Research Article Cite This: ACS Appl. Mater. Interfaces 2018, 10, 13442−13451
www.acsami.org
Utilizing van der Waals Slippery Interfaces to Enhance the Electrochemical Stability of Silicon Film Anodes in Lithium-Ion Batteries Swastik Basu,†,○ Shravan Suresh,†,○ Kamalika Ghatak,‡ Stephen F. Bartolucci,§ Tushar Gupta,† Prateek Hundekar,∥ Rajesh Kumar,⊥ Toh-Ming Lu,# Dibakar Datta,‡ Yunfeng Shi,*,∥ and Nikhil Koratkar*,†,∥ †
Department of Mechanical, Aerospace and Nuclear Engineering, Rensselaer Polytechnic Institute, Troy, New York 12180, United States ‡ Department of Mechanical and Industrial Engineering, Newark College of Engineering, New Jersey Institute of Technology (NJIT), Newark, New Jersey 07102, United States § US Army Armaments Research Development and Engineering Center, Watervliet, New York 12189, United States ∥ Department of Materials Science and Engineering, Rensselaer Polytechnic Institute, Troy, New York 12180, United States ⊥ University School of Basic & Applied Sciences, Guru Gobind Singh Indraprastha University, New Delhi, 110078, India # Department of Physics, Applied Physics and Astronomy, Rensselaer Polytechnic Institute, Troy, New York 12180, United States S Supporting Information *
ABSTRACT: High specific capacity anode materials such as silicon (Si) are increasingly being explored for next-generation, high performance lithium (Li)-ion batteries. In this context, Si films are advantageous compared to Si nanoparticle based anodes since in films the free volume between nanoparticles is eliminated, resulting in very high volumetric energy density. However, Si undergoes volume expansion (contraction) under lithiation (delithiation) of up to 300%. This large volume expansion leads to stress build-up at the interface between the Si film and the current collector, leading to delamination of Si from the surface of the current collector. To prevent this, adhesion promotors (such as chromium interlayers) are often used to strengthen the interface between the Si and the current collector. Here, we show that such approaches are in fact counterproductive and that far better electrochemical stability can be obtained by engineering a van der Waals “slippery” interface between the Si film and the current collector. This can be accomplished by simply coating the current collector surface with graphene sheets. For such an interface, the Si film slips with respect to the current collector under lithiation/delithiation, while retaining electrical contact with the current collector. Molecular dynamics simulations indicate (i) less stress build-up and (ii) less stress “cycling” on a van der Waals slippery substrate as opposed to a fixed interface. Electrochemical testing confirms more stable performance and much higher Coulombic efficiency for Si films deposited on graphene-coated nickel (i.e., slippery interface) as compared to conventional nickel current collectors. KEYWORDS: lithium-ion battery, silicon film anode, van der Waals interface, interfacial slip, stable cycle life
■
INTRODUCTION
high capacity LIBs, much attention has been focused on higher capacity anode materials. Of the various possible materials currently being explored, silicon (Si) exhibits the highest gravimetric (∼4200 mAh/g) and volumetric (∼9800 mAh/ cm3) capacities.4−8 One of the first attempts to deploy Si anodes in LIBs was as films deposited on Cu foil current collector substrates.4 However, mechanical instabilities have consistently led to poor electrochemical performance of such Si film anodes. The principal reason for mechanical failure is stress build up at the interface between the expanding/contracting Si film and the
Lithium-ion batteries (LIBs) have been extensively utilized in portable electronics because of their relatively high energy density and long life cycle.1 The growth in demand for these batteries is increasing with the proliferation of consumer electronics products. In fact, the high storage capacity and better efficiency of Li-ion batteries may allow their use in various electric grid applications, which can improve the reliability of energy harvested from renewable sources.2 Also, Li-ion batteries will significantly reduce greenhouse gas emissions, if a substantial part of the gasoline powered transportation system is replaced by electric vehicles (EVs).3 As the theoretical capacity of traditional graphite anodes (∼370 mAh/g) cannot satisfy the increasing requirements for © 2018 American Chemical Society
Received: January 5, 2018 Accepted: April 5, 2018 Published: April 5, 2018 13442
DOI: 10.1021/acsami.8b00258 ACS Appl. Mater. Interfaces 2018, 10, 13442−13451
Research Article
ACS Applied Materials & Interfaces
Figure 1. a1 and a2 show snapshots of the simulation systems with a rigid nonslip substrate (a1) and a rigid slip substrate (a2) during a lithiation and delithiation cycle. The snapshots are taken prior to lithiation, at a highly lithiated stage (Li/Si = 3.5) and a highly delithiated stage (Li/Si = 0.3), from top to bottom. The red atoms are silicon, while the blue atoms are lithium. b1 and b2 show the normal stress σxx profile at Li/Si = 3.5 (black lines) and Li/Si = 0.3 (orange lines) for the systems with nonslip (b1) and slip substrate (b2), respectively. σxx is averaged over the simulation system excluding the substrate every 0.55 nm along the x-direction. c1 and c2 show the shear stress τxz profile at Li/Si = 3.5 (black lines) and Li/Si = 0.3 (orange lines) for the systems with nonslip (c1) and slip substrate (c2), respectively. τxz is averaged over the region within 1 nm above the bottom substrate (as indicated by the box) every 0.55 nm along the x-direction. The error bar shows the stress fluctuation over 10 independent stress measurements.
