Unveiling Reaction Pathways of Ethylene Carbonate and Vinylene Carbonate in Li-Ion Batteries

: Ethylene carbonate (EC) and vinylene carbonate (VC) are the archetypical electrolyte solvent and additive in Li-ion batteries (LIBs), respectively. However, our understanding of their reaction pathways remains incomplete. Herein, the reaction pathways of EC and VC are explored by using online electrochemical mass spectrometry complemented by nuclear magnetic resonance analysis. For EC, reduction occurs through two distinct pathways <0.8 V vs Li + /Li, one yielding C 2 H 4 and the other yielding CO, depending on the electrode potential and the EC concentration. The CO-releasing pathway does not contribute to the solid electrolyte interphase formation. For VC, reduction commences at <1.9 V, but CO 2 gas evolution proceeds through a chemical step via a nucleophilic attack and VC ring opening. Additionally, VC scavenges H 2 O and reduced protons via hydrolysis and via proton abstraction from the carbon electrode to form EC. Our study uncovers further reaction pathways and underscores the unique properties of EC and VC, both individually and in combination, and elucidates their roles in influencing the formation process in Li-ion batteries.


■ INTRODUCTION
Despite the widespread application of Li-ion batteries (LIBs) today, there are still essential aspects of this energy storage technology that are not fully understood.Particularly, the solid electrolyte interphase (SEI) on the graphite negative electrode remains elusive. 1 The formation of the SEI primarily proceeds through electrochemical reduction reactions of the electrolyte and its additives, a process that is well known to be critical for LIB performance and lifetime.Multiple SEI reaction pathways have been proposed and most of them are still debated. 2Since several reactions trigger gas evolution, careful analysis thereof, as with online electrochemical mass spectrometry (OEMS), has turned out as one of the most powerful approaches to investigate SEI reactions. 3,4thylene carbonate (EC) is the most commonly used LIB electrolyte solvent, largely because it forms an SEI on the negative electrode when reduced at potentials <0.8 V vs Li + /Li.−11 The process behind Li 2 CO 3 formation is still under debate. 12,13While C 2 H 4 is the predominantly observed gas, EC reduction is also reported to generate CO along with (Li + /poly) ethylene glycol [(Li/P)EG] and their derivatives. 5,7,14However, which factors favor which EC reduction reaction pathway is still unknown.
While EC is the most well-known LIB solvent, vinylene carbonate (VC) is the likewise electrolyte additive. 15Both the anode 16,17 and cathode 18,19 are affected by VC, although a positive effect is mainly observed at the anode.−23 A VC-derived SEI is commonly reported to consist primarily of poly(VC) (e.g., via Scheme 2 as suggested by Zhang et al. 20 ) together with other linear oligomeric/polymeric compounds, as observed by infrared (IR), 16,22−26 nuclear magnetic resonance (NMR), 16,22−24 Xray photoelectron spectroscopy (XPS), 22,27 and transmission electron microscopy (TEM) 22,26 studies.Despite the numerous reports and the major impact of VC on the LIB industry, a clear understanding of VC reaction pathways is lacking, and open questions still remain.For instance, why is not poly(VC) always reported? 28Why is there strong CO 2 evolution observed during VC reduction, 20,21,25 which in turn should be unrelated to poly(VC) formation?
Herein, we elucidate further EC and VC reaction pathways by applying a simplified model electrode/electrolyte system.The study aims to add more pieces to the SEI puzzle toward complete understanding of the SEI in modern LIBs.
The electrolytes used in all experiments consisted of 0.2 M LiClO 4 (battery grade >99.99%,Sigma-Aldrich) in 1,2dimethoxyethane (DME, 99.9%, Honeywell) as a base.The desired concentrations of VC (99.5%, Sigma-Aldrich) and EC (>99%, Sigma-Aldrich) were added to the base electrolyte.A single Celgard 2325 separator (dried for 6 h at 70 °C in a vacuum) was used for all cells.
The commercial LiFePO 4 (LFP) electrodes (1 mAh cm −2 , Customcells) were punched into 15 mm disks and dried at 120 °C under vacuum for 12 h.The LFP electrodes were electrochemically delithiated to ∼90% state of charge (SOC) in LFP | 0.2 M LiClO 4 DME | Li cells to achieve a stable potential plateau (herein assumed to be 3.43 V vs Li + /Li) before any experiment (same procedure as described in our earlier study 29 ).All potentials presented from this point onward are vs Li + /Li.
