There has been an active interest in battery systems based on multivalent ions such as Mg 2+ , Zn 2+ , Ca 2+ , and Al 3+ ions. [1][2][3][4] The key motivation behind multivalent-ion battery research is that it offers the promise of obtaining two or three times the energy density of monovalent-ion systems such as Li-ion and Na-ion batteries. 1 Several prototypes have already been developed for Mg-ion, Zn-ion, Ca-ion, and Al-ion batteries. 1,[5][6][7] Among these, magnesium (Mg) batteries are of particular interest owing to the low propensity of Mg-metal to form dendrites, which enables the use of the Mg-metal itself as the battery anode. 8,9 This is an important advantage considering that dendrite growth in Li-metal has been the most prominent detrimental factor preventing the use of Limetal anodes in Li-ion batteries. Mg-metal also possesses a superior volumetric capacity (3862 mAh cm -3 ) in comparison to Li-metal (2062 mAh cm -3 ). Coupling these advantages with the fact that magnesium is a highly abundant metal (eighth most abundant element), Mg batteries are being projected as the most feasible multivalent-ion battery system to complement the role of Li-ion batteries in practical applications. [10][11][12] However, there are significant challenges that need to be resolved before attempting to commercialize Mg batteries. 13 Specifically, two important limitations hinder the development of high-energy-density Mg batteries. First, the choice of Mg battery cathode candidates, with a high specific capacity and capacity retention, is limited in comparison with Li and Na battery systems. This is mainly due to slow solid-state diffusion of Mg 2+ ions. 14 Second, there is a lack of compatible electrolytes that support high cyclability with Mg-metal and simultaneously possess a high oxidative stability (voltage window). 12,15 This limitation has also affected the research progress in magnesium intercalation (cathode) materials. 16 With the limited development of electrolytes with high voltage windows, it has been challenging to experimentally identify and evaluate the reversibility of prospective cathode materials for Mg-ion batteries. 17,18 The Grignard reagents that are compatible with the Mg-metal anode possess a low oxidative stability of <1.5 V, which is not high enough to investigate possible cathode candidates for Mg-ion batteries. 15,19 To improve the oxidative stability and the kinetics of Mg 2+ ions within the electrolyte, several liquid electrolytes were designed with organometallic reagents dissolved in solvents such as tetrahydrofuran (THF), diglyme, and tetraglyme. 20,21 Although these electrolytes exhibit a high degree of reversibility and oxidative stability, the use of flammable, volatile ethereal solvents such as THF poses a safety risk. 22 This conflicts with the notion of projecting Mg-metal batteries as a safer alternative to Li-metal batteries. Therefore, it is also important to choose electrolyte constituents that do not compromise the superior safety aspects of the Mg-metal anode. 12 Polymers like poly(ethylene) oxide (PEO) and polyvinylidene fluoride (PVdF) have been known for a long time to be robust, conductive media for Li-ion battery electrolytes. [23][24][25][26] It has been proposed that their mechanical properties and conductivities could be further enhanced by incorporating ceramic fillers such as SiO 2 , MgO, Al 2 O 3 , and TiO 2 into the polymers. [27][28][29][30][31] An effective use of these polymers can eliminate the need of using liquid solvents for dissolving the salts in a battery electrolyte. Following