Two-Dimensional WS2@Nitrogen-Doped Graphite for High

Oct 11, 2018 - ... and WS2@N-doped graphite (N-graphite) exhibit outstanding cycling ... In the case of WS2@N-graphite, vigorous amorphization of the ...
0 downloads 0 Views 3MB Size
Download PDF
Research Article Cite This: ACS Appl. Mater. Interfaces 2018, 10, 37928−37936

www.acsami.org

Two-Dimensional WS2@Nitrogen-Doped Graphite for HighPerformance Lithium Ion Batteries: Experiments and Molecular Dynamics Simulations Tekalign Terfa Debela,† Young Rok Lim,‡ Hee Won Seo,‡ Ik Seon Kwon,‡ In Hye Kwak,‡ Jeunghee Park,*,‡ Won Il Cho,§ and Hong Seok Kang*,∥

ACS Appl. Mater. Interfaces 2018.10:37928-37936. Downloaded from pubs.acs.org by KAOHSIUNG MEDICAL UNIV on 11/08/18. For personal use only.

†

Institute for Application of Advanced Materials and ∥Department of Nano and Advanced Materials, College of Engineering, Jeonju University, Chonju, Chonbuk 55069, Republic of Korea ‡ Department of Chemistry, Korea University, Sejong 339-700, Republic of Korea § Center for Energy Convergence, Korea Institute of Science and Technology, Seoul 136-791, Republic of Korea S Supporting Information *

ABSTRACT: As promising candidates for anode materials in lithium ion batteries (LIB), two-dimensional tungsten disulfide (WS2) and WS2@(N-doped) graphite composites were synthesized, and their electrochemical properties were comprehensibly studied in conjunction with calculations. The WS2 nanosheets, WS2@ graphite, and WS2@N-doped graphite (N-graphite) exhibit outstanding cycling performance with capacities of 633, 780, and 963 mA h g−1, respectively. To understand their lithium storage mechanism, first-principles calculations involving a series of ab initio NVT−NPT molecular dynamics simulations were conducted. The calculated discharge curves for amorphous phase are well matched with the experimental ones, and the capacities reach 620, 743, and 915 mA h g−1 for WS2, WS2@graphite, and WS2@Ngraphite, respectively. The large capacities of the two composites can be attributed to the tendency of W and Li atoms to interact with graphite, suppressing the formation of W metal clusters. In the case of WS2@N-graphite, vigorous amorphization of the N-graphite enhances the interaction of W and Li atoms with the fragmented N-graphite in such a way that unfavorable Li−W repulsion is avoided at very early stage of lithiation. As a result, the volume expansion in WS2@graphite and WS2@N-graphite is calculated to be remarkably small (only 6 and 44%, respectively, versus 150% for WS2). Therefore WS2@(N-)graphite composites are expected to be almost free of mechanical pulverization after repeated cycles, which makes them promising and excellent candidates for high-performance LIBs. KEYWORDS: tungsten disulfide, nanosheets, lithium ion battery, N-doped graphite, molecular dynamics simulations

1. INTRODUCTION

than that in graphene by guest intercalation, and the resultant weak van der Waals forces between the layers allow Li ions to diffuse easily. The theoretical capacities of MoS2 and WS2 are 670 and 433 mA h g−1, respectively, which are higher than that of Gr. However, TMDs alone exhibit low cycling performance and fast capacity fading. Hence, they are usually used as composites with Gr nanomaterials (for example, graphene and carbon nanotubes), resulting in enhanced cycling performance together with capacities even higher than the theoretical value.5−38 A number of works reported great enhancement of capacities by using the nitrogen (N)-doped Gr nanomaterials.16,27,29,35 It is generally accepted that the Gr nanomaterials not only help to buffer volume expansion but also enhance the

