Direct Cation Exchange in Monolayer MoS

This is the accepted manuscript made available via CHORUS. The article has been

published as:

Direct Cation Exchange in Monolayer MoS_{2} via

Recombination-Enhanced Migration

Shi-Ze Yang, Weiwei Sun, Yu-Yang Zhang (张余洋), Yongji Gong, Mark P. Oxley, Andrew R.

Lupini, Pulickel M. Ajayan, Matthew F. Chisholm, Sokrates T. Pantelides, and Wu Zhou (周武)

Phys. Rev. Lett. 122, 106101 — Published 11 March 2019

DOI: 10.1103/PhysRevLett.122.106101

Direct cation exchange in monolayer MoS

via “explosive”

recombination-enhanced migration

Shi-Ze Yang

1, #

, Weiwei Sun

2,1,#

, Yu-Yang Zhang

3,2

, Yongji Gong

, Mark P. Oxley

Andrew R. Lupini

, Pulickel M. Ajayan

, Matthew F. Chisholm

Sokrates T.

Pantelides

2,3*

, Wu Zhou (周武)

3,1*

Center for Nanophase Materials Science, Oak Ridge National Laboratory, Oak

Ridge, Tennessee 37831, USA

Department of Physics and Astronomy and Department of Electrical Engineering

and Computer Science, Vanderbilt University, Nashville, Tennessee 37235, USA

School of Physical Sciences and CAS Center for Excellence in Topological Quantum

Computation, University of Chinese Academy of Sciences, Beijing 100049, China

School of Materials Science & Engineering, Beihang University, Beijing 100191,

China.

Department of Materials Science & NanoEngineering, Rice University, Houston,

Texas 77005, USA

# These authors contributed equally to this work.

*Corresponding authors: Sokrates T. Pantelides (pantelides@vanderbilt.edu

) or Wu

Zhou ([email protected]

)

Notice of Copyright:

This manuscript has been co-authored by UT-Battelle, LLC, under contract DE-AC05-00OR22725

with the US Department of Energy (DOE). The US government retains and the publisher, by

accepting the article for publication, acknowledges that the US government retains a nonexclusive,

paid-up, irrevocable, worldwide license to publish or reproduce the published form of this

manuscript, or allow others to do so, for US government purposes. DOE will provide public access

to these results of federally sponsored research in accordance with the DOE Public Access Plan

(http://energy.gov/downloads/doe-public-access-plan).

In addition to their unique optical and electronic properties, two-dimensional

materials provide opportunities to directly observe atomic-scale defect dynamics.

Here we use scanning transmission electron microscopy to observe substitutional

Re impurities in monolayer MoS

undergo direct exchanges with neighboring

Mo atoms in the lattice. Density-functional-theory calculations find that the

energy barrier for direct exchange, a process that has only been studied as a

diffusion mechanism in bulk materials, is too large for either thermal activation

or energy directly transferred from the electron beam. The presence of multiple S

vacancies next to the exchanged Re-Mo pair, as observed by electron microscopy,

does not lower the energy barrier sufficiently to account for the observed atomic

exchange. Instead, the calculations find that a Re dopant and surrounding S

vacancies introduce an ever-changing set of deep levels in the energy gap. We

propose that these levels mediate an “explosive” recombination-enhanced

migration via multiple electron-hole recombination events.

As a proof-of-concept,

we also show that Re-Mo direct exchange can be triggered via controlled creation

of sulfur-vacancies. The present experimental and theoretical findings lay a

fundamental framework towards manipulating single substitutional dopants in

two-dimensional materials.

Two-dimensional materials exhibit unique properties that can be used for novel

applications, but just as is the case with semiconductors, it is usually necessary to

introduce impurities to achieve desirable functionalization. “Defect engineering”

raises many challenges and holds many promises. It has been demonstrated that

aberration-corrected scanning transmission electron microscopy (STEM) provides

atomic-resolution images [1-6] that can be used to track impurity atoms in both two-

and three-dimensional materials and can directly monitor the defect reactions that

enable defect engineering. More recently, Lin et al. used the electron beam of a

STEM to sculpt and simultaneously image nanowires that are only three-atoms wide

in transition-metal dichalcogenide monolayers [7]. Ishikawa et al. reported the direct

observation of the migration of Ce and Mn dopant-atoms in a bulk AlN crystal [8].

