DSpace at VNU: Effect of Elasticity of the MoS2 Surface on Li Atom Bouncing and Migration: Mechanism from Ab Initio Molecular Dynamic Investigations

A single Li atom is fired with initial kinetic energy level 0.2 eV or 2.0 eV and various targeting factor x, which determines the collision angle.. Figure 4 demonstrates this trapping p

Migration: Mechanism from Ab Initio Molecular Dynamic Investigations

Thi Huynh Ho, Hieu Cao Dong, Yoshiyuki Kawazoe, and Hung Minh Le

J Phys Chem C, Just Accepted Manuscript • DOI: 10.1021/acs.jpcc.6b09954 • Publication Date (Web): 19 Dec 2016

Downloaded from http://pubs.acs.org on December 24, 2016

Just Accepted

“Just Accepted” manuscripts have been peer-reviewed and accepted for publication They are posted

online prior to technical editing, formatting for publication and author proofing The American Chemical

Society provides “Just Accepted” as a free service to the research community to expedite the

dissemination of scientific material as soon as possible after acceptance “Just Accepted” manuscripts

appear in full in PDF format accompanied by an HTML abstract “Just Accepted” manuscripts have been

fully peer reviewed, but should not be considered the official version of record They are accessible to all

readers and citable by the Digital Object Identifier (DOI®) “Just Accepted” is an optional service offered

to authors Therefore, the “Just Accepted” Web site may not include all articles that will be published

in the journal After a manuscript is technically edited and formatted, it will be removed from the “Just

Accepted” Web site and published as an ASAP article Note that technical editing may introduce minor

changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers

and ethical guidelines that apply to the journal pertain ACS cannot be held responsible for errors

or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Effect of Elasticity of the MoS 2 Surface on Li Atom Bouncing and Migration: Mechanism from Ab Initio

Molecular Dynamic Investigations

Thi H Ho1, Hieu C Dong1, Yoshiyuki Kawazoe2, Hung M Le3,4,*

Faculty of Applied Sciences, Ton Duc Thang University, Ho Chi Minh City, Vietnam

ABSTRACT: Born-Oppenheimer molecular dynamics has been carried out to investigate the evolution of Li-atom trapping on the MoS2 surface A single Li atom is fired with initial kinetic

energy level (0.2 eV or 2.0 eV) and various targeting factor x, which determines the collision

angle After getting trapped, Li is observed to bounce elastically and glide on the MoS2 surface thanks to the "breathing" vibration of MoS2 Both firing energy and targeting factor x are shown

to have a significant effect on the trapping and gliding processes It is found that higher value of

targeting factor x (≥0.6) and initial firing energy (2.0 eV) would enhance Li migration on the

interacting configurations suggests that there is ionic interaction and partial charge transfer

between the absorbed Li atom and MoS2 monolayer during the bouncing and migration process

The HSE calculations for those structures unveils the metallization of MoS2 due to Li

I INTRODUCTION

Over past few decades, transition metal sulfides have become an attractive material due to its considerable properties such as magnetism, superconductivity, fluorescence, and electrical properties.1-4 Among these compound, molybdenum disulfide (MoS2) has been also studied extensively for applications such as electrochemical energy storage and conversion material,5,6catalyst,7-9 and solid lubricant.10,11 Recently, the demand for effective cathode materials of lithium-ion rechargeable batteries leads a great research interest concerning MoS2-Li interactions Like graphite, MoS2 has a hexagonal unit-cell structure and MoS2 nanoparticles can

be classified as an inorganic nanocarbon analogue of structures like plate-like graphene,12 like fullerenes,13 pipe-like nanotubes,14 which exhibits unique properties In MoS2, the atoms are covalently bonded to form a sandwich structure with two-dimensional S-Mo-S trilayers stacked together through weak Van der Waals interactions.15 With high theoretical specific capacity and good raw material abundance,16 MoS2 has been considered a suitable material for developing effective electrodes