their successful integration into Li-battery systems, there have been several reports on Mg-ion-conducting polymer electrolytes possessing high ionic conductivities. [32][33][34][35][36][37][38] Chusid et al. 32 developed a geltype polymer electrolyte consisting of a Mg organohaloaluminate salt: Mg(AlCl 2 -EtBu) 2 (where Et: ethyl, Bu: butyl), PVdF, and tetraglyme, which acted as the plasticizer. This gel electrolyte had a high ionic conductivity of 3.7 mS cm -1 and exhibited a reversible cycling with the Chevrel-phase Mo 6 S 8 cathode for several cycles. Pandey et al. 33 developed a gel polymer electrolyte with magnesium perchlorate (Mg(ClO 4 ) 2 ), poly(vinylidene fluoride-co-hexafluoropropylene) (PVdF-HFP), MgO nanoparticles, and organic solvents such as ethylene carbonate (EC), propylene carbonate (PC), and tetrahydrofuran (THF) incorporated in the polymer blend. Although a high ionic conductivity of 8 mS cm -1 was observed in this gel polymer, no cycling data were reported. There have been other reports on polymer electrolytes based on Mg(ClO 4 ) 2 /PVdF-HFP using conventional carbonate-based solvents such as EC and PC reporting conductivities in the range of 10 -3 S cm -1 . [39][40][41] Alternatively, ionic liquids such as 1-ethyl-3-methylimidazolium trifluoromethanesulfonate (EMITF) has also been used as the plasticizer for Mg-(ClO 4 ) 2 /PVdF-HFP-based polymer electrolytes. 38 However, the cycling performance of such gel-polymer electrolytes against the Mg-metal anode needs improvement. More recently, Shao et al. 37 developed a polymer nanocomposite electrolyte made of Mg(BH 4 ) 2 , PEO, and MgO nanoparticles without the use of any plasticizer or solvent. When this polymer electrolyte was coupled with a Chevrel-phase Mo 6 S 8 cathode, the Mg-metal cells exhibited a highly stable specific capacity for 150 cycles. However, these cycling experiments were conducted at a high temperature of 100 °C. Although the cycling performance is impressive, this high-temperature requirement might make it challenging to incorporate this polymer electrolyte in practical applications. In order to project Mg batteries for practical applications, it is important to develop electrolytes that are capable of Mg plating/stripping at ambient temperature. 12 Here we report a PVdF-HFP-based composite polymer electrolyte (CPE) with Mg(ClO 4 ) 2 salt, 1-butyl-1-methylpyrrolidinium bis(trifluoromethyl)sulfonyl imide (Pyr 14 -TFSI) ionic liquid, and TiO 2 ceramic nanoparticle fillers for application in solid-state Mg-metal batteries at room temperature. We find that the CPE exhibits a high cyclability with a Mg-metal anode at room temperature. Galvanostatic cycling of symmetric Mg | Mg two-electrode cells reveals smooth plating/ stripping profiles with low deposition/dissolution overpotentials in the range of 0.08-0.30 V at current densities of 0.05 and 0.10 mA cm -2 . Moreover, the CPE exhibits a high conductivity of 0.16 mS cm -1 at room temperature. Raman spectra of the polymer composites obtained at different stages of the synthesis reveal a high degree of salt dissociation, which was further confirmed through classical molecular dynamics simulations. Our casting technique using a micrometeradjustable film applicator allows the control of the thickness of the electrolyte layer with a high precision. For the cycling tests, the thickness of the electrolyte was controlled to be within 60-80 μm. The high cyclability of Mg-metal cells coupled with such low deposition potentials appears promising toward the development of safe, long-life, solid-state Mg batteries.
