Molecular and Morphological Engineering of Organic Electrode Materials for Electrochemical Energy Storage

Organic electrode materials (OEMs) can deliver remarkable battery performance for metal-ion batteries (MIBs) due to their unique molecular versatility, high flexibility, versatile structures, sustainable organic resources, and low environmental costs. Therefore, OEMs are promising, green alternatives to the traditional inorganic electrode materials used in state-of-the-art lithium-ion batteries. Before OEMs can be widely applied, some inherent issues, such as their low intrinsic electronic conductivity, significant solubility in electrolytes, and large volume change, must be addressed. In this review, the potential roles, energy storage mechanisms, existing challenges, and possible solutions to address these challenges by using molecular and morphological engineering are thoroughly summarized and discussed. Molecular engineering, such as grafting electron-withdrawing or electron-donating functional groups, increasing various redox-active sites, extending conductive networks, and increasing the degree of polymerization, can enhance the electrochemical performance, including its specific capacity (such as the voltage output and the charge transfer number), rate capability, and cycling stability. Morphological engineering facilitates the preparation of different dimensional OEMs (including 0D, 1D, 2D, and 3D OEMs) via bottom-up and top-down methods to enhance their electron/ion diffusion kinetics and stabilize their electrode structure. In summary, molecular and morphological engineering can offer practical paths for developing advanced OEMs that can be applied in next-generation rechargeable MIBs.


Introduction of Organic Electrode
Materials (OEMs)

Brief History of OEMs Development
The announcement of the first commercial lithium-ion batteries (LIBs) by Sony Corporation in 1991 began the epoch of portable smart electronic devices. In the first generation of LIBs, LiCoO 2 (LCO) and graphite were used as the cathode and anode to deliver high specific capacities of approximately 274 mAh g −1 and 372 mAh g −1 , respectively [1][2][3]. To further improve the energy density, rate capability and safety of LIBs, over 30 years of substantial efforts from the research and development world have been devoted to the exploration of high-performance electrode materials, particularly inorganic electrode materials (IEMs), such as LiMn 2 O 4 [4], LiFePO 4 [5], LiNi II 1/3 Mn IV 1/3 Co III 1/3 O 2 (NMC) [6] cathodes, and silicon [7][8][9] and lithium titanate (LTO) [10] anodes. IEMs were first adopted for LIBs mainly due Zhenzhen Wu, Qirong Liu and Pan Yang have contributed equally to this work. to their commercial availability and apparent advantages, as shown in Table 1. In particular, IEMs can generally maintain structural and property stability in organic electrolytes, partly ensuring a reasonable lifespan. To date, numerous kinds of general charge/discharge mechanisms for IEMs have been proposed and verified, including Li-ion intercalation/extraction, alloying/dealloying, and redox conversion reactions for both cathodes and anodes [1]. In addition, large-scale metal mines for raw materials are readily available to achieve the massive production of IEMs. Finally, morphological engineering over the past 20 years has resulted in the availability of a wide spectrum of nanostructured IEMs, including 0D, 1D, 2D, and 3D nanomaterials, which has also propelled the rapid development of IEMs. However, the present IEMs also face a series of drawbacks.
1. The short lifetime and potential fire risk of IEMs can be attributed to the fact that they often have an inherently rigid structure that is not compatible with significant volume changes along with an irreversible phase transition during the charge/discharge processes. 2. The large-scale production of IEMs demands high consumption of unsustainable natural sources and energy during mining (such as Li, Co, and Ni) and high-temperature treatments. Such high carbon footprint activities can lead to irreversible damage and severe pollution to the ecological environment.
3. The limited amount of natural resources on Earth will not meet the nearly infinite demand of IEMs for largescale LIB applications, such as electric vehicles and smart grids.
With the increasing awareness of the economic and environmental costs of natural resource consumption, the design, production, and application of advanced energy storage devices will promote the growth of the global green economy and meet the requirements of global carbon neutrality in the near future. For this reason, we not only need to further improve the energy density and safety of LIBs but also need to adopt sustainable electrode materials for designing a new generation of LIBs. Organic electrode materials (OEMs) can address the above challenges; therefore, OEMs may play an important role in the next generation of LIBs and even other metal-ion batteries (MIBs), including sodium-ion batteries (SIBs), potassium-ion batteries (PIBs), zinc-ion batteries (ZIBs), and aluminium-ion batteries (AIBs).
Generally, OEMs consist of low-cost and sustainable nonmetallic elements (e.g. C, H, O, S and N) that can have multiple tuneable redox potentials to suit current LIB charge/ discharge processes. Accordingly, OEMs can be classified into six types of compounds ( Fig. 1), i.e. carbonyl OEMs, sulphur-containing OEMs, nitrogen-containing OEMs, conducting polymer OEMs, radical-based OEMs, and overlithiated OEMs. The first OEM-based LIB was reported in 1960, approximately three decades before the commercialization of IEMs (Fig. 1). From the viewpoint of the electrochemical performance and properties of OEMs, it can be concluded that the development history of OEMs can be separated into two paths based on the enhancement of the energy density, i.e. raising the redox potential (the upper part of Fig. 1) and boosting the charge transfer number (the lower part of Fig. 1).
In the first path, electrode materials with a high operating potential are used to advance the energy density of MIBs. The first inherent advantage of OEMs lies in the fact that the molecular structure of OEMs can be designed to tune their redox potentials of OEMs and, therefore, the potential output of the relevant MIBs (see the upper part of Fig. 1). Carbonyl OEMs were the first type of OEM to confirm the mechanistic viability of this approach. In fact, via the introduction of functional groups, the redox potential of the carbonyl OEMs can be tuned between 1.7 and 3.2 V (vs. Li/Li + ), such as dichloroisocyanuric acid (DCA, No. 1) [11], quinone (No. 6) [12], polyimide (No. 10) [13], pyrene 4,5,9,10-tetraone (PYT, No. 12), and cyclohexanehexone (No. 14) [14]. As an example, the redox potential of DCA [11] can be increased to 3.2 V (vs. Li/Li + ) due to the introduction of three chlorine atoms. Conjugated dicarboxylate (No. 15) [15] has a lowvoltage redox potential of 0.8 V because of the aromaticity of its core and resonance of the carboxylates. Overlithiated OEMs, such as 1,4,5,8-naphthalenetetracarboxylic dianhydride (NTCDA,No. 16) [16], polyazaacene analogue (PQL, No. 17) [17], imine-based COFs (No. 18) [18], multicarbonyl polyimides (PMTA,No. 19) [19], and polyimide Schiff base No. 20) [20], can achieve an obvious anodic working plateau between 0 and 1.6 V (vs. Li/Li + ). In addition, the representative sulphur-containing OEM tetraethylthiuram disulphide (TETD, No. 7) [21] delivers a high voltage of approximately 3.0 V (vs. Li/Li + ). Nitrogen-containing OEMs also have a wide working potential between 1.55 and 3.00 V, such as 7,7,8,No. 4) [22], 5,6,11,12,17,18-hexaazatri-naphthylene (HATNA, No. 11) [23], and azo compounds (No. 13) [24]. It is interesting that p-type OEMs, including conducting polymer OEMs and radical-based OEMs, work at a high potential above 3.5 V (vs. Li/Li + ). In 1981, through iodine doping, the conductivity of the conducting polymer polyacetylene (PA, No. 2) was enhanced by approximately 10 orders of magnitude, showing that PA was an excellent OEM with a redox potential of 3.95 V (vs. Li/Li + ) [25,26]. Later, conducting polymers continued to be used as OEMs at high voltages, e.g. from 2.50 to 3.95 V (vs. Li/Li + ) for polyacetylene (No. 3), from 3.0 to 4.3 V (vs. Li/Li + ) for polythiophene (No. 3) [27], from 2.50 to 4.01 V (vs. Li/Li + ) for polyaniline (No. 5) [28], and from 2.0 to 3.5 V (vs. Li/Li + ) for polypyrrole (No. 8) [29]. However, conducting polymers encounter severe barriers in delivering ample capacity in charge/discharge processes because of the limited doping level, which is typically below 50% [26,30]. In 2002, poly(2,2,6,6-tetramethylpiperidinyloxy methacrylate (PTMA, No. 9), a radical OEM associated with a 2,6,6-tetramethyl-1-piperidinyloxy (TEMPO) redox unit, delivered a high potential plateau at approximately 3.6 V (vs. Li/Li + ) and a high voltage range from 2.5 to 4.3 V (vs. Li/Li + ). Additionally, PTMA exhibited a low theoretical capacity of approximately111 mAh g −1 based on a one-electron transfer reaction with anion insertion [31]. After that, in 2012, the above OEM exhibited another voltage range of from 2.5 to 3.2 V (vs. Li/Li + ) based on the secondstage electron transfer of PTMA with a reversible cation (e.g. Li + ) insertion through the construction of an improved electronic conductive network [32]. The strategies for tuning the redox activity of OEMs by molecular engineering, such as tailing the redox potential by grafting functional groups, are reviewed systematically in Sect. 2. In the second path, the charge transfer number (including the electron, cation, or anion transfer number during the energy storage reaction) of electrode materials is another parameter that determines the specific capacity. Through molecular and morphological engineering of OEMs, an increased number of active sites can be realized, thereby increasing the charge transfer number (see the lower part of Fig. 1), which leads to an increased theoretical capacity [33,34]. This roadmap of OEMs implies that their development with more redox electrons can be a promising research direction. Conducting polymers and radical OEMs usually undergo 0-2 electron reactions with relatively low theoretical capacity, while carbonyl, sulphur-containing, nitrogen-containing, and overlithiated OEMs can exhibit a larger number of active sites or redox electrons (from 2 to 22). In 1969, DCA, the first reported carbonyl OEM, could perform a 4-electron transfer in a redox reaction [11]. Carbonyl functional groups became popular in OEMs because of their high reactivity and theoretical capacity of approximately 957 mAh g −1 per C=O unit. Later, significant attention was devoted to increasing the number of redox-active carbonyl groups. Carbonyl compounds, including monomeric carbonyl compounds No. 6 [12], No. 12 [35] and No. 14 [14], equipped with two, four, and six carbonyl groups delivered increasing theoretical capacities of 257 mAh g −1 , 409 mAh g −1 , and 957 mAh g −1 , respectively. In 1988, Visco et al. bestowed a sulphur-containing OEM (No. 7) with high redox activity at the sulphur-sulphur bond (S-S) [21] by using the cleavage and regeneration of covalent disulphide (S-S) bonds to enable a 2-electron reaction when forming sulphide salts. Later, various molecular designs were developed to covalently bind poly(sulphide) s (-S x -) into π-cyclic backbones and achieve a higher number of S-S bond reactions and a corresponding increase in the number of electron transfers [36]. Nitrogen-containing groups, including highly redox-active C≡N, C=N, and N=N bonds, provide another approach to improve the diversity and high charge storage capacity of OEMs. No. 4, a representative nitrogen-based electron-poor molecule incorporated with an unsaturated carbon-nitrogen bond, can transfer up to 4 electrons [22], while No. 11, a Schiff base with a carbon-nitrogen double bond (C=N) and π-conjugated aromatic compounds, can realize the transfer of 6 electrons in a redox energy storage reaction [23]. The hexaazatriphenylene (HAT) shown in the No. 11 molecule is an electron-deficient aromatic system with a planar and rigid structure, playing a critical role as a scaffold for fabricating HAT-based π-conjugated microporous polymers (CMPs) [37][38][39]. These π-conjugated CMPs create a new platform for higher energy storage because of their porous structure and multiple built-in redox-active sites. In 2018, a new family of OEMs containing active azo groups (No. 13) was adopted in LIBs [24]. These azo functional groups pave another way to reversibly use the nitrogen-nitrogen double bond (N=N) to store two lithium ions with a high initial Coulombic efficiency (ICE). Furthermore, overlithiated OEMs are a promising strategy for a large number of redox sites to store metal ions. The polymerization of No. 10 [13] and No. 16 [16] produces inherent conjugated structures that allow overlithiation at the conjugated aromatic rings at low working voltages. Sun et al. [16] was the first to employ a naphthalene-based multiring, aromatic compound (No. 16), illustrating the possibility of electrochemically adding 6 lithium ions to the fused C6 ring. A large π-conjugated ladder polymer (No. 17) realizes a storage mechanism for 12 lithium ions [17]. Larger amounts of Li + -accessible OEMs have also been realized with similar conjugated aromatic compounds, such as No. 18 (14 electrons) [18], No. 19 (22 electrons) [19], and No. 20 (12 electrons) [20].

Advantages of OEMs
It is widely recognized that OEMs have a series of merits in comparison with IEMs (Table 1) [40].
1. Most OEMs can be produced from naturally sustainable resources (such as plants, microorganisms, and animal products) in low energy consuming fabrication processes, thereby achieving a smaller carbon footprint. 2. The entrenched relationships between the electrochemical properties (such as the redox potentials) and functional groups (such as electron donating and attraction of functional groups) allow us to tune the energy storage performance of OEM-based LIBs, including the output voltage, specific theoretical energy density and power capability, by grafting the functional groups onto the stable polymer backbones. Taking TCNQ (No. 4) as an example, in 2013, Morita et al. studied the voltage changes after the substitution of methyl groups and fluorine atoms with TCNQ molecules [41]. The methyl group, an electron donor group, enabled the donation of the electron cloud to the C≡N redox centre, leading to an easier way to extract the electron from the C≡N bond and lowering the output voltage by approximately 0.1 V. In contrast, the organic electron acceptor (fluorine atom) had the opposite effect towards the redox centre, increasing the voltage by approximately 0.2 V. 3. OEMs often have a soft molecular skeleton, which is suitable for flexible electrodes and devices. For example, belt-shaped flexible batteries incorporated with OEMs have been fabricated and illustrated to have good electrochemical performance [42,43]. 4. OEMs can universally accommodate monovalent metal ions, such as Li + , Na + , and K + , and multivalent metal ions, such as Zn 2+ , Mg 2+ , Ca 2+ and Al 3+ [44].

Challenges of OEMs
According to the molecular weight (MW), OEMs can be classified into two main types: small-MW OEMs with short molecular chains and large-MW OEMs with long molecular chains, or more typically, polymers. The latter is often designated as OEM polymers since it has repeated functional units in the long main chain and side chain. OEMs have many critical challenges ( Table 1) that must be addressed in the course of their commercialization. OEMs, especially small-MW OEMs, commonly show poor thermal stability and significant solubility at room temperature. The chemical and electrochemical stability in existing Li-based organic electrolytes is another indispensable parameter that affects the suitability of MIB applications. The high solubility in aprotic electrolytes can dramatically reduce the cycling life of the resultant LIBs, while the poor electronic conductivity of OEMs (except for those with conducting polymers) is another barrier that must be addressed during electrode fabrication to fully utilize the redox sites in the charge/discharge processes. High-degree self-polymerization of polymeric carbonyl materials based on quinone and PYT was performed to address the solubility issue in aprotic electrolytes, leading to improved stability and high capacity [45][46][47][48][49][50]. Moreover, due to the characteristics of polymerization (such as homogeneous chain reactions), the morphology of OEMs is controlled in a much more complicated way compared to the scenarios for synthesizing IEMs. As a result, compared to those of IEMs, the reported morphologies of OEMs are limited, and most of the obtained OEMs are either in powder or bulk form. Finally, the charge/discharge mechanisms, i.e. the general redox schemes of OEMs that will specifically be introduced in this review, are still unclear, vague, and even controversial.

General Strategies of OEMs Development
This work reviews the history, strategies, progress, and achievements and provides prospective views and solutions in addressing the challenges of OEMs and promoting their merits, as concluded in Table 1. Following the historical OEM development trend in the roadmap (Fig. 1), we systematically summarize the strategies of molecular and morphological engineering to boost the electrochemical performance of OEMs in MIBs. The particular aims and operation of the design strategies are summarized as follows.
1. To design a new controllable polymerization reaction to increase the molecular weight of OEMs, thereby addressing their dissolution and stability. 2. To improve redox potential via the adoption of suitable functional groups in a conjugated organic structure. 3. To increase the number of active functional group sites to store more ions and allow a higher electron transfer number. 4. To increase the surface area for storage and accelerate the mass transport of ions and to establish nanostructures that enhance the conduction of electrons in OEMs by morphologically engineering them.
With such coherent efforts, it can be envisaged that the next generation of OEMs for MIBs will eventually achieve a large energy capacity, high power capability, and long cycling life.

Fundamental Redox Mechanism of OEMs
As mentioned above, electroactive OEMs are promising for next-generation sustainable energy storage systems via various electrochemical redox reaction mechanisms [51][52][53][54][55][56][57]. Based on the abilities of OEMs in a neutral state to accept or release electrons during electrochemical processes, OEMs can be categorized into three types: the n-type, the p-type, and the bipolar-type [54,58,59]. As shown in Fig. 2, OEMs have one or more specific redox centres in their organic molecule structure that primarily play the dominant role in their redox mechanism.

Fig. 2
Electrochemical redox reaction mechanisms of OEMs and their electrochemical cell configurations. a The n-type reactions are based on the reversible intercalation/deintercalation of lithium ions (Li + ) into an organic skeleton, where lithium terephthalate is the example. These OEMs are amenable as an anode (charge first) or cathode (discharge first) in a battery. b The p-type reactions revers-ibly occur with bulk anions (B − ), such as the example with diamine radicals, and these OEMs are applied as a cathode (charge first). c Bipolar-type reactions have two oxidative states with the insertion of cations (A + ) or anions (B − ), such as the example with TEMO-based radicals, and these OEMs are employed as a cathode
Generally, n-type OEMs include carbonyl compounds (-C=O), imine compounds (-C=N-), azo compounds (-N=N-), cyano derivatives (-C≡N), organosulphur compounds (-S-S-), conducting polymers (-C=C-), etc. They usually undergo different bond reactions and intramolecular electron transfer in the redox process. During the reduction process, electroactive carbonyl groups convert into enolate monoanions (-C-O-) and then react with active cations. In the oxidation process, a reverse enol-ketone transformation occurs at the carbonyl active sites. For imine compounds [18,[77][78][79], their electroactive imine groups can reversibly accept/donate electrons during the redox process. Simultaneously, the cations combined or left from the functional groups ensure the charge balance between imine groups and active cations. Similar redox chemistries of intramolecular electron transfer can also be authenticated for different electroactive groups in other n-type OEMs, such as the cyano groups in cyano derivatives and carbon bonds in conducting polymers [71,80]. Unlike the above mechanism, the disulphide groups in some organosulphur compounds [81][82][83][84] are subject to breakage and form thiolates during the reduction reactions and reversibly regenerate disulphide bonds in the subsequent oxidation process. In addition, the azo groups in azo compounds [85,86] undergo distinct redox chemistry, namely a direct reaction between the azo groups and cations without intramolecular electron transfer.

p-Type OEMs
The p-type OEMs tend to reversibly release electrons and be oxidized into positively charged states, accompanying their combination with active anions (PF 6 − , ClO 4 − , FSI − , CF 3 SO 3 − , etc.), when charging batteries [87][88][89][90][91]. Generally, p-type OEMs have higher redox potentials because of the relatively low molecular energy levels formed via their transformation from a neutral state to a positively charged state. In contrast, n-type OEMs have higher molecular energy levels when in their negatively charged state. Thus, p-type OEMs have higher operating voltages than n-type OEMs when they work as cathodes in batteries. Because the redox chemistry of p-type OEMs is associated with the combination of active anions instead of conventional cations, the corresponding batteries based on p-type OEMs are also well known as dual-ion batteries (Fig. 2b) involving anion-cation dual-carrier chemistry [92][93][94].
According to previous works, p-type OEMs typically include amine derivatives (-NH-), thioethers (-S-), organic radicals (e.g. -N-O-), and conjugated N-heterocycles (e.g. = N-). Especially for amine-based derivatives [93,95,96], during oxidation reactions, neutral amine groups release electrons to form positively charged N cations while also bonding with active anions. Conversely, the corresponding reduction reactions undergo reversible electron acceptance and anion disassociation. Thioethers are organosulphur compounds but present different redox schemes than disulphides and polysulphides [52,84,97]. Generally, thioethers endow thioradicals (-S-) with redox chemistry, releasing and accepting electrons during the charge/discharge processes, respectively. In addition, radical organic compounds [89,98], mainly nitroxyl radicals, can be oxidized to positively charged oxoammonium radicals and combined with anions during charging. Reverse reduction chemistry occurs in the corresponding discharge process. Finally, the conjugated N-heterocycles [64,96,99] can store and convert energy relating to intramolecular electron transfer at high charge/ discharge potentials. The N atoms are capable of losing one electron, thus generating sp 2 -hybridized ions that bond with anions. The other remaining electron can be shared with an adjacent carbon. However, the p-type reactions of conjugated N-heterocycles generally result in relatively low specific capacities.