For online electrochemical mass spectrometry (OEMS), all cells were assembled in an electrode sandwich configuration, comprising GC | 150 μL of 0.2 M LiClO 4 DME:VC:EC | LFP (90% SOC).The cycling protocol was 5 h open-circuit voltage (OCV) followed by cyclic voltammetry (CV) from OCV (∼3.1 V) with vertex potentials of 0.5 and 1.7 V.The scan rate was 0.1 mV/s.The experiments were conducted at 30 °C, and the gas was purged and analyzed every 10 min.The MS was calibrated with H 2 , CH 4 , C 2 H 4 , and CO, O 2 , and CO 2 calibration gases.A detailed description of the OEMS system is found in our earlier study. 30r nuclear magnetic resonance (NMR) cells, constant current was applied to a GC | 400 μL of 0.2 M LiClO 4 DME (or DMSO):VC:EC (0 or 10 v%: 0 or 10 v%) | LFP (90% SOC) system, until 0.5 V was reached, where the potential was held for 3 h.Thereafter, the electrolyte was extracted to NMR tubes.The cycled cells with EC, VC, and EC:VC-added electrolytes were used for NMR analysis, and deuterated DMSO-d 6 was used as the background solvent.All spectra were recorded at room temperature, and anhydrous DMSO-d 6 was used as the internal standard for 1 H and 13 C NMR.All NMR spectra were obtained using a Bruker 3 channel cryoprobe NMR spectrometer (600 MHz ( 1 H), 162 MHz ( 13 C)).The conditions for the 1 H NMR were 45°pulse angle, 8 s delay between pluses, 20.0 kHz spectral width, and 64 scans.The conditions for the 13 C NMR were 45°pulse angle, 1.8 s delay between pluses, 27 kHz spectral width, and 1024 scans.

■ RESULTS AND DISCUSSION
Figure 1 shows the cyclic voltammogram and associated gas evolution data of C 2 H 4 and CO for the model GC | 0.2 M LiClO 4 DME:EC | LFP (90% SOC) cells containing EC in the range of n EC = 0−1560 nmol EC cm GC −2 .All data are normalized to the GC electrode surface area (A GC ).A GC-based electrode was applied to study electrolyte reduction on carbon and avoid effects associated with solvent cointercalation into graphite.Furthermore, in order to reduce the interference of the solvent and salt, DME and LiClO 4 are employed.DME was chosen as a base electrolyte solvent due to its low reduction potential 31 and the high achievable purity to minimize residual contaminants.LiClO 4 was chosen over the commercially used LiPF 6 , as the latter is known to be reactive toward electrolyte reduction products, 32 which would add further complexity to the system.
Figure 1a shows the cyclic voltammogram of the first cycle, revealing consistent current profiles across all of the ECcontaining cells.A distinct current peak at 1.6 V, attributed to the reduction of H 2 O impurities, marks the first noticeable electrochemical reaction.As the potential drops <1.2 V, there is an increase in current, primarily due to the reversible adsorption of Li + . 33Below 0.8 V, currents corresponding to the reduction of EC emerge.Notably, as n EC increases in the cells, the reversible charge decreases, which apart from lithium loss also suggests an increased cell resistance with higher n EC .
