Research Article Cite This: ACS Appl. Mater. Interfaces 2018, 10, 3470−3478
www.acsami.org
Effect of Binder Architecture on the Performance of Silicon/Graphite Composite Anodes for Lithium Ion Batteries Peng-Fei Cao,*,† Michael Naguib,‡,# Zhijia Du,§ Eric Stacy,∥ Bingrui Li,† Tao Hong,⊥ Kunyue Xing,⊥ Dmitry N. Voylov,⊥ Jianlin Li,§ David L. Wood, III,§ Alexei P. Sokolov,†,⊥ Jagjit Nanda,# and Tomonori Saito*,† †
Chemical Sciences Division, §Energy & Transportation Science Division, and #Materials Science and Technology Division, Oak Ridge National Laboratory, Oak Ridge, Tennessee 37830, United States ‡ Department of Physics and Engineering Physics, Tulane University, New Orleans, Louisiana 70118, United States ∥ Department of Physics and Astronomy, and ⊥Department of Chemistry, University of Tennessee, Knoxville, Tennessee 37996, United States S Supporting Information *
ABSTRACT: Although significant progress has been made in improving cycling performance of silicon-based electrodes, few studies have been performed on the architecture effect on polymer binder performance for lithium-ion batteries. A systematic study on the relationship between polymer architectures and binder performance is especially useful in designing synthetic polymer binders. Herein, a graft block copolymer with readily tunable architecture parameters is synthesized and tested as the polymer binder for the high-mass loading silicon (15 wt %)/graphite (73 wt %) composite electrode (active materials >2.5 mg/cm2). With the same chemical composition and functional group ratio, the graft block copolymer reveals improved cycling performance in both capacity retention (495 mAh/g vs 356 mAh/g at 100th cycle) and Coulombic efficiency (90.3% vs 88.1% at first cycle) than the physical mixing of glycol chitosan (GC) and lithium polyacrylate (LiPAA). Galvanostatic results also demonstrate the significant impacts of different architecture parameters of graft copolymers, including grafting density and side chain length, on their ultimate binder performance. By simply changing the side chain length of GC-g-LiPAA, the retaining delithiation capacity after 100 cycles varies from 347 mAh/g to 495 mAh/g. KEYWORDS: polymer binder, graft copolymer, silicon/graphite anode, grafting density, side chain length
■
INTRODUCTION Increasing the energy density of the lithium-ion battery (LIB) is necessary to meet the demands for their expanding applications from portable electronics to large-scale emerging applications, such as renewable energy storage grids and electric vehicles that require acceptable driving distance upon a single charging.1−4 Among many candidates that can increase the energy density of anode, silicon (Si) is extremely compelling due to its high theoretical capacity (3579 mAh/g for Si compared to 372 mAh/g for commercial graphite anode), low operating potential, nontoxicity and worldwide abundance.5−9 However, the high specific capacity of the silicon-based electrode is typically observed only in the initial cycles and cannot meet the long cycle life required for typical electric vehicle applications. At a fundamental level, the origin of fast capacity fade is mainly due to two main factors: (i) huge volume change (about 320%) during the lithiation and delithiation process that causes the fracturing, pulverization, and electrical isolation of silicon particles from the electrode matrix; (ii) continuous formation and reformation of the solid electrolyte interfaces (SEI) during © 2018 American Chemical Society
charge−discharge, which consume extra lithium and result in lower Coulombic efficiency.10−14 Utilization of polymeric material to hold active materials is a conventional approach, and the polymer binder plays an even more significant role in the cell performance of the siliconbased electrodes because of their enormous volume changes during electrochemical cycling.5,11,15 A high-performance polymer binder can mitigate the volume expansion stress and maintain the electrical contact with conductive additive and current collector to achieve long-term cycling capacity of the LIB.11,16−19 Recent progress in the field of polymer binders for silicon-based anode has clearly demonstrated the importance of adhesion strength and mechanical properties of the polymeric materials.18,20−22 Magasinski et al. reported for the first time improved binder performance of pure poly(acrylic acid) (PAA) compared with conventional polyvinylidene fluoride (PVDF) Received: September 1, 2017 Accepted: January 4, 2018 Published: January 4, 2018 3470
DOI: 10.1021/acsami.7b13205 ACS Appl. Mater. Interfaces 2018, 10, 3470−3478
Research Article
ACS Applied Materials & Interfaces Scheme 1. Synthesis of Graft Copolymer GC-g-LiPAA via RAFT Polymerization
the real application of typically reported silicon based anode materials is their low mass loading of active materials (usually lower than 0.5 mg/cm2) that renders low areal capacity and energy density despite the high specific capacity.37,39 Utilization of silicon/graphite composite anode materials instead of pure silicon nanoparticles (SiNPs) is one of the solutions to increase the mass loading for practical applications, and this approach is especially popular among the manufacturers because the electrode made of graphite has a mature manufacturing process and relatively low price.40−42 Herein, to elucidate the architecture effect of synthetic polymers on the polymer binder performance for silicon-based anodes, graft block copolymers with well-defined architecture were synthesized by reversible addition−fragmentation chain transfer (RAFT) polymerization43−45 and tested as the polymer binder for the high-mass loading silicon/graphite composite electrode.