As the demand for portable electronic devices and electric vehicles continues to rise, developing renewable, environmentfriendly, and cost-effective energy storage systems is a hot research topic. For this purpose, lithium ion batteries (LIBs) have been widely used as commercial energy storage devices.1−4 However, graphite (Gr), which is a common anode material in the LIBs, is unable to satisfy the demand for next-generation applications owing to its low specific capacity (theoretically 372 mA h g−1). Hence, new anode materials with higher capacity and excellent cycling performance are highly desirable. Recently, much research has been devoted to two-dimensional (2D) graphene-like transition-metal dichalcogenides (TMDs), for example, MoS2 and WS2, due to their exceptional physical properties suitable for LIBs.5−12 The spacing between neighboring layers in these TMDs can be made much larger © 2018 American Chemical Society

Received: June 18, 2018 Accepted: October 11, 2018 Published: October 11, 2018 37928

DOI: 10.1021/acsami.8b10133 ACS Appl. Mater. Interfaces 2018, 10, 37928−37936

Research Article

ACS Applied Materials & Interfaces

Figure 1. (a) HRTEM image showing the general morphology of WS2 nanosheets. (b) Lattice-resolved TEM and the corresponding FFT images (zone axis = [0001]), revealing that the distance between the adjacent {110} planes (d110) is 2.7 Å. (c) HRTEM image of WS2@N-doped graphene (NGr) composites. (d) EDX mapping (W M-shell and S K-shell) shows that W and S are distributed homogeneously over the nanosheets. The HAADF STEM image is also shown on the left. (e) EDX spectrum (for the circled area in HAADF STEM image) confirms that the atomic ratio of W and S is 1:2.

specific capacity. The calculation on the MoS2 and Gr system showed that the Gr further enhances the interaction of the anode with Li atoms22 or suppresses the dissociation of MoS2 nanosheets driven by the conversion reaction of MoS2 + 4Li+ + 4e− → Mo + 2Li2S.39 Our group reported that the strong electronegativity of N atoms in the Gr layers could attract Li cations, and the pores around the pyridine-like structures can accommodate more Li cations around them.40,41 As far as the authors’ knowledge is concerned, however, little theoretical work has been reported on the origin of the capacity enhancement in WS2 composites with N-doped Gr (NGr). In this study, we systematically measured the electrochemical performance of LIBs that use WS2 nanosheets and their Gr and NGr composites. The composites (WS2@Gr and WS2@NGr) showed an appreciable enhancement in the cycling performance. The capacity of WS2 in WS2@NGr reached 963 mA h g−1, being significantly larger than 433 mA h g−1 that was predicted using the reactions of WS2 + 4Li+ + 4e− → W + 2Li2S and 2S + 4Li+ + 4e− ↔ 2Li2S. We elucidated the lithium storage mechanism by carrying out extensive and sophisticated calculations using tools such as state-of-the-art density functional theory (DFT), structural prediction algorithm, and ab initio molecular dynamics (AMD) simulations, supplemented by analyses of valuable parameters such as volume expansion and radial distribution functions (RDFs). Our investigation suggested that the incorporation of Gr and NGr increases the performance through strong W−C, Li−C, and Li−N interactions, which accompanies significant reduction of volume expansion. This finding gives an important insight into the role of Gr and NGr in the capacities of LIBs.

