Article pubs.acs.org/cm
Efficient Aluminum Chloride−Natural Graphite Battery Kostiantyn V. Kravchyk,†,‡,§ Shutao Wang,†,‡,§ Laura Piveteau,†,‡ and Maksym V. Kovalenko*,†,‡ †
Laboratory of Inorganic Chemistry, Department of Chemistry and Applied Biosciences, ETH Zürich, Vladimir-Prelog-Weg 1, CH-8093 Zürich, Switzerland ‡ Laboratory for Thin Films and Photovoltaics, Empa-Swiss Federal Laboratories for Materials Science and Technology, Ü berlandstrasse 129, CH-8600 Dübendorf, Switzerland S Supporting Information *
ABSTRACT: The quest for low-cost and large-scale stationary storage of electricity has led to a surge of reports on novel batteries comprising exclusively highly abundant chemical elements. Aluminum-based systems, inter alia, are appealing because of the safety and affordability of aluminum anodes. In this work, we examined the recently proposed aluminum−ionic liquid−graphite architecture. Using 27Al nuclear magnetic resonance, we confirmed that AlCl4− acts as an intercalating species. Although previous studies have focused on graphitic cathodes, we analyzed the practicality of achievable energy densities and found that the AlCl3-based ionic liquid is a capacity-limiting anode material. By focusing on both the graphitic cathode and the AlCl3-based anode, we improved the overall energy density. First, high cathodic capacities of ≤150 mAh g−1 and energy efficiencies of 90% at high electrode loadings of at least 10 mg cm−2 were obtained with natural, highly crystalline graphite flakes, which were subjected to minimal mechanical processing. Second, the AlCl3 content in the ionic liquid was increased to its maximal value, which essentially doubled the energy density of the battery, resulting in a cell-level energy density of ≤62 Wh kg−1. The resulting batteries were also characterized by high power densities of at least 489 W kg−1.
■
INTRODUCTION Secondary (i.e., rechargeable) batteries are deeply integrated into every aspect of human life, including portable devices (smartphones, tablets, and laptops) and transportation (electrical cars, buses, bikes, etc.). Additionally, batteries are increasingly seen as global players in the efficient management of electricity from sustainable and renewable sources as well as for the stabilization and decentralization of the electrical grid. The simplest management concept is a household utility-level system combining a stationary battery with solar or wind electricity.1−3 Even for medium-level storage, the kilowatt-hour cost per charging cycle, in addition to the gravimetric and volumetric energy densities, becomes a major consideration for the commercial success of stationary storage systems.4,5 The volumetric energy densities are the highest for Li-ion batteries. However, for eventual deployment on the terawatt-hour scale, which is the scale of current worldwide hydroelectric storage, low manufacturing costs and high natural abundances of all the chemical elements constituting the battery become critical factors. With these goals in mind, battery technologies based on highly abundant metals, such as Na,6−10 Mg,11−14 and Al,15−30 as charge carriers and/or electrodes have become a major research focus. In particular, batteries that employ metallic Al as an anode can harness numerous advantages, such as its high natural abundance, high charge-storage capacity, and safety.31−34 In addition, Al can be reversibly deposited and stripped in chloroaluminate ionic liquids with a high © 2017 American Chemical Society
Coulombic efficiency of >99.5%, without formation of dendrites.35−38 Batteries employing chloroaluminate ionic liquids have received a great deal of attention since a publication by the group of Dai in 2015,15 wherein a metallic Al anode, two synthetic forms of graphite as cathodes (CVDgrown graphitic foam and pyrolytic graphite), and an AlCl3:EMIMCl (1-ethyl-3-methylimidazolium chloride) ionic liquid were combined into a battery, which is schematically depicted in Figure 1a. This battery design assumes that the intercalating species are AlCl4− ions, as described in early electrochemistry reports from the late 1970s.16,39,40 This assumption is in agreement with our 27Al nuclear magnetic resonance (NMR) spectroscopy results presented below. The graphitic cathodes in ref 15 delivered cathodic capacities of 67 mAh g−1 at an average discharge voltage of 2.0 V. The authors of ref 15 initially stated that natural forms of graphite exhibited performances much poorer than those of both tested synthetic graphitic forms. The high porosity of the synthetic graphite was considered an important factor for the fast and highly reversible intercalation of AlCl4− ions. Subsequent studies by the same group and by others further increased the capacities and power densities using porous synthetic graphitic materials, such as graphitic foams41 and graphene nanoribbons (record-high capacity values of 148 mAh g−1).42 However, the use of costly Received: March 14, 2017 Revised: May 1, 2017 Published: May 1, 2017 4484
DOI: 10.1021/acs.chemmater.7b01060 Chem. Mater. 2017, 29, 4484−4492
Article
Chemistry of Materials
Figure 1. Aluminum chloride−graphite battery. (a) Schematics of the charging process. (b) Comparison of the calculated (curves) and experimental (data points) cell-level energy densities. The curves are computed from eq I, which describes the dependency of the energy density vs Cc at various values of r assuming an average discharge voltage of 2 V. The experimental data points represent this work or estimates from reported voltages and Cc.15,42−45 (c) Graphite ore. (d) Large natural graphite flakes produced commercially from graphite ore.
