# Fileset

[Advanced Science - 2025 - Dutta - Quantitative Porosity Engineering of Carbon Electrode in Lithium Oxygen Batteries with.pdf](https://mdr.nims.go.jp/filesets/683e0188-cddb-4537-97e4-fd94b715632d/download)

## Creator

[Arghya Dutta](https://orcid.org/0000-0002-3769-7820), [Takashi Kameda](https://orcid.org/0000-0003-2080-3540), Junji Takada, Yuuka Nakajima, Takahiro Morishita, [Shoichi Matsuda](https://orcid.org/0000-0002-0640-3404)

## Rights

[Creative Commons BY Attribution 4.0 International](https://creativecommons.org/licenses/by/4.0/)

## Other metadata

[Quantitative Porosity Engineering of Carbon Electrode in Lithium–Oxygen Batteries with Cell‐Level Gravimetric Energy Density Over 1500&nbsp;Wh kg                    <sup>−1</sup>](https://mdr.nims.go.jp/datasets/e356e016-359d-447e-be96-92fac34afa77)

## Fulltext

Quantitative Porosity Engineering of Carbon Electrode in Lithium–Oxygen Batteries with Cell‐Level Gravimetric Energy Density Over 1500 Wh kg−1RESEARCH ARTICLEwww.advancedscience.comQuantitative Porosity Engineering of Carbon Electrode inLithium–Oxygen Batteries with Cell-Level GravimetricEnergy Density Over 1500 Wh kg−1Arghya Dutta,* Takashi Kameda, Junji Takada, Yuuka Nakajima, Takahiro Morishita,and Shoichi Matsuda*Lithium–oxygen batteries (LOBs) offer an extremely high theoretical energydensity; however, their practical realization depends strongly on the design ofporous carbon positive electrodes. Most prior efforts have emphasizedmaterial design while overlooking the role of the electrolyte stored withinpores, leaving the design principles for achieving practicalhigh-energy-density LOBs unclear. In the present study, through simulations,it is quantitatively demonstrated that while increasing pore volume initiallyimproves energy density, it eventually plateaus due to increasing electrolytedemand. The simulations indicate that reduced electrolyte volumes andoptimized mass loading of the positive electrode are crucial for maximizingenergy density. Experimental validation with systematically tuned carbonelectrodes in pouch-type LOBs with realistic mass-loadings supports thesefindings. While large pore volumes enhance capacity, they require excessiveelectrolyte, ultimately counter-balancing energy density. Conversely, loweringelectrolyte volumes in highly porous electrodes leads to incomplete filling,increased impedance, enhanced parasitic reactions, and poor cycling stability.As a result, by tailoring the pore structure, electrodes capable of deliveringcell-level energy density exceeding 1500 Wh kg−1 and maintaining stablecycling under capacity-limited conditions are demonstrated. This workredefines the role of pore engineering in LOB electrodes, highlighting itscrucial contribution to achieving practical, high-energy, and long-lasting LOBs.A. Dutta, T. Kameda, S. MatsudaCenter for Green Research on Energy and Environmental MaterialsNational Institute for Material Science1-1 Namiki, Tsukuba, Ibaraki 305-0044, JapanE-mail: DUTTA.Arghya@nims.go.jp;MATSUDA.Shoichi@nims.go.jpJ. Takada, Y.Nakajima, T.MorishitaKondoTeruhisaMemorial AdvancedCarbonTechnologyCenterToyoTansoCo., Ltd.5-7-12 Takeshima,Nishiyodogawa-ku,Osaka, Japan, 555-0011S.MatsudaCenter for AdvancedBattery CollaborationNational Institute forMaterial Science1-1Namiki Tsukuba, Ibaraki 305-0044, JapanThe ORCID identification number(s) for the author(s) of this articlecan be found under https://doi.org/10.1002/advs.202514406© 2025 The Author(s). Advanced Science published by Wiley-VCHGmbH. This is an open access article under the terms of the CreativeCommons Attribution License, which permits use, distribution andreproduction in any medium, provided the original work is properly cited.DOI: 10.1002/advs.2025144061. IntroductionRechargeable lithium–oxygen (Li–O2) bat-teries (LOBs) are widely recognized as apromising next-generation energy storagetechnology, with theoretical gravimetric en-ergy densities surpassing those of con-ventional lithium-ion batteries (LIBs).[1,2]While the commercial positive electrodematerials in LIBs offer gravimetric en-ergy densities of 500–800 Wh kg−1, LOBscan theoretically achieve ≈3500 Wh kg−1when evaluated solely based on the redox-active material.[3] Despite this potential,practical implementation remains ham-pered by several critical challenges, par-ticularly the design and optimization ofthe porous carbon positive electrode.[4–6]In nonaqueous LOBs, the formation ofinsoluble lithium peroxide (Li2O2) duringdischarge obstructs oxygen and lithium-ion pathways, significantly restricting theeffective utilization of the porous carbonelectrode.[7] Extensive research has focusedon how the structural features of carbonelectrodes, such as surface area, pore size,and pore volume, influence their dischargeperformance.[8–14] For instance, Meini et al.demonstrated that electrodes with larger carbon surface ar-eas typically deliver enhanced discharge capacities.[8] Dinget al. examined diverse carbon materials, including carbonblacks, mesoporous carbons, multiwalled carbon nanotubes, andreduced graphene oxide, revealing a clear positive correlation be-tween pore size and discharge capacity.[9] In contrast, Kubokiet al. underscored the importance of electrode pore volume,showing a stronger link between this parameter and dischargeperformance.[10] Despite these differing emphases, there is broadconsensus that an optimal air electrode for LOBs should com-bine three essential characteristics: (i) high surface area tosupport electrochemical activities and accommodate dischargeproduct deposition, (ii) sufficiently large-sized pores to promoteeffective oxygen and lithium-ion transport and avoid pore block-age, and (iii) high pore volume to provide ample space for thegrowth of discharge products such as Li2O2.Although a lot of work has been done with the emphasis onincreasing the electrode porosity, the optimization of electrolyteAdv. Sci. 2026, 13, e14406 e14406 (1 of 14) © 2025 The Author(s). Advanced Science published by Wiley-VCH GmbHhttp://www.advancedscience.commailto:DUTTA.Arghya@nims.go.jpmailto:MATSUDA.Shoichi@nims.go.jphttps://doi.org/10.1002/advs.202514406http://creativecommons.org/licenses/by/4.0/http://creativecommons.org/licenses/by/4.0/http://crossmark.crossref.org/dialog/?doi=10.1002%2Fadvs.202514406&domain=pdf&date_stamp=2025-10-15www.advancedsciencenews.com www.advancedscience.comvolume used in a cell has, in general, remained an overlookedfactor in LOB research.[15–17] Typically, LOBs have been evaluatedusing an excess of electrolyte (>50 μL cm−2), low areal loading ofcarbon material, and limited areal capacity (<1 mAh cm−2).[18,19]This practice not only limits the attainable cell-level energydensity, often yielding values lower than those of conventionalLIBs, but also obscures a comprehensive analysis by artificiallyinflating performance metrics. Our recent work highlightsthe relevance of the electrolyte-to-capacity (E/C) ratio, a metrictraditionally used in LIBs, as a valuable descriptor for assessingLOB energy density.[20] Specifically, we find that maintaining anE/C ratio below 10 g Ah−1 is essential for achieving practicallyrelevant gravimetric energy densities in LOBs. Moreover, giventhat the electrolyte accounts for a substantial fraction of the totalLOB cell cost, reducing its content also yields a significant eco-nomic advantage. In addition to reducing the electrolyte volume,high gravimetric energy density requires positive electrodeswith substantial mass loadings (>4 mg cm−2) and high arealcapacities (>4 mAh cm−2). However, increasing the electrodethickness exacerbates pore clogging by discharge products, lead-ing to underutilized active material and emphasizing the criticalrole of high porosity in ensuring full electrode utilization.[21] Yet,high porosity introduces a key trade-off: although it enhancesdischarge capacity, it also demands a larger electrolyte volume toensure complete pore filling and adequate ionic transport withinthe structure, which can reduce cell-level gravimetric energydensity. Despite this understanding, systematic and quantitativestudies addressing how to engineer and optimize the porousstructure of carbon-based positive electrodes for practical high-energy LOBs operating under lean electrolyte conditions remainscarce. Addressing this gap is crucial for realizing the full po-tential of LOBs, achieving high gravimetric energy density, andensuring stable cycling under practical conditions.This study aims to quantitatively evaluate the impact ofelectrode porosity on electrolyte demand and gravimetric energydensity in LOBs and to strategically engineer porosity to enableboth high gravimetric energy density and extended cycle lifeunder lean electrolyte conditions. Through simulations, we firstclarify the critical trade-off: while increasing pore volume initiallyboosts gravimetric energy density, this benefit gradually plateausas electrolyte requirements also increase.We then experimentallyvalidate this trade-off using a series of systematically engineeredporous carbon materials with controlled variations in surfacearea, pore diameter, and pore volume. These electrodes, withpractical carbon loadings exceeding 4.0mg cm−2, were integratedinto pouch-type LOBs and evaluated under lean electrolyte condi-tions. Detailed analyses