Han et al. reported observations of migrating iridium adatoms and tri-iridium clusters

on MgO surfaces [9]. Susi et al. reported the observation of silicon dopants in

graphene undergoing beam-induced direct interchange with neighboring C atoms [10],

a key step towards the control of single dopant atoms [11,12]. All these investigations

show that STEM is a powerful tool for studying local atomic movements.

In general, thermal diffusion of substitutional impurities is mediated by native defects

such as vacancies and self-interstitials [13]. These processes are well understood and

are used widely to model dopant diffusion for engineering applications [14]. The

direct exchange between a substitutional impurity and a neighboring host atom has

been discussed in the literature going back to the 1940’s [15-18], but the energy

barrier for such a process is generally believed to be too large. In 1986, Pandey [18]

predicted that the energy barrier for direct exchange (he called it concerted exchange)

for self-diffusion in Si is comparable to the activation energy for self-diffusion

mediated by vacancies or self-interstitials, but verification of such prediction has been

lacking. In only one case, namely boron-doped copper, direct exchange was indirectly

established by ruling out defect-mediated mechanisms [19]. So far, the only

atomic-scale observation of direct exchange was the inversion of Si-C bonds in

graphene, first reported by Susi et al., where they describe a process induced by

momentum transferred from the incident electron beam [10]. However, it is likely that

other mechanisms also play a role for semiconducting materials.

In this Letter, we report the experimental observation of direct exchange between

substitutional Re and Mo host atoms in monolayer MoS

by monitoring Re atoms in

atomically-resolved STEM images. Density-functional theory (DFT) has been

employed to investigate several possible pathways for the direct exchange. We find

that the exchange barrier is initially very large (~11 eV), but it is lowered by

surrounding the cation with sulfur vacancies, which create extra space and weaken the

local bonding. However, even with six sulfur vacancies surrounding a Re atom, the

calculated exchange barrier is only reduced to ~2.4 eV. This value is still much higher

than the maximum energy, 1.45 eV, transferred to Mo atoms (or 0.75 eV transferred to

Re atoms) by the 60 kV electron beam for a single elastic collision and too high to

occur efficiently at room temperature. DFT calculations suggest an alternative

energy-transfer mechanism as follows. The hydrogenic level of an isolated

substitutional Re impurity becomes deeper and deeper with larger and larger

displacements of the Re atom. Moreover, when S vacancies are introduced, additional

deep levels appear in the energy gap, so that the gap contains several deep levels that

constantly move up and down and appear and disappear as S vacancies come and go.

These dynamic energy levels provide multiple paths for the recombination of

beam-generated electron-hole pairs. The energy released by these “explosive”

recombination events is transferred to local vibrations that ultimately enable the direct

exchange of Re dopants as observed in STEM. The required spectator S vacancies

make it possible to control the diffusion process through controlled creation of

sulfur-vacancies as we show in a proof-of-concept experimental demonstration.

The monolayer Re-doped MoS

sample used in our experiments was grown on a

SiO

/Si substrate using a chemical vapor deposition method with molybdenum oxide,

sulfur and ammonium perrhenate powders as Mo, S and Re sources, respectively [20].

The as-grown sample was transferred onto a TEM grid and the STEM imaging was

performed with an aberration-corrected scanning transmission electron microscope

operated at 60 kV using the medium angle annular dark field (MAADF, inner angle

50 mrad, outer angle 300 mrad) imaging mode with a beam current of about 9 pA (see

Methods in the Supplementary Information).