onion-The weak interlayer interaction allows guest atoms and molecules to intercalate reversibly and diffuse through the weakly-bonding stacked layers.17,18 As a result, the intercalation process leads to two main effects: expansion of interlayer spacing and charge transfer from the guest to the MoS2 host.19 Because of such effects, MoS2 has been nominated as a reasonable choice for electrode materials Li et al.20 investigated the adsorption and diffusion of lithium atom on the MoS2, and the results showed that the Li mobility could be significantly facilitated in MoS2nanosheets because Li binding energy decreases Rastogi et al.21 demonstrated that Li was one of the most effective adatoms to enhance the n-type mobile carrier density in MoS2 for battery applications In a previous study, structural transition between the thermodynamically unstable T

phase and the H phase was investigated with the involvement of adsorbing Li atoms Such a

transition was shown to have a barrier, which might be reduced by increasing the concentration

of Li atoms.22 By employing a first-principles calculations, Ersan et al.23 demonstrated a

diffusion of Li on the MoS2(1-x)Se2x, and suggested that the adsorption of Li atoms might

metallize the dichalcogenide layer Concerning the tendency of Li clustering on the MoS2

surface, Putungan and co-workers showed the case of two Li atoms sitting close to each other

was energetically unfavorable because Li dimer would dissociate quickly and re-locate on

nearest Mo top-sites.24 In such a study, the overall migration barrier for Li clustering was

estimated as ~0.5 eV Although there have been several theoretical studies concerning the

interaction between Li atom and single-layer MoS2 based on density functional theory, it is

necessary to find out more about the interacting mechanism during a dynamic process

Considering the fact that it still has limited data on molecular dynamics (MD) mechanism, we

believe it is worthy to conduct a fundamental MD investigation to examine the behavior of Li

atom on the MoS2 surface

In this study, we employ direct ab initio molecular dynamic (MD) simulation of lithium atom collision with MoS2 at two different levels of Li-firing kinetic energy, i.e 0.2 eV and 2 eV,

while the MoS2 is set to thermally vibrate at room temperature (300 K) During the process, we

investigate the role of elasticity of MoS2 in trapping Li atom, and find out how a Li atom

interacts and diffuses on the MoS2 surface Subsequently, we choose several representative

configurations to execute highly qualitative self-consistent calculations for studying the resultant

modification on electronic structure properties We believe our theoretical study provides more

physical insights and disambiguates the attaching process of Li atoms onto the MoS2 surface

II METHODOLOGY

In this study, our main objective is to investigate the progress of Li-atom trapping by the MoS2 surface when a single Li atom is allowed to move toward and collide with the semiconducting layer This objective can be attained by adopting a MD approach In our procedure, there are three primary steps:

i Setting up a randomized configuration of the pure MoS2 surface (without Li) thermally vibrating at 300 K

ii Executing ab initio MD for the Li-MoS2 system

iii Selecting several interesting configurations from the trajectories to perform highly qualitative self-consistent calculations and study the electronic properties

In the following sub-sections, we will describe in details how we set up a trajectory sample and what information should be extracted from the trajectory

In the initial stage, an MoS2 monolayer consisting of 27 atoms (9 Mo and 18 S atoms) in

a (3×3) supercell is allowed to conduct thermal vibration at room temperature for a period of 500 Rydberg time units (Rtu), and a fixed step size of 0.5 Rtu is chosen for integrations In real time, such a period equals 24.19 fs For a (3×3) MoS2 supercell, the a lattice parameter for the two- dimensional system is chosen as 9.59 Å, while 15 Å is assumed to be the length of the c lattice

vector to guarantee the vacuum assumption for the Li atom in the system For simplicity of this case study, during the later MD investigation process, we make an assumption that the defined lattice parameters remain constant during the entire dynamic process of atoms

The Car-Parrinello MD25 (CPMD) technique implemented in the Quantum Espresso package26 is employed in this early stage The cut-off energy is chosen as 35 Rydberg and the

Martin-Troullier norm-conserving pseudo-potentials27 are employed for the involving atoms

(Mo, S, and Li) The MoS2 system is made experimentally realistic when the Mo and S atoms are

allowed to fluctuate at 300 K After the CPMD process, the geometry and velocity configurations

are stored in the database for later use A geometry configuration with well-randomized

velocities is generated by simply choosing a configuration from this CPMD database