nanowires were produced by a hydrothermal reaction at 200 °C (see Supporting Information, Figure S1). The WS2@Gr and WS2@NGr composites were synthesized by ultrasonic treatments of WS2 and reduced graphene oxide (RGO) or Ndoped RGO with a weight ratio of 8:2. X-ray diffraction (XRD) pattern of WS2 nanosheets exhibits peaks corresponding to the hexagonal phase WS2 (JCPDS no. 84-1398, a = 3.153 Å, and c = 12.323 Å), as shown in Supporting Information, Figure S2. The specific surface areas for WS2@ Gr and WS2@NGr are 253.1 and 253.8 m2 g−1, respectively (Supporting Information, Figure S3 and Table S1). X-ray photoelectron spectra revealed that the N composition in NGr is 10 at. % of C atoms and the pyridine-like C−N bonding structure exists as a major one (Supporting Information, Figure S4). Figure 1a displays a high-resolution transmission electron microscopy (HRTEM) image of the WS2 nanosheets. The average thickness is 10 nm and the average size is 100 nm. Lattice-resolved TEM and corresponding fast-Fourier transformed (FFT) images at the zone axis of [0001] reveal a highly crystalline basal plane (Figure 1b). The d-spacing of {110} planes is 2.7 Å, which is in good agreement with the reference value (2.7307 Å). The HRTEM image of WS2@Gr reveals that the WS2 nanosheets are dispersed over the whole graphene sheet (Figure 1c). High-angle annular dark-field scanning TEM (HAADF STEM) image and energy-dispersive X-ray (EDX) fluorescence spectroscopy elemental maps are shown in Figure 1d, proving that the nanosheets are composed of W and S. The mapping of W and S elements was obtained using their M-shell and K-shell peaks, respectively. The corresponding EDX spectrum confirms the atomic ratio of W/S = 1:2 (Figure 1e). The LIB performance of WS2, WS2@Gr, and WS2@NGr as active materials was examined using coin-type half-cells as follows. Figure 2a displays the cyclic voltammetry (CV) curves (10th cycle). All three samples exhibit the lithiation (cathodic) peaks at 1.9 V and delithiation (anodic) peaks at 2.4 V, involving the reversible reaction of WS2 + xLi+ + xe− ↔ LixWS2. The CV data of multiple cycles, including the first one, are shown in Figure S5 (Supporting Information). Once the

2. RESULTS AND DISCUSSION We prepared and characterized three samples (WS2 nanosheets, WS2@Gr composites, and WS2@NGr composites) and then measured their electrochemical properties. All are described in Supporting Information. High-purity WS2 nanosheets were synthesized by sulfurization of monoclinic phase tungsten oxide (WO2.7) nanowires in a chemical vapor deposition reactor, using H2S gas at 400 °C. The WO2.7 37929

DOI: 10.1021/acsami.8b10133 ACS Appl. Mater. Interfaces 2018, 10, 37928−37936

Research Article

ACS Applied Materials & Interfaces

Figure 2d shows the cycling performance with the C rate changing in steps. The capacities in each step are summarized in Supporting Information, Table S3. At 5 C rate, the discharge capacity of WS2 is 430 mA h g−1, corresponding to 70% of that of the initial 1 C rate (5th cycle). In the cases of WS2@Gr and WS2@NGr, the discharge capacities at 5 C rate are 453 and 526 mA h g−1, respectively, which are about 70% of those in their initial step. When the C rate was changed to 0.1 C after 5 C rate, the discharge capacities of WS2, WS2@Gr, and WS2@ NGr exceeded their initial values at 0.1 C rate, that is 628, 676, and 975 mA h g−1 versus 601, 651, and 775 mA h g−1, respectively. Electrochemical impedance spectroscopy measurement provided the charge transfer resistance (Rct) and Li ion diffusion coefficients (Supporting Information, Figure S9 and Table S4). The Rct values of WS2, WS2@Gr, and WS2@ NGr are 31, 22, and 15 Ω, respectively, indicating that the faster charge transfer inside the (N-)Gr composites is responsible for the higher capacity. The values of the Li ion diffusion coefficient clearly suggest that Li ion diffusion is facilitated in the composites compared with that in WS2. After applying 1st discharge potentials, no crystalline phase was identified from ex situ XRD patterns (Supporting Information, Figure S10). The HRTEM images confirmed that the WS2 nanosheets became amorphous after the 1st discharge (Supporting Information, Figure S11). Our data were compared with other reported values for LIB performance of WS2 (Supporting Information, Table S5). The capacity of WS2@NGr (963 mA h g−1) is lower than the best result so far (1130 mA h g−1 for WS2/graphene/carbon nanofiber composite),37 but it is higher than the other result using Ndoped carbon@WS2 composites (801.4 mA h g−1).35 Next, we investigate the origin of the capacity using theoretical methods. For this purpose, we define four lithiated complexes: C-LixWS2, A-LixWS2, LixWS2@Gr, and LixWS2@ NGr. The first two complexes represent the crystalline and amorphous LixWS2 with no Gr. LixWS2@Gr denotes a composite in which two crystalline Gr layers are present in the amorphous LixWS2. In the case of LixWS2@NGr, we investigate two different systems: LixWS2@NGr-C and LixWS2@NGr-A. The former system is similar to LixWS2@ Gr except that the Gr is N-doped. We presume that the spacing between the two crystalline Gr or NGr layers becomes large after only a few cycles, so that they can be modeled as separate graphene or N-doped graphene monolayers. The Li x WS 2 @NGr-A denotes that the NGr is vigorously fragmented, so that its layered structure is completely broken. As the capacity of LixWS2@NGr-A is in better agreement with the experimental data than that of LixWS2@NGr-C, we will focus on the LixWS2@NGr-A system. For the Gr and NGr composites, a graphene sheet (66 C atoms) and a N-doped graphene sheet (60 C and 6 N atoms, corresponding to 10% N-doping) were used for our calculation, respectively. Figure S12 (Supporting Information) shows the structures of (2 × 2) supercells for the crystalline Gr and NGr. The N-doping of NGr in a supercell is modeled by one P-motif and two separate G-motifs, where a P-motif represents a pyridine-like local structure with divacancies (containing 4 N atoms with two C vacancies), and a G-motif denotes an N atom substituting the C atom.39,40 The 10% N doping level and the 2:1 ratio of N atoms in the pyridine- and Gr-like structures are consistent with the X-ray photoelectron spectroscopy (XPS) data as shown above. The graphene or N-doped graphene sheet was inserted into a supercell with 6 WS2 formula units at a weight