show that although larger pore diametersand volumes enhance absolute capacity, they do not necessarilyyield higher specific capacity or gravimetric energy density at thesame rate due to increased electrolyte demand. Conversely, inattempts to lower the amount of electrolyte, cells with high porediameters and volumes suffered from incomplete filling, elec-trolyte depletion, increased impedance, voltage polarization, andaccelerated electrochemical degradation. These findings under-score the importance of precise pore structure optimization tobalance electrolyte use and energy density. By tailoring the porestructure, we demonstrate electrodes capable of delivering gravi-metric energy densities exceeding 1500 Wh kg−1 (normalized tototal cell mass, including electrolyte), maintaining stable cyclingin capacity-limited conditions. This work redefines the conven-tional understanding that higher porosity universally enhancesLOB performance, highlighting the crucial role of optimizedporosity in achieving practical, high-energy LOB systems.2. Results2.1. Simulation of Gravimetric Energy Density Variation withPore VolumeSince solid Li2O2 is deposited during the discharge process ofLOBs, it is intuitive that carbon materials with wide pore diam-eters, large pore volumes, and high surface areas, providing am-ple space for accommodating Li2O2, would result in higher cellcapacities. Typically, the specific capacity of LOBs is normalizedonly to the mass of the carbon material, often neglecting thecontribution of electrolyte mass.[9,16,18,19,22] Consequently, boththe specific capacity and gravimetric energy density of the celltend to increase with pore volume, as illustrated schematically inFigure 1a. However, in practical applications, the electrolyte rep-resents the largest mass fraction among all cell components.[20]Therefore, considering themass of the electrolyte is crucial whendeveloping high-energy-density, practically viable cells. Whilehigh porosity in the carbon material is advantageous for increas-ing the absolute capacity, it also necessitates a larger amount ofelectrolyte to fill the pores.Figure 1b schematically illustrates that as pore volume in-creases, the required electrolyte volume correspondingly in-creases. This added electrolyte mass offsets the capacity gains,suggesting that the positive effect of high porosity on specificcapacity and gravimetric energy density is expected to be con-strained.Next, we have conducted simulations to quantitatively assessthe effects of electrode porosity and electrolyte loading on thegravimetric energy density of a pouch-type LOB cell. The cell con-figuration and the mass distribution of the cell components un-der consideration are shown in Figure 2a,b. The details of thesimulation are provided in theNote S1 (Supporting Information).A practically relevant electrode loading of 4.0 mg cm−2 was se-lected, with the assumption that the electrode pores are fully sat-urated with electrolyte (100% filling). The mass of all other cellcomponents, including the electrolyte, was also included in thegravimetric energy density calculation, and the average dischargevoltage was set at 2.7 V versus Li/Li+. While high electrode load-ing is generally considered advantageous for achieving highergravimetric energy density, a comprehensive quantitative assess-ment that accounts for both pore volume and electrolyte masshas been lacking. We first examined how electrode loading (0.1–10 mg cm−2) and pore volume (0–10 cm3 g−1) simultaneously in-fluence the gravimetric energy density. Figure 2c presents a con-tour map of gravimetric energy density as a function of these twoparameters, under the assumption of complete pore filling withLi2O2 and 100% electrolyte saturation. The results show that in-creasing both pore volume and electrode loading initially leadsto a sharp rise in gravimetric energy density, reaching ≈3000 Whkg−1 at a pore volume of 3 cm3 g−1 and an electrode loadingof 2.5 mg cm−2. However, beyond these values, the increase ingravimetric energy density slows significantly. This trend is fur-ther clarified in Figure 2d, which shows line plots of gravimetricAdv. Sci. 2026, 13, e14406 e14406 (2 of 14) © 2025 The Author(s). Advanced Science published by Wiley-VCH GmbH 21983844, 2026, 1, Downloaded from https://advanced.onlinelibrary.wiley.com/doi/10.1002/advs.202514406 by Shoichi MATSUDA - Argonne National Laboratory , Wiley Online Library on [13/01/2026]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons Licensehttp://www.advancedsciencenews.comhttp://www.advancedscience.comwww.advancedsciencenews.com www.advancedscience.comFigure 1. Schematic illustrations of Li2O2 deposition (discharge process) inside the electrode pores a) without electrolyte and b) with electrolyte inconsideration.energy density versus pore volume at various electrode loadings.Increasing loading from 0.1 to 1.0 mg cm−2 yields a substan-tial improvement in gravimetric energy density; for example, ata pore volume of 4 cm3 g−1, the gravimetric energy density risesfrom ≈400 to ≈2200 Wh kg−1. A further increase to 4.0 mg cm−2boosts the value to ≈4000 Wh kg−1. However, raising the loadingto 10 mg cm−2 results in only a modest gain (≈4600 Wh kg−1),suggesting that excessive electrode mass-loading does not sub-stantially enhance energy metrics. A similar analysis with theelectrode thickness and porosity, in Figure S1a,b (Supporting In-formation), shows the same trend. Increasing electrode thicknessdoes not linearly improve gravimetric energy density; instead,thick electrodes often face mass transport limitations, unevenLi2O2 deposition, and incomplete utilization, resulting in lower-than-expected gravimetric energy density.[21] Based on these in-sights, we fixed the electrode loading at 4.0 mg cm−2 and system-atically varied electrolyte loading to investigate its influence ongravimetric energy density. Figure 2e shows a contour plot de-picting the relationship between gravimetric energy density andelectrode pore volume at various electrolyte-loading levels. Theanalysis assumes that the entire pore volume is initially occupiedby electrolyte and is subsequently filled by Li2O2 upon discharge,with full pore utilization for product deposition.The data indicate that the impact of pore volume on gravimet-ric energy density is highly dependent on the amount of elec-trolyte. When the electrolyte loading exceeds 80%, the gravimet-ric energy density increases sharply with pore volume up to ≈4cm3 g−1, beyond which further gains are minimal. This suggeststhat under high electrolyte-loading conditions, the advantage ofincreasing pore volume is quickly saturated. On the other hand,reducing the electrolyte content has a much stronger influenceon enhancing the gravimetric energy density. Figure 2f simpli-fies these observations: at 100% pore filling, increasing the porevolume from 4 to 10 cm3 g−1 raises the gravimetric energy den-sity from ≈4000 to only ≈5100 Wh kg−1, a modest 27% gain. Incontrast, reducing electrolyte loading to 75%, 50%, and 25% ofthe pore volume at 10 cm3 g−1 leads to dramatic increases ingravimetric energy density, reaching≈6600, 8700, and 12900Whkg−1, respectively. The corresponding contour plot and line plotsfor electrolyte loading and porosity of the electrode are shown inFigure S2a,b (Supporting Information).Taken together, these findings demonstrate that while in-creasing pore volume can initially boost specific capacity andgravimetric energy density, the benefits saturate beyond a cer-tain threshold. Instead, minimizing the electrolyte volume andproperly optimizing the mass-loading/thickness of the elec-trode emerge as essential parameters to maximize gravimetricenergy density. Nevertheless, operating under lean electrolyteconditions with large-pore electrodes introduces additionalcomplexities, such as incomplete electrode filling, electrolyteAdv. Sci. 2026, 13, e14406 e14406 (3 of 14) © 2025 The Author(s). Advanced Science published by Wiley-VCH GmbH 21983844, 2026, 1, Downloaded from https://advanced.onlinelibrary.wiley.com/doi/10.1002/advs.202514406 by Shoichi MATSUDA - Argonne National Laboratory , Wiley Online Library on [13/01/2026]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons Licensehttp://www.advancedsciencenews.comhttp://www.advancedscience.comwww.advancedsciencenews.com www.advancedscience.comFigure 2. a) Schematic representation of the cross-section and b) mass ratio of different cell components of the LOB pouch cell employed in this study.c) Contour map of the simulated gravimetric energy density (total cell mass) against pore volume and electrode mass loading, with 100% of the poresfilled with electrolyte. d) Line plots of simulated gravimetric energy density against pore volume for four specific electrode loading conditions. e) Contourmap of the simulated gravimetric energy density (total cell mass) against pore volume and electrolyte loading, for an electrode loading of 4 mg cm−2. f)Line plots of simulated gravimetric energy density against pore volume for five electrolyte loading conditions.depletion, and long-term stability challenges, all of which mustbe carefully addressed.