Skip to main content
NCBI home page
ACS AuthorChoice logoLink to ACS AuthorChoice
. 2023 May 22;127(21):10001–10011. doi: 10.1021/acs.jpcc.3c00991

NMR Methodology for Measuring Dissolved O2 and Transport in Lithium–Air Batteries

Evelyna Wang 1, Erlendur Jónsson 1, Clare P Grey 1,*
PMCID: PMC10240530  PMID: 37284295

Abstract

graphic file with name jp3c00991_0007.jpg

Similar to fuel cells, poor mass transport of redox active species, such as dissolved oxygen gas, is one of the challenges faced by lithium–air batteries (LABs). Capitalizing on the paramagnetic properties of O2, we used nuclear magnetic resonance (NMR) spectroscopy to measure oxygen concentration and transport in LAB electrolytes. Lithium bis(trifluoromethane) sulfonylimide (LiTFSI) in glymes or dimethyl sulfoxide (DMSO) solvents were investigated with 1H, 13C, 7Li, and 19F NMR spectroscopy, with the results showing that both the 1H, 13C, 7Li, and 19F bulk magnetic susceptibility shifts and the change in 19F relaxation times were accurate measures of dissolved O2 concentration. O2 saturation concentrations and diffusion coefficients were extracted that are comparable to values measured by electrochemical or pressure methods reported in the literature, highlighting the validity of this new methodology. This method also provides experimental evidence of the local O2 solvation environment, with results again comparable to previous literature and supported by our molecular dynamics simulations. A preliminary in situ application of our NMR methodology is demonstrated by measuring O2 evolution during LAB charging using LiTFSI in the glyme electrolyte. While the in situ LAB cell showed poor coulombic efficiency, since no additives were used, the O2 evolution was successfully quantified. Our work demonstrates the first usage of this NMR methodology to quantify O2 in LAB electrolytes, experimentally demonstrate solvation environments of O2, and detect O2 evolution in situ in a LAB flow cell.

Introduction

Lithium–air batteries (LABs) promise extremely high energy densities.1,2 During discharge, lithium metal is oxidized at the anode to form Li+ ions and electrons, the latter traveling through an external circuit to reach the cathode and the former migrating through the electrolyte to the cathode. At the cathode, typically a porous carbon material, O2 is reduced and subsequently reacts with Li+ ions to form solid-phase Li2O2 discharge products, which deposit onto the surface of the cathode.3 The O2 from the environment must diffuse from the gas phase to the 3-phase boundary (solid electrode, liquid electrolyte, gaseous atmosphere) or dissolve into the electrolyte and again react at the electrode surface.4,5 Early work by Read et al. correlated the discharge capacity (and rate capability) in LABs with the O2 solubility in the electrolyte.6−8 Several studies have since endeavored to measure oxygen solubility in electrolytes; typical methods used include pressure and gas uptake experiments or electrochemical methods such as a rotating ring-disk electrode (RRDE).6,9−12 In pressure measurements, the gas overhead pressure above an electrolyte reservoir is monitored as a function of time. In general, a decrease in O2 solubility and diffusivity is observed for electrolytes with increasing viscosity, for example, by increasing the molecular weight of the glyme solvent used. Molecular dynamics (MD) simulations have also been employed to calculate O2 solubility and diffusivity, largely agreeing with experimental measurement trends.10,11

In this work, we employ a nuclear magnetic resonance (NMR) method to measure oxygen in LAB electrolytes. Paramagnetic species, such as radicals or (triplet) O2, in solution, can induce shifts in the measured NMR resonances as well as increase nuclear spin relaxation rates.13,14 The induced shifts in the resonance frequencies are caused by either hyperfine or bulk magnetic susceptibility (BMS) effects. The hyperfine interactions, sometimes termed a local effect, arise due to an interaction between the nucleus of interest and the unpaired electron spins on the paramagnetic species. This interaction can be of two types: contact or pseudo-contact. The first is caused by electron spin delocalization onto, or spin polarization of, the s-orbitals of the nucleus under observation, while the latter involves a through-space dipolar coupling between the electron and nuclear spins. In contrast, the BMS effect is a bulk response that results from the bulk susceptibility of the paramagnetic medium—be it a liquid, gas, or an array of particles. The presence of the paramagnetic species adds to the applied magnetic field felt by the nucleus and affects all nuclei at the same location by the same amount.

Our aim in this work was to explore the effects of paramagnetic O2 in LAB electrolytes using NMR spectroscopy, to establish whether either BMS or hyperfine effects can be used to quantify O2 concentrations and to explore O2 solvation. We start by briefly outlining the relevant theory for BMS shifts. We then compare the hyperfine shifts measured for various nuclei within our electrolytes, namely, various concentrations of lithium bis(trifluoromethane) sulfonylimide (LiTFSI) in glymes or dimethylsulfoxide (DMSO). We measure 19F, 7Li, and 1H solution NMR spectra and, using this theory, obtain information on O2 solvation within the electrolyte. We find that O2 appears to associate more closely with the F atoms of the LiTFSI salt. O2 dissolution from the gas phase into a liquid electrolyte was also measured using 7Li NMR spectroscopy, and the O2 diffusion constants were extracted. Finally, NMR relaxometry measurements were used to monitor O2 evolution during charge in a Li–air flow battery. To the best of our knowledge, these NMR methodologies have not been applied to measure paramagnetic O2 in LAB electrolytes before. The paper also describes the first experimental results on dissolved O2 solvation in LAB electrolytes. Paired with our flow-NMR setup, these methods enable direct observation of dissolved O2 during LAB operation.

Methods

Materials

LiTFSI was obtained from Sigma-Aldrich. Prior to use, LiTFSI salt was dried under a vacuum at 120 °C for 12 h. Diethylene glycol dimethyl ether (diglyme), triethylene glycol dimethyl ether (triglyme), tetraethylene glycol dimethyl ether (tetraglyme), and DMSO were purchased from Sigma-Aldrich. Diglyme, triglyme, and tetraglyme were refluxed under Ar with sodium metal, then distilled, and finally stored in an Ar-filled glovebox over molecular sieves. DMSO was used as is.

Solution NMR

Electrolyte mixtures were prepared in the glovebox. For NMR measurements of the electrolytes under Ar, 0.3 mL of electrolyte was loaded into a 5 mm medium-walled glass solution NMR tube in the glovebox and sealed with a tap. For NMR measurements of oxygen-saturated electrolytes, O2 gas was bubbled into the electrolyte for 3 min with a syringe needle. A sealed capillary with deuterated benzene (C6D6) was added as a reference.

