Research Article www.acsami.org
Nanostructured Carbon Nitride Polymer-Reinforced Electrolyte To Enable Dendrite-Suppressed Lithium Metal Batteries Jiulin Hu,†,‡ Jing Tian,†,‡ and Chilin Li*,† †
State Key Laboratory of High Performance Ceramics and Superfine Microstructure, Shanghai Institute of Ceramics, Chinese Academy of Sciences, Shanghai 200050, China ‡ University of Chinese Academy of Sciences, Beijing 100039, China S Supporting Information *
ABSTRACT: Lithium metal batteries (LMBs) containing S, O2, and fluoride cathodes are attracting increasing attention owing to their much higher energy density than that of Li-ion batteries. However, current limitation for the progress of LMBs mainly comes from the uncontrolled formation and growth of Li dendrites at the anode side. In order to suppress dendrite growth, exploring novel nanostructured electrolyte of high modulus without degradation of Li−electrolyte interface appears to be a potential solution. Here we propose a lightweight polymer-reinforced electrolyte based on graphitic carbon nitride (g-C3N4) mesoporous microspheres as electrolyte filler [bis(trifluoromethanesulfonimide) lithium salt/di(ethylene glycol) dimethyl ether mixed with g-C3N4, denoted as LiTFSI-DGM-C3N4] for the first time. This nanostructured electrolyte can effectively suppress lithium dendrite growth during cycling, benefiting from the high mechanical strength and nanosheet-built hierarchical structure of g-C3N4. The Li/Li symmetrical cell based on this slurrylike electrolyte enables long-term cycling of at least 120 cycles with a high capacity of 6 mA·h/cm2 and small plating/stripping overpotential of ∼100 mV at a high current density of 2 mA/cm2. g-C3N4 filling also enables a separator(Celgard)-free Li/FeS2 cell with at least 400 cycles. The 3D geometry of g-C3N4 shows advantages on interfacial resistance and Li plating/stripping stability compared to its 2D geometry. KEYWORDS: Li metal batteries, Li dendrite suppressing, hybrid electrolyte, nanostructured electrolyte, carbon nitride polymer
1. INTRODUCTION Lithium metal batteries (LMBs) combined with S, O2, or metal fluoride cathodes are receiving increasing attention in view of their much higher energy density than Li-ion batteries (LIBs).1−3 These Li-free cathodes require the use of Li metal anode, which has the highest theoretical capacity of 3860 mA· h/g and low electrochemical potential and can well match with these conversion-type cathodes.4,5 However, the safety issue of facile formation and growth of Li dendrites during cycling still remains unsolved. Recently some strategies have been resorted to in order to inhibit Li dendrite growth or address uneven Li electroplating. Similar to the confinement of sulfur in nanopores of a conductive framework, accommodating melted Li into a conductive scaffold with lithiophilic surface by melt infusion method appears to be a promising route to construct flexible dendrite-free Li metal anode with small volume change.6,7 Modifying the surface shape and coating of Li metal can increase the surface area of Li and lower the local current density, which is beneficial for smoother Li plating/ stripping with better kinetics performance.8−13 The architecture of nanostructured current collectors [e.g., Cu nanowire, threedimensional (3D) porous framework] instead of two-dimensional (2D) planar ones would improve the deposition of Li metal when charging LMBs with desired Coulombic efficiency and cycling stability.14−16 Another strategy to suppress Li © 2017 American Chemical Society
dendrite growth is to modulate the nature, salt concentration, and additives of electrolytes.17 Ionic liquid (IL) or highly concentrated electrolytes based on bis(fluorosulfonyl)imide anion (FSI−) enable a high rate and stable cycling.18−21 In situ formation of a robust solid−electrolyte interface (SEI), as a consequence of FSI− reduction on Li surface, is responsible for nondendritic electroplating. It is found that LiF is the main component of desired SEI, evolved from the reaction between FSI− and Li.20 An intentional addition of lithium halides would lead to a salt-reinforced electrolyte with high modulus.22,23 Also as SEI promoters, lithium polysulfide and LiNO3 additives are often employed in Li−S batteries to prevent the Li anode from corroding.24,25 Their synergetic effect guarantees a longer-life Li deposition/dissolution.26 The strategy from the separator side is to introduce polar surface functional groups on separator fiber surface in order to allow homogeneous Li deposition.27 Solid or slurrylike electrolytes are expected to have higher moduli than separator-supported electrolytes and to enable more effective inhibition of Li dendrite growth.28−30 However, all solid electrolytes often suffer from lower ionic conductivity and unwetted electrode−electrolyte interface. Quasi-solid or Received: January 11, 2017 Accepted: March 14, 2017 Published: March 14, 2017 11615
