Advanced Materials · 2015 · 164 citations · 26 references
EngineeringChemical EngineeringChemistryElectrochemical EngineeringElectrochemistryEnergy MaterialsBattery Cell ChemistryBattery MaterialsElectrochemical CellsElectrochemical Energy StorageEnergy StorageRedox-active MaterialsEnergy DensityFlow BatteriesIonic Liquid ElectrolytesLithium-ion Batteries
A redox-active supercooled liquid is obtained by forming a “solvate ionic liquid” from a mixture of a stabilized organic radical and a specific Li salt and stabilizing the mixture further below their melting points. As a catholyte, the addition of an appropriate amount of water helps to enhance the electrochemical advantage while maintaining its supercooled nature and the liquid shows a high energy density of 200 W h L−1 with reversible charge/discharge. New battery materials are important for a broad range of technology areas ranging from small electronic devices to electrified vehicles to stationary energy storage. In addition to the improvements in state-of-the-art Li-ion batteries, so-called “beyond Li-ion battery” technologies have attracted considerable attention due to their promising potential, their scientific and engineering challenges, and because the demand and expectation for new energy storage devices are quite high. One of the emerging battery technologies is the flow battery, which utilizes liquid active materials, such as catholytes and anolytes. The unique and significant benefits of the flow battery are its simple structure and flexible handling of its liquid active materials as “electrochemical fuels,” as well as its scalability and low maintenance.1 Despite this great “fueling” capability, which is suitable for the rapid charging of electric vehicles, the energy density of the regular catholytes and anolytes is lower (<100 W h L−1) than that of the solid-state active materials used in Li-ion battery. Only three parameters need to be improved to increase the energy density of the catholyte: i) the discharge voltage, ii) concentration, and ii) valence number. A non-aqueous system, which overcomes the 2 V limitation caused by water decomposition, is a simple and effective method of improving the discharging voltage coupling with a highly negative anode close to the Li/Li+ potential.2 However, the concentration of redox species in organic solvents is generally lower than that in aqueous solutions. Despite many trials to dissolve organic redox species, it is difficult to obtain a solution higher than 2 m (Figure 1a, left side). Although the dissolution of corrosive inorganics[2] and the dispersing of small solid particles (slurries)[2] provide an opportunity to have a concentration higher than 2 m, these strategies may also be associated with unwanted safety and/or durability issues. Redox-active ionic liquids are also an attractive approach to reach high energy density; however, the previously reported study has an environmental concern due to the use of heavy metals and lower energetic density caused by the bulky cations.3 In addition, the liquids should have a stable fluidic property even in a low temperature range without freezing, especially for automobile applications. Here, we propose to change the strategy from maximizing the solubility of redox compounds to minimizing the dissolving solvents (Figure 1a, right side). The minimum requirements for the catholyte are stoichiometric couples of a redox species and appropriate supporting salt. Because the valence of the redox species must be changed during discharging and charging, compensating ions must be present in the system. Therefore, a redox couple with a single valence change (i.e., A/A+, A−/A, or A+/A++) requires the same molar of monovalent salt (i.e., X+B−), and their 1:1 molar mixture yields the maximum electronic capacity for batteries. The challenge is then determining how to liquefy this couple. First, we selected 4-methoxy-2,2,6,6-tetramethylpiperidine 1-oxyl (MT or MeO–TEMPO) as a redox compound. MT has excellent chemical, physical, and electrochemical stability,[4] similar to non-substituted TEMPO,5 and also has a low melting point (Tm) (42 °C). Next, we focused on lithium bis(trifluoromethanesulfonyl) imide (LT or LiTFSI) as a supporting salt because TFSI anion was one of the most interesting ions with an unusual plasticizing effect, which often realized low Tm mixtures,6 plastic crystals7 and supercooled liquids, whereas the Tm of LT was rather high (230 °C). In previous studies, the unique effect has been understood by the weak Lewis-basic property due to its charge delocalization and isoenergetic conformation change.8 Then, as shown in Figure 1b (right side), a simple 1:1 molar mixture of MT and LT exhibited a self-melting behavior and formed a smooth viscous liquid. Finally, by the addition of small amount of an appropriate solvent (i.e., acetonitrile (ACN), water, etc.) a highly concentrated (over 2 m) and low-viscosity liquid could be prepared (Figure 1b, center). In this report, the mixture of MT and LT with molar ratios of x and y, respectively, will be described