2 Executing direct DFT molecular dynamics

After constructing the data for thermally equilibrated configurations, we insert the Li atom into the system With the chosen size of unit cell, the distance between two adjacent Li

atoms is 9.59 Å, which guarantees negligible interaction between Li atoms in the periodic

system As a benchmark calculation, we perform a MD simulation with variable unit cell, in

which the Li atom is migrating from one S-Mo-S potential trap to another, and learn that the unit

cell parameter responses insignificantly during the Li migration process Therefore, all

investigated cases herein are conducted with a fixed unit cell

The Li atom is set the move toward MoS2 with two different levels of initial kinetic energy: 0.2 eV and 2 eV Such energy levels are considered “low” because they will not cause a

severe deformation to the MoS2 surface The higher energy level, 2 eV, might be considered as a

hard collision (bombardment) onto the surface

The Li atom is located 6 Å aside from the MoS2 surface, and its projection lies on top of

a Mo atom We hereby consider 21 collision cases at each kinetic energy level In each case, the

angle of striking velocity is varied as described below

In the first case, the Li atom is set to strike perpendicularly to the MoS2 surface, and aim

to an Mo atom (referred as D case) From cases 2 to 11, Li is set to strike 10 different spots of

destination on a projected Mo-S linkage (denoted as spot C in Figure 1) Let us denote R as the projected distance of an equilibrium Mo-S bond A spot of destination is located at the point x 2 R (Å) from the Mo atom as described in Figure 1, where x = 0.1, 0.2, …, 1.0 For convenience, we

refer these cases as the C1, C2, , C10 cases From cases 12 to 21, Li is set to strike 10 different spots of destination resided on the bisecting vector of two Mo-S bonds (denoted as spot B in Figure 1) Again, those B spots of destination are appointed similarly to the previous cases with

the x 2 R factor For convenience, we refer these as the B1, B2, , B10 cases.

In the trajectory integration process, we employ the velocity-Verlet method28 with a standard step size of 0.484 fs The atomic forces of Mo, S, and Li atoms are extracted directly from first-principles self-consistent calculations executed by the Vienna Ab Initio Simulation package.29-32 The well-established Perdew-Burke-Ernzerhof exchange-correlation functional33-35

is employed, while the kinetic-energy cut-off is chosen as 400 eV, which is a standard cut-off level and suitable for the cost of long-time Born-Oppenheimer MD simulations The projector-augmented wave method36,37 is employed to construct the electronic wave-functions for the participant atoms, which describes the valence shells of 5s, 4d for Mo, 3s, 3p for O, and 2s, 2p for Li To save computational expense for the Born-Oppenheimer MD simulations, we only perform Γ-point calculations at each integration step A MD trajectory is terminated after 1,000 integration steps (10,000 Rtu) It should be noted that in the first step, we utilize the input structure produced by PBE calculations within the well-established Quantum Espresso package, which in principle should be produce analogous structural ground state with respect to the PBE calculations with VASP in the second step

3 Performing self-consistent calculations for the chosen Li-MoS 2 complexes

After finishing the MD process, we pay attention to several interacting configurations at the stage of Li movement toward the MoS2 surface, when there are interactions that may cause

modifications to the band structure of MoS2 Therefore, qualitative spin-polarized self-consistent

calculations with a k-point mesh of (12×12×1) are executed for the chosen structures To explore

the partial density of states (PDOS), the Gaussian smearing technique is utilized with a spreading

value of 0.01 eV, and the dipole correction is activated We perform Bader charger analysis38 to

examine the amount of charge transfer between Li and MoS2 Moreover, the hybrid HSE

calculations39 are also employed to investigate the electronic structures of the chosen

configurations The cut-off energy level is chosen as 400 eV, while the k-point mesh of (3×3×1)

is chosen for the HSE calculations

III RESULTS AND DISCUSSION

Initially, the initial kinetic energy of 0.2 eV is chosen because we would like to examine the slow absorption and diffusion processes In a previous study, Ersan et al.23

suggested that a single Li atom could find good settlement on the pure MoS2 surface with an

adsorption energy of 1.92 eV as derived from first-principles calculations Figure 2 and Figure 3

shows the evolution of total kinetic energy terms of the MoS2 single-layer and Li during the