Figure 1 depicts the procedure for synthesizing the composite polymer electrolyte (CPE). A detailed synthesis procedure is given in the Experimental Section. The CPE blend consists of magnesium perchlorate (Mg(ClO 4 ) 2 ) salt dispersed in a mixture of poly(vinylidene fluoride-co-hexafluoropropylene) (PVdF-HFP) polymer base, 1-butyl-1-methylpyrrolidinium bis(trifluoromethyl)sulfonyl imide (Pyr 14 -TFSI) ionic liquid, and TiO 2 ceramic nanoparticle fillers. The polymer PVdF-HFP is known for its superior mechanical properties and high electrochemical stability. 42,43 Addition of ionic liquids to polymer electrolytes is known to improve the safety, hightemperature stability, and conductivity of polymer electrolytes. 44,45 Here, the ionic liquid, Pyr 14 -TFSI, was chosen to leverage its additional advantage of being able to act as a cosolvent for Mg salts. 46,47 To improve the mechanical properties and conductivities of Mg 2+ -conducting polymer electrolytes, addition of nanosized ceramics such as MgO, SiO 2 , and TiO 2 has previously been explored and documented. 33,37,48 In this work, TiO 2 was chosen as the ceramic additive following our earlier demonstration of its beneficial effects in PVdF-based polymer electrolytes. 31 A schematic of the composite polymer electrolyte depicting the role of the constituents is shown in Figure 2a. For material characterization, the viscous CPE blend was cast into circular disks and dried. Upon drying, the viscous blend becomes a thin, stand alone membrane that can be handled using a tweezer as shown in Figure 2b. Morphological characterization of the composite polymer electrolyte was carried out using SEM and EDS analyses. The SEM image of the dried CPE (Figure S1a, Supporting Information) shows the smooth nature of its top surface. Elemental maps (Figure S1b-f, Supporting Information) of the top surface show the uniform distribution of the constituent elements, namely, Mg, C, S, Ti, and Cl, indicating a uniform dispersion of the salt, ionic liquid, and the nanoparticles within the polymer composite. For electrochemical characterization of the CPE using Mg | Mg symmetric cells, the CPE blend was directly coated onto the Mg foil using a micrometer-adjustable film applicator and dried thereafter. Two 100 μm thick Mg foils were used for assembling the Mg | Mg symmetric cell. The surfaces of the foils were scraped with a stainless-steel blade to remove the inherent insulating oxide layer and expose the shiny Mg metal underneath. The thickness of the polymer layer was tuned to be within 60-80 μm, which is sufficient to ensure electrode separation and be capable of achieving facile Mg-ion transport across the electrolyte (Figure S2a,b, Supporting Information).
Thermogravimetric analysis (TGA) was carried out to evaluate the thermal stability of the CPE and compare it with that of the pristine polymer (PVdF-HFP), and the magnesium salt (Mg(ClO 4 ) 2 ) (Figure S3, Supporting Information). The pristine PVdF-HFP shows a high thermal stability for up to 400 °C, after which a sudden loss of up to 60 wt % is observed until 480 °C. This weight loss can be attributed to C-H scission, followed by the formation of HF and the CC bond. 43 The anhydrous Mg(ClO 4 ) 2 salt is known to undergo decomposition between 390 and 440 °C, 49 which was observed in our case as well. The initial weight loss of ∼10 wt % between 200 and 300 °C prior to this decomposition could be due to the loss of absorbed water molecules. 50 For the case of CPE, there is weight loss initiating at 90 °C. This ∼10% loss could be attributed to the removal of remnant solvent from the polymer mixture. From 200 to 300 °C, there is a continuous weight loss of up to 50 wt %, similar to the trends reported in PVdF-salt electrolyte blends. 33,43 Therefore, the CPE is thermally stable until at least 200 °C after which it undergoes degradation.
Measurement of the ionic conductivity is of paramount importance while characterizing polymer and inorganic solidstate electrolytes. 51 The ionic conductivity was measured through an electrochemical impedance spectroscopy (EIS) technique, wherein the polymer electrolyte was sandwiched between two stainless-steel blocking electrodes. 33,51 The roomtemperature conductivity of the electrolyte was then calculated using the formula
where l and A denote the thickness and surface area of the electrolyte layer, respectively, and R b denotes the bulk resistance (the x-intercept at the high-frequency region) obtained from the Nyquist plot. 52 The conductivity of the as-synthesized CPE was calculated to be 1.6 × 10 -4 S cm -1 at 30 °C, a value that is close to the ionic conductivities of previously reported Mg polymer electrolytes. 33,34,48 The activation energy of a polymer electrolyte represents the height of the energy barrier for ion migration and therefore serves as a measure of how energy-expensive it is for the ions to migrate within the polymer matrix. 