Bipolar-Type OEMs
Bipolar-type OEMs in their neutral state combine the merits of n-type and p-type OEMs because they can form negatively and positively charged states depending on the operating potential.
In the reverse redox process, these negatively (or positively) charged OEMs can be reduced/oxidized to their initial neutral state and even to positively (or negatively) charged states by attracting and releasing electrons. This redox process is achieved via successive two-step reduction/oxidation reactions accompanying cation (anion) dissociation and subsequent anion (cation) combination. Accordingly, bipolar-type OEMs can be used as anodes and cathodes in conventional rechargeable batteries and dualion batteries (Fig. 2c). [89][90][91] Furthermore, when paired with appropriate anodes, they can be assembled into hybrid batteries (Fig. 2c) in conjunction with the "rocking-chair" cation shuttle characteristic (stage I) and anion-cation dualion mechanism (stage II) [100,101].
Due to the features of the bipolar charging process, bipolar-type OEMs usually have two compositions: free radicals with good electron-withdrawing and electron-donating capabilities [89] and implants with different functional groups that individually undergo n-type and p-type reactions [91]. Typically, these OEMs include radical polymers (e.g. nitronyl nitroxide) and conducting polymers (e.g. polyaniline and polypyrrole). Taking conducting polymers as an example, they participate in redox chemical reactions based on delocalization along their main chain and then incorporate anions and cations during oxidation and reduction. In addition, for radical compounds, when neutral radical polymers undergo p-type and n-type reactions, their free radicals, such as nitroxyl radicals, can be oxidized to positively charged oxoammonium cations and reversibly reduced to negatively charged aminoxy anions [32,89]. Interestingly, these electrochemical redox processes are carried out without intramolecular electron transfer.

Classification of OEMs
Since the first demonstration of OEMs in 1969, a number of organic materials containing diverse electroactive organic functions have been successfully exploited for electrochemical energy storage. Based on electroactive redox centres, OEMs can be classified into six categories (Table 2), including carbonyl compounds, sulphur-containing OEMs, nitrogen-containing compounds, conducting polymers, radicalbased polymers, and overlithiated compounds [51,52,56,96,99,102]. They show variable electrochemical properties at the material level, including their operating potential, specific capacity, conductivity, and cycling stability, which will be introduced in the following section.
Nevertheless, carbonyl compounds, particularly those with low molecular mass, present severe dissolution issues in aprotic electrolytes because of their facile interaction with electrolyte solvents [50,127,128]. In addition, their discharge products can dissolve in aqueous electrolytes even though carbonyl compounds have low solubility in water solvents. Accordingly, these electrodes based on carbonyl compounds deliver high electrochemical irreversibility and fast capacity fading during charge/discharge processes [70]. For example, although BQ features a high theoretical capacity via a two-electron reaction, it generally undergoes a severe dissolution problem, which correlates with its relative polarity, donor number, and intermolecular interaction energy [110,111]. 9,10-Anthraquinone (AQ) is another typical carbonyl compound with good electrochemical redox reversibility and high theoretical capacity; however, its one main challenge is its inferior cycling performance due to its serious solubility in electrolytes [129,130]. Yang et al. [129] observed the poor cycling performance of AQ with less than 50% capacity retention after 80 cycles.
The low electronic conductivity of carbonyl OEMs is another hurdle that leads to low energy densities and rate capabilities [20,99]. In the preparation of OEM electrodes, the extra addition of a conductive agent, e.g. up to 50 wt% carbon black (wt% means the weight percentage), is needed, which inevitably lowers the specific capacity of the electrode and resultant energy density [131]. Thus, the current studies

Sulphur-Containing OEMs
The low specific capacity of early conductive polymer cathodes called for the exploration of alternative OEMs with high capacity. Sulphur-containing OEMs are defined as compounds including redox-active sulphur atoms, such as disulphides, polysulphides and thioethers; furthermore, these promising OEMs are inexpensive, environmentally friendly, and biodegradable. The bond length of the S-S bond is longer than that of the C-S bond, indicating that the bond energy of the S-S bond is smaller than that of the C-S bond. Thus, disulphides and polysulphides are based on the reversible breakage and regeneration of S−S bonds that undergo n-type two/ multielectron redox, making them capable of delivering KGaA, Weinheim. c Imides: redox mechanisms of (N,N′-bis(2-anthraquinone))-perylene-3,4,9,10tetracarboxydiimide (PTCDI-DAQ). Reproduced with permission form Ref. [122]. Copyright 2020, Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim. d Anhydrides: the simulated PTCDA unit cell incorporating two H 3 O + ions and a comparison of the charge/discharge profiles for the storage of Na + ions at 1 A g −1 , K + ions at 0.5 A g −1 , and hydronium ions (the 10th cycle) at 1 A g −1 in PTCDA. Reproduced with permission from Ref. [125]. Copyright 2017, Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim higher capacity [82,84,132]. In contrast, thioethers are subject to the reversible p-type reaction between C−S−C and C−S + −C because the large atomic radius of the sulphur atoms has less binding force for electrons outside the nucleus. Hence, thioether cathodes are prone to bind with active anions from electrolytes when charging and deliver relatively high discharge potentials [133,134]. In addition, there is no breakage or regeneration of S−S bonds during the electrochemical redox reaction of thioethers; thus, it is expected to present better cycling stability than disulphides and polysulphides. In addition, owing to the existence of the π-electron delocalization effect between the aromatic rings and S atoms with a lone pair of electrons, the thioether bond can contribute to faster electron transfer. Since Visco et al. reported the first n-type sulphur-containing OEMs in 1988 [21], increasing efforts have been devoted to developing organosulphur compounds as cathode materials. For example, Fu et al. [135] developed an organotrisulphide cathode material (DMTS, CH 3 −S−S−S−CH 3 ) with a four-electron redox reaction mechanism, which delivered an initial capacity of up to 720 mAh g −1 (Fig. 4a). Ren et al. synthesized two thianthrene-containing poly(phenylacetylene)s, in which thianthrene groups with C−S−C bonds served as redox-active centres. Both polymer cathodes could deliver high discharge potentials of 4.1 V versus Li/Li + , unfortunately, their specific capacities were below 100 mAh g −1 (Fig. 4b) [134]. Some thioethers, involving a four-electron reaction mechanism of "thioether-sulfoxide-sulfone", are capable of obtaining high specific capacities ranging from 500 to 800 mAh g −1 [136].
Disulphides and polysulphides usually demonstrate poor cycling performance due to the dislocation of broken S-S bonds and the severe dissolution of these organosulphides and their products in aprotic electrolytes. When discharging, the products resulting from the breakage of the -S-S-bonds in sulphides commonly have lower molecular weights and are more soluble in electrolytes [84,135]. For example, poly(2,5-dimercapto-1,3,4-thiadiazole) (PDMcT), a typical polymeric organodisulphide, has a high theoretical capacity of 362 mAh g −1 [137]. Due to the breakage of S-S bonds in the main chains, the polymer in the cathode is converted into a small organic molecule of DMcT 2− when discharging, which undergoes severe dissolution in the electrolyte and causes poor cycling stability. Additionally, Wang et al. [81] improved the cycling performance of organodisulphides via the substitution of N-containing heterocycles for aromatic rings in diphenyl disulphide (DPDS) (Fig. 4c). Compared with DPDS, the designed 2,2′-dipyridyl disulphide (2,2′-DpyDS) showed superior cycling performance due to the coordination of the N/ Li/S bridge lowering the solubility of the discharge product of 2,2′-DPyDS. In the case of thioethers, the unusual redox reaction mechanism involving the formation of sulphoxide and sulphone groups generally results in poor cycling performance after dozens of cycles.
In addition, organosulphur compounds suffer from low electronic conductivity and thus sluggish kinetics for energy storage applications. For example, only when the temperature is above 100 °C can DMcT be utilized for the reversible storage of Li + [138]. Considering the poor conductivity of organosulphur electrode materials and the high solubility of small-molecule products, researchers have developed a type of organodisulphide (poly(2,2′-dithiodianiline), PDTDA) consisting of polymeric main chains and side chains that contain S-S bonds [138]. The polymeric main chains are generally insoluble and electronically conductive, which boosts cycling performance. Nevertheless, the dislocation of different chains in the discharged state gives rise to a low rebonding efficiency of the broken S-S bonds on the side chains. To surmount this problem, S-S bonds are designed to decorate the sides of the same chains and enhance the rebonding capability. Deng et al. [139] developed such an organosulphur compound, namely poly(5,8-dihydro-1H,4H-2,3,6,7-tetrathia-anthracene) (PDTTA), which markedly presented a high specific capacity of 422 mAh g −1 and enhanced cycling performance. However, owing to the intrinsically slow kinetics of organosulphur compounds, their electrochemical properties need to be further improved.

Nitrogen-Containing OEMs
It is well known that reversible electrochemical redox reactions are prone to occur on organic moieties with conjugated structures or at redox-active heteroatom centres (e.g. N, S, and O) with a lone pair of electrons. These conjugated structures can facilitate electron transfer during electrochemical redox reactions and the charge delocalization of redox products. Regarding redox-active heteroatom centres, their intrinsic lone-pair electrons bestow them with higher redox activity and enhanced electronic conductivity [78,136,140]. Compared with S and O heteroatoms, N heteroatoms exhibit several unique characteristics and different electrochemical behaviour. On the one hand, N atoms have strong bonds with adjacent C/O/N atoms so that the breakage of N-C/N-O/N-N bonds rarely takes place during redox reactions, thereby ensuring good electrochemical reversibility. On the other hand, compared with C atoms, the N atoms of saturated amines uniquely feature sp 3 hybridization and have one lone pair of electrons, which can be shared with a cation to produce quaternary ammonium ions. Interestingly, the N atoms allow the loss of one electron and interact with one anion [102]. Such behaviour indicates bipolar electrochemical character. N-containing compounds are defined as OEMs in which their electrochemical redox-active centres contain N [96]. N-containing compounds can be divided into various types according to their different redox-active centres: imine compounds, containing N=C bonds; cyano derivatives, containing N≡C bonds; azo compounds, containing N=N bonds; and arylamine compounds, containing arylamine groups.
Imine compounds, which contain redox-active N=C bonds, have served as OEMs for electrochemical energy storage since 2014 [141,142]. The redox reaction mechanism is associated with single/multielectron redox chemistry, namely the conversion of C=N bonds into C-N bonds, intramolecular electron transfer in the conjugated structure, and insertion of active ions (e.g. Li + , Na + ); notably, this mechanism corresponds to the acceptance of 1 or 1.4 electrons per C=N group [141,142]. Generally, redoxactive C=N centres in common aromatic groups, except for strong electron-deficient groups such as pyrazine rings, show relatively low redox potentials, implying that they are promising anode materials. Imine compounds, typically including pyrazinyl compounds [140], triazinyl compounds [100,143], Schiff bases [142], and pteridine derivatives [144], are characterized by tunable electrochemical redox activity, which is strongly dependent on their conjugated structure and geometric plane structure [79,145]. For example, a phenazine molecule consisting of pyrazine bonding with two aromatic rings is usually subject to an n-type reaction of active cation insertion, in which the N atoms are redox-active centres. During the n-type reaction process of phenazine at moderate potentials (generally below 2.0 V vs. Li/Li + ), there is the transformation of C=N bonds into C-N bonds and the simultaneous formation of N-M bonds (M denotes active metal cations) [146]. Such an electrochemical process can yield N-substituted phenazine with p-type character, which can undergo a p-type reaction to produce positively charged N atoms accompanied by the re-formation of N=C without the breakage of N-M (nitrogen-metal cation) bonds at relatively high potentials (above 3.0 V vs. Li/Li + ) [145]. The positively charged state of the N atoms is stable due to Electrochemical performance of some typical nitrogen-containing OEMs. a Imine compounds: structural evolution of 3Q during lithiation. Reproduced with permission from Ref. [78]. Copyright 2017, Macmillan Publishers Limited, part of Springer Nature. b Cyano derivatives: the charge/discharge profile of DCA. Reproduced with permission from Ref. [151]. Copyright 2016, Elsevier B.V. c Azo compounds: c 1 charge/discharge curves, c 2 rate capability, and c 3 cycling performance of ADALS. Reproduced with permission from Ref. [85]. Copyright 2018, Proceedings of the National Academy of Sciences. d Arylamine compounds: d 1 long-term cycling stability of a potassium-ion battery with a PTPAn cathode (Reproduced with permission from Ref. [155]. Copyright 2017, American Chemical Society.) and d 2 charge/discharge characteristics of potassium-ion batteries with a p-DPPZ cathode. Reproduced with permission from Ref. [156]. Copyright 2019, American Chemical Society the increased conjugation and balanced anions stemming from the electrolyte. However, some high-capacity pyrazinyl compounds usually exhibit high solubility in electrolytes, resulting in poor cycling performance [140]. For example, small-molecule hexaazatrinaphthalene (HATN) with six C=N groups per unit has a high specific capacity but exhibits fast capacity fading (Fig. 5a) [78,140] and that generally restricts its practical application. Polymer HATN (PHATN) has been proposed to limit the solubility and simultaneously boost electronic conductivity [147]. As a result, PHATN not only presents enhanced cycling performance but also achieves fast reaction kinetics and a high rate capability [79,140,148]. Triazinyl compounds generally consist of polymerized triazine rings with different linkers (e.g. benzene) [100] and share a similar redox mechanism with pyrazinyl compounds. Interestingly, in comparison with pyrazinyl groups, triazinyl groups theoretically have a higher reversible capacity. Additionally, the porous structure of polymeric triazinyl compounds results in fast ionic transfer, and thus, a high rate capability.
Nevertheless, it is noteworthy that a wide voltage window is necessary for triazinyl compounds to fully deliver their specific capacities because of their sloping potentials during the charge/discharge processes. In addition, regarding pyrazinyl and triazinyl compounds, their working potentials for n-type redox reactions are located in the range of 1-2 V vs. Li/Li + ; however, their capacities for p-type redox reactions are still insufficient even though they can achieve relatively high redox potentials. Schiff bases with a configuration of R 1 -N=CH-R 2 (R 1 and R 2 typically denote aromatic groups) can undergo a two-electron redox reaction at 1 V versus Li/Li + , implying that they are a potential anode material for batteries. Compared with monomeric Schiff bases, polymeric Schiff bases feature low solubility, high thermal stability, and mechanical strength. Commonly, polymeric Schiff bases, consisting of the Hückel-stabilized repeating unit of -N=CH-Ar-HC=N-(Ar denotes an aromatic group) with ten-π-electron electrochemical activity, can accommodate more than one Na + per redox-active -C=N-bond. The repeating unit complies with the Hückel rule of aromaticity, i.e. cyclic coplanar units with (4n + 2) π electrons (n is a positive integer) have extra stability [141,149]. Accordingly, the existence of inactive Ar aromatic groups is highly conducive to boosting the stability of the Schiff-base structure; otherwise, the Schiff bases present poor cycling performance. It should be emphasized that the isoelectronic inverse unit of -CH=N-Ar-N=HC-is electrochemically inactive because the strong interaction between N atoms and the neighbouring aromatic ring results in the loss of its planarity. Pteridine derivatives are also a typical classification of imine compounds. Distinct from other imine compounds, pteridine derivatives have biologically occurring redox centres of conjugated diazabutadiene moieties, which are capable of facilitating proton-coupled electron transfer reactions even when the N atoms have been reduced [144]. Flavin, a typical pteridine derivative, can reversibly store two electrons and two balancing active cations per molecule, in which the n-type redox reaction reversibly occurs at the conjugated diazabutadiene moiety of the isoalloxazine ring, accompanied by the reversible conversion of the isoalloxazinic structure to an alloxazinic structure, via the two successive protoncoupled single-electron transfer processes of N5 and N1 [144,150]. Note that, the carbonyl groups in flavin do not work as redox-active sites.
Cyano derivatives were reported as cathode materials for lithium-ion batteries in 1984 [22]. 7,7,8,8-Tetracyanoquinodimethane (TCNQ) is a typical cyano derivative. Its redox reaction mechanism involves two-step n-type reactions in which TCNQ is converted into TCNQand then into TCNQ 2followed by the insertion of Li + . The corresponding potential is approximately 3.0 V versus Li/Li + , which is among the highest values for n-type OEMs. Further reducing the operating potential to below 1 V versus Li/Li + will possibly cause an irreversible electrochemical redox reaction relating to the formation of LiCN. In addition, some cyano derivatives are composed of aromatic rings decorated with cyanide groups, which usually deliver relatively low redox potentials of 1.0-2.0 V (Fig. 5b) [151]. However, these cyano derivatives also suffer from high solubility in electrolytes, which results in poor electrochemical reversibility.
Azo compounds are an emerging type of OEM, where azo (N=N) groups act as redox-active centres for reversible electrochemical redox reactions; these compounds were first reported by Luo et al. [24,85]. Azo compounds undergo n-type reactions, where N=N bonds are added into N-N bonds followed by the insertion of two active cations (e.g. Li + and Na + ). Compared with every imine group that can accept 1 or 1.4 electrons, every azo group can achieve a higher theoretical capacity with 2 electrons [24,85,152]. However, the azo groups connected with two phenyl rings generally result in a redox potential of ~ 1.5 V versus Li/Li + (Fig. 5c) [24,85,152]. Nevertheless, azo compounds also have undergo dissolution in organic electrolytes, which results in possible electrochemical redox irreversibility at low operating potentials.
Unlike other redox-active centres in N-based redox compounds, the N atoms in arylamine compounds [93,153] can serve as p-type redox-active centres, simultaneously donating electrons and being stabilized by anions. Triphenylamine polymers, such as poly(triphenylamine) (PTPAn), are a typical type of arylamine compound in which the N atoms situated at the centre link with three phenyl rings [154]. This triphenylamine group results in a nonplanar structure, and positive charges in the oxidized state are prone to localize at these N atoms [95,155]. For example, the redox reaction mechanism of the PTPAn cathode involves the N atom donating one electron and being converted to positively charged N + , which is stabilized by the insertion of one PF 6 − anion from the electrolyte [154,155]. An organic potassium-ion battery with PTPAn as the cathode has delivered an encouragingly high rate capability and demonstrated stable cycling performance (capacity retention of 85% after 500 cycles) (Fig. 5d 1 ) [155]. Interestingly, although the N atoms in poly(N-phenyl-5,10-dihydrophenazine) (p-DPPZ) have a similar location structure as PTPAn, its redox mechanism is remarkably different from the latter. The redox reaction of p-DPPZ during oxidation, which involves the redox chemistries of intramolecular electron transfer, can be divided into two steps [156]. The first step is associated with the transformation of every second 1,4-phenylenediamine moiety in the main polymer chain to quinoneimine. The second step corresponds to the further conversion of all 1,4-phenylenediamine fragments into quinoneimines. In addition, carbazole polymers are typical arylamine compounds in which conjugated arylamine groups are appended on nonconjugated polymer backbones. Not only are these separated redox-active centres beneficial for stabilizing the working voltage but also the intramolecular π-π stacking interaction between arylamine groups favours charge transfer, which is conducive to boosting the reaction kinetics and cycling performance (Fig. 5d 2 ) [51]. For example, poly(N-vinylcarbazole) (PVK) was exploited by Li et al. as a cathode material for a potassium-ion battery [157]. The PVK cathode underwent a similar p-type reaction relating to the electron-donating/accepting process of the N atom in PTPAn. The battery delivered a high rate performance and a high medium discharge voltage of 4.05 V versus K/K + . Nevertheless, the obvious disadvantage of these arylamine compounds is their low specific capacity.