reduced to Al atoms for electrodeposition and simultaneously generates AlCl4− for intercalation. Therefore, severe requirements are posed on the amount of used ionic liquid, as the ionic liquid, in fact, acts as a capacity-limiting liquid anode and not just as an electrolyte. Thus, we tested AlCl3-rich formulations of this battery up to the highest achievable AlCl3:EMIMCl ratio of 2 and thereby improved the cell-level energy density to 62 Wh kg−1. We discuss the reasons why a further substantial increase in energy density is likely impossible. However, we highlight the high energy efficiencies (≤91%) and excellent power densities of the Al chloride− graphite batteries with values of at least 489 W kg−1 at high areal loadings of graphite (≥10 mg cm−2).
CVD processes could pose an unsurmountable obstacle for the further practical use of such batteries, fully neutralizing the cost benefits associated with the simplicity of the battery and the abundance of the constituent elements. In parallel, similar charge-storage capacities were obtained from electrodes made of carbon paper,25,27,43 which presumably was produced from natural graphite,27 and from a few-layer graphene aerogel.44 The latter was also derived from natural graphite through its oxidation into graphite oxide followed by reduction. Encouraged by both the initial report and some of these recent publications, herein, we sought to thoroughly test the most basic and inexpensive form of natural graphite, graphite flakes, as the cathode material in such Al batteries. We found that natural graphite flakes, with minimal processing by sonication, delivered a charge-storage capacity of 148 mAh g−1, at an average voltage of 2 V. We observed that open graphite edges and flaky morphology with preserved pristine crystalline structures (low density of crystalline defects) were crucial for achieving such high charge-storage capacity. Notably, Dai et al. also reported on natural graphite flakes as cathodes during the preparation of this work, using a somewhat different electrode formulation and achieving a high charge-storage capacity of 110 mAh g−1 at a current density of 99 mA g−1.45 We also present direct spectroscopic evidence of the intercalation of AlCl4− using solution- and solid-state 27Al NMR spectroscopy. Furthermore, we analyze and discuss the challenges related to the eventual practical use of this battery technology. Moreover, we emphasize that an Al chloride− graphite battery is not a rocking-chair battery, contrary to Liion batteries, and should not be called an “Al-ion battery” because the species that intercalate and/or deintercalate into the graphitic cathode (AlCl4−) are different from those involved in the simultaneous anodic process (i.e., exclusive deposition and stripping of Al atoms and/or ions from the Al foil). This is not only a matter of correct definitions. A portion of AlCl3 acts as an anode material during the charging process, whereby it is
■
RESULTS AND DISCUSSION Operation Mechanism and Energy Density of the Al Chloride−Natural Graphite Flake Battery. Contrary to a Li-ion battery utilizing graphite as an anode, an Al chloride battery exploits the reversible oxidation of the graphite network, i.e., its cathodic functionality. The non-rocking-chair, i.e., mixed-ion, operation of the battery is clearly seen from Figure 1a. The following half-reactions take place upon charging: on anode: 4Al 2Cl 7− + 3e− ↔ 7AlCl4 − + Al on cathode: xC + AlCl4 − ↔ Cx (AlCl4 −) + e−
(I) (II)
There is no unidirectional flow of Al ions or any other Al species from one electrode to the other. For this reason, it is misleading to call this battery an “Al-ion battery”, although this term has become commonplace since the publication by Dai et al.15 The actual operation mechanism is, however, fundamentally different from that of metal-ion batteries: Al species are depleted from the liquid phase during the charge process (Figure 1a). AlCl3:EMIMCl (i.e., EMIM+ + AlCl4−), a mixture of two solids that become a liquid at room temperature (an 4485
DOI: 10.1021/acs.chemmater.7b01060 Chem. Mater. 2017, 29, 4484−4492
Article
Chemistry of Materials
Figure 2. Electrochemical characterization of the pristine flakes (black) and processed smaller flakes (blue) compared to potato-shaped particles produced commercially (red) and in house (light green). (a) Representative SEM images of the pristine graphite flakes and of two products of their mechanical processing by sonication (smaller flakes) and by knife-milling (partially spheroidized, i.e., potato-shaped). (b and d) Galvanostatic charge and discharge voltage curves and (c and e) cycle stability measurements for the pristine, processed smaller flakes and both potato-shaped specimens with respect to Li+- or AlCl4−-ion storage. The same color in the legend represents the same material throughout all plots. Electrochemical measurements in Li-ion half-cells [with 1 M LPF6 in ethylene carbonate/dimethyl carbonate (1:1 vol %) as an electrolyte and metallic lithium as a counter electrode] and in AlCl3−graphite batteries (r = 1.3). Panel c also plots the Coulombic efficiency, whereas panel e presents the energy efficiency. Typical Coulombic efficiencies of the AlCl3−graphite batteries are presented in Figure S3.