48 The value of activation energy could be obtained by plotting the temperature dependence of the ionic conductivity. Figure 2c shows the Nyquist plots obtained at different temperatures for a cell assembled with two blocking electrodes. Figure 2d shows the temperature-conductivity traces obtained for three samples as follows: (i) CPE, i.e., the polymer composite sample containing the ionic liquid and TiO 2 , (ii) Mg(ClO 4 ) 2 -PVdF-HFP-IL, i.e., the polymer sample containing the ionic liquid, but without TiO 2 , and (iii) Mg(ClO 4 ) 2 -PVdF-HFP-TiO 2 , i.e., the polymer composite sample containing TiO 2 , but without the ionic liquid. The conductivities of the CPE are remarkably higher than those of the samples without either the ionic liquid or the TiO 2 nanoparticles at all temperatures. This shows the beneficial effect of having both the ionic liquid and the TiO 2 nanoparticles in the CPE. The temperature dependence curves of all three samples show a nonlinear behavior. Therefore, the curves have been fit with the Vogel-Tammann-Fulcher (VTF) equation given by
where A is a prefactor; E a is a pseudoactivation energy; R is the gas constant; and T 0 is the Vogel temperature, which is related to the glass transition temperature (T 0 = T g -50 °C). 53,54 For PVdF-based electrolytes, T 0 is typically taken as 179 K. 33 The activation energy of the CPE was calculated to be 0.13 eV from the fit profile. The calculated value of activation energy is comparable to those of previously reported composite polymer electrolytes for both Li and Mg batteries. 30,48,55,56 Another important parameter is the transport number, which quantifies the fraction of ionic transport and the cationic contribution to the conductivity of any metal-ion-conducting electrolyte. The transport number of the CPE was determined using the steady-state current method developed by Bruce and Vincent, 57 wherein the electrolyte is sandwiched between two nonblocking electrodes (Mg foils, in this case). A constant potential bias, ΔV, is applied to this setup, and the current, I, is obtained as a function of time. The current response of such a symmetric Mg | Mg cell under a constant potential bias of 0.05 V is shown in Figure 2f. AC impedance of this cell was measured both before and after the polarization experiment, and the Nyquist plots thus obtained are shown in Figure 2e. The values of the electrode-electrolyte interfacial resistances were obtained from these Nyquist plots. The cationic transport number t + was then obtained from the formula
where I 0 and I ss are the initial and steady-state currents, respectively, and R i,0 and R i,ss are the initial and steady-state resistances of the interface, respectively. 58,59 The cationic transport number measured using this relation was found to be 0.23. This t + value is notably close to those of the existing polymer electrolytes that use liquid plasticizers. 33,34 The polarization experiment showing a flow of steady positive current in the Mg | Mg symmetric cell for a period of 1 h also serves as a confirmation for Mg-ion transport through the electrolyte. Although this method for computing the transport number was originally proposed for electrolytes containing univalent salts (such as Li + or Na + ), Balsara et al. noted in a separate analysis that this equation is in fact applicable to all binary salts, irrespective of the charge of the cation. 60 Therefore, the value of t + obtained here indeed denotes the transport number of Mg 2+ cations in the electrolyte. Additionally, the transport number could also be estimated using the solid-state pulsed-field gradient nuclear magnetic resonance (PFG-NMR) technique. This technique has been previously carried out successfully for estimating the transport number of Li + polymer electrolytes. 59 However, for the case of Mg, obtaining accurate spectra is notably difficult because of the low natural abundance of the 25 Mg isotope and sensitivity issues resulting from its quadrupolar nucleus. 61 The ability of the polymer electrolyte to deposit Mg onto a substrate was verified through elemental analysis using energydispersive X-ray spectroscopy (EDS) and X-ray photoelectron spectroscopy (XPS). To deposit Mg onto a stainless-steel disc, a discharge current of 0.10 mA cm -2 was applied for 10 h onto a two-electrode Mg | stainless-steel setup, with the polymer electrolyte sandwiched between the electrodes. The Mgdeposited stainless-steel disc was then recovered and used for SEM, EDS, and XPS analyses. The SEM image (Figure S4a, Supporting Information) of the surface of a stainless-steel disc showed planar deposits, free of any dendritic growth. A uniform distribution of Mg was observed in the elemental map (Figure S4b, Supporting Information), confirming the elemental composition of the deposits on the stainless-steel disc. A sharp peak corresponding to the Mg 2p state was observed in the high-resolution XPS spectrum of the Mg deposit on the stainless-steel disc (Figure S4c, Supporting Information). However, the deconvolution of the XPS spectrum revealed the presence of additional peaks corresponding to MgO. This oxidation of the Mg deposits could have resulted from an exposure to the ambient atmosphere during sample transfer.