Conducting Polymer OEMs
Conjugated conducting polymers are a class of attractive OEMs because of their inherently high electronic conductivity and chemical stability. Since the first investigation of polyacetylene conductive polymers for cathodes in lithium-based dual-ion batteries in 1981 [158], tremendous efforts have been devoted to developing electroactive conducting polymers for OEMs. In view of the different electroactive redox groups, conjugated conducting polymers can be classified into four major groups: conjugated hydrocarbons with an electronic conductivity of approximately 10 −5 S cm −1 (e.g. polyacetylene), conjugated benzene (approximately 10 −3 -10 −4 S cm −1 , e.g. polyparaphenylene), conjugated amines (approximately 10 −2 -10 −4 S cm −1 , e.g. polyaniline and polypyrrole), and conjugated thioethers with a high electronic conductivity of 100 S cm −1 (e.g. polythiophene) (Fig. 6a) [52,56,99,136]. Generally, electron delocalization in the conjugated structure enables either the acceptance or donation of electrons (bipolar-type reactions) and stabilizes the resultant products. Some conjugated conducting polymers (such as polyaniline, polypyrrole, and polythiophene) are mainly used as p-type redox cathodes. On the one hand, most conducting polymers have an electron-rich character, meaning they have more stable oxidized states than reduced states. On the other hand, their p-type reactions have relatively high electrochemical redox potentials (2.5-4.5 V vs. Li/ Li + ) [159][160][161][162]. In addition, if the working voltage windows are sufficiently wide, conjugated conducting polymers with electrochemical bipolar characteristics (such as polyacetylene and polyparaphenylene) can successively be subject to n-type and p-type reactions, thus making them appropriate for use as cathodes in hybrid batteries (generally on the basis of alkali(ne) metal anodes) [100,101]; additionally, this property is beneficial for improving the energy density of MIBs. Compared with small-molecule carbonyl compounds suffering from low electronic conductivity and high solubility in aprotic electrolytes, conducting polymers as electrodes present superior characteristics of a relatively high electronic conductivity and limited solubility in electrolytes; these properties are beneficial for enhancing their rate capability and cycling stability [159,162]. For example, Wan et al. [101] reported an aqueous rechargeable zinc-organic battery using conducting polyaniline (PANI) with a hybrid mechanism as the cathode, which exhibited excellent rate capability and cycling performance (Fig. 6b). Nitrogen was doped (=NH + -) and undoped (=N-) in PANI because it was mostly in the half-oxidation state. During the 1st discharge, the doped nitrogen (=NH + -) was reduced, and the Cl − ions dropped off from the PANI. Simultaneously, the -NH-moiety was reduced to -N − and interacted with Zn 2+ . In the rate test, the capacity at the high current density of 5 A g −1 was as high as 95 mAh g −1 , which was approximately 47.5% of that at 0.05 A g −1 . After 35 cycles at different rates, a high capacity of 191 mAh g −1 could be recovered when the current density was returned to 0.05 A g −1 .
However, some drawbacks for conjugated conducting polymers remain to be further studied, including their low discharge capacity, sloped operating voltage profile, and self-discharge issue. Unlike small-molecule OEMs, conducting polymers have a high molecular mass, and their energy storage mechanism is based on ion doping. As a result, their specific capacity is theoretically determined by the doping ability of the repeating structural units and the type of active anions. The doping degree can be estimated by (P x+ ·xA -) n (0 ⩽ x ⩽ 1), where P and Adenote the repeating structural unit (e.g. aniline groups and acetylene groups) and the active anion (e.g. PF 6 and ClO 4 -) [136]. Generally, conducting polymers with higher degrees of doping are capable of higher specific capacities. However, the process for achieving a high degree of doping at the neighbouring repeating structural units causes severe charge repulsion interactions, leading to inactive structural units and low reversibility [84,136]. For example, a polyacetylene cathode delivers a relatively high Coulombic efficiency of 65.6% at a doping degree of 0.097. The value sharply decreases to 35.5% when the doping degree is increased to 0.194 [163]. Thus, the low doping level (approximately 0.3-0.5) is the main reason for the low specific capacities of conducting polymers [84].
In addition, the electrochemical working potentials of conjugated conducting polymers present sloped profiles because the continuously increasing/decreasing doping degrees persistently change the equilibrium potentials during electrochemical reactions. Figure 6c 1 schematically illustrates a typical evolution of the electrochemical working potential against the doping degree for a p-type conductive polymer cathode [136]. As observed, there is an approximately linear correlation between the electrochemical working potential and doping degree. Pristine conducting polymers are in an intermediate state, which can be discharged to a full-reduction state along with a decreasing working potential and charged to a semioxidation state followed by an increasing working potential. Unfortunately, a full-oxidation state during the charge process is generally inaccessible for conducting polymers because fully oxidized conducting polymers at high potentials are usually unstable and tend to induce severe parasitic reactions with electrolytes as soon as the electrochemical oxidation process passes through the C point; this restricts the practical capacity of conductive polymer cathodes. Figure 6c 2 shows a sloped working potential profile ranging from 2.0 to 4.0 V as a function of its specific capacity in a potassium-based dual-ion battery, in which polyaniline serves as the cathode having a p-type reaction with PF 6 − doping during the charging process [164]. Similar sloped potential profiles have been reported in other studies [165]. In addition, although the polyaniline cathode has a theoretical capacity of 295 mAh g −1 (fulloxidation state), its experimental capacity is below 140 mAh g −1 even at an ultralow current density of 10 mA g −1 . Furthermore, although the capacitive-controlled response boosts the reaction kinetics, anion dopants are inclined to disassociate from conducting polymers during long-term storage periods, potentially causing self-discharge issues when used in energy storage systems.

Radical-Based OEMs
Radical polymers with pendant free radicals per repeating unit as the redox-active centres take advantage of a high rate capability and good cycling stability. These radical polymers have densely popularized unpaired electrons, and their electrochemical redox during the charge/discharge processes is generally associated with slight structural variations and electron rearrangements. A considerable number of studies have focused on radical polymers for electrode materials since Nakahara et al. [31] reported the first radical polymer of poly(2,2,6,6-tetramethyl-piperidinyl-1-oxy-4-yl methacrylate) (PTMA) as a cathode material to the battery community (Fig. 7a). During the anodic process, the nitroxyl radical was oxidized into a cation and interacted Electrochemical redox mechanism and performance of radicalbased OEMs. a Typical p-type reaction of radical-based OEMs and their corresponding charge/discharge profiles. Reproduced with permission from Ref. [31]. Copyright 2002, Elsevier Science B.V. b Different modes of anion doping into radical-based OEMs: three modes of anionic doping in a swollen PTMA electrode. Reproduced with permission from Ref. [89]. Copyright 2018, Springer Nature Limited. c Working potential and specific capacity of some typical radicalbased OEMs: the energy diagram of p-type (red) and n-type (blue) radical polymers as a map of the redox potentials versus the formula weight-based capacities. Reproduced with permission from Ref. [166]. with the electrolyte anion to generate oxoammonium salt. During the cathodic process, a reverse reaction occurred. Regarding the cells using the PTMA cathode and Li anode, a cut-off charge voltage of 4.0 V was observed along with obvious charge/discharge profiles at current densities of 0.1 and 1.0 mA cm −2 . The plateau voltage of approximately 3.5 V corresponds to the redox potential of PTMA, indicating the presence of oxoammonium salt in the charged state and pristine state in the subsequent discharge process. Wang et al. [89] demonstrated that the electrochemical redox reaction mechanism of PTMA was dominated by dual-doping modes, which probably consisted of lithium expulsion and anion uptake I and II, as shown in Fig. 7b. In the first mode, the PTMA in the electrolyte was doped by the anion, resulting in lithium cation expulsion into the bulk electrolyte. In the second mode, the anion was exchanged from the external electrolyte. In the third mode, the doped anion came from the internal electrolyte and was balanced by the anion taken from the external electrolyte. Anion uptake I and II coexisted in the doping reactions. The doping in PTMA was controlled by the anion type, electrolyte concentration, and timescale. In addition, among the redoxactive radicals in these radical polymers, nitroxyl groups have attracted the most attention, such as nitronyl nitroxide radicals, 2,2,6,6-tetramethylpiperidinyl-N-oxyl (TEMPO), and 2,2,5,5-tetramethylpyrrolidin-N-oxyl. In addition, there are some other radicals, including dialkoxyaryl, O-centred phenoxyls and galvinoxyls. Although nitroxyl radicals in the neutral state are bipolar-type functional groups and can be either oxidized to form oxoammonium cations (> 3.0 V vs. Li/Li + ) or reduced to yield aminoxy anions (~ 2.0 V vs. Li/Li + ) based on one-electron transfer (Fig. 7c) [166], they mainly work as p-type cathode materials along with redox reactions that utilize active anions in the electrolyte due to the competitive electrochemical redox potentials.
Despite the insulating feature of their main chains, radical polymer electrode materials result in superior rate capability and small voltage polarization, which is probably ascribed to the outer-sphere electron self-exchange reaction mechanism between adjacent radical redox sites induced by redox gradient-driven electron transport (Fig. 7d 1 ) [166]. For example, PTMA can maintain a capacity retention that is above 90% at 12C relative to that at 1.2C and deliver a stable discharge potential plateau of ~ 3.5 V versus Li/Li + . Nishide et al. [167] reported a hydrophilic radical polymer of poly(2,2,6,6-tetramethylpiperidinyloxy-4-yl vinyl ether) (PTVE) bearing pendant TEMPO radicals to serve as a cathode material in an aqueous zinc-ion battery; this polymer simultaneously exhibited a low overpotential of less than 0.1 V (even at an ultrahigh rate of 60C) and relatively stable working voltage plateaus. Moreover, radical polymers have also been reported as n-type electrode materials such as poly(galvinoxylstyrene) (PGVS). Pairing the PGVS anode with a poly(TEMPO-substituted norbornene) (PTN) cathode yields an all-radical polymer battery, which can be operated at an extremely high rate of 360C ( Fig. 7d 2 ) [168]. Similar to nitroxyl radical polymers, other radical-based polymers also present the advantage of fast reaction kinetics. In addition, with an appropriate electrolyte, radical polymers exhibit excellent cycling stability due to their stable polymer structure. PTMA can maintain a stable specific capacity without apparent fading for 500 cycles. Another radical polymer of poly(2,2,5,5-tetramethyl-3-oxiranyl-3-pyrrolin-1-oxyl ethylene oxide) (PTEO) presents extremely small electrochemical polarization, a superior rate capability, and excellent longterm cycling performance with no substantial capacity decay after 1 000 cycles (Fig. 7d 3 ) [169].
Nevertheless, the practical application of radical polymer electrode materials is still restricted by their low specific capacity of less than 150 mAh g −1 due to the intrinsic singleelectron redox reaction mechanism and the high large molecular mass of these radical polymers. For example, PTMA, PTVE, PTEO, PGVS, and PTN have theoretical capacities of 112, 135, 147, 51, and 109 mAh g −1 , respectively [136]. It is noteworthy that further improvement in the specific capacities of these radical polymers is highly challenging via structural optimization. In addition, the intrinsically low conductivity of radical polymers generally requires a high ratio of conductive additives to guarantee smooth electron transfer inside the electrodes, which inevitably imposes a further restriction on their full specific capacities [170].

Overlithiated OEMs
Despite the structural diversity and tunability of OEMs with different electrochemical redox-active centres, the large majority cannot deliver a practical capacity of over 400 mAh g −1 , which renders them insufficient to compete with their inorganic counterparts [171]. Rational molecular engineering, such as the introduction of redox-active moieties and conjugated structures, can increase the specific capacity [71]; however, the extra weight scarcely ensures a remarkable improvement in the specific capacity of OEMs. Interestingly, there is another classification of OEMs, namely multifunctional overlithiated OEMs; these demonstrate overlithiated energy storage processes due to the reversible reduction in their unsaturated C=C or C≡C bonds and other redox groups [16,17,20,90,96,[172][173][174][175]. The overlithiation reactions can greatly boost electron transfer and realize an exciting 1:1 Li/C ratio, and theoretically, the corresponding maximum specific capacity is approximately 2 232 mAh g −1 . Han et al. [16] reported the overlithiated behaviour of 1,4,5,8-naphthalenetetracarboxylic dianhydride (NTCDA) containing a naphthalene ring structure (Fig. 8a). The electrochemical Li + -storage process of NTCDA included two successive steps. The first step was the enolization of four carbonyl groups, and the second step was the reversible reduction in the unsaturated C=C bonds in NTCDA. The enolization reaction step corresponded to a four-electron lithiation process, while the second reduction step referred to a 14-electron overlithiation process. Thus, the whole lithiation process was accompanied by electrochemical reduction with the transfer of 18 electrons, resulting in NTCDA having a high specific capacity of up to ~ 1 800 mAh g −1 . He et al. [19] constructed a multicarbonyl polyimide material (PMTA) containing six carbonyl groups per repeated unit. On the basis of the similar two-step reaction of the enolization of carbonyl groups and the overlithiation of unsaturated carbon bonds, each unit of PMTA was subject to a reversible 22-electron reduction reaction, which corresponded to a theoretical capacity of 1 704 mAh g −1 . Consequently, PMTA could deliver a high experimental capacity of 1 638 mAh g −1 . Through a condensation reaction between 1,4-diaminobenzene and 1,3,5-benzenetricarboxaldehyde, Lei et al. [18] designed a covalent organic framework (COF) material anchored with carbon nanotubes (CNTs) to form a composite, namely COF@CNT. It was demonstrated that the electrochemical redox reaction involved a transfer of 14 electrons per repeated unit. The method also underwent two steps, individually relating to the lithiation of two C=N groups and the overlithiation of unsaturated C 6 rings. The overlithiation reaction included four stages due to the variable free energy (ΔG) at various stages of the lithiated COF monomer (Fig. 8b 1 ). Accordingly, the Electrochemical redox process, performance and structural evolution of overlithiated OEMs. a Electrochemical redox process: a 1 the electrochemical insertion of Li ions onto NTCDA during the discharge process, a 2 potential profile of NTCDA during the discharge process, exhibiting five distinct discharge plateaus corresponding to the insertion of 1, 2, 4, 8, and 18 Li ions, respectively, and a 3 reaction free energy (ΔG) for the addition of lithium onto the aromatic rings of NTCDA (red), pyrene (black), and naphthalene (green). Reproduced with permission from Ref. [16]. Copyright 2012, Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim. b Specific capacity and structural evolution: b 1 structural evolution of the COF monomer during the lithiation process and b 2 its capacity contribution (Reproduced with permission from Ref. [18]. Copyright 2018, Springer Nature.). c Oversodiated and overpotassiated OEMs: the electrochemical c 1 oversodiation process (Reproduced with permission from Ref. [177]. Copyright 2014, Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim.) and c 2 overpotassiation process of PTCDA (Reproduced with permission from Ref. [176]. Copyright 2015, Elsevier Science B.V.) COF in COF@CNT could deliver a high reversible capacity of 1 536 mAh g −1 (Fig. 8b 2 ).
High specific capacities can also be achieved for the storage of Na + and K + via similar oversodiation and overpotassiation of specific OEMs (Fig. 8c) [176,177]. Chen et al. [176] demonstrated that 3,4,9,10-perylene-tetracarboxylicacid-dianhydride (PTCDA) underwent the fourelectron potassium enolization reaction between carbonyl groups and K + and further underwent another transfer of 7 electrons during the overpotassiation reaction between K + and unsaturated C 6 /anhydride rings. The whole reduction process of K + storage corresponded to the transfer of 11 electrons per PTCDA molecule, which experimentally bestowed PTCDA with a high specific capacity of 753 mAh g −1 . Similarly, the sodium enolization reaction and successive oversodiation reaction of PTCDA could generate Na 15 PTCDA, which involved a transfer of 15 electrons and delivered a high capacity of 1 157 mAh g −1 [177]. Compared with Li 24 PTCDA [16], the smaller numbers of inserted Na + and K + in the same molecular structure were probably attributed to the larger ionic radius of the latter two [171].
It should be highlighted that functional groups play significant roles in the electrochemical processes of the overlithiation reaction. For example, aromatic compounds with anhydride groups (e.g. PTCDA and NTCDA) can demonstrate a performance efficiency (the ratio of experimental capacity to theoretical capacity) of 100%; in contrast, for the counterparts with carbonyl groups (e.g. 1,4-naphthoquinone (NQ) and anthraquinone), the performance efficiencies are in the range of 30%-40% [16]. Furthermore, aromatic compounds without anhydride and carbonyl groups (e.g. perylene and pyrene) present extremely low performance efficiencies of below 25% [16]. Anhydride groups have a significantly lower free energy for the Li + insertion reaction than unsubstituted aromatics. In addition, once the sp 3 -hybridized oxygen atoms are substituted by the carbonyl reduction process, there is no overlithiation reduction in unsaturated C=C bonds in the diimide derivatives of NTCDA [171].
Furthermore, despite their high specific capacities, multifunctional overlithiated OEMs still suffer from some practical challenges. The overlithiation/sodiation/potassiation process may destroy the crystal structure, thus leading to inferior cycling performance and a low initial Coulombic efficiency. For example, although the oversodiation of PTCDA can form Na 15 PTCDA corresponding to a total specific capacity up to 1 017 mAh g -1 , only a charge capacity of 227 mAh g -1 can be achieved after five cycles, and the initial Coulombic efficiency is approximately 43% [177]. In addition, the overlithiation process inevitably damages the π-conjugated structure and lowers the conductivity, giving rise to high electrochemical polarization. Renault et al. [171] reported a multifunctional overlithiated OEM of dilithium Fig. 9 Strategies for molecularly engineering OEMs to improve the specific energy density, rate capability and cycling stability of the corresponding energy storage devices benzenedipropiolate (Li 2 BDP), which delivered an initial specific capacity of 1 363 mAh g −1 ; this corresponded to an 11.5-electron reduction process. However, fast capacity fading could be observed in the initial stage, stemming from its intrinsic instability that resulted in structural damage. In addition, there was a large polarization (above 1.0 V) between the charge/discharge processes. Therefore, surmounting these challenges is of great significance for the practical application of multifunctional overlithiated OEMs.

Molecular Engineering of OEMs
As mentioned above, although OEMs have structural diversity, showing different electrochemical properties for energy storage applications, they also individually have a series of challenges, such as their low output voltage, insufficient capacity, poor conductivity, or high solubility. These challenges stem from the inherent features of OEMs, including a relatively low redox potential, a limited number of transferred electrons per redox-active centre, high solubility, or an insufficient electronic conductivity, all of which are remarkably dominated by the molecule-level structure and material-level morphology. Nevertheless, the molecular tunability of OEMs makes it feasible to controllably design their molecular structure, allowing these challenges to be targeted for improvement. This section aims to review the sophisticated molecular design strategies that have been regarded as fundamental approaches to tuning the electrochemical performance of OEMs. As shown in the overall scheme ( Fig. 9) for molecular engineering, these strategies cover the introduction of functional groups, the rearrangement of molecular skeletons, and the polymerization of small molecules, depending on the requirements of the OEMs.

Specific Energy Density
The specific energy density is one of the most critical indexes in evaluating the electrochemical performance of energy storage devices. The specific energy density (E) is directly proportional to the working voltage (V) and specific capacity (C), i.e. E = V × C. The theoretical specific capacity (C) of electrode materials can be determined by the following formula: C = nF/(3.6M w ), where n, F and M w denote the electron transfer number per molecule or repeated unit, the Faraday constant, and the molecular weight of the molecule or repeated unit, respectively. Therefore, E = nFV/ (3.6M w ). Theoretically, it is clear that rationally tuning the electrochemical redox potential of electrode materials and designing a molecular structure with a high n and low M w are effective approaches to promoting the specific energy densities of OEMs.