ionic liquid) due to the acid−base reaction forming AlCl4−, is not only an electrolyte but also a source of electroactive species. AlCl3 is added in excess to allow Al2Cl7− ions to form. The electrodeposition of Al ions from chloride-based ionic liquids has been thoroughly studied in the past, leading to the consensus that Al2Cl7− ions allow the electroplating of Al.35,36,46−53 Al does not deposit from neutral or basic melts (excess of EMIMCl). Clearly, the amount of Al2Cl7− ions, and hence the whole mass of the liquid phase, must match the capacity of the graphitic cathode. For each electrodeposited Al atom, three AlCl4− anions simultaneously intercalate into graphite. The charging process ends when there are only AlCl4− ions left, i.e., the neutral melt forms (AlCl3:EMIMCl = 1), or when the maximal capacity of the graphitic cathode is reached. The highest molar ratio (r) of AlCl3 to EMIMCl that still forms a usable ionic liquid is ∼2:1; above this ratio, AlCl3 no longer dissolves in the solution. The operation of such a battery is not uniquely limited to EMIM-based formulations: AlCl3−urea28,54 or fully inorganic NaAlCl4 melts27 have been recently presented as cost-efficient alternatives. We also note that starting with an Al anode is not a necessity. Any current collector supporting the initial electroplating of Al or coated with a minimally thin Al film as a seed layer is sufficient. Therefore, we call this
battery an Al chloride−graphite battery, highlighting the fact that AlCl3 can be viewed as an actual anode material dissolved in a stoichiometric mixture of AlCl3 and EMIMCl. Therefore, anodic reaction I can be reduced to AlCl3 +
3 − 3 1 e ↔ AlCl4 − + Al 4 4 4
(III)
The quantity of the neutral AlCl3:EMIMCl melt is linked to the excessive AlCl3 via r. Considering the whole mass of the liquid anode, the theoretical charge-storage capacity of this battery on a cell level can be determined starting from the standard relationship Ctotal = (CACC)/(CA + CC),55,56 wherein the weights of both the cathode and liquid anolyte are factored in C total = =
Fx(r − 1)CC Fx(r − 1) + CC(rMAlCl3 + MEMIMCl) 20.1(r − 1)CC × 103 20.1(r − 1) × 103 + CC(rMAlCl3 + MEMIMCl) (1) −1
where F = 26.8 × 10 mAh mol (Faraday constant), x = 3/4 (number of electrons used to reduce 1 mol of the anodic 3
4486
DOI: 10.1021/acs.chemmater.7b01060 Chem. Mater. 2017, 29, 4484−4492
Article
Chemistry of Materials
have focused exclusively on natural graphite flakes and the use of minimal processing or the simplest processing of these flakes by mechanical means. We started with large commercial natural graphite flakes (pristine; 0.2−1 mm in lateral size), which were commercially produced from graphite ore by mechanical means (crushing, wet milling, and floatation). The production is often combined with acidic chemical treatment for additional purification. In addition to using these pristine flakes as received, we also fragmented them into smaller particles, resulting in either a retained flat morphology with unfolded edges [processing by sonication; ∼40 μm in lateral size (Figure 2a, left)] or an imparted partial spheroidization [∼50 μm; processed by knife-milling (Figure 2a, right)]. The particle shape after spheroidization resembled the “potato” graphite particles commonly used in Li-ion batteries. A custom-made cell design with Swagelok fittings was employed for the electrochemical tests (Figure S2). The cell consisted of aluminum (anodic side) and tungsten or glassy carbon (cathode side) current collectors combined inside a fully plastic housing. The cathodes consisted of pure graphite (10 mg cm−2), without the use of any binder or conductive additives, to ensure that the measured electrochemical characteristics could be fully attributed to graphite itself. The electrodes were moderately pressed against each other with a glass-fiber separator between them. No preferential orientation of the graphite flakes with respect to the current collectors was observed by optical microscopy. All graphite electrodes were tested not only as cathodes in the AlCl3−graphite cells but also in standard Li-ion cells. All such comparative tests were conducted at moderate current densities of 100 mA g−1. Our findings are indeed orthogonal between the experiments of Li-ion storage (Figure 2b,c) and intercalation of AlCl4− ions (Figure 2d,e; all Al cells tested at r = 1.3). In particular, the