manipulation,” Stephen Jesse of ORNL, another author on the article, points out.

Kalinin concludes, “Until now, we considered two primary pathways toward atom-by-atom nanotechnology—either via scanning probe microscopy assembly, or chemical synthesis. Now electron beam manipulation provides the third paradigm of nanotechnology.”

Paul Weiss of the California NanoSystems Institute at the University of California, Los Angeles, who was not involved in this work, notes: “This result is exciting in that it will enable building and then measuring test structures for precisely doped graphene, in analogy to the early days of when we first moved atoms one by one on surfaces with the scanning tunneling microscope.”

Read the article in Applied Physics Letters.