Voltage Output
The working voltage plays a decisive role in the energy density of batteries, which is determined by the difference in redox potentials between the anode and cathode during charge/discharge processes. Thus, rationally tuning the redox potentials of OEMs, i.e. increasing/decreasing the electrochemical potential of redox-active centres in organic cathodes/anodes can result in batteries with an increased energy density. For example, p-type OEMs are generally used as cathode materials due to their high redox potentials (typically greater than 3.5 V vs. Li/Li + ) during the p-type reactions, e.g. conducting polymers, radical polymers and arylamine compounds. Unfortunately, despite their higher theoretical capacities, n-type OEMs are usually subject to relatively low redox potentials of less than 3.0 V versus Li/ Li + when they are used as cathodes. Moreover, the redox potentials of these n-type OEMs are too high to be applied as anodes. Therefore, the rational tuning of n-type OEMs is of great significance according to their practical application scenarios in batteries, which can be realized by different molecular design strategies, such as the introduction of electron-withdrawing/donating groups and modulating the degree of conjugation.
The redox potential of OEMs is associated with their molecular orbital energy level, which is determined by the electron cloud of the OEMs [178]. Hence, the addition of molecular-level functional groups with electron-withdrawing/donating characteristics is an effective approach to modulate the redox potential of OEMs. Specifically, electron-withdrawing groups (e.g. the halogen groups [41,150,[179][180][181][182], cyano groups [182][183][184], perfluoroalkyl groups [185], sulphonyl groups [186,187], and heteroaromatic ring groups [188][189][190] in Fig. 10a, b) are capable of tuning the lowest unoccupied molecular orbital (LUMO), thus controlling the reduction potential of OEMs because theoretical computations suggest that the redox potential has a linear correlation with the LUMO energy (i.e. reduction energy of molecules) [189,190]. Of course, different redox-active structures inherently feature variable electron cloud distributions; thus, different molecules comply with diverse linear correlations. For example, compared with the common atoms in OEMs, halogen groups have relatively high electronegativity, i.e. a higher electron affinity, which will result in an electron-withdrawing effect on the electron cloud of the redox-active centre and decrease the LUMO energy level [182]. As a result, a higher redox potential is necessary for the redox-active centres to donate electrons. Kim et al. [180] introduced halogen groups onto 1,4-benzoquinone. Due to the fluorine group having the highest electronegativity among the halogen groups, the redox potentials of the C=O redox-active centres for Na + storage showed a tendency of C 6 F 4 O 2 > C 6 Cl 4 O 2 > C 6 Br 4 O 2 > C 6 H 4 O 2 . As expected, when Page 23 of 65 26 C 6 Cl 4 O 2 was applied as the cathode in sodium-ion batteries, it delivered two higher redox potential plateaus (2.9 V and 2.6 V vs. Na/Na + ) than those of C 6 H 4 O 2 (2.4 V and 2.2 V vs. Na/Na + ); notably, even C 6 F 4 O 2 achieved slightly higher redox potentials [180]. Li et al. [76] introduced two Br-or -NO 2 groups into pyrene-4,5,9,10-tetraone (PT), which clearly lowered the LUMO energy level (3.86 eV for PT-2Br and 4.52 eV for PT-2NO 2 vs. − 3.56 eV for PT).

Fig. 10
Molecular engineering towards optimizing the voltage outputs of OEMs. a Grafting functional groups: a 1 some typical electron-withdrawing/donating groups for increasing/decreasing the voltage output, respectively, and a 2 calculated redox potential of 9,10-anthracenedicarbonitrile derivatives grafted with different functional groups. Reproduced with permission from Ref. [80]. Copyright 2019, Elsevier Science B.V. b Number and position of functional groups. b 1 LUMO energy levels of C 6 R 4 O 2 molecules (R = F, Cl, Br) using DFT calculations. Reproduced with permission from Ref. [180]. Copyright 2015, American Chemical Society. b 2 Lithiation reaction potentials of anthracene with aromatic groups at different positions. Reproduced with permission from Ref. [198]. Copyright 2015, The Royal Society of Chemistry. c Relative position of redox-active moieties: different voltage profiles of Li 4 -p-DHT and Li 4 -o-DHT. Reproduced with permission from Ref. [199]. Copyright 2014, American Chemical Society. d Changing conjugated structures. d 1 Calculated LUMO/HOMO energy of BBQ, BBQB, and TBQB. (Reproduced with permission from Ref. [110]. Copyright 2020, Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim. d 2 Working voltage of the OEMs against ΔC 2Li . Reproduced with permission from Ref. [198]. Copyright 2015, The Royal Society of Chemistry. e Cation substitution: correlation between the antisymmetric vibrational modes of the carboxylate functional groups ν as (COO − ) and the average operating potential against the ionic potential of M n+ .Reproduced with permission from Ref. [106]. Copyright 2018, Springer Nature Banda et al. [182] also demonstrated that the introduction of two Br-or cyano groups enabled perylene diimide (PDI) to increase the reduction potentials from 2.16 to 2.27 V and 2.61 V versus Na/Na + , respectively. PDI-CN 2 having a higher reduction potential than PDI-Br 2 should be attributed to the higher electronegativity of the cyano group than the Br-group. The addition of halogen groups and cyano groups to naphthalene diimide also resulted in homologous variation in the redox potential [183]. Yokoji et al. [185] synthesized electron-deficient benzoquinones (BQs) bearing strongly electron-withdrawing perfluoroalkyl groups. Compared with electron-rich BQ derivatives, such as CH 3 -BQ (− 1.08 V vs. Ag/Ag + ), these electron-deficient BQ derivatives (e.g. CF 3 -BQ, Rf 4 -BQ, and Rf 6 -Cl-BQ) as cathode materials delivered higher redox potentials (individually − 0.41, − 0.38 and − 0.26 V vs. Ag/Ag + , respectively). Moreover, the molecular orbital energy levels can be tuned by utilizing electron-deficient/rich aromatic cores [189]. Substituting one or more carbon atoms in an aromatic ring with heteroatoms (e.g. N, S, and O) with stronger electronegativity can induce nonuniform electron delocalization, which leads to a shift in negative charges towards these electron-withdrawing heteroatoms [191]. Liang et al. [189] introduced heteroaromatic rings with O, S, and N heteroatoms in a carbonyl compound of anthraquinone (AQ), individually corresponding to benzofuro [5,6- In comparison with AQ (2.27 V vs. Li/Li + ), the latter three presented reduction potentials that increased by 0.25, 0.34, and 0.44 V. The evolution trend of the redox potential for BFFD, BDTB and PID could be ascribed to the electron-withdrawing ability of the heteroaromatic rings induced by the increased electron affinity of the O, S, and N heteroatoms. Wang et al. [81] applied N-containing heterocycles to tune the working potential of organodisulphide cathodes in lithium-ion batteries. Compared with diphenyl disulphide (DPDS), dipyridyl disulphides (DpyDSs) derived from N substitution at the ortho or para positions in the aromatic rings of DPDS delivered increasing discharge potential plateaus from 2.20 to 2.45 V. Furthermore, after the substitution of nitro groups for N atoms at the ortho positions, the obtained 2,20-dipyridyl disulphide-N,N′-dioxide (DpyDSDO) presented elevated discharge plateaus of 2.80 V. Because the N atom and nitro group had electron-withdrawing features, their introduction reduced the LUMO, corresponding to − 1.02 and − 1.60 eV, respectively.
In contrast, when n-type OEMs are developed as anodes in batteries, it is favourable to further decrease their redox potentials. The introduction of electron-donating groups (e.g. amino groups [181], alkyl groups [41,192,193], methoxy groups [193][194][195], -OLi [192,195,196], and -ONa [195,197] in Fig. 10a) is capable of donating extra electron clouds to the redox-active centres in the molecular structure of the OEMs. Hence, the LUMO energy level of these OEMs can be increased, and their corresponding redox potential can be reduced. For instance, Park et al. [181] substituted one hydrogen in disodium terephthalate (Na 2 TP) with one amino group to yield a Na 2 TP derivative of NH 2 -Na 2 TP, which presented a lower operating potential than that of bare Na 2 TP. Different from the inductive electron-withdrawing effect exerted by halogen on the conjugated carbon backbone, the lone-pair electron of the amino group could contribute electron density to the conjugated carbon backbone via a π-bond in the corresponding resonance structure. Despite the electronegative character of the amino group, the π-donating effect played a dominant role in NH 2 -Na 2 TP. Consequently, a lower redox potential could be achieved for the NH 2 -Na 2 TP anode. Nishida et al. introduced different functional groups into 7,7,8,8-tetracyanodimethoquinone (TCNQ), including methyl groups and F-groups. Compared with pristine TCNQ, the methyl-bearing TCNQ derivative delivered a lower redox potential of ~ 0.1 V [41]. Similarly, OLi-form lawsone (2-hydroxy-1,4-naphthoquinone) produced by the addition of -OLi into 1,4-naphthoquinone (NQ) presented a lower reduction potential than that of NQ due to the increased electron localization destabilizing the elevated LUMO energy [192].
In addition, the number and relative position of electronwithdrawing/donating functional groups in organic molecules are critical for the modulation of their redox potentials [80]. For instance, an increased redox potential of the C=O redox-active centres in 1,4-benzoquinone derivatives can be achieved along with an increasing number of halogen groups [180]. Similarly, the introduction of a pair of F-groups results in the TCNQ derivative having a redox potential that increases by ~ 0.2 V; for two pairs of F-groups, the potential value is increased by ~ 0.3 V [41]. As shown in Fig. 10b 1 , theoretical calculations suggest that the addition of halogen groups is capable of reducing the LUMO energy, causing a linear downshift of ~ 0.2 eV per halogen group. Similar to the BQ derivatives bearing halogen groups, the PDI derivatives decorated with Br-groups present an increasing reduction potential after increasing the number of Br-groups from 2 to 4. A large number of Br-groups can bring about an enhanced structural twist of the PDI-derivative molecule. Such a change in molecular structure can alleviate the increasing electrostatic repulsion inside the molecules and, interestingly, results in control over their discharge potential profile [182]. In the case of the relative position of functional groups, the different positions have an influence on the electron cloud distribution of redox-active centre moieties. Theoretical calculations suggested that different positions of aromatic functional groups can change the lithiation reaction potential of anthracene (Fig. 10b 2 ) [198]. The relative position of redox-active moieties can affect the electrochemical redox potential of OEMs. Taking lithiated dihydroxyterephthaloyl derivatives as an example, a positive shift in the operating potential (~ 300 mV) can be observed when the location of their active groups is transformed from a para-(2.55 V vs. Li/Li + for Li 4 -p-DHT) to an ortho-positioned structure (2.85 V vs. Li/Li + for Li 4 -o-DHT) (Fig. 10c) [199]. Patil et al. also reported a similar position shift in the redox potential of para-quinone groups and ortho-counterparts in a single-ion conducting redox-active polymer cathode [200]. These phenomena were associated with the intrinsic electrostatic interaction between the O atoms in the different regioisomers. During the charging process, a closer distance between the O atoms in the delithiated ortho-positioned structure gave rise to the destabilization of the molecular structures, thus increasing their operating potentials. Jung et al. [80] comprehensively studied the effects of introducing different functional groups on the redox potential of two representative organic classes of carbonitriles and quinones and emphasized the importance of their relative electronwithdrawing strength on tuning the redox potential.
Furthermore, conjugated structures play a critical role in the electrochemical performance of OEMs. As mentioned previously, the molecular structures of OEMs generally contain one or more conjugated structures, and even the conjugated effect is a prerequisite to realize the good reversibility and stability of the electrochemical energy storage properties of OEMs. Conjugated moieties in molecular structures are prone to promote the delocalization of electrons via the π-π orbital interactions induced by the resonance effect. Many works have analysed the effect of tuning conjugated structures on the output potentials of OEMs, reporting on extending the conjugation length, relocating the conjugation moiety, and utilizing different conjugated structures. For instance, Nishida et al. [41] elongated the π-conjugated motif to decrease the on-site Coulombic repulsion inside TCNQ derivatives. Compared with pristine TCNQ, which has two plateau voltages of 3.2 V/2.6 V, the addition of the π-extended structure shortened the gap between the two plateau voltages of the TCNQ derivatives (3.1/2.8 V for 9,9,10,10-tetracyano-2,6-naphthoquinodimethane (TNAP)), while lengthening the gap could cause a smaller plateau gap (2.8/2.7 V for 11,11,12,12-tetracyano-2,6-anthraquinodimethane (TANT)). Larger conjugation could cause a lower bandgap [201]. Through the alteration of aromatic systems, Wang et al. [202] tuned the conjugated backbone of a series of dianhydride-based polyimides (PIs), including pyromellitic dianhydride (PMDA), 1,4,5,8-naphthalenetetracarboxylic dianhydride (NTCDA), and perylene 3,4,9,10-tetracarboxylic dianhydride (PTCDA). Their individual average discharge potentials were 1.73, 1.89, and 1.94 V versus Na/Na + , respectively, which was mainly ascribed to the fact that the variation in the conjugated aromatic systems lowered the LUMO energy level from PMDA to NTCDA and PTCDA-based PIs, thus increasing the average discharge potential. Luo et al. [114] added four phthalimide (C 8 H 4 NO-) groups into the BQ molecule to produce 2,3,5,6-tetraphthalimido-1,4-benzoquinone (TPB), where four rigid aromatic phthalimide groups were symmetrically grafted on the BQ framework. The introduction of phthalimide groups with a conjugated structure could lower the LUMO of the BQ molecule. As a result, the TPB with enhanced structural conjugation delivered a high discharge potential of 3.05 V. Yang et al. [110] introduced a benzene ring to extend the conjugated structure of BQ derivatives (2,2′-bis-p-benzoquinone, BBQ), resulting in 1,4-bis(p-benzoquinonyl)benzene (BBQB) and 1,3,5-tris(p-benzoquinonyl)benzene (TBQB) (Fig. 10d1). The extension of the conjugated structure lowered the average discharge potentials (2.92, 2.61, and 2.82 V for BBQ, BBQB, and TBQB, respectively) and lowered the corresponding gap between the two plateau potentials, which was due to the increased LUMO energy levels (− 3.86, − 3.64, and − 3.68 eV for BBQ, BBQB, and TBQB, respectively) and reduced energy gaps (3.43 eV for BBQ, 3.13 eV for BBQB and 3.33 eV for TBQB). The slightly lower LUMO energy level and larger energy gap of TBQB could be attributed to the electron-donating effect of the conjugated aromatic ring shared by the three BQ units in TBQB, instead of the two units in BBQB. In contrast, Nokami et al. [35] applied more conjugated aromatic rings to add extra π-bonds in molecules, including benzocyclobutenedione (BBD), acenaphthenequinone (ANQ), and pyrene-4,5-dione (PYD). Along with the increased conjugated aromatic rings, BBD, ANQ and PYD delivered an increased reduction peak potential. Liang et al. [188] designed a series of conjugated carbonyl compounds by embedding pre-aromatic 1,2-dicarbonyl moieties in different extended conjugated structures, which realized the efficient regulation of the LUMO energy level. Experimental and theoretical analyses indicated that the average operating potentials had excellent linear dependence on the LUMO energy level. Theoretical computations further suggested that the conjugated structures played a crucial role in tuning the redox potentials [198]. Regarding carbonyl-containing polycyclic aromatic hydrocarbons, carbonyl groups were tightly attached to aromatic carbon skeletons, and the conjugation effect boosted electron delocalization in the whole molecular structure. The correlation between the electron delocalization (aromaticity) and operating potential followed Clar's aromatic sextet theory, which can be evaluated by an index, ΔC 2Li = ΔC/ (0.5 × ΔLi), where ΔC 2Li denotes the average change in the Clar sextet numbers accompanied by the insertion of two Li atoms, and ΔC and ΔLi individually represent the changes in the number of Clar sextets and inserted Li + during lithiation. The molecules with a higher ΔC 2Li generally had a higher redox potential (Fig. 10d 2 ).
It has also been demonstrated that cation substitution is capable of tuning the redox potential of carboxyphenolatebased OEMs. Jouhara et al. [106] substituted Mg 2+ , Ca 2+ , and Ba 2+ cations for some Li + in dilithium (2,5-dilithiumoxy)-terephthalate (Li 4 -p-DHT), which resulted in redox potentials of 3.40, 2.90 and 2.45 V versus Li/Li + , respectively; these values were distinct from that of Li 4 -p-DHT (2.55 V vs. Li/Li + ). The substitution of cations in the host structure was conducive to modulating inductive effects in the redox-active organic skeleton involving electronic charge perturbation. Compared with Ca 2+ and Ba 2+ , Mg 2+ substitution resulted in a higher potential gain of approximately + 800 mV, which could be ascribed to its stronger interaction with carboxylate anions; thus, the stabilization of the aromatic structure was improved. It was also suggested that there was an approximately linear correlation between the average operating potential and ionic potential of the substituting cations (Fig. 10e).

Fig. 11
Tuning the electron transfer number of OEMs. a Increasing the number of redox-active centres, e.g. electrochemical redox mechanisms of BQ and C 6 O 6 ; their theoretical capacities are 496 and 957 mAh g −1 , respectively. Reproduced with permission from Ref. [14]. Copyright 2019, Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim. b Enhancing the electron-accommodating ability of redoxactive centres, e.g. schematic illustration of sulphur doped into the Na 2 TP anode. Reproduced with permission from Ref. [203]. Copyright 2017, Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim. c Introducing other redox-active centres: reversible electrochemical redox mechanisms of TPB/Li 6 TPB (TPB M w = 689 g mol −1 , theo-retical specific capacity = 233 mAh g −1 ). Reproduced with permission from Ref. [114]. Copyright 2017, Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim. d Dependence of specific capacity on molecular weight: chemical structure of some quinones and their reduction potential and theoretical capacity based on one-carbonyl one-electron reduction. Reproduced with permission from Ref. [193]. Copyright 2017, Macmillan Publishers Limited, part of Springer Nature. e Influence of phase transformation on the de/sodiation mechanism of Na 2 C 6 O 6 . Reproduced with permission from Ref. [107]. Copyright 2017, Macmillan Publishers Limited, part of Springer Nature

Electron Transfer Number
Increasing the number of redox-active centres in the structural units is a promising approach to increase the electron transfer number of OEMs. BQ is a typical carbonyl compound and bears two redox-active groups of C=O, which can deliver a theoretical capacity of 496 mAh g −1 . When four more redox-active groups of C=O are introduced to yield cyclohexanehexone (C 6 O 6 ), its theoretical specific capacity is as high as 957 mAh g −1 (Fig. 11a) [14]. Similarly, compared with dipotassium terephthalate (K 2 TP), which has two redox-active groups of -COOK [61], tetrapotassium pyromellitic (K 4 PM), which has four -COOK groups [72], presents a higher specific capacity. The extra addition of some redox-active centres will inevitably increase the molar mass of molecules, thereby limiting the increase in the specific energy density.
There is another feasible method to enhance the electron-accommodating ability of redox-active centres. Zhao et al. [203] replaced O atoms with S atoms in the carboxylate groups of Na 2 TP to produce sulphur-containing sodium salts, such as sodium tetrathioterephthalate. Different from the redox reaction mechanism of Na 2 TP involving the storage of Na + only at redox-active carboxylate groups, sodium tetrathioterephthalate was capable of accommodating six electrons per molecule at the thiocarboxylate groups and on both sides of the benzene ring (Fig. 11b). Because the sulfur-substituted structure had a much higher electron density, which was conducive to improving electron delocalization and thus promoting electron transfer. Such a molecular design could remarkably increase the theoretical capacity from 255 mAh g −1 for Na 2 TP to 586 mAh g −1 for sodium tetrathioterephthalate. It follows that the rational design of the molecular structure can affect the number of transferred electrons per structural unit and thus promote the specific capacity of OEMs.
Moreover, grafting redox-active functional groups increases not only the number of redox-active centres but also the molecular weight of the whole structural unit. For example, the phthalimide group results in two electrochemically redox-active carbonyl groups (−C=O) [114]. As mentioned above, TPB, with a BQ framework grafted with four rigid aromatic phthalimide groups, has ten redox-active carbonyl sites, which is much higher than the two of pristine BQ. However, TPB has a lower theoretical specific capacity of 233 mAh g −1 (vs. 496 mAh g −1 for BQ) due to the increase in molecular weight (689 g mol −1 for TPB vs. 108 g mol −1 for BQ) and the possible steric effects inhibiting the accessibility of some redox-active sites (Fig. 11c).