Ex situ SEM analysis of the CPE membrane taken from the cycled Mg | Mg cells was also carried out. The digital images of the CPE layer before and after cycling (Figures S5a andS5b, respectively, Supporting Information) show that the electrolyte did not undergo any significant structural change after cycling for 100 h. The minimal wear that is observed on the surface (Figure S5b) is the result of its strong adherence to the Mg foil within the cell. Comparison of the SEM images of the CPE before and after cycling (Figures S5c andS5d, respectively, Supporting Information) shows that the distribution of TiO 2 nanoparticles remains the same even after cycling for 100 h.
The magnesium deposition and stripping processes on the Mg electrode surface were confirmed through cyclic voltammetry (CV) performed using a symmetric two-electrode Mg | Mg cell (Figure S6a, Supporting Information). Occurrence of cathodic and anodic peaks denoting the deposition and stripping, respectively, could be clearly seen in the voltammogram. The CV plot resembles the ones obtained for Mg 2+ -conducting polymer electrolytes developed by Kumar et al. 62 and by Pandey et al. 33 The decomposition voltage of the CPE was determined using a linear sweep voltammetry (LSV) test with an asymmetric Mg | SS two- electrode cell within the voltage range of 1.0-5.0 V (Figure S6b, Supporting Information). The onset of a decomposition phenomenon begins at 3.1 V evidenced by the gradual increase in current. After 4.0 V, the current increases steeply, indicating the instability of the polymer electrolyte beyond this voltage. In contrast, the polymer electrolyte without the ionic liquid undergoes decomposition at a lower voltage of 2.4 V. This demonstrates the beneficial effect of the presence of the Pyr 14 -TFSI ionic liquid in the CPE.
Although the occurrence of peaks in a cyclic voltammogram and a steady current in DC polarization test indicate Mg-ion transport in the electrolyte, galvanostatic cycling tests must be carried out to provide clear evidence of the reversible nature of the Mg plating/stripping from Mg-metal. The overpotential values observed in such galvanostatic cycling tests are a measure of the ease of Mg deposition/stripping from the Mgmetal surface and, therefore, help in determining the electrolyte's suitability in Mg-metal batteries. One of the earliest reports on Mg liquid electrolytes by Aurbach et al. 63 reported Mg dissolution/deposition overpotentials of 0.10 V for Mg(AlCl 2 EtBu) 2 /THF electrolyte at an applied current density of 0.5 mA cm -2 . More recently, Tutusaus et al. 64 reported overpotentials of 0.12 V at 0.1 mA cm -2 for the stateof-the-art liquid electrolyte [Mg(CB 11 H 12 ) 2 ]/tetraglyme developed by Toyota. Son et al. 65 reported galvanostatic cycling results with overpotentials of up to 0.10-0.30 V at 0.01 mA cm -2 to assert the favorable nature of their artificial interphase in Mg | Mg(TFSI) 2 | Mg cells. Such high overpotentials in the range of hundreds of millivolts are in sharp contrast to the few-mV overpotentials commonly observed in the case of Li metal. 66 This can be attributed to the significantly stronger cation-solvent and ion-ion interactions in divalent-metal ions compared to monovalent Li + ions. 67 Although there have been several reports of Mg 2+conducting polymer electrolytes, there is little evidence for their reversible plating/stripping with Mg-metal anodes at room temperature. To find out if our composite polymer electrolyte is capable of cycling, galvanostatic cycling tests were carried out with symmetric Mg | Mg two-electrode cells at different current densities, and their overpotentials were observed. Figure 3a-c shows the overpotential vs time plots of the Mg | Mg cells carried out at areal current densities (J A ) of 0.05, 0.10, and 0.20 mA cm -2 , respectively. The plating/ stripping time (T C ) for each cycle was set to 0.25 h. For low and moderate current densities of 0.05 and 0.10 mA cm -2 , respectively, the deposition/stripping overpotentials were observed to be ≤0.15 V for the first 50 cycles. It must be noted that this value is comparable to the overpotential values of 0.12 V obtained with the most recent state-of-the-art liquid electrolyte, magnesium monocarborane [Mg(CB 11 H 12 ) 2 ]/ tetraglyme. 