For orientation-dependent measurements, 7 mm sections of polyether ether ketone (PEEK) tubing, 1/16″ outer diameter, were used to hold electrolyte, ∼75 μL, and sealed with epoxy. The PEEK tubing with the electrolyte sample was then placed into a 10 mm solution NMR tube either horizontally or vertically and loaded into the probe.

NMR spectra were measured using a Bruker 300 MHz Advance III Spectrometer, with a magnetic resonance imaging (MRI) probe, using a single 90° pulse excitation for the different nuclei. Pulse lengths were 18.5 μs for 19F, 15 μs for 7Li, and 21 μs for 1H. Longitudinal T1 relaxation measurements were performed using a saturation recovery experiment, and transverse T2 relaxation measurements were performed using a Carr–Purcell–Meiboom–Gill (CPMG) pulse sequence.

Flow Cell

A Li–air flow cell was set up following specifications from Milshtein et al.;15 however, instead of flowing both anolyte and catholyte, only one electrolyte was flown, and a polytetrafluoroethylene (PTFE) seal was placed on the Li metal anode side. A stainless-steel current collector was used with a Li metal anode, and a graphite current collector with etched interdigitated flow fields was used with the carbon electrode. Carbon electrodes, Sigracet 39BC, comprising carbon fiber paper coated with microporous carbon, were purchased from Fuel Cell Store and cut into 1 cm2 squares. Borosilicate glass fiber (GF) separators from Whatman were used. Lithium metal, 99.5%, was purchased from LTS Research and stored in an Ar-filled glovebox. The 0.25 M LiTFSI in the diglyme electrolyte was used. The total electrolyte volume was 20 mL.

Perfluoroalkoxy (PFA) tubing, 1/16 in. outer diameter, was used for most connections, and MasterFlex tubing 77202-60 #14 Chem Bio was used with the peristaltic pump. The electrolyte flow rate was ∼0.06 mL s–1. The LAB cell was initially assembled in the glovebox, where all components were sealed, before being placed onto a lab bench with the pump and Biologic SP-150 potentiostat. The electrolyte reservoir was then bubbled with O2 gas for 15 min. The cell was subjected to a 2.5 mA h discharge at a rate of 0.05 mA cm–2 to form Li2O2 at the carbon cathode. Subsequently, the electrolyte was purged with N2 to remove the O2 in the electrolyte. The PFA tubing was then connected to the NMR spectrometer to enable in situ measurement. The online flow-through design is described in previous reports.16 The LAB NMR measurements were performed using a Bruker 300 MHz Advance III Spectrometer, with an MRI probe and a custom flow-through NMR sampling tube.16

BMS Theory

The BMS effect and hyperfine interactions experienced by a nucleus are additives, and the total shift caused by a paramagnetic species can be expressed as13

graphic file with name jp3c00991_m001.jpg 1

where δlocal is the shift caused by local hyperfine interactions between the unpaired electrons of the paramagnetic species and the nuclei of interest.13 δlocal does not distinguish between the types of hyperfine interactions: Fermi contact or pseudo-contact. The δbulk is the shift caused by the BMS given by17

graphic file with name jp3c00991_m002.jpg 2

where χv is the bulk volumetric magnetic susceptibility and α is the shape factor. This shape factor accounts for the effect of sample shape and orientation relative to the applied magnetic field (produced by the NMR magnet) on the magnetic field observed by the nuclei of interest. Non-isotropic macroscopic effects such as induced dipoles at the boundaries of the sample contribute to the shape factor.

The shape factor at a point within the sample, Inline graphic can be calculated from the surface integral which gives the total magnetic flux through a point on the surface

graphic file with name jp3c00991_m004.jpg 3

where x′ is the location of the surface element, ẑ is the unit vector normal to the surface, and β is the angle between the applied magnetic field and the normal to the surface. For simple geometries, the shape factors are constant throughout the sample and have been explicitly solved: the shape factors for a sphere or infinite cylinder oriented either perpendicular or parallel to the applied magnetic field are summarized below.17

graphic file with name jp3c00991_m005.jpg 4

If we approximate our samples as infinite cylinders, we can see from eq 2 that the shift caused by a BMS effect for samples in a perpendicular orientation is δbulk⊥ = −2/3πχv and that for samples in a parallel orientation, δbulk = 4/3πχv. We can then write out eq 1 for a sample in either perpendicular or parallel orientations

graphic file with name jp3c00991_m006.jpg 5
graphic file with name jp3c00991_m007.jpg 6

where Δobs⊥ and Δobs are the experimentally observed shifts caused by adding paramagnetic species into the solution; that is, Δobs⊥ = δparamag – δnoparamag⊥ and likewise, Δobs = δparamag∥ – δnoparamag. Solving the system of equations allows the calculation of χv and δloc from experimentally measured values

graphic file with name jp3c00991_m008.jpg 7
graphic file with name jp3c00991_m009.jpg 8

Thus, orientation-dependent measurements can be used to deconvolute the BMS and hyperfine effects caused by paramagnetic dissolved oxygen in LAB electrolytes. This method was described by Delpuech et al. to compare O2 solubility in fluorinated benzene with benzene18 and built upon earlier paramagnetic NMR studies first described by Evans and later by Becconsall et al., as well as Garroway.14,19,20

Once the volumetric magnetic susceptibility χv is determined, it is used to calculate the concentration of paramagnetic species16

graphic file with name jp3c00991_m010.jpg 9

where Cpar is the concentration of paramagnetic species, kB is the Boltzmann’s constant, T is the temperature, NA is Avogadro’s number, μB is the Bohr magneton, and S is the total spin quantum number (S = 1/2 for each unpaired electron). For O2 with two unpaired electrons, S = 1, and solving for the constants, we end with

graphic file with name jp3c00991_m011.jpg 10

Equations 7 and 10 allow us to directly calculate the concentration of dissolved O2 in mM from our orientation-dependent NMR measurements.

Results

Oxygen Solubility—NMR Shifts and Relaxometry

Figure 1 shows the 19F, 7Li, 1H, and 13C NMR spectra of a 0.25 M LiTFSI in diglyme electrolyte after it is prepared in the glovebox (blue trace) as well as after saturating the electrolyte with O2 (red trace). The 19F and 7Li resonances result from the TFSI– salt anion and Li+ cation, respectively, while the 1H and 13C spectra result from the diglyme solvent. The measured NMR shifts caused by the addition of O2 into the LiTFSI-diglyme electrolyte were 0.156 ppm for 19F, 0.12 ppm for 7Li, 0.133 ppm for 1H, and 0.125 ppm for 13C. The shifts caused by O2 are not uniform across the different nuclei, suggesting that O2 does not interact with the nuclei present in the electrolyte solely through a bulk effect: there are also local interactions between the paramagnetic electrons on the O2 molecules and the nearby salt or solvent molecules, eq 1. The O2-induced shifts observed in either the 13C or 1H spectra are the same for the −CH3 and −CH2 peaks, within the resolution of this method.