Specific Capacity
In addition to the electron transfer number, the specific energy density of batteries is associated with the specific capacity involving the molecular weight of molecules or repeated units. Although the introduction of redox-inactive functional groups (e.g. halogen groups and alkyl groups) or conjugation moieties is beneficial to tune the redox potentials of OEMs, it causes an increase in molecular weight, inevitably compromising their theoretical specific capacity. For example, similar to BQ, 2,5-dihydroxy-1,4-benzoquinone (DHBQ), naphthoquinone (NQ), 2,3,5,6-tetramethyl-1,4-benzoquinone (TMBQ), 2,5-dimethoxy-1,4-benzoquinone (DMBQ) and AQ undergo a two-electron transfer process per molecule in the n-type reaction. However, increasing the molecular weight results in a decreasing theoretical specific capacity (Fig. 11d) [193]. Similar phenomena are the same as the dianhydrides of PMDA (246 mAh g −1 ), NTCDA (200 mAh g −1 ), and PTCDA (137 mAh g −1 ) on the basis of the theoretical two-electron transfer per molecule. Therefore, it is crucial for OEMs to reduce the unnecessary structural moieties in molecular structures.
In addition, the practical specific capacity of OEMs not only relies on the theoretical value but also depends on the phase transformation, morphology and conductivity of the electrode, and electrolyte. Herein, we mainly focus on the intrinsic nature of OEMs. Disodium rhodizonate (Na 2 C 6 O 6 ) exhibits a high theoretical specific capacity of 501 mAh g −1 via four-electron redox chemistry. However, it generally delivers a much lower practical capacity after the first cycle. Lee et al. [107] demonstrated that the irreversible transformation from the α-phase to γ-phase deteriorated the redox activity of Na 2 C 6 O 6 , which led to its low practical specific capacity. Combining rational designs of active particle size, electrolyte conditions, and cut-off potentials enabled the enhanced reversibility of transforming the α-phase to the γ-phase of Na 2 C 6 O 6 (Fig. 11e), and this increased the practical capacity up to 484 mAh g −1 , which was close to the theoretical capacity. Walter et al. [87] introduced -NO 2 groups into a polypyrene cathode, yielding a poly(nitropyrene-copyrene) cathode material. Although such a design scarcely impacted the redox potential, the resulting structural difference was likely to result in a higher fraction of active poly(nitropyrene-co-pyrene) than polypyrene in the cathode material. Hence, compared to polypyrene (~ 70 mAh g −1 ), the poly(nitropyrene-co-pyrene) cathode delivered a superior specific capacity of 100 mAh g −1 for the storage of aluminium tetrachloride anions (AlCl 4 − ).

Rate Capability
The power density is a crucial factor in comprehensively evaluating the electrochemical properties of a battery system, which is not only determined by the output voltage but also depends on the rate capability of the battery. In practical applications, it is favourable that the energy capacity of the battery system be charged to 80% within 15 min [58]; as a result, the pursuit of a high rate capability is of great significance for electrode materials. However, OEMs generally have some intrinsic deficiencies, such as their low electronic conductivity (especially for small-molecule OEMs), which limits their rate capability; thus, it is necessary to add a high fraction of redox-inactive conductive additives to the electrodes. The availability of a high rate capability requires OEMs with high conductivity, fast redox reaction kinetics, and small active particles [170]. Herein, this section pays attention to reviewing the molecular design strategies towards improving the rate capability of OEMs. The rate capability of OEMs is strongly dependent on their intrinsic conductivity, which is directly pertinent to the bandgap between the LUMO and the highest occupied molecular orbital (HOMO) energy levels. As mentioned previously, the introduction of functional groups is a feasible approach to tune molecular orbital energy levels. Lee et al. [204] added two amino groups into naphthoquinone (NQ) to yield 2,3-diamino-1,4-naphthoquinone (DANQ) via a simple molecular substitution. Compared with NQ (3.95 eV), DANQ results in a much lower bandgap of 2.74 eV (Fig. 12a). The introduction of the strong electron-donating amino groups significantly caused the destabilization of the HOMO energy level and lowered the corresponding bandgap. The low bandgap facilitated fast electron/Li + transfer and boosted the rate capability of the DANQ cathode. Lee et al. [192] introduced different functional groups into NQ molecules. Compared with the -CH 3 and -OH groups, the -OLi group resulted in a molecule (OLi-form lawsone) with a smaller HOMO-LUMO band gap (2.82 eV) than pristine NQ (4.01), lawsone (3.81 eV) and menadione (4.02). Furthermore, when used as the cathode in lithiumion batteries, the OLi-form lawsone exhibited superior rate capability over the others. In addition, the introduction of two -COONa groups into azobenzene (AB) has been demonstrated to clearly reduce the bandgap. When used as the cathode for sodium-ion batteries, the formed azobenzene-4,4′-dicarboxylic acid sodium salt (ADASS) demonstrated a superior rate performance of up to 40 C [86].
The presence of π-conjugated structures directly links to the rearrangement of electron clouds in the OEMs. Extending π-conjugated structures is capable of strengthening the intermolecular interactions (e.g. π−π or C−H···π interactions) and enhancing the electron transfer kinetics [207]. For example, due to the extended π-conjugated systems, fused N-heteroaromatic diquinoxalinylene (2Q) and triquinoxalinylene (3Q) present much lower bandgaps (Fig. 12c), individually corresponding to 3.39 eV and 3.66 eV, compared to quinoxaline (4.65 eV), indicating that the former two have higher intrinsic electronic conductivities than that of quinoxaline. Hence, 2Q and 3Q can deliver excellent rate capabilities of up to 20C with ~ 60% capacity retention [78]. Similarly, Wang et al. [207] demonstrated that the extension of the π-conjugated system was a promising approach to enhancing the rate capability of OEMs. In their work, the π-conjugated system of Na 2 TP was extended using a C=C bond as the linkage to form sodium 4,4′-stilbene-dicarboxylate (SSDC). Compared with Na 2 TP, SSDC showed a smaller energy gap of 3.49 eV (vs. 4.18 eV for Na 2 TP), implying higher electronic conductivity, stronger intermolecular interactions, and faster charge transfer. When working as the cathode in sodium-ion batteries, SSDC could deliver a high rate capability of up to 10 A g −1 . In addition, Xu et al. [208] demonstrated that branched conjugated polymers (BCPs) had extra advantages over linear conjugated polymers (LCPs) because the branched conjugated structure with regular regional pores facilitated electrolyte infiltration and ionic transfer, which contributed to its excellent rate capability.
In addition, the influence of introducing nonconjugated linkages on electrochemical performance has always been neglected. Different from the π-conjugated structure that facilitates electronic conductivity, nonconjugated linkages (e.g. carbonyl groups) can specifically promote the adsorption of Na + for superior rate capability in sodium-ion batteries. Gao et al. [209] applied electrochemically inactive nonconjugated diketone as a linkage to construct an OEM consisting of perylenediimide (PDI)-based polyimide (OAP) (Fig. 12d). The theoretical calculation suggested that the diketone linkage in OAP had a lower absorption energy for Na + (− 3.38 eV) than that of the counterpart in the reference sample (UP) (− 3.17 eV) using a single carbonyl group as the linkage, which indicated that the former promoted the adsorption of Na + and resulted in OAP having a discharged state with higher stability than the latter. Because there was a shorter distance between the Na + and O atoms in the diketone moieties of OAP (~ 3.07 Å) than that of UP (~ 3.7 Å), OAP showed a superior rate capability than UP.
The tunability of the rate capability can also be realized by the substitution of metal cations. Xue et al. [210] substituted Ag + for Li + in lithium terephthalate (Li 2 TP) molecules, in which Ag + could be in situ converted into conductive Ag particles residing in the electroactive terephthalate moiety due to the high redox potential of Ag + /Ag (0.799 V vs. the standard hydrogen electrode). The obtained silver terephthalate (Ag 2 TP) presented satisfactory rate capability in Li-ion and Na-ion batteries. The same group [211] also studied the effect of K + cation substitution on the Li 2 TP cathode (Fig. 12e). It has been demonstrated that the larger radius of K + resulted in potassium terephthalate (K 2 TP) having a stabler lattice structure. In addition, the K-O bond was more ionic, while the Li-O counterpart featured a more covalent character. When both were separately applied to serve as the cathode in Li-ion batteries, K 2 TP was superior to Li 2 TP with regard to their rate capabilities.
In addition, the increase in the surface area of active materials is beneficial to the contact area between active materials and electrolytes and boosts the ionic transfer kinetics. Zhang et al. [153] synthesized a series of microporous organic polymers (i.e. OPTPA, SPTPA and YPTPA) with the same redox-active triphenylamine units for lithium-ion batteries via different molecular designs. Compared to OPTPA with a relatively low surface area (66 m 2 g −1 ), SPTPA and YPTPA with higher surface areas of 544 and 1 557 m 2 g −1 , respectively, showed much better rate performance. Both SPTPA and YPTPA exhibited no capacity fading even at 2 A g −1 , in comparison with that at 0.05 A g −1 , which was mainly attributed to the higher surface area allowing the active material to have full contact with the electrolyte and shorten the Li + diffusion paths (Fig. 12f). Similarly, Molina et al. [212] designed a new microporous anthraquinonebased conjugated polymer cathode with an ultrahigh specific surface area of 2 200 m 2 g −1 via a two-step pathway combining miniemulsion and solvothermal method; the resulting structure contained a redox-active anthraquinone moiety and an inactive 1,3,5-triethynylbenzene moiety. The design of this molecular structure significantly enhanced the cycling performance and rate capability but compromised the specific capacity. This was because the high surface area of this material lowered the tap density.
Developing organic charge-transfer (CT) complexes has proven to be an efficient method to tune the energy bandgap involving the donor (D) and acceptor (A) band structures, thus improving the rate capability of OEMs. As shown in Fig. 12g1, I D and E A denote the ionization potential of the donor and the electron affinity of the acceptor, respectively. It has been indicated that the HOMO of CT complexes is mainly dominated by the HOMO of the donor, while their LUMO is strongly dependent on the LUMO of the acceptor. Hence, the energy band structure can be effectively modulated by the rational selection of donor and acceptor units [213]. Lee et al. [214] designed two CT complexes, namely phenazine-7,7,8,8-tetracyanoquinodimethane (PNZ-TCNQ) and dibenzo-1,4-dioxin-7,7,8,8-tetracyano-quinodimethane (DD-TCNQ), by self-anchoring two types of constituents ( Fig. 12g 2 ), namely 7,7,8,8-tetracyanoquinodimethane (TCNQ) and phenazine (PNZ)/dibenzo-1,4-dioxin (DD)). The molecules in the CT complexes, originating from the electron-donating and electron-accepting molecules, were bound to each other. The interlayered π-π interaction could produce dense electron clouds, which was beneficial for increasing their electronic conductivities by orders of magnitude (Fig. 12g 3 ). Both CT complexes presented excellent rate capabilities of ~ 73% capacity retention at 0.5 A g −1 for PNZ-TCNQ and ~ 76% capacity retention at 1 A g −1 for DD-TCNQ (vs. that at 0.05 A g −1 ).

Cycling Performance
The operational lifespan is a vital parameter to assess the whole performance of batteries for practical application. However, the cycling performance of OEMs, particularly small-molecule organics, is generally restricted by their inherent solubility or redox reaction products in electrolytes. Thus, surmounting the solubility challenge has been regarded as one of the research hot spots of OEMs. In addition, for different OEMs, the insufficient reversibility of electrochemical redox reactions, volume variation, structural Page 31 of 65 26

Fig. 13
Molecular engineering towards OEMs with a high cycling stability. a Grafting functional groups, e.g. a 1 molecular structures of quinones and LC-quinones with high-polarity functional groups, and their corresponding cycling performance. Reproduced with permission from Ref. [215]. Copyright 2014, Elsevier. a 2 Electron density maps of the optimized structures of PNZ and DAP and their solubilities in the solvent. Reproduced with permission from Ref. [146]. Copyright 2017, The Royal Society of Chemistry. b Introduction of heteroatoms in phenyl rings, e.g. electrochemical behaviour summary of DPDS, 2,2′-DpyDS, 4,4′-DpyDS, and DpyDSDO. Reproduced with permission from Ref. [81]. Copyright 2019, The Royal Society of Chemistry. c Designing di-/trimeric oligomers, i.e. π-conjugated heteroaromatic molecules of 2Q and 3Q, and the lithiation mechanism and cycling performance of 3Q. Reproduced with permission from Ref. [80]. Copyright 2017, Macmillan Publishers Limited, part of Springer Nature. d 1 Polymerization of small-molecule OEMs, e.g. molecular structures with different extended π-conjugated systems from BQ and AQ to PT and poly(benzoquinone sulphide) (PBQS) and poly(anthraquinone sulphide) (PAQS) to PPTS, along with the comparison of their cycling performance. Reproduced with permission from Ref. [219]. Copyright 2018, Elsevier. d 2 Chemical structure and possible electrochemical redox mechanism of TQBQ-COF and its long cycling stability at different current densities. Reproduced with permission from Ref. [77] Copyright 2018, Springer Nature change, etc., are also some underlying factors that affect cycling performance [58]. Aiming at improving the cycling performance, different molecular engineering strategies have been proposed to address these aforementioned issues.
According to the rule of like dissolves like, changing the similarity in polarity results in lowering the solubility of OEMs in electrolytes. Grafting functional groups (-COOLi, -SO 3 Na, -OLi, -ONa, and -COOK, etc.) with high polarity on organic molecules enables OEMs to increase their molecular polarity. Thus, such a molecular design contributes to restricting their solubility in aprotic electrolytes and enhancing their cycling performance. Shimizu et al. [215] introduced two -COOLi groups into quinones and produced different quinone derivatives, namely 2,6-bis(lithiooxycarbonyl)-9,10-anthraquinone (LCAQ), 2,7-bis(lithiooxycarbonyl)-9,10-phenanthrene quinone (LCPQ), and 2,7-bis(lithiooxycarbonyl) pyrene-4,5,9,10-tetraone (LCPYT). The introduction of the -COOLi groups was conducive to enhancing the intermolecular interactions and thus limited dissolution, which remarkably boosted the cycling performance (Fig. 13a 1 ). Similarly, the introduction of -OLi groups could inhibit the solubility of poly(2,5-dihydroxy-p-benzoquinonyl sulphide) (PDHBQS) [216]. The formed lithium salt of PDHBQS (Li 2 PDHBQS) resulted in a high cycling stability for up to 1 500 cycles while retaining 90% capacity, which was ascribed to the synergistic effect of the O···Li···O coordination bond and high molecular weight giving rise to the insolubility of Li 2 PDHBQS. Luo et al. [86] compared the electrochemical performance of three types of organic compounds, including AB, 4-(phenylazo)benzoic acid sodium salt (PBASS), and ADASS, for sodium-ion batteries. To suppress the high solubility of AB, one/two -COONa groups were introduced to form PBASS/ ADASS. Compared with AB and PBASS, ADASS presented the lowest solubility in the applied electrolyte and showed the best cycling stability. Deng et al. demonstrated that the K 2 TP cathode produced by the substitution of -COOK for -COOLi in the Li 2 TP molecule could greatly enhance the dissolution resistance of the cathode and promote the cycling stability of the material due to the higher polarity induced by the ionic K-O bonds in K 2 TP than the Li-O bonds in Li 2 TP [211]. In addition, -SO 3 Na groups were introduced to inhibit the dissolution of pristine AQ, which resulted in increasing the cycling performance of SO 3 Na-substituted AQ (AQS) [186].
The structural stability also plays a significant role in determining the cycling capability of OEMs. The introduction of functional groups has proven to be an effective method to enhance the structural stability of OEMs. For example, BQ is generally subject to serious dissolution issues in an electrolyte. Yokoji et al. [217] introduced alkyl groups with various degrees of bulkiness to the BQ skeleton, which resulted in enhanced cyclability and increasing substitution degrees of bulky alkyl groups. The presence of bulky substituents effectively protected the reactive BQ anion radical and suppressed side reactions. In addition, a BQ derivative bearing a methoxy group has been demonstrated to present better cycling performance than pristine BQ because π-π interactions and hydrogen bonding limit the solubility [194]. In addition, unlike phenazine (PNZ) with its solubility of ~ 125 mg mL −1 , the presence of amino groups in 2,3-diaminophenazine (DAP) contributed to the suppression (< 2 mg mL −1 ) of its dissolution problem in an aprotic electrolyte, thus enhancing its cycling performance (Fig. 13a 2 ) [146]. The introduction of ring-type functional groups is also beneficial to enhance the structural stability and cycling performance of OEMs [189].
The introduction of heteroatoms in phenyl rings is also capable of tuning the solubility of OEMs. Organodisulphides, such as DPDS, are generally subject to high solubility in electrolytes [81]. N substitution at the ortho position (ortho-N) in the phenyl rings greatly enhances the cycling performance of DPDS. In contrast, a similar N substitution at the para position (para-N) results in poor cycling performance that is similar to that of DPDS. Ortho-N results in a strong interaction among the N atoms, with its lone pair of electrons; the S atom, with its negative charge; and the Li ions in lithium pyridine-2-thiolate. This results in a tight, clustered structure of N···Li···S bridges due to the short distance between the N and S atoms. However, the para-N atoms are far from these Li and S atoms in lithium pyridine-4-thiolate. Thus, the unique molecular structure of 2,2′-dipyridyl disulphide (2,2′-DpyDS) with ortho-N substitution leads to reduced solubility in electrolytes and enhanced cycling performance for 500 cycles, demonstrating 69% capacity retention; this result is superior to that of DSDP (100 cycles with 57% capacity retention) and 4,4′-DpyDS (100 cycles with 57% capacity retention) (Fig. 13b) [81].
Although the introduction of some specific functional groups is beneficial to boost the cycling performance of OEMs, the addition of extra mass can lower their specific capacities. Therefore, dimeric/trimeric oligomers based on small-molecule OEMs have been exploited to reduce their solubility and enhance their structural stability without comprising their specific capacity. To overcome the dissolution problem of BQ, Yokoji et al. [195] synthesized a dimeric BQ derivative of 2,2′-bis-p-benzoquinone (BBQ), which presented superior cycling performance to that of BQ. Similar to other functional groups, one BQ unit in BBQ molecules could serve as a protective substitute to prevent the singly occupied molecular orbital (SOMO) of another BQ anion radical from approaching other molecules. Peng et al. [78] designed a trimeric 3Q cathode by fusing quinoxaline building blocks. Compared with quinoxaline, the trimerization extended the π-conjugated structure and bestowed the 3Q molecule with enhanced structural stability. This molecular design enabled the 3Q cathode material to deliver an excellent cycling performance of over 10 000 cycles (Fig. 13c). Similarly, Hu et al. [122] designed a novel organic oligomer of PTCDI-DAQ by hybridizing one perylene-3,4,9,10-tetracarboxydiimide (PTCDI) moiety and two AQ moieties. The increased molecular weight resulted in the PTCDI-DAQ cathode having poor solubility in electrolytes; however, each moiety maintained its intrinsic redox activity. Thus, the PTCDI-DAQ cathode in a potassium-ion battery delivered excellent cycling stability and a satisfactory specific capacity.
Commonly, the polymerization of small-molecule OEMs has been demonstrated to be a promising method to improve the stability and reduce the solubility of OEMs in electrolyte solutions, which is beneficial to long-term cycling performance. For example, owing to its unfavourable dissolution in electrolytes, BQ generally exhibits inferior cycling performance. Song et al. [97] designed a poly(benzoquinonyl sulphide) (PBQS) consisting of abundant BQ units linked with thioether bonds, to be used as a cathode in lithium-ion and sodium-ion batteries. Compared with the BQ cathode that had poor cycling stability of 20 cycles with 32% capacity retention, the PBQS cathode in a lithium-ion battery could maintain capacity retention of 86% after 1 000 cycles. Analogously, AQ is also subject to the serious dissolution problem in electrolytes. Song et al. [50] coupled AQ rings to generate poly(1,4-anthraquinone) (P14AQ) with a high molecular weight (approximately 230 000) via a one-step condensation polymerization method. Compared with AQ and poly(1,5anthraquinone) (P15AQ), with their low molecular weights (approximately 2 300), the P14AQ cathode in a lithium-ion battery presented remarkably superior cycling performance due to its lower solubility. It was demonstrated that the different dissolution behaviour played a dominant role in capacity fading. When used as a cathode for magnesium-ion batteries, P14AQ was also endowed with highly superior cycling performance [218]. Regarding polymer-based OEMs, the extension of π-conjugated structures is also conducive to enhancing their cycling performance. Tang et al. [219] designed a poly(pentacenetetrone sulphide) (PPTS) using extended π-conjugated structures as a rigid backbone. This design was beneficial to enhance π-π intermolecular interactions and charge transport and boost insolubility. Compared with PBQS and poly(anthraquinone sulphide) (PAQS), PPTS with an extended π-conjugated structure presented superior cycling performance (Fig. 13d 1 ), particularly at high current densities. Covalent organic frameworks (COFs) have designable skeletons and stable nanoporous structures, which can promote the electrochemical stability of OEMs. Shi et al. [77] designed a COF consisting of triquinoxalinylene and benzoquinone units (TQBQ-COF) in the main skeletons by condensing tetraminophenone (TABQ) and cyclohexanehexaone (CHHO). The stable structure and massively open ion transport channels bestowed the TQBQ-COF cathode with excellent cycling stability, reversible Na + storage, and 96.4% capacity retention after 1 000 cycles at 1.0 A g −1 (Fig. 13d 2 ). Finally, grafting redox-active moieties on the conductive skeletons can also enhance the cycling stability, but increasing the inactive weight will clearly reduce the specific capacity of the whole OEM.