Electron microscopy characterization reveals that Re dopants are substitutional in the

MoS

lattice [20]. The intensity in MAADF images is roughly proportional to Z

2-x

where Z is the atomic number and x is a fractional number depending on the exact

experimental parameters [5]. The STEM images in Fig. 1 show that single Re atoms

occupy Mo sites in the hexagonal MoS

lattice and they appear as bright spots in the

MAADF image. Although double Mo atoms (i.e. a surface Mo adatom on top of a

lattice Mo atom) have similar MAADF image contrast as that from a substituted Re

atom, they can be easily identified in experiment as adsorbed Mo adatoms hop around

easily under electron beam irradiation [21]. The stable configurations of the bright

cation before and after direct exchange, therefore, rule out the possibility that these

features are due to surface Mo adatoms, and the good agreement between the

simulated image intensities with the experimental MAADF image verifies that the

brightest features in the experimental MAADF images are substitutional Re atoms

(Fig. 1).

FIG 1. Comparison between experimental and simulated STEM-MAADF images.

The simulated MAADF image (indicated by the blue rectangle) is superimposed on

top of the experimental image. Substitutional Re site, double Mo column, and single S

vacancy (V

) are highlighted by red, blue, and orange, respectively, in the simulated

image.

In addition to the observation of substitutional Re dopants in the MoS

lattice, a

Re-Mo exchange event is identified by analyzing the recorded sequential MAADF

images in areas where Re dopants are isolated. The exchange process is illustrated by

the different Re positions in the MAADF images recorded at time t

and t

+3.5 second

(s) as shown in Fig. 2, where the Re atom exchanged positions with a neighboring Mo

atom. In the images that document the exchange event (Figs. 2a and 2b), we observe

several S vacancies. The number and arrangements of S vacancies before and after the

exchange event are different. It is inferred that sulfur vacancies may play a role in the

Re-Mo exchange. The contribution of S vacancies will be further investigated using

theoretical calculations as described in the following paragraphs. Mo vacancies were

not observed during the cation exchange.

FIG 2. Direct Re-Mo exchange in Re-doped monolayer MoS

. (a, c)

STEM-MAADF images showing the position of the Re relative to other Re atoms (a)

and (c) the local area of Re doped MoS

with the corresponding structural model (e),

recorded at time t

before the exchange event. (b, d) STEM-MAADF images showing

the position of the Re relative to other Re atoms (b) and (d) the local MAADF image

with the corresponding structural model (f) recorded after the exchange event at

+3.5 seconds. The exchange of Re and Mo atoms is highlighted by blue and red

arrows, while sulfur mono-vacancies are indicated by orange dashed circles and sulfur

double-vacancies are indicated by yellow dashed circles. Re atoms are in red, Mo in

blue, and sulfur in yellow.

In order to elucidate the mechanism for the Re-Mo exchange process and the pertinent

activation energy, we performed systematic DFT calculations using the projector

augmented wave (PAW) method [22,23], as implemented in the Vienna Ab-initio

Simulation Package (VASP) [24-26]. Total energies were obtained in the generalized

gradient approximation (GGA) using the PBE functional formulated by Perdew,

Burke, and Ernzerhof [27]. For Mo, S, and Re atoms, Mo-4p, 4d, and 5s; S-3s, and 3p;

and Re-5p, 5d, and 5s states were treated as valence states. A cutoff energy of 500 eV

was used in all calculations. We used a 7×7 supercell with a 2×2×1 Monkhorst-Pack

k-point mesh. The Re-Mo exchange barrier was calculated in the presence of different

numbers of sulfur vacancies.

To determine migration barriers, we used the nudged elastic band (NEB) method

implemented within density-functional theory. In the calculations, three migration

pathways (see Fig. S1) were considered, with the initial and final cation

configurations derived from the experimental MAADF images. The Re-Mo

direct-exchange barrier as a function of the number of sulfur vacancies

) is shown

in Fig. 3a. The Re-Mo exchange barrier is calculated to be ~11 eV when there are no

neighboring S vacancies. The barrier drops significantly with increasing numbers of

vacancies, but even with six S vacancies, the exchange barrier is still 2.4 eV, which is

larger than the possible energy transfer from the 60 keV electron beam to Mo and Re

atoms (1.45 eV and 0.75 eV, respectively; see Fig. S2 [28]). Obviously, the S

vacancies surrounding the Re and Mo sites

provide more space for the exchange and

weaken the binding with surrounding Re and Mo atoms.