Figure 1.

Figure 1

Open in a new tab

(a) 19F, (b) 7Li, (c) 1H, and (d) 13C solution NMR spectra of a 0.25 M LiTFSI in diglyme electrolyte before (blue) and after (red) saturation with O2, acquired in a standard, sealed NMR tube.

In order to untangle the shift induced by a bulk effect from any local effects caused by the paramagnetic O2 molecules, solution NMR measurements were made with the samples oriented at two different angles to the external magnetic field (Figure 2). The electrolyte samples were first oriented perpendicular to the applied magnetic field as shown in the schematic in Figure 2a: a 7 mm PEEK tube was filled with the electrolyte (shown in dark red) and then positioned horizontally at the bottom of a 10 mm glass solution NMR tube. Figure 2b shows the NMR measurements for the electrolyte samples oriented parallel to the magnetic field, with the 7 mm PEEK tube positioned vertically into the glass tube. This positioning was possible due to static cling to the glass tube (see Supporting Information for an image of the sample). The electrolyte samples were measured with and without O2 saturation, and the 19F, 7Li, and 1H spectra for the 0.25 M LiTFSI in diglyme electrolyte are shown in Figure 2, where only the peaks corresponding to the −CH3 protons are shown in the 1H spectra. The spectra were referenced by setting the shift positions measured under Ar (without O2) to the values observed inside a standard NMR tube (i.e., Figure 1), which accounts for any diamagnetic contributions to the BMS. The measurements were then repeated for different electrolytes: 0.25 M LiTFSI in triglyme, tetraglyme, and DMSO. A monoglyme (also known as DME—dimethoxyethane)-based electrolyte was not measured due to its volatility since bubbling in O2 gas vaporizes the solvent and changes the LiTFSI concentration.

Figure 2.

Figure 2

Open in a new tab

(a) 19F, 7Li, and 1H NMR spectra of the 0.25 M LiTFSI in diglyme electrolytes where the sample is oriented perpendicular to the applied magnetic field, before (blue) and after (orange) O2 saturation. The 1H NMR peak shown is the −CH3 peak of the diglyme solvent. (b) 19F, 7Li, and 1H NMR spectra of the same electrolyte where the sample is oriented parallel to the applied magnetic field.

In order to calculate the volumetric susceptibility and the hyperfine shift, the observed change in shift caused by adding O2 was first calculated for the 19F, 7Li, and 1H (CH3−) NMR spectra by subtracting the NMR resonance of a chosen peak for the non-O2 electrolyte (Ar) from the NMR resonance of the O2-saturated electrolyte, defined for the perpendicular electrolyte sample as Inline graphic. Similar calculations were performed for the parallel orientation to extract Δobs∥. Once Δobs and Δobs∥ were calculated, eqs 7 and 8 could be used to determine the shift caused by the BMS effect (and from that χv) and the hyperfine shifts (δloc) for 19F, 7Li, and 1H (Table 1). It is encouraging to note that the calculated volumetric magnetic susceptibilities were similar in value for the different nuclei, as expected for a bulk effect.

Table 1. Calculated Volumetric Susceptibility (χv, Unitless) and Hyperfine Shift (ppm) from Orientation Measurements of Electrolytes before and after O2 Saturationa.

  19F
 7Li
 1H
solvent (0.25 M LiTFSI) χv δloc χv δloc χv δloc
diglyme 0.031 ± 0.001 –0.35 ± 0.01 0.030 ± 0.001 –0.01 ± 0.01 0.028 ± 0.005 –0.05 ± 0.02
triglyme 0.018 ± 0.003 –0.18 ± 0.01 0.017 ± 0.001 –0.01 ± 0.01 0.016 ± 0.005 –0.03 ± 0.01
tetraglyme 0.011 ± 0.005 –0.15 ± 0.01 0.011 ± 0.003 –0.02 ± 0.01 0.008 ± 0.007 +0.01 ± 0.02
DMSO 0.006 ± 0.001 –0.23 ± 0.01 0.007 ± 0.005 –0.11 ± 0.01 0.007 ± 0.006 –0.05 ± 0.01
Open in a new tab
a

Errors included were calculated from the variance in the Δobs⊥ and Δobs values from repeat measurements (each measurement was carried out twice).

Using the calculated χv, we applied eq 10 to calculate the concentration of O2 for the different electrolytes and compared them to literature values (Table 2). For the electrolytes measured in this work, the calculated O2 concentration values are comparable to those previously reported and measured via different methods: Gittleson et al. used electrochemical methods,9 while Hartmann et al. and Schürmann et al. used pressure uptake measurements.11,21 For the diglyme-based electrolyte, increasing salt concentration also led to a decrease in O2 saturation. These trends are again consistent with those reported in the literature.9 The relatively large errors of the method reflect the difficulty in ensuring that the solution is saturated with O2 and errors in the orientations of the samples. We note that it would be straightforward to improve the method to reduce the errors by designing bespoke NMR cells.

Table 2. O2 Saturation Concentrations in mM from NMR Measurements (Calculated from χυ from Measurements of the Three Nuclei) and the Literaturea.

    this work: concentration of O2 (mM)
      
solvent LiTFSI concentration (M) 19F 7Li 1H from ref (9) from ref (21) from ref (11)
diglyme 0.125 9.3 ± 0.4 8.9 ± 0.8 8.5 ± 0.4   6.4* 7.1*
  0.25 9.0 ± 0.4 8.7 ± 0.4 8.2 ± 0.4      
  0.5 8.7 ± 0.4 8.8 ± 0.9 6.2 ± 0.8      
  1 8.6 ± 1.4 8.3 ± 0.9 5.6 ± 1.4      
triglyme 0.25 5.3 ± 0.9 5.0 ± 0.4 4.7 ± 1.4   5.6* 5.6*
tetraglyme 0.25 3.3 ± 1.4 3.3 ± 0.8 2.3 ± 2.1 2.1** 4.3* 4.3*
DMSO 0.25 1.7 ± 0.4 1.9 ± 1.4 2.1 ± 2.0 0.99**    
Open in a new tab
a

The experimental errors were again calculated from the variance in the Δobs⊥ and Δobs values from repeat measurements (each measurement was carried out twice). * indicates literature values for the solvent only (no salt added). ** indicates a value for a 0.5 M LiTFSI salt concentration in the electrolyte.

