Letter pubs.acs.org/JPCL
Dominant Decomposition Pathways for Ethereal Solvents in Li−O2 Batteries Jeannette M. García,* Hans W. Horn,* and Julia E. Rice IBM Almaden Research Center, 650 Harry Road, San Jose, California 95120, United States S Supporting Information *
ABSTRACT: The promise of high specific energies for Li−O2 batteries has driven research toward the development of new compatible materials for this emerging technology. Obtained energies, however, fall short of the theoretical values partly due to parasitic chemistries arising from organic solvent decomposition during battery cycling. Electrolyte solvent and salt decomposition have also been identified as limiting factors for rechargeability of the battery. Although lithium trifluorosulfonamide (LiTFSI) dissolved in 1,2-dimethoxyethane (DME) has been shown to be a promising solvent/ electrolyte candidate for Li−O2 batteries, significant challenges remain, namely minimizing decomposition of both the solvent and electrolyte salt during battery cycling. Herein, we provide spectroscopic labeling studies to identify the source of H2 at high potentials during charge and propose a decomposition pathway for DME to lithium formate and acetate products at low potentials. NMR studies were preformed to show that DME decomposes to lithium formate and acetate in aqueous Li2O2, products which are also observed after D2O workups on cathodes after discharge. Finally, we use density functional theory (DFT) to elucidate a mechanistic pathway for DME decomposition that is based on known organic oxidation processes. in DME on discharge were ∼2.01 e−/O2 (theoretical e−/O2 = 2.00). [Despite its appearance, the discrepancy of +0.01 is still rather large. An ideal solvent would hit the target value of e−/ O2 = 2.00. However, compared to other solvents such as DMSO, where discrepancies of +0.05 are observed, this discrepancy is low, indicating more dominant parasitic electrochemistries during discharge in nonethereal solvents. See ref 9.] Spectroscopic methods to identify discharge products in the battery have included XRD, SEM, AFM, and in-situ and ex-situ SERS10,11 to name a few. Isotopic labeling techniques have been used to determine that oxygen evolved during the charging process is dominated by Li2O2 decomposition, and not by other Li-containing solid species such as Li2O or LiOH. In order to probe this, Li−O2 batteries were cycled7 under a 50:50 mixture of isotopically labeled oxygen (18O2) and evaluated using differential electrochemical mass spectrometry (DEMS). If radical processes were operative that involved O− O bond cleavage during oxygen evolution on charge, the oxygen evolved would contain a mixture of the two isotopes, which was not observed. Additionally, a small amount of heavy water and CO2 were introduced into the system, but the oxygen evolved never contained more than one labeled atom.6 Alternatively, when 13C carbon was used as a conductive cathode material, the CO2 evolved was found to contain ∼50/
T
he rechargeable lithium−oxygen (Li−O2) battery has gained significant interest due to its high theoretical specific energy densities in manufactured batteries. However, the theoretical energy density calculated for Li−O2 batteries has not yet been realized, in part due to parasitic chemistries that depart from the functional Li + O2 ↔ Li2O2 chemistry.1 There are many factors that contribute to the deviation from the operative electrochemical reaction, including dendrite formation on the lithium metal anode, separator and binder stability during cycling, cathode stability, as well as solvent and electrolyte salt stability.2 Parasitic side reactions when an ethereal solvent with LiTFSI is used as an electrolyte in Li−O2 batteries result the formation of lithium formate (HCO2Li), lithium acetate (LiOAc), lithium fluoride (LiF) at low potentials (< ∼3.0 V) and Li2CO3, and gaseous CO2 and H2 at high potentials (∼4.0 V). Thus, requisite operating conditions for Li−O2 batteries would require the use of materials inert to reduction from the lithium metal anode as well as stability to the oxidative discharge product and intermediates, namely, lithium peroxide (Li2O2) and lithium superoxide (LiO2).3,4 Previously, it has been shown that different organic solvents exhibit different variability from the ideal 2-electron process as well as a diminishment in oxygen evolution reaction to oxygen reduction reaction ratio (OER/ORR). A stable organic solvent evaluated (based on OER/ORR values and electrons per oxygen molecule studies5−7) was found to be a 1 M solution of a lithium salt, specifically lithium trifluorosulfonamide (LiTFSI), dissolved in 1,2-dimethoxyethane (DME) solvent.8 Typical electron consumed per oxygen molecule values for 1 M LiTFSI © XXXX American Chemical Society
Received: March 12, 2015 Accepted: April 29, 2015
1795
DOI: 10.1021/acs.jpclett.5b00529 J. Phys. Chem. Lett. 2015, 6, 1795−1799
Letter
The Journal of Physical Chemistry Letters
Figure 1. Cycling and DEMS analysis for 1 M LiTFSI in THF. Conditions: AvCarb P50 Carbon Cathode, Celgard separator, 200 μA discharge and charge to 1 mAh.