64 At the end of 100 cycles, the deposition overpotential gradually increased to 0.23 V for the case of 0.05 mA cm -2 and to 0.31 V for the case of 0.10 mA cm -2 . However, for the cycling test carried out with a high current density of 0.20 mA cm -2 , a relatively higher overpotential was observed (Figure 3c). Toward the end of 50 cycles with 0.20 mA cm -2 , the overpotential stabilized to a value of 0.39 V. From then on, a gradual increase to 0.65 V was observed at the end of 100 cycles. Figure 3d-f shows the representative cycle numbers 42-50 of the cycling tests carried out at current densities of 0.05 mA cm -2 , 0.10 mA cm -2 , and 0.20 mA cm -2 , respectively. Flat voltage profiles, indicative of a smooth Mg plating/stripping process on the surface of Mg-metal, could be clearly seen in all three cases. To explore the effect of longer plating/stripping times, a symmetric Mg | Mg cell was cycled at 0.05 mA cm -2 with the plating/stripping time set to 0.50 h. Figure 3g shows the latter cycling results, where it could be seen that the deposition overpotential remains low at 0.20 V for up to 80 cycles, despite the longer plating/stripping process. Thereafter, a slight increase was observed, and the overpotential remained stable at ∼0.30 V for up to 400 cycles. The representative cycle numbers 330-340 shown in Figure 3h illustrate a voltage profile similar to that of earlier tests with flat plating/stripping features. Such a high degree of cycling reversibility for up to several hundred cycles is unique among the existing Mg-ion-conducting polymer electrolytes. Galvanostatic cycling tests were also carried out with a longer plating/ stripping time of 1.0 h (Figure S7a,b, Supporting Information). The overpotential values were slightly higher than those of the tests done with 0.5 h, which could be a consequence of the increased amount of metal that needs to be stripped from the electrodes.
To understand the high performance of the CPE, the chemical environments of the Mg 2+ ions and the polymer chains in the CPE were analyzed through Raman spectroscopy. Raman spectra were obtained at different stages of the synthesis to observe the effects of progressive addition of the precursors. Figure 4a shows the Raman spectra of the (i) pristine polymer (PVdF-HFP), (ii) Mg(ClO 4 ) 2 /ionic liquid (I.L) mixture, (iii) Mg(ClO 4 ) 2 /PVdF-HFP blend, (iv) Mg-(ClO 4 ) 2 /PVdF-HFP/I.L blend, and, the final CPE composition, (v) Mg(ClO 4 ) 2 /PVdF-HFP/I.L/TiO 2 nanoparticles. Normalized intensities were used to compare and quantify the presence of constituent ions and functional groups at different stages during the synthesis. It is widely accepted that an ample dissociation of the salt into its cations and anions is essential to obtain high cationic conductivity in polymer electrolytes. Raman spectra in the region 800-1100 cm -1 in Figure 4a clearly show the gradual increase in the intensity of the peak at 937 cm -1 which corresponds to the ν 1 mode of vibration in the (ClO 4 ) -ion. This indicates a progressive increase in (ClO 4 ) -ions which could have resulted only due to the increased dissociation of the Mg(ClO 4 ) 2 salt into (ClO 4 ) - and Mg 2+ ions. Therefore, the addition of ceramic fillers seems to have further improved the salt dissociation in the electrolyte (which is discussed in more detail in Figure 5).