The longitudinal (R1) and transverse (R2) relaxation values were then measured as a function of O2 concentration for the 0.25 M LiTFSI in diglyme electrolyte, as plotted in Figure 3a,b, respectively. 19F was used for relaxation measurements due to its increased sensitivity to O2 concentration compared to other nuclei (see Figure S3). The observed 7Li shift was then directly used to calculate O2 concentrations using eqs 6 and 10 since the hyperfine shift for this nucleus is essentially zero (see Table 1). Thus, relaxation rates can be compared with O2 concentrations. A linear behavior with respect to oxygen concentration was observed, allowing a calibration curve for estimating O2 concentration in the electrolyte from measured relaxation rates to be obtained. This method of using sensitive 19F relaxometry to quantify O2 has previously been used in biological systems.22

Figure 3.

Figure 3

Open in a new tab

(a) 19F longitudinal and (b) transverse relaxation rates plotted as a function of O2 concentration in the 0.25 M LiTFSI in diglyme electrolyte, where O2 concentration was measured via 7Li BMS.

Oxygen Diffusivity

We then designed experiments to measure oxygen diffusivity via NMR: rather than saturating the electrolyte with O2, the gas overhead space in the NMR tube was filled with O2, Figure 4a inset. In this way, O2 from the gas phase needs to dissolve into the electrolyte, and the subsequent change in NMR chemical shift is measured as a function of time. Figure 4a shows the change in 7Li chemical shift caused by O2 diffusion into the electrolyte for the different electrolytes. Relatively small changes in the chemical shifts are observed in the 7Li spectra, on the order of only 0.01–0.02 ppm per time step. These small changes, in conjunction with possible errors such as minor temperature fluctuations, gave rise to the noise observed in Figure 4. Nonetheless, an increasing 7Li shift as a function of time caused by the dissolution of paramagnetic O2 is clearly observed. Furthermore, an increase in O2 dissolution/diffusion rate with decreasing molecular weight glymes is consistent with literature reports and increasing viscosity effects. Plotting the 19F shift in LiTFSI peak position as a function of time for the same experimental setup results in a larger shift resolution (Figure S2); however, the 19F peak shifts include hyperfine contributions that are difficult to deconvolute in the experimental setup.

Figure 4.

Figure 4

Open in a new tab

(a) Change in 7Li NMR LiTFSI peak position as a function of time for various electrolytes. The gas overhead space in a solution NMR tube was filled with O2 gas. For each sample, 0.3 mL of electrolyte was used, (b) exponential fitting of the O2 concentration versus time for the diglyme electrolyte as measured from the change in 7Li chemical shift.

The diffusion data was fit using an exponential growth function. Fitting oxygen dissolution data is well studied in the bio-reactor literature23

graphic file with name jp3c00991_m013.jpg 11

where eq 11 can be integrated to give

graphic file with name jp3c00991_m014.jpg 12

C(t) is the concentration of O2 as a function of time, Csat is the saturated oxygen concentration, KL is related to the oxygen transfer coefficient (from the gas phase to dissolving into the solvent), which is directly proportional to diffusivity, and A is the surface area to volume ratio of the liquid. In two film theories, it is assumed that molecular diffusion is the only mechanism of O2 transfer, and therefore, KL is equal to the diffusion coefficient divided by the distance over which the O2 needs to dissolve.23 Knowing the geometric parameters (volume to surface area ratio and diffusion distance), KL and diffusivity can be extracted from the fitting (see Supporting Information for fitting other electrolytes). Table 3 lists the calculated O2 diffusivity from the 7Li NMR diffusion experiments and compares them with literature values.

Table 3. Calculated O2 Diffusion Coefficients from Fitting Chemical Shift Dataa.

solvent LiTFSI concentration (M) this work: diffusivity of O2(cm s–1) from ref (9) (cm s–1) from ref (21) (cm s–1) from ref (11) (cm s–1)
diglyme 0.25 7.8 × 10–5± 2.9 × 10–5   4.4 × 10–5* 4.6 × 10–5*
triglyme 0.25 4.8 × 10–5± 2.6 × 10–5   3.2 × 10–5* 3.5 × 10–5*
tetraglyme 0.25 4.6 × 10–5± 2.0 × 10–5 3.97 × 10–7** 2.6 × 10–5* 2.6 × 10–5*
DMSO 0.25 4.0 × 10–5± 1.2 × 10–5 1.40 × 10–5**    
Open in a new tab
a

Errors listed in the table for the diffusivity values calculated in this work indicate the 95% confidence interval calculated for the parameter fitting. * indicates a measurement for the solvent only (no salt added). ** indicates a measurement for a 0.5 M LiTFSI electrolyte.

LAB Cells

The effects of O2 evolution on NMR parameters in a working LAB device were then explored. A flow-cell design setup with in situ NMR measurements was used where the electrolyte flows from the cell into the spectrometer and then back into the reservoir during cell operation to monitor dissolved O2 concentrations (Figure 5a). The cell was charged under galvanostatic conditions to oxidize the Li2O2 formed during discharge, and O2 was detected via NMR relaxometry measurements. Relaxometry methods were used rather than measuring changes in the chemical shifts due to the small amounts of O2 evolved during charge; changes in the chemical shifts of the 7Li spectra would be very small and difficult—although not impossible—to measure accurately.

Figure 5.

Figure 5

Open in a new tab

(a) Schematic of the Li–air flow battery during charge with electrolyte flowing through the NMR spectrometer. (b) Voltage versus time profile for the Li–air flow battery under galvanostatic charge. Every 4 h, the current and pump are paused, as highlighted in the gray regions. The purple circles indicate the points when relaxometry measurements were taken. The final 4 h charge and OCV were not included in this plot. (c) Longitudinal, R1, and (d) transverse, R2, relaxation rates measured during charge.

Figure 5b shows the voltage profile during charge for the Li–O2 battery. The galvanostatic charge was paused every 4 h during relaxation measurements, seen as the drops in voltage down to open-circuit potential values. During the pause (relaxation measurements took an hour in total), the pump was also paused to halt the electrolyte flow. The large overpotentials observed during charge and drops during pauses are typical of a LAB without redox mediators. Figure 5c,d shows the measured R1 and R2 rates, respectively, collected every 4 h during charge. These rates increase with increasing state of charge, as expected due to increased paramagnetic O2 concentration in the electrolyte. Using the calibration curves from Figure 4, the O2 concentrations can be calculated from the relaxation rates; although the value of R1 measured before charging is larger than observed in the ex situ electrolyte measurements, the change in O2 concentration measured via R1 and R2 is similar (Table 4).