Li-ion tests readily confirm that the commercial potato-shaped particles and our spheroidized particles (also termed potato-shaped in Figure 2) deliver capacities of at least 360 mAh g−1, near the theoretical chargestorage capacity (372 mAh g−1). The 2−3-fold poorer Li-ion storage in the flaky graphite particles is in good agreement with results of numerous kinetic studies of the effects of their orientation and microstructure,57−61 which have all highlighted the accessibility of the edges and particle sizes as key factors for Li-ion storage. In this context, a recent study by Billaud et al.62 showed that dense vertical staking of graphite flakes perpendicular to the current collectors allowed for fast intercalation into laterally large flakes. In the absence of such favorable orientations, the potato-shaped particles remain an optimal morphology for industrial applications. However, a strikingly different scenario is found for the insertion of AlCl4− ions. With large pristine flakes, a high capacity of 95 mAh g−1 (Figure 2d,e) and a high energy efficiency [90% (Figure 2e)] at an average discharge voltage of ∼2 V were observed. A further capacity increase to 132 mAh g−1 was obtained with smaller flakes, retaining the superb flatness of the voltage profiles. On the other hand, commercial and in-house potato-shaped graphite particles exhibited lower capacities for AlCl4− ions, which is in striking contrast to the results for Li-ion storage. Importantly, while the graphite anodes are prone to irreversible capacity loss, as seen from the reduced Coulombic efficiencies of 70−80% (Figure 2d) in the initial cycles as a result of the formation of a solid−electrolyte interface (SEI), the operation of AlCl3−graphite batteries proceeds from the beginning with
material, i.e., AlCl3), r is the AlCl3:EMIMCl molar ratio, CC and CA are the specific capacities of the cathode and anode, respectively, in milliampere-hours per gram, MAlCl3 is the molar mass of AlCl3 in grams per mole, and MEMIMCl is the molar mass of EMIMCl or of any other Cl− source (i.e., HCl in the simplest case) in grams per mole. Ctotal is an upper bound for a capacity, as in the commercial battery it is further reduced by 25−50% because of the weight of current collectors, separators, and packaging (determined by the battery design). Using eq I, we analyzed the effects of increasing r and CC on the overall energy density of this battery (E = CtotalV). In Figure 1b, we assume an average voltage of the battery to be constant at 2 V for all theoretical curves. This value of the average discharge voltage is the highest value we obtained experimentally at r = 1.3. At r > 1.3, the measured average voltage dropped progressively by up to 0.2 V [from ∼2 to ∼1.8 V (Figure S1)]. The primary conclusion is that parameter r is the most useful variable, in addition to the cathodic capacity of graphite, for improving the energy density of the battery because of the capacity-limiting effect of the ionic liquid. This capacity-limiting effect can be illustrated by calculating the theoretical CA, considering the whole mass (volume) of the ionic liquid as a liquid anode (i.e., as an anolyte). The resulting values are 48 mAh g−1 (63 Ah L−1) and 19 mAh g−1 (24 Ah L−1) for r values of 2 and 1.3, respectively. Hence, to achieve an optimal energy density, the highest capacity of the cathode (CC) must be harnessed and a concomitant increase in r must also be included. As a reference point, we analyzed all available literature on AlCl3−graphite batteries and identified the highest experimental energy density of ∼45 Wh kg−1 for an AlCl3:EMIMCl anolyte at r = 1.8, a graphitic capacity of 57 mAh g−1 (obtained at r = 1.8), and a corresponding discharge voltage of 1.88 V (see the Supporting Information of ref 15). All other reported data in the literature were obtained at r = 1.3. At this lower value of r, the energy density does not exceed 30 Wh kg−1 for experimental graphitic capacities of 70−110 mAh g−1. An increase in r leads to a concomitant substantial decrease in CC and the average discharge voltage, as also mentioned in ref 15, and is also apparent from our measurements (Figure S1). As detailed below, we have experimentally examined this trade-off for natural graphite flakes, leading to the highestreported energy density of 62 Wh kg−1 (91 Wh L−1) calculated from a capacity of 124 mAh g−1 (273 Ah L−1) at r = 2 and