rigid (fixed) current collector substrate.5 The most popular research direction to counter this challenge has been focused on studying various kinds of nanostructured Si materials. Various studies have been done by (i) synthesizing sophisticated nanostructured Si, including nanowires,6,7 nanotubes,8 nanopillars, nanospheres,9,10 etc., (ii) embedding of nanostructured Si particles into an active or inactive matrix,11−15 (iii) hybridizing with carbonaceous materials to electrochemically improve the performance,16−20 (iv) exploring new polymer binding materials,21,22 and (v) core−shell Si nanoparticles.23−25 Perhaps the most successful approach to date has involved conventional carbon coating on nano-Si materials.26−31 This can be accomplished by directly mixing Si nanopowder with graphite, mesocarbon microbeads or hard carbon to achieve better cycle life.26 Another variation of this approach is a yolk−shell structure, where Si nanoparticles were sealed inside carbon shells with the void space allowing the Si to expand and contract, leading to a more stable solid electrolyte interface.27 Many such core−shell
structures have been further designed using different forms of carbon.28−31 Even though improvement in the design of nanostructured Si has led to better performance and cycle life, several disadvantages persist. Si nanoparticle based anodes offer much less packing density than Si films, which results in lower volumetric energy density.32 Moreover, yolk−shell type nano-Si architectures are complex in design and expensive to manufacture. Further, it should be noted that portable electronics and even batteries for automotive applications are “volume” limited rather than weight limited.32 Therefore, if Si film anodes can be rendered stable in a LIB, then this has important implications for the development of batteries with high volumetric performance. Here, we demonstrate a pathway to achieving such an outcome by using graphene sheets to engineer a van der Waals “slippery” interface between the Si-film anode and the metal current collector substrate. 13443
DOI: 10.1021/acsami.8b00258 ACS Appl. Mater. Interfaces 2018, 10, 13442−13451
Research Article
ACS Applied Materials & Interfaces
Figure 2. Setup for the interface system of a-Si on a-Si, Cu (111), Ni (111), and monolayer graphene substrates used for carrying out the first-principles (DFT) calculations.