as “MTLT(x/y).” For example, MTLT(1/1) denotes a 1:1 molar mixture of MT and LT. Mixtures ranging from MTLT(1/1) to MTLT(20/1) are found to yield orange-colored smooth liquids at room temperature. Because i) a series of sulfonylimide-based salts, such as Li bis(fluorosulfonyl) imide (LiFSI) or Li bis(pentafluoroethanesulfonyl) imide (LiBETI), exhibit similar properties and ii) other inorganic Li-salts, such as LiPF6, or LiBF4, do not exhibit this unique feature (Figure 1b, left), the sulfonylimide structure could be critical to understanding this unique property. With regard to TEMPO, even non-substituted TEMPO forms a similar liquid when mixed with LT; however, the liquid phase is not stable can be crystallized more easily than the MT system. We first describe the properties and melting mechanism of solvent-free MTLT systems followed by the solvent-added systems, and then demonstrate the battery performance using MTLT liquid as a catholyte in both static and flowing conditions, which is the first approach to liquefy organic redox-active materials by using TFSI− for a high-energy-density catholyte of flow battery. To investigate the details of the melting mechanism of solvent-free molten MTLT mixtures, the thermal behaviors of MTLT(x/y) were monitored using differential scanning calorimetry (DSC). The heat flows during heating are shown in Figure 2a, and the behaviors during cooling are shown in Figure S1, Supporting Information. Whereas pure MT has a clear Tm at 41.9 °C and its Tm decreases with a decrease in x due to the “Tm depression” effect, when x ranges from 6 to 10, the mixtures have two peaks corresponding to the crystalline temperature (Tc) and Tm. In the temperature region between the glass transition temperature (Tg) and Tc, the mixtures behave like liquids and can be recognized as supercooled liquids. In the x region below 6, no further Tc appears by either heating or cooling, and only Tg is observed. The obtained phase diagram (Figure 2b) indicates the appropriate composition and temperature range to prepare the supercooled liquids and proves the potential of MTLT(x/y) for application over a wide temperature range. The supercooled liquids in this composition range are highly stable and can be used as regular liquids. For example, MTLT(5/1) kept around 8 °C longer than 100 days indicated a stable supercooled liquid phase nature (Figure 2b, photo) and MTLT(3/2) also did not show any crystallization at least over one year in ambient condition (Figure S2, Supporting Information). We employed Raman spectroscopic analysis to elucidate the melting mechanism and structure, and the spectra of MTLT(x/y) are presented in Figure 3. Highly MT rich samples (x > 10) became solid in the measurement condition. The peak at 1386 cm−1, which is attributed to MT’s N–O• stretching vibration, decreased with the appearance of a new peak at ≈1402–1408 cm−1 with decreasing x (Figure 3a), and the peaks were nearly completely switched at MTLT(1/1) (Figure 3c). Meanwhile, in Figure 3b, TFSI−′s entire stretching vibration in MTLT(x/y) (x > 2) at ≈741 cm−1 can be assigned to solvent-separated ion pairs (SSIPs) by the excess amount of MT, and the peak linearly switches to 747 cm−1 when x approaches 1 (Figure 3d), which is identical to that of crystalline LT, suggesting the likelihood of an ordered ionic structure with Li+ (contacted ion-pair).[8],9 These facts indicate 1:1 molar coupling between MT (at the N–O• site) and Li+, leading to the formation of a new ionic couple between (Li–MT)+ and TFSI−. A similar example is the 1:1 molar coupling between tetraglyme and LT, which are classified as rare examples of “solvate ionic liquids.”[8],10 This coupling also has a scientifically interesting solvation phase recognized as a liquefied ion pair stabilized by the neutral third compound. For the next step, we investigated an effect of the addition of solvents in molten MTLT because MTLT(1/1), which has the highest energy density as a catholyte in this couple, exhibits a high viscosity (η) of more than 10 000 mPa s at room temperature and cannot be easily employed in battery applications. To lower η to within an acceptable range effectively, we considered an addition of water by keeping the system highly concentrated. From this point, “MTLT(1/1) + H2O(z)” denotes a liquid consisting of MTLT(1/1) and z wt% of water. Based on actual measured density values of the liquids, the relationship between the water contents and the MT concentration in MTLT(1/1) + H2O can be expressed in Figure 4a and indicates an excellent potential to have higher MT concentration than 2 m. Theoretical capacity of the catholyte can also be calculated using the concentration values by taking into account the expected battery mechanism (Figure 4a, inset). It is important to note that the solubility of MT in water is only ≈0.3 m, and then the water should not be considered as a simple dissolving solvent. The thermal behaviors of MTLT(1/1) + H2O(17) clearly indicate the lack of bulk water properties, such as icing, and the MTLT(1/1)+H2O maintains a highly stable