Figure 2. (a) CV data of WS2, WS2@Gr, and WS2@NGr at the 10th cycle, scanned over the 0.01−3 V range at a rate of 0.1 mV s−1. (b) Charge and discharge voltage profiles at the 100th cycle, determined between 0.01 and 3 V at a rate of 0.1 C (43 mA g−1). (c) Charge/ discharge capacity vs cycle number at the rate of 0.1 C. (d) Cycling performance with an increasing C rate from 0.1 C to 5 C (2.2 A g−1) and then down to 0.1 C. As the C rate changed in steps (0.1 C, 0.2 C, 0.5 C, 1 C, 2 C, 5 C, and 0.1 C), 10 discharge/charge cycles were performed for each step.

solid electrolyte interphase were formed during the 1st cycle, reversible reactions persisted over all subsequent cycles. Figure 2b shows the charge/discharge voltage profiles at the 100th cycle, measured using 0.1 C rate (=43 mA g−1). For convenience, the 1 C rate was defined as 433 mA g−1 for all samples. The charge/discharge curves of all three samples show plateaus between 2.0 and 2.4 V, which are matched with the anodic/cathodic peaks of the CV curves. Corresponding voltage profiles at other cycles are shown in Figure S6 (Supporting Information). The specific capacity (mA g−1) of WS2 in the Gr (or NGr) composite was calculated by subtracting the capacity contribution of Gr or NGr in the LIB (330 or 360 mA h g−1 as shown in Supporting Information, Figure S7) from the measured capacities based on the total mass of the composites (see the formula in Supporting Information, Experimental Details). The charge/discharge capacity versus the cycle number is plotted in Figure 2c. The WS2 nanosheets have a discharge capacity of 633 mA h g−1 after 100 cycles, corresponding to Li5.9WS2, which is higher than the theoretical capacity (Li4WS2), while WS2@Gr and WS2@NGr show even higher capacities of 780 and 963 mA h g−1, respectively. All three samples exhibit increased capacities on cycling; WS2, WS2@Gr, and WS2@NGr show 7, 20, and 27% increase between the 5th and 100th cycles, respectively. The capacities versus cycle numbers are summarized in Supporting Information, Table S2. Clearly, WS2@NGr exhibits the most significant capacity increase. Figure S8 (Supporting Information) displays the capacity of WS2@NGr on 1000 cycles, showing a rise for up to 600 cycles. The rise in the capacity has frequently been observed in other anode materials and is explained by further activation of the active materials with cycling.44,45 The following calculation will show that the vigorous fragmentation of N-doped graphene is one of the major factors that cause the capacity rise. 37930

