# Fileset

[A data-driven elucidation of the challenges in designing stable quinone derivatives for aqueous organic redox flow batteries  correlations between the.pdf](https://mdr.nims.go.jp/filesets/814694d4-b2d6-433c-8d50-8c02ff6ec73a/download)

## Creator

Ning Lu, Shiyao Ju, Daniel Willimetz, Shengming Tang, Abhishek Khetan

## Rights

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

## Other metadata

[A data-driven elucidation of the challenges in designing stable quinone derivatives for aqueous organic redox flow batteries: correlations between thermodynamics of chemical degradation and energy capacity metrics](https://mdr.nims.go.jp/datasets/194eeebf-865e-4afd-bcc5-f96ab57019ca)

## Fulltext

Microsoft Word - AMO_TSTA_A_2698227.docxScience and Technology of Advanced MaterialsISSN: 1468-6996 (Print) 1878-5514 (Online) Journal homepage: www.tandfonline.com/journals/tsta20A data-driven elucidation of the challengesin designing stable quinone derivativesfor aqueous organic redox flow batteries:correlations between thermodynamics ofchemical degradation and energy capacitymetricsNing Lu, Shiyao Ju, Daniel Willimetz, Shengming Tang & Abhishek KhetanTo cite this article: Ning Lu, Shiyao Ju, Daniel Willimetz, Shengming Tang & Abhishek Khetan(16 Jul 2026): A data-driven elucidation of the challenges in designing stable quinonederivatives for aqueous organic redox flow batteries: correlations between thermodynamicsof chemical degradation and energy capacity metrics, Science and Technology of AdvancedMaterials, DOI: 10.1080/14686996.2026.2698227To link to this article:  https://doi.org/10.1080/14686996.2026.2698227© 2026 The Author(s). Published by NationalInstitute for Materials Science in partnershipwith Taylor & Francis Group.View supplementary material Accepted author version posted online: 16Jul 2026.Submit your article to this journal View related articles View Crossmark dataFull Terms & Conditions of access and use can be found athttps://www.tandfonline.com/action/journalInformation?journalCode=tsta20https://www.tandfonline.com/journals/tsta20?src=pdfhttps://www.tandfonline.com/action/showCitFormats?doi=10.1080/14686996.2026.2698227https://doi.org/10.1080/14686996.2026.2698227https://www.tandfonline.com/doi/suppl/10.1080/14686996.2026.2698227https://www.tandfonline.com/doi/suppl/10.1080/14686996.2026.2698227https://www.tandfonline.com/action/authorSubmission?journalCode=tsta20&show=instructions&src=pdfhttps://www.tandfonline.com/action/authorSubmission?journalCode=tsta20&show=instructions&src=pdfhttps://www.tandfonline.com/doi/mlt/10.1080/14686996.2026.2698227?src=pdfhttps://www.tandfonline.com/doi/mlt/10.1080/14686996.2026.2698227?src=pdfhttp://crossmark.crossref.org/dialog/?doi=10.1080/14686996.2026.2698227&domain=pdf&date_stamp=16%20Jul%202026http://crossmark.crossref.org/dialog/?doi=10.1080/14686996.2026.2698227&domain=pdf&date_stamp=16%20Jul%202026https://www.tandfonline.com/action/journalInformation?journalCode=tsta20Publisher: Taylor & Francis & The Author(s). Published by National Institute for Materials Science in partnership with Taylor & Francis Group. Journal: Science and Technology of Advanced Materials DOI: 10.1080/14686996.2026.2698227 A data-driven elucidation of the challenges in designing stable qui-none derivatives for aqueous organic redox flow batteries: correla-tions between thermodynamics of chemical degradation and energy capacity metrics Ning Lua, Shiyao Jua, Daniel Willimetza,b, Shengming Tanga, Abhishek Khetana*  aMODES, The Integrated Fuel & Chemical Science Centre, RWTH Aachen Universi-ty, 52062 Aachen, Germany bFraunhofer Institute for Algorithms and Scientific Computing SCAI, Fraunhofer-Gesellschaft, Schloss Birlinghoven 1, 53757 Sankt Augustin, Germany *Corresponding author: Abhishek Khetan; Email: askhetan@modes.rwth-aachen.de    ACCEPTED MANUSCRIPThttps://crossmark.crossref.org/dialog/?doi=10.1080/14686996.2026.2698227&domain=pdfAbstract: Aqueous organic redox flow batteries (AORFBs) offer significant potential for grid-scale storage of renewable energy. However, their commer-cial viability is often limited by the chemical instability of the organic electro-lytes. Quantifying and assessing the effects of the various degradation mecha-nisms from a molecular design perspective is very challenging both experimen-tally and computationally. Here, we propose the use of simple thermodynamic descriptors for seven degradation mechanisms for a diverse virtual library of ca. 2000 monofunctionalized quinones built from seven core structures. In the process, we developed a cheminformatics-based workflow that can reliably and automatically generate reactants and mechanism-specific degradation products. Using DFT-calculated reaction Gibbs free energies as thermodynamic de-scriptors and a calibrated model for redox-potential prediction, we systemati-cally analyse seven degradation mechanisms across this library. We find clear relationships between redox potential and degradation thermodynamics for six of the seven mechanisms, with opposite trend directions for oxidized and re-duced forms, revealing stability–potential trade-offs that constrain molecular design. The analysis further shows how functional groups modulate degrada-tion thermodynamics, helps rationalize the relative stability of anthraquinone-based outliers. Finally, redox active molecules from recent experimental stud-ies are evaluated within the proposed thermodynamics-based framework, and we comment on the implications of electrochemical reversibility of some criti-cal degradation mechanisms. Overall, this work provides a physically motivat-ed framework for the multi-objective screening of quinone-based AORFB elec-trolytes and helps clarify the design criteria needed to identify favourable mo-lecular outliers. Keywords: aqueous organic redox flow batteries; quinone stability; degrada-tion; molecular design; high-throughput screening; durability; grid-scale ener-gy storage Impact Statement: This work comprehensively analyses major quinone degradation mechanisms, reveals pathway-specific stability trade-offs, and identifies structural distortion and reversibility as critical yet previously overlooked stability descriptors.  ACCEPTED MANUSCRIPT1 Introduction Redox flow batteries (RFBs) offer a promising and scalable solution for storing large amounts of electrical energy.[1-3] Their distinctive design allows the independent scaling of power and energy by adjusting the electrode active surface area and the concentration of redox-active materials (RAMs) in the electrolyte.[4-6]  While vana-dium redox flow batteries lead the market in terms of share,[7]  Their limited natural availability and high cost of vanadium restrict their scalability and long-term viabil-ity.[4,8,9] As a result, the next-generation RFBs need to be developed that can operate durably for decades, offer high round-trip efficiency, respond rapidly to grid fluctua-tions, and are scalable for sustainable large-scale energy storage. Aqueous organic redox flow batteries (AORFBs) present a greener and more cost-effective alternative to metal-ion-based redox flow batteries for intermittent en-ergy storage.[10] [4,11,12] The RAMs in AORFBs are organic molecules dissolved in aqueous electrolytes that ideally undergo reversible redox reactions at the electrodes during charge and discharge. For practical deployment in AORFBs, RAMs must sim-ultaneously exhibit suitable redox potentials, high chemical and electrochemical sta-bility,[13,14] sufficient aqueous solubility,[15-17] and low cost through feasible and scalable synthesis.[18] Among various candidates, quinones are some of the most ex-tensively studied organic RAMs for AORFBs due to their tuneable redox potentials, favourable aqueous solubility, and fast redox kinetics.[19-22]  The aromatic π-systems embedded in quinone structures contribute to their relatively high intrinsic stability and reversible electrochemical behaviour by providing delocalization of the radicals formed during redox processes.[19]  ACCEPTED MANUSCRIPTFrom the point of view of commercial application, however, quinone-based molecules still exhibit insufficient stability.[23,24] They are prone to decomposition under aqueous conditions via various mechanisms, such as Michael addition,[25-27] gem-diol formation,[26,28,29] anthrone formation,[30-32] tautomerization,[33-35] disproportionation,[36,37] or dimerization,[29,33] thereby compromising their long-term electrochemical performance. The degradation of redox-active molecules can result in irreversible capacity loss, posing a significant challenge to their commercial application.[25] As a result, enhancing the stability of AORFB electrolytes has be-come an increasingly important focus in the field.[7,19,20,25,38-40] Over the past few years, extensive experimental investigations have deepened our understanding of the stability of quinone-based electrolytes in AORFBs.[30,39]  Many efforts have focused on inhibiting degradation reactions,[40-43] and achieving stable cycling under operational conditions.[44] Substitution strategies such as phos-phonate-functionalization,[8] sulfonation,[45] and amino-acid zwitterionic modifica-tion,[39] have been employed to suppress side reactions. Structural strategies such as π-conjugation extension, [19] asymmetric substitution, and branching [41] have prov-en effective in increasing redox reversibility and molecular robustness.  Notably, full substitution of reactive positions on hydroquinone rings has been shown to signifi-cantly mitigate Michael addition reactions, leading to decay rates as low as 0.45%/day in symmetric cell tests.[46] In alkaline media, particularly with anthraquinone deriva-tives, stable cycling over more than 1000 cycles has been achieved, with capacity fade rates below 0.01% per cycle.