due to hydrogen bonding or covalent connection of the carboxylic groups with the hydroxyl groups on the silicon surface.23 Direct evidence for the formation of a covalent bond between the SiNPs and PAA was provided by the Koo et al.24 The mechanical strength of the polymeric materials was also demonstrated to be important in retaining the capacity of silicon-based anodes. Li et al. reported improved binder performance of a stiff binder, that is, sodium carboxymethoxy cellulose (CMC), compared with the elastic PVDF and CMC/ styrene−butadiene rubber (SBR).25 Bao and co-workers also demonstrated a significant role of the elastic mechanical strength in the polymer binders for microsize silicon anode.26,27 Recently, Choi et al. also incorporated polyrotaxane in the conventional PAA binder to impart mechanical elasticity to the obtained polymer network.28 Despite this progress in natural and synthetic polymer binder development, few studies have investigated the macromolecular architecture effect on polymer binder performance for silicon based electrodes. Among the polymers with different architectures,29−33 the graft polymers, which are called either combs or bottle brushes depending on grafting density, have been attracting significant attention due to their efficient control over the structure and the corresponding properties of obtained materials.32,34−36 The graft copolymer with chemically distinguished grafted side chains and polymer backbones can retain their desired properties from both polymer components and enable efficient control of both chemical−physical interaction and mechanical performance,34,36 which are extremely important for polymer binder applications. PAA grafted polyvinylidene fluoride (PVDF) and NaPAA grafted CMC are reported to exhibit improved cycling performance compared with linear analogues, while their architecture effect on polymer binder performance was not investigated due to the limitation of synthetic strategy, which did not allow a defined polymer architecture.37,38 Conventional knowledge of “structure−property” relationships in polymers indicates that the structural parameters, such as the side chain length and grafting density, significantly affect the properties of graft polymers.34−36 Another important issue in
■
EXPERIMENTAL SECTION
Chemicals and Materials. Glycol chitosan (≥60%-NH2 by titration, degree of polymerization (DPn) ≥ 400, see compound 1 in Scheme 1), N-(3-(dimethylamino)propyl)-N′-ethylcarbodiimide hydrochloride (EDC), 4,4′-azobis(4-cyanopentanoic acid) (ACPA), lithium hydroxide (LiOH), and N-hydroxysuccinimide (NHS) were obtained from Sigma-Aldrich and used directly. 2-(((Butylthio)carbonothioyl)thio)propanoic acid (compound 2) was synthesized following previous literature.44 Acrylic acid was passed through a column with alternative inhibitor remover and aluminum oxide (basic) to remove any inhibitor before the polymerization. Semipermeable membranes made of regenerated cellulose with molecular weight cutoff equal to 3.5 kDa and 14 kDa were ordered from SpectrumLabs and Ward’s Science, respectively. Synthesis of the Macro-RAFT-CTA (Compound 3). A typical synthesis of GC22%-RAFT was performed as follows: 800 mg (3.60 mmol repeating units) of GC was dissolved in 120 mL of mixed solvent V(H2O)/V(CH3OH) = 2:1 by vigorously stirring overnight at 45 °C. EDC (0.552 g, 2.88 mmol) and NHS (0.332 g, 2.88 mmol) were added to the solution by the addition of 0.514 g (2.16 mmol) of carboxylic acid terminated RAFT-CTA (compound 2) in 10 mL of mixed solvent V(H2O)/V(CH3OH) = 1:1. The mixture was stirred at 35 °C for 2 days before it was purified by dialysis using a 3471
DOI: 10.1021/acsami.7b13205 ACS Appl. Mater. Interfaces 2018, 10, 3470−3478
Research Article
ACS Applied Materials & Interfaces
Figure 1. (A) Comparative 1H NMR spectra of GC (1) and GC22%-RAFT (3) and (B) IR spectra of GC (1), GC22%-RAFT (3), GC22%-g-PAA62 (4), and GC22%-g-LiPAA62 (5).