Morphological Engineering of OEMs
Systematic and comprehensive reviews of the effects of OEM morphology on battery performance are still rarely reported. In this section, we will introduce the design strategies and preparation methods of OEMs with different morphologies, the properties of the resultant OEMs and the benefits of morphologically engineering OEMs for improved battery performance to achieve an in-depth understanding of the morphological engineering of OEMs.
It is well established that morphological engineering can create nanostructures to boost the surface area and active sites and to create more channels and networks for electron conduction and ion transport, leading to significant benefit to the specific energy density and rate capability [222][223][224]. Accordingly, the design strategies of the morphological engineering of OEMs for improving the specific energy density, rate capability, and cycling performance of OEM-based MIBs are summarized in Fig. 14. In particular, the strategies of the self-assembly of 0D-2D nanomaterials, in situ/ex situ growth of 0D-2D nanomaterials on porous inorganic templates, and in situ/ex situ OEM-encapsulating 3D nanostructures can not only boost the surface area and active sites but also enhance the porosity for rapid ion transport, which is bound to improve the energy density and rate capability. The strategies of in situ/ex situ growth of OEMs on/in 1D CNTs, in situ/ex situ growth of OEMs on 2D templates, and in situ/ex situ OEM encapsulation of OEMs in 3D frameworks can be beneficial to the construction of electron-conducting networks, which enhance the rate capability. For example, the rate capability of MIBs based on OEMs can be enhanced by facilitating electron conduction and ion diffusion through integration with 2D inorganic frameworks and the exfoliation of 2D nanomaterials, respectively [225,226]. Furthermore, cycling performance can be improved by alleviating parasitic dissolution and aggregation of OEMs through the in situ/ex situ growth of OEMs on/in 1D-2D frameworks and ex situ infusion of OEMs into 3D inorganic nanostructures [223,227].
As shown in Table 3, the synthesis methods for OEMs can be separated into two main categories, i.e. bottom-up methods and top-down methods. The bottom-up methods piece the system together into a more complex system, making the original system a subsystem of the emerging system. Top-down methods, or step-by-step design, essentially decompose the system to obtain the information of subsystems through reverse engineering. Bottom-up fabrication is a strategy to synthesize nanostructure materials in situ with a wide range of morphologies (e.g. 0D, 1D, 2D, or 3D) in high yield by stacking atoms/molecules in an orderly manner on building blocks [228]. The bottom-up method is regarded as a promising approach to fabricate nanomaterials due to formation of few defects, uniform chemical composition and short-/long-range order of prepared nanomaterials [228]. Conventional OEMs present bulky amorphous morphologies, which hinder the diffusion of metal ions and electrons among OEMs, decreasing the utilization ratio of redox-active sites on the OEMs [229]. Therefore, the creation of the nanostructured morphology of OEMs facilitates electron conduction and ionic transfer and subsequently improves battery performance, such as higher energy densities and better rate capabilities. In situ polymerization [222] Ball-milling [231] Antisolvent methods [232,243] 1D Controlled precipitation [78,246] In situ polycondensation [247,249] In situ growth on the 1D template [103,148] Ex situ integration with CNT227 [233,250] Antisolvent methods [223,251] 2D In situ polymerization [77,100,256,[259][260][261] In situ growth on 2D templates [230] Controlled precipitation [72,262] Ball-milling [49,272] Chemical exfoliation [275,276] Supersonication [257,277] Antisolvent methods [147] 3D In situ polymerization with 3D inorganic frameworks [281][282][283] Ex situ integration with 3D inorganic frameworks [200,289,290]     153 mAh g −1 /60C 180 mAh g −1 /30C/5 100 [200] As shown in Table 3, conventional bottom-up methods for preparing OEM nanostructures include in situ polymerization [222] and controlled precipitation [78]. Interestingly, template synthesis, such as in situ growth on 1D, 2D, and 3D inorganic templates, can be readily adapted to the fabrication of 1D, 2D, and 3D OEMs, respectively [103,230]. In contrast, top-down methods can be readily used in the fabrication of OEMs with 0D, 1D, 2D and 3D morphologies by scaling down bulky OEMs to nanoscale size [228]. As illustrated in Table 3, most reported top-down methods for fabricating OEM nanomaterials, including ball milling [231], antisolvent methods [223,232], and ex situ integration with 1D/2D inorganic frameworks [233,234], which can be readily used to produce nanostructured OEMs from bulky OEMs, are commonly available from traditional organic synthesis methods. Top-down methods are often able to preserve the physical and chemical properties of bulky OEMs. Table 4 lists dozens of representative OEMs with different morphologies and their electrochemical performance. From 0D to 3D, OEMs have been widely applied in many battery systems, such as SIBs, LIBs, and PIBs. Notably, carbon additives are necessary to enhance the electronic conductivity of organic electrodes. Interestingly, there are some successful  75 mAh g −1 /5 A g −1 89%/250 mA g −1 /500 [290] CNT, carbon nanotube; rGO, reduced graphene oxide; AB, acetylene black; CB, carbon black; KB, Ketjen black; ASSSB, all-solid-state sodium battery; ASSLB, all-solid-state lithium battery; SSLB, solid-state lithium battery; COF, covalent organic framework; CCP-HATN, carbon-linked conjugated polymer framework; 2D-PAI, 2D polyarylimide; TThPP, 2D COF polyporphyrin; PTMA, poly(2,2,6,6-tetramethylpiperidinyloxy-4-vinylmethacrylate); LIBs, SIBs, PIBs and ZIBs represent lithium-ion, sodium-ion, potassium-ion and zinc-ion batteries, respectively examples of OEMs in all-solid-state batteries, which will be the next popular research direction of OEMs because of their advantages for use in wearable devices.

0D OEMs
Zero-dimensional (0D) OEMs are generally defined as organic cathode/anode materials made of nanosized particles, which typically are spherical [220]. In MIBs, 0D OEMs with a nanoparticle morphology can significantly enhance the surface area for electrochemical reactions and facilitate electron and metal-ion transport, thereby leading to fast redox reaction kinetics and high reversible capacities. Compared to bulky OEMs, 0D OEMs with small grain sizes and large surface areas allow for increased infiltration of electrolytes into electrodes, resulting in faster (shorter) ion diffusion rates (pathways) and batteries with higher power densities [222,235]. In addition, unlike bulky OEMs, 0D OEMs with small diameters result in electrodes with a lower tendency of particle cracking, which effectively increases the reversible specific capacity and capacity retention of batteries [236]. Therefore, it is of great importance to design and control the size and morphology of 0D OEMs.

Bottom-Up Synthesis for 0D OEMs
0D OEMs usually exhibit a smaller size and larger surface area than bulky OEMs, resulting in their close contact with electrolytes and a high tendency to be dissolved [220]. Therefore, the most reported 0D OEMs are polymer dots, which are fabricated by in situ polymerization from molecular precursors; these have a high molecular weight that can restrain unwanted dissolution in electrolytes [237]. The TFPB-TAPT covalent organic framework (COF) can be synthesized through a Schiff-base reaction from molecular reagents, including 1,3,5-tris(4-formyl phenyl) benzene (TFPB) and 1,3,5-tris(4-amino phenyl)-triazine (TAPT) [238]. The resultant COF shows a nanospherical morphology (Fig. 15a) and a large surface area (120 m 2 g −1 ) [238]. Benefiting from this 0D nanospherical morphology, a sodium-ion battery (SIB) with TFPB-TAPT COF active material exhibits good rate capability (Fig. 15b) [238]. 0D OEMs can also decrease the interfacial resistance between electrodes and electrolytes. For example, Yao et al. synthesized 0D-structured Na 4 C 6 O 6 with a diameter of 200-300 nm (Fig. 15c) [239]. This Na 4 C 6 O 6 nanoparticle decreased the interfacial resistance between the electrode and sulphide SSEs due to the inherent flexibility of the organic structure and high surface areas [239]. The electronic conductivity of Na 4 C 6 O 6 was 1 × 10 −4 S cm −1 , which could compete with that of LiMn 2 O 4 and LiFePO 4 [239]. The chemically and electrochemically stable Na 4 C 6 O 6 nanoparticles were used as a cathode for an all-solid-state sodium battery structured as Na 4 C 6 O 6 -sulphide SSE‖sulphide SSE‖Na-Sn alloy anode [239]. This novel Na 4 C 6 O 6 electrode exhibited excellent electrochemical performance, including a high reversible capacity (184 mAh g −1 ), specific energy density (395 Wh kg −1 ), and stable long-term cycling performance when used in this sulphide-based SSE battery (Fig. 15d) [239].
The fabrication of 0D OEMs is more difficult than that of inorganic electrode materials. Therefore, it is crucial to comprehensively investigate the reaction conditions that can affect the morphologies of OEMs. Wang et al. reported that a polymer cathode material called poly(pyrrole-squaraine) (PPS) with a 0D morphology in various particle sizes ranging from 200 to 1 200 nm could be prepared by controlling the polarity of the reaction solvent, the ratio of reactive monomers, and the concentration of monomers in the solvent [222]. PPS (PPS-XS), having the smallest particles (200 nm), was prepared through polycondensation between two monomers at a molar ratio of 1 in a pure 1-butanol solvent (Fig. 15e); this material exhibited the highest capacity among the five PPS samples after 2 000 cycles at a current density of 1 000 mA g −1 , thereby demonstrating the positive effect of the 0D nanostructure on increasing metal-ion diffusion and electron conduction (Fig. 15f) [222].
Even though 0D OEMs prepared through bottom-up methods exhibit good battery performance, they can aggregate into larger interconnected particles, necklaces or chainlike structures, thus burying their active sites and ultimately leading to relatively fast reversible capacity deterioration [38,240]. Therefore, it is important to develop novel and controllable bottom-up methods to prepare 0D OEMs with high monodispersity. Regarding the application of 0D OEMs obtained through bottom-up methods, the aggregation or stacking of OEMs due to intermolecular interactions, such as van der Waals forces, hydrogen bonds and π-π interactions, among 0D nanoparticles needs to be prevented [241].

Top-Down Synthesis for 0D OEMs
Top-down synthesis is a versatile and straightforward method for the preparation of 0D OEMs. The first top-down synthesis for 0D OEMs is ball milling, which utilizes mechanical force to downsize bulky OEMs without varying their chemical compositions and structures [231,242]. For instance, a simple 3,4,9,10-perylene-bis(dicarboximide) (PTCDI) 0D OEM can be fabricated by ball milling, and the diameters of the PTCDI nanoparticles range from 100 nm to 1 μm [231]. A sodium-ion battery (SIB) based on a 0D PTCDI active material exhibits a high specific capacity of 103 mAh g −1 at a current density of 600 mA g −1 , which is equivalent to 75% of the theoretical capacity of PTCDI [231]. The impressively fast rate capability of 0D PTCDI results from fast electron conduction and Na + diffusion, which is facilitated by the small PTCDI nanoparticles (close contact with the electrolyte) and π-conjugated structure of PTCDI molecules [231]. In addition, SIBs based on the 0D PTCDI active material show stable long-term cycling performance with capacity retention of 90% after 300 cycles; this result is uncommon among SIBs based on OEMs, especially for OEMs with low molecular weights [231]. However, the particle size of 0D PTCDI prepared by ball milling is uneven, which may be due to the difficulty of controlling the synthesis conditions during the ball milling process; thus, this should be solved in the future to optimize battery performance [231].  Another top-down synthesis method for 0D OEMs is the antisolvent method: first, bulky OEMs are dissolved in one solvent (which the OEMs have a high solubility in) and then 0D OEMs are precipitated after mixing this solution with another solvent (which is defined as the antisolvent and the one OEMs have a low solubility in) [243]. Therefore, the antisolvent method is regarded as a promising top-down approach to fabricate 0D OEMs due to its many advantages, such as its simplicity for controlling the size, shape, and morphology of 0D OEMs (by adjusting the concentration of OEMs and the type and content of the two solvents) and versatility for many OEMs (including some polymer active materials) [17,107]. For example, two kinds of a conjugated ladder polymer poly(benzobisimidazobenzophenant hroline) (BBL) and its derivative (SBBL) have been synthesized through one-pot polycondensation [243]. However, the bulky morphology of BBL and SBBL results in difficult charge transfer within electrodes due to the existence of the large voids within the BBL and SBBL particles (Fig. 16a) [243]. The antisolvent method is proposed to decrease particle sizes by utilizing methanesulphonic acid as the solvent and ethanol as the antisolvent, leading to the formation of 0D BBL and SBBL nanoparticles with diameters of 30 and 70 nm (Fig. 16b), respectively [243]. LIBs based on BBL and SBBL can retain a high reversible capacity of 496 mAh g −1 and 320 mAh g −1 after 1 000 cycles at 50 °C and a current rate of 3C (Fig. 16c), respectively [243].
Although disodium rhodizonate (Na 2 C 6 O 6 ) has a high theoretical capacity (501 mAh g −1 ), the practical capacities of batteries based on the Na 2 C 6 O 6 active material in most literature are much less than 501 mAh g −1 [107]. Bao et al. reported that the structure of Na 2 C 6 O 6 could change from α-Na 2 C 6 O 6 to γ-Na 2+n C 6 O 6 during the first desodiation process [107]. However, γ-Na 2+n C 6 O 6 changed into γ-Na 2+x C 6 O 6 instead of returning to α-Na 2 C 6 O 6 due to the significant activation barrier of transformation from γ-Na 2+n C 6 O 6 to α-Na 2 C 6 O 6 when using bulky Na 2 C 6 O 6 Fig. 18 (Fig. 17a) as the active material in the PC electrolyte, thereby leading to limited sodium-ion storage [107]. As displayed in Fig. 17b, Na 2 C 6 O 6 nanoparticles were fabricated through the antisolvent method with water as the solvent and ethanol as the antisolvent [107]. After being utilized as an active material in SIBs with DEGDME as an electrolyte, the crystal structure of Na 2 C 6 O 6 was changed from α-Na 2 C 6 O 6 to γ-Na 2+n C 6 O 6 during the first desodiation process [107]. As illustrated in Fig. 17c, the reversible phase transformation from γ-Na 2+n C 6 O 6 to α-Na 2 C 6 O 6 occurred after charging to 3.2 V in the first cycle with the formation of nanosized grains (smaller than 20 nm) within the Na 2 C 6 O 6 nanoparticles [107]. It was reported that the phase could transform successfully from γ-Na 2+n C 6 O 6 to α-Na 2 C 6 O 6 during subsequent cycles even at 2.2 V, demonstrating the notably decreased activation barrier of the phase transformation from γ-Na 2+n C 6 O 6 to α-Na 2 C 6 O 6 [107]. The SIB based on the 0D Na 2 C 6 O 6 active material with DEGDME as the electrolyte exhibited a high reversible capacity of 498 mAh g −1 at 50 mA g −1 (Fig. 17d), which was equivalent to 95% of its theoretical capacity [107].
Interestingly, the reversible redox process of 0D Na 2 C 6 O 6 nanoparticles with a few tens of nanometres (ex-Nano) in a PC electrolyte does not deteriorate with an increase in the cycle number, indicating the transformation from γ-Na 2+n C 6 O 6 to α-Na 2 C 6 O 6 can also occur in the PC electrolyte by decreasing the nanoparticle size of Na 2 C 6 O 6 ; this can highly decrease the activation barrier of the phase transformation due to the 0D morphology of Na 2 C 6 O 6 [107].

1D OEMs
1D OEMs include nanorods, nanotubes, nanowires, and nanofibres, which usually have small widths (less than one μm) and long lengths [220]. 1D OEMs have been regarded as one of the most promising candidates of advanced electrodes for energy storage systems due to their several advantages, such as their high surface area, stress alleviation and continuous ultrafast electronic conduction and ionic diffusion in oriented directions, which results in batteries with an excellent rate capability [220,245]. In addition, it is an effective approach for some OEMs with small molecular weights to grow on/in 1D CNTs to mitigate the high dissolution of OEMs, benefiting stable long-term cycling performance [227]. Moreover, 1D OEMs can be fabricated by either a bottom-up synthesis method or 1D material-assisted top-down treatment, such as nanorods [246], CNTs [18], and nanofibres [78].