[30] Collectively, these experimental advancements in-spire molecular design of stable quinone-hydroquinone pairs, thereby reducing the gap between laboratory demonstrations and practical, scalable AORFB applications.[28] ACCEPTED MANUSCRIPTRecent computational studies have provided detailed insights into the stability of quinone-based molecules in AORFBs.[47,48] These investigations have elucidated how specific functional groups and molecular structures influence several degradation reactions. For example, tertiary ammonium groups in biphenol derivatives can signifi-cantly improve stability by inhibiting Michael addition reactions,[48] while sulfonic acid groups can promote degradation in acidic media, and methyl groups can enhance chemical inertness.[47] The position of functional groups on the quinone core struc-ture (ortho vs. para) further modulates stability, highlighting the critical role of mo-lecular architecture in determining susceptibility to tautomerization,[34] Michael ad-dition,[25] and gem-diol formation.[28] Collectively, these studies demonstrate that functionalization strategies and core structure design can be leveraged to improve the chemical stability of quinone-based redox-active molecules. While several large-scale computational efforts have explored subsets of qui-none derivatives, [28,33,49] most studies have focused on individual degradation mechanisms or specific molecular core structures. For example, Tabor et al. created a large database of quinone-based structures and analysed their stability toward Michael addition and gem-diol formation, finding that very few molecules simultaneously achieve a high reduction potential and thermodynamic protection against these degra-dation pathways.[28] Similarly, Tong et al. demonstrated that dimerization and tai-lored functional groups can enhance stability by mitigating tautomerization.[33] These studies highlight the importance of both core-structure selection and functional-ization in improving quinone robustness; however, a comprehensive virtual library spanning diverse quinone structures and multiple degradation pathways is still lacking. In this work, we construct a comprehensive virtual library of 1,956 quinone-derived molecules by systematically generating reactants and their corresponding deg-ACCEPTED MANUSCRIPTradation products using a combined RDKit-[50] and SMIRKS-[51] based approach, enabling consistent coverage of different core structures, functionalization, and deg-radation pathways reported in the literature. Building on this library, we provide a thermodynamic description of quinone degradation by calculating the Gibbs free en-ergy of reaction for each mechanism, which serves as a simple and physically moti-vated descriptor of a molecule’s susceptibility to degradation. We then investigate the relationship between redox potential and stability across several degradation pathways to identify structure-property trends, which have not yet been reported and that must be considered in the rational design of molecules. We also discuss the absence of a relationship between stability and solubility. Finally, we rationalize our findings in the context of experimentally extracted metrics of stability, including an analysis of outli-ers, and propose new degrees of freedom for exploring the design space.  2 Methods 2.1 Generation of initial structures To investigate the electrochemical properties of candidate quinones, we constructed a virtual molecular library starting from seven quinone core structures derived from ar-omatic backbones (benzene, naphthalene, and anthracene) (Fig. 1a). The core struc-tures are labelled 1-7 (Fig. 1b). They are generated by installing carbonyl (C=O) groups at different combinations of substitution sites on the corresponding backbones (Fig. 1a). For example, core structure 1 is formed by placing carbonyl groups at sites 1 and 4 of the benzene backbone, and the remaining core structures are defined analo-gously. Based on the number of carbonyl pairs, the core structures can be divided into two classes: single-pair (SO) core structures (1, 2, 4, and 5), which contain one car-bonyl pair, and double-pair (DO) core structures (3, 6, and 7), which contain two car-bonyl pairs.ACCEPTED MANUSCRIPTIt is well established that both the type[52] and position[53] of functional groups can substantially affect the redox thermodynamics and kinetics of quinones, including functional group‐dependent shifts in redox potential[15,54], and changes in proton‐coupled electron transfer(PCET) behaviour in aqueous media. To build a chemically tractable yet electrochemically relevant screening library for aqueous qui-none-based redox electrolytes, we systematically functionalized each core structure with five representative functional groups (-SO₃H, -COOH, -NH₂, -OH, and -Cl) to span a broad range of electronic effects (electron-withdrawing vs. electron-donating) and battery-relevant physicochemical properties, notably ionizability, hydrogen-bonding capability, and enhanced aqueous solubility. In this set, -SO₃H and -COOH serve as strongly solubilizing/ionizable electron-withdrawing groups that are com-monly used to mitigate precipitation and enable high active-material concentrations, whereas -NH₂ and -OH provide electron-donating and hydrogen-bonding functionali-ty that can tune redox potential and intermolecular interactions in aqueous electro-lytes. Chlorine was selected as a small, non-ionizable electron-withdrawing functional group and as a practical representative of halogen substitution (synthetically accessi-ble and electronically well-defined). An illustrative example of the functionalization workflow for anthraquinone is shown in Fig. 2. Functionalized derivatives were ex-haustively enumerated using RDKit[50,55,56] (via Python[57]).  In total, the seven quinone core structures yield 87 functionalized reactant quinones. In our virtual li-brary, each entry is defined by its structural representation, i.e. a SMILES string. We then compiled several degradation mechanisms from the literature (Table 1) to enumerate degradation products alongside the main redox reaction.[49] In acidic ACCEPTED MANUSCRIPTaqueous solutions, quinones typically undergo two-electron, proton-coupled redox reactions in redox flow batteries.[58] The main electrochemical reaction is listed in the first row of Table 1. The degradation mechanisms can differ depending on whether the quinone is in its oxidized or reduced state. For the oxidized (quinone) form, reported pathways include Michael addition,[26,30,59] gem-diol formation,[26,28] proto-desulfonation[60], and anthrone formation.[30,40] Pathways involving the reduced (quinol) form include tautomerization[46], disproportionation[40,61], and dimerization.[35,36,62] 2.2 Extension of the virtual library using reaction-based enumeration The next step is to extend the virtual library to enumerate the SMILES strings of all products from the degradation reactions shown in Table 1, and this step comprises two parts: (i) designing SMIRKS⁶² reaction templates to encode the transformation rules, and (ii) implementing a SMIRKS-based enumeration algorithm to generate product SMILES along with exception handling for incorrect transformations. To represent each reaction type with a manageable number of SMIRKS tem-plates, we restricted the design to the minimal transforming substructures-i.e., only the atoms and bonds that change during the reaction-while leaving nonreacting parts unspecified. For the main electrochemical reaction, five distinct SMIRKS templates were created, as shown in the first column of Table 2. The second column of Table 2 illustrates how these five SMIRKS templates match quinone compounds with differ-ent core structures. The key difference among these templates is whether the reacting carbon atoms are aromatic. Note that in the corresponding RDKit depictions, aroma-ticity is explicitly reflected in the drawing style: aromatic atoms/bonds are rendered with aromatic bond representations (e.g., dashed/alternating bonds), whereas non-ACCEPTED MANUSCRIPTaromatic (Kekulé) structures are shown with explicit single and double bonds. When an input reactant matches a given SMIRKS template, all possible products consistent with that template are enumerated. This SMIRKS design strategy, which distinguishes aromatic from non-aromatic carbons, effectively captures quinone redox reactions and applies to other reaction types as well.  In total, we designed 35 SMIRKS templates to represent the eight reaction types considered in this work, comprising the seven deg-radation reactions and the main electrochemical redox reaction listed in Table 1. The additional SMIRKS templates corresponding to the degradation reactions are provid-ed in the Supplementary Information (Table S1). Next, the reactant SMILES are processed by an RDKit-based reactor module, which applies the relevant SMIRKS templates to enumerate all possible products. The resulting structures are then filtered by removing duplicates and discarding invalid molecules that fail RDKit sanitization (e.g., valence, aromaticity, and charge/valence-state consistency checks). Finally, the output SMILES are organized by reaction type and exported to separate CSV files. In summary, using this procedure, we first systematically enumerated 113 functionalized quinols from the 87 quinones. After further enumeration using degra-dation reactions associated with either quinones or quinols, the virtual library com-prises 1,956 unique structures. This set provides a reliable and comprehensive founda-tion for subsequent computational analyses and mechanistic investigations. 2.3. Computational methods All electronic structure calculations were performed using the ORCA (v5.0.3) soft-ware.