Due to the limited energy transfer from electron beam bombardment in elastic

collisions (Fig. S2), other energy transfer mechanisms for triggering the exchange

need to be explored. It has long been known that, in semiconductors under

non-equilibrium conditions, defect migration can be enhanced by electronic processes

[29]. More specifically, under electron or laser irradiation, large concentrations of

electron-hole pairs are present. If a defect has localized energy levels in the gap, these

levels can mediate electron-hole recombination: an electron in the conduction band is

first captured at the defect level and subsequently annihilates a hole in the valence

bands. In each of the two steps, the energy is dissipated to local vibrations that

enhance the migration rate. Each recombination event transfers an amount of energy

that is equal to or larger than the band gap. The transfer of the

electron-hole-recombination energy to defect-atom vibrations is instantaneous

(electronic-transition time scale), while its effect on the defect, i.e. the exchange

process, occurs in phonon time scales. The phenomenon is extremely fast, as

documented by experimental observations of boron and silicon-interstitial migration in

Si at 20K [30]. In the 1980’s, several cases of such “recombination-enhanced defect

migration” (REDM) processes were explained in detail by theoretical calculations

[31-38]. The most intriguing one is the case of the Si self-interstitial, which, in the

presence of electron-hole pairs, migrates with an effective zero barrier (athermal

migration) [17,38].

The combination of those reports with our observations motivated

us to propose the dynamically changing defect levels to mediate direct exchange.

FIG 3. Energy barriers for Re-Mo exchange in the presence of S vacancies and

defect energy levels of Re-doped MoS

. (a) The calculated Re-Mo exchange

migration barriers in the presence of different numbers of sulfur vacancies. (b)-(f)

defect energy levels (at Γ) in the gap region of Re-doped monolayer MoS

. (b) The

energy gap of perfect MoS

; (c) the defect energy level of a single substitutional Re in

MoS

; (d) the defect energy level of a Re dopant atom when the atom is displaced by

0.2 Å; (e) defect energy levels of a Re atom displaced by 0.2 Å with a nearby S

vacancy; (f) defect energy levels of Re atom displaced by 0.6 Å with two nearby S

vacancies.

In order to explore the REDM mechanism for the Re-Mo direct exchange in MoS

we performed a series of calculations of defect energy levels. Typical results are

shown in Fig. 3. In Fig. 3b, a clean band gap for the pure MoS

is shown. When one

Re substitutional impurity is included, one shallow donor level is found below the

conduction band minimum (CBM) as shown in Fig. 3c. If the Re atom is displaced by

0.2 Å (these are representative small displacements that may be induced by the

transfer of energy from the electron beam [10]), the shallow donor level moves deeper

in the gap as shown in Fig. 3d. In Fig. 3e, when a sulfur vacancy is placed near the Re

impurity, two split donor levels are present together with an acceptor level emerging

above the valence band minimum (VBM). If we further impose a larger displacement

of 0.6 Å and place two sulfur vacancies near the Re atom, the defect levels move even

deeper in the gap as shown in Fig. 2f. Clearly, as more vacancies are added, even

more defect levels appear in the band gap.

The results shown in Figs. 3e and 3f and similar results for other defect configurations

make it clear that, as a Re atom initiates a displacement and S vacancies appear and

disappear, an array of defect levels move up and down in the band gap, appear and

disappear, and provide a plethora of electron-hole recombination paths in a dynamical

fashion. The fact that we captured snapshots of S vacancy motion in Fig. 2 and Movie

S1 is strong evidence that the number of S-vacancies around the targeted Re atom and

thus the levels in the gap are in a constant flux. Therefore, several recombination

events can occur during a single Re-Mo exchange, providing energy equal to several

times the band gap. This “explosive” recombination-enhanced defect migration or

multivacancy-assisted REDM is the most likely mechanism that makes possible the

observed Re-Mo exchange in MoS

lattice. Since Re-Mo exchanges are induced by the

electron beam via the “explosive” recombination mechanism

rather than a thermally

activated process, the activation barriers cannot be determined by Arrhenius plots.