BOMD processes for 21 investigated cases at 0.2 eV We observe that the MoS2 monolayer

vibrates periodically at the average initialized temperature of 300 K, in which the S atoms tend to

move up and down This seems more or less like a "breathing" behavior, and the vibrational

period of MoS2 almost remains constant In the cases presented in this section, before Li collides

with MoS2, the average periodic time for the thermal vibration is approximately 444.61 Rtu (21.51 fs) at 300 K However, even after Li successfully establishes bonding with MoS2, the vibrational period of the layer does not seem to be affected significantly In our MD process, MoS2 vibrates around its equilibrium position for about 2,200 Rtu (106.43 fs) while Li moves closer to MoS2 According to our kinetic energy examination (see Figure 2), at the average distance of 4.78 Å from the surface, Li seems to start getting attracted by the layer as the kinetic energy of Li increases dramatically The attracting effect becomes gradually intensive, and reaches the maximum level of attraction at the distance of 1.78 Å, where we conceive the largest momentum of Li moving toward MoS2 In more details, the kinetic energy magnitude of Li increases dramatically from 0.2 eV at the beginning to over 5 eV thanks to the assistance of MoS2 breathing and strong attractions of MoS2 upon Li After that, the repulsive force begins to occur rapidly during the collision It is true that Li absorbs kinetic energy from MoS2 and there are effects of strong attractive and repulsive interactions MoS2 seems to vibrate stronger as proved by a significant increase of kinetic energy (more than 3.94 eV) at 4,200 Rtu (203.19 fs) This is due to the establishment of a stable bonding configuration between MoS2 and Li During the collision process, we also observe that Li can rebound several times However, the Li atom is quickly pulled back and joins the vibration with MoS2 It should be noted that the bouncing behavior does not provide enough momentum for Li to escape from the great attraction from the layer There are two circumstances that can occur then:

i Li is trapped around the triangular region formulated by three S ions, or

ii Li glides from a trap created by three nearest neighbor S atoms to the most nearby triangular trap We refer this behavior to as “migration”

In the D case (where Li is set to move toward to center of the triangular trap as described in Figure 1), the C1-C5 cases (in which Li moves toward one Mo-S bond), and the B1-

B6, B8 and B10 cases (where the collision is projected to the bisector of two Mo-S bonds), we

observe that Li is trapped in the triangular valley constituted by three S atoms Indeed, the Li

atom fluctuates continuously, but it cannot find a way to get out of the valley during the whole

investigating period with a maximum examination time of 10,000 Rtu (483.78 fs) Figure 4

demonstrates this trapping process in the C1 case, in which we show snapshots, x-, y-, and

z-directional kinetic energies, distances of Li to S atoms on the top layer as well as the

approximated distance of Li to the surface In the snapshots, Li is attracted strongly and

establishes interactions with three S atoms as it moves closer to MoS2 After that, Li rebounds

up-and-down for several times because of its large kinetic energy absorbed from the MoS2 layer

movement The rebound of Li decreases gradually until almost kinetic energy is transferred to

MoS2 At 6,590 Rtu (318.81 fs), the y-directional kinetic energy increases as shown in Figure 4

(a) The distance of Li-S(1) decreases significantly, which is an evidence that Li is now pulled

dominantly by S(1) and moves off-center At the same time, Li seems to run out of kinetic energy

and collides obliquely with respect to the S-S-S center However, there are still attractive

interactions from S(2) and S(5) toward Li as well as the elastic collision with S(1), Li can be

attracted back, trapped, and circulate around triangular position for at least another bouncing

period

In the remaining cases, the sliding translation is observed after the elastic collisions

There are two mechanisms that can lead to this behavior First, Li strongly collides with one of

three S atoms and rebounds elastically to jump out of the trap as observed in the C6-C10 cases