Table 4. Summary of the LAB Charge (from Figure 5) and O2 Evolution as Calculated from Capacity or Relaxometry Measurements.

time (h) charge passed (mA h) R1 rate measured (s–1) R2 rate measured (s–1) [O2] from charge (mM) [O2] from R1 rate (mM) [O2] from R2 rate (mM)
0 0 0.31 0.37 0 0.27 0
5.5 0.22 0.32 0.39 0.16 0.45 0.20
11 0.44 0.34 0.41 0.33 0.55 0.30
16.5 0.66 0.35 0.42 0.49 0.70 0.50
22 0.88 0.37 0.44 0.66 1.0 0.65
27.5 1.1 0.37 0.44 0.82 1.05 0.70
33 1.3 0.38 0.45 0.97 1.1 0.80
38.5 1.5 0.385 0.46 1.1 1.15 0.90
44 1.8 0.39 0.47 1.3 1.2 1.0
Open in a new tab

These O2 concentrations are compared with the theoretical O2 evolution calculated from the charge passed (assuming 100% coulombic efficiency) in Table 4, with the concentrations calculated via relaxometry measurements being lower than the expected theoretical O2, calculated assuming 100% coulombic efficiency. Discrepancies in the measured O2 and the expected concentration of O2 evolved can be accounted for by parasitic reactions that consume electrons rather than oxidizing O2.39,40Figure S5 similarly shows poor coulombic efficiency of O2 evolution as measured via pressure, whereby not all of the O2 consumed on discharge was recovered on charge. Figure S6 shows the decreased capacity for the subsequent discharge and charge cycle of the LAB cell shown in Figure 5; the decreased capacity in Figure S6 is again indicative of discharge products (desired or parasitic) that were not removed upon charge in the first cycle and can therefore help explain the lack of O2 evolution.

Discussion

The oxygen diffusivities and solubilities in the electrolytes used, as determined via NMR in this work, were comparable to values previously recorded in the literature. This suggests that measuring oxygen concentrations using the chemical shift induced by the BMS effect is an effective method for estimating concentrations. Across the measurements, the NMR peaks broadened when O2 was introduced because of increased relaxation rates.13 The relaxometry measurements plotted in Figure 3a,b also demonstrate an increase in relaxation rates with increasing O2 concentration. This is expected as the relaxation rate is proportional to the spatial distance between paramagnetic species and the nuclei of interest.

General O2 solubility and diffusivity trends in the glyme series are reproduced with this method, where the decrease in O2 diffusivity with increasing molecular weight of the glyme solvent, going from diglyme to triglyme and tetraglyme, is likely a viscosity effect as described by the Stokes–Einstein equation.9 Possible explanations for variations in solubilities and diffusivities between our measurements and literature values may be a result of approximating sample geometries as infinite cylinders, as well as the differences in concentrations of electrolyte salts used or differences in O2 saturation (i.e., our samples or theirs may not have been fully saturated due to loss of O2 on transfer between tubes), as well as possible temperature effects.

There appears to be an overestimation in value for the diffusion measured via NMR compared to reported literature values;9,11,21 although oxygen is known to have higher solubilities and diffusivities in nonaqueous solvents compared to aqueous solvents,24,25 the oxygen diffusion value measured here for diglyme electrolyte approaches the diffusion coefficient of protons in water.26 Potential causes for the overestimation in diffusion coefficients may be a result of the large diffusion length used in the calculation for KL and any inaccuracies in fitting caused by the resolution limits of the NMR spectra when measuring small shifts in peak positions caused by dissolving oxygen.

In addition to providing matches to previously measured O2 solubilities and diffusivities, the NMR methodology also reveals information on the O2 solvation environments within the electrolytes. As observed in the chemical shifts and calculated local hyperfine shifts of different nuclei (Table 1), O2 appears to interact or situate more closely, on average, with F atoms in LiTFSI-glyme electrolytes than it does with Li+ ions or the glyme protons. There is also a larger effect of O2 concentration on relaxation rates in 19F compared to other nuclei, as observed via the larger slope calculated when plotting relaxation rates as a function of O2 concentration for 19F (see Figure S3). The F atoms may be more accessible to O2 than Li+ ions in the glyme electrolytes studied since glymes are known to solvate Li+ ions, forming chelation complexes,27,28 and making it more difficult for O2 to bind to Li+.

In contrast, in the LiTFSI-DMSO electrolyte case, significant hyperfine shifts were seen for both 19F and 7Li (−0.23 and −0.11 ppm, respectively), for a saturated concentration of O2 of approximately 2 mM (Table 1). These compare to hyperfine shifts of −0.15 and −0.02 ppm for 19F and 7Li, respectively, observed for tetraglyme, despite tetraglyme having a higher concentration of dissolved O2 (3–8 mM, Table 1). DMSO does not chelate effectively with the Li+ ions,10 which may result in increased O2–Li+ binding, resulting in larger hyperfine interactions and thus 7Li hyperfine shifts (as compared to glymes).

Paramagnetic oxygen has been used to probe, for example, cation-binding sites in molecular sieves.29−32 Previous literature using 6,7Li and 133Cs NMR have shown large positive shifts for cations that are able to bind directly to O2.29−31 These observed positive shifts are caused by Fermi-contact or through-bond interactions.30 By contrast, a small negative shift was observed for protons in Brønsted acid sites of zeolites coordination to O2.32 While further calculations are required, the negative shifts seen here for Li+ in DMSO-containing electrolytes suggest that a dipolar (pseudo-contact) mechanism may be important in the coordination of dissolved O2 to Li+ in these electrolytes, with dipolar coupling being inversely proportional to the third power of the distance (between O2 and Li+, in our case). Similarly, a pseudo-contact mechanism likely plays a role in O2 to TFSI binding.

