Communication pubs.acs.org/JACS
Unleashing the Power and Energy of LiFePO4‑Based Redox Flow Lithium Battery with a Bifunctional Redox Mediator Yun Guang Zhu,† Yonghua Du,‡ Chuankun Jia,† Mingyue Zhou,† Li Fan,† Xingzhu Wang,† and Qing Wang*,† †
Department of Materials Science and Engineering, Faculty of Engineering, National University of Singapore, Singapore 117576 Institute of Chemical and Engineering Sciences, 1 Pesek Road, Jurong Island, Singapore 627833
‡
S Supporting Information *
reported, greatly simplifying the catholyte composition.17 Upon charging, LiFePO4 is delithiated with the oxidation of polyiodide; upon discharging, FePO4 is lithiated with the reduction of iodide. Although interesting, polyiodides are corrosive and the voltage efficiency of the cell is compromised by the large potential difference of the two-step redox reactions. Therefore, it is desired to explore alternative redox mediators to surmount these issues. Here we demonstrated a Li-LiFePO4 RFLB cell by using a bifunctional redox mediator, 2,3,5,6-tetramethyl-p-phenylenediamine (TMPD), with which much enhanced voltage efficiency and cycling performance have been achieved. In addition, operando X-ray absorption near-edge structure (XANES) measurements were conducted to monitor the reactions between LiFePO4 and TMPD, from which an insightful understanding of the redox-targeting reactions has been obtained. As a thermodynamic basis of the redox targeting reaction, the potentials of the bifunctional redox molecule should either match or straddle that of the solid material. The cyclic voltammogram (CV) of TMPD presents two pairs of peaks at 3.20 and 3.60 V vs Li/Li+ (Figure 1a) corresponding to the
ABSTRACT: Redox flow batteries, despite great operation flexibility and scalability for large-scale energy storage, suffer from low energy density and relatively high cost as compared to the state-of-the-art Li-ion batteries. Here we report a redox flow lithium battery, which operates via the redox targeting reactions of LiFePO4 with a bifunctional redox mediator, 2,3,5,6tetramethyl-p-phenylenediamine, and presents superb energy density as the Li-ion battery and system flexibility as the redox flow battery. The battery has achieved a tank energy density as high as 1023 Wh/L, power density of 61 mW/cm2, and voltage efficiency of 91%. Operando X-ray absorption near-edge structure measurements were conducted to monitor the evolution of LiFePO4, which provides insightful information on the redox targeting process, critical to the device operation and optimization. edox flow batteries (RFBs), distinct from the battery technologies with enclosed configuration, have decoupled power generation and energy storage, and thus superior scalability, operation flexibility, and safety for large-scale application.1,2 Despite these advantages, the deployment of RFBs is currently hindered by the high capital cost which is largely associated with the low energy density of the battery system. The energy density of RFBs is constrained by the concentration of soluble redox species in the electrolytes. For instance, the concentration of vanadium redox species in vanadium redox-flow battery (VRB) is around 2 M, which leads to an energy density of 25−30 Wh/L.3 The narrow voltage window is another limitation for aqueous RFBs. Therefore, efforts have been devoted to the development of nonaqueous RFBs with larger operating voltage and meanwhile higher redox concentration.4−9 These new systems generally employ Li+ as the charge balancing ion using soluble or semisolid redox species in the