Table 1. Composite Anodes with Different Polymer Binders anodes with different polymer bindersa PVDF GC GC21%-m-LiPAA76 LiPAA76 GC22%-g-LiPAA62 GC22%-g-LiPAA27 GC22%-g-LiPAA17 GC12%-g-LiPAA76
porosity (%) 63.8 62.7 59.2 56.1 73.2 64.5 47.7 68.7
± ± ± ± ± ± ± ±
0.5 0.9 6.3 7.0 1.5 0.9 1.2 0.7
initial delithiation capacity (mAh/g) 445 558 729 723 745 771 671 758
± ± ± ± ± ± ± ±
100th delithiation capacity (mAh/g)
2.7 18 13 17 12 10 21 6.4
278 155 356 411 495 425 346 400
± ± ± ± ± ± ± ±
1.6 5.6 3.6 8.9 2.9 3.2 4.7 1.2
initial Coulombic efficiency (%) 63.8 76.1 88.1 65.3 90.3 92.9 92.8 91.1
± ± ± ± ± ± ± ±
0.9 0.1 0.3 0.4 0.1 0.0 0.2 0.1
20th Coulombic efficiency (%) 99.7 99.7 99.1 99.3 99.4 99.2 99.2 99.1
± ± ± ± ± ± ± ±
0.1 0.0 0.1 0.1 0.0 0.1 0.0 0.0
a
Subscript following GC indicates the molar ratio of LiPAA to the monomer of GC; m means physical mixing; g means chemical grafting; subscript following LiPAA is the DPn of LiPAA.
semipermeable membrane (Mw = 3.5 kDa) against the mixed solvent V(H2O)/V(CH3OH) = 3:1. The 1H NMR and IR spectra are shown in Figure 1. Synthesis of the Graft Block Copolymer GC-g-LiPAA (Compound 5 in Scheme 1). A typical synthesis of GC22%-gLiPAA62 was performed as follows: 452 mg of GC22%-RAFT (0.367 mmol RAFT-CTA) was dissolved in 60 mL of mixed solvent V(H2O)/ V(CH3OH) = 3:1. To that solution, 3.97 g of acrylic acid (55.1 mmol) and 10.3 mg of ACPA (3.67 × 10−4 mol) in 40 mL of DI water were added. The RAFT polymerization was performed at 70 °C for 10 h. Further dialysis using the semipermeable membrane (Mw = 14 kDa) against water afforded the purified GC22%-g-PAA62 with the 1H NMR and IR spectra shown in Figure S1 and Figure 1. The GC22%-g-LiPAA62 was obtained by dropping the aqueous solution of LiOH into that of GC22%-g-PAA62 until the pH = 7. See Figure S7 and Figure 1 for the 1 H NMR and IR spectra. Characterization. 1H Nuclear magnetic resonance (NMR) spectra were recorded on a Bruker 500 MHz NMR spectrometer using chloroform-d or DMSO-d. Infrared (IR) spectra were recorded on a Cary 600 series FT-IR spectrometer (Agilent Technologies), and the scanning range was 4000−400 cm−1. The solution viscosity of polymers was measured on AR2000ex (TA Instruments) through small amplitude oscillatory shear measurements at the shear rate of 1 s−1. For peer test, a 1-in.-wide and 3-in.-long electrode sample was attached to 3 M tape, and the peel strength of the electrode specimens was measured with a high-precision micromechanical test system (FPT-H1 Horizontal Dedicated Friction, Peer and Tear Tester; Mecmesin). While pulling the tape at a constant displacement rate of 100 μm/s, the applied load was continuously monitored, and force− displacement plots were made. The atomic force microscopy (AFM) indentation was performed using OmegaScope, AIST-NT (Novato,
US) scanning probe microscope. The measurement was performed within indentation depth of up to 200 nm to reduce the surface effect in elastic modulus of the studies polymers.46 A Hertz approach was employed to calculate the Young’s modulus by assuming the Young’s modulus of silicon E = 150 GPa and Poisson’s ration ν = 0.22.47 Adhesion forces analysis was based on measurement of Force− distance curves using CP-PNPS-SiO-E-5 cantilever with spherical SiO2 tip of 15 μm size (resonance f = 67 kHz, spring constant k = 0.32 N/ m, length L = 100 μm). Preparation of Electrodes and Assembly of Coin Cells. All of the coin cells (stainless steel CR-2032, Hohsen Corp., Osaka, Japan) were prepared and assembled in an argon-filled glovebox. The synthetic polymers (10% in weight) were dissolved in DI water (1methyl-2-pyrrolidine (NMP) for PVDF), followed by the addition of SiNPs (15% in weight, Nanostructured & Amorphous Materials Inc., 70−120 nm), graphite (73% in weight, TIMCAL), and carbon black (2% in weight, TIMICAL SUPER C65). The composite was mixed using a homogenizer for 2 h, and the slurry was coated on the copper foil by using a doctor blade (8 mil). The coated electrode was placed in the vacuum oven for 18 h at 120 °C before it was transferred into the glovebox for coin cell assembly. The coin cells consisted of polyprolylene (Celgard 2400) as separator, 1.2 M lithium hexafluorophosphate (LiPF6) in ethylene carbonate (EC), ethyl methyl carbonate (EMC) (EC/EMC = 3:7 wt) with 10 wt % of fluoroethylene carbonate (FEC) as electrolyte, and Li metal as counter electrode. Galvanostatic cycling between 50 mV and 1.5 V was conducted using Series 4000 MACCOR batteries cycler (MACCOR Inc. Tulsa OK, USA). The cycling was carried out at 25 °C by keeping the cells in a temperature control chamber while testing. Cyclic voltammetry (CV) between 50 mV and 3 V vs Li/Li+ was executed by using a VSP300 potentiostat (Biologic, Claix, France) with a fixed 3472
DOI: 10.1021/acsami.7b13205 ACS Appl. Mater. Interfaces 2018, 10, 3470−3478
Research Article
ACS Applied Materials & Interfaces
Figure 2. (A) Photograph of the silicon/graphite/GC22%-g-LiPAA62 electrode on copper foil (film width is 10 cm) and (B) SEM image, (C) energy-dispersive X-ray spectroscopy (EDX) mapping, and (D) silicon distribution of the same electrode.