[63] Geometry optimizations were carried out using the B3LYP exchange-correlation functional[64] and the 6-311++G(d,p) basis set, with implicit solvation ACCEPTED MANUSCRIPTtreated using the conductor-like polarizable continuum model (CPCM) for water. [65-67] Optimization convergence was achieved when the maximum force was be-low3×10-4 Hartree/Bohr, and the root mean square force was below 1×10-4 Har-tree/Bohr. Gibbs free energies were also obtained under CPCM by including entropic corrections and zero-point energy contributions. We estimated the aqueous solubility of the reactant molecules using AqSolPred v1.0,[68] a state-of-the-art consensus ML model trained on AqSolDB, a large open-access curated dataset of experimentally measured aqueous solubilities. [69] For correlating degradation thermodynamics with aqueous solubility, we used either the reduced- or oxidized-state of the molecule, depending on which state the degradation was evaluated. Aqueous solubility was estimated using AqSolPred v1.0,[68] a consensus ML model trained on AqSolDB, which is a curated open-access dataset of experimentally measured aqueous solubilities.[69] This ML-predicted solu-bility is reported in log 𝑆 units, i.e., the base-10 logarithm of the molar aqueous solu-bility, at 25oC. AqSolPred is one of the state-of-the-art models for predicting aqueous solubilities of drug-like molecules with a log 𝑆 error < 0.4 units.  It simply uses the SMILES representations of the molecules for a direct prediction of solubility. In addition to log 𝑆, the CPCM-based solvation free energy,[66] ∆𝐺 , was used as an approximate descriptor representing the aqueous solvation contribution to the solubility. ∆𝐺  was calculated as the difference between the Gibbs free energy obtained with CPCM water solvation and the corresponding gas-phase Gibbs free en-ergy: ∆𝐺 = 𝐺 − 𝐺                                  (1) In their current usage, neither log 𝑆  nor ∆𝐺  explicitly accounts for pH-dependent speciation, supporting-electrolyte composition, salt formation, concentra-ACCEPTED MANUSCRIPTtion-dependent aggregation, or other condensed-phase effects that may influence ex-perimentally measured solubility in AORFB electrolytes. Therefore, both quantities should be interpreted as descriptors for high-level screening rather than a direct pre-diction of experimentally validated solubilities under specific electrolyte formulations. The redox potential is defined as the reaction Gibbs free energy change per unit charge transferred.[49,70,71]  The calculated reaction energies are often calibrat-ed empirically against experimental redox potentials using a simplified linear relation-ship (Eq. 2).𝐸   =   𝑚 𝐺 𝑄  −  𝐺 𝑄𝐻 − 𝐺 𝐻 + 𝑐        (2) In this equation, m and c denote the slope and intercept of the linear calibra-tion model, respectively, and 𝐺  represents the Gibbs free energy.[71]  Therefore, an empirical linear calibration was performed between the experimental redox poten-tials of 25 quinone couples (Supporting Information Table S2) and their calculated reaction Gibbs free energies. [21,25,28,72] The resulting calibration equation for pre-dicting redox potentials versus SHE at pH = 0 is given in Eq. (3) and shows excellent agreement with experiment (R² = 0.98; MAE = 33mV; Supporting Information Fig. S1).                𝐸 =  −0.48 ∆𝐺 +  0.18                                 (3) Here 𝐸° (V) represents the redox potential, and ∆𝐺 (eV) is defined as the Gibbs free energy difference between the products and reactants of the electrochemical reac-tion. In our study, we use ∆𝐺 , defined as the Gibbs free energy difference between the products and reactants of different degradation mechanisms, as a thermo-ACCEPTED MANUSCRIPTdynamic descriptor of degradation susceptibility in aqueous solution. Accordingly, for a given degradation reaction, a more negative  ∆𝐺  indicates a stronger thermo-dynamic driving force toward instability, implying a higher propensity for decomposi-tion.[73] Although  ∆𝐺  does not capture kinetic barriers or reaction rates, it offers a physically motivated and internally consistent metric for comparing the in-trinsic thermodynamic tendency of different molecules to degrade. 3 Results and DiscussionWe first investigated the relationship between redox potential and thermodynamic stability for the various degradation mechanisms, as shown in Fig. 3. In these plots, Δ𝐺reaction_type denotes the thermodynamic stability descriptor for a specific reaction, as defined in Table 1. The various molecules in the dataset are colour-coded according to the type of functional group and shape-coded according to the number of rings. Fur-thermore, filled symbols represent molecules with a single pair of C=O groups, whereas molecules with two pairs of C=O groups. Figs. 3a (Δ𝐺Mic), 3b (Δ𝐺Gem), 3c (Δ𝐺Ant) show the correlations for the oxidized (quinone) forms, and Figs. 3d (Δ𝐺Tau), 3e (Δ𝐺Dim), 3f (Δ𝐺Dis) show those for the reduced (quinol) forms. First, it can be observed that, for every mechanism, there is a very clear corre-lation between Δ𝐺reaction_type and 𝐸°. This correlation indicates that, for planar mole-cules with single functional groups, such as those considered in our dataset, the elec-tron-donating/withdrawing character has a decisive influence on thermodynamic sta-bility, much like it does on redox potential. It is also clearly observable that molecules with two pairs of C=O groups have higher redox potentials than those with a single pair (Fig. S2). [28,74] However, the quinone (oxidized) and quinol (reduced) forms ACCEPTED MANUSCRIPTexhibit opposite trends. For Michael addition, gem-diol formation, and anthrone for-mation, a more positive Δ𝐺reaction_type (i.e., greater thermodynamic resistance to degra-dation) is generally associated with a lower redox potential, indicating a favourable design space for anolytes. This trend is consistent with Tabor et al.’s analysis of Mi-chael addition degradation in quinones,[28] which links higher 𝐸∘to increased elec-trophilicity of the quinone core, thus resulting in greater thermodynamic susceptibility to nucleophilic attack. In contrast, tautomerization, disproportionation, and dimerization show the re-verse behaviour, with lower redox potentials corresponding to more negative Δ𝐺reaction_type, indicating a trade-off that must be balanced when designing anolytes. In the case of tautomerization, the positive correlation is consistent with the study by Tong et al.[33] on reduced naphthoquinones, in which compounds with lower reduc-tion potentials were found to be more prone to tautomerization. Notably, we also find a small subset of systematic outliers that deviate from the dominant trend, motivating a closer structure-based analysis in the following section. For protodesulfonation, no clear correlation between redox potential and Δ𝐺reaction type is observed (Supporting Information Fig. S3). A likely reason is that this reaction is governed primarily by the local electronic environment of the sulfonic acid group rather than by the overall electrophilicity of the quinone core. Therefore, this mechanism is not discussed further. Overall, these observations indicate that both redox potential and degradation thermodynamics are strongly dependent on electron density changes induced by func-tional groups. A cursory look at the quinol forms would suggest that designing stable, low-potential anolytes is a futile task. However, thermodynamic susceptibility to deg-radation may fortunately not translate directly into long-term instability. A striking ACCEPTED MANUSCRIPTexample of this is the recently demonstrated electrochemical reversibility of the dis-proportionation/dimerization reactions in the case of 2,6-dihydroxy-anthraquinone (DHAQ) in alkaline media.[36] In this study, Jing et al. were able to demonstrate non-invasive regeneration of the desired redox-active components by stimulating oxidative processes using an externally controlled potential that was only slightly outside the operating limits. The authors further confirmed the same regeneration in the case of anthraquinone-2,7-disulfonic acid (AQDS) in acidic media. These promising evidenc-es point to the need to develop new stability descriptors that capture the reversibility of such degradation reactions rather than one-directional thermodynamic susceptibil-ity.             Another such example is the case of tautomerization (Fig. 3d), in which a small outlier subset appears in the lower-right region of the plot, deviating from the main correlation line. To investigate these deviations in more detail, we first exam-ined the molecular structures corresponding to the outliers and found that they are ex-clusively associated with anthraquinone cores (Fig. 1b, structure 4) and their func-tionalized derivatives. Our thermodynamic analysis predicts this subset of anthraqui-none anolytes to be relatively resistant to tautomerization, which is consistent with numerous literature studies that have collectively established anthraquinones as benchmark molecules.[43,44,61]  To rationalize the enhanced thermodynamic resistance of the anthraquinone-derived outliers to tautomerization, we investigated another metric of stability, i.e., structural distortion of the molecules upon reaction. The structural distortion was computed using the carbon-skeleton RMSD between DFT-optimized reactant and product geometries with the LS-align (Rigid-LS-align)[75] tool. For this analysis, we evaluated the four single-carbonyl parent core structures (1, 2, 4, 5) shown in Fig. 1b. ACCEPTED MANUSCRIPTSingle-carbonyl structures were selected in order to enable a fair comparison and avoid the effects of intramolecular hydrogen bonding. The RMSD values for tautom-erization were 0.213, 26.316, 56.432, and 37.742 Å for structures 1, 2, 4, and 5, re-spectively. Clearly, structure 4 shows the largest geometric rearrangement, consistent with a higher reaction Gibbs free energy and a less favourable tautomerization process, which plausibly explains its deviation from the main trend. The RMSD values for the cores with double carbonyls, namely structures 3, 6, and 7, are 22.194, 6.989, and 49.843 Å. Despite having relatively high RMSD values, they do not appear as outliers because their redox potentials are also high. Taken together, these observations indi-cate that tautomerization thermodynamics is governed not only by electron-density effects but also by the structural distortion of the resulting tautomerized product. Ac-cordingly, thermodynamic screening should be interpreted as providing a baseline metric that would benefit from complementary kinetic and geometrical considerations. 