[20,23] In the following sections, we presenta systematic approach to engineer porous carbon materials withcontrolled pore structures and investigate how this optimizationof electrode architecture and electrolyte volume governs both thegravimetric energy density and cycling stability of LOBs, criticalfactors for translating this technology into practical, high-energyapplications.2.2. Synthesis of MgO Templated Porous Carbon PowdersNanoporous carbon materials were synthesized through a hard-templating approach, employing phenol resin as the carbonsource and magnesium oxide (MgO) as the hard template. Thismethodology affords precise control over the resulting mesoporesize of the carbon materials, which can be tailored by adjustingthe dimensions of the MgO template. The synthesis procedureinvolved subjecting a mixture of phenol resin and MgO to heattreatment at 900 °C under nitrogen (N2) atmosphere, leading tothe carbonization of the precursormaterial. Subsequently, the se-lective dissolution of the MgO template was carried out at roomtemperature using diluted sulfuric acid. This step resulted in theisolation of the carbon structure, followed by a secondary heattreatment at 1800 °C. The resultant mesoporous carbons, charac-terized by varying pore diameters, are identified asMPC-5, MPC-10, MPC-18, MPC-33, andMPC-38, with each number indicatingthe respective pore diameter in nanometers. The synthesis pro-cess of the carbons is schematically shown in Figure S3 (Support-ing Information).2.3. Physico-Chemical Characterization of the Carbon MaterialsA comprehensive investigation into the physicochemical prop-erties of the porous carbon materials was conducted usingvarious analytical techniques. N2 adsorption and desorptionmeasurements at -196 °C were utilized to determine the surfaceAdv. Sci. 2026, 13, e14406 e14406 (4 of 14) © 2025 The Author(s). Advanced Science published by Wiley-VCH GmbH 21983844, 2026, 1, Downloaded from https://advanced.onlinelibrary.wiley.com/doi/10.1002/advs.202514406 by Shoichi MATSUDA - Argonne National Laboratory , Wiley Online Library on [13/01/2026]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons Licensehttp://www.advancedsciencenews.comhttp://www.advancedscience.comwww.advancedsciencenews.com www.advancedscience.comFigure 3. a) N2 adsorption/desorption isotherms and b) BJH pore size distribution (from adsorption data) of different carbon powders. c) TEM imageof the MPC-10 powder sample. d) Schematic illustration of the NIPS method employed to fabricate the porous carbon electrodes. e) Cross-sectionalSEM image of the MPC-10 electrode. Comparison of f) surface area and g) pore volume of different porous carbon electrodes.area, pore volume, and pore diameter of the powder samples.The N2 adsorption/desorption isotherms and the Barrett-Joyner-Halenda (BJH) pore size distribution curves are depicted inFigure 3a,b, respectively, with summarized results provided inTable S1 (Supporting Information). The analysis reveals thatthe samples exhibited a range of pore diameters from 5 to38 nm. Interestingly, there was a gradual decrease in the BET(Brunauer−Emmett−Teller) surface area of the samples as thepore diameter increased, following this sequence: MPC-5 (1423m2 g−1) > MPC-10 (1233 m2 g−1) > MPC-18 (1117 m2 g−1) >MPC-33 (757 m2 g−1) >MPC-38 (623 m2 g−1). However, the porevolume generally showed an increasing trend with the enlarge-ment of the pore diameter, except for MPC-38. The trend of porevolume follows this sequence: MPC-5 (2.0 cm3 g−1) < MPC-10(2.98 cm3 g−1) < MPC-38 (3.04 cm3 g−1) < MPC-18 (3.71 cm3g−1) < MPC-33 (3.93 cm3 g−1). We qualitatively examined theoxygen-containing functional groups using X-ray photoelectronspectroscopy (XPS), focusing on MPC-10 as a representativecase. The XPS C1s spectrum (Figure S4, Supporting Infor-mation) reveals the presence of various functional groups. Inaddition to the components at 284.6 and 285.5 eV correspondingto sp2 and sp3 hybrid forms of carbon, we observed C−O func-tional groups, including phenol, ether, and epoxy, with bindingenergies ≈286.5 eV.[24] Furthermore, C═O functional groups,such as carbonyl and quinone, were detected at ≈287.5 eV,and carboxyl carbon at 288.9 eV.[24] The corresponding peaksfor these functional groups in the O1s spectrum are shown inFigure S5 (Supporting Information).Adv. Sci. 2026, 13, e14406 e14406 (5 of 14) © 2025 The Author(s). Advanced Science published by Wiley-VCH GmbH 21983844, 2026, 1, Downloaded from https://advanced.onlinelibrary.wiley.com/doi/10.1002/advs.202514406 by Shoichi MATSUDA - Argonne National Laboratory , Wiley Online Library on [13/01/2026]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons Licensehttp://www.advancedsciencenews.comhttp://www.advancedscience.comwww.advancedsciencenews.com www.advancedscience.comConcerning microstructural characterization, scanning elec-tron micrographs (SEM) in Figure S6 (Supporting Information)illustrate that at lower magnifications, all carbon samples displaya particle-like morphology, with sizes distributed across severalmicrometers. However, upon closer examination at higher mag-nification (as depicted in Figure S7, Supporting Information), itbecomes apparent that these particles are comprised of interwo-ven carbon flakes. Figure S8 (Supporting Information) illustratesthe Raman spectrum of MPC-10, serving as a representativematerial and offering significant insights into the structuralproperties of carbon. The presence of the D band at 1350 cm−1signifies vibrations associated with sp3-bonded carbon atoms ordefects, while the G band at 1580 cm−1 suggests the presenceof sp2-bonded carbon atoms within a more graphitized carbonframework.[25] Furthermore, the ID/IG intensity ratio of 1.49indicates a turbostratic structure for the carbon. Figure S9(Supporting Information) depicts the X-ray diffraction (XRD)pattern of carbon sample MPC-10, indicating the developmentof a graphitic structure. The XRD patterns of the carbons showa prominent diffraction peak at 2𝜃 = 24°, corresponding tothe (002) crystallographic plane of graphite crystallites. Anotherpeak at ≈43° is considered as an overlap of two peaks. The peakat 42.4°, corresponding to the (100) plane, while the other peakat 44.3°, is associated with the (101) plane.[26] The randomlystacked graphitic layers in the carbon material MPC-10 arevisually confirmed through the high-resolution transmissionelectron micrograph (TEM) shown in Figure 3c. The appearanceof graphitic layers provides direct evidence of graphitization inthe carbon. Additionally, the TEM image confirms the presenceof mesoporosity in the sample.2.4. Fabrication of Self-Standing Porous Carbon MembraneThe synthesized carbon materials were employed in the fabri-cation of a self-standing porous carbon membrane using a tech-nique known as nonsolvent-induced phase separation (NIPS).[23]In this process, a solvent that does not dissolve the polymeris introduced into a polymeric film, leading to the creation ofinterconnected voids.[27] This ultimately results in the formationof a sponge-like structure within the polymer film. By employingthis methodology, we successfully produced carbon membranesfeaturing a hierarchical macro-mesoporous interconnectednetwork, which is expected to enhance both the transport prop-erties and capacity of the electrode. To create these self-standingmembranes, we combined the synthesized porous carbonpowders with carbon fibers (CF), carbon nanotubes (CNTs),and polymeric materials. The incorporation of CF and CNTsserved to enhance the mechanical strength of the membranein preventing the disintegration of the electrode during cycling.We prepared a typical slurry mixture comprising 75 wt.% carbonpowder, 5 wt.% CF, 5 wt.% CNTs, and 15 wt.% PAN (polyacry-lonitrile). This slurry was utilized to generate a uniform carbonfilm using a wet film-forming method involving a doctor blade.Subsequently, the film was immersed in methanol, acting as apoor solvent, to induce the formation of a porous film throughthe NIPS process. The resulting membranes were subsequentlydried and subjected to an appropriate heat treatment, followedby carbonization at 1050 °C within N2 atmosphere. Figure 3ddisplays a schematic illustration of the fabrication of the hierar-chical porous carbon membrane using the NIPS method. Thecross-section of themembrane prepared withMPC-10 as a repre-sentative was observed using a focus ion beam scanning electronmicroscopy (FIB-SEM). The micrograph in Figure 3e clearlydemonstrates the interparticle macroporosity in the membrane.This SEM analysis complements the observed mesoporosityin the TEM, confirming a hierarchical porous structure of themembrane.2.5. Analysis of the Porosity of the Self-Standing CarbonMembraneTo characterize the surface area, pore volume, and pore di-ameter of the membranes, N2 adsorption and desorptionmeasurements were performed at -196 °C. The resulting N2adsorption/desorption isotherms and the pore size distribu-tion curves, analyzed using the BJH method, are depicted inFigures S10 and S11 (Supporting Information), respectively.Examination of the pore size distribution curves indicates thatthe carbon membranes maintain pore sizes similar to those ofthe powder samples. However, it is noted that the surface areaand pore volume of the membranes are slightly lower comparedto the powders, which can be attributed to the incorporationof CNT, carbon fibers, and PAN with lower porosity into themembrane structure. The