DOI: 10.1021/acsami.8b10133 ACS Appl. Mater. Interfaces 2018, 10, 37928−37936

Research Article

ACS Applied Materials & Interfaces

that the progressive lithiations induce structural rearrangements in such a way that the Li ions gradually occupy the existing free volume without causing appreciable volume expansion in the composites. Therefore, LixWS2@Gr and LixWS2@NGr-A can be almost free of mechanical pulverization after repeated cycles. Next, we describe the stepwise volume change, δV, which is defined as V(LixWS2) − [V(Lix−1WS2) + V(Li)], where V(Li) represents the volume of a Li atom in its bcc crystal. As shown in Figure 3d, the lithiated C-LixWS2 with x ≤ 3 suffers from unfavorable volume contraction (δV < 0). In contrast, the ALixWS2 experiences volume expansion (δV > 0) for x ≤ 3. The volume difference between the two phases, δV(A-LixWS2) − δV(C-LixWS2), is the largest at x = 2 and negligible at x ≥ 3. This observation leads us to conjecture that amorphization can occur during the early lithiation steps (x = 1−2), unless the experimental charging/discharging rate is sufficiently slow. The large volume contraction (δV < 0) of LixWS2@Gr and LixWS2@NGr-A indicates that the lithiation induces significant mixing of atoms and formation of chemical bonds between them. LixWS2@Gr shows a larger fluctuation in δV with significant volume contraction at x = 3 (where ΔV/V decreases slightly) but expansion at x = 7. In contrast, LixWS2@NGr-A generally displays smaller volume contraction with less fluctuation. This observation suggests that the stepwise lithiation induces more efficient mixing of chemical species in LixWS2@NGr-A than in LixWS2@Gr. Figure 4 shows the theoretically calculated discharge curves in comparison with the experimental curves (at the 100th

ratio of WS2/(C + N) = 2:1. This ratio is larger than the value of 8:2 used in the experiments. To measure the thermodynamic stability of the system, we define the formation energies (Ef) of C-LixWS2 and A-LixWS2 according to the equation: Ef(LixWS2) = Etot(LixWS2) − xEtot(Li) − Etot(WS2), where Etot is the total energy of each component, as described in our previous work.42 A more negative value of Ef means higher thermodynamic stability. Figure 3a shows the Ef in a stepwise lithiation process. The

Figure 3. (a) Formation energy (Ef), (b) volume expansion (ΔV/V), (c) nonfree volume expansion ΔVnf/Vnf, and (d) stepwise volume contraction (δV) of LixWS2, LixWS2@Gr, and LixWS2@NGr-A as functions of x.

stability increases with x for both C-LixWS2 and A-LixWS2, reaching a maximum at x = 4. However, the negative Ef values indicate that further lithiation is still possible at least up to x = 7, especially for A-LixWS2. This result is consistent with the experimental capacity of 633 mA h g−1 (=Li5.9WS2) after 100 cycles. Now, we discuss the volume expansion, ΔV/V = (V − V0)/ V0, by defining V and V0 as the supercell volumes of LixWS2 and WS2, respectively. Figure 3b shows that the volume expands almost linearly with x for C-LixWS2 and A-LixWS2. Their volume expansions are about 150% at x = 6, where the maximum discharge capacity is achieved by the experiments. However, the expansion is smaller than the calculated values (∼280%) of other materials such as Li3P and LixMoS2 (x = 4).20,43 Remarkably, the volume expansions of LixWS2@Gr and LixWS2@NGr-A are much smaller, being 6 and 44% at the maximum capacity we calculated, as shown later. Furthermore, we introduce another measure of volume change based on the nonfree volume Vnf, which is defined by Vnf = Vs − Vol, where Vs is the summed volumes of spherical atoms (with covalent radii of rW = 1.62 Å, rS = 1.05 Å, rLi = 1.28 Å, rC = 0.76 Å, and rN = 0.71 Å) and Vol is the sum of overlapped volume between the atoms.46 The volume expansion is defined as ΔVnf/Vnf = (Vnf − V0,nf)/V0,nf, where Vnf and V0,nf are the nonfree volumes of LixWS2 and WS2, respectively. ΔVnf/Vnf displays nearly linear increase with x for all four systems (Figure 3c). This behavior is different from that of ΔV/V for LixWS2@Gr and LixWS2@NGr-A, suggesting