3.1 Effect of the functional groups on stability against specific degradation reactions To assess functional group effects on quinone stability, we defined, for each degrada-tion mechanism, the relative free energy difference (∆∆𝐺 _ ) between the reaction Gibbs free energies of each functionalized derivative (∆𝐺 ) and that of its corresponding parent core structure (∆𝐺 _ ), as follows:  ∆∆𝐺 _ = ∆𝐺 _ − ∆𝐺 _             (4) A larger ∆∆𝐺 _  indicates greater thermodynamic resistance of the function-alized derivative toward the corresponding degradation pathway relative to the parent core structure. We then analysed functional group effects across the various degrada-tion reactions using this descriptor, as shown in the violin plots in Fig. 4. The corre-ACCEPTED MANUSCRIPTsponding detailed, core-wise decomposition of ∆∆𝐺 _  for each functional group and degradation mechanism, enabling direct comparison both within and across core structures, is provided in the Supporting Information (Figs. S4-S9). It is not surprising that, for Michael addition, gem-diol formation, and an-throne formation, the trends are broadly similar, with the electron-donating (-OH, -NH2) functionalized cores generally showing higher values of ∆∆𝐺 _  than the electron-withdrawing (-SO3H, -COOH) functionalized cores. Despite this, there are again striking differences with important consequences for molecular design. For example, for Michael addition (Fig. 4a) and anthrone formation (Fig. 4c), nearly all the considered functional groups lead to decreased stability, whereas for gem-diol formation (Fig. 4b), a significant number of -SO3H, -OH, and -NH2-functionalized molecules exhibit higher stability than their parent cores. Nevertheless, there are for-tunate exceptions and outliers, for example, naphthoquinone and anthraquinone cores seem to achieve higher stability across all degradation mechanisms when functional-ized with -NH2 (Figs. S3-S5 in the Supporting Information). As expected, tautomeri-zation, disproportionation, and dimerization show the opposite dependence on func-tional groups (Figs. 4d-f). For all three mechanisms, the electron-withdrawing -COOH, -SO3H, and -Cl groups are broadly associated with positive ΔΔ𝐺  values, whereas -OH and -NH2 lie mainly below zero. A closer look at the more detailed distributions (Figs. S6-S8) again shows a large degree of coupling between the functional group and the core type, because even the same functional group has very different effects on benzo-, naphtho-, and anthraquinone cores.   While the overall landscape of quinone thermodynamic stability suffers from the bespoke trade-offs, a fortunate observation is that the range over which stability is influenced extends in both the positive and the negative direction. This implies that ACCEPTED MANUSCRIPTthere may still be functional groups (or a combination of them, as discussed in section 3.3) that can be employed to balance these trade-offs. These results have direct implications for molecular design and screening. Be-cause functional group effects are pathway-specific and may reverse across mecha-nisms, no single functional group rule is universally favourable. Instead, functional group effects should be assessed by considering multiple degradation pathways along-side redox properties.3.2 Relations between stability and solubility descriptors Next, we examined the relationship between thermodynamic stability and two de-scriptors relevant to aqueous solubility: (a) the ML-predicted aqueous solubility (log 𝑆) and (b) the solvation free energy (Δ𝐺 ). Because aqueous solubility depends not on-ly on solvation but also on intermolecular packing, aggregation, and other condensed-phase effects, Δ𝐺  is treated here as a complementary descriptor rather than a direct surrogate for log 𝑆. Fig. 5 reveals weak and highly dispersed relationships between predicted aqueous solubility and Δ𝐺reaction type  across all six degradation pathways. Although slight visual differences between the pathway-specific distributions may be present, no clear or robust monotonic trend can be identified for any individual mech-anism. The corresponding analysis based on solvation free energy (Supporting Infor-mation, Fig. S10) leads to the same general conclusion. The relationship between Δ𝐺  and Δ𝐺reaction type is likewise weak and broadly scattered, and no robust stability-dependent trend can be identified across the degradation pathways. More broadly, the ML-predicted aqueous compatibility is governed more strongly by structural features, particularly ring count and carbonyl count. In compari-ACCEPTED MANUSCRIPTson, functional group-specific effects are less clearly resolved. Overall, these results do not support a clear relationship between degradation thermodynamics and aqueous compatibility in the present dataset. It should be noted, however, that even state-of-the-art solubility prediction models do not have well-established uncertainty esti-mates,[76,77] and their predictions can often be erroneous by several orders of magni-tude. In any case, the absence of any specific correlation bodes well for the molecular design space.  3.3 Approximate metrics of stability from experimentally studied molecules and their thermodynamic stability evaluation A solid validation of our approach would ideally involve verification of the proposed stability descriptors against experimental stability data. As a starting point, we com-piled a representative collection of quinone-based molecules reported in the literature as comparatively stable (Fig. S11),[19,20,26,30,39,40,44,46,78-80] thereby placing the present work in the context of experimentally studied compounds with promising electrochemical performance. Their reported performance was extracted in terms of various durability metrics, such as capacity retention, cycle number, energy efficiency, etc., as tabulated in Table S3 in the Supporting Information. A cursory look at the sparseness of Table S3 shows that the reported systems were tested under different solvent chemistries, pH, active-molecule concentrations, and current densities. This, combined with the fact that their stability can only be interpreted qualitatively (as op-posed to being inferred quantitatively) and with the availability of disparate durability metrics, makes a direct comparison between theory and experiment very challenging. To the best of our knowledge, there are no experimental data on the individual reac-tions considered in our work that would enable a direct comparison between experi-ACCEPTED MANUSCRIPTmentally observed reaction rates and computed thermodynamic descriptors. Nevertheless, most of the literature molecules collected align well with the broad stability-potential correlations identified in our analysis, with only a few nota-ble outliers, as seen in Fig. S12 in the Supporting Information. A large fraction of the experimentally collected molecules has two or more functional groups, which entail intramolecular interactions and effects beyond simple electron donation/withdrawal. On closer inspection, we find that the more outlying molecules are multi-functionalized species, which points to multi-functionalization as another degree of freedom that warrants further exploration. The same has also been evidenced in recent literature, for example, in the study by Alfaraidi et al.,[80] who showed that combin-ing a neighbouring -NH2 group with an ether-containing functional group  (-O-CH(CH₃)-COOH) in an anthraquinone electrolyte can simultaneously suppress side-chain loss and anthrone formation while maintaining high solubility. They reported an extremely low-capacity fade rate of 0.01%/day and a very high Coulombic efficiency of >99.8%.4 Conclusion In this work, we performed a systematic thermodynamic analysis of multiple degrada-tion pathways in quinone–quinol redox couples and established clear structure–property relationships linking redox potential, functionalization, and stability. Across all investigated mechanisms, strong correlations were identified between the reaction Gibbs free energy of various degradation reactions and the redox potential, confirm-ing that electron-donating and electron-withdrawing effects govern both electrochem-ical properties and thermodynamic susceptibility to degradation. Notably, oxidized (quinone) and reduced (quinol) forms exhibit fundamentally different trends, high-ACCEPTED MANUSCRIPTlighting an intrinsic asymmetry in the stability landscape that must be considered in molecular design. For Michael addition, gem-diol formation, and anthrone formation, higher re-dox potentials correlate with increased thermodynamic susceptibility, consistent with enhanced electrophilicity of the quinone core. In contrast, tautomerization, dispropor-tionation, and dimerization display the opposite trend, leading to inherent trade-offs between achieving low redox potential and maintaining thermodynamic stability. These competing relationships demonstrate that optimizing quinone-based anolytes cannot rely on a single descriptor but instead requires a multi-objective perspective that accounts for multiple degradation pathways simultaneously. A key finding of this work is the identification of systematic outliers, particu-larly among anthraquinone derivatives, which exhibit enhanced resistance to tautom-erization despite their redox characteristics. Detailed