of different adsorption states in (a), (c), (e). (h) Statistical dwell time of different adatom structures. (i) DFT revealed energetics for evolution between ground state TMo, transition state TS, and metastable state H. (j) Detailed atomic dynamics of transition from TMo to H with top view and side view of the 3D atomic structure, respectively. Note the TMo1 → H → TMo2 transition in (i) is symmetric, and hence, only the first half process (left red arrows in (i)) is drawn. Reproduced from Hong et al. (2017) with permission.
Figure 7. Atomic-scale migration of vacancy defects in monolayer MoS2. (a)–(d) Time-lapse series of experimental ADF-STEM images of an Mo vacancy. Scale bar: 0.5 nm. (e)–(h) DFT-relaxed atomic structures corresponding to the evolution in a–d from the Mo vacancy (VMo in (a)) to its metastable state (
The experimental ADF-STEM images of the ground-state TMo, first-metastable H, and second-metastable TS also provide important configurations as the input for the DFT calculation, to reveal more details of the structure transition and energetics involved in the surface migration. Figures 6(i) and 6(j) show the energy profile and corresponding structural evolution in the primary kinetic pathway TMo1 → H → TMo2 with a migration barrier of 0.62 eV. Another secondary kinetic pathway TMo1 → TS → TMo2 is also found but with a much lower frequency in the experimental observations, with a barrier calculated to be 1.1 eV. The TMo1 → H → TMo2 pathway on MoS2 surface is preferred and acts as the dominant pathway, due to the tendency of the d-electrons of the Mo adatom to form covalent bonds with the surface S in a most-stable triangular–prismatic or metastable octahedral coordination [32].
This difference in energy barriers for different adatom migration pathways also agrees well with the contrasting experimental statistics of TMo and TS. As both the energy barriers are not so large, thermal activation could still induce the surface migration but influenced by the electron beam irradiation.
3.2.2. Mo vacancy
Compared to the mobile Mo adatom defects, Mo vacancies are less frequently observed to migrate within the monolayer lattice. Jin et al. utilized high acceleration voltage to observe the evolution of the vacancies through time-elapsed ADF-STEM imaging series [31]. Figure 7 shows one time-sequential example of the migration of Mo vacancies with initial, metastable, and final states all imaged in one series of vacancy hopping. This Mo vacancy migration is actually the movement of neighboring Mo lattice atom. In the corresponding structure models in Figs. 7(e)–7(h), the migrating Mo lattice atom neighboring the Mo vacancy is highlighted in blue and purple, with arrows indicating the distance and direction of the next hopping. Again, in the vacancy hopping, the defect migration still obeys the random walk behavior without directional preference, typical of the Brownian motion of particles. Also, the Mo vacancy hopping only occurs within the sublattice of Mo and would never enter the S sublattice. Consistent with the DFT calculation, statistical analysis confirms the ground-state VMo and metastable
As shown in Figs. 8(c) and 8(d), the DFT-calculated energetics in the VMo →
Figure 8. States of vacancy and their evolution. (a) and (b) Statistical counts and dwell time of vacancies and their metastable states during the migration of an Mo vacancy. (c) DFT-calculated migration pathway of Mo vacancy. The dynamic process is shown by the inset atomic models with arrows illustrating the migration pathway of the neighboring Mo atom. (d) Detailed atomic dynamics of Mo vacancy migration with top-view and side-view, respectively. Note the VMo →
Such a high energy barrier of 2.9 eV also indicates that the vacancy migration must be induced by the beam–atom scattering interaction, since this order of energy barrier is not accessible by thermal activation at room temperature. Hence, the observed vacancy migration is a process driven by the electron beam which transfers enough energy to excite the target atom/defect into its metastable states.
3.3. Grain/domain boundaries in MBE-grown MoSe2
Domain/grain boundaries are very common defects in polycrystalline materials, playing a dominating role in their mechanical and electric properties. Xie et al. found that ordered grain boundaries existed quite commonly in the atomically thin transition metal dichalcogenides (TMDs) synthesized by molecular beam epitaxy (MBE) [33–40]. These defects are recently characterized in atomic resolution by STM and ADF-STEM, named as inversion domain boundary (IDB) or mirror twin boundary. They emerge in the matrix of the as-grown monolayer MoS2 and MoSe2 and link one with another in a triangular network and run along the zigzag directions. Figure 9(a) shows an STM image of the IDB-decorated MoSe2 surface, while the close-up image of Fig. 9(b) reveals some fine details where each IDB defect manifests by two closely spaced mirror-symmetric