C 2 H 4 evolves <0.8 V as expected for cells with n EC > 0 nmol EC cm GC −2 in the electrolyte.The registered total C 2 H 4 amounts reach an upper limit of ∼3 nmol C 2 H 4 cm GC −2 for cells with n EC ≥ 80 nmol EC cm GC −2 (Figure 1b,d), which means that a similar number of mols of EC are needed to passivate the GC surface, according to Scheme 1.The amounts of C 2 H 4 drop off <80 nmol EC cm GC −2 , indicating that n EC is lower than needed for a fully passivating SEI, which is also evident from the second cycle where additional C 2 H 4 evolution is detected in the dilute EC cells (Figure S1).The C 2 H 4 onset shifts from ∼0.5 to >0.7 V as n EC increases, which is attributed to at least two factors: a concentration-dependent EC reaction pathway and limited mass transport of EC to the electrode surface at lower n EC .Although the LEDC or Li 2 CO 3 reaction pathways (Scheme 1) are indistinguishable here, the strongly shifting C 2 H 4 potential onset indicates that the more concentration-dependent LEDC formation pathway (2 EC/C 2 H 4 compared to Li 2 CO 3 formation pathways) is likely the governing SEI process.Evidence of competing EC reaction pathways is found in analysis of the concomitant CO evolution (Figure 1c).Unlike C 2 H 4 , the CO gas evolution always sets in at ∼0.6 V regardless of n EC and the total amount of CO displays a different trend compared to C 2 H 4 (Figure 1d), thus indicating that the processes behind the evolution of these two gases follow different kinetics.For instance, the cells with ∼3 nmol C 2 H 4 cm GC −2 (colored blue in Figure 1) evolution show an increase in CO evolution as n EC decreases, which demonstrates that the EC reduction pathways are n EC -dependent.However, the COevolving EC reaction pathway, with lithium Li-glycol (Li-EG) as a product (Scheme 1), does likely not contribute to the SEI as ∼3 nmol C 2 H 4 cm GC −2 is still detected, i.e., the SEI formation process would be less efficient at lower n EC .
H 2 gas evolution was observed in all cells, as shown in Figure S1.The presence of EC in the electrolyte consistently suppressed a portion of the H 2 evolution, even though a definitive trend was not evident.
Figure 2 shows the cyclic voltammogram and gas evolution trends of CO ).The first observable electrochemical reaction is distinguishable in the range of 1.9−1.6V in the current profile (Figure 2a).All VC-containing cells have a similar negative current slope down to ∼1.4 V without a clear H 2 O reduction plateau compared to that of the pure DME cell.The two highest n VC cells (≥865 nmol VC cm GC −2 ) reach a current peak at 1.4−1.3V, indicating an early fully covering SEI layer, which does not develop to the same extent in the ≤180 nmol VC cm GC −2 cells, as seen in the inset of Figure 2a.A new current peak is observed at ∼0.8 V for the VC-containing cells with ≥865 nmol VC cm GC −2 and is related to CO 2 consumption <1.0 V as discussed below. 25,29Similar current intensity profiles are present for lower n VC cells too but to a lesser extent.A general trend of lower reversible charge per A GC with increasing n VC is observed, as seen in Figure S4.Indeed, the SEI layer formed by VC has previously been reported to increase the internal resistance of Li-ion cells, which explains such a trend. 21,34,35A  The Journal of Physical Chemistry C less efficient Li-adsorption process as a result of the buildup of the VC-based SEI can, however, not be ruled out.
CO 2 gas evolution is detected already >2.0V before the current onset at ∼1.9 V for the cell with the highest n VC of 1280 nmol VC cm GC −2 .However, CO 2 gas evolution is shifted to lower potentials and decreased gas evolution rates are detected with lesser n VC until no CO 2 is present in the pure DME cell (Figure 2b).CO 2 has previously been suggested to originate from an initial electrochemical reduction process of VC along with a chemical propagation step. 20However, we hypothesize that CO 2 from VC at these potentials rather stems from VC ring opening via a nucleophilic attack by reduced surface oxygens on the carbon electrode, which are generated at potentials <2.6 V, 29 as virtually no reduction current is detected at this stage�an observation that agrees with the study by Schwenke et al. 25 A CO 2 consumption reaction is present <1.0 V and is seen as an indent in the gas evolution rate, resulting in a decline in gas evolution rate in all VCcontaining cells.The first CO 2 consumption reactions are associated with the parallel consumption of surface hydrogens as a similar indent in the gas rate profile is present in the H 2 gas evolution data in Figure 2c, likely forming lithium formate (HCO 2 Li, Scheme 3) as discussed below.
