On the incompatibility of lithium–O2 battery technology with CO2

The peroxide dianion reacts with CO2 in polar aprotic organic media to afford the hydroperoxycarbonate and carbonate radical anions. These highly reactive species, if formed in lithium–O2 cells, can lead to cell degradation via oxidation of the electrolyte and electrode.


Materials and methods
All manipulations were carried out either in a Vacuum Atmospheres model MO-40M glovebox under an atmosphere of N 2 or using standard Schlenk techniques. 1 H, 13 C, 31  to stand for at least 3 days before use. Celite 435 (EMD Chemicals). All glassware were oven dried at 220 • C prior to use. 17  To probe the fate of the missing oxygen atom, the reaction of CO 2     was conducted and indicated there was no observable reaction at 25 • C over the course of 4 hours without added CO 2 ( Figure S6).   Equiv. of DHA 2 yield anthraquinone yield 1 eq 72% 18% 5 eq 87% 55% 10 eq 88% 72% S14 Figure (Table S3). Afterward, vials 2, 3 and 4 were degassed and brought back into the glovebox. The suspensions in each vial were filtered to afford colorless homogeneous solutions.
The formation of methyl methoxyacetate (MMA), one of the possible products of the DME solvent oxidation, was analyzed by 1 H NMR spectroscopy. The filtrate (0.100 mL) was transferred to an NMR tube with DMSO-d 6 (0.500 mL). A benzene solution in DMSO-d 6 (0.010 mL, 0.500 M) was added to each tube as an internal standard. To obtain reliable 1 H NMR integrations, the spectra were measured with a presaturation pulse sequence 1 to suppress the residual DME solvent peaks.
The yield of methyl methoxyacetate (MMA) was calculated based on the integrations of the peak located at 4.04 ppm ( Figure S11, Table S3 and     Table S5).    (Table S6 and Table S7).  Figure S14).
This protocol was found to be highly reproducible (Table S8)   UV-vis spectrometer, and a UV-vis spectrum was taken. The absorbance at 438 nm is proportional to the concentration of Li 2 CO 3 present in the solution. A calibration curve was constructed using standard solutions with known amounts of Li 2 CO 3 ( Figure S15).
The protocol outlined above was used to determine the yield of Li 2 CO 3 from the reaction of Li 2 O 2 with CO 2 (1 atm) in DME. In some cases, the yield of Li 2 CO 3 exceeded 100%, perhaps due to the fact that commercial Li 2 O 2 is often contaminated with Li 2 O and LiOH, which can be converted to

Control experiment A
In a glovebox, a J Young NMR tube was charged with TMSOOTMS (8.7 mg, 0.049 mmol) and 0.500 mL DMF. The J Young NMR tube was capped, the headspace was evacuated for 1 min and back filled with 13 CO 2 (ca. 1 atm). The sample was allowed to thaw and a 13 C NMR spectrum was taken at 21 • C ( Figure S23). No reaction was observed between TMSOOTMS and 13 CO 2 .
S36 Figure S23: 13 C{ 1 H} NMR (DMF-d 7 , 126 MHz, 21 • C) spectrum of the mixture obtained after treatment of TMSOOTMS with 13 CO 2 in DMF at 21 • C. The DMF peak is indicated by a yellow circle and CO 2 by a blue circle.

Control experiment B
In a glovebox, a J Young NMR tube was charged with KOtBu (12.4 mg, 0.111 mmol) and 0.500 mL DMF. The J Young NMR tube was capped, the headspace was evacuated for 1 min and back filled with 13 CO 2 (ca. 1 atm). The sample was allowed to thaw and a 13 C NMR spectrum was taken at 21 • C ( Figure S24). The 13 C NMR data indicate formation of KO 2 COtBu at 157.5 ppm as the only 13 C containing product ( Figure S24).