Figure 2. Cycling and DEMS analysis for 1 M LiTFSI in d8-THF. Conditions: AvCarb P50 Carbon Cathode, Celgard separator, 200 μA discharge and charge to 1 mAh.
50 mixture of 13CO2 and 12CO2 at high potentials (only 12CO2 at low potentials from electrolyte), indicating that the cathode was at least partially the source of CO2 observed at high potentials.12 Grey et al. determined though labeling experiments that enriched peroxide O atoms were transferred to lithium formate by solid state 17O NMR and also observed products that appeared to originate from reduction of the O− CH3 groups on DME.14 Because isotopic labeling coupled with DEMS has proved to be a useful method for the evaluation of chemistries and electrochemistry operative during cell cycling, we decided to employ a similar method through the use of deuterium labeling of ethereal solvents to evaluate the source of H2 often observed as a byproduct of charging at high potentials (> ∼ 4.0 V). Because very little H2 is ever observed in the case of DME,
tetrahydrofuran (THF) was chosen as a model system for this study (Figure 1). [H2 is typically evolved in batteries at voltages greater than ∼4.0 V on charge. In systems in which this potential is not reached (including LiNO3 based systems13) no H2 is evolved.] Interestingly, when deuterated THF was used, the amount of deuterium gas evolved was approximately halved compared to the H2 production in the nondeuterated solvent, perhaps indicative of a primary kinetic isotope effect during the hydrogen evolution reaction (Figure 2). In addition, only D2 and no H2 was observed. This implies that in the Li−O2 cell, the source of the hydrogen gas that is evolved at high potentials is the organic solvent and not other proton-containing sources such as the separator or cathode. Intrigued by these results, we decided to explore some of the decomposition pathways of 1796
DOI: 10.1021/acs.jpclett.5b00529 J. Phys. Chem. Lett. 2015, 6, 1795−1799
Letter
The Journal of Physical Chemistry Letters Scheme 1. Proposed Mechanism for DME Oxidation to Lithium Formate and Lithium Acetate in Aqueous Solution
DME, a solvent that typically does not evolve H2 even at high potentials but has nonetheless been shown to decompose during battery operation. We and others9,14−16 have reported previously that when an ethereal solvent (such as DME) is utilized, lithium acetate and lithium formate are formed as primary side products during discharge in the battery (see Supporting Information Figure S1). Perplexed by the apparent stability of DME to solid Li2O2 over long periods of time, we studied the reactivity of DME with solvated Li2O2 in deuterated water (D2O) by 1H NMR (see Supporting Information Figures S2 and S3). Notably, in the absence of the electrolyte salt (typically LiTFSI) and electrodes, lithium formate and acetate were successfully produced under these conditions. Additionally, dimethyl sulfone was generated from dimethyl sulfoxide (DMSO) at a higher rate with aqueous Li2O2 than with hydrogen peroxide alone (see Supporting Information Figure S6). This suggests that the presence of lithium ions together with peroxide nucleophiles enhances the rate of reaction with the solvent, not unlike mechanisms proposed for organolitiation reactions in the presence of diamines,17,18 or the directed deprotonation of DME with n-BuLi.19 Contrary to computational studies that require accounting for liquid-electrolyte interfaces with solid Li2O2,20,21 this enabled the use of in silico methods to identify the most probable reaction pathway in a homogeneous environment. In order to rationalize the small amounts of formate and acetate formed during discharge in DME, and also observed after D2O extraction on cathodes, we turned to molecular modeling. We explored several redox mechanisms that would transform DME via successive one-electron steps such as lithium peroxide nucleophilic attack on DME (Scheme 1A) and homolytic bond cleavage and radical-based mechanisms (Scheme 1B). The nucleophilic attack mechanism produced intermediates that resulted in unproductive pathways. The radical-based mechanism such as that previously reported for DME oxidation was ruled out partially since the e−/O2