The volume of the evolved CO 2 gas in the initial cycle increases with increasing n VC (Figure 2d), in agreement with the study by Petibon et al. 36 The amount of CO 2 evolved scales virtually with the square root of n VC as could be expected if the rate of evolution of CO 2 in VC-containing cells is controlled by the distance between VC molecules.No clear evidence regarding how much VC is reacting can be garnered from the OEMS data.However, as the total amount of CO 2 evolved is much lesser than the amount of VC present, it can be assumed that reaction pathways where no CO 2 is released take place, e.g., via the formation of different oligomers/ polymers where the carbonate groups may be intact.Such pathways have been reported in the literature 16,23,24 and will be discussed later herein.
H 2 evolution onsets at ∼1.6 V in all EC-containing cells (Figure S1), the pure DME cell, and the cell containing 5 nmol VC cm GC −2 (Figure 2c).All other VC-containing cells have the H 2 gas evolution significantly suppressed at ∼1.6 V. Two distinct H 2 evolution peaks are observed, the first reaching its maximum at ∼1.0 V with the second at ∼0.5 V (Figure 2c).The presence of two peaks is likely attributed to the formation of HCO 2 Li at around 1.0 V (Scheme 1), which causes a decrease in the gas evolution rate.This suggests a single main H 2 evolving process rather than two distinct H 2 -evolving reactions, which agrees with the study by Metzger et al.where H 2 evolution was reported to mainly originate from H 2 O reduction. 37The H 2 gas evolution rate is not normalized to the surface area but is reported as the absolute value since H 2 O is expected to be found throughout the cell and not contributing significantly to the SEI on the GC.The formation of a VCbased blocking layer has been suggested as the reason for the lowered H 2 evolution. 38However, we hypothesize that VC when ring-opened (and releasing CO 2 ) scavenges H 2 O, which would as a result be the reason for decreased mols of evolved H 2 .Notably, irrespective of the n VC , the H 2 evolution was never completely suppressed.Instead, a plateau of 0.3 μmol H 2 is reached at 70 nmol VC cm GC −2 and is not affected by higher n VC .Therefore, another H 2 source other than H 2 O reduction is likely present, where surface groups on the carbon are possible candidates.
Another observation is that cells containing VC as the sole cyclic carbonate noticeably evolve CO 2 and cells containing EC as the sole cyclic carbonate show no clear sign of CO 2 evolution (Figure S3).However, a similar amount of CO 2 evolves in cells where VC and EC are mixed together (Figures 2d and S3) as for cells containing only VC, if the CO 2 gas evolution is normalized to the total number of mols of cyclic carbonates in the cells, i.e., n VC+EC .In other words, the VC reaction pathway releasing CO 2 does not discriminate between EC and VC during the propagation step (Figure 4), and ringopened VC apparently can react with EC in the electrolyte.Notably, mixing EC and VC does not alter the trend of H 2 gas evolution (Figure 2e) as the same plateau of 0.3 μmol is reached for the mixed EC:VC cells as for the VC-only cells.
The products from EC and VC reactions were further investigated by NMR as shown in Figure 3.No clear evidence of soluble reduction products was detected in EC, VC, or EC:VC cells.However, the cell containing only VC (∼1.6 M) before cycling contains 3 mM EC after cycling.A similar effect was detected with OEMS (Figure S2), where C 2 H 4 was detected in cells at low n VC even though no EC (n EC = 0) was added.We therefore propose that VC can abstract reduced protons from the GC surface and turn into EC.The reason for only detecting C 2 H 4 at low n VC in the OEMS is due to the VCderived SEI not fully covering the GC surface, and the EC thus formed can be reduced at the uncovered GC surface to the extent at which it is detectable.