catholyte and anolyte.7,10−12 Besides the above approaches, a redox flow lithium battery (RFLB) based on the redox-targeting concept has recently been reported that stores energy in Li-ion battery materials statically kept in energy tanks while operating as a RFB.13 With suitable pairs of redox mediators, LiFePO4 and TiO2 have been demonstrated as the cathodic and anodic active materials in RFLB, respectively, and delivered an energy density 10 times as high as VRB.14−16 More recently, a Li-LiFePO4 RFLB cell based on iodide, a single 2-electron redox mediator has been
R
© 2017 American Chemical Society
Figure 1. (a) CV measurement of 10 mM TMPD•+ (red curve) in 0.5 M LiTFSI/DEGDME electrolyte at a scan rate of 50 mV/s. The working electrode was a Pt disc. The black curve is that of LiFePO4 measured in a Swagelok cell. The inset shows the molecular structure of TMPD. (b) CV measurements of in 0.5 M LiTFSI/DEGDME with or without 2.5 mM TMPD•+ at a scan rate of 10 mV/s. The working electrode was FTO glass coated with Al2O3 or Al2O3-LiFePO4/FePO4 layer. The counter and reference electrodes for both CV measurements were Pt wire and Li metal, respectively. The inset shows the double-layer structure of working electrode and the associated redox targeting reactions during the CV measurement. Received: February 2, 2017 Published: April 24, 2017 6286
DOI: 10.1021/jacs.7b01146 J. Am. Chem. Soc. 2017, 139, 6286−6289
Communication
Journal of the American Chemical Society reversible reactions of TMPD/TMPD•+ and TMPD•+/ TMPD•2+, respectively, which straddle that of LiFePO4 (3.45 V vs Li/Li+). To examine further the redox targeting reactions of TMPD toward LiFePO4, a specially designed electrode was employed for CV measurement (see inset of Figure 1b).18 Two pairs of reversible peaks arising from the reactions of TMPD on the exposed FTO underneath Al2O3 were observed, consistent with Figure 1a. The presence of LiFePO4/FePO4 on top of Al2O3 leads to a dramatic enhancement of the reduction wave of TMPD•+ to TMPD and the oxidation wave of TMPD•+ to TMPD•2+, whereas the oxidation of TMPD and the reduction of TMPD•2+ are notably eliminated, as shown in Figure 1b. Compared with the reduction process, the oxidation current is significantly lower, presumably due to the relatively small driving force, and/or bimolecular reactions between TMPD-based reaction intermediates in solution which lowers the effective concentration of TMPD•2+.19,20 The electrochemical reversibility and cycling stability of TMPD were also evaluated by CV measurements (Figure S1). The CV curves kept nearly unchanged for 500 consecutive cycles, revealing the robustness of TMPD for the condition tested. The above have validated TMPD as a bifunctional redox mediator for Li-LiFePO4 RFLB cells. The reactions in the cell and tank are summarized below: Charging process:
Figure 2. (a) Schematic structure of a Li-FePO4 RFLB cell. It consists of a stack cell, energy tank loaded with LiFePO4 granules, and circulation system driven by a peristaltic pump. (b) Voltage profiles of RFLB cells with 25 mM TMPD and LiI in the catholyte while with/ without LiFePO4 in the tank. The current density was 0.125 mA/cm2. (c) Voltage profiles of a RFLB cell at fixed LiFePO4 utilizations and different current densities. The catholyte was 20 mM TMPD in 0.5 M LiTFSI/DEGDME with 100 mM LiFePO4 loaded in the tank. The membrane was Nafion-PVDF composite membrane. (d) Polarization curves of RFLB cells at 100% SOC. The catholyte was 0.5 M TMPD in 1 M LiTFSI/DEGDME with/without equivalent concentration of FePO4 in the tank.