structural analysis revealed that this behaviour is linked not only to electronic effects but also to significant geometric distortion upon reaction, as quantified by RMSD metrics. These results emphasize that molecular stability is governed by a complex interplay of electronic, structural, and functionalization effects, and that rational design must navigate competing trends across multiple reaction pathways. Functional group effects were shown to be strongly dependent on the reaction-type and highly coupled to the underlying quinone core. While electron-donating groups generally stabilize against nucleophilic degradation pathways, they tend to de-stabilize reduced-form reactions, with the opposite behaviour observed for electron-withdrawing substituents. Importantly, no universally beneficial functional group was identified, reinforcing the need for balanced, context-specific design strategies. Nev-ertheless, the broad distribution of functional group effects in both stabilizing and de-ACCEPTED MANUSCRIPTstabilizing directions suggests that careful functionalization—and particularly multi-functionalization—offers viable routes to mitigate these trade-offs. In contrast, the relationship between thermodynamic stability and aqueous compatibility (as described by ML-predicted solubility and solvation free energy) was found to be weak and non-systematic. This lack of correlation indicates that stability and solubility can, to a first approximation, be tuned independently, thereby expand-ing the accessible design space for aqueous organic redox flow battery electrolytes. Comparison with experimentally studied quinone systems further supports the qualitative validity of the proposed descriptors, despite the challenges associated with heterogeneous experimental conditions and the absence of direct kinetic data for indi-vidual degradation pathways. Importantly, literature examples demonstrate that some thermodynamically favourable degradation reactions, such as disproportionation and dimerization, can be electrochemically reversible.  Overall, this study establishes a thermodynamic framework for understanding and screening degradation mechanisms in quinone-based systems. Future work should therefore focus on integrating reversibility, kinetic modelling, explicit solvent effects, and multi-functional group interactions, as well as developing descriptors that capture dynamic stability under operating conditions. Disclosure statement The authors report there are no competing interests to declare. Acknowledgement N.L. acknowledges the support of the China Scholarship Council (No. 202208410105). A.K. and S.T. gratefully acknowledge funding by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) under Germany’s Excellence Strategy – Cluster of Excellence 2186/2, “The Integrated Fuel & Chemical Science Center” (ID: 390919832). A.K. also grate-ACCEPTED MANUSCRIPTfully acknowledges financial support through the project KI-LECTROLYTE (EFRE-20801094) granted jointly by the European Union and the State of North-Rhein Westfalia under the NEXT.IN.NRW program. A.K. and N.L. gratefully acknowledge the computing time provided by the NHR Center NHR4CES at RWTH Aachen University under project numbers p0024808 and p0027076. NHR4CES is funded by the Federal Ministry of Education and Research and the state governments participating based on the resolutions of the GWK for national high-performance computing at universities (www.nhr-verein.de/unsere-partner). Author contribution list N.L.: Conceptualization; Data curation; Formal analysis; Methodology; Software; Writing – original draft; Writing – review & editing. S.J.: Software; Methodology. D.W.: Writing – review & editing. S.T.: Validation. A.K.: Conceptualization; Data curation; Formal analysis; Funding acquisition; Supervision; Writing – review & edit-ing. Supporting Information Table S1. 35 different SMIRKS to represent eight types of reactions Table S2. A summary of 2D structures, SMILES representations, and experimental redox potentials of the compounds. Table S3. Summary of experimental redox couples and performance metrics of AORFBs from literature Figure S1. The simulation between the free energy and the redox potential in experi-ments. Figure S1. Redox potential property relationships. Figure S3. The relationship between the predicted redox potential and the Gibbs free energy of Proto-desulfonation. Figure S4. Effect of functional groups on stability based on core structure under Mi-chael addition mechanisms. ACCEPTED MANUSCRIPTFigure S5. Effect of functional groups on stability based on core structure under Gem-Diol Formation mechanisms. Figure S6. Effect of functional groups on stability based on core structure under the anthrone formation mechanism. Figure S7. Effect of functional groups on stability based on core structure under the Tautomerization mechanism. Figure S8. Effect of functional groups on stability based on core structure under the Disproportionation mechanism. Figure S9. Effect of functional groups on stability based on core structure under the Dimerization mechanism. Figure S10. The relationship between the solvation reaction Gibbs free energy and the reaction Gibbs free energy of the six degradation mechanisms. Figure S11. Structures of literature compounds used for validation. Figure S12. Experimental Validation of Theoretical Predictions: The relationship be-tween the stability and the redox potential with the structures of the experimental points. 5. Reference [1] Chen H, Cong G, Lu Y-C. Recent progress in organic redox flow batteries: Active materials, electrolytes and membranes. Journal of Energy Chemistry. 2018;27(5):1304-1325. [2] Zhao Z, Zhang C, Li X. Opportunities and challenges of organic flow battery for electrochemical energy storage technology. Journal of Energy Chemistry. 2022;67:621-639. [3] Mao J, Ruan W, Chen Q. Understanding the Aqueous Solubility of Anthraquinone Sulfonate Salts: The Quest for High Capacity Electrolytes of Redox Flow Batteries. Journal of The Electrochemical Society. 2020;167(7). ACCEPTED MANUSCRIPT[4] Winsberg J, Hagemann T, Janoschka T, et al. Redox-Flow Batteries: From Metals to Organic Redox-Active Materials. Angew Chem Int Ed Engl. 2017;56(3):686-711. [5] Bauer S, Namyslo JC, Kaufmann DE, et al. Evaluation of Options and Limits of Aqueous All-Quinone-Based Organic Redox Flow Batteries. Journal of The Electrochemical Society. 2020;167(11). [6] Forner-Cuenca A. Bringing redox organics back to life. Nat Chem. 2022;14(10):1091-1092. [7] Sharmoukh W. Redox flow batteries as energy storage systems: materials, viability, and industrial applications. RSC Adv. 2025;15(13):10106-10143. [8] Ji Y, Goulet MA, Pollack DA, et al. A Phosphonate‐Functionalized Quinone Redox Flow Battery at Near‐Neutral pH with Record Capacity Retention Rate. Advanced Energy Materials. 2019;9(12). [9] Soloveichik GL. Flow Batteries: Current Status and Trends. Chem Rev. 2015;115(20):11533-58. [10] Fontmorin J-M, Guiheneuf S, Godet-Bar T, et al. How anthraquinones can enable aqueous organic redox flow batteries to meet the needs of industrialization. Current Opinion in Colloid & Interface Science. 2022;61. [11] Han Y-K, Jin C-S. Computational screening of electroactive indolequinone derivatives as high-performance active materials for aqueous redox flow batteries. Current Applied Physics. 2018;18(12):1507-1512. [12] Ye R, Henkensmeier D, Chen R. Imidazolium cation enabled reversibility of a hydroquinone derivative for designing aqueous redox electrolytes. Sustainable Energy & Fuels. 2020;4(6):2998-3005. [13] Goulet M-A, Aziz MJ. Flow Battery Molecular Reactant Stability Determined by Symmetric Cell Cycling Methods. Journal of The Electrochemical Society. 2018;165(7):A1466-A1477. [14] Mazur P, Charvat J, Mrlik J, et al. Evaluation of Electrochemical Stability of Sulfonated Anthraquinone-Based Acidic Electrolyte for Redox Flow Battery Application. Molecules. 2021;26(9). [15] Fornari RP, Mesta M, Hjelm J, et al. Molecular Engineering Strategies for Symmetric Aqueous Organic Redox Flow Batteries. ACS Materials Letters. 2020;2(3):239-246. ACCEPTED MANUSCRIPT[16] Kristensen SB, van Mourik T, Pedersen TB, et al. Simulation of electrochemical properties of naturally occurring quinones. Sci Rep. 2020;10(1):13571. [17] Li J, Xu H, Wang J, et al. Theoretical insights on the hydration of quinones as catholytes in aqueous redox flow batteries. Chinese Journal of Chemical Engineering. 2021;37:72-78. [18] Cannon CG, Klusener PAA, Brandon NP, et al. Aqueous Redox Flow Batteries: Small Organic Molecules for the Positive Electrolyte Species. ChemSusChem. 2023;16(18):e202300303. [19] Lu H, Yu J, Zhang J, et al. Designing p‐π Conjugated Naphthoquinone‐Derivatives as High‐Performance Cathodes for Aqueous Proton Batteries. Advanced Functional Materials. 2025;35(24). [20] Ozouf I, Fontmorin JM, Lebeuf R, et al. Long-cycling of a water-soluble quinizarin derivative in redox flow batteries: Role of the cut-off voltage on the stability. Electrochimica Acta. 2024;475. [21] Gerhardt MR, Tong L, Gómez‐Bombarelli R, et al. Anthraquinone Derivatives in Aqueous Flow Batteries. Advanced Energy Materials. 2016;7(8). [22] Huang Z, Lee J, Henkensmeier D, et al. Effect of Molecular Structure and Coordinating Ions on the Solubility and Electrochemical Behavior of Quinone Derivatives for Aqueous Redox Flow Batteries. Journal of The Electrochemical Society. 2020;167(16). [23] Kwabi DG, Wong AA, Aziz MJ. Rational Evaluation and Cycle Life Improvement of Quinone-Based Aqueous Flow Batteries Guided by In-Line Optical Spectrophotometry. Journal of The Electrochemical Society. 2018;165(9):A1770-A1776. [24] Li Z, Lu YC. Material Design of Aqueous Redox Flow Batteries: Fundamental Challenges and Mitigation Strategies. Adv Mater. 