an averaged discharge voltage of 1.77 V (Figure S1). For comparison, the highest graphitic capacity of ∼150 mAh g−1 (330 Ah L−1) was obtained at r = 1.3; nevertheless, this corresponds to a much lower energy density of 33 Wh kg−1 (45 Wh L−1). With the most inexpensive form of graphite (natural flakes) as a cathode and all other components being composed of highly abundant elements (Al, Cl, C, N, and H), the AlCl3− graphite battery might become a viable technology for largescale stationary storage purposes. In this domain, this battery may find use as a nontoxic alternative to the primary battery technologies of similar energy densities: lead−acid (30−50 Wh kg−1; 50−80 Wh L−1) and vanadium redox-flow batteries (VRB; 10−30 Wh kg−1; 50−80 Wh L−1).4 AlCl3−graphite batteries exhibit high energy efficiencies (∼90%), which are, again, on par with those of lead−acid batteries (90%) and VRB (85%) or even Li-ion batteries (90−95%). Testing Natural Graphite Flakes as Inexpensive Cathodes. To comply with stringent cost requirements, we 4487
DOI: 10.1021/acs.chemmater.7b01060 Chem. Mater. 2017, 29, 4484−4492
Article
Chemistry of Materials
charge, followed by various discharge rates). As seen in the standard CC measurements (Figure 3c; same rates for charge and discharge), higher discharge capacities were obtained at charging/discharging current densities of ≤200 mA g−1, which decreased at higher rates. A slower CCCV (50 mAh g−1 for a CC step) protocol indicates the unique rate capability of these batteries: the same high discharge capacities were obtained up to the highest tested discharge rates of 2.5 A g−1, i.e., 3.6 min for the complete discharge. The corresponding power density was as high as 489 W kg−1, which compares favorably with those of the other stationary battery technologies such as lead acid (75−300 W kg−1)4 and VRB (60−100 W L−1).63 We then examined the effect of the AlCl3 molar fraction by adjusting parameter r in the range of 1.1−2.0 (highest possible) while applying the CCCV charging protocol. A clear trend was observed in which a higher r resulted in a greater loss of voltage (by ≤0.2 V). Also, the graphitic capacities dropped from ∼150 mAh g−1 (r = 1.3) to ∼124 mAh g−1 (r = 2). These losses are lower than the theoretically expected energy densities (stars as data points in Figure 1b). Despite these effects, the highest energy density of 62 Wh kg−1 was obtained at r = 2, highlighting the importance of the concerted focus on both the cathode and the anode. Benefits of the Flaky Morphology for Achieving Higher Capacities from Natural Graphite. Powder X-ray diffraction (XRD) patterns of the charged graphite flakes [a capacity of ≤132 mAh g−1 (Figure 4a)] indicate that the (002) graphite peak fully vanishes, and the appearance of three additional peaks indicates that the intercalation of the ions occurs in an alternating manner. Such a “staging” intercalation mechanism is known for various intercalation compounds with graphite (see the illustration of stages 0−6 in Figure S4).64,65 A lower stage number indicates a higher intercalant concentration. Recent theoretical calculations24 have shown that intercalation of AlCl4− into graphite can be observed by the formation of stages 4, 3, and 2. Simulated XRD peak positions qualitatively fit the experimental pattern, suggesting a mixed stage 3−4 intercalation (Figure 4a). The insertion of ions leads to the expansion of the graphite volume. Larger AlCl4− anions (295 pm) lead to a drastic expansion by up to 80%, as seen from recent experimental and theoretical works,24,42,45,54,66−68 which is in sharp contrast to the minor (∼10%) effect from smaller Li+ ions (90 pm).69 This may explain the highly beneficial role of the flake morphology: large volumetric changes are easily accommodated by the homogeneous increase in thickness. On the other hand, folding, bending, or other geometric constrains present in other shapes, such as in potatoshaped graphite, might hamper the high loading of intercalating anions. It is natural to expect that variations in the synthesis/ processing conditions not only modify the overall shape of the graphite particles but also introduce various degrees of structural disorder into the idealized lattice of crystalline graphite, which might also contribute to the electrochemical performance. For example, powder XRD patterns of pristine, flake, and spherical graphite particles revealed an increase in the interplanar graphene−graphene distance from 3.354 to 3.369 Å (see Table S1), as can be clearly observed from the shift of the (002) peak to smaller 2Θ values (Figure 4b). Similarly, the Raman spectra of the spherical particles were characterized by an increased magnitude of the signal at 1250− 1400 cm−1 (D-band), which is usually attributed to structural disorder. Higher defectiveness, as seen from the larger d spacing