surface area and pore volume data arepresented in Figure 3f,g, respectively. The trend in BET surfacearea shows the same pattern observed in the powder samples,demonstrating a decreasing trend as pore diameter increases.However, the surface area contributed by large pores (diameter> 20 nm) shows a gradual increase from MPC-5 to MPC-33,followed by a decline in MPC-38. Furthermore, we carried out adetailed analysis of the pore volume of the membrane electrodes.To better understand the pore volume characteristics of themem-brane electrodes, we conducted a detailed analysis using bothN2 adsorption/desorption and mercury porosimetry. As the porediameter increases, the pore volume associated with micropores(diameter < 2 nm) gradually decreases, dropping from 0.46 cm3g−1 in MPC-5 to 0.22 cm3 g−1 in MPC-38. Conversely, the porevolume for mesopores initially increases from MPC-5 to MPC-33, before decreasing in MPC-38. More specifically, the volumeof the small mesopores (diameter 2–20 nm) increases, from 1.57cm3 g−1 inMPC-5 to 3.26 cm3 g−1 inMPC-33, and then decreasesto 2.67 cm3 g−1 in MPC-38. The same trend is observed for largemesopores (diameter > 20 nm), although the changes are morepronounced. For example, the pore volume of MPC-33 (2.68 cm3g−1) is more than five times greater than that of MPC-5 (0.41cm3 g−1). Since the NIPS method was applied consistently underidentical conditions following the preparation of carbon pow-ders, the differences in macropore (> 200 nm) volume betweensamples were minimal, as shown in Figure 3g. Consequently,the total pore volume of the carbons generally increased, exceptfor MPC-38. These results indicate that for MPC-33 andMPC-38,not only is the total pore volume high, but a significant portion ofthis volume is derived from largemesopores. All the BET surfacearea and pore volume data of the membrane electrodes are sum-marized in Table 1. These findings unequivocally suggest thatthese carbon membranes share similar macroporous structures,Adv. Sci. 2026, 13, e14406 e14406 (6 of 14) © 2025 The Author(s). Advanced Science published by Wiley-VCH GmbH 21983844, 2026, 1, Downloaded from https://advanced.onlinelibrary.wiley.com/doi/10.1002/advs.202514406 by Shoichi MATSUDA - Argonne National Laboratory , Wiley Online Library on [13/01/2026]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons Licensehttp://www.advancedsciencenews.comhttp://www.advancedscience.comwww.advancedsciencenews.com www.advancedscience.comTable 1. Surface area and pore volume of different membrane electrodes measured by N2 adsorption/desorption and Hg porosimetry.Membrane electrode BET surface area [m2 g−1] Pore volume [cm3 g−1]Micropore Small mesopore Large mesopore Macropore TotalMPC-5 1118 0.46 1.57 0.41 3.29 5.73MPC-10 1070 0.4 2.63 1.05 2.9 6.98MPC-18 1035 0.37 3.11 1.58 3.42 8.48MPC-33 913 0.25 3.26 2.68 3.42 9.61MPC-38 595 0.22 2.67 2.18 2.87 7.94providing a consistent framework for the diffusion of oxygen andthe deposition of Li2O2 during discharge. However, variationsin the mesoporous structure of membrane electrodes couldpotentially lead to differences in electrolyte loading, capacity,and energy density.2.6. Application of Free-Standing Membrane Electrodes in LOBs2.6.1. Discharge Capacity Estimation of the Carbon ElectrodesThe discharge performance of positive electrodes comprisingvarious mesoporous carbon materials was systematically in-vestigated using pouch-type LOB cells under three controlledelectrolyte loadings, corresponding to 100%, 80%, and 65% of thetotal pore volume of each electrode. Detailed cell configurationsare provided in the Experimental methods, and the masses of thevarious cell components are shown in Table S2 (Supporting Infor-mation). For the cell tests, an electrolyte containing 0.5 m lithiumbis(trifluoromethanesulfonyl)imide (LiTFSI), 0.5 m lithium ni-trate (LiNO3), and 0.2 m lithium bromide (LiBr) dissolved intetraethylene glycol dimethyl ether (TEGDME) was employed.TEGDME was chosen for its stability against both electrodes andits high boiling point, which prevents solvent loss in the open-cellconfiguration of LOBs.[28] Each salt served a distinct function.LiTFSI in the electrolyte ensures high ionic conductivity, LiNO3stabilizes the Li anode and promotes solution-phase discharge,suppressing carbon surface passivation, and LiBr reduces chargeoverpotential while further stabilizing the Li surface.[29–32]Galvanostatic discharge was performed at a current density of0.4 mA cm−2 to a cut-off voltage of 2.0 V versus Li/Li+. The dis-charge voltage profiles obtained under each electrolyte conditionare presented in Figures S12–S14 (Supporting Information),while the resulting specific capacities, normalized to the carbonmass, are summarized in Figure 4a. All samples, with the excep-tion of MPC-5 (pore volume < 6 cm3 g−1), delivered specific ca-pacities exceeding 3500 mAh g−1. Figures S15–S17 (SupportingInformation) further illustrate the correlation between dischargecapacity and total pore volume. Interestingly, the results revealthat an increase in pore volume does not linearly translate tohigher specific capacity. FromMPC-5 to MPC-18, via MPC-10, anincrease in specific capacity is observed. However, beyond thisrange, as both pore diameter and pore volume further increase(MPC-33), no additional capacity improvement is noted. Never-theless, in the case of MPC-38, the reduction in capacity can beattributed to lower pore volume compared to MPC-18 and MPC-33. This non-monotonic trend suggests that excessively largepore volumesmay lead to premature cell failure, primarily due toearly passivation of the carbon surface by electronically insulatingLi2O2, well before complete pore filling. Furthermore, larger porevolumes inherently necessitate increased electrolyte volumes toensure adequate filling, which can offset the energy density gainsachieved via higher capacities. Figure 4b and Table S3 (Support-ing Information) present the electrolyte-to-carbon mass ratioscorresponding to three electrolyte loading levels, 100%, 80%, and65% of the total pore volume, of the respective carbon electrodes.As expected, an increase in electrode pore volume leads to arise in electrolyte mass per unit mass of electrode. To assess thepractical implications of this, the specific capacities of the LOBcells were recalculated by incorporating the actual electrolytemasses under each condition. With the exception of MPC-5,which has the lowest pore volume, the inclusion of electrolytemass in the specific capacity normalization yields minimal varia-tion across other electrode types (Figure 4c; Table S3, SupportingInformation). Notably, a declining trend in specific capacityis observed from MPC-10 to MPC-38 when normalized to thecombined mass of carbon and electrolyte, indicating that porevolumes beyond an optimum limit may lead to diminishingreturns in capacity enhancement.This trend is also evident in the gravimetric energy densityprofiles shown in Figure 4d and detailed in Table S3 (SupportingInformation). These results reveal that increasing electrodeporosity does not inherently translate into higher gravimetricenergy density at the cell level, where masses of all the cell com-ponents, including the electrolyte mass (Table S2, SupportingInformation), were considered. On the contrary, the gravimetricenergy density exhibits a more pronounced dependence on elec-trolyte mass. For instance, in the case of MPC-18, the gravimetricenergy density increases markedly from ≈1050 Wh kg−1 at 100%electrolyte-loading to an unprecedented value of 1500 Wh kg−1under 65% electrolyte-loading. This emphasizes the critical roleof electrolyte volume minimization in achieving high energydensities. Therefore, instead of continuously increasing porevolume, which enhances electrolyte demand, a more effectivestrategy for practical LOB design is to optimize electrode archi-tecture in tandem with lean electrolyte conditions. Figure 4eand Figure S18 (Supporting Information) present a comparativeanalysis of the gravimetric energy densities achieved in this studyagainst those reported for LOB cells in the literature.