determined through DEMS analysis of Li−O2 batteries with DME/LiTFSI electrolytes is always >2.00 and also because production of high-energy radical species in protic environments is disfavored.22 Thus, we premised through experiments that a parasitic one-electron or radical-based mechanistic pathway was unlikely under these conditions. Furthermore, no one-electron pathway was found computationally to have rate-limiting barriers that would be accessible at room temperature or would lead to the observed products. We surmised that perhaps a two-electron redox mechanism (such as the Dakin oxidation,23 see Supporting Information Scheme S1) might be operative. Here, we model a possible pathway leading to formate and acetate computationally in implicit water as solvent using the protocol outline in the Supporting Information (Section C). Scheme 1C shows our proposed mechanism for DME oxidation in aqueous solution. Computational modeling of the first step (E2 elimination) yields a rate-limiting barrier of 36 kcal/mol.24 The corresponding energy profile is shown in the Supporting Information (Figure S7). Computational modeling of the second step (water-catalyzed hydrolysis of enolate) yields a rate-limiting barrier of 33 kcal/mol. The corresponding energy profile is shown in the Supporting Information (Figure S9). Computational modeling of the third step (Dakin oxidation23 of acetaldehyde) yields rate-limiting barriers of 31 kcal/mol (CH3 shift, leading to formate) and 24 kcal (H shift, leading to acetate). The corresponding energy profiles are shown in the Supporting Information (Figure S10). Note that, if modeled in implicit DME, assuming there is a partially dissociated lithium peroxide present (Li+ + −OOLi), the corresponding barriers for the first step (E2 elimination) are significantly lower (barrier ∼7 kcal/mol). This suggests that in an aprotic environment such as in the Li−O2 battery, E2 elimination is likely, although subsequent oxidative mechanisms may be different from that shown in Scheme 1C. Isolation and characterization of the enolate after battery discharge is difficult due to its high reactivity; however, distinct products can be 1797
DOI: 10.1021/acs.jpclett.5b00529 J. Phys. Chem. Lett. 2015, 6, 1795−1799
Letter
The Journal of Physical Chemistry Letters observed by 1H NMR depending on extraction solvent used (Scheme S4, Supporting Information). In aqueous conditions, the barrier for elimination is higher; however, the elimination is spontaneous due to higher concentration of the dissociated ions. In contrast, in anhydrous conditions, due to the low solubility of Li2O2, the ionic concentrations will be considerably lower, thus affecting the rate of the reaction. Because the cathodes from the battery are typically extracted with D2O after discharging, a possible explanation for this is that the observed lithium acetate and formate are formed during workup or postbattery (ex-situ) analysis. The rate-limiting barriers for steps 1, 2, and 3 are summarized in Table 1.
involving a Dakin-type oxidation mechanism. These experiments and calculations suggest that the effect of cathode workup after battery operation may not be insignificant to observe volatile or reactive decomposition products as well as formate and acetate. The development of electrolyte solvents with enhanced stability to decomposition pathways such as that outlined here is currently underway in our laboratories.
