Novel polymer coating for chemically absorbing CO2 for safe Li-ion battery

Gas evolution in Li-ion batteries remains a barrier for the implementation of high voltage materials in a pouch cell format; the inflation of the pouch cell is a safety issue that can cause battery failure. In particular, for manganese-based materials employed for fabricating cathodes, the dissolution of Mn2+ in the electrolyte can accelerate cell degradation, and subsequently gas evolution, of which carbon dioxide (CO2) is a major component. We report on the utilization of a mixture of polymers that can chemically absorb the CO2, including the coating of aluminum foils, which serve as trapping sheets, introduced into two Ah pouch cells—based on a LiMnFePO4 (cathode) and a Li4Ti5O12 (anode). The pouch cells with trapping sheets experienced only an 8.0 vol% inflation (2.7 mmol CO2 per gram of polymers) as opposed to the 40 vol% inflation for the reference sample. Moreover, the cells were cycled for 570 cycles at 1 C and 45 °C before reaching 80% of their retention capacity.

www.nature.com/scientificreports www.nature.com/scientificreports/ been used as separators in Li-ion batteries for trapping M 2+ with some success, significantly improving the cycle life of cells [12][13][14][15] . Crown-ether derivatives have been very efficient for trapping; the grafting on of carbon (acetylene black) was demonstrated to be efficient for capturing Mn 2+ , and extended the cycle life of a NMC-graphite Li-ion cell 14 . A. Banerjee et al. introduced a multifunctional separator based on a poly(ethylene-alt-maleic anhydride) commercial polymer, to capture Mn 2+ by chelation with -COOH groups 16 . The amount of Mn 2+ detected on the negative electrode was 1.5 times lower than that of plain separators, and a higher capacity was retained by cells (LMO-graphite) at 30 °C and 55 °C. A novel binder for fabricating cathodes was also investigated, with H. Lee et al. reporting an alginate-based polymer being used as a binder for LMO-based batteries 17 . Dissolving Mn 2+ in the electrolyte was reduced drastically compared with a PVDF binder at elevated temperatures, the chemical coordination between the secondary alcohol groups from the binder and Mn 2+ impeding the diffusion of ions in the electrolyte.
Chemical modification of separators, binders or carbons are elegant methods to produce functional materials efficient for trapping ions. However, the application of new materials in commercial products is difficult because integration into battery production lines is so costly.
Instead of using new materials for the fabrication of a cathode or battery, we propose, for the first time, the use of active polymers to capture the CO 2 produced during cycles 6,18,19 , to chemically prevent pouch cell inflation instead of mechanically controlled. Combining two polymers and one catalyst in the solid phase offers a good system for trapping CO 2 20-23 . This system can be used as a polymeric coating on aluminum foil introduced as a layer in a pouch cell. This approach is unique and does not require any modification of the production line, offering a cost-effective approach for battery manufacturers. We introduced this concept in two Ah pouch cells composed of a stack of cathodes of carbon-coated LiMnFePO 4 (LMFP), anodes of carbon-coated Li 4 Ti 5 O 12 (LTO) and a liquid electrolyte.

Results and Discussion
S.-I. Yamamoto et al., reported in 2005 a polymer mixture which can capture gaseous CO 2 under atmospheric pressure 21 . This system was unique as the polymers were applied in a solid form with glycidyl-methacrylate-based polymers (active polymer), poly(N-vinyl-2-pyrrolidone) (PVP) as a solid solvent and LiBr as a catalyst. The epoxy groups reacted with CO 2 to form cyclic carbonate and a conversation of 94% was reported. PVP was applied as a good phase transfer agent, acting as a "solid solvent" for the catalyst and permitting a mixture with glycidyl based polymers. To our knowledge, that was the only example of trapping CO 2 with polymers in the solid phase. We decided to use a more efficient and polymer-soluble catalyst in our system 22 , which is highlighted in Fig. 1.
Because of the nature of the process, the injection of electrolytes into cells wet all parts, and as such the coating has to meet several requirements: it should be insoluble in solvents, it should capture gas, and it should exhibit excellent adhesion to aluminum (Al) foil. The polymers were modified by incorporation of monomers into the backbone in order to increase adhesion to Al and to evaluate effect on gas capture. Three (3) monomers with glycidyl methacrylate were tested as copolymers in order to evaluate the adhesion to the Al foil and the effect of CO 2 capture. Figure 2 shows these monomers. Examples of the 1 H NMR spectra (Figure S1-S3) and GPC traces ( Figure S4-S5) can be found in the Supplementary Information.