[20,22,33–36]The areal capacities and areal current densities of the cells men-tioned in Figure 4e are shown in Table S4 (Supporting Informa-tion). Notably, the pore-optimized carbon electrodes developedAdv. Sci. 2026, 13, e14406 e14406 (7 of 14) © 2025 The Author(s). Advanced Science published by Wiley-VCH GmbH 21983844, 2026, 1, Downloaded from https://advanced.onlinelibrary.wiley.com/doi/10.1002/advs.202514406 by Shoichi MATSUDA - Argonne National Laboratory , Wiley Online Library on [13/01/2026]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons Licensehttp://www.advancedsciencenews.comhttp://www.advancedscience.comwww.advancedsciencenews.com www.advancedscience.comFigure 4. a) Specific capacity (normalized to carbon mass), b) Electrolyte to carbon mass ratio, c) Specific capacity (normalized to carbon + electrolytemass), and d) Gravimetric energy density (normalized to total cell mass) of different electrodes with different electrolyte loading amounts. e) Comparisonof the gravimetric energy density (normalized to total cell mass) reported in this work with a few selected earlier reports with high values.here deliver cell-level gravimetric energy densities that surpassnearly all previously reported values to a large extent, with onlyone exception. This clear performance advantage underscoresthe effectiveness and novelty of the pore-structure engineeringstrategy in advancing practical, high-energy LOB systems.2.6.2. Discharge/Charge Cycling TestFollowing the initial capacity assessment under different levelsof electrolyte loading, the carbon electrodes underwent repeateddischarge/charge cycling experiments. Pouch-type LOBs wereemployed for these cycling tests, with cell capacities restricted to4mAh cm−2. To evaluate the impact of electrode pore volume andlean amount of electrolyte loading on the cycle life, the electrolytemass was meticulously adjusted to three different levels: ≈5, ≈7,and ≈9 g Ah−1. Figure 5a–f compares the galvanostatic voltageprofiles of cells with different electrodes under varying levels ofelectrolyte loading for both the 1st cycle and the specific cycle atwhich either the cell stopped functioning due to reaching the cut-off voltage or the discharge capacity dropped below 80% of the setcapacity (depicted as cell death). In Figure 5a–c, it is evident thatall carbon electrodes exhibited a consistent discharge voltage at≈2.65 V versus Li/Li+ in the 1st cycle. Nevertheless, there weredifferences in charge voltage profiles of the carbons dependingon the electrolyte loading.When 9 g Ah−1 electrolyte was employed, throughout mostof the charge, for all the electrodes, the voltage remained sta-ble at 3.5–3.6 V versus Li/Li+, which is more evident from themagnified image in Figure S19 (Supporting Information). How-ever, as the electrolyte amounts decreased, the charge voltagebecame higher for carbons with a larger pore diameter, as de-picted in Figures S20 and S21 (Supporting Information). Specif-ically, the differences in the charge voltage among different elec-trodes are quite apparent in the case of 5 g Ah−1 in Figure S21(Supporting Information). Figures S22–S26 (Supporting Infor-mation) further illustrate that the electrolyte loading amount hasminimal effect on the discharge/charge potential in the cases ofMPC-5 and MPC-10. However, for MPC-18, MPC-33, and MPC-38, the charge voltage is higher for a lower amount of electrolyte.These results indicate that for carbons with larger pore diameterand pore volume, the resistance is higher during charging undera lean amount of electrolyte 5 g Ah−1. N2 adsorption/desorptionanalysis showed that the carbon electrodes with a larger pore sizehave a lower surface area. Consequently, the interfacial area be-tween the electrode and electrolyte, specifically between Li2O2Adv. Sci. 2026, 13, e14406 e14406 (8 of 14) © 2025 The Author(s). Advanced Science published by Wiley-VCH GmbH 21983844, 2026, 1, Downloaded from https://advanced.onlinelibrary.wiley.com/doi/10.1002/advs.202514406 by Shoichi MATSUDA - Argonne National Laboratory , Wiley Online Library on [13/01/2026]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons Licensehttp://www.advancedsciencenews.comhttp://www.advancedscience.comwww.advancedsciencenews.com www.advancedscience.comFigure 5. First cycle discharge/charge potential profiles of different carbon electrodes with a limited capacity of 4 mAh cm−2 under a) 9 g Ah−1, b) 7 gAh−1, and c) 5 g Ah−1 electrolyte loadings. Discharge/charge potential profiles for the failed cycle of the same electrodes under d) 9 g Ah−1, e) 7 g Ah−1,and f) 5 g Ah−1 electrolyte loadings. Comparison of average discharge voltage over the cycles of different electrodes under g) 9 g Ah−1, h) 7 g Ah−1, andi) 5 g Ah−1 electrolyte loadings. j–l) Comparison of the cycle numbers of the electrodes with different electrolyte loadings.and the electrolyte (the reaction site for Li2O2 decomposition),is also decreased. This reduction in interfacial area leads to anincrease in effective current density. Moreover, as the amountof electrolyte decreases, the interfacial area further diminishes.This is why carbon electrodes with larger pore diameters andvolumes exhibit higher charging voltages under lean electrolyteconditions.More interesting results emerge regarding the failure modesof the cells depending on the electrolyte loading. Figure 5d showsthe voltage profiles of all the cells in 9 g Ah−1 electrolyte-loadingAdv. Sci. 2026, 13, e14406 e14406 (9 of 14) © 2025 The Author(s). Advanced Science published by Wiley-VCH GmbH 21983844, 2026, 1, Downloaded from https://advanced.onlinelibrary.wiley.com/doi/10.1002/advs.202514406 by Shoichi MATSUDA - Argonne National Laboratory , Wiley Online Library on [13/01/2026]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons Licensehttp://www.advancedsciencenews.comhttp://www.advancedscience.comwww.advancedsciencenews.com www.advancedscience.comfor the specific cycle where the cell died. While cells with MPC-5,MPC-10, MPC-18, and MPC-33 showed a gradual increase inboth discharge and charge voltage polarizations and capacitydecay, the cell with MPC-38 showed a sudden vertical spike incharge voltage, leading to cell failure. This sudden increase inthe charge voltage can be attributed to the loss of electrolytefrom the electrode pore and increased resistance in the cell.This observation becomes more obvious when the electrolyteloading amounts were decreased to 7 and 5 g Ah−1. In Figure 5e,it is observed that the cells with both the larger pore carbons,MPC-33 and MPC-38, failed due to a sudden increase in chargevoltage. However, the cells with relatively smaller pore carbonsshowed a gradual decay in capacity. When the electrolyte loadingwas minimum at 5 g Ah−1, except for the cell with the smallestpore carbon MPC-5, all other cells in Figure 5f showed a voltagespike during charge. Figure 5g–i presents the evolution ofaverage discharge voltage over cycling for cells incorporatingdifferent carbon electrodes under electrolyte loadings of 9, 7,and 5 g Ah−1. At the highest electrolyte loading (9 g Ah−1), theaverage discharge voltages remain relatively consistent acrossall electrode types, indicating sufficient ionic conductivity andelectrode filling. However, under leaner electrolyte conditions (7and 5 g Ah−1), a marked decrease in average discharge voltage isobserved for MPC-33 and MPC-38. Therefore, in the cases of thecells with carbons having a large pore diameter and pore volume,lean amounts of electrolyte impart higher voltage polarizationand lead to sudden cell failure due to large resistance originatingfrom the loss of electrolyte from the pores. The energy efficiencyof the cells, defined as the ratio of discharge energy to charge en-ergy, is presented in Figures S27–S29 (Supporting Information).A gradual decrease in energy efficiency is observed as the carbonpore diameter increases. Since the cycling was performed undera fixed capacity condition, these differences can be attributed tovariations in the average discharge and charge voltages. In par-ticular, the lower efficiency of the larger-pore carbons arises fromtheir lower average discharge potential and higher average chargepotential. The apparent sudden increase in energy efficiencyfor cells with the larger pore carbons results from incompletecharging caused by voltage spikes. The cycling performance dataof various carbon electrodes with different electrolyte loadingsare shown in Figure 5j–l. When the electrolyte loading wasmaintained at 9 and 7 g Ah−1, a general decreasing trend incycling stability was observed as the pore size of the carbonincreased: MPC-5 ≥ MPC-10, ≈ MPC-18 > MPC-33 > MPC-38.Nevertheless, the most substantial differences in cycling stabilitywere observed under the lowest electrolyte loading of ≈5 g Ah−1.While MPC-5 maintained stability for 38 cycles, MPC-10, MPC-18, MPC-33, and MPC-38 could only cycle for 13, 11, 10, and 7cycles, respectively. Considering the importance of both reducingthe electrolyte content and increasing the cyclable capacity forhigh energy density of the cells, cells were cycled between 2.0-4.5 V versus Li/Li+ with 100%, 80%, and 65% electrolyte loadingswithout any capacity limit. As shown in Figures S30–S32 (Sup-porting Information), capacity fade became progressively worseas the electrolyte content decreased, particularly for electrodeswith larger pores. In most cases, cell failure was triggered bysudden voltage spikes during charging. These observations indi-cate that electrolyte depletion-induced polarization, rather thangradual degradation of the electrode or electrolyte, is the primarycause of failure in LOB cells under lean-electrolyte conditions,and the situation becomes worse as the electrolyte displacementbecomes higher at higher cyclable capacity.2.7. Investigation of Cell Failure2.7.1. Impedance Analysis During CyclingEnhanced voltage polarization and sudden voltage spikes ob-served during charging cycles signify a progressive increasein cell impedance, ultimately leading to premature cell failure.To elucidate the origins of this impedance rise under lean-electrolyte conditions, electrochemical impedance spectroscopy(EIS) was conducted across different cycles. To minimize theinfluence of the Li electrode, a solid ceramic separator was usedtogether with a stable 4 m lithium bis(fluorosulfonyl)imide in1,2-dimethoxyethane (DME) electrolyte on the Li side, which isknown to sustain several hundred cycles with negligible degrada-tion compared to the positive electrode. Under