Figure 4. Discharge voltage vs capacity for (a) C-LixWS2 and ALixWS2 and (b) LixWS2@Gr and LixWS2@NGr-A. The experimental discharge (Exp.) voltage profiles at the 100th cycle are plotted for comparison.

cycle). The theoretical voltage was calculated by employing the centered-difference scheme: V(Li x WS 2 ) = μ(Li) − [Etot(Lix+ΔxWS2) − Etot(Lix−ΔxWS2)]/2Δx. Here, we used Δx = 1, although a smaller Δx can be used in a more accurate calculation. The theoretical and experimental curves of CLixWS2 (x = 1−7) show close agreement for x ≤ 2. However, this is not the case for x ≥ 3, probably due to the amorphization, as suggested by the volume change (Figure 4a). In contrast, the theoretical curve of A-LixWS2 shows a good match with the experimental data of WS2. The calculated capacity is 620 mA h g−1 (Li5.7WS2 at V = 0), which is quite consistent with our experimental data (633 mA h g−1). The calculated discharge curves of LixWS2@Gr and LixWS2@NGrA are also matched with the experimental data (Figure 4b). The specific capacity of WS2 was obtained using the same method as that used in the experiments; the total capacity of LixWS2@Gr (or NGr) composites was subtracted by the 37931

DOI: 10.1021/acsami.8b10133 ACS Appl. Mater. Interfaces 2018, 10, 37928−37936

Research Article

ACS Applied Materials & Interfaces

Figure 5. Structures of (a) C-LixWS2 (x = 1, 2, 5, and 6), and A-Li6WS2, (b) LixWS2@Gr, and (c) LixWS2@NGr-A at (x = 1, 6, and 9).

Figure 6. gi−j(r) for various (i,j) pairs: (a) A-LixWS2, (b) LixWS2@Gr, and (c) LixWS2@NGr-A.

contribution of 33 wt % Gr or NGr (109 and 119 mA h g−1 using the experimental data of 330 or 360 mA h g−1, respectively) and divided by 0.67. The calculated discharge curve of LixWS2@NGr-A mimics the plateau at ∼2 V in the region of 200−300 mA h g−1. The capacities of LixWS2@Gr and LixWS2@NGr-A are 743 and 915 mA h g−1, which are in reasonable agreement with our experimental data values of 780 and 963 mA h g−1, respectively. The optimized structures of C-LixWS2 (x = 1, 2, and 5) and A-LixWS2 (x = 6), predicted by our calculation, are shown in Figure 5a. The structures for all x values are shown in Figures S13 and S14 (Supporting Information). C−LiWS2 does not preserve the hexagonal phase but instead undergoes a phase transition to the monoclinic phase, in which the W−S distance increases from 2.42 to ∼2.55 Å. A Li ion is placed at the center

of four S atoms. The formed Li−S bonds have a length of ∼2.50 Å, which is longer than that of Li2S (2.08 Å), probably due to the partial W−Li bond formation (∼3.05 Å).47 The phase transition can be driven by the Li−S ionic bonds replacing the W−S bonds. C−Li2WS2 belongs to a space group (triclinic) that has the lowest symmetry. The two W atoms have a distance of 2.50 Å, corresponding to a strong quadruple W−W bond.46 The two S atoms bridge two W atoms (one above and one below) via W−S−W bonds (with the W−S bond length of 2.44 Å), and Li ions between two W chains form Li−S bonds (2.40 Å). The W atoms in C−Li5WS2 eventually form linear chains, indicating that the maximum charge capacity of C-LixWS2 is limited by the phase separation of the W chain from Li5S2. A similar formation of Mo layers on the lithiation of MoS2 was demonstrated by calculations.20 37932