In the second mechanism, Li elastically collides at a bisectional point of two Mo-S vectors, then

bounces back to the third S atom in the trap, and finally leaps to another trap, as can be observed

in the B7 and B9 cases To some extent, this process is similar to the first mechanism Li moves toward the MoS2 layer with larger deflective targeting factor, then collides with S atoms for several times Li can bounce around three S atoms and finally jump to another S-S-S trap due to high kinetic energy Figure 5 presents snapshots, x-, y-, and z-directional kinetic energy, distances of Li to S atoms on the top layer and distance of Li to surface for the trapping process

in the B7 case In this case, Li collides elastically at the bisecting position between S(1) and S(5) at the beginning of the trapping process Therefore, the collision direction is changed, which makes the sliding migration occur easier In Figure 5 (a), the x-directional kinetic energy line goes up almost after the Li collision Also, there is evidence of deflective collisions of Li with the S atoms At 6,880 Rtu (332.84 fs), the gliding process occurs when Li begins to move from the

S(1)-S(2)-S(5) trap to the S(1)-S(4)-S(5) trap because the Li-S(2) and Li-S(5) distances are shorter than the Li-S(4) distance (illustrated in Figure 5 (b) when we see that the Li-S(4) goes down while the other two lines go up)

In this section, we observe that the effects depend much on the targeting factor x After

investigating 21 cases at the kinetic energy level of 0.2 eV, we observe the migration of Li

occurs easier if factor x is greater than or equal to 0.6 (i.e the shooting angle is over 6.3o) In

Figure 2, the total kinetic energy of Li for 11 investigated cases changes gradually when x

increases This leads to the conclusion that there is a deep valley of potential well, which tends to pull Li down and establish a stable Li-MoS2 complex This was also revealed by two previous theoretical studies Before Li collides MoS2, the kinetic terms seem to be similar, but the

movement of Li decreases faster with x ≥ 0.6, which proves that Li possesses more bouncing

collisions For convenience, we summarize the elapsed time before the first Li collision, number

of rebounds and elapsed time before Li gliding with respect to each shooting angle for seven

gliding cases (C6-C10, B7 and B9) at 0.2 eV in Table 1 In the C cases, the gliding process

occurs more easily with probability of 50% (5/10 cases), while there are just 2/10 gliding cases

in the bisecting collision cases (B cases) For the B cases, this happens due to the fact that the

kinetic energy of Li can decrease significantly after two collisions at the bisecting position and at

the opposite S atom In the C6-C7 cases as well as B7 case, the gliding process of Li happens

after bouncing three times on the surface, while C8, C9, C10 and B9 cases, the gliding process

occurs rigorously after only one bouncing period This shows a tendency that the number of

rebounds is lessened when we increase the shooting angles Li can elastically collide and jump

out of the potential trap directly at high values of the targeting factor x If the targeting factor is

low, Li can only escape after several interaction collisions with the three surrounding S atoms

With the above results, we conclude that the glide of Li would occur easier when increasing the

targeting factor Interestingly, we observe that the kinetic energy of Li is low (0.13-0.26 eV) in

two gliding moments in the B8 and C9 cases In those trajectories, migration happens when Li

jumps relatively far from the MoS2 surface In such circumstances, the potential trap will pull Li

back to the surface while gliding transition occurs almost at the same time In the other gliding

cases where migration occurs with short Li-MoS2 distances, we observe that the required kinetic

energy of Li is very diverse in a wide range (1.3-4.2 eV)

Previously, at the kinetic energy of 0.2 eV, we observe that there are two behaviors occurring when Li move to the MoS2 surface: trapping and gliding (migrating) However, with

this level of kinetic energy, trapping is still dominant, and gliding is somewhat difficult to occur

(only 33% of chance in the investigated cases) We now re-investigate those 21 cases with the initial Li kinetic energy of 2 eV Considering the same configuration condition applied for 21 new cases in the MD investigation, we will focus on clarifying how the effect of targeting factor

x enhances Li diffusing ability on the surface at the higher kinetic energy.

