Challenges for Scaling Up Electron-Beam Manipulation of Graphene Impurities

• DOI: 10.1093/micmic/ozad067.704
• Publication Date: 2023-07-22T19:58:28Z
• Authors: Andreas Postl; Eugen Kozyrau; Jacob Madsen; Toma Susi
• Keywords: 103042 Elektronenmikroskopie; 103030 Radiation physics; 103018 Materialphysik; 210004 Nanomaterials; 103030 Strahlenphysik; 210004 Nanomaterialien; 103018 Materials physics; 103042 Electron microscopy

Since the possibility to manipulate lattice impurity atoms using a focused electron beam in scanning transmission microscope was discovered 2014 [1] and first experimentally demonstrated graphene 2017 [2,3], progress has been made extend manipulation different elements even other materials including single-walled carbon nanotubes bulk silicon, as discussed recent review [4].However, despite advances automation machine-learning, scaling up of technique build atomically precise patterns out multiple proven challenging.Electron-beam is challenging for two reasons: same must be used both visualize structures, energy transfer from electrons target stochastic.Together these mean that it not possible prevent unwanted interactions whereby an moves undesired direction [2, 3], or worse yet, atom knocked instead desired movement [1].However, although interaction does appear fully elastic [5], tuning effective increasing number successful manipulations [6].The greatest initial challenge fabrication samples with high local concentration impurities otherwise undamaged lattice, but techniques such vacancy-mediated substitution silicon adatoms [3,7] able produce highly suitable more than dozen potentially manipulable close proximity (Fig. 1a).However, attempts led surprising disappointing result: most were quickly replaced by 1b), preventing creation multi-atom structures.This so-called replacement process observed germanium [8], particularly problematic phosphorus where its probability measured on par [9].The common at thought susceptible problem, turned cleaner samples, likely becameprecisely issue mentioned above shown Fig. 1.We note data failed experiments are usually published.Although atomistic details this mechanism still murky, cannot purely thermally activated.In our own experiments, phosphorus-and silicon-doped have heated temperatures 1000 °C tens cumulative hours, can easily found, placing stringent constraints related reaction barriers suggesting irradiation play pivotal role.Recent studies radiation stability elevated shed light source atoms: diffusing whose migration barrier estimated 0.33 eV [10].It seem heating unclear, they ubiquitous.However, considering value barrier, ultra-stable cryogenic stages could expected provide solution, recently proposed [4].A second tackled before scaled sample drift.Although active drift compensation enabled small subscan windows [3], we believe ideal leads significant part dose missing triggering processes.Further, machine learning structure recognition use keep track drift.Active parked real-time monitoring converged-beam diffraction pattern, center-of-mass displacement results real-space beam.This compensated via opposing shifts, keeping atom, detecting jumps, correcting following scan frames.Image simulations work practical implementation will [11].