Having established that S vacancies are essential for the Re migration process, we

performed experiments to explore the possibility of controlling the S-vacancy

generation process, aiming to steer Re migration. We found that the number of

electron-beam-induced sulfur vacancies follows a linear relationship with the total

electron dose (Fig. S3). This suggests that the creation of sulfur vacancies at

designated positions could be achieved by precise control of the position of the

electron beam in STEM, as already demonstrated in Ref [10,11], and the electron dose.

As a proof-of-concept experiment, we further demonstrated that the direct exchange

of Re and Mo atoms in the MoS

lattice can be triggered by controllably creating

sulfur vacancies at specific sites surrounding the designated Re-Mo pair (Figs. 4 and

S4). As shown in Fig. 4b, during the controlled scanning, the electron beam was

scanned over a small region within the red dashed circle and was then intentionally

parked at specific S sites, as indicated by the red dots, to create sulfur vacancies

around the selected Re-Mo pair (highlighted by the yellow and red arrows in Fig. 4a).

Figure 4c shows the structure after this controlled scanning experiment, where the

designated Re-Mo pair exchanged positions.

FIG 4. Controllable Re-Mo exchange triggered by controlled creation of sulfur

vacancies around the selected Re and Mo atoms. (a) and (c) show the

STEM-MAADF images before and after the controlled scanning. The atomic

positions in exchange are highlighted by the yellow and read arrows. (b) illustrates the

region scanned by the electron beam and the S sites where the electron beam was

parked. Related sulfur vacancies are highlighted by orange (V

) and yellow (V

)

dash circles in (c).

In summary, we report atomic-scale observations of direct Re-Mo exchange events in

a monolayer MoS

lattice. The high diffusion barrier for exchange diffusion is

effectively lowered by the presence of multiple spectator sulfur vacancies. However,

energy transfer from the electron beam is still not sufficient to drive the direct

exchange. Electron-hole recombination via dynamically changing deep levels in the

gap is proposed to explain the observed phenomenon. Overall, we have demonstrated

that, in addition to the direct energy transfer from the electron beam to atoms, energy

can also be transferred indirectly, i.e. the beam generates electron-hole pairs, which

then undergo defect-mediated recombination and transfer the energy to local phonons,

causing or enhancing defect migration. We also demonstrate that by controllably

creating sulfur vacancies around a selected Re-Mo pair this Re-Mo direct exchange

can be triggered in a controlled manner. The present work has clarified the exchange

mechanism of cations in monolayer MoS

, which lays the foundation for future work

towards manipulating single atoms using an electron beam.

Acknowledgments:

Electron microscopy at ORNL (S.Z.Y., M.P.O., A.R.L., M.F.C. and W.Z.) was

supported by the U.S. Department of Energy, Office of Science, Basic Energy

Sciences, Materials Sciences and Engineering Division and performed in part as a

user proposal at the ORNL Center for Nanophase Materials Sciences, which is a DOE

Office of the Science User Facilities. Research at Vanderbilt (W.W.S., Y.Y.Z., and

S.T.P.) was supported by Department of Energy grant DE-FG02-09ER46554 and by

the McMinn Endowment. W.Z. and Y.Y.Z. acknowledge support from the National

Key R&D Program of China (2018YFA0305800), the Natural Science Foundation of

China (51622211) and the Key Research Program of Frontier Sciences, CAS. This

research used resources of the National Energy Research Scientific Computing Center,

a DOE Office of Science User Facility supported by the Office of Science of the U.S.

Department of Energy under Contract No. DE-AC02-05CH11231. This work also

used the Extreme Science and Engineering Discovery Environment (XSEDE), which

is supported by National Science Foundation grant number ACI-1053575.

References:

[1] P. E. Batson, N. Dellby, and O. L. Krivanek, Nature 418, 617 (2002).

[2] P. D. Nellist et al., Science 305, 1741 (2004).

[3] D. A. Muller, L. F. Kourkoutis, M. Murfitt, J. H. Song, H. Y. Hwang, J. Silcox, N. Dellby, and O. L.

Krivanek, Science 319, 1073 (2008).

[4] O. L. Krivanek et al., Nature 464, 571 (2010).

[5] W. Zhou, M. P. Oxley, A. R. Lupini, O. L. Krivanek, S. J. Pennycook, and J.-C. Idrobo, Microscopy

and Microanalysis 18, 1342 (2012).