After finishing the MD investigations for 10,000 Rtu (483.78 fs), we observe that Li diffusion occurs in most of the cases (18/21 cases, except C5, B5 and B7) Recall that there are just 7 gliding cases with the firing energy of 0.2 eV In the three cases with no Li diffusion (C5, B5 and B7 cases), Li is observed to be trapped in the potential well and does not have sufficient time to escape the trap at the end of our MD investigation operation (10,000 Rtu) However, Li still fluctuates strongly on the surface, and we believe that if we extend the MD trajectory time, diffusions probably occurs in those three cases

For the cases where Li diffusions clearly occur, the mechanism for Li gliding on the MoS2 surface is fairly similar to the cases at 0.2 eV:

i Li first approaches a triangular hole, rebounds for several times and finally glides to the most neighboring hole after the alternating collisions with three S atoms

ii Li collides strongly with one S atom on the surface and leap to another trap very

rapidly This happens when the firing angle is large (high value of x).

At the higher firing energy level (considered as hard collision), Li rebounds for less periods than the cases at 0.2 eV as listed in the Table 1 and Table 2 This demonstrates that the trapping potential well on MoS2 surface has high elastic behavior Such a potential well keeps Li bouncing many times due to the resilience of three symmetric S atoms The S atoms can elevate and take down Li measuredly until Li loses most of its kinetic energy Still, it is hard for the

bouncing process to retain for a long time, and migration finally occurs Overall, the gliding

process of Li can take place much easier than the cases at 0.2 eV Besides, we also observe the

intensive impact of targeting factor When x is greater than or equal to 0.6, the number of

rebounds reduces from 5 times to only once and the gliding process of Li changes from indirect

to the direct way At low targeting factor (0 ≤ x < 0.6), Li only escapes the trap after rebounding

and then colliding elastically with three nearest neighbor S atoms During this bouncing process,

the kinetic energy of Li is much lower than the kinetic energy at the colliding moment The

gliding process occurs indirectly depending upon the appropriate collisions with S atoms

Contradictorily, for cases with high targeting factor (x ≥ 0.6), Li collides directly at one of three

S atoms or bisecting site, and then jump out of the trap with the aide of high kinetic energy

Moreover, as showed in the Figure 6 and Figure 7, the kinetic energies in the D, C1-C4 and

B1-B4 cases retain almost similar until 6,200 Rtu (299.94 fs) Li can still rebound for several times

before Li glides away For the C6-C10, B5, B6 and B8-B10 cases, the kinetic energies change at

the early stage, which shows the gliding process can occur after the first collision with MoS2

Comparing cases at two kinetic energy levels of 0.2 eV and 2 eV, we still see there is a threshold of the strong attracting forces when Li moves closer to MoS2 Li seems to be pulled at

the distance of 4.70 Å and the kinetic energy of Li reaches the maximum value of about 6.5 eV

at the Li-S layer distance of 1.65 Å After that, the repulsive forces increase dramatically while

Li continues to approach closer to collide with the surface We also observe that the gliding

movement of Li for most cases tends to follow the following path: at the beginning, Li jumps out

of the traps constituted by three S atoms above the Mo atom, then it moves to the hollow site of

MoS2 and finally turns to another trapping site Actually, there are several theoretical studies

concerning the diffusion path of Li adsorbed on the MoS2 surface The diffusion process occurs

due to the migrating of Li from a trapping site to another site by passing through a hollow site

Xu et al.41 reported that the T site of MoS2 is more energetically stable to bind with Li than the H site (those adsorption sites are defined in Figure 1) Therefore, Li can be trapped easier in the T site for a long time than in the H site before jumping out of the trap According to our observation from MD trajectories, Li moves to the H site and quickly glides to the T trap by following the zigzag-like path Overall, our MD evidence establishes a good agreement with this study

In addition to clarifying the impact of firing energy and targeting factor x, we carefully

examine the kinetic energy of Li as well as the elapsed time at the gliding moment, when Li escapes one trap for another, for all gliding cases as shown in Figure 8 With 2.0 eV kinetic energy, the gliding process occurs more frequently than the cases of 0.2 eV Besides, in the cases

with lower targeting factors x (< 0.6), gliding seems to occur after about 320 fs for both firing

kinetic energy levels During the short period before gliding, the kinetic energy of Li is low due

to the "indirect" gliding of Li, i.e Li only moves to another potential trap after bouncing and

