Combining Accurate O2 and Li2O2 Assays to ... - ACS Publications

Aug 20, 2013 - IBM Almaden Research Center, San Jose, California 95120, United States. ‡ ... appropriately assigned, which is critically important i...
0 downloads 3 Views 495KB Size
Letter pubs.acs.org/JPCL

Combining Accurate O2 and Li2O2 Assays to Separate Discharge and Charge Stability Limitations in Nonaqueous Li−O2 Batteries Bryan D. McCloskey,*,† Alexia Valery,†,‡ Alan C. Luntz,†,§ Sanketh R. Gowda,†,# Gregory M. Wallraff,† Jeannette M. Garcia,† Takashi Mori,†,⊥ and Leslie E. Krupp† †

IBM Almaden Research Center, San Jose, California 95120, United States University of Grenoble, Phelma Grenoble INP, Minatec, CS 50257, 38016 Grenoble, Cedex 1, France § SUNCAT, SLAC National Accelerator Laboratory, Menlo Park, California 94025-7015, United States ⊥ Central Glass International Inc., 2033 Gateway Place, Suite 569, San Jose, California 95110, United States ‡

S Supporting Information *

ABSTRACT: Li−air batteries have generated enormous interest as potential high specific energy alternatives to existing energy storage devices. However, Li− air batteries suffer from poor rechargeability caused by the instability of organic electrolytes and carbon cathodes. To understand and address this poor rechargeability, it is essential to elucidate the efficiency in which O2 is converted to Li2O2 (the desired discharge product) during discharge and the efficiency in which Li2O2 is oxidized back to O2 during charge. In this Letter, we combine many quantitative techniques, including a newly developed peroxide titration, to assign and quantify decomposition pathways occurring in cells employing a variety of solvents and cathodes. We find that Li2O2-induced electrolyte solvent and salt instabilities account for nearly all efficiency losses upon discharge, whereas both cathode and electrolyte instabilities are observed upon charge at high potentials. SECTION: Energy Conversion and Storage; Energy and Charge Transport

L

and the electrolyte, which have been proposed to occur via nucleophilic attack or hydrogen abstraction,8 would still yield a 2 e−/O2 process (given that Li2O2 would be formed electrochemically via a 2 e−/O2 process and then chemically react), yet the oxidation of the product formed in this reaction would lead to a reduction in OER during charge. Furthermore, a loss of Li2O2 due to a reaction between carbon-based cathodes and Li2O2 or its reactive intermediates could also contribute to the observed reduction in rechargeability in Li− O2 cells.5−7 Therefore, certain chemical decomposition processes occurring during discharge may actually only be observable during charge when employing DEMS. Combining quantitative DEMS with ex situ spectroscopic analyses of the discharge products provides a more complete picture of battery rechargeability. To this end, discharge reaction products have been identified using a variety of techniques, including XPS, XRD, and Raman spectroscopy.5,9,12,13 However, these measurements are only qualitative in nature; that is, they do not quantify the amount of Li2O2 present during battery operation. Quantifying Li2O2 formed and oxidized during either discharge or charge should allow parasitic processes occurring during each to be decoupled and

i−air batteries have received significant attention as possible rechargeable energy storage devices for electric vehicles and other applications demanding high specific energy.1−4 However, numerous scientific challenges, such as the well-known instability of carbon-based cathodes5−7 and nonaqueous organic electrolytes,8−11 limit both the cycle life and energy efficiency in rechargeable Li−air cells. These instabilities are a consequence of parasitic reactions that occur in the presence of the active reversible cathode electrochemistry, which for nonaqueous aprotic Li−O2 batteries is the net electrochemical reaction 2(Li+ + e−) + O2 ↔ Li2O2 (U0 = 2.96 V vs Li/Li+). In previous reports, DEMS was used to compare Faradaic charge to the moles of O2 consumed/evolved during galvanostatic discharge/charge in nonaqueous Li−O 2 cells.8,12−14 Many nonaqueous electrolytes yielded a ∼2 e−/ O2 process during discharge, which is consistent with the expected stoichiometry of the cell reaction shown above. However, during charge, all cells, regardless of their electrolyte or carbon cathode composition, evolved less oxygen during charge (OER) than was consumed during discharge (ORR), such that OER/ORR < 0.9. Given these observations, most stability issues should seemingly be ascribed to the charge process, and, in fact, many electrolyte solvents are observed to be unstable in the presence of Li2O2 at potentials significantly below their expected oxidation potentials.8 However, thermal chemical reactions occurring during discharge between Li2O2 © 2013 American Chemical Society

Received: August 5, 2013 Accepted: August 20, 2013 Published: August 20, 2013 2989

dx.doi.org/10.1021/jz401659f | J. Phys. Chem. Lett. 2013, 4, 2989−2993

The Journal of Physical Chemistry Letters

Letter

operation conditions). In each case, a slightly higher than 2 e−/ O2 process is observed on discharge (Figure 1b and 1e). However, less Li2O2 is formed than O2 is consumed. Li formate, Li acetate, Li carbonate, and Li-salt decomposition have been observed to form during discharge using FT-IR, NMR, and gas evolution measurements9,15−17 (see Figures S2− S5), making parasitic side reactions the likely cause of Li2O2 loss during discharge. LiOH formation could also occur if cell components are not sufficiently dry, but it has never been observed as a discharge product in our laboratory.8,18 We now consider the general nature of these parasitic reactions (i.e., electrochemical versus thermal decomposition, reaction between the electrolyte or carbon and Li2O2/LiO2, etc.). Figure 2 presents YLi2O2 as a function of discharge rate and

appropriately assigned, which is critically important information when considering the design of a stable cathode or electrolyte. In this Letter, we have developed a simple modified iodometric titration to quantify Li2O2 formation and oxidation, which we then couple to O2 consumption/evolution (using DEMS8,12) to more completely understand stability limitations of various components of a Li−O2 cell. The protocol of the iodometric titration, including the numerous control experiments performed to ensure accuracy of the assay, is outlined in the Supporting Information (SI). Other measurements, such as NMR using internal standards, were also employed to quantify decomposition product formation during discharge and charge. We find that although a ∼2 e−/O2 process may occur during discharge in cells employing various electrolytes and cathodes, the yield of Li2O2 (YLi2O2, the amount of Li2O2 produced divided by the amount of Li2 O 2 expected given the Coulometry; See table S1, SI) is a function of the cathode and electrolyte composition. In the best combination (see Figure 1a−c and the TOC Abstract graphic), YLi2O2 = 91%, with

Figure 2. (a) Yield of Li2O2, YLi2O2, as a function of the Li−O2 battery discharge rate. (b) O2 consumption during the first 10 h of cell discharge at various discharge rates.

the corresponding O2 consumption at lower rates for a P50/ DME/LiTFSI-based cell. The Li2O2 yield decreases as the rate decreases, yet the O2 Coulombic efficiency remains constant within experimental error (2.03 ± 0.02 e−/O2) at all rates. These data imply that freshly made Li2O2 remains exposed to the electrolyte for a longer time at lower discharge rates, allowing a slow chemical reaction to occur between the Li2O2 and electrolyte. A small amount of parasitic electrochemistry may also occur given the slightly higher than 2 e−/O2 process, but this would account for only a small fraction (∼10%) of the total decomposition observed. In support of this proposal, we note that Gallant et al. observed that at relatively high rates of discharge, many small (