2020;32(47):e2002132. [25] Wedege K, Drazevic E, Konya D, et al. Organic Redox Species in Aqueous Flow Batteries: Redox Potentials, Chemical Stability and Solubility. Sci Rep. 2016;6:39101. [26] Kwabi DG, Lin K, Ji Y, et al. Alkaline Quinone Flow Battery with Long Lifetime at pH 12. Joule. 2018;2(9):1894-1906. ACCEPTED MANUSCRIPT[27] Sivanadanam J, Murugan R, Khan H, et al. Investigation of Alkyl Amine Substituted Quinone Derivatives for the Redox Flow Battery Applications in Acidic Medium. Journal of The Electrochemical Society. 2022;169(2). [28] Tabor DP, Gómez-Bombarelli R, Tong L, et al. Mapping the frontiers of quinone stability in aqueous media: implications for organic aqueous redox flow batteries. Journal of Materials Chemistry A. 2019;7(20):12833-12841. [29] Liu Y, Chen Q, Zhang X, et al. Degradation of electrochemical active compounds in aqueous organic redox flow batteries. Current Opinion in Electrochemistry. 2022;32. [30] Wu M, Jing Y, Wong AA, et al. Extremely Stable Anthraquinone Negolytes Synthesized from Common Precursors. Chem. 2020;6(6):1432-1442. [31] Beck F, Heydecke G. On the Mechanism of the Cathodic Reduction of Anthraquinone to Anthrone. Berichte der Bunsengesellschaft für physikalische Chemie. 1987;91(1). [32] Jing Y, Fell EM, Wu M, et al. Anthraquinone Flow Battery Reactants with Nonhydrolyzable Water-Solubilizing Chains Introduced via a Generic Cross-Coupling Method. ACS Energy Letters. 2021;7(1):226-235. [33] Tong L, Goulet M-A, Tabor DP, et al. Molecular Engineering of an Alkaline Naphthoquinone Flow Battery. ACS Energy Letters. 2019;4(8):1880-1887. [34] Fain VY, Zaitsev BE, Ryabov MA. Tautomerism of anthraquinones: III. Tautomerization and rotational isomerization as processes responsible for the appearance of several π1,π*-bands in the absorption spectra of hydroxy-substituted quinones. Russian Journal of Organic Chemistry. 2006;42(10):1464-1468. [35] Carney TJ, Collins SJ, Moore JS, et al. Concentration-Dependent Dimerization of Anthraquinone Disulfonic Acid and Its Impact on Charge Storage. Chemistry of Materials. 2017;29(11):4801-4810. [36] Jing Y, Zhao EW, Goulet MA, et al. In situ electrochemical recomposition of decomposed redox-active species in aqueous organic flow batteries. Nat Chem. 2022;14(10):1103-1109. [37] Goulet MA, Tong L, Pollack DA, et al. Extending the Lifetime of Organic Flow Batteries via Redox State Management. J Am Chem Soc. 2019;141(20):8014-8019. ACCEPTED MANUSCRIPT[38] Zhang X, de Silva P. Computational framework for discovery of degradation mechanisms of organic flow battery electrolytes. Chem Sci. 2025;16(19):8422-8434. [39] Liu Y, Zhang P, Wu Z, et al. Biomimetic Naphthoquinone Zwitterion Derivative with Water-Solubilizing Amino Acid Side Chain for High-Stability Aqueous Redox Flow Batteries. ACS Energy Letters. 2024;9(2):586-593. [40] Wu M, Bahari M, Jing Y, et al. Highly Stable, Low Redox Potential Quinone for Aqueous Flow Batteries**. Batteries & Supercaps. 2022;5(6). [41] Kerr EF, Tang Z, George TY, et al. High Energy Density Aqueous Flow Battery Utilizing Extremely Stable, Branching-Induced High-Solubility Anthraquinone near Neutral pH. ACS Energy Letters. 2022;8(1):600-607. [42] Jethwa RB, Zhao EW, Kerber RN, et al. Designing for conjugate addition: an amine functionalised quinone anolyte for redox flow batteries. Journal of Materials Chemistry A. 2021;9(27):15188-15198. [43] Fenini F, Drazevic E, Bentien A. Impact of pH management on utilization and performance of anthraquinone/ ferrocyanide flow batteries. Journal of Power Sources. 2022;540. [44] Guiheneuf S, Godet-Bar T, Fontmorin JM, et al. A new hydroxyanthraquinone derivative with a low and reversible capacity fading process as negolyte in alkaline aqueous redox flow batteries. Journal of Power Sources. 2022;539. [45] Gerken JB, Anson CW, Preger Y, et al. Comparison of Quinone‐Based Catholytes for Aqueous Redox Flow Batteries and Demonstration of Long‐Term Stability with Tetrasubstituted Quinones. Advanced Energy Materials. 2020;10(20). [46] Yang X, Garcia SN, Janoschka T, et al. Novel, Stable Catholyte for Aqueous Organic Redox Flow Batteries: Symmetric Cell Study of Hydroquinones with High Accessible Capacity. Molecules. 2021;26(13). [47] Nandi S, de Sousa LE, Vegge T, et al. Degradation of Quinone‐based Flow Battery Electrolytes: Effect of Functional Groups on the Reaction Mechanism**. Batteries & Supercaps. 2022;6(2). [48] Liu W, Zhao Z, Li T, et al. A high potential biphenol derivative cathode: toward a highly stable air-insensitive aqueous organic flow battery. Sci Bull (Beijing). 2021;66(5):457-463. ACCEPTED MANUSCRIPT[49] Khetan A. High-Throughput Virtual Screening of Quinones for Aqueous Redox Flow Batteries: Status and Perspectives. Batteries. 2022;9(1). [50] Bento AP, Hersey A, Félix E, et al. An open source chemical structure curation pipeline using RDKit. Journal of Cheminformatics. 2020;12(1). [51] Bannan CC, Mobley DL. ChemPer: An Open Source Tool for Automatically Generating SMIRKS Patterns. Theoretical and Computational Chemistry. 2019. [52] Song Y, Buettner GR. Thermodynamic and kinetic considerations for the reaction of semiquinone radicals to form superoxide and hydrogen peroxide. Free Radic Biol Med. 2010;49(6):919-62. [53] Manisankar P, Valarselvan S. Electrocatalytic properties of glassy carbon electrodes modified with hydroxy derivatives of 9,10-anthraquinone for oxygen reduction reaction. Ionics. 2012;18(7):679-686. [54] Schwan S, Schroder D, Wegner HA, et al. Substituent Pattern Effects on the Redox Potentials of Quinone-Based Active Materials for Aqueous Redox Flow Batteries. ChemSusChem. 2020;13(20):5480-5488. [55] Coley CW, Green WH, Jensen KF. RDChiral: An RDKit Wrapper for Handling Stereochemistry in Retrosynthetic Template Extraction and Application. Journal of Chemical Information and Modeling. 2019;59(6):2529-2537. [56] Lovrić M, Molero JM, Kern R. PySpark and RDKit: Moving towards Big Data in Cheminformatics. Molecular Informatics. 2019;38(6). [57] Srinath KR. Python – The Fastest Growing Programming Language. International Research Journal of Engineering and Technology (IRJET) 2017;04(12):354-357. [58] Guin PS, Das S, Mandal PC. Electrochemical Reduction of Quinones in Different Media: A Review. International Journal of Electrochemistry. 2011;2011:1-22. [59] Hoober-Burkhardt L, Krishnamoorthy S, Yang B, et al. A New Michael-Reaction-Resistant Benzoquinone for Aqueous Organic Redox Flow Batteries. Journal of The Electrochemical Society. 2017;164(4):A600-A607. [60] Murali A, Nirmalchandar A, Krishnamoorthy S, et al. Understanding and Mitigating Capacity Fade in Aqueous Organic Redox Flow Batteries. Journal of The Electrochemical Society. 2018;165(7):A1193-A1203. ACCEPTED MANUSCRIPT[61] Huang S, Zhang H, Salla M, et al. Molecular engineering of dihydroxyanthraquinone-based electrolytes for high-capacity aqueous organic redox flow batteries. Nat Commun. 2022;13(1):4746. [62] Zhao EW, Liu T, Jonsson E, et al. In situ NMR metrology reveals reaction mechanisms in redox flow batteries. Nature. 2020;579(7798):224-228. [63] Neese F. The ORCA program system. WIREs Computational Molecular Science. 2011;2(1):73-78. [64] Medvedev MG, Bushmarinov IS, Sun J, et al. Density functional theory is straying from the path toward the exact functional. American Association for the Advancement of Science. 2017;355:49-52. [65] Barone V, Cossi M. quantum-calculation-of-molecular-energies-and-energy-gradients-in-solution-by-a-conductor-solvent-model. The Journal of Physical Chemistry A. 1998;102(11). [66] Cossi M, Rega N, Scalmani G, et al. Energies, structures, and electronic properties of molecules in solution with the C-PCM solvation model. J Comput Chem. 2003;24(6):669-81. [67] Truong TN, Stefanovich EV. A new method for incorporating solvent effect into the classical, ab initio molecular orbital and density functional theory frameworks for arbitrary shape cavity.pdf. Chemical Physics Letters. 1995;240(4):253-260. [68] Sorkun MC, Koelman J, Er S. Pushing the limits of solubility prediction via quality-oriented data selection. iScience. 2021;24(1):101961. [69] Zhang Q, Khetan A, Sorkun E, et al. Data-driven discovery of small electroactive molecules for energy storage in aqueous redox flow batteries. Energy Storage Materials. 2022;47:167-177. [70] Zhang Q, Khetan A, Er S. A quantitative evaluation of computational methods to accelerate the study of alloxazine-derived electroactive compounds for energy storage. Sci Rep. 2021;11(1):4089. [71] Zhang Q, Khetan A, Er S. Comparison of computational chemistry methods for the discovery of quinone-based electroactive compounds for energy storage. Sci Rep. 2020;10(1):22149. [72] Wass JRTJ, Ahlberg E, Panas I, et al. Quantum Chemical Modeling of the Reduction of Quinones. The Journal of Physical Chemistry A. 2006;110(5). ACCEPTED MANUSCRIPT[73] Lin J, Wang Y, Chen M, et al. Regulating the Gibbs Free Energy to Design Aqueous Battery‐Compatible Robust Host. Advanced Energy Materials. 2024;14(31). [74] Zhou X, Khetan A, Zheng J, et al. Discovery of lead quinone cathode materials for Li-ion batteries. Digit Discov. 2023;2(4):1016-1025. [75] Hu J, Liu Z, Yu DJ, et al. LS-align: an atom-level, flexible ligand structural alignment algorithm for high-throughput virtual screening. Bioinformatics. 2018;34(13):2209-2218. [76] Fowles DJ, Connaughton BJ, Carter JW, et al. Physics-Based Solubility Prediction for Organic Molecules. Chem Rev. 2025;125(15):7057-7098. [77] Reber D, Wang Z, Amini K, et al. Solubility Challenges in Battery Electrolytes. Chem Rev. 2025;125(23):11216-11259. [78] Yang G, Zhu Y, Hao Z, et al. An Aqueous All‐Quinone‐Based Redox Flow Battery Employing Neutral Electrolyte. Advanced Energy Materials. 2024;14(26). [79] Ding Y, Li Y, Yu G. Exploring Bio-inspired Quinone-Based Organic Redox Flow Batteries: A Combined Experimental and Computational Study. Chem. 2016;1(5):790-801. [80] Alfaraidi AM, Xi D, Ni N, et al. An Extremely Stable and Soluble NH2-Substituted Anthraquinone Electrolyte for Aqueous Redox Flow Batteries. ACS Applied Energy Materials. 