The growing concerns with regard to environment and sustainability have promoted the development of renewable energy resources, such as wind and solar energy [1]. Due to the intermittent nature of these renewable energy sources, the development of energy storage devices has become an actual necessity. Alkali ion batteries, such as lithium ion batteries (LIBs), sodium ion batteries (SIBs) and potassium ion batteries (PIBs), are considered as promising energy storage devices because of their high energy and power densities. Particularly, SIBs receive great attention as the next generation low cost alternative by virtue of abundant sodium resources ^[2][3][4][2,3,4]. In addition, the abundance of potassium in the crust is close to that of sodium, and the redox potential of K⁺/K (−2.92 V) is similar to that of lithium and much lower than that of Na⁺/Na at −2.714 V. Therefore, PIBs have also greatly attracted the interest of researchers ^[5][6][5,6]. At present, a series of anode materials such as carbon and derivatives, oxides and alloys ^[7][8][7,8] have been exploited. Nevertheless, carbon materials, due to their low cost, suitable working potential and high stability, were always the first choice for anode materials since the first generation of commercial LIBs was developed. Graphite has already made a wide and profound impact as the first generation of commercial anode materials. Additionally, hard carbon is considered to be the most promising and probably the only alternative anode material for SIBs, as sodium ions could not be inserted due to the narrow layer spacing of graphite [9]. Hard carbon can exhibit a high capacity benefiting from its expanded interplanar spacing and significant porosity for ion accommodation. However, carbon anode materials still suffer from drawbacks such as low initial coulombic efficiency (ICE) and sluggish rate capability in many practical applications ^[10][11][12][10,11,12].

During the operation of alkali ion batteries, a solid electrolyte interphase (SEI) forms on the carbon material surface rising from the decomposition of electrolyte at low potential, which is beyond its stable window [13]. It has been widely accepted that SEI is essential to alkali metal-ion batteries and the initial coulombic efficiency (ICE), calendar life, cycle life, rate capability as well as safety of the batteries are dramatically influenced by it ^{[14][15][16][17][18]}[14,15,16,17,18]. Therefore, it is vital to control its formation and growth with suitable chemical composition, morphology and stability, so as to maintain a good passivating and electrochemical stability over long-term cycling. Researchers have extensively worked on the formulation of electrolytes, including solutes, solvents and additives, etc., aiming to obtain an advanced SEI on carbon materials. Some excellent reviews have summarized the research progress of SEI for carbon materials with a focus on the impact of electrolytes ^[19][20][21][19,20,21].

Carbon materials prepared by the pyrolysis of organic compounds can usually be divided into two main categories: graphitizable and non-graphitizable carbon ^[22][52]. The most important structural difference between graphitizable and non-graphitizable carbon is the nano-texture. The former has a relatively well orientated graphitic domain, while the latter has a randomly oriented nanotexture of graphitic nano-domains and contains a large number of micropores. Increasing heat treatment temperature leads to the growth of crystal size with expansion and stacking of graphitic layers. For non-graphitizable carbon, the crystal growth is limited by the local orientation, even at high temperatures above 3000 °C. Thus, non-graphitized carbon, often described as hard carbon (HC), has low graphitization, low apparent density and high porosity. In general, HC is amorphous and cannot be graphitized even at very high temperatures. On the contrary, soft carbon, which can also be referred to as graphitizable carbon, has an ordered structure. Artificial graphite with good crystal structure can be obtained after graphitization of soft carbon. Graphite has a layered structure consisting of sp² hybridized carbon of a hexagonal packing. Hexagonal graphite or rhombic graphite is formed by the orderly stacking of carbon layers in different rules ^[23][53]. While hard carbon does not have a long-range ordered structure with planar extensions or carbon layer stacks, it is characterized by randomly arranged sp² graphitic microregions that are partially cross-linked with amorphous regions of sp³ hybridized carbon atoms. Different topologies lead to diverse electrochemical behaviors of the materials. Additionally, the exposure of a distinct crystal surface has an important impact on the catalytic production of SEI on it ^[24][54].