■
fatigue damage, leaving the Si film susceptible to fracture and pulverization. On the other hand, one can see that the graphene substrate creates a slippery van der Waals interface with the a-Si layer, which allows the a-Si network to expand freely upon lithiation and shrink freely upon delithiation (Figure 1, right). Thus, there is considerably lower buildup of normal and shear stress in comparison to that of the no-slip substrate. Consequently, the resulting stress cycling during lithiation/delithiation (Figure 1, right) is much less, which would greatly alleviate fatigue induced crack propagation and failure. These results suggest that controlled mechanical failure (i.e., interfacial slippage) at the interface between the Si film and the current collector can greatly alleviate stress buildup and stress cycling in the Si film, which is expected to enhance its stability. It should be noted that such an interfacial slip would not result in electrical disconnection of the Si film as it would remain in electrical contact with the substrate. Interface Properties. Having established that the structural integrity of the a-Si electrode is sensitive to the a-Si/substrate interface (strong adhesion for no-slip substrate, while low adhesion for slippery substrate), we will examine the interfacial adhesion properties of a number of a-Si/substrate combinations. In general, the work of separation Wsep provides a useful measure of the strength with which two materials adhere to each other, and the larger the strength of adhesion, the higher will be the stress buildup at the interface due to lithiation induced volume expansion and vice versa. The standard definition of Wsep for an interface between two materials is
RESULTS AND DISCUSSION Stress Evolution. The evolution of the normal and shear stress profile is analyzed using molecular dynamics (MD) simulations upon lithiation and delithiation of an amorphous Si (a-Si) film deposited on different substrates. In order to compute the profiles of the normal and shear stress, the system was divided into thin slabs with equal thickness of 0.55 nm along the xdirection of the simulation box. For computation of normal stress, the height of the slabs along the z-direction is the height of the system (2.15 nm) excluding the fixed substrate, while for computation of the shear stress τxz slabs of height 1 nm above the substrate are chosen. Figure 1 shows the profiles of the stress tensor components along the normal (σxx) and shear (τxz) directions in two systems with no-slip substrate (fixed a-Si substrate) and slip substrate (monolayer graphene). Here, positive (negative) values of the normal stress components indicate compressive (tensile) stresses that act to contract (expand) the film, while positive (negative) values of the shear stress components indicate shear stress tending to turn the element counterclockwise (clockwise). It is observed from the σxx profiles that at a highly lithiated stage (Li/Si = 3.5) there is a considerable buildup of compressive stress in the central region of the film where the Li−Si is in contact with the fixed substrate (Figure 1, left panels). This is because the a-Si network tries to expand in volume upon lithiation, while the fixed substrate constrains this volume expansion, resulting in considerable shear stress distribution between the substrate and the a-Si. The shear stress from the top and bottom substrate together lead to overall compressive stress in the a-Si in the middle. This rather high compressive stress from the substrate acts to counter the thermodynamic driving force toward lithiation, such that the a-Si interior region contains much less lithium atoms than the outer region. The stress distribution roughly reverses its sign during delithiation. This is because the expanded a-Si network tries to revert to its original density, while the sticky substrate prevents the a-Si from shrinking. Thus, the shear distribution between the no-slip substrate and the a-Si changes sign, while the overall stress along the x-direction becomes tensile at the center of the aSi film. It should be noted that both the compressive stress reached during lithiation and the tensile stress reached during delithiation are significant. Such large stress cycling over extended lithiation/delithiation cycles will invariably lead to
Wsep =
tot E1tot + E2tot − E12 A
tot where Etot i is the total energy of slab i, E12 is the total energy of the interface system with the slab materials, and A is the total area of the interface.33 We have used a-Si/a-Si, a-Si/Cu, a-Si/Ni, and aSi/Graphene vacuum interface models (Figure 2) to calculate Wsep for a-Si as a function of different substrates using firstprinciple density functional theory (DFT) calculations. The number of metal layers used in the calculations were determined after studying the convergence of Wsep as a function of slab thickness (Figure S1 in Supporting Information). The results for Wsep for the different interfaces are summarized in Figure 3 and indicate that the interface energy for the a-Si/
13444
DOI: 10.1021/acsami.8b00258 ACS Appl. Mater. Interfaces 2018, 10, 13442−13451
Research Article
ACS Applied Materials & Interfaces
the cycle life of the a-Si film when compared to that of traditional metal current collectors such as Ni and Cu. Electrochemical Testing. To validate our prediction that graphene substrates could be used to alleviate stress buildup and prolong the cycle life of Si films, we performed constant current charge−discharge cycling of a-Si films deposited on metal (Ni) and graphene-coated Ni current collector substrates. Since the work of separation of a-Si is similar for both Ni and Cu substrates, one can expect similar results on Cu as well, and this should be the subject of future study. The graphene-coated Ni foils used in this study were purchased from ACS Materials Inc. Crosssectional SEM imaging of the graphene-coated Ni current collectors (Figure S2A) was carried out. The SEM imaging indicates that the multilayer graphene coating is ∼1 μm thick. The graphene film is of high quality as indicated by its Raman spectra. From Figure S2B, it is evident that the Raman D peak, related to disorder, is miniscule compared to the G peak (ID/IG = 0.0014). In addition, there is no D′ peak (another disorder related peak), nor is the D + D′ peak present in the spectrum. All these signs indicate that the defect density in the graphene film is negligibly low. Figure 4a and b shows the schematics of the Si sputter deposition on bare Ni foil and graphene-coated Ni foil. It should be noted that Si films deposited by sputtering are amorphous in nature.32 Scanning electron microscopy (SEM) imaging of the Si film deposited on the Ni foil (Figure 4c) shows that the Si film mimics the roughness of the Ni foil. On the other hand, Si films deposited on graphene-coated Ni (Si-Gr) in Figure 4d show graphene-like morphology with grain boundaries. For the theoretical calculations and molecular dynamics (MD) simulations, we used a monolayer graphene sheet at the interface between the Si film and the Ni current collector. This is because
Figure 3. Interfacial work of separation for relaxed a-Si/a-Si, a-Si/Cu, aSi/Ni, and a-Si/graphene interfaces.