Being highly charge dense, Mg 2+ ions tend to strongly interact with TFSI -anions to form bulky coordination complexes. 46 Prior reports on ionic-liquid electrolytes synthesized using Mg salt and Pyr 14 -TFSI ionic liquid have shown a coordination between Mg 2+ and TFSI -ions. 46,47 To know if this interaction is present in the case of our composite polymer electrolyte, the Raman spectra in the region 700-800 cm -1 (Figure 4b) were analyzed. The peak at 742 cm -1 corresponds to the TFSI -expansion-contraction normal mode of vibration and denotes the "free" TFSI -anions that are not coordinated to Mg 2+ cations. 47 The interaction between Mg 2+ and TFSI - ions would be typically indicated by the origin of peaks at 746 and 752 cm -1 corresponding to their coordination in bridging and bidentate geometries, respectively. 47 However, in the case of our composite polymer, this characteristic peak did not show any new shoulders originating at higher wavenumbers upon addition of the ionic liquid. We believe that their coordination between Mg 2+ and TFSI -ions (as seen in ionicliquid electrolytes) might have been prevented by the presence of polymer chains and TiO 2 nanoparticles (also discussed in Figure 5d). Figure 4c shows the comparison of the Raman spectra in the region of the pristine polymer and the final composite. The peak at 796 cm -1 , corresponding to the crystalline phase (α-phase) of the PVdF, 68,69 gradually decreases upon addition of Mg salt and almost disappears in the final composite indicating the favorable loss of crystallinity of the polymer upon the addition of the salt and ionic liquid.
We also carried out classical molecular dynamics simulations to fully understand the migration behavior of Mg 2+ ions in the polymer electrolyte (Figure 5a). Mean square displacement calculations demonstrate that the addition of ionic liquid significantly enhances the mobility of Mg 2+ ions in PVdF from almost no ion mobility. Figure 5b shows that the addition of ceramic fillers, which can adsorb free (ClO 4 ) -ions, also increases the diffusivity of Mg 2+ ions from 3.5 × 10 -10 to 1.0 × 10 -9 cm 2 s -1 . Such an increase of ion mobility is also supported by the calculated radial distribution functions (RDFs) (Figure 5c), which indicate that ceramic fillers lead to stronger electrostatic interaction between Mg 2+ ions and F atoms of PVdF chains (larger and sharper peaks in RDFs) and further facilitate the mobility of Mg 2+ ions through inter-and intramolecular PVdF chains. These computational results are in good agreement with Raman spectra analysis, confirming the role of both ionic liquid and ceramic fillers in improving the mobility of Mg 2+ ions. Evaluated RDF calculations of Mg-TFSI ion pairs in Figure 5d show a decrease of the RDF peak of around 2 Å, confirming a weak coordination between TFSI - and Mg 2+ ions as observed in the Raman spectra (Figure 4b).
These results suggest that a favorable Mg-ion transport within the polymer and deposition onto the electrode as observed could be explained by the combination of the following factors: (1) increased dissociation of salt into its constituent cations and anions, (2) absence of any significant coordination between Mg 2+ ions of the salt and TFSI -ions of the ionic liquid, and (3) amorphization of the polymer upon addition of Mg salt and TiO 2 fillers. Further optimization of these constituents and a systematic investigation into specific interactions between the ionic species and polymer can aid in extending this technique to other magnesium salts and polymer candidates.
In summary, a composite polymer electrolyte (CPE) has been developed for application in room-temperature all-solid-state rechargeable magnesium batteries. Galvanostatic cycling with symmetric cells shows that this polymer electrolyte supports excellent cyclability with Mg-metal. For cycling tests carried out at current densities of 0.05 and 0.10 mA cm -2 , low deposition/stripping overpotentials of 0.3 V for up to 400 cycles and of 0.2 V for up to 80 cycles, respectively, were observed. Interestingly, these solid-state Mg-deposition overpotential values are similar to those obtained for the present state-of-the-art liquid Mg electrolytes. Raman spectra analysis indicated an increased dissociation of the precursor salt upon the addition of ionic liquid and ceramic fillers. Furthermore, no significant interaction was observed between the TFSI -anions of the ionic liquid and Mg 2+ cations, which would otherwise impede the motion of Mg 2+ ions across the polymer. An optimization of the composition and further investigation into the deposition mechanism are necessary to understand the increasing overpotential values for high current densities (0.20 mA cm -2 ). Nevertheless, we believe that the development of this composite polymer electrolyte is a significant step toward enabling highly cyclable room-temperature solid-state magnesium batteries.