■
S Supporting Information *
Materials and methods, alternative pathways explored, NMR spectra, Cartesian coordinates, and reaction coordinate diagrams. The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/ acs.jpclett.5b00529.
Table 1. Summary of Rate-Limiting Barriers for Steps 1, 2, and 3
■
rate-limiting barrier ΔG [kcal/mol] M11/ACCPVTZ Step 1. E2 elimination of DME Step 2. Hydrolysis of enolate Step 3. Dakin oxidation of CH3CHO
36 33 31 (CH3 shift), forms formate 24 (H shift), forms acetate
ASSOCIATED CONTENT
AUTHOR INFORMATION
Corresponding Authors
B3LYP-D3/ ACCPVTZ
*E-mail:
[email protected]. Tel.: 1-408-927-1602. *E-mail:
[email protected]. Tel.: 1-408-927-2519. Notes
The authors declare no competing financial interest.
21 (CH3 shift) 16 (H shift)
■
ACKNOWLEDGMENTS The authors are grateful to Bryan McCloskey and Winfried Wilcke for helpful and informative discussions and Naga Phani Aetukuri for valuable experimental assistance.
Overall, the rate-limiting barriers of all three steps are large (Table 1), which is consistent with the small amount of formate and acetate observed experimentally. Of concern are the barriers for step 3 (Table 1), which are not consistent with the experimental observation that more formate is formed than acetate. This may be a computational artifact of the DFT functional (M1125,26) used in this study. A recent publication by Radom et al.27 indicates that Mxx25,26 functionals underestimate proton transfer barriers. To this end we computed key stationary points of step 3 (reactants, RC0, INT1, TS2) using the B3LYP28-D329 functional. Even though the B3LYP-D3 barriers are significantly lower than the M11 barriers (see Table 1, row 3), their trend is very similar. From the partial charges of ≈ +0.1e at the transferred CH3 and H units (M11 and B3LYPD3) in Supporting Information Figure S11, we see that in both cases a virtually neutral species is being transferred, which would indicate that Radom’s findings are not applicable for the Dakin oxidation step. Recently, a paper by Curtiss et al.30 visited the subject of DME decomposition in Li−O2 batteries. Their computational study on (Li2O2)n clusters with DME in implicit solvent concluded that for n ≥ 4 high spin states are more stable and that DME decomposition preferentially be caused by H abstraction at −CH2− of DME in the presence of O2 by a radical pathway with the predicted products being aldehydes and carboxylates as well as LiOH. In conclusion, we have determined through DEMS on battery cells in which isotopically labeled THF is utilized that the origin of hydrogen gas evolved at high potentials during Li−O2 charging is the organic solvent. There may be a primary kinetic isotope effect associated with this process, because the amount of D2 evolved in the case of d8-THF is approximately half that of H2 in the nondeuterated electrolyte. In addition, decomposition products of DME through the addition of Li2O2 were modeled in aqueous medium using DFT calculations. DFT calculations support a pathway for DME decomposition
■
REFERENCES