DOI: 10.1021/acsami.8b10133 ACS Appl. Mater. Interfaces 2018, 10, 37928−37936

Research Article

ACS Applied Materials & Interfaces

It is also noteworthy that gmax W−Li decreases with increasing x significantly at the very early lithiation stage (x > 1). This is in strong contrast with the case of A-LixWS2 (as shown in Figure 6a), in which the peak maximum increases monotonically with x. These observations indicate that unfavorable Li−W repulsion is avoided in the LixWS2@NGr-A. Furthermore, the repulsion can be also indirectly compensated for by favorable W−C or W−N interaction, which is not possible in the case of A-LixWS2. Although less pronounced, a similar argument holds for LixWS2@Gr. Therefore, we can conclude that the enhanced charge capacities and reduced volume expansion in the two composites can be ascribed to the efficient reduction of the Li−W repulsion at high lithiation content. We have compared the results of LixWS2@NGr-A with those for LixWS2@NGr-C, as shown in Supporting Information, Figure S17. The capacity of LixWS2@NGr-C is 790 mA h g−1, which is appreciably smaller than the experimental value. This observation indicates that the higher capacity of LixWS2@NGr can be achieved through the vigorous fragmentation of the NGr layers. It is reasonable to conjecture that the significant capacity increase of WS2@NGr on cycling can be ascribed to the fragmentation of NGr leading to the amorphization of the whole composite. Our separate analysis of gmax W−Li for LixWS2@ NGr-C indicates that it behaves quite similarly to that for LixWS2@Gr in that it also increases with x monotonically, thereby indicating that the reduction of unfavorable Li−W repulsion in the LixWS2@NGr-A can be ascribed to this fragmentation (see Supporting Information, Figure S18).

For the amorphous phase, the RDF between a pair of atoms (i, j), gi−j(r), was evaluated by averaging the trajectories extracted from NVT3 simulations for 6 ps at 300 K. Figure 6a displays the gW−W(r), gW−Li(r), gW−S(r), and gLi−S(r) data for ALixWS2 (x ≤ 5), and the strong peak intensities manifest the short-range order in the corresponding atomic pairs. As the degree of lithiation increases, the W−W, S−Li, and W−Li pairs are formed following the order: gW−W > gLi−S > gW−Li, whereas max the W−S pairs are separated. The gW−W value exhibits a significant increase from 5.0 to 18.3, and its peak position (rmax) gradually decreases from 2.70 to 2.55 Å (W−W quadrupole bond length).48 The strong W−W interaction causes the W−S bonds to be replaced by the Li−S ionic interaction. Meanwhile, unfavorable W−Li repulsion becomes more and more significant with increasing x, which seems to be responsible for the stability decrease of A-LixWS2 at x > 4. This is indicated by a monotonic increase of the gmax W−Li value with x, suggesting that the repulsion plays a major role in limiting its capacity. Now, let us examine the local structural ordering in LixWS2@Gr based on the analysis of RDFs. The gW−W(r), gW−Li(r), gW−C(r), and gLi−C(r) data for x = 1, 4, 7, and 9 are displayed in Figure 6b, and the gW−S(r) and gLi−S(r) data are shown in Figure S15 (Supporting Information). The attenuation of gW−W shows that the W−W interaction is significantly weaker than that in A-LixWS2, and the W−W bond formation is insignificant. The increase of gW−C and gLi−C suggests that the weakening of W−W interaction is compensated by the competing W−C and Li−C interactions. The value of gW−Li monotonically increases, and the rmax (at the maximum g) = ∼2.8 Å, which is shorter than that for A-LixWS2 (∼3.0 Å). The unfavorable repulsion between the W−Li pairs seems to be offset by their interactions with Gr. All these observations suggest that the high discharge capacity of LixWS2@Gr correlates with the interaction of W and Li with Gr. The rmax(Li−C) value converges to ∼2.2 Å, and rmax(W− C) decreases as x increases, indicating that further lithiation leads to a denser packing through the packing of W atoms near the Gr plane. This could be the reason why the volume expansion is so small. The structure of LixWS2@Gr is shown in Figure 5b. The Gr sheet becomes rippled but mostly preserves the crystalline planarity after the composite is heated to 2000 K and then cooled down to 300 K during the simulation. The W atoms are attached to the Gr plane, and the Li ions are also packed near the Gr plane. Figure 6c shows the RDFs of LixWS2@NGr-A (x = 1, 4, 7, and 9): gW−W(r), gW−Li(r), gW−C(r), and gLi−C(r). The gW−S(r), gW−N(r), gLi−S(r), and gLi−N(r) are shown in Supporting Information, Figure S16. Figure 5c clearly shows the fragmentation of the NGr sheets. The weak W−W interaction max is comparable to that in the LixWS2@Gr. Although the gW−C max and gLi−C are further enhanced, the Li−N interaction is even stronger than the Li−C interaction, which is clearly manifested max by gmax Li−N > gLi−C and rmax(Li−N) < rmax(Li−C). In addition, the W−N interaction is weaker than the W−C interaction, which max also correlates well with gmax W−N < gW−C and rmax(W−N) ≫ rmax(W−C). Therefore, the highest capacity of the WS2@NGrA can be ascribed to the vigorous fragmentation of N-Gr leading to the amorphization of the whole composite in such a way that weakening of W−W interaction is compensated for by W−C, Li−N, and Li−C interactions, which avoids W−N interaction in the optimal way.