these conditions,the major changes observed in Ohmic, interphasial, and charge-transfer resistances can be reasonably attributed to processes oc-curring at the positive electrode. The correspondingNyquist plotsand fitted equivalent circuits are provided in Figures S33–S48(Supporting Information). Figures S49a–c and S50a–c (Support-ing Information) show the evolution of Ohmic resistance (ROh)and charge-transfer resistance (RCT), respectively, measured atfull charge. At an electrolyte loading of 9 g Ah−1, all cells exhib-ited a gradual increase in ROh, except for MPC-38, which showeda steep rise starting around the 11th cycle, which is consistentwith the abrupt voltage behavior observed in Figure 5d. Underlower electrolyte loadings (7 and 5 g Ah−1), three cells withhigher-pore-volume carbons (MPC-18, MPC-33, and MPC-38)demonstrated a sudden increase in ROh, each marking the onsetof cell death. It is worth mentioning that it is the sudden increasein ROh, rather than its absolute value, that correlates with thevoltage spike in the respective cell. These results indicate thatOhmic resistance growth, driven by electrolyte depletion in high-pore-volume electrodes, is the dominant contributor to cell deathunder lean-electrolyte regimes. By contrast, RCT did not show adirect correlation with cycle life and instead fluctuated during cy-cling. This behavior is attributed to the dynamic deposition andpartial decomposition of insulating side products (e.g., lithiumcarbonate (Li2CO3) and lithium carboxylates), which transientlyalter interfacial charge transfer properties. While RCT generallytrends upward with cycle number, its variability suggests a morecomplex interfacial degradation mechanism, distinct from thebulk electrolyte-mediated failure reflected by ROh.2.7.2. Online Electrochemical Mass Spectrometry for Gas AnalysisSubsequently, we focused on investigating cell degradation un-der different conditions. In our recent analytical investigation, wehave indeed demonstrated enhanced parasitic reactions in LOBsunder lean electrolyte conditions. The core principle of LOBs in-volves the reversible electrochemical reduction and evolution ofO2 during discharge and charge cycles, respectively. As such, theassessment of O2 evolution serves as a standard measure to eval-uate the reversibility of LOBs. However, alongside the desired O2Adv. Sci. 2026, 13, e14406 e14406 (10 of 14) © 2025 The Author(s). Advanced Science published by Wiley-VCH GmbH 21983844, 2026, 1, Downloaded from https://advanced.onlinelibrary.wiley.com/doi/10.1002/advs.202514406 by Shoichi MATSUDA - Argonne National Laboratory , Wiley Online Library on [13/01/2026]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons Licensehttp://www.advancedsciencenews.comhttp://www.advancedscience.comwww.advancedsciencenews.com www.advancedscience.comFigure 6. Charging voltage profiles a,b), and the corresponding O2 c,d) and CO2 e,f) evolution rates of MPC-5 and MPC-10 for the first and tenth cycles,respectively. Comparison of the quantities of evolved g) O2 and h) CO2 for different electrodes in the first and 10th charge.evolution, the occurrence of parasitic side reactions from boththe electrode and electrolyte often results in the release of car-bon dioxide (CO2) gas in LOB systems. Therefore, quantifyingthe evolved gases offers a comprehensive insight into the reac-tions within the positive electrode and their impact on the cyclingstability of the cells.To quantify the evolved gases, we conducted an online electro-chemical mass spectrometry (online MS) analysis. For the onlineMS experiments, we utilized a specialized two-compartment celldesign, wherein the Li negative electrode is isolated from thepositive electrode using a glass ceramic separator. Figure S51(Supporting Information) shows a schematic design of the flow-type cell used for this MS analysis. This cell design ensures theexclusive estimation of evolved gases from the positive electrode.Figures S52–S61 (Supporting Information) present the voltageprofiles of LOBs using different electrodes, along with the corre-sponding rates of gas evolution for the 1st and 10th cycles. Theoutcomes depicted in Figure 6a,b compare the voltage profiles,and Figure 6c–f compare the gas evolution rates of the carbonswith the smallest and largest pore diameter, MPC-5 andMPC-38,as examples, for the 1st and 10th charge processes, respectively.Figure 6c–e indicate that during the 1st charge, there were no sig-nificant differences in the rates of O2 and CO2 evolution betweenMPC-5 and MPC-38. However, at the 10th charge, a substantialdecrease in O2 evolution rate is observed in Figure 6d and f forMPC-38 compared to MPC-5. Additionally, the charge voltagefor the cell with MPC-38 in the 10th charge (see Figure 6b) ishigher than that with MPC-5, potentially triggering enhancedAdv. Sci. 2026, 13, e14406 e14406 (11 of 14) © 2025 The Author(s). Advanced Science published by Wiley-VCH GmbH 21983844, 2026, 1, Downloaded from https://advanced.onlinelibrary.wiley.com/doi/10.1002/advs.202514406 by Shoichi MATSUDA - Argonne National Laboratory , Wiley Online Library on [13/01/2026]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons Licensehttp://www.advancedsciencenews.comhttp://www.advancedscience.comwww.advancedsciencenews.com www.advancedscience.comdecomposition of both the electrolyte and the electrode. Indeed,the cell with MPC-38 exhibited a much earlier onset of CO2evolution at ≈60% state of charge (SOC), compared to 75% SOCin the case of MPC-5. Figure 6e and f illustrate the quantitativeestimation of the total amounts of O2 and CO2 evolved duringthe 1st and 10th cycles for all samples. The results in Figure 6gindicate that during the 1st charge, all cells exhibited similaramounts of O2 evolution, with an estimated yield of 80±2% O2(see Figure S62, Supporting Information). However, notabledifferences in O2 evolution profiles emerged during the 10thcharge. A gradual decrease in O2 yield was observed as the porediameter of the carbon increased. For instance, the LOB withMPC-5 displayed an O2 evolution yield of 69% during the 10thcharge, decreasing to 56±1% for MPC-10 and MPC-18, and46±1% for MPC-33 and MPC-38. Consistent with the O2 evo-lution profiles, all cells exhibited similar CO2 evolution duringthe 1st charge, as shown in Figure 6h. However, a significantincrease in the quantity of evolved CO2 during the 10th chargeindicated enhanced parasitic reactions, particularly pronouncedfor electrodes with larger pore carbons, MPC-33 and MPC-38.This gas analysis provides compelling evidence of the superiorO2 reversibility and increased resistance to side reactions in pos-itive electrode materials with relatively smaller pore diametersunder lean electrolyte conditions. Depletion of electrolytes fromthe electrodes with larger pore carbons under these conditions,which worsens during cycling, leads to increased ROh and voltagepolarization, triggering enhanced electrolyte decomposition.2.7.3. Analysis of the Carbon Electrode Pore Structure During CyclingThe N2 adsorption/desorption isotherms and BJH pore size dis-tribution curves for all the carbon electrodes at various cyclestates are shown in Figures S63–S67 (Supporting Information).Figure S68a,b (Supporting Information) compare the BET sur-face area and pore volume, respectively, of all the electrodes intheir pristine state and after selected cycles. The results indicatethat, except for MPC-33 and MPC-38, there is a gradual decreasein both BET surface area and pore volume during cycling forthe other carbon electrodes. During prolonged cycling of LOBcells, electrolyte degradation leads to the formation and accumu-lation of solid lithium-containing byproducts that progressivelyobstruct the porous structure of the electrode. This effect is partic-ularly exacerbated in carbon electrodes with smaller pore diam-eters and limited pore volumes (e.g., MPC-5, MPC-10, MPC-18),where the confined pore space is more easily blocked. The highersurface area of these smaller-pore carbons exposed to the oxidiz-ing environments increases their susceptibility to parasitic reac-tions, further accelerating electrode passivation. Moreover, dueto more effective electrolyte filling, smaller-pore electrodes showrelatively lower ROh and reduced charge voltages, which are insuf-ficient to decompose accumulated side products, as reflected byminimal CO2 evolution. This results in progressive pore block-age over repeated cycles. Conversely, larger-pore electrodes suchas MPC-33 and MPC-38 exhibit higher charge voltages, whichfacilitate partial oxidative decomposition of the parasitic species,thereby preserving porosity to a greater extent. Nonetheless, theretention of open pore structure in these electrodes does not sig-nify improved reversibility or long-term stability, but rather indi-cates a continuous and more aggressive interfacial degradation.3. DiscussionsThe findings presented above highlight a critical trade-off be-tween electrode porosity, gravimetric energy density, and cyclingstability in LOBs. Electrodes with low pore volume, such asMPC-5, consistently exhibit low gravimetric energy density, irre-spective of the electrolyte-loading level, due to limited space forLi2O2 deposition. As the pore volume increases to intermediatelevels (MPC-10 and MPC-18), the