graphene system is around one-fifth of that for the Cu and Ni substrates and less than half of that for the a-Si/a-Si substrate. This indicates that compared to traditional metal current collectors such as Cu and Ni, graphene offers a much weaker (slippery) interface with the amorphous Si film. The greater work of separation of a-Si/Cu and a-Si/Ni relative to a-Si/Si indicates that the “non-slip” condition is more valid for a-Si/Cu and a-Si/ Ni relative to a-Si/Si. Since MD results (Figure 1) indicate lower stress buildup during the cycling of an a-Si/Graphene interface compared to that of the a-Si/a-Si interface, one can conclude that the performance of the a-Si/Graphene interface would be even better when compared to that of the nonslipping (i.e., fixed) a-Si/ Cu and a-Si/Ni interfaces during cycling. Therefore, slippage during lithiation/delithiation of the amorphous Si film with respect to the graphene substrate should greatly alleviate stress buildup and stress cycling at the interface and therefore prolong
Figure 4. Schematics of Si sputter deposition on bare Ni foil (Si−Ni) (a) and multilayer graphene coated Ni foil (Si-Gr) (b). SEM images of the (c) Si− Ni electrode and (d) Si-Gr electrode before electrochemical cycling, mimicking the surface roughness of Ni and graphene, respectively. 13445
DOI: 10.1021/acsami.8b00258 ACS Appl. Mater. Interfaces 2018, 10, 13442−13451
Research Article
ACS Applied Materials & Interfaces
Figure 5. Electrochemical characterization of Si-Gr electrode. (a) Cyclic voltammograms acquired at a scan rate of 0.5 mVs−1. The CVs show that Si is the active anode material in both Si−Ni and Si-Gr. (b) Comparison of the areal average capacity of the graphene coating with and without the Si film indicating negligible contribution of the graphene coating to the charge capacity of the anode. (c) Plots indicating an improvement in Coulombic efficiency of the Si-Gr. (d) Galvanostatic charge−discharge curves. (e) Electrochemical cycling performance of Si-Gr showed an enhanced capacity of ∼800 mAh/g compared to a capacity of ∼190 mAh/g exhibited by Si−Ni. The galvanostatic cycling was performed at a current density of ∼1.8 A/g.
van der Waals forces are short-range, and hence, there is negligible interaction between Si and the second graphene layer (i.e., the layer below the nearest graphene layer). Therefore, purely from a “theoretical” perspective, a monolayer graphene
sheet is sufficient to capture the interfacial stress evolution and illustrate the principle of van der Waals slip. While theoretically we can use a defect-free and perfect graphene sheet, experimentally this is not the case. Monolayer 13446
DOI: 10.1021/acsami.8b00258 ACS Appl. Mater. Interfaces 2018, 10, 13442−13451
Research Article
ACS Applied Materials & Interfaces
Figure 6. (a) Schematic representation of a fracture near the interface due to electrochemical cycling of Si film on bare Ni foil. (b) Slip at the interface due to the graphene coating on Ni suppresses Si film delamination near the interface on electrochemical cycling. (c) Wavelength dispersive spectra acquired at ∼10 keV show a strong presence of Ni in Si−Ni film in comparison to the Si-Gr film after cycling. (d) Postcycling SEM image of the Si-Gr electrode after 150 cycles indicating the toughened mud-cracked structure. (e) Postcycling SEM image of the Si−Ni electrode after 150 charge/ discharge cycles. Si has delaminated exposing the bare Ni surface with large debris on the surface.