colliding elastically with the nearby S atoms for several times When the targeting factor x

increases, the number of rebounds seems to decrease and Li is able to migrate faster with elapsed time varying from 170 fs to 50 fs, which is much shorter than the above 320 fs Especially, the gliding process can also occur directly after Li collides with the MoS2 surface and bounces for only one time We also observe that the elapsed time before gliding is less than 100 fs (see Figure 8) In such circumstances, the kinetic energy of Li is high (3-4.5 eV) because of energy adsorption from the vibrational movement of MoS2 With the two last cases of B9 and B10, we observe that the gliding process can even occur before Li collides with MoS2 as Li seems to be more attracted to the bisecting position Therefore, in the association of high targeting factor and

high initial kinetic energy, it is easier for Li to deflect and the gliding process can take place

more spontaneously Overall, we can conclude at this point that the gliding behavior relies

heavily on the firing energy, beside the effect of targeting factor x

In this section, qualitative self-consistent calculations are executed for six configurations to examine the electronic structure when Li moves and collides with MoS2 The

configurations that we choose for the investigation include: (a) a starting configuration at the

beginning of MD simulation, (b) the configuration at the moment Li starts to get attracted by

MoS2 (kinetic energy of Li starts to increase) (c) the configuration with highest kinetic energy

(repulsion begins to occur), and (d), (e), (f) three collision configurations in the D, C8, and B8

cases with the initial firing energy of 2.0 eV The density of states (DOS) of these configurations

is shown in the Figure 9 At the beginning, there is an evidence of a weak Li-MoS2 interaction

due to the insignificant hybridization of Li and MoS2 orbitals near the Fermi level (Figure 9 (a))

At the approximate distance of 4.70 Å from the surface, Li seem to be pulled by MoS2 and the

Li-2s state becomes delocalized and a hybridized eigenstate shows up quite below the Fermi

level as showed in Figure 9(b), and indicates partial charge is transferred to the MoS2 Looking at

the DOS of pure MoS2 form given by HSE calculations (Figure 10), we observe that the

conduction band maximum remains almost unaltered with respect to the Fermi level even when

Li is introduced into the system For the cases where Li is approaching closely to the surface,

there is a significant change in the valence band as seen in Figure 9(c)-(f) In terms of Bader

charge, we also notice a significant increase of charge of Li from +0.37 to +0.86 as listed in

Table 3 Then, Li is pulled stronger and stronger by the surface; as a result, the kinetic energy of

Li begins to increase and finally reaches its maximum At the highest kinetic energy, the projected DOS shows Li-2s overlapping with the orbitals of MoS2, which features for the ionic hybridization interactions Meanwhile, the Bader charge now increases to +0.86 and seems to get quite larger than the values at the collision moment In the three later cases, the DOS distributions of MoS2 also occupy energy states below the Fermi level In these situations, Li approaches very closely to the surface S atoms, and it can get a small amount of electronic charge back, thereby becomes quite less positive (see Table 3) The positive charge of Li is lowest for the D case (+0.77), while the C8 case gives the highest positive charge (+0.84).

The results of spin-polarized PBE calculations show that all investigated systems do not exhibit magnetic moments However, when we perform the HSE calculations for those configurations, magnetism is found The starting configuration (a) has a magnetic moment of 0.29 µB, which mainly arises from the 2s orbital of Li (35%) and 4d orbitals of Mo (56%), and the electronic contribution from Li seems not to affect the overall band gap of MoS2 When Li begins to get attracted (configuration (b)), the spin polarization term is stronger (0.58 µB) with the major contribution from the 4d orbitals of Mo (>80%) In this configuration, an in-gap state occurs at the Fermi level, which is not observed by the PBE calculations In configuration (c), Li approaches very close to the MoS2 layer, and a magnetic moment of 0.87 µB is found (90% contributed by 4d orbitals of Mo), and the in-gap state is now very clear due to the strong interaction between Li and MoS2 In last three cases of configurations (d), (e), and (f), the magnetic moments are reported as 0.54 µB, 0.88 µB, and 0.56 µB, respectively In configuration (e), we obtain the highest magnetic moment as Li is set to strike at an S atom Also, the in-gap states show up very clearly in configurations (d) and (e), which reveals metallicity of the structure This observation is in good agreement with the metallization of MoS2 discussed by