Synthesis of the Composite Polymer Electrolyte (CPE). An amount of 1.1 g of magnesium perchlorate (Mg(ClO 4 ) 2 , Sigma-Aldrich, Inc.) was dissolved in 3 mL of acetone. The ratio of the following constituents is expressed as weight percentage (wt %) with respect to Mg(ClO 4 ) 2 . In a separate vial, 30 wt % of poly(vinylidene fluoride-co-hexafluoropropylene) (PVdF-HFP) pellets (Sigma-Aldrich, Inc.) were dissolved in a few mL of acetone at 40 °C and added to the Mg(ClO 4 ) 2 solution. To this mixture, 75 wt % of ionic liquid, 1-butyl-1-methylpyrrolidinium bis(trifluoromethyl)sulfonyl imide (Pyr 14 -TFSI, Iolitec Ltd.), was added and stirred well. This was followed by the addition of 10 wt % of TiO 2 nanoparticle powder (US Research Nanomaterials, Inc.). This mixture was stirred in a magnetic stirrer at 60 °C for 12 h to get a white, viscous CPE blend. Prior to the synthesis, Mg(ClO 4 ) 2 , PVdF-HFP, and Pyr 14 -TFSI were dried in a vacuum for 12 h. TiO 2 nanoparticles were annealed at 150 °C in a vacuum for 24 h to remove any surface-adsorbed moisture.
Casting. Circular Mg disks of 15 mm diameter each were punched from magnesium foil (0.1 mm thick, MTI Corporation). The polymer was cast onto these circular Mg disks using a micrometer-adjustable film applicator. Prior to coating, the oxide layer on the Mg disk surface was scraped off using a blade until the shiny Mg-metal layer was revealed. The thickness of the polymer melt was adjusted to be within 60-80 μm. The casting was dried at 80 °C.
Morphological, Structural, and Thermal Characterization. The surface morphologies and elemental maps of the polymer sample and the cross-section of Mg foil | CPE were examined through scanning electron microscopy (Hitachi VPSEM S-3000N) linked with energy-dispersive X-ray analysis. Raman spectra of all the samples were obtained using a confocal Raman microscope (Renishaw InVia Reflex) employing a green 532 nm/50 mW diode pumped solid state laser. Thermal stabilities of the samples were analyzed using a TA Instruments model Q5000IR thermogravimetric analyzer (TGA). Classical Molecular Dynamics Simulations. All our molecular dynamics simulations were carried out using the parallelized LAMMPS package. 70 The force fields developed by Chaban 71 and Lopes et al. 72 as well as Lachet et al. 73 were utilized to describe both intramolecular and intermolecular interaction of ionic liquids Pyr 14 -TFSI and PVdF, respectively. A recently developed Lennard-Jones potential 74 was used to describe interaction of Mg-salt Mg(ClO 4 ) 2 . Cross-term interaction among the ionic liquid, PVdF, and Mg salt was described by the Lorentz-Berthelot rule. All the force field parameters are given in Tables S1-3 ) equals 2:1. A melt-and-quench procedure 75,76 was utilized to relax all amorphous models to their steady-state mixed states at different temperatures. Since the migration of Mg 2+ ions is too slow at room temperature (300 K), which requires extremely long simulation time, all the simulation models were run under NPT ensemble at temperatures of 350, 400, and 450 K to accelerate the Mg 2+ transport. Mean square displacements (MSDs) of Mg 2+ ions as a function of simulation time were calculated according to the equation where N is the total atom number; ⟨K⟩ denotes the average value over all atoms; and r i (0) and r i (t) are i-axis positions at simulation time 0 and t.
ACS Appl. Energy Mater. 2019, 2, 7980-7990