The peak at ∼0.2 V during the cathodic scan and ∼0.56 V during the anodic scan represent the electrochemical lithiation/ delithiation reactions of Si. This indicates that only Si is the active material and that the contribution of graphene is negligible. To verify the capacity contribution of graphene, we also performed electrochemical tests on pristine graphene films and compared the capacity of Si-Gr with the pristine graphene films at a current density of 1.8 A g−1. As shown in Figure 5b, the capacity of the pristine graphene film was negligible as compared to the Si-Gr films. The CV curves in Figure 5a also show that the area under the curve of Si-Gr films is higher and that the redox
graphene grown by chemical vapor deposition (CVD) is defective and imperfect with a multitude of topological and vacancy defects32 not to mention wrinkles/folds and tears. Therefore, for the experiments, the graphene film that was deployed was multilayered. The use of such a multilayered and relatively defect-free (Figure S2) graphene film ensures that there is minimal interaction between the Ni substrate and the Si film, and therefore, a “slippery” interface between the Si layer and the graphene film is ensured. The cyclic voltammograms (CV) of Si and Si-Gr films were acquired at a scan rate of 0.5 mV s−1 and are shown in Figure 5a. 13447
DOI: 10.1021/acsami.8b00258 ACS Appl. Mater. Interfaces 2018, 10, 13442−13451
Research Article
ACS Applied Materials & Interfaces
show a strong presence of Ni in the Si film, which is unlike the SiGr film. This indicates that the Si film which remains on the graphene-coated Ni after cycling is thicker than the Si film that remains on the Ni foil. Furthermore, the SEM images in Figures 6d,e that were acquired post-electrochemical cycling (after 150 charge−discharge steps) indicate the presence of pulverized debris on the Si films on Ni, whereas mud-cracked Si films that are still intact can be seen in the case of cycled Si-Gr films. This was also confirmed by high resolution SEM imaging (Figure S5) of the cycled Si-Ni and Si-Gr electrodes. Mud-cracked Si films that are still intact can be seen in the case of cycled Si-Gr films, whereas the Si film is seen to be delaminated, exposing the Ni substrate in the case of cycled Si-Ni films (Figure S5). The presence of the graphene layer below the mud-cracked Si film was confirmed by Raman spectroscopy acquired after electrochemical cycling (150 charge−discharge cycles) showing the characteristic D (∼1360 cm−1) and G (∼1580 cm−1) band peaks of graphene in the Si-Gr film (Figure S6). Both the Si and Si-Gr films are unprotected from the top, which implies that there will be losses due to pulverization and an unstable solid electrolyte interface (SEI). Hence, a capacity fade can be observed in both Si and Si-Gr films. The reduction of specific capacity and CE at ∼25 cycles can be ascribed to the loss of active Si due to electrical disconnections. Further, the fractured surfaces near the interface of Si−Ni also re-expose the surface of Si film to the electrolyte, thereby augmenting the growth of SEI, which further drops the CE. On the contrary, SiGr films which slip at the interface remain unconstrained, and therefore, the stresses at the interface are minimized. Hence, in the case of Si-Gr films the fade of the specific capacity and the CE is significantly reduced. Thus, we see that the bottom interface (with the current collector) plays a major role in the electrochemical stability of Si films in LIB anodes. Further improvements in electrochemical stability and Coulombic efficiency can be expected as the top interface with the electrolyte is also protected by draping a graphene film over the a-Si film as shown in ref 32.