Ersan et al.23 In terms of Bader charge, the average charge for each atom type obtained HSE

calculations is in good agreement with the Bader charge derived from PBE calculations (see

Table 3) For comparison of charge analysis, we also conduct PBE calculations for the

hydrogen-bound isolated Li-MoS2 systems, which resemble the above structures, using the Gaussian 09

package42 with the 6-31G basis set43 for H, Li, S atoms and the LanL2DZ basis set44 for Mo

Only the H atoms are optimized, while we keep the other atoms frozen to resemble the above

structure of interest obtained from MD simulations In general, the charge of Li given by

Mulliken analysis is higher than the Bader charge as shown in Table 3 We find a good

agreement between Bader charge and Mulliken charge in predicting the trend of charge with

respect to different Li-MoS2 interacting configuration When Li is still far away from the MoS2

surface (configurations (a) and (b)), the charge of Li is relatively low (0.97-0.99 proton charge)

As it approaches MoS2, Li tends to give more electrons and becomes more positive (1.04-1.15

proton charge) However, in configuration (d) where Li approaches closely to the Mo atom (and

far away from the surrounding S atoms), the Mulliken charge of Li is almost neural The

Mulliken charge in this case is contradicting to that observed in Bader charge analysis

IV SUMMARY

In this study, we perform Born-Oppenheimer MD to investigate the evolution of Li-atom trapping on the MoS2 surface The single Li atom is allowed to move toward and collide with

MoS2 with variable targeting factors x and two firing kinetic energy levels of 0.2 and 2.0 eV

Interestingly, as we investigate the trapping mechanism during the MD processes, we also

observe a gliding (migration) behavior of the Li atom on the MoS2 surface Such an interesting

feature is delivered due to the elastic "breathing" vibration of the semiconducting monolayer The removal of S atom from the surface is not observed in our study.

The trapping and gliding behaviors are observed to rely heavily on the firing energy and

targeting factor x Higher value of targeting factor x (≥0.6) as well as initial firing energy (2.0

eV) would enhance the gliding probability; per contra, Li can be trapped in the potential hole created by three nearest S atoms for a longer period before Li can finally escape to another More specifically, with an initial kinetic energy of 2.0 eV, Li is more probable to translate from one triangular trap to another in 18/21 cases (85.7%), while there are just 7 gliding cases in the 0.2

eV case (33.3%) Besides, we observe a kinetic energy threshold for the Li movement when Li moves closer to MoS2 at both investigated levels of firing energy (0.2 eV and 2 eV) Even though the introduced firing energy does play a decisive role in Li gliding, it seems that such initial energy is significantly lower than the kinetic energy at the later stage as Li absorbs heat from MoS2 It should be noted that in all investigated cases herein, we do not observe a bounce-off behavior of Li from the surface

In the last section, the electronic structure examination for six representative configurations is performed by PBE and HSE calculations The PDOS and Bader charge are analyzed to examine the interactions and electron transfer between Li and MoS2, which can be employed to clarify the change of electronic behaviors of the Li-MoS2 system The electronic result reveals that Li is mostly attracted when it comes closer to the MoS2 surface due to ionic interactions At the same time, Li transfers most of its electronic charge to MoS2 and Li consequently becomes cationic The DOS evidence given by HSE calculations show that when

Li approaches closer to the MoS2 surface, there exist in-gap states Such eigenstates indicate the metallization of the layer, which is in good agreement with a previous study.15 Overall, our MD

trajectories offer two basic mechanisms for trapping and gliding when firing a single Li atom

onto the single-layer MoS2 surface We believe that further computational studies are necessary

to supplement how firing multiple Li atoms at the same time (atomic beam) would affect ion

trapping and migration

SUPPORTING INFORMATION

The trajectory configuration data for all collision cases are provided in the associated supplementary material Those files are in the Xcrysden structural format and compressed as

axsf.zip A trajectory video file is made for illustration of the Li migration This material is

available free of charge via the Internet at http://pubs.acs.org