Article pubs.acs.org/JPCC
Using Rotating Ring Disc Electrode Voltammetry to Quantify the Superoxide Radical Stability of Aprotic Li−Air Battery Electrolytes Juan Herranz,*,† Arnd Garsuch,‡ and Hubert A. Gasteiger† †
Institute of Technical Electrochemistry, Technische Universität München, D-85748 Garching, Germany BASF SE, GC, Ludwigshafen am Rheim 67056, Germany
‡
ABSTRACT: Despite the promising high specific energy density of lithium−air batteries, their commercialization remains hindered by numerous issues, including the poor stability of the electrolyte due to its reaction with the superoxide radical (O2•−) produced upon discharge at the battery’s cathode. In this work, we have used rotating ring disc electrode (RRDE) voltammetry to study this reaction and to quantify the stability of the electrolyte against O2•− by its pseudo-first-order reaction constant, k. Our results confirm the recently reported reactivity of propylene carbonate (PC, which was used in many of the initial works on Li−air batteries), while unveiling the enhanced stability of 1-butyl-1-methylpyrrolidinium bis(trifluoromethylsulfonyl)imide (Pyr14TFSI), with a k value at least 3 orders of magnitude lower than that estimated in PC. Moreover, our RRDE-transient measurements indicate that the diffusion of O2•− in this ionic liquid is ≈70 times slower than that in PC, which could partially explain the poor discharge capacities observed in Li− air battery tests using Pyr14TFSI. with the superoxide radical (O2•−) produced in the first step of the cathodic reduction of O2 upon battery discharge,11,12 yielding Li-carbonate, propylene dicarbonate, acetate, and formate as well as CO2 and H2O9 instead of the desired Li2O2 and Li2O products. Moreover, even if the presence of this superoxide radical in Li+-containing electrolytes remains hypothetical, the reactivity of tetrabutylammonium-containing electrolytes (where O2 reduction reportedly results in O2•−) in cyclic voltammetry (CV) measurements shows strong analogies with the nature of the discharge products in the actual Li−O2 battery. Indeed, while in CV tests, dimethoxyethane (DME)13,14 and propylene carbonate (PC)13 are, respectively, stable and unstable against O2•−, upon discharge of Li−O2 cells, the same electrolytes yield primarily Li2O210 or Li-carbonate9 (i.e., using Li+-containing DME of PC). These results suggest that the reaction of superoxide with the electrolyte does play a crucial role on the nature of the discharge products, which, when derived from the chemical decomposition of the electrolyte, will likely have a negative effect on the recharge potential, cyclability, and round-trip efficiency of the Li−air battery. In addition to the potential reaction between O2•− and the electrolyte, the possible degradation of other battery components, such as the Li anode15 and certain binders used in typically carbon-based cathodes,16 can negatively affect Li− air battery cyclability. Furthermore, the presence of atmospheric contaminants, such as H2O17 and CO2,18 in the cathode feed has also been shown to affect the nature of the discharge
1. INTRODUCTION In recent years, concerns about our excessive energy dependence on oil resources and the negative impact of their combustion products on the environment have boosted the interest in novel energy production, conversion, and storage systems. For transportation applications, which account for ≈50% of the overall oil demand,1 electricity storage in lithiumion batteries appears as the alternative of choice for a first fleet of hybrid or even fully electric vehicles. Unfortunately, the specific energy of fully packaged battery systems based on lithium-ion batteries (currently at ≈120 Wh/kg with a potential to reach ≈200 Wh/kg)2 is insufficient for the desired driving range of at least 300 miles, typical of today’s automotive systems based on internal combustion engines.2 Alternatively, a roughly 3-fold greater specific energy (fully packaged battery system) could be attained by replacing the lithium-intercalation compounds at the cathode of the Li-ion battery with an air (O2) electrode,3 where, upon discharge in aprotic electrolytes, Li+ ions would react with oxygen to yield Li2O2 and Li2O.4−6 These discharge products should then get electrooxidized back into Li+ and O2 upon battery recharge.7 Although doubtlessly rewarding if finally accomplished, the development of rechargeable Li−air batteries remains hindered by multiple issues, including their low round-trip efficiency and rate capability, and a poor cycle life. The latter appears to be related to the electrolyte instability in lithium−air cathodes (positive electrodes), which also renders the understanding and eventual solving of the remaining problems more difficult. As an example of these difficulties, a number of recent publications8−10 have dealt with the inadequacy of the carbonate-based electrolytes used in many of the initial works on Li−air batteries. These carbonates are now believed to react © 2012 American Chemical Society
Received: May 3, 2012 Revised: August 3, 2012 Published: August 7, 2012 19084
dx.doi.org/10.1021/jp304277z | J. Phys. Chem. C 2012, 116, 19084−19094
The Journal of Physical Chemistry C
Article