Bottom-Up Synthesis for 1D OEMs
Bottom-up synthesis for 1D OEMs can be realized by three general methods in this work, including controlled precipitation [78,246], in situ polycondensation [247,248] and in situ growth on a 1D template [103,148]. Conventional bottom-up synthesis for 1D OEMs is a controlled precipitation method, which is an approach in which precipitates of 1D OEMs will be formed after the removal of a solvent [78,246]. For example, an organic material with a nanorod Fig. 19 a TEM image of COF@CNT. b Long-term cycling performance of COF and COF@CNT. Reproduced with permission from Ref. [18]. Copyright 2018, Springer Nature. c TEM image of COF-10@CNT. d Long-term cycling performance of COF-10 and COF-10@CNT. Reproduced with permission from Ref. [249]. Copyright 2019, American Chemical Society morphology, named tetrasodium salt of 2,5-dihydroxyterephthalic acid (Na 4 C 8 H 2 O 6 ), was first fabricated through a facile and green one-pot neutralization method, and then the solvent was subsequently removed under vacuum distillation [246]. Na 4 C 8 H 2 O 6 with a nanorod morphology (Fig. 18a) was obtained through the π-stacking forces among the numerous Na 4 C 8 H 2 O 6 molecules in an ordered direction, and the surface area of Na 4 C 8 H 2 O 6 was 103.8 m 2 g −1 [246]. As shown in Fig. 18b, the SIB based on the Na 4 C 8 H 2 O 6 nanorod exhibited a high reversible capacity (186 mAh g −1 ), which was quite close to its theoretical capacity (187 mAh g −1 ) [246]. SIBs utilizing the 1D Na 4 C 8 H 2 O 6 active material displayed stable and high capacity retention and Coulombic efficiency (CE) due to the high dissolution resistance of the tetrasodium salt unit [246]. In addition, a full SIB was assembled by utilizing Na 4 C 8 H 2 O 6 as both the cathode and anode active materials, retaining 76% of the reversible capacity after 100 cycles at 0.1C; this demonstrated the promising potential of Na 4 C 8 H 2 O 6 for application in SIBs [246].
Quinoxaline-based heteroaromatic molecule (3Q) was first synthesized through a facial Schiff-base reaction, and single-crystalline 3Q was obtained through slow evaporation from a mixed solvent, which showed a uniform nanofibre morphology (Fig. 18c) derived from π-π interactions among adjacent 3Q molecules [78]. LIBs based on the 3Q active material displayed excellent rate capability, demonstrating fast charge transfer kinetics between the extended π-conjugated 3Q molecules and lithium ions (Fig. 18d) [78]. In addition, LIBs with 3Q active materials exhibited a high initial reversible capacity (215 mAh g −1 ) and capacity retention (67%) after 10 000 cycles at a high current density of 20C [78]. This excellent battery performance was quite extraordinary, especially for OEMs with low molecular weights, demonstrating the positive effect of the extended π-conjugated system and 1D nanofibre morphology on achieving an ultrafast-rate capability and stable long-term cycling performance [78]. However, the loading of active material was too low (30%), being unable to meet the requirements of practical battery applications [78].
In addition, 1D OEMs can be prepared by the in situ polycondensation method. For instance, two polyimide molecules (PI and PI-IMI) with the same chemical structure but different morphologies have been synthesized through polycondensation between 3,4,9,10-perylene tetracarboxylic dianhydride (PTCDA) and hydrazine monohydrate in N-methyl-2-pyrrolidone solvent and imidazole medium, respectively [248]. The diameter of the PI nanorod (100 nm) is much smaller than that of the PI-IMI nanorod (the diameter: 500 nm), and the surface area of the PI nanorod (102 m 2 g −1 ) is higher than that of the PI-IMI nanorod (41 m 2 g −1 ) [248]. Therefore, the large surface area and small diameter of this 1D morphology of PI resulted in SIBs with a high rate capacity (maintained 75% capacity at 800 mA g −1 in comparison with the specific capacity measured at 25 mA g −1 ) [248].
Another bottom-up synthesis method for 1D OEMs is in situ growth on a 1D template [103,148]. For instance, even though COFs have open channel π-electron systems that can benefit electron conduction and ion diffusion, the electroactive sites of COFs cannot be fully utilized because the strong π-π interactions between COF molecules will form closely packed layers that deeply bury their interior redox-active sites [55]. Therefore, it is critical to fabricate OEMs with more exposed redox-active sites to increase battery performance. COF@CNT with 1D morphology has been fabricated by the in situ growth of few-layered COFs (~ 5 nm) on CNTs through π-π interactions between the CNTs and COFs (Fig. 19a) [18]. The CNTs can facilitate the dispersion of COFs into its layered exterior surface, which leads to the exposure of more of the redox-active sites of the COF [18]. In comparison with the pristine COF, the 1D morphology of COF@CNT is beneficial for activating the interior redox-active groups of C=C on the benzene ring of the COF, which highly boosts the lithium-ion storage when COF@CNT is utilized as the active material, thereby leading to a high reversible capacity (Fig. 19b) [18]. In addition, COF-10@CNT has been prepared by in situ polycondensation between 4,4-tobiphenyldiboronic acid and 2,3,6,7,10,11-hexahydroxytriphenylene in the presence of CNTs, which grow uniformly on the exterior surface of the CNTs (Fig. 19c) [249]. The specific capacity of PIB based on COF-10@CNT is 68 mAh g −1 at a high current density of 5 A g −1 , demonstrating fast electron conduction and a high potassium-ion diffusion rate [249]. In addition, a PIB with the 1D COF-10@CNT nanomaterial maintains a higher specific capacity (288 mAh g −1 ) than pristine COF-10 (57 mAh g −1 ) after 500 cycles at a current density of 0.1 A g −1 [249]. Moreover, CCP-HATN@CNT and PAI@ CNT have been prepared by the in situ growth of thin polymer layers on the exterior surface of CNTs and both exhibit faster charge transfer kinetics and stabler long-term cycling performance in comparison with pristine COFs, indicating the significant effectiveness of the 1D morphology for improving the battery performance of OEMs [103,148].

Top-Down Synthesis for 1D OEMs
Top-down synthesis for 1D OEMs can be realized by two general methods in this work, including ex situ integration with CNTs [227,233,250] and the antisolvent method [223,251]. As mentioned above, OEMs with low molecular weights usually have two drawbacks, namely their high dissolution in conventional electrolytes and electron-insulating features [252,253]. Although many strategies have been developed to tackle these problems, there is a trade-off between high active material loading and a high rate capability due to the low electronic conductivity of OEMs [252]. Organic/polymer materials can exhibit a high electronic conductivity with a high content of active material by fabricating 1D OEMs through ex situ growth on the surface of CNTs [227]. For example, lumiflavine (LF) shows high dissolution in electrolytes and a low rate capability due to its low molecular weight and electron-insulating feature, which limits its further application in MIBs [227]. A 1D LF-SWNT composite nanofibre (Fig. 20a) has been fabricated by sonicating LF with SWNTs in acetone, resulting in LF anchoring on the surface of the SWNTs via π-π interactions between the LF and SWNTs [227]. The LIB based on the 1D LF-SWNT nanofibres exhibits a higher reversible capacity (203 mAh g −1 ) and capacity retention (99.7%) than the LIB with a pristine LF electrode after 100 cycles due to less 1D LF-SWNT nanofibres dissolving in the electrolyte; this result stems from the π-π interactions between the LF and SWNTs along with the fast electron conduction and lithium-ion diffusion within the 1D LF-SWNT nanofibres (Fig. 20b) [227].
Apart from growing on the exterior surface of CNTs, electro-active organic/polymer can also infuse into the interior tubular space of the CNTs. Although poly(2,2,6,6tetramethyl piperidinyl oxy-4-vinyl methacrylate) (PTMA) has a high power density due to its fast reversible oxidation/ reduction process, PTMA suffers from severe self-discharge and a low energy density resulting from its high dissolution in electrolytes and high electron-insulating feature [250]. 1D PTMA-impregnated CNT nanofibres (Fig. 20c) have been fabricated through the facile mixing of PTMA and CNTs in N-methyl-2-pyrrolidone (NMP) solvent; the PTMA diffuses into the CNTs and impregnates in the vacant space within the CNTs [250]. The surface area of the CNTs decreases after being infused with PTMA, which demonstrates that the interior space of the CNTs can be filled with PTMA [250]. SIBs based on the 1D PTMA-impregnated nanofibres exhibit a much better rate capability (Fig. 20d) than that based on the PTMA-CNT active material prepared with the traditional Fig. 20  slurry method due to the better electron-conducting network and dissolution resistance of the 1D PTMA-impregnated CNT nanofibres [250]. In addition, SIBs with the 1D PTMAimpregnated CNT nanofibres show stable long-term cycling performance, demonstrating that the 1D morphology can effectively prevent the dissolution of PTMA in electrolytes [250].
Solid-state batteries (SSBs), which are composed of electrodes and solid-state electrolytes (SSEs), exhibit a low rate capability due to the low metal-ion diffusion rate in SSEs and high interfacial resistances at the electrode/SSE interfaces [233]. Therefore, achieving fast electron transfer and metal-ion diffusion rates are prerequisites to develop SSBs with high energy and power densities [233]. 1D PTMA-impregnated nitrogen-doped CNT (NCNT) nanofibres have been prepared by infusing PTMA into NCNTs in a mixture of NMP and acetone (Fig. 20e) [233]. A solidstate lithium battery (SSLB) has been assembled with 1D PTMA-filled NCNT nanofibres and a polyimide-based gel polymer electrolyte [233]. Even though PTMA is electron insulating, a fast electron transfer rate can be obtained due to the well-formed electron-conducting network of 1D PTMAimpregnated NCNT nanofibres [233]. The gel polymer electrolyte has an acceptable lithium-ion diffusion speed and low interfacial resistance due to the existence of a liquid electrolyte within the polyimide matrix, thereby guaranteeing a fast lithium-ion diffusion rate [233]. By virtue of the gel polymer electrolyte and 1D PTMA-filled NCNT nanofibres, the SSLB exhibits an impressively high reversible capacity (89.6 mAh g −1 ) at 20C and high capacity retention (higher than 80%) after 3 000 cycles at a current density of 1C (Fig. 20f), thus demonstrating the potential of using 1D PTMA-filled NCNT nanofibres for developing SSLBs with high power density [233].
Another top-down synthesis method for 1D OEMs is an antisolvent method. In addition to enhancing the rate capability of batteries, the 1D morphological structure of OEMs can benefit from the close electrode/electrolyte interfaces, thereby alleviating strain within the electrodes [223]. For instance, 1D croconic acid disodium salt (CADS) nanowires (Fig. 21a) with a mean diameter of approximately 150 nm have been synthesized through an antisolvent method by using water as the solvent and acetone as the antisolvent [223]. The formation of CADS nanowires may result from their low dissolution in the mixed water/acetone solution and the π-π interactions among CADS molecules, causing the self-assembly of CADS nanowires in an ordered direction [223]. LIBs based on the CADS nanowires exhibit a higher specific capacity (Fig. 21b) and better rate capability (Fig. 21c) than those based on CADS micropillars and microwires, which may result from the tight electrode/ electrolyte and CADS/carbon black interfaces, ultimately leading to a shorter lithium-ion diffusion distance [223]. In addition, the interface resistances of the CADS micropillar (300 Ω) and microwire (750 Ω) are much higher than that of the CADS nanowire (50 Ω), which is in accordance with the higher rate capability of the CADS nanowire [223]. Moreover, severe pulverization and microcracks are detected on the CADS micropillar and microwire electrodes after 110 cycles due to the large strain during the charge/discharge processes [223]. Nevertheless, no pulverization or microcracks appear on the CADS nanowire, demonstrating the advantage of the 1D CADS nanowire morphology for alleviating strains during the long-term cycling process [223].
In addition, 1D sodium rhodizonate dibasic (SR) nanorods with a uniform diameter of 200 nm have been prepared through the antisolvent method with water as the solvent and ethanol as the antisolvent (Fig. 21d) [251]. SIBs based on 1D SR nanorods exhibit a higher capacity (Fig. 21e) and better rate capability (Fig. 21f) than those with SR microbulks and microrods because 1D SR nanorods have a better contact interface with the electrolyte, which facilitates sodium-ion diffusion and decreases the length of the sodium-ion diffusion pathway [251]. Furthermore, the diameter of 1D SR nanorods can be easily controlled by adjusting the concentration of the SR solution, paving an effective approach to fabricate 1D OEMs with various sizes [251].

2D OEMs
2D OEMs are defined as those that have nanosheet, nanoplate and nanofilm morphologies [220,254]. 2D OEMs are regarded as promising candidates for preparing high-performance batteries due to their impressively high surface area, massive exposed active sites, and ultrafast electronic and ionic in-plane transport, all of which results in excellent rate capabilities and high specific capacities and energy densities [221,226,255,256]. However, most 2D OEMs undergo restacking, which hinders electrolyte penetration and ion diffusion and ultimately leads to a decrease in the power density and specific capacity of batteries [245,257,258]. Therefore, in this section, we summarize recent fabrications of 2D OEMs and their application in MIBs, providing effective and controllable approaches to fabricate highperformance 2D OEMs for constructing batteries with high power and energy densities.

Bottom-Up Synthesis for 2D OEMs
Bottom-up synthesis for 2D OEMs can be realized by three general methods in this work, including in situ polymerization [77,100,256,[259][260][261], in situ growth on a 2D template [230] and controlled precipitation [72,262]. The straightforward bottom-up synthesis method for 2D OEMs is in situ polymerization [77]. As one of the most promising 2D OEMs in MIBs, COFs usually exhibit a fast ion diffusion rate and stable long-term cycling performance due to their open ordered 2D morphology, large surface area, stable molecular structure, and π-π interactions among COF layers [96,256,259]. Therefore, many applications of COFs in MIBs have been reported due to the versatility and tunability of COFs, which can be found in published reviews [55,258,263]. Among these COFs, one of the most frequent synthesis methods of COFs is the Schiff-base reaction due to their widespread raw sources. A novel TQBQ-COF has been synthesized through a Schiff-base reaction in a mixed acetic acid/ethanol solvent, which has abundant electroactive sites, Fig. 22 [100]. Copyright 2013, Springer Nature. g TEM images of a COF (the inset is the HR-TEM image of the COF). h Rate capability of a magnesium-ion battery with a COF cathode. Reproduced with permission from Ref. [67]. Copyright 2020, American Chemical Society such as C=O and C=N, and 2D morphology with moderate crystallinity (Fig. 22a) [77]. 2D TQBQ-COF exhibits a large surface area (94.36 m 2 g −1 ), high electronic conductivity (1.973 × 10 −9 S cm −1 ) and good sodium-ion conductivity (5.53 × 10 −4 S cm −1 ), leading to the excellent rate capability of SIBs (Fig. 22b) based on the TQBQ-COF active material (134.3 mAh g −1 at 10 A g −1 ) [77]. In addition, the abundant C=O and C=N electroactive sites and decreased inactive parts endow TQBQ-COF with a surprisingly high specific capacity (452.0 mAh g −1 at 0.02 A g −1 ) and stable long-term cycling performance (96% capacity retention after 1 000 cycles at 1 A g −1 ) [77]. In addition, Wu et al. synthesized BQ1-COF with a 2D honeycomb-like morphology that had the same chemical structure as TQBQ-COF by a Schiff-base reaction with a mixed NMP/sulphuric acid solvent [264]. This BQ1-COF displayed good rate capability and long-term cycling performance [264]. Nevertheless, Li et al. reported that the Schiff-base reaction in the mixed acid/solvent would result in irreversible mishaps in the connectivity between vicinal diamines and vicinal diketone, thereby leading to the formation of OEMs with amorphous crosslinked structures and unsatisfactory battery performance [265]. In contrast, the polycondensation reaction between vicinal diamines and vicinal diketone is reversible in a basic solvent, such as sodium hydroxide (NaOH)/potassium hydroxide (KOH) aqueous solvent, facilitating the growth of COFs with crystalline structures through an error-checking process [265]. For example, PGF-1 COF, with high crystallinity and a 2D stacked-sheet morphology, has been prepared successfully in 4 M KOH aqueous solvent, showing a large surface area of 101 m 2 g −1 , a high specific capacity of 842 mAh g −1 at a current density of 100 mA g −1 and excellent rate capability (189 mAh g −1 at 5 A g −1 ) [265].
Another example of in situ polymerization is the distribution of 2D polyporphyrin (TThPP) COF nanofilms (3.8 nm) onto the surface of copper foil (Fig. 22c) [256]. The in-plane electronic conductivity of the TThPP nanofilm reaches 2.38 × 10 −4 S m −1 , indicating the fast electron conduction and excellent power density of the TThPP nanofilm [256]. Owing to the high electronic conductivity and enhanced transport of lithium ions and open nanopores of the TThPP nanofilm, LIBs based on 2D TThPP nanofilms exhibit a high specific capacity of 195 mAh g −1 at a current density of 4 A g −1 (Fig. 22d), which is higher than most reported LIBs based on OEMs and IEMs [256]. In addition, a high specific capacity (381 mAh g −1 ) is obtained after 200 cycles at a current density of 1 A g −1 , demonstrating the promising potential of 2D TThPP nanofilms for high power and energy density MIB applications [256].
Covalent triazine frameworks (CTFs), as a representative COF, can also be used to fabricate amorphous morphologies. This can be achieved by controlling the selfpolymerization reaction of cyano-containing monomers with a ZnCl 2 catalyst at a relatively high temperature (350-450 °C). For example, a BPOE COF containing benzene rings and triazine rings has been synthesized nanosheet. f Rate capabilities of SIBs based on bulky Na 2 TP and Na 2 TP nanosheets. Reproduced with permission from Ref. [262]. Copyright 2015, Elsevier through a self-polymerization method catalysed by ZnCl 2 , exhibiting a 2D nanosheet morphology with a width of 300 nm, a thickness of 4 nm (Fig. 22e), and high electronic conductivity (4.09 × 10 −6 S cm −1 ) [100]. In addition, the SIB based on the BPOE COF retains 80% of its initial capacity after 7 000 charge-discharge cycles and exhibits a reasonable rate capability ranging from 0.1 to 10 A g −1 (Fig. 22f), demonstrating the stable molecular structure and fast sodium-ion diffusion rate of the BPOE COF [100]. Moreover, another COF with the same chemical structure as the BPOE COF has been prepared, also displaying a 2D lamellar nanosheet morphology (Fig. 22g) [67]. Generally, OEMs have fast ion diffusion kinetics in LIBs and SIBs, but may not display fast Mg 2+ diffusion kinetics due to the radius of Mg 2+ being larger than Li + and Na + [67]. However, the amorphous 2D morphology with a large surface area (428 m 2 g −1 ) of this COF not only facilitates the filtration of liquid electrolyte into the interior space of the COF but also shortens the Mg 2+ diffusion distance, ultimately leading to ultrafast Mg 2+ diffusion and excellent rate performance (Fig. 22h) [67]. Additionally, the magnesium-ion battery based on this COF active material displays an ultraslow rate of capacity loss (0.019 6% per cycle) at a high current density of 5C, making it one of the best rechargeable magnesium-ion batteries [67].
Even though COFs have abundant electroactive sites, most COFs exhibit serious aggregation due to the strong π-π interactions among COF layers, which can block redoxactive sites and limit the reversible capacity and ion diffusion rate of batteries [266]. This issue can be addressed by in situ growth on 2D inorganic templates. The resultant 2D OEMs not only effectively prevent the agglomeration of COFs but also significantly enhance the electron conduction and ion diffusion rate within COFs [226]. As shown in Fig. 23a, multiple layers of a pristine poly(imide-benzoquinone) (PIBN) COF without the addition of graphene agglomerate as nanoflakes [226]. In contrast, 2D PIBN-graphene (PIBN-G) nanosheets have been fabricated via in situ polymerization between tetramino-benzoquinone (TABQ) and pyromellitic dianhydride (PMDA) precursors, which grow on the surface of graphene [226]. The thin, 2D crystalline nanosheet morphology of PIBN-G (Fig. 23b) has an effective inhibitory effect on the agglomeration of PIBN, ultimately facilitating metal-ion diffusion into the micropores of PIBN-G [226]. LIBs based on 2D PIBN-G nanosheets display a much better rate capability (Fig. 23c) than those based on pristine PIBN owing to the more exposed electroactive sites and enhanced electronic conductivity of 2D PIBN-G nanosheets [226]. However, the energy density of the battery decreases with an increasing addition of graphene, which is a trade-off that should be taken into careful consideration. Therefore, the ideal method for preventing the agglomeration of COFs may be self-exfoliation, i.e. fabricating COFs with fewer layers that resist agglomeration due to the forces within their molecular structures instead of adding redox-inactive templates [260,[267][268][269][270].
Regarding controlled precipitation, 2D nanosheets with lamellar morphology can be formed by strong in-plane covalent forces and relative out-of-plane van der Waals forces [72]. For instance, bulky disodium terephthalate (Na 2 TP) has been synthesized by directly adding terephthalic acid into a benzene-containing solution, exhibiting a spheroidal structure with a diameter of approximately several micrometres (Fig. 23d) [262]. In strong contrast, 2D Na 2 TP nanosheets ( Fig. 23e) have been prepared by adding a dimethylformamide solution of terephthalic acid into a benzene-containing mixed solution [262]. SIBs based on the 2D Na 2 TP nanosheets display a high reversible capacity (248 mAh g −1 ) at a current density of 25 mA g −1 , which is very close to its theoretical capacity (255 mAh g −1 ) [262]. In addition, SIBs with a Na 2 TP active material exhibit a much better rate capability (Fig. 23f) than those with bulky Na 2 TP due to the 2D Na 2 TP nanosheet structure [262].