The polymers were mixed using the following method. First, a solid solvent (PVP) and catalyst were mixed with ethanol until complete dissolution (0.15 g, 50.0 mg (5.0 mol% versus copolymer) and 1.00 g., respectively). A second solution of copolymer (0.50 g. and 1.5-2.0 g of NMP) was heated at 80 °C under vigorous stirring. Drop  www.nature.com/scientificreports www.nature.com/scientificreports/ by drop, the first solution was added to the second under vigorous stirring, the final solution was then cooled at 22 °C to obtain a homogenous mixture. This method ensured the perfect mixed of polymers without phase separation. The initial evaluation of CO 2 trapping was done using a simply dropping mixture (80.00-100.00 mg) on an Al-plastic bag used for making the pouch cell; the testing for the chemical absorption of gas was done by pressurizing with CO 2 (5.00-10.00 mg) an Al-plastic bag as shown in Fig. 3. All weights were measured using an electronic balance with an accuracy of 0.01 mg. Sealed Al-plastic bags were left in an oven at 25 °C, 45 °C, and 60 °C for different periods of time. This simple method was selected because a high pressure did not occur in the pouch cell, and cells may well vent at low pressures which presents safety concerns.
The conversion of the epoxy groups in cyclic carbonate was followed by Fourier transform infrared spectroscopy (FTIR) 21 . The conversion rate was maybe overestimated due to the limited penetration depth by IR. Therefore, we cannot scrutinize the 30 μm tick of coatings, we assumed a depth around 5 μm. The disappearance of signals at 900-905 cm −1 (epoxy groups) and the appearance of C=O characteristic of cyclic carbonate at 1790-1800 cm −1 allowed an easy calculation of the conversion using the ratio of the two areas (Fig. 4). Solid-state 13 C NMR was also performed to confirm the presence of cyclic carbonate, and the appearance of C=O groups at 155 ppm (see Figure S6 in Supplementary Information). Figure 5 shows the results of the conversion calculated using data collected by FTIR for the three polymers at different temperatures after 24 h of gas exposure. Chemical absorption of CO 2 by the polymers was calculated using conversion (area of peaks by FTIR) and the amount of active polymer coated (eq. S1 in Supplementary  Information).
The rate of conversion of oxirane moieties with CO 2 seemed to be correlated with the nature of the monomer. Indeed, the methyl acrylate (monomer 1) based polymer (polymer 1) had a rapid conversion compared to monomer 2 (polymer 2), despite the fact that polymer 2 had a higher level of glycidyl methacrylate incorporated (80 mol% vs 57 mol%, see Table 1). We deem it to be plausible that this was related to the proximity between the polymer chains because the short pendant groups diminished the free volume between the polymers. Therefore, PVP and the solubilized catalyst were closer to poly(glycidyl methacrylate-co-methyl acrylate) (polymer 1) thus oxirane moieties reacted rapidly because of the catalysts in this environment 21 . Monomer 3 based polymer (Polymer 3), with a bulky pendant group (ethylene glycol methyl ether), had the slowest conversion of the three polymers, which corroborated our previous observations. Moreover, the logically augmented levels of trapping groups in the polymer backbones (polymer 2 vs. polymer 1 and 3) permitted the capture of more CO 2 (polymer 2: 0.79 molCO 2 mol −1 polymer). The time of exposure was also evaluated; polymer 1 had a conversion of 87% after 12 h at 45 °C and 98% after 24 h. An important fact is that the conversion of the epoxy groups was almost complete at 60 °C for all polymers after 24 h. As reported by Yamamoto et al. 21 . NMP is a better solvent for ionic catalyst  www.nature.com/scientificreports www.nature.com/scientificreports/ and PVP. To validate that we changed NMP with tetrahydrofuran, and experienced a conversion decrease of 17% (45 °C, 24 h). Moreover, a decrease in the catalyst quantity in the mixture reduced the conversion of the oxirane groups to 48% with 2.5 mol% from 65% with 5.0 mol%, as expected.