MD simulations were performed on an O2-saturated electrolyte of 0.25 M LiTFSI in diglyme to further verify the O2 solvation environment observed via the NMR. Closer F–O2 as compared to Li+–O2 interactions are seen, where the extracted radial distribution functions show Li+ density at distances of greater than 5 Å from O2 (Figure S4). Our simulations also suggest an increase in the normalized density of F atoms around O2 compared to the H atoms around dissolved O2 (Figure S4). These agree with MD simulations performed by Haas et al., who observed more TFSI– anions surrounding dissolved O2 than solvent molecules.10

Our NMR observations also align with previous work by Hamzah et al., where they observed O2 affinity for F atoms when using 1H and 19F NMR to probe dissolved oxygen in benzene vs fluorinated benzene.33 The paramagnetic effects of O2 on 19F NMR spectra and the oxyphilic nature of F atoms have also been well studied, and the approach is often used in biological samples.22,34,35 For example, Taylor et al. used 19F relaxometry to study oxygen uptake in a suspension of bacterial cells.34 Furthermore, fluorinated solvents are a class of artificial oxygen carriers and can be used in emulsions for blood replacement due to their high O2 solubilities.25 As such, increasing fluorination of LAB electrolytes has been proposed as an approach to improve O2 transport.36−38 While some work has suggested that fluorinated compounds can be used as additives or as co-solvents, the challenges of low Li salt solubility and Li+ ion conduction in fluorinated compounds remain to be solved. The NMR methodology in this work could be applied to these systems to characterize the hyperfine interactions between different nuclei with O2, to better understand how O2 is solvated, and to correlate solvation modes with solubility limits.

We also used our NMR methodology to perform operando measurements to quantify evolved O2 during cell operation. Discrepancies in the measured O2 and the expected concentration of evolved O2 can be accounted for by parasitic reactions that consume electrons rather than generating O2:39,40 the evolution of O2 from Li2O2 oxidation does not occur at a constant rate, nor at 100% faradaic efficiency, as shown using differential electrochemical mass spectroscopy, suggesting parasitic reactions consuming electrons, for instance, to form/oxidize Li2CO3.41 In Figure S5, we calculated the ratio of the moles of electrons consumed on charge (from the current) to the moles of O2 determined via pressure measurements: 2.8 moles of e– were consumed per mole of O2 evolved, deviating from the ideal ratio of 2 moles of e– per mol of O2 (Li2O2 → 2Li+ + 2e– + O2(gas)). Calculating the ratio of moles of electrons consumed on charge to moles of O2 measured via NMR relaxometry for the LAB flow cell (Figure 5 and Table 4) gives a ratio of 2.2 moles e– per mole of evolved O2 (see Supporting Information for calculation). Thus, the flow cell demonstrates an improved performance compared to the static LAB, which we ascribe to the improved mass transport of dissolved O2.4

Further explanations for the difference between measured and expected O2 concentrations may result from the chemical shift resolution limits when using the BMS effect in NMR and the errors in fitting the calibration curve. Oxygen may have also been released from the electrolyte and entered the gas overhead space in the electrolyte reservoir rather than remaining dissolved. Although more work is needed to understand the origin of some of the discrepancies, this experiment serves as a proof of concept for an alternative method to measure operando O2 evolution in a LAB electrolyte. Future work exploring MRI techniques using this method could potentially map out distributions in O2 concentrations within Li–O2 battery electrodes, which could feed into improved electrode design.

Conclusions

In conclusion, this work presents a new methodology to measure dissolved O2 in LAB electrolytes via either the BMS shift or change in relaxation (T1 or T2) time of the observed nuclei. The measured solubilities were comparable to previously reported values in the literature, while the diffusivity values we measured were slightly overestimated but in the correct order of magnitude. The shifts induced by hyperfine interactions were separated from the BMS shifts by measuring samples aligned at different orientations to the field; the hyperfine shifts revealed preferential solvation of the dissolved O2 by F atoms in the TFSI– salt anion, particularly in glyme electrolytes. To the best of our knowledge, experimental results showing O2 solvation in LAB electrolytes have not been reported prior to this work. Of the electrolytes measured, we observed the highest O2 solubility and diffusivity when using the lower molecular weight glyme as a solvent, as well as the largest shifts induced by hyperfine interactions with the F atoms, indicating that the O2 molecules spend more time, on average, near these atoms. Although DMSO-based electrolytes had lower O2 solubilities, large hyperfine interactions were observed for both 19F and 7Li, indicating that O2 interacts with both the Li+ ions and F atoms of the TFSI anions, suggesting competitive binding interactions and possibly also more TFSI– in the Li+ coordination shell. Characterizing the O2 solubilities, diffusivities, and solvation environments in electrolytes is important in designing electrolytes for increased LAB capacities and rate capabilities. Finally, relaxometry methods were also used to quantify dissolved O2 during LAB cell operation using in situ measurements, where we observed limited coulombic efficiency. Our NMR method could therefore be applied to monitor O2 evolution in LAB systems with better coulombic efficiencies, such as with redox mediator additives, and validate the improvements.

Acknowledgments

The authors thank EPSRC-EP/M009521/1 and the Cambridge Trust (E.W.) for research funding. C.P.G. acknowledges support from the Centre of Advanced Materials for Integrated Energy Systems (CAM-IES), via EPSRC grant no. EP/P007767/1. E.J. and C.P.G. acknowledge support from the ERC grant BATNMR no. 835073. The authors thank Fikile R. Brushett (MIT) for sharing the design of the flow battery that is compatible with non-aqueous solvents.

Glossary

Abbreviations

LAB

lithium–air battery

NMR

nuclear magnetic resonance

LiTFSI

lithium bis(trifluoromethane) sulfonylimide

BMS

bulk magnetic susceptibility

MD

molecular dynamics

RRDE

rotating ring-disk electrode

CPMG

Carr–Purcell–Meiboom–Gill

GF

glass fiber

PEEK

polyether ether ketone

PFA

perfluoroalkoxy alkanes

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.jpcc.3c00991.

  • Additional NMR spectra, calculations, and MD simulation details (PDF)

Author Present Address

† Chemical Sciences and Engineering Division, Argonne National Laboratory, Lemont, Illinois 60439, United States

Author Contributions

The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript.

The authors declare no competing financial interest.