2023;6(24):12259-12266.  Figure 1. (a) Carbon atom numbering scheme used to define substitution positions on benzene, naphthalene, and anthracene. (b) quinone core structures investigated in this study (1-7).  ACCEPTED MANUSCRIPT Figure 2. A schematic representation of the anthraquinone functionalization applied during the enumeration process for generating the virtual molecular library   ACCEPTED MANUSCRIPT Figure 3. Correlation between predicted half-cell redox potential and Δ𝐺reaction type for six degradation mechanisms (excluding proto-desulfonation) of quinone-hydroquinone pairs.   ACCEPTED MANUSCRIPT Figure 4. Effect of functional group on the stability of different degradation mecha-nisms.                ACCEPTED MANUSCRIPT Figure 5. The relationship between the ML-predicted solubility (AqSolPred) and the reaction Gibbs free energy of the six degradation mechanisms.      ACCEPTED MANUSCRIPTTable 1. The schemes of the reaction mechanismACCEPTED MANUSCRIPT Electrochemical reaction: ∆𝐺 = 𝐺 − 𝐺 − 𝐺   Micheal Addition: ∆𝐺 = 𝐺 − 𝐺 − 𝐺   Gem-diol Formation: ∆𝐺 = 𝐺 − 𝐺 − 𝐺  Proto-desulfonation: ∆𝐺 = 𝐺 + 𝐺 − 𝐺 − 𝐺   Anthrone Formation: ∆𝐺 = 𝐺 + 𝐺 − 𝐺 − 2𝐺  Tautomerization: ∆𝐺 = 𝐺 − 𝐺  Disproportionation:  ∆𝐺 = 𝐺 + 𝐺 − 𝐺 + 𝐺  ACCEPTED MANUSCRIPT Dimerization: ∆𝐺 = 𝐺 + 2𝐺 − 2𝐺 − 𝐺                       Table 2. Example of five SMIRKS templates corresponding to the main electrochemical reaction Electrochemical reaction SMIRKS template   SMIRKS1:[O:0]=[C:1]1-[C:2]=[C:3]-[C:4](=[O:5])-[C:6]=[C:7]-1 >>[O:0]-[c:1]1:[c:2]:[c:3]:[c:4](-[O:5]):[c:6]:[c:7]:1 ACCEPTED MANUSCRIPT  SMIRKS2:[O:0]=[C:1]1-[c:2]:[c:3]-[C:4](=[O:5])-[C:6]=[C:7]-1 >>[O:0]-[c:1]1:[c:2]:[c:3]:[c:4](-[O:5]):[c:6]:[c:7]:1   SMIRKS3:[O:0]=[C:1]1-[c:2]:[c:3]-[C:4](=[O:5])-[c:6]:[c:7]-1 >>[O:0]-[c:1]1:[c:2]:[c:3]:[c:4](-[O:5]):[c:6]:[c:7]:1   SMIRKS4:[O:0]=[c:1]1:[c:2]=[c:3]:[c:4](=[O:5]):[c:6]:[c:7]:1 >>[O:0]-[c:1]1:[c:2]:[c:3]:[c:4](-[O:5]):[c:6]:[c:7]:1   SMIRKS5:[O:0]=[c:1]1:[c:2]:[c:3]:[c:4](=[O:5]):[c:6]:[c:7]:1 >>[O:0]-[c:1]1:[c:2]:[c:3]:[c:4](-[O:5]):[c:6]:[c:7]:1    ACCEPTED MANUSCRIPT Graphical abstract   ACCEPTED MANUSCRIPTStatement of Novelty  The work presents the most comprehensive analysis to date of all major quinone degra-dation mechanisms, revealing pathway-specific trade-offs and identifying structural distortion and reversibility as critical, previously overlooked stability descriptors.   ACCEPTED MANUSCRIPTSupplementary information  A data-driven elucidation of the challenges in designing stable quinone derivatives for aqueous organic redox flow batteries: correlations between thermodynamics of chemical degradation and energy capac-ity metrics Ning Lua, Shiyao Jua, Daniel Willimetza,b, Shengming Tanga, Abhishek Khetana* a MODES, The Integrated Fuel & Chemical Science Centre, RWTH Aachen Universi-ty, 52062 Aachen, Germany  bFraunhofer Institute for Algorithms and Scientific Computing SCAI, Fraunhofer- Gesellschaft, Schloss Birlinghoven 1, 53757 Sankt Augustin, Germany  *Corresponding author: Abhishek Khetan; Email: askhetan@modes.rwth-aachen.de       Explanation of the abbreviations used in the below tables:  SMILES: simplified molecular-input line-entry system DFT: density functional theory R2: the coefficient of fit  ACCEPTED MANUSCRIPTRMSE: the root mean squared error MSE: the mean squared error MAE: the mean absolute error RMSD: root mean square deviation ACCEPTED MANUSCRIPTTable S1. The 35 SMIRKS templates used to represent the eight reaction categories considered in this work, comprising seven degradation reactions and the main electro-chemical redox reaction. SMIRKS1 : [O:0]=[C:1]1-[C:2]=[C:3]-[C:4](=[O:5])-[C:6]=[C:7]-1>>[O:0]-  [c:1]1:[c:2]:[c:3]:[c:4](-[O:5]):[c:6]:[c:7]:1   SMIRKS 2 : [O:0]=[C:1]1-[c:2]:[c:3]-[C:4](=[O:5])-[C:6]=[C:7]-1>>[O:0]-  [c:1]1:[c:2]:[c:3]:[c:4](-[O:5]):[c:6]:[c:7]:1   SMIRKS 3 : [O:0]=[C:1]1-[c:2]:[c:3]-[C:4](=[O:5])-[c:6]:[c:7]-1>>[O:0]-  [c:1]1:[c:2]:[c:3]:[c:4](-[O:5]):[c:6]:[c:7]:1   ACCEPTED MANUSCRIPT SMIRKS 4 : [O:0]=[c:1]1:[c:2]=[c:3]:[c:4](=[O:5]):[c:6]:[c:7]:1>>[O:0]-  [c:1]1:[c:2]:[c:3]:[c:4](-[O:5]):[c:6]:[c:7]:1   SMIRKS 5 : [O:0]=[c:1]1:[c:2]:[c:3]:[c:4](=[O:5]):[c:6]:[c:7]:1>>[O:0]-  [c:1]1:[c:2]:[c:3]:[c:4](-[O:5]):[c:6]:[c:7]:1   SMIRKS 6 : [O:0]=[C:1]1-[C:2]=[C:3]-[C:4](=[O:5])-[C:6]=[C:7]-1>>[O-:0]-  [c:1]1:[c:2]:[c:3]:[c:4](-[O-:5]):[c:6]:[c:7]:1   ACCEPTED MANUSCRIPT SMIRKS 7 : [O:0]=[C:1]1-[c:2]:[c:3]-[C:4](=[O:5])-[C:6]=[C:7]-1>>[O-:0]-  [c:1]1:[c:2]:[c:3]:[c:4](-[O-:5]):[c:6]:[c:7]:1   SMIRKS 8 : [O:0]=[C:1]1-[c:2]:[c:3]-[C:4](=[O:5])-[c:6]:[c:7]-1>>[O-:0]-  [c:1]1:[c:2]:[c:3]:[c:4](-[O-:5]):[c:6]:[c:7]:1   SMIRKS 9 : [O:0]=[c:1]1:[c:2]=[c:3]:[c:4](=[O:5]):[c:6]:[c:7]:1>>[O-:0]-  [c:1]1:[c:2]:[c:3]:[c:4](-[O-:5]):[c:6]:[c:7]:1   ACCEPTED MANUSCRIPT SMIRKS 10 : [O:0]=[c:1]1:[c:2]:[c:3]:[c:4](=[O:5]):[c:6]:[c:7]:1>>[O-:0]-  [c:1]1:[c:2]:[c:3]:[c:4](-[O-:5]):[c:6]:[c:7]:1   SMIRKS 11 : [O:0]=[C:1]1-[C:2]=[C:3]-[C:4](=[O:5])-[C:6]=[C:7]-  1.[O;H2;+0:8]>>[O:0]=[C:1]1-[C:2]=[C:3]-[C:4](-[O:5])(-[O:8])-[C:6]=[C:7]-1   SMIRKS 12 : [O:0]=[C:1]1-[c:2]:[c:3]-[C:4](=[O:5])-[C:6]=[C:7]-  1.[O;H2;+0:8]>>[O:0]=[C:1]1-[C:2]:[C:3]-[C:4](-[O:5])(-[O:8])-[C:6]=[C:7]-1    ACCEPTED MANUSCRIPT SMIRKS 13 : [O:0]=[C:1]1-[c:2]:[c:3]-[C:4](=[O:5])-[c:6]:[c:7]-  1.[O;H2;+0:8]>>[O:0]=[C:1]1-[C:2]:[C:3]-[C:4](-[O:5])(-[O:8])-[C:6]:[C:7]-1   SMIRKS14 : [O:0]=[c:1]1:[c:2]=[c:3]:[c:4](=[O:5]):[c:6]:[c:7]:1.[O;H2;+0:8]>>[O:0]=[c:1]1:[c: 2]=[c:3]:[c:4](-[O:5])(-[O:8]):[c:6]:[c:7]:1  SMIRKS 15 : [O:0]=[c:1]1:[c:2]:[c:3]:[c:4](=[O:5]):[c:6]:[c:7]:1.[O;H2;+0:8]>>[O:0]=[C:1]1- [C:2]:[C:3]-[C:4](-[O:5])(-[O:8])-[C:6]=[C:7]-1  ACCEPTED MANUSCRIPT SMIRKS 16 : [C;H0;D3;+0:0]-[S:1](- [O:2])(=[O:3])=[O:4]>>[C;H0;D3;+0:0].[S:1](-[O:2])(=[O:3])=[O:4]  SMIRKS 17 : [c:0]-[S:1](-[O:2])(=[O:3])=[O:4]>>[c:0].[S:1](- [O:2])(=[O:3])=[O:4]  SMIRKS 18 : [O:0]=[C:1]1-[C:2]=[C:3]-[C:4](=[O:5])-[C:6]=[C;D2;+0:7]-  1.[O;H2;+0:8]>>[O:0]=[C:1]1-[C:2]=[C:3]-[C:4](-[O:5])=[C:6]-[C:7]-1-[O:8]   SMIRKS 19 : [O:0]=[C:1]1-[c:2]:[c:3]-[C:4](=[O:5])-[C:6]=[C;D2;+0:7]-  1.[O;H2;+0:8]>>[O:0]=[C:1]1-[C:2]=[C:3]-[C:4](-[O:5])=[C:6]-[C:7]-1-[O:8]   ACCEPTED MANUSCRIPT SMIRKS 20 : [O:0]=[c:1]1:[c:2]=[c:3]:[c:4](=[O:5]):[c:6]:[c;H1;D2;+0:7]:1.[O;H2;+0:8]>>[O:0] =[C:1]1-[C:2]=[C:3]-[C:4](-[O:5])=[C:6]-[C:7]-1-[O:8]   SMIRKS 21 : [O:0]=[c:1]1:[c:2]:[c:3]:[c:4](=[O:5]):[c:6]:[c;H1;D2;+0:7]:1.[O;H2;+0:8]>>[O:0]= [C:1]1-[C:2]=[C:3]-[C:4](-[O:5])=[C:6]-[C:7]-1-[O:8]   SMIRKS 22 : [C:0]-[C:1](=[O:2])- [c:3]1:[c:4]:[c:5]:[c:6]:[c:7]:[c;H1;D2;+0:8]:1.[O;H2;+0:9]>>[C:0]/[C:1](- [O:2])=[C:3]1\[C:4]=[C:5]-[C:6]=[C:7]-[C:8]-1-[O:9]   SMIRKS 23 : [c:0]-[C:1](=[O:2])- [c:3]1:[c:4]:[c:5]:[c:6]:[c:7]:[c;H1;D2;+0:8]:1.[O;H2;+0:9]>>[C:0]/[C:1](- ACCEPTED MANUSCRIPT [O:2])=[C:3]1\[C:4]=[C:5]-[C:6]=[C:7]-[C:8]-1-[O:9]   SMIRKS 24 : [c:0]:[c:1](=[O:2]):[c:3]1:[c:4]:[c:5]:[c:6]:[c:7]:[c;H1;D2;+0:8]:1.[O;H2;+0:9]>>[C :0]/[C:1](-[O:2])=[C:3]1\[C:4]=[C:5]-[C:6]=[C:7]-[C:8]-1-[O:9]   SMIRKS 25 : [O:0]=[C:1]1-[C:2]=[C:3]-[C:4](=[O:5])-[C:6]=[C:7]-  1>>[O:0]=[c:1]1-[c:2]=[c:3]-[c:4]-[c:6]=[c:7]-1.[O;H2;+0:5]   SMIRKS 26 : [O:0]=[C:1]1-[c:2]:[c:3]-[C:4](=[O:5])-[C:6]=[C:7]-  1>>[O:0]=[c:1]1-[c:2]:[c:3]-[c:4]-[c:6]=[c:7]-1.[O;H2;+0:5]   ACCEPTED MANUSCRIPT SMIRKS 27 : [O:0]=[C:1]1-[c:2]:[c:3]-[C:4](=[O:5])-[c:6]:[c:7]-1>>[O:0]=[c:1]1- [c:2]:[c:3]-[c:4]-[c:6]:[c:7]-1.[O;H2;+0:5]    SMIRKS 28 : [O:0]=[c:1]1:[c:2]=[c:3]:[c:4](=[O:5]):[c:6]:[c:7]:1>>[O:0]=[C:1]1- [c:2]:[c:3]-[C:4]-[C:6]=[C:7]-1.[O;H2;+0:5]  SMIRKS 29 : [O:0]=[c:1]1:[c:2]:[c:3]:[c:4](=[O:5]):[c:6]:[c:7]:1>>[O:0]=[C:1]1- [c:2]:[c:3]-[C:4]-[C:6]=[C:7]-1.[O;H2;+0:5]  ACCEPTED MANUSCRIPT SMIRKS 30 : [O;v2;+0:0]-[c:1]1:[c:2]:[c:3]:[c:4](- [O;v2;+0:5]):[c:6]:[c:7]:1>>[O:0]-[C:1]1=[C:2]-[C:3]-[C:4](=[O:5])-[C:6]=[C:7]-1  SMIRKS 31 : [O;v2;+0:0]-[c:1]1:[c;D3;+0:2]:[c;D3;+0:3]:[c:4](- [O;v2;+0:5]):[c:6]2:[c:7]:[c:8]:[c:9]:[c:10]:[c:11]:1:2>>[O:0]-[C:1]1-[C:2]=[C:3]- [C:4](=[O:5])-[C:6]2-[C:7]=[C:8]-[C:9]=[C:10]-[C:11]=1-2    SMIRKS 32 : [O:0]=[C:1]1-[C:2]=[C:3]-[C:4](-[O:5])=[C:6]-[C:7]-  1>>[O:0]=[C:1]1-[C:2]=[C:3]-[C:4](=[O:5])-[C:6]-[C:7]-1   ACCEPTED MANUSCRIPT SMIRKS 33 : [O:0]=[C:1]1-[c:2]:[c:3]-[C:4](-[O:5])=[C:6]-[C:7]-  1>>[O:0]=[C:1]1-[C:2]=[C:3]-[C:4](=[O:5])-[C:6]-[C:7]-1   SMIRKS 34 : [O-:0]-[c:1]1:[c:2]:[c:3]:[c:4](-[O-  :5]):[c:6]:[c:7]:1>>[O;H0;+0:0]=[c:1]1-[c:2]=[c:3]-[c-:4]-[c:6]=[c:7]-1.[O;H2;+0:5]   SMIRKS 35 : [O:0]=[c:1]1:[c:2]:[c:3]:[cH-  :4]:[c:5]:[c:6]:1.[O:7]=[c:8]1:[c:9]:[c:10]:[cH-:11]:[c:12]:[c:13]:1>>[O:0]=[C:1]1-  [C:2]=[C:3]-[C;H1;+0:4](-[C;H1;+0:11]2-[C:12]=[C:13]-[C:8](=[O:7])- [C:9]=[C:10]-2)-[C:5]=[C:6]-1  ACCEPTED MANUSCRIPT    Figure S1. The simulation between the free energy and the redox potential in exper-iments. ACCEPTED MANUSCRIPT        Figure S2. Redox potential property relationships. The reduction potential depends on both the redox pattern (the relative positions of the carbonyl groups) and the functional groups. The numbers refer to the molecules of the redox pattern and correspond to the labels in Fig. 1. The median of each set of data is indicated with a black line. The regions ACCEPTED MANUSCRIPTbounded by the colours represent the range of the 5th and 95th percentiles of reduction potentials. The lines extend out to the maximum and minimum reduction potential for each substitution and ketone   Figure S3. The relationship between the predicted redox potential and the Gibbs free energy of Proto-desulfonation. ACCEPTED MANUSCRIPT      Figure S4. Effect of functional groups on stability based on core structure under Mi-chael addition mechanisms ACCEPTED MANUSCRIPT      Figure S5. Effect of functional groups on stability based on core structure under Gem- Diol Formation mechanisms. ACCEPTED MANUSCRIPT      Figure S6. Effect of functional groups on stability based on core structure under the Anthrone formation mechanism. ACCEPTED MANUSCRIPT      Figure S7. Effect of functional groups on stability based on core structure under the Tautomerization mechanism. ACCEPTED MANUSCRIPT      Figure S8. Effect of functional groups on stability based on core structure under the Disproportionation mechanism. ACCEPTED MANUSCRIPT      Figure S9. Effect of functional groups on stability based on core structure under the Dimerization mechanism. ACCEPTED MANUSCRIPT    Figure S10. The relationship between the solvation reaction Gibbs free energy and the reaction Gibbs free energy of the six degradation mechanisms. ACCEPTED MANUSCRIPT    Figure S11. Structures of literature compounds used for validation. Label colour indicates relative cycling stability reported experimentally (green: more stable; red: less stable); see Table S3 for conditions and performance metrics. ACCEPTED MANUSCRIPT    Figure S12. Experimental Validation of Theoretical Predictions: The relationship be-tween the stability and the redox potential with the structures of the experimental points. ACCEPTED MANUSCRIPTTable S2. A summary of 2D structures, SMILES representations, experimental redox potential(V versus SHE) of the compounds.  #  Molecule  SMILES 𝐸 (V 𝑒𝑥𝑝 vs. SHE)  Ref.  1   O=C1C(=O)C=C C=C1 0.831  [1]  2   O=C1C(=O)C=Cc  (c12)cccc2  0.547  [1]   3    O=C1C=CC(=O) C=C1   0.699   [1]   4    O=C1C=CC(=O)c  (c12)cccc2   0.470   [1]   5    c1cccc(c12)C(=O) c3c(C2=O)cccc3   0.090   [1]   6    c1cccc(c12)c3c(C (=O)C2=O)cccc3   0.442   [1]   7   c1cccc(C2=O)c1C (=O)c(c23)cc(cc3) S(=O)(=O)O   0.187   [2] ACCEPTED MANUSCRIPT   8   c1cccc(c12)C(=O) C(=C(C2=O)O)C C=C(C)C   0.333   [3]   9    c1cccc(c12)C(=O) C(O)=CC2=O  0.308   [3]   10   c1c(O)ccc(c12)C(  =O)c3c(C2=O)ccc (c3)O   0.039   [3]   11   Oc1c(O)ccc(c12) C(=O)c3c(C2=O) cccc3   0.005   [3]   12    OC1=CC(=O)C(O  )=CC1=O   0.382   [3]   13    O=C1C(F)=C(F)C (=O)C(F)=C1F   0.707   [3]   14    O=C1C(O)=C(Cl) C(=O)C(=C1Cl)O   0.394   [3]   15   O=C1C(Cl)=C(Cl  )C(=O)C(Cl)=C1  Cl   0.700   [3] ACCEPTED MANUSCRIPT    16     c1ccc(S(=O)(=O)  O)c(c12)C(=O)c3  c(C2=O)cccc3S(= O)(=O)O    0.206    [4]    17     c1ccc(S(=O)(=O)  O)c(c12)C(=O)c3  c(C2=O)c(S(=O)(  =O)O)ccc3    0.239    [4]    18     O=S(=O)(O)c(cc1  )cc(c12)C(=O)c3c  (C2=O)cc(S(=O)(  =O)O)cc3    0.228    [4]   19   O=S(=O)(O)c(cc1  )cc(c12)C(=O)C= CC2=O   0.534   [4]   20    O=C1C=CC(=O) C(O)=C1   0.594   [4]   21    CC1=CC(=O)C= CC1=O   0.641   [4]   22    O=C1C=CC(=O) C(Cl)=C1   0.71   [4] ACCEPTED MANUSCRIPT    23     O=c1c(=O)c(S(= O)(=O)O)cc(c2= O)c1c(=O)c(c23)c ccc3    1.21    [4]     24      c1cccc(c12)c(=O) c3c(c2=O)c(=O)c( SCCS(=O)(=O)O) c(c3=O)SCCS(=O )(=O)O     1.08     [4]    25     O=c1c(S(=O)(=O  )O)cc(=O)c(c2=O  )c1c(=O)c(c23)cc cc3    1.05    [4] ACCEPTED MANUSCRIPT    Table S3. Summary of experimental redox couples and performance metrics of AORFBs from literature  ID RFB  positive/electrolyte/negative pH Current density Cycle no. Coulombic efficiency Capacity fading rate Capacity retention at cycle no. Energy ef-ficiency at cycle no. Loss in discharge capacity Discharge capacity Charge efficiency 1[5] 2Cl-NQ/1 M  H2SO4/PTCDA(perylene- 3,4,9,10-tetracarboxylic dianhydride)    2.78 1 Ag-1 1800 100%  98.60%     2[6] 1,4-dihydroxyphenylsulfonate potassium (HQS)/Na2SO4/9,10- anthraquinone-2,7-disulfonic salt (2,7-AQDS) 7 60  mAcm-2 120 96.00%   56%    3[7] K4Fe(CN)6/KOH/1,4-dihydroxy- 2-carboxymethyl-9,10- anthra-quinone (1,4-CDHAQ) 13 40  mAcm-2 2500 97% 0.28% /  day 55% 82%    ACCEPTED MANUSCRIPT   ID RFB  positive/electrolyte/negative pH Current density Cycle no. Coulombic efficiency Capacity fading rate Capacity retention at cycle no. Energy ef-ficiency at cycle no. Loss in discharge capacity Discharge capacity Charge efficiency 4[8] K3Fe(CN)6([Fe(CN)6]3−/4)/KOH  / 1,5-dihydroxyanthraquinone(1,5- DHAQ-PAQS) 14 80  mAcm-2 500 99%  94.30% 48.00%    5[9] ferrocyanide/NaOH/ sodium 3,3′,3′′,3′′′-((9,10-anthraquinone- 2,6diyl)bis(azanetriyl))tetrakis(pro pane-1-sulfonate)(2,6-N-TSAQ) 14 40  mAcm−2 900 99.9 % 0.025%/d  ay        6a[5  ] 2,6-dimethyl-3,5-bis(morpholino methylene)benzene-1,4-diol (asym-O-5)/H2SO4/2,6-dimethyl- 3,5- bis(morpholinomethylene)benzene  -1,4-diol (asym-O-5) 1.54 20  mAcm-2 300  1.8%  /day      ACCEPTED MANUSCRIPT   ID RFB  positive/electrolyte/negative pH Current density Cycle no. Coulombic efficiency Capacity fading rate Capacity retention at cycle no. Energy ef-ficiency at cycle no. Loss in discharge capacity Discharge capacity Charge efficiency   6b[5  ] 2,5-dimethyl-3,6-bis(morpholino methylene)benzene-1,4-diol (O- 5)/H2SO4/2,5-dimethyl-3,6- bis(morpholinomethylene)benzene -1,4-diol (O-5) 1.54 50  mAcm-2 300  0.87%  /day        6c[5  ] 2,3,5-trimethyl-6-(morpholino methylene)benzene-1,4-diol (O- 1)/H2SO4/2,3,5-trimethyl-6- (morpholinomethylene)benzene- 1,4-diol (O-1) 1.54 50  mAcm-2 300  7.3%  /day        6d[5  ] 2,5-  bis(morpholinomethylene)benzene  -1,4-diol (O-3)/H2SO4/2,5- 1.54 50  mAcm-2 300  52.7%  /day      ACCEPTED MANUSCRIPT   ID RFB  positive/electrolyte/negative pH Current density Cycle no. Coulombic efficiency Capacity fading rate Capacity retention at cycle no. Energy ef-ficiency at cycle no. Loss in discharge capacity Discharge capacity Charge efficiency  bis(morpholinomethylene)benzene  -1,4-diol (O-3)           7a[1  0] a tetrasubstituted quinone contain-ing four sulfonated thioether sub-stituents/H2SO4/anthraquinone -2,7-disulfonate (AQDS) 2.75 50  mAcm-2 50    75% 2% 90% 100% 7b[1  0] 4,5-Dihydroxy-1,3- benzenedi-sulfonate/H2SO4/anthra quinone-2,7-disulfonate (AQDS) 2.75 50  mAcm-2 50    50% 48% 50% 95% 7c[1  0] para-Hydroxy benzenesulfonic acid/H2SO4/anthraquinone-2,7- disulfonate (AQDS) 2.75 50  mAcm-2 50    40% 52% 30% 95% ACCEPTED MANUSCRIPT   ID RFB  positive/electrolyte/negative pH Current density Cycle no. Coulombic efficiency Capacity fading rate Capacity retention at cycle no. Energy ef-ficiency at cycle no. Loss in discharge capacity Discharge capacity Charge efficiency 7d[1  0] 2,4-Dimethyl-3-sulfo-1,5- dihy-droxybenzene/H2SO4/anthraq ui-none-2,7-disulfonate (AQDS) 2.75 50  mAcm-2 50    70% 40% 20% 99.80% 8[11  ] ferrocyanide/KCl/2-2PEAQ 7 40  mAcm-2 2000 ≈100% 0.09%  /day       ferrocyanide/KOH/2-2PEAQ 14 40  mAcm-2 800 ≈100% 0.05%  /day      9[9] ferro-/ferricyanide//2,2′-((9,10- dioxo-9,10-dihydroanthracene-2,6- diyl)bis(oxy))dipropionic acid (2,6-D2PEAQ) 7 100  mAcm-2 170 99.70% 0.02%  /day      10[1  2] potassium ferrocyanide//4,4′- ((9,10-anthraquinone-2,6- 12 100  mAcm-2 880  0.01%  /day      ACCEPTED MANUSCRIPT   ID RFB  positive/electrolyte/negative pH Current density Cycle no. Coulombic efficiency Capacity fading rate Capacity retention at cycle no. Energy ef-ficiency at cycle no. Loss in discharge capacity Discharge capacity Charge efficiency  diyl)dioxy)dibutyrate (2,6- DBEAQ)           11[1  3] ferri/ferrocyanide//2,6-DPPEAQ, (((9,10-dioxo-9,10- dihydroanthracene-2,6- diyl)bis(oxy))bis(propane-3,1- diyl))bis(phosphonic acid) 9 100  mAcm-2 480 99.90% 0.014%  /day  65%    15[1  4] potassium ferrocyanide/NaOH/2,3- dihydroxylated anthraquinone(2,3- DHAQ) 14 50  mAcm-2 3000 100.00% 0.8% /  day  75%    19[1  5] ferrocyanide/NaOH/3-NH2-2- 2PEAQ 14 80  >98% 0.01% /  day  >70%    ACCEPTED MANUSCRIPT Reference:  [1] Wass JRTJ, Ahlberg E, Panas I, et al. Quantum Chemical Modeling of the Reduction of Quinones. The Journal of Physical Chemistry A. 2006;110(5). [2] Gerhardt MR, Tong L, Gómez‐Bombarelli R, et al. Anthraquinone Derivatives in Aqueous Flow Batteries. Advanced Energy Materials. 2016;7(8). [3] Wedege K, Drazevic E, Konya D, et al. Organic Redox Species in Aqueous Flow Batteries: Redox Potentials, Chemical Stability and Solubility. Sci Rep. 2016;6:39101. [4] Tabor DP, Gómez-Bombarelli R, Tong L, et al. Mapping the frontiers of quinone stability in aqueous media: implications for organic aqueous redox flow batteries. Journal of Materials Chemistry A. 2019;7(20):12833-12841. [5] Lu H, Yu J, Zhang J, et al. Designing p‐π Conjugated Naphthoquinone‐ Derivatives as High‐Performance Cathodes for Aqueous Proton Batteries. Advanced Functional Materials. 2025;35(24). [6] Yang G, Zhu Y, Hao Z, et al. An Aqueous All‐Quinone‐Based Redox Flow Battery Employing Neutral Electrolyte. Advanced Energy Materials. 2024;14(26). [7] Ozouf I, Fontmorin JM, Lebeuf R, et al. Long-cycling of a water-soluble quin-izarin derivative in redox flow batteries: Role of the cut-off voltage on the sta-bility. Electrochimica Acta. 2024;475. [8] Liu Y, Zhang P, Wu Z, et al. Biomimetic Naphthoquinone Zwitterion Deriva-tive with Water-Solubilizing Amino Acid Side Chain for High-Stability Aqueous Redox Flow Batteries. ACS Energy Letters. 2024;9(2):586-593. [9] Wu M, Bahari M, Jing Y, et al. Highly Stable, Low Redox Potential Quinone for Aqueous Flow Batteries**. Batteries & Supercaps. 2022;5(6). [10] Kwabi DG, Lin K, Ji Y, et al. Alkaline Quinone Flow Battery with Long Lifetime at pH 12. Joule. 2018;2(9):1894-1906. [11] Guiheneuf S, Godet-Bar T, Fontmorin JM, et al. A new hydroxyanthraquinone derivative with a low and reversible capacity fading process as negolyte in al-kaline aqueous redox flow batteries. Journal of Power Sources. 2022;539. ACCEPTED MANUSCRIPT[12] Ding Y, Li Y, Yu G. Exploring Bio-inspired Quinone-Based Organic Redox Flow Batteries: A Combined Experimental and Computational Study. Chem. 2016;1(5):790-801. [13] Wu M, Jing Y, Wong AA, et al. Extremely Stable Anthraquinone Negolytes Synthesized from Common Precursors. Chem. 2020;6(6):1432-1442. [14] Yang X, Garcia SN, Janoschka T, et al. Novel, Stable Catholyte for Aque-ous Organic Redox Flow Batteries: Symmetric Cell Study of Hydroqui-nones with High Accessible Capacity. Molecules. 2021;26(13). [15] Alfaraidi AM, Xi D, Ni N, et al. An Extremely Stable and Soluble NH2- Substituted Anthraquinone Electrolyte for Aqueous Redox Flow Batteries. ACS Applied Energy Materials. 2023;6(24):12259-12266.  ACCEPTED MANUSCRIPT