gravimetric energy density im-proves significantly. However, further increases in pore volume(MPC-33 and MPC-38) result in a decline in gravimetric energydensity. This reduction is attributed to the greater electrolytemass required to saturate the expanded pore network, which off-sets the capacity gain and lowers the gravimetric energy density.This trend is consistent across varying electrolyte-loading levels.Although decreasing the electrolyte volume improves gravimet-ric energy density across all samples, MPC-10 and MPC-18 stillmaintain higher values. Besides, the highly porous electrodes(MPC-33 and MPC-38) suffer from electrolyte depletion, highimpedance growth, increased parasitic reactions, and shortenedcycle life under lean electrolyte conditions. Overall, MPC-10and MPC-18 represent an optimal balance between porosity andelectrolyte utilization, resulting in superior gravimetric energydensity and cycling stability. At 80% pore filling, they delivergravimetric energy densities exceeding 1200 Wh kg−1, and sus-tain stable cycling for over 30 cycles under the capacity limitingcondition of 4 mAh cm−2. These results suggest that tuning thepore volume and precisely controlling electrolyte loading are keydesign parameters for practical LOB electrodes. This interplay isqualitatively illustrated in Figure S69 (Supporting Information).4. ConclusionOur study underscores the critical importance of optimizingcarbon porosity to achieve high capacity while balancing theelectrolyte amount necessary for adequate electrode filling,thereby ensuring stable cycling of LOBs with high gravimetricenergy densities. Through simulation and experimental vali-dation, we found that although increasing electrode porositycorrelates with higher absolute capacity, it does not guaranteehigh specific capacity or gravimetric energy density due to thesignificant electrolyte volumes required to saturate the pores.Besides, the cycling stability tests under three different leanelectrolyte conditions revealed a declining trend in stability withreduced electrolyte content and increased carbon pore volume.As pore diameter and volume increased, cell failure occurreddue to insufficient electrode filling and electrolyte depletion. EISmeasurements during cycling showed that this depletion causedincreased cell impedance, voltage polarization, and electrolytedegradation. Conversely, for carbons with very small pores andlow pore volume, the issue of electrolyte depletion is less severe,but pore blockage and decreased porosity during cycling posesignificant challenges. Therefore, our findings emphasize theneed to optimize carbon electrode porosity to balance energydensity and cycle life in LOBs. Subsequently, a pore-optimizedAdv. Sci. 2026, 13, e14406 e14406 (12 of 14) © 2025 The Author(s). Advanced Science published by Wiley-VCH GmbH 21983844, 2026, 1, Downloaded from https://advanced.onlinelibrary.wiley.com/doi/10.1002/advs.202514406 by Shoichi MATSUDA - Argonne National Laboratory , Wiley Online Library on [13/01/2026]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons Licensehttp://www.advancedsciencenews.comhttp://www.advancedscience.comwww.advancedsciencenews.com www.advancedscience.comcarbon electrode showed a very high gravimetric energy densityexceeding 1500 Wh kg−1 and stable cycling under capacity-limiting conditions. These insights offer substantial promisefor advancing the development of carbon materials for the nextgeneration of practical LOBs, providing enhanced performanceand energy density.5. Experimental SectionSynthesis of Carbons: Mesoporous carbon structures were synthesizedusing a hard-template approach, where magnesium oxide (MgO) parti-cles of various sizes acted as the template, and the details of the synthesisprocedure can be found elsewhere.[37–39] MgO nanoparticles were syn-thesized, and the particle size was controlled by pyrolyzing magnesiumacetate (Mg(CH3COO)2) andmagnesium citrate (Mg3(C7H7O7)2) for dif-ferent durations. Phenol resin was employed as the carbon precursor. TheMgO and phenol resin mixture was thermally treated at 900 °C in a nitro-gen atmosphere to form a carbon–MgO composite. TheMgO componentwas subsequently removed by acid treatment with 1 m sulfuric acid, result-ing in the formation of mesoporous carbons. The removal of MgO wasconfirmed by XRF (X-ray Fluorescence) analysis, showing that the residualamount of MgO was 100 ppm or less. This carbon material underwent afinal high-temperature treatment at 1800 °C under nitrogen to improve itsstructural integrity.Fabrication of Self-Standing Carbon Membrane Electrode: A self-standing carbon gel-based membrane was developed by utilizing carbonpowder samples. The fabrication process involved the following steps:Fabrication of Self-Standing Carbon Membrane Electrode–Slurry Prepara-tion: The process starts with preparing a slurry consisting of 75 wt.%carbon powder, 5 wt.% single-walled carbon nanotube (OCSiAl, TUBALL,CNT, average fiber diameter 1.6 nm, average length 5 μm), 5 wt.% carbonfiber (Nippon Polymer Sangyo Co., Ltd., CF, average fiber diameter 6 μm,average length 3 mm), 15 wt.% PAN, and NMP as a solvent to ensureuniform dispersion.Fabrication of Self-Standing CarbonMembrane Electrode–Film Formation:The prepared slurry was then uniformly spread onto a sheet using a doctorblade technique. This method helps maintain a consistent film thicknessacross the entire surface.Fabrication of Self-Standing Carbon Membrane Electrode–Pore Genera-tion: Afterward, the sample underwent immersion in methanol (a poorsolvent) and was transformed into a porous film through the nonsolvent-induced phase separation (NIPS) method.Fabrication of Self-Standing Carbon Membrane Electrode–Drying and Sta-bilization of the Film: The resulting film underwent solvent removal bydrying at 80 °C for 10 h, followed by an infusibilization treatment at 230 °Cfor 3 h in an air-circulating atmosphere using an oven DN411 (YamatoScientific Co., Ltd.).Fabrication of Self-Standing Carbon Membrane Electrode–Carbonization:Carbonization was the final step that was conducted in a box-type furnace(Denken High Dental Co., Ltd.) under nitrogen flow (800 mL min−1) bygradually increasing the temperature to 1050 °C at a rate of 10 °C min−1,maintaining it at 1050 °C for 3 h, and then allowing the sample to cool toroom temperature.Characterization of Self-Standing Carbon Membrane Electrode: Thepore structures of the samples underwent characterization using nitrogenadsorption/desorption (3 FLEX, Micromeritics Instrument Corp.), whilemacropore size distributions were determined via mercury porosimetry(AutoporeIV 9505, Shimadzu Co.). X-ray diffraction (XRD) patterns of thecarbon powders were obtained using an X-ray diffractometer (SmartLab,Rigaku). Morphological analysis of the carbon powders was conducted us-ing a Field-emission scanning electron microscope (FE-SEM, S-4800, Hi-tachi). A laser Raman microscope (RamanTouch-VIS-NIR, Nanophoton)was employed to assess the graphitization of the carbons. Surface chemi-cal characterization of the carbon samples was carried out using X-ray pho-toelectron spectroscopy (ULVAC-PHI, VersaProbe II). The transmissionelectron micrographs (TEM) were obtained from a Jeol JEM-ARM200F.Lithium–Oxygen Cell Assembly andDischarge Performance Test: An elec-trolyte comprising of 0.5 m lithium bis(trifluoromethanesulfonyl)imide(LiTFSI; Kishida Chemical Co., Ltd., purity >99.9%), 0.5 m lithium nitrate(LiNO3; Sigma–Aldrich Co., LLC, purity 99.99% trace metals basis), and0.2 m lithium bromide (LiBr; Sigma–Aldrich Co., LLC, purity 99.995% tracemetals basis) dissolved in tetraethylene glycol dimethyl ether (TEGDME;Kishida Chemical Co., Ltd., purity >99%) was utilized for cell tests. BothLiNO3 and LiBr salts were dried under vacuum at 120 °C for over 3 daysbefore electrolyte preparation to ensure dryness. The water content of theelectrolyte was verified to be less than 30 ppm through Karl Fischer titra-tion. Self-standing carbon membranes were utilized as the positive elec-trodes after undergoing vacuum drying at 100 °C for 12 h. Pouch-typelithium–oxygen cells (2 × 2 cm2) were assembled in a dry room with a wa-ter content of less than 10 ppm. The cell assembly process involved stack-ingmetallic lithium foil (HonjoMetal Co., Ltd.), a polyolefin-based separa-tor, a carbon electrode, and a gas diffusion layer (TGP-H-060, Toray, Japan)sequentially. Electrolyte injection into carbon electrodeswas accomplishedusing the vacuum impregnation method, maintaining 3 different levels ofelectrolyte loading:≈65%, 80%, and 100% of the pore volumes of the elec-trodes. A pressure of 100 kPa was applied to the cell by a spring coil. Thelithium–oxygen cells were kept within an oxygen-filled box, with oxygen gascontinuously flowing at a rate of 80 mL min−1. Electrochemical dischargeexperiments were conducted using a battery test equipment (SD8, HokutoDenko Corp.). The cutoff voltage was set at 2.0 V versus Li/Li+ at a currentdensity of 0.4 mA cm−2.Lithium–Oxygen Cell Assembly and Discharge/Charge Cycling Test: Forthe cycling tests, the same ternary salt electrolyte was utilized. Lithium–oxygen cells were assembled inside a dry room where the water contentwas maintained below 10 ppm. The cell assembly comprised stacking alithium–metal foil (2 × 2 cm2, thickness of 0.1 