(1) Luntz, A. C.; McCloskey, B. D. Nonaqueous Li−Air Batteries: A Status Report. Chem. Rev. 2014, 114, 11721−11750. (2) Lu, J.; Amine, K. Recent Research Progress on Non-Aqueous Lithium-Air Batteries from Argonne National Laboratory. Energies 2013, 6, 6016−6044. (3) Adams, B. D.; Black, R.; Williams, Z.; Fernandes, R.; Cuisinier, M.; Berg, E. J.; Novak, P.; Murphy, G. K.; Nazar, L. F. Towards a Stable Organic Electrolyte for the Lithium Oxygen Battery. Adv. Energy Mater. 2015, 5, n/a−n/a. (4) Bryantsev, V. S.; Giordani, V.; Walker, W.; Blanco, M.; Zecevic, S.; Sasaki, K.; Uddin, J.; Addison, D.; Chase, G. V. Predicting Solvent Stability in Aprotic Electrolyte Li−Air Batteries: Nucleophilic Substitution by the Superoxide Anion Radical (O2•−). J. Phys. Chem. A 2011, 115, 12399−12409. (5) McCloskey, B. D.; Valery, A.; Luntz, A. C.; Gowda, S. R.; Wallraff, G. M.; Garcia, J. M.; Mori, T.; Krupp, L. E. Combining Accurate O2 and Li2O2 Assays to Separate Discharge and Charge Stability Limitations in Nonaqueous Li−O2 Batteries. J. Phys. Chem. Lett. 2013, 4, 2989−2993. (6) Gowda, S. R.; Brunet, A.; Wallraff, G. M.; McCloskey, B. D. Implications of CO2 Contamination in Rechargeable Nonaqueous Li− O2 Batteries. J. Phys. Chem. Lett. 2013, 4, 276−279. (7) McCloskey, B. D.; Scheffler, R.; Speidel, A.; Girishkumar, G.; Luntz, A. C. On the Mechanism of Nonaqueous Li−O 2 Electrochemistry on C and Its Kinetic Overpotentials: Some Implications for Li−Air Batteries. J. Phys. Chem. C 2012, 116, 23897−23905. (8) Xu, W.; Xiao, J.; Zhang, J.; Wang, D.; Zhang, J.-G. Optimization of Nonaqueous Electrolytes for Primary Lithium/Air Batteries Operated in Ambient Environment. J. Electrochem. Soc. 2009, 156, A773−A779. (9) McCloskey, B. D.; Bethune, D. S.; Shelby, R. M.; Mori, T.; Scheffler, R.; Speidel, A.; Sherwood, M.; Luntz, A. C. Limitations in Rechargeability of Li-O2 Batteries and Possible Origins. J. Phys. Chem. Lett. 2012, 3, 3043−3047. (10) McCloskey, B. D.; Bethune, D. S.; Shelby, R. M.; Girishkumar, G.; Luntz, A. C. Solvents’ Critical Role in Nonaqueous Lithium− 1798
DOI: 10.1021/acs.jpclett.5b00529 J. Phys. Chem. Lett. 2015, 6, 1795−1799
Letter
The Journal of Physical Chemistry Letters Oxygen Battery Electrochemistry. J. Phys. Chem. Lett. 2011, 2, 1161− 1166. (11) Peng, Z.; Freunberger, S. A.; Hardwick, L. J.; Chen, Y.; Giordani, V.; Bardé, F.; Novák, P.; Graham, D.; Tarascon, J.-M.; Bruce, P. G. Oxygen Reactions in a Non-Aqueous Li+ Electrolyte. Angew. Chem., Int. Ed. 2011, 50, 6351−6355. (12) McCloskey, B. D.; Speidel, A.; Scheffler, R.; Miller, D. C.; Viswanathan, V.; Hummelshøj, J. S.; Nørskov, J. K.; Luntz, A. C. Twin Problems of Interfacial Carbonate Formation in Nonaqueous Li−O2 Batteries. J. Phys. Chem. Lett. 2012, 3, 997−1001. (13) Walker, W.; Giordani, V.; Uddin, J.; Bryantsev, V. S.; Chase, G. V.; Addison, D. A Rechargeable Li−O2 Battery Using a Lithium Nitrate/ N,N -Dimethylacetamide Electrolyte. J. Am. Chem. Soc. 2013, 135, 2076−2079. (14) Leskes, M.; Moore, A. J.; Goward, G. R.; Grey, C. P. Monitoring the Electrochemical Processes in the Lithium−Air Battery by Solid State NMR Spectroscopy. J. Phys. Chem. C Nanomater. Interfaces 2013, 117, 26929−26939. (15) Ottakam Thotiyl, M. M.; Freunberger, S. A.; Peng, Z.; Chen, Y.; Liu, Z.; Bruce, P. G. A Stable Cathode for the Aprotic Li−O2 Battery. Nat. Mater. 