Supplementary Material

jp3c00991_si_001.pdf (396.8KB, pdf)

References

  1. Aurbach D.; McCloskey B. D.; Nazar L. F.; Bruce P. G. Advances in Understanding Mechanisms Underpinning Lithium–Air Batteries. Nat. Energy 2016, 1, 16128. 10.1038/nenergy.2016.128. [DOI] [Google Scholar]
  2. Liu T.; Vivek J. P.; Zhao E. W.; Lei J.; Garcia-Araez N.; Grey C. P. Current Challenges and Routes Forward for Nonaqueous Lithium–Air Batteries. Chem. Rev. 2020, 120, 6558–6625. 10.1021/acs.chemrev.9b00545. [DOI] [PubMed] [Google Scholar]
  3. Abraham K. M.; Jiang Z. A Polymer Electrolyte-Based Rechargeable Lithium/Oxygen Battery. J. Electrochem. Soc. 1996, 143, 1–5. 10.1149/1.1836378. [DOI] [Google Scholar]
  4. Li X.; Huang J.; Faghri A. Modeling Study of a Li–O2 Battery with an Active Cathode. Energy 2015, 81, 489–500. 10.1016/j.energy.2014.12.062. [DOI] [Google Scholar]
  5. Huang J.; Faghri A. Capacity Enhancement of a Lithium Oxygen Flow Battery. Electrochim. Acta 2015, 174, 908–918. 10.1016/j.electacta.2015.06.071. [DOI] [Google Scholar]
  6. Read J. Characterization of the Lithium/Oxygen Organic Electrolyte Battery. J. Electrochem. Soc. 2002, 149, A1190. 10.1149/1.1498256. [DOI] [Google Scholar]
  7. Read J. Ether-Based Electrolytes for the Lithium/Oxygen Organic Electrolyte Battery. J. Electrochem. Soc. 2006, 153, A96. 10.1149/1.2131827. [DOI] [Google Scholar]
  8. Read J.; Mutolo K.; Ervin M.; Behl W.; Wolfenstine J.; Driedger A.; Foster D. Oxygen Transport Properties of Organic Electrolytes and Performance of Lithium/Oxygen Battery. J. Electrochem. Soc. 2003, 150, A1351. 10.1149/1.1606454. [DOI] [Google Scholar]
  9. Gittleson F. S.; Jones R. E.; Ward D. K.; Foster M. E. Oxygen Solubility and Transport in Li-Air Battery Electrolytes: Establishing Criteria and Strategies for Electrolyte Design. Energy Environ. Sci. 2017, 10, 1167–1179. 10.1039/c6ee02915a. [DOI] [Google Scholar]
  10. Haas R.; Murat M.; Weiss M.; Janek J.; Natan A.; Schröder D. Understanding the Transport of Atmospheric Gases in Liquid Electrolytes for Lithium-Air Batteries. J. Electrochem. Soc. 2021, 168, 070504. 10.1149/1945-7111/ac0d66. [DOI] [Google Scholar]
  11. Schürmann A.; Haas R.; Murat M.; Kuritz N.; Balaish M.; Ein-Eli Y.; Janek J.; Natan A.; Schröder D. Diffusivity and Solubility of Oxygen in Solvents for Metal/Oxygen Batteries: A Combined Theoretical and Experimental Study. J. Electrochem. Soc. 2018, 165, A3095–A3099. 10.1149/2.0601813jes. [DOI] [Google Scholar]
  12. Lindberg J.; Endrődi B.; Åvall G.; Johansson P.; Cornell A.; Lindbergh G. Li Salt Anion Effect on O2 Solubility in an Li–O2 Battery. J. Phys. Chem. C 2018, 122, 1913–1920. 10.1021/acs.jpcc.7b09218. [DOI] [Google Scholar]
  13. Chu K.-C.; Xu Y.; Balschi J. A.; Springer C. S. Jr Bulk Magnetic Susceptibility Shifts in Nmr Studies of Compartmentalized Samples: Use of Paramagnetic Reagents. Magn. Reson. Med. 1990, 13, 239–262. 10.1002/mrm.1910130207. [DOI] [PubMed] [Google Scholar]
  14. Evans D. F. 400. The determination of the paramagnetic susceptibility of substances in solution by nuclear magnetic resonance. Nucl. Magn. Reson. 1959, 2003–2005. 10.1039/JR9590002003. [DOI] [Google Scholar]
  15. Milshtein J. D.; Kaur A. P.; Casselman M. D.; Kowalski J. A.; Modekrutti S.; Zhang P. L.; Harsha Attanayake N.; Elliott C. F.; Parkin S. R.; Risko C.; et al. High Current Density, Long Duration Cycling of Soluble Organic Active Species for Non-Aqueous Redox Flow Batteries. Energy Environ. Sci. 2016, 9, 3531–3543. 10.1039/C6EE02027E. [DOI] [Google Scholar]
  16. Zhao E. W.; Liu T.; Jónsson E.; Lee J.; Temprano I.; Jethwa R. B.; Wang A.; Smith H.; Carretero-González J.; Song Q.; Grey C. P. In Situ NMR Metrology Reveals Reaction Mechanisms in Redox Flow Batteries. Nature 2020, 579, 224–228. 10.1038/s41586-020-2081-7. [DOI] [PubMed] [Google Scholar]
  17. Hoffman R. E. Measurement of Magnetic Susceptibility and Calculation of Shape Factor of NMR Samples. J. Magn. Reson. 2006, 178, 237–247. 10.1016/j.jmr.2005.09.009. [DOI] [PubMed] [Google Scholar]
  18. Delpuech J.-J.; Hamza M.; Serratrice G. Determination of oxygen by a nuclear magnetic resonance method. Nucl. Magn. Reson. 1979, 36, 173–179. 10.1016/0022-2364(79)90091-X. [DOI] [Google Scholar]
  19. Becconsall J. K.; Daves G. D.; Anderson W. R. Reference-Independent Nuclear Magnetic Resonance Solvent Shifts. J. Am. Chem. Soc. 1970, 92, 430–432. 10.1021/ja00705a632. [DOI] [Google Scholar]
  20. Garroway A. N. Magic-Angle Sample Spinning of Liquids. J. Magn. Reson. 1982, 49, 168–171. 10.1016/0022-2364(82)90308-0. [DOI] [Google Scholar]
  21. Hartmann P.; Grübl D.; Sommer H.; Janek J.; Bessler W. G.; Adelhelm P. Pressure Dynamics in Metal–Oxygen (Metal–Air) Batteries: A Case Study on Sodium Superoxide Cells. J. Phys. Chem. C 2014, 118, 1461–1471. 10.1021/jp4099478. [DOI] [Google Scholar]
  22. Taylor N. G.; Chung S. H.; Kwansa A. L.; Johnson R. R. III; Teator A. J.; Milliken N. J. B.; Koshlap K. M.; Yingling Y. G.; Lee Y. Z.; Leibfarth F. A. Partially Fluorinated Copolymers as Oxygen Sensitive 19F MRI Agents. Chem.—Eur. J. 2020, 26, 9982–9990. 10.1002/chem.202001505. [DOI] [PubMed] [Google Scholar]