Top-Down Synthesis for 2D OEMs
2D OEMs, such as COFs, tend to agglomerate due to the intermolecular π-π and van der Waals interactions among layers, which ultimately leads to a decrease in the number of exposed electroactive sites and in the specific energy density [258,271]. Although 2D OEMs can be fabricated by the in situ growth of COFs on 2D inorganic frameworks, the addition of electroinactive components can decrease the overall energy density of a battery. Notably, 2D OEMs can be prepared via top-down synthesis, i.e. the postexfoliation of COFs [224]. Four top-down synthesis methods for 2D OEMs are introduced in this work, including ball milling, chemical exfoliation, supersonication, and antisolvent methods [225,257,272].
Ball milling could be the most reported exfoliation method, and it utilizes mechanical force to overcome the weak π-π interactions among layers while not affecting the strong covalent bonds of layered materials. For example, although an anthraquinone-based COF (DAAQ-TFP-COF) has a 2D morphology, it exhibits a bulky and stacked morphology due to the π-π interactions of DAAQ-TFP-COF (Fig. 24a) [49]. In contrast, DAAQ-ECOF (Fig. 24b) is prepared by delaminating DAAQ-TFP-COF through ball milling at a vibration frequency of 50 Hz for 0.5 h, resulting in an ultrathin and transparent nanosheet structure with a thickness of approximately 5 nm and a high surface area of approximately 216 m 2 g −1 [49]. The specific capacity of LIBs based on DAAQ-ECOF (Fig. 24c) is much higher than that of LIBs with DAAQ-TFP-COF during 1 800 cycles because more of the electroactive sites of DAAQ-ECOF are exposed after ball milling [49].
The delamination of the bulky fluorinated covalent triazine framework (FCTF) (Fig. 24d) by ball milling is another excellent example to demonstrate the effectiveness of the ball milling method [273]. The exfoliated FCTF (E-FCTF) displays an ultrathin and layered structure (Fig. 24e), which facilitates fast lithium-ion diffusion kinetics and a large surface area (583 m 2 g −1 ) [273]. Moreover, LIBs based on E-FCTF exhibit a much improved specific capacity (1 035 mAh g −1 ) compared with that based on pristine FCTF (576 mAh g −1 ) after 300 cycles at a current density of 0.1 A g −1 (Fig. 24f) [273].
Ball milling can also be used to delaminate COF-CNT composite OEMs [274]. As shown in Fig. 24g and h, E-CIN-1/CNT displays a more ultrathin and transparent layered structure than CIN-1/CNT, leading to E-CIN-1/CNT (525.1 m 2 g −1 ) having a higher surface area than CIN-1/ CNT (265.3 m 2 g −1 ) [274]. The specific capacity of LIBs with E-CIN-1/CNT is higher than that of LIBs with CIN-1/ CNT during 250 cycles at 0.1 A g −1 (Fig. 24i), which possibly results from the improved lithium-ion diffusion kinetics and better electrolyte infiltration within E-CIN-1/CNT [274].
In addition, 2D OEMs can undergo chemical exfoliation. As an example of chemical exfoliation, IISERP-COF7 prepared through a Schiff-base reaction between 2,4,6-triformylresorcinol and 2,6-diaminoanthracene exhibits the stacking aggregation of several layers (Fig. 25a) [275]. To alleviate the severe stacking among layers, the Diels-Alder reaction between maleic anhydride and anthracene moieties on IISERP-COF7 successfully exfoliates IISERP-COF7 into a less aggregated 2D sheet-like morphological structure, resulting in IISERP-CON2 (Fig. 25b) [275]. The LIB based on IISERP-CON2 displays a much higher specific capacity than that based on pristine IISERP-COF7 (Fig. 25c) [275]. The applicability is dependent on the functional groups on the layer structure. In this case, this chemical exfoliation method is not suitable for COFs without anthracene moieties.
A new imine-based bulky TFPB-COF has been prepared through a Schiff-base reaction, which displays a multilayer stacked morphological structure (Fig. 25d), leading to an increase in buried electroactive sites for redox reactions with lithium ions [276]. After a reduction reaction occurs between KMnO 4 and HClO 4 in the mixture containing TFPB-COF, a 2D E-TFPB-COF/MnO 2 composite nanostructure is fabricated after the intercalation of MnO 2 nanoparticles into the TFPB-COF layers; this effectively prevents the severe stacking of E-TFPB-COF/MnO 2 layers (Fig. 25e) [276]. With the few-layered 2D nanostructure of E-TFPB-COF/MnO 2 , the LIB exhibits a higher specific capacity (1 359 mAh g −1 ) than the total specific capacities of MnO 2 and E-TFPB-COF after 300 cycles (Fig. 25f), thereby demonstrating the positive effects of chemical exfoliation on improving battery capacity [276].
Supersonication is another effective and facile approach to fabricate 2D OEMs [277]. For instance, organic tetralithium salts of 2,5-dihydroxyterephthalic acid (Li 4 C 8 H 2 O 6 , Li 4 DHTPA) with a bulky morphology have been prepared by reacting 2,5-dihydroxyterephthalic acid (DHTPA) with lithium methoxide [277]. Li 4 DHTPA nanoparticles with varying diameters of ~ 50-200 nm can be obtained by decreasing the size of bulky Li 4 DHTPA through ball milling, which still exhibits severe agglomeration [277]. Subsequently, 2D lamellar Li 4 DHTPA, in which the thickness of each sheet is several nanometres, is fabricated by supersonication in three different solvents, i.e. methanol, ethanol and diethyl ether [277]. A better nanosheet morphology of Li 4 DHTPA can be obtained by supersonication in methanol than in ethanol and diethyl ether [277]. In addition, LIBs based on 2D Li 4 DHTPA nanosheets exhibit better overall battery performance than LIBs with bulky Li 4 DHTPA and Li 4 DHTPA nanoparticles [277].
2D OEMs can also be fabricated by antisolvent methods, which have been described in the top-down synthesis of 0D and 1D OEMs [147]. As an example, after NG-HCP, a bulky COF is synthesized by polymerization, and then, a 2D NG-HCP nanosheet is successfully prepared by the antisolvent method with methanesulphonic acid as the solvent and deionized water as the antisolvent [147]. LIBs based on the as-prepared 2D NG-HCP nanosheet active material exhibit a high specific reversible capacity of 257 mAh g −1 at 2.3 A g −1 and stable long-term cycling performance, demonstrating the superior advantages of the antisolvent method for preparing 2D OEMs for use in MIBs [147].

3D OEMs
The unsatisfactory rate capability and insufficient electroactive sites stemming from a low electronic conductivity (including conducting polymers) and an insufficient surface area for exposing active electrochemical reactive sites are major challenges towards the large-scale application of OEMs in MIBs [278]. The construction of 3D nanostructured OEMs can simultaneously address both of these challenges. Although the 3D nanostructure of OEMs can be fabricated via the careful control of reaction conditions (Table 3), it is too sophisticated and costly for scaled-up production. Fortunately, a wide spectrum of conductive inorganic templates, including CNTs, graphene, graphene oxide (GO), and reduced graphene oxide (rGO), are readily available for the construction of 3D OEMs [113,279,280]. Through bottom-up and top-down synthesis, 3D OEMs can be fabricated by integrating organic materials with these conductive inorganic frameworks to build an electron-conducting network and enlarge the surface area, which leads to batteries with enhanced rate capabilities and reversible capacities [170,235].

Bottom-Up Synthesis for 3D OEMs
3D OEMs are commonly synthesized via the assistance of templates, such as inorganic nanostructured frameworks (graphene, graphite or CNTs). In the bottom-up synthesis, polymer precursors (i.e. small organic molecules) can be evenly distributed onto the templates, and the resultant 3D OEMs establishes extensive contact with the inorganic substrates after in situ polymerization [174,[281][282][283]. Therefore, the high electronic conductivity and fast metalion diffusion rate of 3D OEMs are expected when used in MIBs. For example, polyimide@CNT (PI@CNT) has been fabricated through in situ polymerization of PI monomers in NMP solvent in the presence of CNTs [120]. The CNTs are distributed randomly within the 3D PI/CNT OEM, and a 3D conductive network is formed among the different domains of PI (Fig. 26a) [120]. Impressively, the reversible capacities of multivalent-ion batteries (Mg 2+ and Al 3+ ) at 20C are closely equivalent to that of LIBs, while retaining 55% of their reversible capacities at 0.5C due to fast electron conduction and high ion diffusion through the 3D conductive matrix (Fig. 26b) [120]. Functionalized graphene sheets (FGSs) have been added into the precursor of poly(anthraquinonyl sulphide) (PAQS), which are then grown uniformly by PAQS after in situ polymerization (Fig. 26c) [284]. Compared with the electronic conductivity (1 × 10 −11 S cm −1 ) and surface area (30 cm 2 g −1 ) of pristine PAQS, PAQS-FGS-b (26 wt% FGS) exhibits a much higher electronic conductivity (6.4 × 10 −3 S cm −1 ) and larger surface area (153 cm 2 g −1 ), leading to a higher utilization ratio (95%) of electroactive sites [284]. Due to these merits, LIBs based on 3D PAQS-FGS OEMs exhibit better rate capabilities than those based on pristine PAQS electrodes (Fig. 26d), especially at high current densities [284].
The 2D material rGO is another popular template for the construction of 3D OEMs. As an example, rGO has been added to the monomers of polyimide (PI); in this case, rGO is wrapped by the PI and forms a 3D graphene-polyimide (GF-PI) OEM after in situ polymerization (Fig. 26e) [285]. The size of PI in GF-PI is approximately 100-150 nm, which is much smaller than that of pristine PI (0.5-1.0 μm) [285]. This suggests that template synthesis can be used to control the growth of polymers [285]. Furthermore, the as-prepared 3D GF-PI OEM can be compressed into a 30-μm-thick electrode [285]. Most impressively, without the addition of any binder and conductive carbon, the fabricated 3D GF-PI OEM still displays high electronic conductivity (2.38 S cm −1 ) [285]. Moreover, the resultant 3D GF-PI OEM displays a better rate capability (102 mAh g −1 at 4 A g −1 ) than LIBs based on corresponding PI cathodes (Fig. 26f) [285].
Even though the addition of conductive carbon additives benefits the increase in reversible capacities due to the enhanced electronic conductivities of 3D OEMs, the excessive addition of conductive carbon additives can significantly decrease the energy densities of batteries [286]. In reality, the preparation of 3D OEMs with a small amount of conductive carbon additives is extremely challenging with regard to achieving decent electrochemical performance. Sun et al. fabricated a 3D carbonyl-based OEM (PI10G) with only 10 wt% graphene (Fig. 26g) through the in situ polymerization of polyimide with graphene [286]. Such an OEM exhibits a much better rate capability (159.8 mAh g −1 ) than PI5G (71.1 mAh g −1 ) and pristine PI (18.9 mAh g −1 ) due to the significant increase in the electronic conductivity of PI10G (Fig. 26 h) [286]. Moreover, LIBs with PI10G can retain 86.6 mAh g −1 after 1 000 cycles at a current density of 50C, demonstrating their excellent power density and high specific capacity [286].

Top-Down Synthesis for 3D OEMs
3D OEMs can be prepared by top-down methods, i.e. to disperse bulky OEMs into the 3D structures of other materials. This process not only mitigates the dissolution of organic material in electrolytes but also realizes the fabrication of free-standing, flexible 3D OEMs (without the addition of any binder) [234,287,288]. For instance, Vat green 8 (VG 8) is a promising organic material to fabricate MIBs with high specific energy density due to its abundance of electroactive sites [289]. However, the severe dissolution of VG 8 in electrolytes and the decrease in the number of electroactive sites by being buried in the stacked structures hinder their practical application as OEMs [289]. Through intercalating VG 8 into graphene layers, VG and graphene are connected via weak π-π interactions, forming a 3D VG-GF OEM (Fig. 27a) [289]. Such a structure prevents the severe aggregation and stacking of VG 8, significantly shortening the lithium-ion diffusion distance, providing fast electron conduction, and leading to a better rate capability compared to pristine VG 8 (Fig. 27b) [289]. As another example, a 3D free-standing P(DA 87 -stat-LiAMPS 13 )/CNT OEM has been fabricated through facile vacuum filtration of a mixture of P(DA 87 -stat-LiAMPS 13 ) and CNTs [200]. Without the  addition of binders, 3D P(DA 87 -stat-LiAMPS 13 )/CNT OEM is prepared via the bioadhesive function of the catechol pendant groups on P(DA 87 -stat-LiAMPS 13 ) (Fig. 27c); this OEM exhibits an excellent rate capability and high reversible capacity because the resulting 3D continuous conductive network affords enhanced ion diffusion within the electrode during the charge/discharge processes (Fig. 27d) [200].
With ideal OEMs, binder-free and current collector-free electrodes can be fabricated, which can further improve the energy densities of batteries. A 3D PDHBQS-SWCNT film has been fabricated by vacuum filtration of a mixture of PDHBQS and SWCNTs, forming 3D interconnected networks (Fig. 27e) [290]. 3D PDHBQS-SWCNT films have such high electronic conductivities (125 S cm −1 for PDHBQS-30% SWCNTs) that a current collector (such as aluminium foil or copper foil) is no longer needed [290]. 3D PDHBQS-SWCNT films display a better rate capability (Fig. 27f) than conventional film electrodes (with a binder and current collectors) in LIBs [290]. The as-prepared 3D interconnected network of PDHBQS-SWCNT films is not only electronically conductive but also mechanically robust, as evidenced by its high Young's modulus (2.8 GPa) and tensile strength (20.1 MPa) [290].

Conclusion
In summary, significant efforts have been devoted to exploring new charge/discharge mechanisms, novel active bonds, suitable electrolytes, and compatible aqueous and nonaqueous battery systems, for extending the application of OEMs in MIBs. The development of next-generation OEMs for MIBs can be achieved by molecular and morphological engineering. Molecular engineering of OEMs can be carried out based on the following two aspects: (1) the introduction of the redox mechanisms of n-type, p-type, and bipolartype OEMs and (2) the grafting of various redox groups (resulting in 6 categories of OEMs, including carbonyl, sulphur-containing, nitrogen-containing, conducting polymer, radical-based, and overlithiated OEMs) to systematically reveal their molecular reactions in battery processes, material merits and challenges. Notably, the molecular engineering of OEMs facilitates the design and tuning of molecular backbones (i.e. small OEMs and polymeric OEMs), which is associated with different functional groups (i.e. redox groups, electron-withdrawing, or electron-donating); the results of this engineering are regarded based on the energy density, power density, and long-term cycle life.
Morphological engineering can tailor the interfacial properties of OEMs and electrolytes, forming 0D, 1D, 2D, and 3D nanostructured OEMs through bottom-up and top-down methods; these methods can facilitate fast electronic conduction and ion diffusion along with a stable electrode structure. The challenges of low capacity, high solubility, low electronic conductivity, and low stability can also be addressed by forming composites with other functional materials. This provides effective and inspiring guidance to construct high-performance OEMs for practical application in MIBs application.

Perspectives
Affordable, recyclable, and electrochemically reversible OEMs have shown great potential as energy storage and conversion materials with high capacity, safety, and cost-effectiveness. In the aforementioned sections, we established that significant progress in addressing the challenges of OEMs was achieved via molecular and morphological engineering during the synthesis of OEMs.
From the viewpoint of the practical application of OEMs, the strategy, i.e. the incorporation of OEMs with functional and conductive nanomaterials, can be applied to address these challenges in the course of fabricating the electrodes, especially for the large-scale application of OEMs in rechargeable battery systems.
(i) The dissolution of OEMs in electrolyte solution remains an inherent drawback, especially after nanosizing. During electrode fabrication, OEMs can be incorporated with other functional materials. The incorporation of functional additives, such as polar conductive materials, to promote intra/intermolecular forces (e.g. extended conjugation, mixing with inorganic conductive substrates) may solve/mitigate the dissolution problems of OEMs [291]. (ii) The low conductivity of OEMs and the aggregation of nanostructured OEMs lead to a decrease in capacity and rate capability [292][293][294]. The incorporation of OEMs with conducting nanomaterials, such as graphene and carbon nanotubes, can inhibit aggregation and solve the low conductivity dilemma of OEMs [295,296].
From the viewpoint of the design and development of OEMs, strong theoretical guidance must be adopted to improve the usefulness and efficiency of OEMs. The traditional trial-and-error method has large operation time and resource costs during the selection, design, and synthesis of OEMs. Thus, the application of theoretical and computational chemistry towards OEMs could offer a series of potential benefits, including the prediction of ideal molecular structures, functional groups, morphologic configurations, and even synthesis costs. Over nine million organic compounds and corresponding synthesis methods have been used to form a solid statistical foundation for machine learning in artificial intelligence (AI). AI enables the extraction of the design, characterization and electrochemical performance of OEMs from different publications and other domains, making OEM structures and properties easily accessible. Furthermore, AI can automatically research, design, and optimize novel OEMs with the most straightforward synthesis steps and the lowest cost. Theoretical models of OEMs could be built and run through computational methods (i.e. the density functional theory and molecular dynamics) to predict the physical properties and battery behaviour, and this method has already been used in previous literature [34]. Therefore, the broad utilization of computational chemistry can accelerate the promotion of a new generation of OEMs for energy storage.
From the viewpoint of practical applications, OEMs could be used for fabricating symmetric or asymmetric electrodes for solid-state electrolyte utilization in all-solidstate organic battery (ASSSB) configurations due to their sustainability, high flexibility, high capacity, and abundant resources (from biomass to industrial products). ASSOBs will be one of the most promising flexible battery systems for smart energy devices because of their high competitiveness with regard to safety, whole gravimetric/volumetric energy density, and low price. The cycling stability, energy capacity, and power density could be enhanced in ASSOBs because the inherent high solubility issue in organic electrolytes can ultimately be overcome in solidstate electrolytes. This novel battery design has already had a successful case, namely an ASSOB configuration of Na 4 C 6 O 6 cathode/Na 3 PS 4 (NPS) solid electrolyte/Na 4 C 6 O 6 anode [239]. Na 4 C 6 O 6 enables the storage of 2 Na + at the two phenolate groups at a high working potential (acting as the cathode), while it stores an additional 2 Na + at the two carbonyl groups at a low potential (acting as the anode).
OEMs could be universal electrode materials for dualion batteries (DIBs), simultaneously undergoing the reversible intercalation/extraction of anions (e.g. PF 6 − , ClO 4 − , BF − and TFSI − ) at the positive electrode and cations (e.g. Li + , Na + , and K + ) at the negative electrode. These unique DIBs could significantly increase the rate capability and energy density because they allow simultaneous reactions on the two sides of the electrode. OEMs, such as thianthrene [297], coroene [298], and terephthalate [299], could act as anion-accepting cathode materials in DIBs to replace traditional graphite. The unique anion intercalation reaction at the cathode is favourable to a high working potential and fast battery kinetics, leading to enhanced energy density and power density. According to the redox mechanism of OEMs, high voltage stability and active OEMs, i.e. p-type radical OEMs and conducting polymers, are recommended to store different anions with high capacity and stability.
Many kinds of advanced characterization techniques are critical to obtaining material information on OEMs. Mass spectrometry (MS), nuclear magnetic resonance (NMR) spectroscopy, Fourier transform infrared (FT-IR) spectroscopy, and Raman spectroscopy are the usual methods to characterize the molecular structure of OEMs. Scanning electron microscopy (SEM) and transmission electron microscopy (TEM) are essential to study the morphological features of OEMs at the microlevel. In addition to these methods, in situ characterization techniques (e.g. in situ FT-IR spectroscopy, in situ Raman spectroscopy, and in situ X-ray diffraction (XRD)) have been utilized to effectively study the redox mechanism of OEMs during the charge/discharge processes. Moreover, in situ synchrotron X-ray techniques are another good choice to obtain adequate and reliable information.