2013, 12, 1050−1056. (16) McCloskey, B. D.; Garcia, J. M.; Luntz, A. C. Chemical and Electrochemical Differences in Nonaqueous Li−O2 and Na−O2 Batteries. J. Phys. Chem. Lett. 2014, 5, 1230−1235. (17) Chadwick, S. T.; Rennels, R. A.; Rutherford, J. L.; Collum, D. B. Are n-BuLi/TMEDA-Mediated Arene Ortholithiations Directed? Substituent-Dependent Rates, Substituent-Independent Mechanisms. J. Am. Chem. Soc. 2000, 122, 8640−8647. (18) Chadwick, S. T.; Ramirez, A.; Gupta, L.; Collum, D. B. NButyllithium/N,N,N′,N′-Tetramethylethylenediamine-Mediated Ortholithiations of Aryl Oxazolines: Substrate-Dependent Mechanisms. J. Am. Chem. Soc. 2007, 129, 2259−2268. (19) Vedejs, E.; Fang, H. W. An E-Selective 1,3-Diene Synthesis from Moderated Ylides and Aldehydes. J. Org. Chem. 1984, 49, 210−212. (20) Khetan, A.; Pitsch, H.; Viswanathan, V. Solvent Degradation in Nonaqueous Li-O2 Batteries: Oxidative Stability versus H-Abstraction. J. Phys. Chem. Lett. 2014, 5, 2419−2424. (21) Khetan, A.; Luntz, A.; Viswanathan, V. Trade-Offs in Capacity and Rechargeability in Nonaqueous Li−O2 Batteries: Solution-Driven Growth versus Nucleophilic Stability. J. Phys. Chem. Lett. 2015, 6, 1254−1259. (22) Litwinienko, G.; Beckwith, A. L. J.; Ingold, K. U. The Frequently Overlooked Importance of Solvent in Free Radical Syntheses. Chem. Soc. Rev. 2011, 40, 2157−2163. (23) Dakin, H. The Oxidation of Hydroxy Derivatives of Benzaldehyde, Acetophenone and Related Substances. Chem. J. 1909, 42, 498. (24) We note that one should consider only trends in rate-limiting barrier heights from DFT calculations, rather than absolute barriers. (25) Peverati, R.; Truhlar, D. G. Improving the Accuracy of Hybrid Meta-GGA Density Functionals by Range Separation. J. Phys. Chem. Lett. 2011, 2, 2810−2817. (26) Peverati, R.; Truhlar, D. G. Performance of the M11 and M11-L Density Functionals for Calculations of Electronic Excitation Energies by Adiabatic Time-Dependent Density Functional Theory. Phys. Chem. Chem. Phys. 2012, 14, 11363−11370. (27) Chan, B.; Gilbert, A. T. B.; Gill, P. M. W.; Radom, L. Performance of Density Functional Theory Procedures for the Calculation of Proton-Exchange Barriers: Unusual Behavior of M06Type Functionals. J. Chem. Theory Comput. 2014, 10, 3777−3783. (28) Becke, A. D. Density-functional Thermochemistry. III. The Role of Exact Exchange. J. Chem. Phys. 1993, 98, 5648−5652. (29) Grimme, S.; Antony, J.; Ehrlich, S.; Krieg, H. A Consistent and Accurate Ab Initio Parametrization of Density Functional Dispersion Correction (DFT-D) for the 94 Elements H-Pu. J. Chem. Phys. 2010, 132, 154104−154119. (30) Assary, R. S.; Lau, K. C.; Amine, K.; Sun, Y.-K.; Curtiss, L. A. Interactions of Dimethoxy Ethane with Li2O2 Clusters and Likely
Decomposition Mechanisms for Li−O2 Batteries. J. Phys. Chem. C 2013, 117, 8041−8049.
1799
DOI: 10.1021/acs.jpclett.5b00529 J. Phys. Chem. Lett. 2015, 6, 1795−1799