3. CONCLUSIONS We synthesized the WS2 nanosheets by gas-phase sulfurization of WO2.7 nanowires using H2S at 400 °C and the WS2@Gr and WS2@NGr composites by ultrasonic treatments of WS2 and RGO (or N-doped RGO). We examined the electrochemical performance of WS2, WS2@Gr, and WS2@NGr in LIBs. The WS2 nanosheets exhibited a capacity of 633 mA h g−1 (at 43 mA g−1) after 100 cycles, whereas the WS2@Gr and WS2@ NGr showed higher capacities, which are 780 and 963 mA h g−1, respectively. These systems were analyzed using extensive calculations based on the crystal structure prediction, DFT method, and AMD simulations for crystalline/amorphous LixMoS2 and amorphous LixWS2@Gr and LixWS2@NGr-A complexes. The analyses of the volume expansion, the RDFs, and the discharge capacity suggest that the amorphization occurs during the initial lithiation steps. The calculated discharge curves of the amorphous phase are well matched with the experimental ones. The calculated capacities are 620, 743, and 915 mA h g−1 for WS2, WS2@Gr, and WS2@NGr, respectively, which are close to the experimental values. The higher capacity of WS2@Gr and WS2@NGr originates from the strong interaction of W and Li atoms with Gr or NGr, which (1) suppresses the aggregation of W atoms and (2) reduces unfavorable Li−W repulsion at early lithiation stages. All these interactions result in a better packing of atoms than that in WS2, as shown by the small volume expansion (theoretically calculated to be 150, 6, and 44% at maximum capacities for WS2, WS2@Gr, and WS2@NGr, respectively). The small volume expansion should cause little mechanical pulverization after repeated cycles. Our findings should stimulate further experimental work on the capacity improvement of 2D TMD materials. 37933

DOI: 10.1021/acsami.8b10133 ACS Appl. Mater. Interfaces 2018, 10, 37928−37936

Research Article

ACS Applied Materials & Interfaces

4. CALCULATION METHODS The search for the crystal structure of LixWS2 (x = 1−7) with the lowest enthalpy was performed using the evolutionary structure predictor method in the USPEX package.49−51 At each stoichiometry of LixWS2, ∼900 initial structures were generated, and the total energy for each was calculated based on the DFT implemented in the Vienna ab initio simulation package (VASP).52,53 The projected augmented plane wave approach was employed with a plane-wave kinetic energy cutoff of 520 eV and reciprocal-space k-point meshes of 2π × 0.045 Å−1.54 Ten most stable structures were selected, for which more accurate structure optimization was done with denser k-point meshes with a resolution of 2π × 0.03 Å−1. During the optimization, the structures were refined until the average force was