mm; Honjo Metal Co.,Ltd.), a polyolefin-based separator (2.2 × 2.2 cm2, thickness of 0.02 mm),a porous carbon electrode (2 × 2 cm2), and a gas-diffusion layer. The gas-diffusion layer used in the cell consisted of an array of carbon fibers having≈10 μm diameter (thickness of 110 μm, TGP-H-030, Toray). Moreover, anarray of Ni/Cu-coated PET fibers ≈30 μm in diameter (thickness of 45 μm,Sui-40-9027, SEIREN Electronics Materials) served as the current collec-tor. A ceramic-based solid-state separator (LICGC, thickness of 0.50 mm;Ohara, Inc.) was used to protect the lithium–metal negative electrode.This ceramic-based solid-state separator was placed between polyolefin-based separators, and the same electrolyte was used for both the positiveand negative electrode sides. Electrolyte injection into carbon electrodeswas accomplished using the vacuum impregnation method, maintaining3 different levels of electrolyte loading: ≈5, 7, and 9 g Ah−1. A pressureof 100 kPa was applied to the cell by a spring coil. Electrochemical experi-ments were conducted using a battery test equipment (TOSCAT, Toyo Sys-tem Co., Ltd.). The limiting capacity and cutoff voltage were set at 4.0 mAhcm−2 and 2.0 V/4.5 V versus Li/Li+, respectively, with a current density of0.4 mA cm−2. The lithium–oxygen cells were kept inside an oxygen-filledbox, with oxygen gas continuously flowing at a rate of 80 mL min−1.Online Mass Spectroscopic (MS) Analysis: High-resolution mass spec-trometry (MS) analysis was conducted using an MS instrument (M-401GA, CANON ANELVA Corp.) configured in an online setup. An elec-trochemical flow cell, specifically designed for this purpose and with aninner volume of ≈24 mL (diameter = 70 mm; depth = 15 mm), was uti-lized for the analysis. This cell incorporated the same components asthose detailed for the lithium-oxygen cells mentioned earlier. The nega-tive electrode was separated from the positive electrode with a glass ce-ramic separator. While the same ternary-salt electrolyte was used for thepositive electrode side, a 4 m lithium bis(fluorosulfonyl)imide (LiFSI) indimethoxyethane (DME) electrolyte was used for the negative electrode.Prior to measurement, the cells were filled with oxygen (O2) gas and sub-jected to a discharge process until reaching a discharge capacity of 4 mAhcm−2. Subsequently, the test cell was purged with excess helium (He) gasat a flow rate of 50 mL min−1 for 1 min to remove any remaining O2. Dur-ing the charging process, gas evolution was continuously monitored usingHe as the carrier gas at a flow rate of 5 mL min−1. For online MS analysis,the generated gases were directly transferred to the MS detector througha capillary tube (internal diameter: 0.05 mm, length: 7 m).Adv. Sci. 2026, 13, e14406 e14406 (13 of 14) © 2025 The Author(s). Advanced Science published by Wiley-VCH GmbH 21983844, 2026, 1, Downloaded from https://advanced.onlinelibrary.wiley.com/doi/10.1002/advs.202514406 by Shoichi MATSUDA - Argonne National Laboratory , Wiley Online Library on [13/01/2026]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons Licensehttp://www.advancedsciencenews.comhttp://www.advancedscience.comwww.advancedsciencenews.com www.advancedscience.comThe reproducibility of the results was verified by performing all experi-ments in at least two independent runs.Supporting InformationSupporting Information is available from the Wiley Online Library or fromthe author.AcknowledgementsThe present work was partially supported by JSPS KAKENHI (Grant Num-ber 24K08590). This work also received support from theNational Institutefor Materials Science (NIMS) Battery Research Platform.Conflict of InterestThe authors declare no conflict of interest.Data Availability StatementThe data that support the findings of this study are available from the cor-responding author upon reasonable request.Keywordscarbon electrode, high energy battery, lean electrolyte, Lithium-oxygen bat-tery, porosity optimizationReceived: July 29, 2025Revised: September 29, 2025Published online: October 15, 2025[1] K. Song, D. A. Agyeman, M. Park, J. Yang, Y.-M. Kang, Adv. Mater.2017, 29, 1606572.[2] W.-J. Kwak, Rosy, D. S., C. Xia, H. Kim, L. R. Johnson, P. G. Bruce,L. F. Nazar, Y.-K. Sun, A. A. Frimer, M. Noked, S. A. Freunberger, D.Aurbach, Chem. Rev. 2020, 120, 6626.[3] W.-J. Kong, C.-Z. Zhao, S. Sun, L. Shen, X.-Y. Huang, P. Xu, Y. Lu, W.-Z.Huang, J.-Q. Huang, Q. Zhang, Adv. Mater. 2024, 36, 2310738.[4] A. A. Franco, K.-H. Xue, ECS J. Solid State Sci. Technol. 2013, 2, M3084.[5] M.-K. Song, S. Park, F. M. Alamgir, J. Cho, M. Liu, Mater. Sci. Eng. RRep. 2011, 72, 203.[6] O. L. Li, T. Ishizaki, in Emerging Materials for Energy Conversion andStorage, (Eds.: K. Y. Cheong, G. Impellizzeri, M. A. Fraga), Elsevier,New York 2018, pp. 115–152.[7] M. D.Womble, K. R.McKenzie,M. J. Wagner, Sci. Rep. 2025, 15, 5868.[8] S. Meini, M. Piana, H. Beyer, J. Schwämmlein, H. A. Gasteiger, J. Elec-trochem. Soc. 2012, 159, A2135.[9] N. Ding, S. W. Chien, T. S. A. Hor, R. Lum, Y. Zong, Z. Liu, J. Mater.Chem. A 2014, 2, 12433.[10] T. Kuboki, T. Okuyama, T. Ohsaki, N. Takami, J. Power Sources 2005,146, 766.[11] S. B. Ma, D. J. Lee, V. Roev, D. Im, S.-G. Doo, J. Power Sources 2013,244, 494.[12] S. R. Younesi, S. Urbonaite, F. Björefors, K. Edström, J. Power Sources2011, 196, 9835.[13] M. Kim, E. Yoo, W.-S. Ahn, S. E. Shim, J. Power Sources 2018, 389, 20.[14] K. Sakai, S. Iwamura, S. R. Mukai, J. Electrochem. Soc. 2017, 164,A3075.[15] B. Sun, S. Chen, H. Liu, G. Wang, Adv. Funct. Mater. 2015, 25, 4436.[16] Z.-L. Wang, D. Xu, J.-J. Xu, L.-L. Zhang, X.-B. Zhang, Adv. Funct. Mater.2012, 22, 3699.[17] H. Wang, X. Wang, M. Li, L. Zheng, D. Guan, X. Huang, J. Xu, J. Yu,Adv. Mater. 2020, 32, 2002559.[18] A. Dutta, K. Ito, Y. Kubo,Mater. Adv. 2021, 2, 1302.[19] W. Yu, T. Yoshii, A. Aziz, R. Tang, Z.-Z. Pan, K. Inoue, M. Kotani,H. Tanaka, E. Scholtzová, D. Tunega, Y. Nishina, K. Nishioka, S.Nakanishi, Y. Zhou, O. Terasaki, H. Nishihara, Adv. Sci. 2023, 10,2300268.[20] S. Matsuda, E. Yasukawa, T. Kameda, S. Kimura, S. Yamaguchi, Y.Kubo, K. Uosaki, Cell. Rep. Phys. Sci. 2021, 2, 100506.[21] A. Dutta, K. Ito, Y. Kubo, J. Mater. Chem. A 2019, 7, 23199.[22] W. Yu, Z. Shen, T. Yoshii, S. Iwamura, M. Ono, S. Matsuda, M. Aoki,T. Kondo, S. R. Mukai, S. Nakanishi, H. Nishihara, Adv. Energy Mater.2024, 14, 2303055.[23] J. Saengkaew, T. Kameda, M. Ono, S. Matsuda, Mater. Adv. 2022, 3,3536.[24] A. Dutta, K. Ito, A. Nomura, Y. Kubo, Adv. Sci. 2020, 7, 2001660.[25] X. Zheng, W. Lv, Y. Tao, J. Shao, C. Zhang, D. Liu, J. Luo, D.-W. Wang,Q.-H. Yang, Chem. Mater. 2014, 26, 6896.[26] V. S. Babu, M. S. Seehra, Carbon 1996, 34, 1259.[27] K.-V. Peinemann, V. Abetz, P. F. W. Simon, Nat. Mater. 2007, 6, 992.[28] A. Dutta, K. Matsushita, Y. Kubo, Adv. Sci. 2024, 11, 2404245.[29] C. M. Burke, V. Pande, A. Khetan, V. Viswanathan, B. D. McCloskey,Proc. Nat. Acad. Sci. 2015, 112, 9293.[30] A. Dutta, Y. Kubo, A. Nagataki, K. Matsushita, ACS Appl. Mater. Inter-faces 2023, 15, 15467.[31] K. Shi, A. Dutta, Y. Hao, M. Zhu, L. He, Y. Pan, X. Xin, L.-F. Huang, X.Yao, J. Wu, Adv. Funct. Mater. 2022, 32, 2203652.[32] X. Xin, K. Ito, Y. Kubo, ACS Appl. Mater. Interfaces 2017, 9, 25976.[33] W. Chen, W. Yin, Y. Shen, Z. Huang, X. Li, F. Wang, W. Zhang, Z. Deng,Z. Zhang, Y. Huang, Nano Energy 2018, 47, 353.[34] H. C. Lee, J. O. Park, M. Kim, H. J. Kwon, J.-H. Kim, K. H. Choi, K.Kim, D. Im, Joule 2019, 3, 542.[35] Y. J. Lee, S. H. Park, S. H. Kim, Y. Ko, K. Kang, Y. J. Lee, ACS Catal.2018, 8, 2923.[36] S. Zhao, L. Zhang, G. Zhang, H. Sun, J. Yang, S. Lu, J. Energy Chem.2020, 45, 74.[37] T. Morishita, T. Tsumura, M. Toyoda, J. Przepiórski, A. W. Morawski,H. Konno, M. Inagaki, Carbon 2010, 48, 2690.[38] T. Morishita, Y. Soneda, T. Tsumura, M. Inagaki, Carbon 2006, 44,2360.[39] M. Inagaki, M. Toyoda, Y. Soneda, S. Tsujimura, T. Morishita, Carbon2016, 107, 448.Adv. Sci. 2026, 13, e14406 e14406 (14 of 14) © 2025 The Author(s). Advanced Science published by Wiley-VCH GmbH 21983844, 2026, 1, Downloaded from https://advanced.onlinelibrary.wiley.com/doi/10.1002/advs.202514406 by Shoichi MATSUDA - Argonne National Laboratory , Wiley Online Library on [13/01/2026]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons Licensehttp://www.advancedsciencenews.comhttp://www.advancedscience.com Quantitative Porosity Engineering of Carbon Electrode in Lithium9040�Oxygen Batteries with Cell-Level Gravimetric Energy Density Over 1500Wh kg9042�1 1. Introduction 2. Results 2.1. Simulation of Gravimetric Energy Density Variation with Pore Volume 2.2. Synthesis of MgO Templated Porous Carbon Powders 2.3. Physico-Chemical Characterization of the Carbon Materials 2.4. Fabrication of Self-Standing Porous Carbon Membrane 2.5. Analysis of the Porosity of the Self-Standing Carbon Membrane 2.6. Application of Free-Standing Membrane Electrodes in LOBs 2.6.1. Discharge Capacity Estimation of the Carbon Electrodes 2.6.2. Discharge/Charge Cycling Test 2.7. Investigation of Cell Failure 2.7.1. Impedance Analysis During Cycling 2.7.2. Online Electrochemical Mass Spectrometry for Gas Analysis 2.7.3. Analysis of the Carbon Electrode Pore Structure During Cycling 3. Discussions 4. Conclusion 5. Experimental Section Supporting Information Acknowledgements Conflict of Interest Data Availability Statement Keywords