DOI: 10.1021/acsami.7b00478 ACS Appl. Mater. Interfaces 2017, 9, 11615−11625
Research Article
ACS Applied Materials & Interfaces
Figure 1. (a) Scheme for synthesizing g-C3N4 porous microspheres by thermal polycondensation of melamine−cyanuric acid supramolecular aggregates with their precursor solutions kept at 60 °C. (b, c) SEM images of g-C3N4 spheres at different scales. (d−f) TEM images of g-C3N4 spheres at different scales. (g) Photos of as-synthesized g-C3N4 powder and its composite with LiTFSI-DGM to form slurrylike electrolyte. g-C3N4 hollow microspheres are about 5 μm in size and consist of numerous nanosheets with a thickness of ∼10 nm. The surface of g-C3N4 aggregate is also void-rich.
semiconductor often used as photocatalyst or chemical catalyst.38 Since the s-triazine ring (C3N3) is aromatic, the conjugated, two-dimensional polymer of s-triazine would tend to form p-conjugated planar layers like that of graphite. The tris-triazine ring structure and its high degree of condensation make g-C3N4 possess high thermal (stable up to 600 °C in air) and chemical stability (e.g., invulnerable against acid, base and organic solvents). Furthermore, g-C3N4 is metal-free and therefore lightweight, and its synthetic methods and functional decoration strategies are rich. The intrinsic polarity and semiconductor property of g-C3N4 may be helpful to modulate the spatial and interface distribution of ionic charge carriers in electrolyte.39 The desired surface doping and grafting of functional groups are expected to immobilize dissolved electroactive species from the cathode side.38 Its mechanical strength (shear modulus ∼21.6 GPa) is much higher than that of most conventional polymer materials (e.g., shear modulus 900 Ω·cm2 for garnet, Li7La3Zr2O12),50 and is about a third that of LiF-reinforced electrolyte proposed by Choudhury and Archer.23 It indicates a well-matched interface contact and favors SEI formation between bare lithium and soft LiTFSI-DGM-C3N4 during aging. The activation energy for Ri in a temperature range from 30 to 70 °C is estimated to be 0.45 eV from the near-linear behavior of Arrhenius plots (Figure 2c), which is comparable to that referring to electrode−electrolyte interface even with negligible space charge layer effect.51 Note that it requires much longer time to stabilize the Li−electrolyte interface in the absence of gC3N4. Its Ri increases from 13 to 75 Ω·cm2 after 96 h and is stabilized at 55 Ω·cm2 after 120 h (Figure 2b). Since the formation of SEI between Li and LiTFSI-DGM is unstable, especially when changing ambience temperature, it is not successful in achieving satisfactory Arrhenius behavior (Figure S9). The addition of g-C3N4 can accelerate the stabilization of Li−electrolyte interface with resistance decrease during aging. After stabilization at room temperature for 120 h, the interface resistance of Li/LiTFSI-DGM-C3N4 is ∼60 Ω·cm2 larger than that of Li/LiTFSI-DGM without g-C3N4 filler, in view of steric hindrance by insulating g-C3N4 (Figure S10). However, it would not influence the cell polarization, as will be discussed. 11618
DOI: 10.1021/acsami.7b00478 ACS Appl. Mater. Interfaces 2017, 9, 11615−11625
Research Article
ACS Applied Materials & Interfaces
Figure 4. SEM images of cycled Li anodes from Li/LiTFSI-DGM-C3N4/Li symmetric cells (a) after 100 cycles at 0.5 mA/cm2 and (b, c) after 120 cycles at 2 mA/cm2. (d) SEM image of packing of g-C3N4 microspheres in slurrylike electrolyte after 120 cycles at 2 mA/cm2. (e−g) SEM images of cycled Li anodes from Li/LiTFSI-DGM/Li symmetric cells (e) after 80 cycles at 0.5 mA/cm2 and (f, g) after 10 cycles at 2 mA/cm2. (h) Scheme of Li dendrite growth and inhibition depending on Li symmetric cells with g-C3N4 or without addition. The cycled Li surface is flat, dense, and dendrite-free in the presence of g-C3N4, whereas it becomes rough and porous with the appearance of fiberlike dendrites in the absence of g-C3N4 filler.