[6] H. Sawada, T. Sasaki, F. Hosokawa, and K. Suenaga, Physical Review Letters 114, 166102 (2015).

[7] J. Lin et al., Nature Nanotechnology 9, 436 (2014).

[8] R. Ishikawa, R. Mishra, A. R. Lupini, S. D. Findlay, T. Taniguchi, S. T. Pantelides, and S. J. Pennycook,

Physical Review Letters 113, 155501 (2014).

[9] C. W. Han, H. Iddir, A. Uzun, L. A. Curtiss, N. D. Browning, B. C. Gates, and V. Ortalan, The Journal

of Physical Chemistry Letters 6, 4675 (2015).

[10] T. Susi et al., Physical Review Letters 113, 115501 (2014).

[11] M. Tripathi, A. Mittelberger, N. A. Pike, C. Mangler, J. C. Meyer, M. J. Verstraete, J. Kotakoski, and

T. Susi, Nano Letters 18, 5319 (2018).

[12] O. Dyck, S. Kim, S. V. Kalinin, and S. Jesse, Applied Physics Letters 111, 113104 (2017).

[13] C. Nichols, C. Van de Walle, and S. Pantelides, Physical Review B 40, 5484 (1989).

[14] P. M. Fahey, P. Griffin, and J. Plummer, Reviews of Modern Physics 61, 289 (1989).

[15] H. Huntington and F. Seitz, Physical Review 61, 315 (1942).

[16] F. Seitz, Acta Crystallographica 3, 355 (1950).

[17] E. Kaxiras and K. C. Pandey, Physical Review B 47, 1659 (1993).

[18] K. C. Pandey, Physical Review Letters 57, 2287 (1986).

[19] B. Ittermann et al., Physical Review Letters 77, 4784 (1996).

[20] S.-Z. Yang et al., Adv. Mater. 30, 1803477 (2018).

[21] J. Hong, Y. Pan, Z. Hu, D. Lv, C. Jin, W. Ji, J. Yuan, and Z. Zhang, Nano Letters 17, 3383 (2017).

[22] G. Kresse and D. Joubert, Physical Review B 59, 1758 (1999).

[23] P. E. Blöchl, Physical Review B 50, 17953 (1994).

[24] G. Kresse and J. Hafner, Physical Review B 48, 13115 (1993).

[25] G. Kresse and J. Hafner, Physical Review B 49, 14251 (1994).

[26] G. Kresse and J. Furthmüller, Physical Review B 54, 11169 (1996).

[27] J. P. Perdew, K. Burke, and M. Ernzerhof, Physical Review Letters 77, 3865 (1996).

[28] See Supplemental Material [url] for calculation of maximum transferrable energy from the

electron beam, which includes Ref. [39].

[29] S. T. Pantelides, Deep centers in semiconductors (CRC Press, 1992).

[30] G. D. Watkins, Physical Review B 12, 5824 (1975).

[31] J. R. Troxell, A. P. Chatterjee, G. D. Watkins, and L. C. Kimerling, Physical Review B 19, 5336

(1979).

[32] K. Maeda, M. Sato, A. Kubo, and S. Takeuchi, Journal of Applied Physics 54, 161 (1983).

[33] G. A. Baraff, M. Schluter, and G. Allan, Physical Review Letters 50, 739 (1983).

[34] S. T. Pantelides, A. Oshiyama, R. Car, and P. J. Kelly, Physical Review B 30, 2260 (1984).

[35] H. Sumi, Journal of Physics C: Solid State Physics 17, 6071 (1984).

[36] K. H. Chow and G. D. Watkins, Physical Review Letters 81, 2084 (1998).

[37] J. W. Steeds, W. Sullivan, A. Wotherspoon, and J. M. Hayes, Journal of Physics: Condensed Matter

21, 364219 (2009).

[38] R. Car, P. J. Kelly, A. Oshiyama, and S. T. Pantelides, Physical Review Letters 52, 1814 (1984).

[39] F. Banhart, Reports on Progress in Physics 62, 1181 (1999).