  23. Clarke K. G.8 - The Oxygen Transfer Rate and Overall Volumetric Oxygen Transfer Coefficient. In Bioprocess Engineering; Clarke K. G., Ed.; Woodhead Publishing, 2013; pp 147–170. [Google Scholar]
  24. Narita E.; Lawson F.; Han K. N. Solubility of Oxygen in Aqueous Electrolyte Solutions. Hydrometallurgy 1983, 10, 21–37. 10.1016/0304-386X(83)90074-9. [DOI] [Google Scholar]
  25. Ferenz K. B.; Steinbicker A. U. Artificial Oxygen Carriers—Past, Present, and Future—a Review of the Most Innovative and Clinically Relevant Concepts. J. Pharmacol. Exp. Ther. 2019, 369, 300–310. 10.1124/jpet.118.254664. [DOI] [PubMed] [Google Scholar]
  26. Mccall D. W.; Douglass D. C. The Effect of Ions on the Self-Diffusion of Water. I. Concentration Dependence. J. Phys. Chem. 1965, 69, 2001–2011. 10.1021/j100890a034. [DOI] [Google Scholar]
  27. Peng J.; Carbone L.; Gobet M.; Hassoun J.; Devany M.; Greenbaum S. Natural Abundance Oxygen-17 NMR Investigation of Lithium Ion Solvation in Glyme-Based Electrolytes. Electrochim. Acta 2016, 213, 606–612. 10.1016/j.electacta.2016.07.144. [DOI] [Google Scholar]
  28. Zhang C.; Ueno K.; Yamazaki A.; Yoshida K.; Moon H.; Mandai T.; Umebayashi Y.; Dokko K.; Watanabe M. Chelate Effects in Glyme/Lithium Bis(Trifluoromethanesulfonyl)Amide Solvate Ionic Liquids. I. Stability of Solvate Cations and Correlation with Electrolyte Properties. J. Phys. Chem. B 2014, 118, 5144–5153. 10.1021/jp501319e. [DOI] [PubMed] [Google Scholar]
  29. Plévert J.; de Ménorval L. C.; Di Renzo F.; Fajula F. Accessibility of Cation Site in Zeolites by 6Li MAS NMR Spectroscopy Using Paramagnetic O2 as a Chemical Shift Agent. The Example of Zeolite Li-LSX. J. Phys. Chem. B 1998, 102, 3412–3416. 10.1021/jp9802937. [DOI] [Google Scholar]
  30. Liu H.; Grey C. P. Probing Cs+ Cation Accessibility with O2 and 133Cs MAS NMR Spectroscopy. Microporous Mesoporous Mater. 2002, 53, 109–120. 10.1016/S1387-1811(02)00330-X. [DOI] [Google Scholar]
  31. Accardi R. J.; Lobo R. F. Accessibility of Lithium Cations in High-Silica Zeolites Investigated Using the NMR Paramagnetic Shift Effect of Adsorbed Oxygen. Microporous Mesoporous Mater. 2000, 40, 25–34. 10.1016/S1387-1811(00)00234-1. [DOI] [Google Scholar]
  32. Liu H.; Kao H. M.; Grey C. P. 1H MAS and 1H/27Al TRAPDOR NMR Studies of Oxygen-Zeolite Interactions at Low Temperatures: Probing Bronsted Acid Site Accessibility. J. Phys. Chem. B 1999, 103, 4786–4796. 10.1021/jp983686e. [DOI] [Google Scholar]
  33. Hamza M. A.; Serratrice G.; Stébé M. J.; Delpuech J. J. Solute-Solvent Interactions in Perfluorocarbon Solutions of Oxygen. An NMR Study. J. Am. Chem. Soc. 1981, 103, 3733–3738. 10.1021/ja00403a020. [DOI] [Google Scholar]
  34. Taylor J.; Deutsch C. 19F-Nuclear Magnetic Resonance: Measurements of [O2] and PH in Biological Systems. Biophys. J. 1988, 53, 227–233. 10.1016/S0006-3495(88)83084-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Luchette P. A.; Prosser R. S.; Sanders C. R. Oxygen as a Paramagnetic Probe of Membrane Protein Structure by Cysteine Mutagenesis and 19F NMR Spectroscopy. J. Am. Chem. Soc. 2002, 124, 1778–1781. 10.1021/ja016748e. [DOI] [PubMed] [Google Scholar]
  36. Kwak W.-J.; Chae S.; Feng R.; Gao P.; Read J.; Engelhard M. H.; Zhong L.; Xu W.; Zhang J.-G. Optimized Electrolyte with High Electrochemical Stability and Oxygen Solubility for Lithium–Oxygen and Lithium–Air Batteries. ACS Energy Lett. 2020, 5, 2182–2190. 10.1021/acsenergylett.0c00809. [DOI] [Google Scholar]
  37. Wang Y.; Zheng D.; Yang X. Q.; Qu D. High Rate Oxygen Reduction in Non-Aqueous Electrolytes with the Addition of Perfluorinated Additives. Energy Environ. Sci. 2011, 4, 3697–3702. 10.1039/c1ee01556g. [DOI] [Google Scholar]
  38. Zhang S. S.; Read J. Partially Fluorinated Solvent as a Co-Solvent for the Non-Aqueous Electrolyte of Li/Air Battery. J. Power Sources 2011, 196, 2867–2870. 10.1016/j.jpowsour.2010.11.021. [DOI] [Google Scholar]
  39. Black R.; Shyamsunder A.; Adeli P.; Kundu D.; Murphy G. K.; Nazar L. F. The Nature and Impact of Side Reactions in Glyme-Based Sodium–Oxygen Batteries. ChemSusChem 2016, 9, 1795–1803. 10.1002/cssc.201600034. [DOI] [PubMed] [Google Scholar]
  40. Schwenke K. U.; Meini S.; Wu X.; Gasteiger H. A.; Piana M. Stability of Superoxide Radicals in Glyme Solvents for Non-Aqueous Li--O2 Battery Electrolytes. Phys. Chem. Chem. Phys. 2013, 15, 11830–11839. 10.1039/c3cp51531a. [DOI] [PubMed] [Google Scholar]
  41. Temprano I.; Liu T.; Petrucco E.; Ellison J. H. J.; Kim G.; Jónsson E.; Grey C. P. Toward Reversible and Moisture-Tolerant Aprotic Lithium-Air Batteries. Joule 2020, 4, 2501–2520. 10.1016/j.joule.2020.09.021. [DOI] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

jp3c00991_si_001.pdf (396.8KB, pdf)

Articles from The Journal of Physical Chemistry. C, Nanomaterials and Interfaces are provided here courtesy of American Chemical Society

RESOURCES