Next, the DME electrolyte was exchanged for DMSO-d 6 during cycling in an attempt to identify possible SEI products (Figure S6).The GC | 0.2 M LiClO 4 DMSO:VC:EC (0 or 10 v%: 0 or 10 v%) | LFP (90% SOC) cells were cycled to 1.0 V, and the potential was held for 3 h before the electrolyte was extracted for NMR analysis.Lithium formate was detected, which agreed with the OEMS conclusions.Additionally, vinoxyl groups (CH(�O)−CH 2 R) are present along with a Scheme 3. Lithium Formate Formation

■ CONCLUSIONS
The reduction reaction pathways of EC and VC on carbon have been investigated by operando gas analysis, and our conclusions are schematically illustrated in Figure 4. First, EC reduction follows two distinct reduction pathways, one releasing C 2 H 4 and the other releasing CO.The extent to which each reaction pathway occurs is governed by electrode potential and EC concentration.While the C 2 H 4 -evolving pathway contributes to the SEI, likely by forming LEDC, the CO-evolving pathway does not.Second, CO 2 generated from VC does not originate from VC reduction but from a ring opening process.Third, VC efficiently suppresses H 2 evolution by scavenging reduced protons in two distinct processes.Rather than forming a blocking SEI layer restricting access for H 2 O to the reductive carbon surface, VC supports the consumption of reduced protons on the surface by forming EC and generating CO 2 , which, in turn, scavenges protons to form Li-formate. Fourth, regardless of n VC , a constant H 2 gas evolution plateau is reached, indicating that different H 2 sources from H 2 O are present in the cells.We hypothesize that they are protic surface groups, which cannot be scavenged by VC.Fifth, the species formed from EC and VC reduction and decomposition reactions differ significantly, with different oligomers/polymers forming.Sixth, by mixing EC and VC, the resulting products are more similar to products in the VC-only cells compared to the EC-only cells, hence indicating that VC governs the SEI formation process when present.The top potential axis represents the potential at the GC.Gaseous products are colored (CO 2 is blue, H 2 is dark yellow, C 2 H 4 is pink, and CO is red).Boxed-in products are solid/SEI products.Further end products dissolve in the electrolyte and do not contribute to the SEI.

Figure 1 .
Figure 1.OEMS data for GC | 0.2 M LiClO 4 DME:EC | LFP (90% SOC) cells with different n EC cycled with CV at 0.1 mV/s rate.The gray vertical dashed line represents the vertex potential 0.5 V.The legend shows n EC in every cell.(a) Current profiles for the initial cycle.(b) C 2 H 4 gas evolution trend from EC reduction in the initial cycle.(c) The CO gas evolution trend from EC reduction in the initial cycle.(d) Total C 2 H 4 and CO evolution for different n EC .The maximum and minimum GC surface areas are labeled for the experiment series.The markers are closed for C 2 H 4 and open for CO.

Figure 2 .
Figure 2. OEMS data for a GC | 0.2 M LiClO 4 DME:VC | LFP (90% SOC) system cycled with CV at 0.1 mV/s rate.The gray vertical dashed line represents the vertex potential of 0.5 V.The legend shows n VC in every cell.(a) Electrochemical profiles for the different cells during the initial cycle, with a zoom-in on the >1.2 V electrochemical profiles.(b) CO 2 gas evolution trend from VC in the initial cycle.(c) H 2 gas evolution trend in the initial cycle.(d) Total CO 2 evolution for different n VC .A GC is labeled for selected cells (full list in Table S1).(e) Total H 2 evolution for different n VC .The cells denoted 65 (130) and 80 (225) are a mix of VC and EC (only shown in (d) and (e)), where the first value is n VC and the value in parentheses is n VC+EC .
Our results deepen the understanding of the SEI formation reactions involving VC and EC.The processes identified herein are easily overlooked and challenging to identify in a commercial battery cell environment where multiple reactions overlap and overshadow these processes, thus showcasing importance of studying model systems to understand and model the SEI in LIBs.■ASSOCIATED CONTENT* sı Supporting InformationThe Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.jpcc.4c00927.Online electrochemical mass spectrometry data of GC | 0.2 M LiClO 4 DME:EC | LFP, GC | 0.2 M LiClO 4 DME:VC | LFP, and GC | 0.2 M LiClO 4 DME:EC:VC | LFP cells, generated charge during CV sweeps for ECand VC-containing cells, calculations to determine the H 2 O content in cells and tabulated values comparing the H 2 O content to the VC content in VC-containing cells, extended NMR data of DME-based electrolytes, and NMR data of DMSO-based electrolytes (PDF)

Figure 4 .
Figure 4. Schematic of suggested VC and EC reaction pathways on a carbon electrode based on the findings in this study and literature.9,25,29The top potential axis represents the potential at the GC.Gaseous products are colored (CO 2 is blue, H 2 is dark yellow, C 2 H 4 is pink, and CO is red).Boxed-in products are solid/SEI products.Further end products dissolve in the electrolyte and do not contribute to the SEI.