supercooled liquid phase in the lower temperature range (<−50 °C) compared with the water-less MTLT(1/1) system (Figure S3, Supporting Information). These results indicate that the added water has the ability to plasticize the MTLT system while maintaining its “solvate ionic liquid” nature. A water content higher than 29 wt% caused unwanted icing behavior (crystallization of water), which narrows the temperature ranges of the liquid phase and the supercooled liquid. The η and conductivity (σ) of MTLT(3/1) and MTLT(1/1) + H2O were measured in the temperature range of −30 to 60 °C and are summarized as a Walden plot (Figure 4b). Arrhenius plots of η and σ are also shown in Figure S4a–c, Supporting Information. In any case, for the MTLT(x/1) with and without water, the slopes are nearly identical, indicating that the fundamental ion-conducting mechanism is nearly identical, and for MTLT(3/1), the plot is classified as a “poor” ionic liquid[8] because of its high η and low σ. However, the MTLT(1/1) + H2O exhibits better performance and is classified as a “good” ionic liquid, and MTLT(1/1) + H2O(17), as an example, is characterized by η = 72 mPa s and σ = 3.1 mS cm−1. Figure S5, Supporting Information presents the Raman spectra of the water-containing mixtures and the new peak shoulder near 1390 cm−1 indicates the water coordinated (solvated) structure of MTLT(1/1). The details of the coordination effect are now being evaluated using analytical and computational methods. Because a large amount of water lowers the energy density at the expense of a reduced η, the composition must be optimized to meet the application's demands. For the catholyte, MTLT(1/1) + H2O(17) was selected due to its well-balanced energy density and viscosity. Before evaluating the actual battery performance, the electrochemical properties were measured using cyclic voltammetry. As observed in Figure S6, Supporting Information in the potential range from 2.75 to 4.75 V, reversible redox peaks appear near 3.5 and 4.0 V as reduction and oxidation reactions were observed, respectively. The redox can be considered as the redox reaction between neutral (radical) MT and oxidized (cationic) MT (Figure 1b). In the wider voltage range from 1.75 to 4.75 V, the reduction peak at 2.2 V appears difficult to be oxidized again reversibly, which predicts the instability of reduced (anionic) MT. Therefore, the former MT/MT+ redox couple was selected for the following battery performance evaluations. The electrochemical stability of water in this potential range was also confirmed using the same technique using a MT-less electrolyte (Figure S6, green dotted line, Supporting Information). For the battery tests, we designed both static and flow cells with a Li+ conducting architecture by applying a solid-state Li-ion conductor to separate the cathodic and anodic rooms (Figure S7 and S10, Supporting Information). The charge/discharge curves using the static cell are presented in Figure 5a with different current densities. The successful formation of MT+ after charging was confirmed by IR identification (Figure S8, Supporting Information) by comparison with the spectrum of chemically prepared MT+ (MT+ClO4−). The charged and discharged capacities at 0.1 mA cm−2 were 93% and 92% of the theoretical capacity (1.21 mA h), respectively, and indicated the good solubility and Coulombic efficiency (99%) of MT+. The energy density and specific energy of MTLT(1/1) + H2O(17) were calculated to be 200 W h L−1 (= 3.6 (V) × 1.11 (mA h)/20 (μL)) and 155 W h kg−1 ( = 200 (W h L−1)/1.288 (kg L−1)) using the above values, respectively, and these values prove that this catholyte has the highest energy densities using organic redox compounds at the moment. Although the rate capability will be strongly dependent on the current collecting structure, a simple carbon-paper-electrode allows keeping high capacity even in higher current density such as 1.0 mA cm−2. The cycling property shown in Figure 5a (inset) indicated a good capacity retention (84% retention after 20 cycles), and based on our observation of a slight precipitation after charging, increasing the solubility of MT+ salt could promote the coulombic efficiency and capacity retention after cycling. In this report, we mainly selected the water added system as the most excellent example; however, the water can be replaced by the other appropriate organic and the battery performance using ACN is also shown in Figure S9, Supporting Information. In this case, conventional porous polymer separators may be applicable as a replacement from the Li-ion conducting ceramics to have a simpler battery structure due to the compatibility with the Li-based anodes. We also demonstrated an actual circulating type flow batteries using the same catholyte (Figures 5b and S11, Supporting Information), and the details of the setup and evaluation can be found in Figure S10, Supporting Information. The charge/discharge behaviors clearly show stable flowing capability with a wide range of flow-rate and prove facile scalability of the developed catholyte. We