peaks are sharper. The area under the CV curves indicates charge storage capacity, and the intensity of the CV peaks is proportional to the charge transfer due to the electrochemical reactions. Thus, the enhancement of the area under the curve and the increase in intensity of the currents are due to the improved charge transfer kinetics because of the graphene buffer layer at the interface. Further, due to the presence of graphene, the lithiation/delithiation reactions of Si films are also enhanced by reducing the interfacial charge transfer resistance. Hence, although the direct capacity contribution of the graphene buffer layer is negligible, it indirectly increases the charge storage capacity due to an enhanced “electrical” interface with the Si film. Such an enhanced electrical interface has also been reported for Si nanowire anodes deposited on graphene substrates.33 The galvanostatic charge/discharge cycling test results are shown in Figure 5d,e. The baseline Si−Ni electrode exhibited a rapid drop in capacity after ∼25 charge/discharge steps. To verify this, several coin cells were tested to check for repeatability (Figure S3,S4). Interestingly, it can also be seen in Figure 5c that the Coulombic efficiency (CE) followed the same trend as the curve in Figure 5e, i.e., after ∼25 cycles it dropped rapidly for Si films (average CE of ∼98% after 50 cycles). On the other hand, the Si-Gr films exhibited an excellent average CE of ∼99.2%. As seen in Figure 5e, the specific capacity of Si films decayed rapidly, retaining only ∼190 mAh g−1 after 150 cycles. The Si-Gr films were far more stable with a capacity retention of ∼800 mAh g−1 after 150 cycles. The volumetric capacity of the Si-Gr film averaged over 150 cycles is ∼5462 mAh cm−3, which is far superior to that of conventional32 graphitic anodes (∼400 mAh cm−3). The electrochemical characterizations confirm the superior cycle life of Si films due to the presence of a graphene buffer layer. The graphene layer creates a slippery interface, which plays a key role in preventing fracture due to the volume expansion of Si films. After the first few cycles, Si films form a mud-cracked surface, which is typically observed during their electrochemical cycling. This results in the formation of isolated Si islands, which will expand/contract during lithiation/delithiation. In the case of Si films deposited on Ni foil, there is strong adhesion at the Si−Ni interface. During lithiation of Si films, the volume expansion is restrained by interfacial adhesion, which results in the buildup of stress in Si films. This causes fracture near the Si−Ni interface and accelerates the delamination of Si film (Figure 6a). In the case of Si-Gr films, the graphene buffer layer allows the Si film to slip. Hence, the expansion of Si in the presence of a graphene buffer is relatively unconstrained, and the fracture due to adhesion stresses is mitigated (Figure 6b). Thus, we can see that due to strong adhesion on metal current collectors, Si films would be eviscerated rapidly due to mechanical stresses. On the other hand, a slippery van der Waals interface due to its low interfacial adhesion mitigates fracture. Our results indicate that such low energy interfaces can be engineered using graphene buffer layers and that this concept could also be extended to other high specific capacity anode materials (such as As, Sb, P, Al, etc.) that undergo large volume expansion on lithiation. Electron microprobe analysis was performed after electrochemical cycling to confirm the delamination of Si films. The composition maps of Si and Si-Gr films exhibited a striking difference. In Si films, the Ni current collector can be clearly seen after cycling, which means that the Si film has delaminated during electrochemical cycling. In contrast, a thick Si film can be seen in Si-Gr film samples after 150 charge/discharge cycles. As shown in Figure 6c, the wavelength dispersive spectra acquired at ∼10 keV
■
CONCLUSIONS To summarize, we have used MD and DFT simulations to explore the fundamental reasons behind the failure of bulk Si on a typical metal (i.e., rigid or fixed) substrate. Further, a graphene draped metal substrate has been introduced for the a-Si film, which serves to “unconstrain” the Si film and has been shown to significantly reduce the stress buildup and stress cycling at the interface. The concept has been proven experimentally where Si has been deposited on multilayer graphene-coated Ni, which provides a better specific capacity and a more stable Coulombic efficiency as an anode than Si deposited on Ni foil. Such a Si film with a slippery van der Waals interface has been shown to be stabilized for up to 150 cycles with a capacity of ∼800 mAh g−1, whereas the Si film deposited on Ni could retain only 190 mAh g−1. With such densely packed Si films, an average volumetric capacity of ∼5462 mAh cm−3 was achieved with the Si film deposited on graphene, which is ∼10-fold higher than that of conventional graphitic anodes. Graphene was used as the substrate on top of the current collector, and no protective layer on the Si-electrolyte interface was used in order to demonstrate that even without attempting to stabilize the SEI with a protective graphene drape, it is possible to improve the performance and cyclic stability by preventing delamination at the Si/current−collector interface, as predicted by the MD and DFT simulations. 13448
DOI: 10.1021/acsami.8b00258 ACS Appl. Mater. Interfaces 2018, 10, 13442−13451
Research Article
ACS Applied Materials & Interfaces In the MD simulations, the fixed substrate was taken to be Si, rather than the typical Ni and Cu substrates which are used in experiments. This is because in MD, we are yet to develop a reactive potential involving Cu/Ni, Si, and Li, but the strong Si− Si interaction essentially demonstrates the same physical restraints of a fixed (or rigid) interface. A comparative study of the evolution of stress profile and the interface properties gives quantitative evidence of the fundamental reasons that lead to better performance of the Si anode when lithiated on a substrate with weak van der Waals interaction. Our results indicate that (i) “slip” is observed between the graphene layer and the expanding silicon, (ii) less stress buildup is present, and (iii) less stress “cycling” is observed when Si is supported on graphene. These predictions were validated by experiments and indicate that Si films with superior volumetric capacity, Coulombic efficiency, and cycling stability can be engineered by exploiting van der Waals slippery interfaces between the Si films and the current collector substrates. In this work, we have focused on thin Si films (mass loading of ∼0.05 mg/cm2) to show the proof-of-principle. For thicker Si films, a layer-by-layer deposition strategy with multiple graphene films as interlayers might be necessary, and this should be the focus of future work. The concept of using van der Waals slippery interfaces to enhance electrochemical stability could also in principle be extended to other high specific capacity anode materials (such as As, Sb, P, Al, etc.) that undergo large volume expansion on lithiation.