higher Coulombic efficiencies of >92% (Figure S3), suggesting SEI-free operation. CCCV Charging Protocol for Combining High Capacities and High Power Densities. Having established that the mechanically processed graphite flakes offer higher chargestorage capacities, we examined this form of graphite in greater detail (Figure 3). First, we found that the CCCV (constant
Figure 3. Electrochemical performance of the processed graphite flakes in the AlCl3−graphite battery (r = 1.3). (a and b) Galvanostatic charge−discharge voltage curves and cycling performance measured using CC and CCCV protocols. (c) Rate capability measurements obtained with two protocols: standard (same current densities for charge and discharge during the cycle) and CCCV (50 mAh g−1 CCCV charge and different discharge currents).
current−constant voltage) charging protocol, which involves constant charging up to 1.9 V followed by one or more constant voltage steps (in the range of 1.9−2.1 V; terminated at a current drop of 90%), boosts the subsequent discharge capacity to 150 mAh g−1. It should be noted that constant voltage steps at higher voltages (>2.1 V) were detrimental because of side reactions that reduced the Coulombic efficiency. Constant voltage steps at lower voltages (150 mAh g−1 would have a marginal effect on the overall energy density. We suggest that future work should focus on the other issues associated with this technology, one being the incompatibility of most metallic current collectors with the corrosive AlCl3-based ionic liquids. Even gold slowly dissolves in AlCl3:EMIMCl ionic liquid when electrochemically polarized up to 2.5 V versus Al. Thus far, only tungsten and molybdenum have been identified as electrochemically stable current collectors in such batteries. These metals do not fulfill the criterion of high natural abundance, when compared to the abundance of the other components of the AlCl3−graphite battery or to Li ions. There is a pressing need to identify current collectors made of alternative alloys or compounds comprising elements of higher natural abundance and available at low manufacturing costs.
■
EXPERIMENTAL SECTION
Chemicals and Battery Components. Pristine graphite flakes (99.9%, ∼10 mesh, Alfa Aesar), spherical graphite powder (AE 104L, TIMCAL), 1-ethyl-3-methylimidazolium chloride (EMIMCl, 99%, Iolitec), AlCl3 (99%, granules, Acros), Al foil (MTI Corp.), a W plate (Bocheng Molybdenum Co., Ltd), a glassy carbon plate (GoodFellow), and a glass microfiber separator (GF/D, catalog no. 1823257, Whatman) were used as received. Preparation of Electrolytes. Ionic liquid electrolytes based on EMIMCl were prepared by slowly mixing EMIMCl solid powder and AlCl3 granules in an argon-filled glovebox. During mixing, a highly isothermal reaction occurs, eventually forming a light-yellow liquid. Subsequently, the ionic liquid electrolyte was treated with Al foil at 150 °C for 6 h until a nearly colorless liquid was seen. Sonication of Pristine Graphite Flakes. Pristine graphite flakes (0.2 g) were placed into a 4 mL glass vial, and the vial was filled with 3.5 mL of ethanol and sonicated for 30 min using a model HD2200 Sonopuls ultrasonic homogenizer operated at 30% power. The resulting smaller flakes were washed three times with ethanol and dried under vacuum at 80 °C for 12 h. Knife-Milling of Pristine Graphite Flakes (spheroidization). Pristine graphite flakes (10 g) were placed in a kitchen blender (Betty Bossi from Fust MixFIT) and knife-milled for 30 min. During the milling, a nitrogen flow was added to the container to cause more efficient stirring of the flakes. The knife-milled graphite particles were dried under vacuum at 80 °C for 12 h. Assembly and Testing of AlCl3−Graphite Batteries. All graphite samples were dried at 200 °C under vacuum overnight. No binders or solvents were used for the preparation of the electrodes. Homemade plastic cells were assembled in an argon-filled glovebox (