capacity of 6 mA·h/cm2. If there is no filling of g-C3N4, higher current density would accelerate the roughening and dendrite growth of Li anode, as indicated from the polarization deterioration during earlier cycles (>500 mV after 10 cycles). The stable Li plating/striping in LiTFSI-DGM-C3N4 benefits from the suppression of Li dendrite growth, as shown in the SEM images of cycled Li anode (Figure 4). At 0.5 mA/cm2, the Li metal surface is almost flat and dense, and the grain boundaries are not easy to discern even after 100 cycles by
using Li/LiTFSI-DGM-C3N4/Li symmetric cell (Figure 4a). When the current density is increased to 2 mA/cm2, the Li surface becomes corrugated more or less (Figure 4b), and the grain boundaries become evident after 120 cycles (Figure 4c). The electroplated grains are as small as ∼50 nm in size and no dendrite signal is observed. Some sheet-shaped species are occasionally observed and vertically insert in Li anode. They should be the residual g-C3N4 nanosheets peeled off the g-C3N4 micropheres, which were removed before the SEM measure11619
DOI: 10.1021/acsami.7b00478 ACS Appl. Mater. Interfaces 2017, 9, 11615−11625
Research Article
ACS Applied Materials & Interfaces
Figure 5. (a) Galvanostatic charge−discharge curves of Li/FeS2 cell based on LiTFSI-DGM-C3N4 electrolyte of at 0.1C in a voltage range of 1−3 V. (b) Cycling performance of LiTFSI-DGM-C3N4-based Li/FeS2 cell (red circles) and its comparison with Li/LiTFSI-DGM/FeS2 cell (purple triangles) and with Li/FeS2 cell containing g-C3N4 prepared by keeping its precursor solutions at 120 °C (maroon circles). The former displays the best capacity reversibility (400 mA·h/g in the 100th cycle) and cyclability (at least 400 cycles), benefiting from inhibition of Li dendrite growth and electrolyte penetration through the hierarchical structure filler built by g-C3N4 nanosheets.