Preparation of [K(Kryptofix 222)][mBDCA-5t-H 5 ]
We decided to sequester the potassium ion of [K (18- (4) A) in comparison with protonated moieties (C-N av : 1.34Å; C-O av : 1.24Å, Figure S30). Low-temperature (100 K) diffraction data (φ and ω) were collected on a Bruker-AXS X8 Kappa Duo diffractometer coupled to a Smart APEX2 CCD detector with Mo Kα radiation (λ = 0.71073 A) from an IµS micro-source. Absorption and other corrections were applied using SADABS. 6 The structure was solved by direct methods using SHELXT 7 and refined against F 2 on all data by fullmatrix least squares with SHELXL-2015 8 using established refinement approaches. 9 All hydrogen atoms were included in the model at geometrically calculated positions and refined using a riding model. The isotropic displacement parameters of all hydrogen atoms were fixed to 1.2 times the U eq value of the atoms they are linked to (1.5 times for methyl groups). Details about crystal S44 properties and diffraction data can be found in the table below. The program SQUEEZE 10 as implemented in PLATON 5 was used to account for the contribution of disordered solvent contained in voids within the crystal lattice. The solvent contribution was added to the model in a separate file (the .fab file) by SHELXL. Squeeze identified two crystallographically independent solvent accessible voids with a volume of 1106Å 3 . In these voids, Squeeze identified the equivalent of 179 electrons, corresponding to about 8 MeCN molecules.

EPR spectroscopy
Continuous wave EPR experiments were performed on an ECS 106 (Bruker) at X-band (9.8 GHz).
Spectra were collected using 1 G modulation amplitude at 100 kHz.

Mass spectrometry
Electrospray ionization high-resolution mass spectra (ESI-HRMS) of the spin trap samples were measured using an LTQ Orbitrap XL mass spectrometer equipped with an electrospray ionization source (ThermoFisher, San Jose, CA) operating in positive ion mode. For experiments involving CO 2 , the BMPO solution and the EPR tubes were prepurged with CO 2 and the reaction was initiated by adding the peroxide cryptate and BMPO solution described above to this EPR tube containing CO 2 . EPR analysis of the CO 2 purged sample showed a mixture of BMPO-OH, BMPO-O, and BMPO-OCO 2 − (Figure 3C-E, Table 2). Simulations and least squares (LSQ) fittings of the spectra were carried out with EasySpin toolbox (version 4.5.5) in Matlab (Mathworks Inc., Natick, MA). 11 A summary of simulation parameters are given in Table S10.

EPR standard preparation
The standard BMPO-OH solution was prepared by mixing iron(II) sulfate heptahydrate     Table S10 below for simulation parameters.

Experimental solid-state 17 O NMR details
The relevant 17 O SSNMR details can be found in the caption of Figure 5 of the main paper.    was taken. Then the stopcock between the cell and the CO 2 bulb was opened. Spectra ( Figure S40) were taken at −80, −50, −30, 0, and 25 • C ( Figure S40). After acquisition of the spectra, a 1 H NMR spectrum ( Figure S41) was taken to confirm that carbonate cryptate had formed.
A meaningful interpretation of the Raman data could not be made on account of the similarity of the spectrum of the starting and ending points. In addition, low temperature spectra exhibited broadening or disappearance of some peaks, also making in situ interpretation and identification of intermediates impossible. Figure S40: Raman spectra of solid [K(18-crown-6)] 2 [O 2 ⊂mBDCA-5t-H 6 ] exposed to excess CO 2 at various temperatures. The starting spectrum at 25 • C and at −80 • C before adding CO 2 are represented as the black spectra. The red spectra are representative after the addition of CO 2 and due to the lack of differentiation between the spectra, some temperatures are omitted. appears to react with CO 2 based on the differential amount of CO 2 left in the headspace, oxygen gas could not be detected, indicating that the product of the reaction is not oxygen.

13 C NMR calculation
To provide an accurate NMR prediction, an experimental chemical shift (δ) vs. calculated absolute chemical shielding (σ) plot ( Figure S43) was generated using known carbon containing compounds as calibration standards in order to minimize errors due to basis set effects.  Figure S43: Experimental 13 C NMR chemical shifts (δ) vs. calculated absolute chemical shielding (σ).

17 O NMR calculation
To provide an accurate NMR prediction, an experimental chemical shift (δ) vs. calculated absolute chemical shielding (σ) plot ( Figure S44) was generated using known oxygen containing compounds as calibration standards in order to minimize errors due to basis set effects.  27 To determine the bond dissociation energy of symmetrical peroxydicarbonate, the optimized equilibrium structures of peroxydicarbonate dianion and carbonate radical anion were determined and frequency calculations were performed to determine both calculated structures corresponded to local minima using the program Gaussian 09 at the MP2/6-311G++(2d,2p) 28 level of theory. An IEFPCM solvation model (DMSO) was used. 29,30 The coordinates of the optimized structures are deposited in the appendix of this supplementary material. The homolytic BDE, defined as the enthalpy change of equation 7 at 298 K depicted below and includes the electronic energy and zero-point correction: The O-O BDE (∆H 298K ) of symmetrical peroxydicarbonate was found to be 15.9 kcal/mol.