The dynamic viscosity was measured at a shear rate of γ = 100, using a rotational rheometer (Physica MCR 101, Anton Paar, Graz, Austria). A 0.75 mL aliquot of electrolyte solution was applied on a sample plate onto which a rotational plate was lowered, thus creating a thin liquid film between both plates. The measurements were carried out at 25 °C. The densities of the electrolyte solutions were determined using a precision scale and sealed volumetric flasks. The electrochemical measurements were performed in a four-neck, jacketed glass cell assembled and sealed inside the Ar-filled glovebox. The working electrode consisted of a PTFEembedded glassy carbon disc of 5.0 mm in diameter surrounded by a gold ring with an internal diameter of 6.5 mm and an external diameter of 7.5 mm (Pine Research Instrumentation, Durham, NC). Prior to its use, the ring disc electrode was polished with a 0.05 μm alumina suspension (Buehler, Düsseldorf, Germany), cleaned by sonication in ultrapure water, screwed onto a PEEK shaft that was fed through a stopper equipped with a ceramic ball-bearing seal (Pine Research Instrumentation), and then dried for 1 h in an oven at 70 °C. The counter electrode was a platinum wire (99.99+%, Advent, Oxford, England) sealed with a glass fitting and immersed in a glass tube partially filled with electrolyte and terminated by a medium-porosity (nr. 3) frit that prevented the diffusion of the species evolved at the Pt counter electrode into the main electrolyte compartment. The reference electrode consisted of a glass tube filled with 0.1 M AgNO3 (99.9999% metals basis, Sigma Aldrich) in acetonitrile (99.8% anhydrous, Sigma Aldrich; also predried over zeolites and containing ≤4 ppm H2O) and sealed with a Vycor 7930 frit (Advanced Glass & Ceramics, Holden, MA) at its tip, which was immersed in the AgNO3/CH3CN solution; the silver wire was embedded in a plastic cap, which sealed the reference compartment against the ambient. The reference electrode was assembled in the glovebox at least 90 min before putting together the rest of the electrochemical cell; once ready, it was partially immersed into a beaker containing the electrolyte of interest, along with a piece of Li foil (99.9%, Chemetall, Frankfurt, Germany) connected to a Ni wire (99.98%, Advent). The potential difference between both electrodes was subsequently measured for a minimum of 90 min, and all potentials in this work are corrected using this potential difference (3.48 and 3.30 V in PC and Pyr14TFSI, respectively) and are referred to as “volts vs Li+/Li” or VLi. The RRDE working electrode, the Pt wire counter electrode, the reference electrode, and a glass bubbler allowing for direct flow of gas into the solution or blanketing atop the electrolyte were all assembled inside the glovebox. Once assembled, the electrochemical cell was taken outside of the glovebox, the working electrode rod was mounted onto the rotator, and the bubbler was connected to a gas line constantly fed with argon or oxygen (6.0 quality, Westfalen-AG, Münster, Germany). This allowed for a constant overpressure inside the cell and prevented contamination from the atmosphere. Moreover, the cell’s jacket was also connected to a precalibrated thermostat (Julabo, Seelbach, Germany) set to 25 °C. All subsequent electrochemical measurements were performed using an AFCBP1 bipotentiostat (Pine Research Instrumentation) controlled with Aftermath software. Prior to the RRDE measurements, ac impedance measurements to determine the Ohmic drop between working and reference electrodes were recorded with a VMP3 multichannel potentiostat (BioLogic, Grenoble, France), applying a 10 mV voltage perturbation (1
products, yielding larger capacities, but compromising the rechargability.18 Considering this multitude of possible contributions to Li−air battery degradation (and the possible existence of others that still remain unknown), it is of crucial importance to develop methodologies that allow for an independent evaluation and quantification of the stability of the electrolyte toward superoxide radical without interference from other degradation processes, thereby enabling a quantitative comparison of the stability of different electrolytes. On the basis of this motivation, we have used rotating ring disc electrode (RRDE) voltammetry to quantify the reactivity of O2•− in two selected solvents: propylene carbonate (PC) and the ionic liquid (IL) 1-butyl-1-methylpyrrolidinium bis(trifluoromethylsulfonyl)imide (Pyr14TFSI), the chemical structures of which are shown in Figure 1. PC was chosen as
Figure 1. Chemical structure of the compounds used in this work: (A) propylene carbonate (PC), (B) tetrabutylammonium cation (TBA+), (C) bis(trifluoromethylsulfonyl)imide anion (TFSI−), and (D) 1butyl-1-methylpyrrolidinium cation (Pyr14+).
a benchmark solvent due to its now well-known instability8−10,13 and its widespread use in many of the early Li−air battery works. On the other hand, Pyr14TFSI was selected due to the fact that ILs have already attracted some attention as possible electrolytes for Li−air batteries,19−23 whereby particularly Pyr14TFSI is suggested to be stable toward superoxide radical24,25 and has indeed been used as a solvent in several recent studies.22,23 In addition to quantifying the reactivity of superoxide radical with these two solvents, the here-presented RRDE experiments and analysis also allowed us to quantify the diffusion coefficients of superoxide radical and oxygen as well as oxygen solubility, all of which are critical parameters for fundamental models of the mass transport limitations in Li−air batteries.
2. EXPERIMENTAL SECTION All chemicals were dried prior to their use and stored in an Arfilled glovebox (Jacomex, Dagneux, France) containing