Figure 3. (A) Solution viscosity of different polymer binders at the same polymer concentration (1 mg/mL), shear rate (10 s−1), and temperature (20 °C): (I) GC22%-g-LiPAA16, (II) GC22%-g-LiPAA27, (III) GC22%-g-LiPAA62, (IV) GC12%-g-LiPAA76, (V) GC21%-m-LiPAA76, and (VI) GC. (B) CV curves of Si/graphite electrode with GC22%-g-LiPAA62 as polymer binder; inset is the zoom-in image to show the small peak in the first cathodic scan. (C) Peer test result of anode films with different polymer binders. (D) Related adhesion force of different polymer binders measured by AFM; the curve with lower relative force value (more negative) means higher adhesion force.
■
voltage rate (10 mV/s) at room temperature. Three coin cells were made from each of the composite anodes, and average specific capacity was reported in the manuscript. Error analysis was included in Table 1. All of the specific capacities reported in this study are per gram of active materials (SiNPs + graphite). The C rate was also determined based on the theoretical capacity of active materials (SiNPs + graphite).
RESULTS AND DISCUSSION
Synthesis of Graft Copolymer. Systematic investigation on the architecture effect of synthetic polymer on binder performance requires the synthesis of graft copolymers with well-controlled structure parameters, such as side chain length and grafting density. “Graft from” strategy was selected here
3473
DOI: 10.1021/acsami.7b13205 ACS Appl. Mater. Interfaces 2018, 10, 3470−3478
Research Article
ACS Applied Materials & Interfaces
Figure 4. Cycling performance of silicon/graphite electrodes from the polymer binder of PVDF, GC, LiPAA76, GC21%-m-LiPAA76, and GC22%-gLiPAA62 at a current rate of C/10 (A) and C/1 (B). (C) Coulombic efficiency of the electrodes from different polymer binders; inset is the zoom-in area showing the comparative efficiency of electrodes from GC21%-m-LiPAA76 and GC22%-g-LiPAA62 in the first 30 cycles. (D) Voltage profiles of the electrodes with GC22%-g-LiPAA62 binders.
instead of “graft to” method because it can overcome the steric hindrance and provide a controlled grafting density and side chain length.48,49 A hydrophilic derivative of chitosan, glycol chitosan (GC) was selected as the polymer backbone due to its high-modulus nature, good water solubility, and abundant reactive groups for facile modification.50 The macro-RAFTCTA (Scheme 1, compound 3) was synthesized via the amidation reaction between the GC and carboxylic acid terminated RAFT-CTA (compound 2) in aqueous solution. The comparative integration of peaks e, f and h in Figure 1A allows the calculation of the average grafting density of the RAFT-CTA on the GC backbone: 22.1% of the repeating units have been functionalized, which corresponds to 88 arms per graft copolymer. RAFT polymerization of acrylic acid from the macro GC22%RAFT (3) allows the growth of polymer side chains. The significant absorption peak lying at 1692 cm−1 attributed to the typical stretching mode of CO bond in the carboxylic acid suggests the successful grafting of PAA side chains. Moreover, comparative integration of the peak from terminal methyl groups in the RAFT-CTA compared with that from the repeating units in the PAA provides the degree of polymerization (DPn) of the PAA being 62 (See Figure S1). Neutralization of the obtained GC22%-g-PAA62 by addition of LiOH produces the GC22%-g-LiPAA62, which is confirmed by the significant growth of the carboxylate peak at 1544 cm−1 at the expense of the carboxylic acid peak as shown in Figure 1B. Comparative Binder Performance with Linear Analogue. The composite electrode with 73% graphite and 15% silicon as the active materials at a relatively higher mass loading (loading of active materials >2.5 mg/cm2; thickness >30 μm) than the typical silicon based electrode (