to as small as 100 mA·h/g after 200 cycles and is negligible after 250 cycles. In contrast, the C3N4-containing FeS2 cell still has a high discharge capacity of 350 mA·h/g after 200 cycles, 300 mA·h/g after 300 cycles, and 250 mA·h/g after 400 cycles, benefiting from dendrite inhibition. Note that the change of gC3N4 microstructure would influence its effect on preserving the capacity of Li/FeS2 cell (Figure S12b). When the MCA precursor is precipitated at 120 °C, the resultant thicker g-C3N4 primary units (as shown in Figure S4) cause a quick capacity fading (below 200 mA·h/g after the first 20 cycles). It indicates that fine and thin nanostructures are crucial to better wet electrolyte as well as favorable mass transport. However, the long-term cyclability is still guaranteed with a reversible capacity around 200 mA·h/g after 300 cycles. It appears that the rougher g-C3N4 microstructure is still effective in alleviating anode dendrite growth to a certain degree. In order to further show the advantages of 3D mesoporous sphere geometry of g-C3N4 over 2D geometry, we also prepared carbon nitride nanosheets (O-g-C3N4) as a reference filler material by thermal oxidation to exfoliate bulk carbon nitride.46 These nanosheets are constructed by numerous crosslinked branched grains with a diameter of ∼25 nm, and they have a thickness less than 30 nm as shown in SEM and TEM images (Figure 6a−d and Figure S13). XRD and FTIR spectra of O-g-C3N4 display similar patterns as those of g-C3N4 (Figures S14 and S15). O-g-C3N4 shows more intense diffraction peaks and clearer SAED rings (Figure 6e), indicating better crystallinity than g-C3N4. Apart from three sets of characteristic FTIR peaks assigned to g-C3N4 structure, there is an additional peak at 1090 cm−1 remarkably observed in O-gC3N4. It corresponds to C−O vibration, confirming the presence of O-containing group after thermal oxidation.46 The O signal is also observed in the XPS of O 1s, which is
ment of cycled Li anode surface. Figure 4d also shows the layout of g-C3N4 microspheres in slurrylike electrolyte after 120 cycles. Large current density (2 mA/cm2) does not seriously break these g-C3N4 microspheres; their porosity and close packing are well-maintained (Figure S11). In the absence of gC3N4 filler, the Li anode becomes much rougher even at a relatively low current density of 0.5 mA/cm2 after 80 cycles (Figure 4e). Some stick-shaped grains with a width of 250−500 nm appear at surface. The dendrite growth becomes remarkable at 2 mA/cm2 (Figure 4f,g). The sticklike grains evolve into fiberlike ones with a narrower diameter of 100−250 nm, leading to a porous Li surface after merely 10 cycles. The high hardness, dense packing, and hierarchical structure of microsized g-C3N4 spheres are responsible for the change of Li deposition behavior to limit dendrite growth (Figure 4h). We tested Li/FeS2 cells based on both electrolytes, LiTFSIDGM-C3N4 and LiTFSI-DGM, at 0.1C in the voltage range 1− 3 V (Figure 5 and Figure S12). They display good capacity reversibility, with a discharge capacity around 400 and 350 mA· h/g for LiTFSI-DGM-C3N4 and LiTFSI-DGM, respectively, in the 100th cycle (Figure 5b). The capacity of C3N4-containing FeS2 cell is 50−100 mA·h/g higher than that of Li/LiTFSIDGM/FeS2 at the corresponding cycling stages (before 150 cycles). The same magnitude of interface resistance of slurrylike electrolyte as that of liquid electrolyte does not compromise its voltage polarization (main discharge at 1.5 V and charge at 1.8 V) and capacity performance (with first charge/discharge capacity >700 mA·h/g; Figure 5a). Degradation of the upper discharge plateau (∼2 V) is remarkably alleviated for C3N4containing FeS2 cell, owing to the potential effect of g-C3N4 on decreasing polysulfide loss. After 150 cycles, the capacity degradation is accelerated for the Li/LiTFSI-DGM/FeS2 cell due to Li dendrite growth, and its discharge capacity decreases 11620
DOI: 10.1021/acsami.7b00478 ACS Appl. Mater. Interfaces 2017, 9, 11615−11625
Research Article
ACS Applied Materials & Interfaces
Figure 6. (a, b) SEM images of O-g-C3N4 nanosheets at different scales. (c, d) TEM images of O-g-C3N4 nanosheets at different scales. (e) SAED pattern of O-g-C3N4. (f) Photos of as-synthesized O-g-C3N4 powder and its composite with LiTFSI-DGM to form slurrylike electrolyte. These nanosheets with thickness