Moreover, the physical absorption of CO 2 was related to the nature of the comonomer, and not with the level of incorporation in the main chains of glycidyl methacrylate alone. Many research teams have reported that the polymer itself physically interacted with the gas; reports of the solubilizing of supercritical CO 2 by polymers have suggested that the solubility of CO 2 was driven by the free volume of polymers 24,25 . This behavior can change the nature of the coating, assuming that n-butyl acrylate (Monomer 2) based polymers have a higher free volume compared with methyl acrylate based polymers; free volume being related to the temperature of the glass transition (T g ). Statistical random copolymers with Monomer 2 had lower T g versus Polymer 1 established on the Fox equation 26 , calculated at 32 °C and 43 °C respectively. Moreover, at low pressures the solubility of the gas increased with pressure with a diluent effect, acting like a plasticizer (increasing the distance between polymer chains), and therefore decreasing T g 27 . Therefore, in the presence of CO 2 , the polymer incorporating n-butyl acrylate becomes very fluid and exhibits a lack of adhesion after CO 2 exposure, which could be related to that along with the hydrophobicity of Monomer 2.
We assumed that the thickness of the coating had an influence on the absorption, and as such we evaluated the coating of polymer 1 using the Doctor Blade method on a large Al foil. We decided to use poly(glycidyl methacrylate-co-methyl acrylate) (polymer 1) due to its better adhesion on Al foil and fast trapping was observed. The solution was coated using the Doctor Blade method on Al Foil and dried at 80 °C for 12 h. The coating (surface = 45 cm 2 , thickness = 81 μm, 123 mg of polymers) was then inserted into an Al-Plastic bag pressurized with CO 2 . We evaluated the coating at 45 °C, the testing temperature of the cells. After 12 h, a conversion of 69% of the epoxy groups was observed by FTIR. However, the coating showed a lack of adhesion when soaked in battery electrolyte (20 mL of electrolyte for 2 h at 22 °C), and the polymers were soluble. Consequently, 3-(trimethoxysilyl)propyl methacrylate was selected as a monomer for polymerization with glycidyl methacrylate. This monomer is known to form a strong bond with aluminum oxide preventing the leaching of the coating from the Al surface 28 . Moreover, we incorporated a large percentage of glycidyl methacrylate (92 mol%) in polymer backbone in order to enhance its chemical absorption. The coating on the Al foil was homogenous and did not exhibit any dissolution in the battery electrolyte (the same proportion of solid solvent and catalyst which were used previously, were mixed).
The coating on the Al foil is shown in Fig. 6. In brief, two sheets of 27.5 cm 2 were introduced into the pouch cell, 61.0 mg of polymers by sheet with a thickness of 90 μm. A pouch cell of 2 Ah was assembled with an LMFP (cathode) and LTO (anode), aluminum foils with polymer coatings were added to the top and the bottom of the cell (two sheets) (Fig. 6). Details of the electrode preparations and cell configurations are highlighted in the Methods Section.
Prior to being tested, the cells were cycled at 0.2 C for 3 cycles (charge and discharge) at 25 °C (30 h of cycling), the charge-discharge curves of which can be found in the Supplementary Information (Figure S7). Images of the cells after cycling and their respective diagram bars are shown in Fig. 7. Visually (Fig. 7a), we observed the inflation of the cell without trapping sheets. We expected a major gas evolution during the first cycles as reported in Figure 5. Diagram bars of the conversion of epoxy groups calculated using the area of peaks by FTIR and lines showing chemical absorption calculated using eq. S1. www.nature.com/scientificreports www.nature.com/scientificreports/ the literature 6,19 . We measured the volume of the cells using Archimedes' principle, initially reported by J. R. Dahn and team to evaluate the gas evolution in NMC-graphite cells 29 . The initial volumes of the cells prior to being cycled were 25.0 mL, the cell without sheets grew 40 vol% (10.0 mL of gas was generated in the reference) while the cell with trapping sheets grew only 8.0 vol% which was not visually detectable. Based on Fig. 7b, we calculated the amount of gas trapped by the two sheets (i.e. 8.0 mL of gas, which we assumed was CO 2 since the polymers did not chemically interact with the other gas). Using the ideal gas law (or general gas equation) for our calculations at 1 bar pressure 19 , we estimated that 2.7 mmol of CO 2 per gram of polymers was trapped.