Highly oriented pyrolytic graphite (HOPG) is a kind of “crystalline” graphite with good crystalline structure, excellent orientation and a high degree of graphitization. Time-of-flight secondary-ion mass spectrometry (TOF SIMS) and X-ray photoelectron spectroscopy (XPS) measurements were carried out for SEI formed on the different planes of HOPG ^[25][55]. The SEI on both the basal planes (carbon layer extension direction) and cross-sectional (edge) planes (carbon layer stacking direction) are present with C₂H (or other C_xH_y-based fragments), O, C₂H₃O₂ and C₂H₃O, which cover the material surface intactly and show a fairly high degree of homogenization. The SEI on the basal plane is dominated by organic matter, with oxygen and C₂H₃O being the main components. Lithium and fluorine are concentrated in a large region (100 μm), with some smaller micron-sized particles as well ^[26][56]. The distinctive feature of the basal SEI is the presence of 10 to 30 atomic% Li₂CO₃ on the surface and in the bulk. The SEI on the edge plane consisted mainly of lithium and fluorine with a Li/F ratio close to one, but there are also regions of one to several tens of microns in size where both elements are almost absent. Moreover, E. Peled et al. ^[27][57] showed that the thickness of SEI on the basal plane is thinner relative to that on the edge plane. SEI thickness measured by XPS is 7 nm for basal-SEI and 35 nm for edge plane SEI. Thick SEI at the edge plane was explained by hydrogen permeation and solvent co-intercalation and exfoliation. A study reported by Eshkenazi et al. ^[28][58] showed a possible co-intercalation layer of protons as HF dissociation products, resulting in a thick SEI. Partial exfoliation of graphite by Li₂PF₆⁺ and hydrogen permeation may cause an increase in SEI thickness, which is similar to the results on HOPG cross section in LiPF₆ electrolyte. The content of PO₃⁻ species on the edge plane was three times higher than that on the basal plane revealed by TOF SIMS.

The influence of active surface area (ASA) on graphite materials was investigated by Nova’k et al. ^[29][61] and their study demonstrated that ASA is the key parameter affecting graphite surface passivation and graphite layer exfoliation. Graphite with smaller particle sizes have a larger specific surface area and more prismatic surfaces (edge plane), promoting the catalytic decomposition of electrolyte to produce SEI. The electrolyte preferentially decomposes to form SEI at the edge plane, where it has more active sites and provides higher reactivity. It makes it possible to complete stable passivation of graphite before exfoliation occurs, thus suppressing graphite exfoliation. Spahr et al. ^[30][62] demonstrated that the heat-treated material, which has high crystallinity, low superficial defect concentration and low ASA value, has a low surface reactivity toward electrolyte concentration, thus hindering the SEI generation. Ion insertion and de-insertion without complete formation of the passivation layer will result in a greater tendency for the graphite layer to exfoliate. S.H. Ng et al. ^[31][63] showed that the critical value of ASA is 0.2 m² g⁻¹ and the tendency of graphite layer exfoliation is suppressed when the value of ASA is higher than 0.2 m² g⁻¹. The exfoliation of the graphite can also be suppressed to some extent by the formation of chemically bonded SEI at the cross section, which was confirmed by the study of Buqa et al. ^[32][64]. Under the mild oxidation with carbon dioxide and oxygen, the morphology changes at the prismatic surfaces ranging between “nano-roughness” after 15 min and “channel-like” after 69 h of treatment, which provides a favorable nucleation site for the chemically bonded SEI.

The carbon material has a higher reactivity at the edge plane, with a large number of SOG at the edges when the treatment temperature is below 1000 °C ^[33][71]. Firstly, the wettability of the carbon material surface with the electrolyte is improved by the SOG, which will facilitate the diffusion of the electrolyte inside the material and contribute to the formation of intact SEI ^[34][72]. Secondly, surface oxygen groups can act as a link between the electrode material and SEI, and improve surface charge transfer and influence the exchange current density during SEI formation, thereby reducing electrode overpotential during operation ^[35][59].

SOG exhibits a high activity and can catalyze or even directly participate in reactions in electrochemical processes. In earlier studies, it has been demonstrated that some components, such as -COO-Li⁺ and -O-Li⁺, of the inchoate SEI monolayers could be obtained through the treatment of surface carboxyl groups with aqueous solutions of alkali metal hydroxides ^[33][71]. When the first insertion process begins, these acid groups are converted into surface lithium carboxylates and surface O-Li groups, which become components of SEI. These species have been shown to be nucleation sites for electrolyte decomposition ^[31][63], and the absence of these species may hinder decomposition reactions as well as increase the potential for graphite layer exfoliation. The changes in SEI of SIBs during charging and discharging were studied using FTIR in the report by Chen et al. ^[36][73] and they demonstrated the involvement of SOG in the sodium intercalation reaction.