■
Once the lithium atoms taken out are in vacuum, they are deleted, and the incrementally delithiated system is allowed to relax under an NVT ensemble for 10 ps. Because of the computational expense associated with the ReaxFF force field, the initial system size is limited to around 1000 atoms. The temperature of 300 K is maintained by a Nose Hoover thermostat with a damping parameter of 0.01 fs−1. DFT Calculations. First principle calculations were carried out to study the interface properties of a-Si with different substrates. In order to create the a-Si, ab initio molecular dynamics (AIMD) simulations were performed at a very high temperature of 1200 K in an NVT ensemble with a Nose thermostat for around 5000 MD time steps, with an interval of 1 fs. The heating process was followed by a rapid cooling process to 298 K (room temperature) within 5000 MD time steps. Upon the end of the final RT AIMD calculation, the structure was further relaxed using DFT in order to achieve an energy minimized structure. The mass density and structural signatures such as the radial distribution function (RDF) of the a-Si were observed with the dominant features consistent with previous quantum calculations.38 To calculate the interface energy, individual slab models of a-Si and the different substrates were created by adding ∼12 Å vacuum to the z-direction in the simulation cell. The supercell containing the interface is constructed by placing the a-Si and different substrates together and adding ∼14 Å of vacuum along the zdirection. Both interface models were relaxed using DFT without any cell relaxation. This methodology was adapted from a previously reported study.39 In this study, four different types of substrates have been considered, namely, a-Si, Cu (111), Ni (111), and graphene. We generated a-Si, graphene, Ni, and Cu substrates first with similar cell dimensions, and then we accordingly created a-Si samples of 64 Si atoms each with slightly different dimensions according to the different substrates. This gives rise to distinctive a-Si samples for a-Si, graphene, Ni (111), and Cu (111). After the end of the DFT relaxation, each of the a-Si’s RDF was checked and verified in order to get the confirmations of their amorphous nature. An accurate estimate of long-range van der Waals (vdW) interactions is important to study the a-Si/graphene interface, which is not properly calculated by conventional DFT methods, using the standard (semi)local exchange correlation functionals such as the local density approximation (LDA)40 or the generalized gradient approximation (GGA).41 An important vdW-correction approach is the vdW-density functional (vdW-DF)42 which combines nonlocal correlations directly within a DFT functional. All DFT based calculations in this study have been performed using the optPBE functional within the vdW-DF family, implemented in the VASP package by Klimes and co-workers.43−45 The plane-wave energy cutoff for different interfaces (for all the calculations) was taken as 550 eV. The convergence tolerance for the electronic relaxation was 10−5 eV/cell, and the total energy was calculated with the linear tetrahedron method with Blochl corrections. All structural relaxations employed conjugate gradient methods to minimize the total energy of the structure, and the required Hellmann−Feynman force on each atom was less than 0.02 eV/Å. For all our interface calculations, 3 × 3 × 1 Γ-centered k-meshes were employed. In this study, we relaxed all atoms and have not fixed any atomic positions. Experiment. Silicon films of ∼200 nm thickness were deposited on bare Ni foils and graphene-coated Ni foils by DC sputtering in an AJA sputter system. An N-type Silicon (Plasmaterials Inc.) target (∼2 in. in diameter) was used. The base pressure of the chamber was ∼1 × 10−7 Torr, and the working pressure was ∼3 mTorr. Scanning electron microscopy was performed on an FEI Nanolab 600i Dual Beam SEM operating at 10 kV. The mass loading of the silicon electrode is ∼0.05 mg/cm2. The deposited Si films were tested as anodes in 2032 type coin cells. The cells were assembled inside a glovebox (Mbraun Labstar) with an oxygen and moisture content of