The cells were extensively cycled to age the active materials and accelerate the side reactions. Although only a minimum of gas was generated after the first cycle, we deemed it valuable to evaluate the technology over a long cycle time. The cells had a capacity of 1.90 Ah at 1 C and 45 °C (first cycle), and their retention capacity was still at 99% over an extended 290 cycles. A depletion of capacity was recorded up to 1.52 Ah after 570 cycles (retention capacity <80%), (see Figure S8 for the cycle-life curve), and no cell inflation was visually observed while reference had inflation (see Figure S9 in Supplementary Information). The disassembly of the cell allowed the recovery of the trapping sheets, which exhibited no visual degradation or leaching; subsequent analysis by FTIR showed a clean spectrum without trace of degradation, presenting a large peak at approximately 1800 cm −1 (Fig. 8).

conclusion
We demonstrated, for the first time, the effectiveness of applying a coating of polymers on Al foil for chemically absorbing CO 2 during Li-ion pouch cell operations. This strategy was effective in preventing pouch inflation. We observed a decrease in the cell volume (after cycling) by 80% as the sheets captured CO 2 . No safety issues were experienced due to cell inflation. Moreover, this technology was applied to a 2 Ah Li-ion battery which was cycled over more than 500 cycles at 1 C and 45 °C in order to accelerate the aging process (with associated side reactions). Therefore, we can conclude that our approach is economically viable, industrially applicable, and permits the fabrication of safe high-power Li-ion batteries based on large format cell with significant cycle life potential.

Methods
Materials and method. All chemicals were purchased from Sigma Aldrich, VWR chemicals or Acros. They were used without purification unless mentioned. 2,2′-azobis(2-methylpropionitrile) (AIBN) was crystallized in hot methanol. Poly(N-vinyl-2-pyrolidinnone) (PVP) from Acros had a molecular weight of 3500 gmol −1 . Polymers were made using the same method as described below, the details of which are shown in Table 1.
Before use, the glycidyl methacrylate (GMA) and monomer were passed through a plug of basic Al 2 O 3 . To a round-bottom 100 mL flask, 50 mL of tetrahydrofuran (THF), 5.8 g. (40.8 mmol) of GMA, and a co-monomer were added. The solution was stirred for 30 min and through which nitrogen was bubbled. Next, AIBN (radical initiator) was added. The flask was equipped with a condenser and heated at 65 °C for 12 h under nitrogen. The resultant solution was cooled down at 22 °C and poured into 10 volumes of diethyl ether. The supernatant was decanted, and the polymer was dried under vacuum at 60 °C for 12 h.  www.nature.com/scientificreports www.nature.com/scientificreports/ Characterization. FTIR spectra were recorded on an Agilent Cary 630 equipped with an ATR accessory.
Liquid 1 H NMR analysis was performed on a Bruker 300 instrument operating at 300 MHz with an acquisition time of 1.9 s. Samples were dissolved in CDCl 3 or DMSO d-6 (a TMSPA-based polymer) and the residual solvent peak was used for calibration. The solid-state 13 C NMR spectrum was recorded on a Bruker Avance III HD operating at 100.60 MHz, using a triple-resonance 1.9 mm MAS probe in double resonance mode. The sample had a mass of 15 mg and the spinning frequency was 20 kHz. 13 C spectra were obtained using 1.5 ms cross-polarization ramped from 70-100% of the maximum amplitude or a 90° pulse. The molecular weight distribution (Mn, Mw and PDI) was determined with a GPC (1260 Infinity II Multi-detector from Agilent) equipped with triple detection (LS at 690.0 nm, viscosity, and RI) and operating at 25 °C with THF as the eluent. The columns were Agilent (Varian) PL-Gel mixed-B 10 mm.
Carbon-coated LiMn 0.75 Fe 0.20 Mg 0.05 PO 4 (LMFP) was graciously supplied by Sumitomo Osaka Cement and carbon-coated Li 4 Ti 5 O 12 (LTO) was made using the method published 30 . LiPF 6 , solvents and poly(vinylene difluoride) (PVDF) were purchased from BASF. Electronically conductive material such as Acetylene black and VGCF were purchased from Denka and Showa Denko, respectively.
Prior to extensive cycling, the cells were cycled at 0.2 C for 3 cycles at 25 °C. Subsequently, cycling was done at 1 C using a Biologic BCS-815 in a climate chamber operated at 45 °C. Cycling was stopped when the cells retained 80% of their initial capacity.