Another advantage of SOG is the formation of chemical bonds between the SEI and the groups (such as -COO and -CO) attached to the surface, which improves the stability of the SEI. HC was obtained from carbonized sucrose and placed in an oxygen plasma cleaner in the study by Xie et al. ^[37][74]. Oxygen-containing functional groups, including carbonyl and hydroxyl groups, were successfully introduced on the surface of HC. The sample prepared in this way was named hydrophilic hard carbon (HHC). The defects of hard carbon, especially C=C/C−C at the edges and nanopores, exhibit high activity in decomposing sodium salts and electrolyte at low voltages, leading to irreversible capacity loss from electrolyte decomposition. Due to the high oxidation of oxygen plasma, the active carbon sites at the nanopores or on the edges were converted to ether or carbonyl groups, which were less chemically active and inhibited the decomposition of the electrolyte. The electron paramagnetic resonance (ERP) test showed that the unpaired electrons in the HHC electrode are stable, while the HC can easily generate free radicals to attack the electrolyte due to the high concentration of sp³ C and sp³ O, causing a large amount of electrolyte decomposition. A clear plateau was seen in the first discharge curve for HC at around 0.5 V, indicating the formation of SEI, while such a plateau disappeared for HHC. Additionally, the reduction peak at 0.7 V for HC in the cyclic voltammetry (CV) curves was assigned to the formation of SEI, while the reduction peak at 0.82 V for HHC indicates the reaction of carbonyl with sodium ions.

The presence of oxygen species on graphite surfaces increases the electrolyte reduction potential vs. Li/Li⁺ and contributes to the early SEI formation before lithium intercalation ^[38][75]. Solvent co-intercalation was inhibited by the SOG to avoid collapse of the graphite structure ^[33][71]. Additionally, the amount of surface oxide groups seems to be a key factor affecting graphite exfoliation during electrochemical lithium insertion. A study reported by Spar et al. ^[39][76] demonstrated that graphite materials exposed to air (room temperature) after heat treatment chemisorbed a small amount of oxygen on the surface. The SOG formed by this process exhibited a basic character, such as hydroxyl and ether groups. The SEI formed on the graphite surface with these basic SOGs cannot prevent graphite exfoliation. In contrast, when graphite is oxidized with large amounts of oxygen at high temperatures, the carbon surface becomes increasingly acidic due to the production of carboxyl and phenolic groups. The presence of acidic SOG contributed to the formation of favorable SEI layers and inhibited solvent co-intercalation, thus avoiding the tendency of graphite layers to exfoliate during the initial lithium intercalation.

Many studies have demonstrated that electrolyte decomposition and side reactions can be suppressed by the nitrogen on the surface, which facilitates the formation of favorable SEI ^[40][41][42][77,78,79]. The introduction of nitrogen-containing groups modifies carbon materials in two ways: one is to regulate the electronic structure of the carbon material through polarization caused by the difference in electronegativity between the nitrogen and carbon atoms, thus changing the formation process and chemical composition of the SEI; the second is to improve the ion storage capacity through a pseudocapacitance mechanism ^[43][80].

Interfacial nitrogen bonds catalyze the production of more inorganic compounds and the increase in the inorganic layer ensures the stability of SEI ^[44][81]. This is very beneficial to prolong the cycle life of the carbon material. Yuan et al. ^[45][82] showed that the nitrogen groups catalyze the decomposition of potassium bis(tri-fluoromethylsulfonyl) imide (KFSI) to produce more inorganic sulfur-containing compounds and inorganic substances containing S-F and K-F bonds. XPS measurements on the pristine P/N co-doped three-dimensional interconnected carbon nanocage (denoted as PN-CNC) electrode used as the anode of potassium-ion batteries (PIBs) do not obtain peaks of S or F. However, after the first discharge, the spectrum of S indicates the decomposition of KFSI and the formation of S-F and K-F. Additionally, a very low intensity RO-COOK signal was observed in the spectrum of C indicating that the amount of organic substances in the SEI was very small, while inorganic components were predominant. The SEI with more inorganic components endows the electrode with extremely stable cycling performance, maintaining an ultra-high capacity of 188.7 mAh g⁻¹ for 3000 cycles at a current density of 2 A g⁻¹. Additionally, XPS studies have shown that the process of discharging to 0.01 V and then charging to 2.5 V corresponds to a significant weakening of the P and N peak until it disappears and then recurs. This suggests that P and N directly participates in the electrochemical process of potassium ion storage. A similar study was reported by Yi et al. ^[46][83] in the DME electrolyte with a higher KFSI concentration, in which a multi-dimensional N-doped carbon nanopolyhedron@nanosheet (M-NC) anode of PIBs achieved excellent cycling stability, maintaining a capacity of 289 mAh g⁻¹ at 0.5 A g⁻¹ after 1600 cycles, and the coulombic efficiency was maintained at near 100% in PIB.