TMPD•+ − e− → TMPD•2 + (on electrode) TMPD•2 + + LiFePO4 → TMPD•+ + Li+ + FePO4 (in tank)
excellent cycling performance and Coulombic efficiency, much superior to the LiI/LiFePO4 cell (Figure S4). This is further corroborated by the good stability of both the electrolyte and LiFePO4 (Figures S5 and S6) after testing for 45 cycles. The power performance of the RFLB cell was examined at 100% state-of-charge (SOC) by steady-state polarization measurement (Figure 2d). The electrolyte was 0.5 M TMPD•2+ in 1 M LiTFSI/DEGDME loaded with equimolar FePO4. To assess the influence of redox targeting reactions, the resistive Nafion-PVDF membrane was removed from the cell. Instead, the lithium foil anode was treated with 5 vol % vinylene carbonate to protect it from the parasitic reactions with the redox mediators. This is acceptable considering the polarization measurement was typically not lengthy. As shown in Figure 2d, the J−V curve shows two voltage plateaus arising from the two-step reductions of TMPD•2+ to TMPD•+ and TMPD, which thus lead to a bimodal power vs current density curve. The maximum power density of the FePO4-loaded RFLB reached 61 mW/cm2, significantly higher than that without FePO4 (36 mW/cm2). Apparently, the redox targeting reaction of TMPD (formed during the discharge process) with FePO4, which oxidizes TMPD and instantaneously replenishes the consumption of TMPD•+, accounts for the enhancement of power density. This is an encouraging result for the development of redox targeting-based flow batteries, with which both the energy density and power density are boosted. In comparison, the power performance of LiI/LiFePO4 system is considerably lower and only marginal improvement was observed in the presence of FePO4 (Figure S7). Several redox systems have been reported for RFLB thus far. However, the redox targeting reactions between the redox mediators and solid active materials have yet been scrutinized
Discharging process: TMPD•+ + e− → TMPD (on electrode)
TMPD + Li+ + FePO4 → TMPD•+ + LiFePO4 (in tank)
To assess the performance of TMPD in RFLB, a RFLB halfcell illustrated in Figure 2a was fabricated, in which the lithium anode and carbon felt cathode are separated by a NafionPVDF composite membrane. The catholyte consisting of 25 mM TMPD in 0.5 M LiTFSI/DEGDME is circulated by a peristaltic pump between the cathodic compartment and energy tank filled with granules of LiFePO4. Galvanostatic charge and discharge measurements were conducted (Figure 2b) at a current density of 0.125 mA/cm2, from which two voltage plateaus were observed, broadly consistent with the two oxidation potentials of TMPD. TMPD in the catholyte has a discharge capacity of ∼2.5 mAh. After 158 mg of LiFePO4 (∼13 times equivalent concentration to that of TMPD) was added in the energy tank, the discharge time was extended by ∼7 times, corresponding to 73% utilization of LiFePO4 (in terms of the capacity 160 mAh/g). The voltage efficiency of the cell is ∼91%, considerably higher than the LiI system,17 which is only 61% at the same current density (Figure 2b), and hitherto among the highest reported. It is noted the Coulombic efficiency is ∼76%, which presumably stems from the degradation of TMPD•2+ at high charging voltages (Figure S2). In contrast, when the charging capacity is controlled with fixed LiFePO4 utilization ratios so that TMPD will not be entirely oxidized at the end of charging, the cell revealed much improved Coulombic efficiency and reasonably good cycling stability at different current densities (Figure 2c and S3). At 40% utilization ratio of LiFePO4, the cell demonstrated 6287
DOI: 10.1021/jacs.7b01146 J. Am. Chem. Soc. 2017, 139, 6286−6289
Communication
Journal of the American Chemical Society
and 7153 eV indicate that the two-phase reaction occurs at the second charging plateau.24 After 2 h charging, no further change was observed, implying LiFePO4 has been oxidized to FePO4 by TMPD•2+. The near unity redox targeting reaction is substantiated by the comparison of XAENS spectra with a reference FePO4 sample (Figure S9). The subsequent charging until the end is attributed to the reaction of TMPD•+ to TMPD•2+. The discharging process broadly mirrors that of the charging. The first short discharging plateau corresponds to the reaction of TMPD•2+ to TMPD•+. When the voltage reached the second plateau, the absorption edge of Fe quickly shifted to lower energy in less than 0.5 h (Figure 3e), suggesting FePO4 was promptly reduced by TMPD to form LiFePO4. Thereafter, the phase change became relatively slow, presumably a result of the incommensurate Li+ diffusion in the bulk FePO4. Based on the fitting result, ∼90% FePO4 was converted back to LiFePO4, revealing good reversibility of the redox targeting reactions. The above operando XAENS measurements have several important implications to the operation and optimization of RFLB: (1) The redox targeting reaction occurs as soon as TMPD•2+ is generated during the charging process. (2) The rate of phase transformation of active material in the tank varies at different stages of discharge, of which the fast component (