have proposed and demonstrated a new class of catholyte employing a stabilized supercooled liquid through the combination of a specific organic radical (MT) and plasticizing Li salt (LT). The catholyte has the unique feature of forming an MT–LT coupled structure to stabilize its supercooled state over a wide temperature range and can be recognized as a “solvated ionic liquid.” The addition of water to this supercooled liquid does not damage its unique feature of having one of the highest energy densities as a pure liquid using organic redox compound but improves the parameters required for battery applications, such as viscosity and conductivity. The performance of the battery with a Li-ion-conducting architecture has proved this battery's potential for high-energy-density flow battery applications because the developed supercooled liquid consists of environmentally friendly compounds (non-corrosive and heavy-metal-free). The flow battery demonstrated in this report has also proved facile scalability and high adaptability for actual flowing architectures. The simple molecular structure of MT and facile preparation capability of the catholyte also give a large potential to reduce a production cost of the battery system. The strategy for creating a high-energy-density catholyte can be used as a universal method to prepare other organic redox-active liquids and will be reported elsewhere in the near future. Materials: MT (>98.0%, Tokyo Chemical Industry), LT (99.80%, Kishida Chemical), and all the other materials were used as purchased without further purification. To prepare the mixture of MT and LT, an equivalent molar amount of MT and LT were mixed in an Ar glove box. Slight heat treatment (e.g., 60 °C, 3 h) of the mixtures ensured the homogeneity of the liquids. The water-containing samples were prepared by adding an appropriate amount of deionized water in an ambient atmosphere. Measurement of the Liquid Characteristics: DSC measurements were performed using a TA Instruments DSC-Q100. A 10–20 mg sample was placed in an aluminum pan (TA, TZero) and hermetically sealed. The temperature sweep was started by cooling to −90 °C and then heating up to 80 °C with a sweeping rate of 2 °C min−1. Raman spectroscopy measurements were performed using a Horiba Scientific LabRAM HR800 with a 532 nm laser. The viscosity was measured using a Thermo Electron HAAKE RheoStress 600 rheometer with a cone/plate geometry in combination with a temperature control unit. The conductivity was measured using the AC impedance technique with a BioLogic VMP3 multichannel potentiostat that included frequency response analyzer modules. The density was measured using a Mettler Toledo Densito 30PX density meter. The electrochemical cell has Pt disk electrodes (13 mm of diameter) facing each other with a gap of 5 mm to be filled with liquid samples. The cell constant was determined by measuring the standard KCl solution in advance. The temperature of the cell was controlled in an ESPEC SH-241 temperature chamber. Cyclic voltammograms were also acquired using a VMP3 with a three-electrode cell (WE: glassy carbon, CE: Pt wire, RE: Ag/AgCl aq.) at 25 °C. Battery Tests: The battery characteristics were determined using an Aska Electronic ACD-01 battery performance analyzer. A sheet of carbon paper (Toray, TGP-H-030) with a thickness of 110 μm and a diameter of 15 mm was used as a current collector of the cathode. 20 μL of catholyte was used for all evaluations. Li metal and 1 mLT in propylene carbonate were used as the anode and its electrolyte, respectively. Li ion-conducting glass ceramics (LIC–GC) (Ohara) with a 0.15 mm of thickness were used as the separators and were located between the cathode and anode rooms (Figure S7, Supporting Information). The voltage range for cycling is 3.0–4.2 V. Charge/discharge cycling tests were performed using following conditions; 50 μL of catholyte, LIC–GC with a thickness of 1 mm, carbon paper with a thickness of 190 μm, and a current density of 0.1 mA cm−2. The voltage range for cycling is also 3.0–4.2 V. All the battery tests were performed at 25 °C. Details of structure and evaluation for the preliminary flow battery demonstration can be found in the caption of Figure S10, Supporting Information. Note: Reference citations on pages 2502 and 2503 were corrected to read reference 8 on April 14, 2015, after initial publication As a to our and this supporting by the materials are and may be for but are not or from supporting than should be to the The is not for the content or of any supporting by the than should be to the corresponding for the
26
A metal-free organic–inorganic aqueous flow battery
Brian Huskinson, Michael P. Marshak, Changwon Suh et al. · Nature · 2014 · 1.6K citations · Full text
Recent Progress in Redox Flow Battery Research and Development
Wei Wang, Qingtao Luo, Bin Li et al. · Advanced Functional Materials · 2012 · 1.5K citations
Rechargeable batteries with organic radical cathodes
Kentaro Nakahara, Shigeyuki Iwasa, Motonobu Satoh et al. · Chemical Physics Letters · 2002 · 638 citations