HC is usually obtained by the pyrolysis of organic materials, which generates small molecules such as CO, CO₂ and H₂O during pyrolysis, resulting in a large number of mesopores and micropores in the structure ^[23][53]. Numerous studies ^{[47][48][49][50]}[84,85,86,87] have shown that as the pyrolysis temperature increases, the porosity of carbon materials first increases because of the vaporization of these small molecules. When exceeding a certain temperature (around 1000 °C), the specific surface area of HC decreases, small pores merge into large pores and the total pore volume decreases. The structural evolutions of the HC at various pyrolysis temperatures has been reviewed by Zhao et al. ^[51][88]. The conditions of heat treatment plays a crucial role in determining the specific surface area and porosity of carbon materials. In general, as the heat treatment temperature increases and the time prolongs, the sp³ hybridization of the carbon material decreases, while the sp² hybridization increases. It results in a reduction in the specific surface area of the carbon material and a decrease in the porosity.

During cycling, electrons are transported to the surface of the material, leading to the reduction of the electrolyte components, and the decomposition products form a SEI layer on the surface ^[52][89]. This SEI layer should be able to bar the electrons and prevent the electrolyte from being continuously decomposed. Therefore, materials with larger specific surface areas should be passivated by a larger area of SEI in the first cycle, resulting in greater irreversible capacity loss and lower ICE.

By increasing the heat treatment temperature of the carbon material, the specific surface area, porosity and heteroatom content can be reduced, resulting in a flatter surface and thus reducing the generation of SEI and increasing ICE ^[53][54][55][90,91,92]. Firstly, increasing the heat treatment temperature benefits the increasing crystallinity of the carbon material, thus reducing the specific surface area. According to Sun et al. ^[56][93], the specific surface area of the obtained HC reaches a maximum at the pyrolysis temperature between 700 °C and 800 °C. As the pyrolysis temperature increases, the specific surface area decreases continuously, which is due to the more ordered structure of hard carbon formed by high temperature pyrolysis. Secondly, heat treatment at high temperature was able to reduce the amount of surface functional groups and avoid the massive decomposition of electrolyte. Jiang et al. ^[57][94] reported an ICE of 58.9% for hard carbon obtained by heat treatment at 800 °C, while the ICE of hard carbon obtained at 600 °C was only 46.5%, which was due to the reduction in the number of heteroatoms in the hard carbon structure caused by high temperature. Finally, reducing the porosity by heat treatment will also reduce the specific surface area. Alvin et al. ^[47][84] obtained hard carbon by a two-step carbonization approach, specifically by first pyrolyzing at 800 °C and then carbonizing at higher temperatures (1100 °C, 1300 °C and 1500 °C, respectively). The hard carbon obtained by heat treatment at 1300 °C in the second step has a lower specific surface area (23.7 m² g⁻¹) and smaller porosity (0.03 cm³ g⁻¹), thus obtaining a higher ICE (67%). In contrast, the hard carbon that underwent only one step of pyrolysis had a large specific surface area (35.5 m² g⁻¹) and a low ICE of 48%. Therefore, by reducing the specific surface area and porosity with heat treatment, the generation of SEI can be effectively controlled, thus avoiding large irreversible capacity loss.

The carbon material has a long enough time to adjust the internal structure, surface morphology and impurity composition in the case of a slow heating rate and long treatment time. Xiao et al. ^[54][91] used four different heating rates (0.5 °C/min, 1 °C/min, 2 °C/min and 5 °C/min, named, respectively, as HC-0.5, HC-1, HC-2 and HC-5) to pyrolyze sucrose precursors at 1300 °C. The slow heating rate is favorable to the hard carbon crystallinity as shown by the smaller d₀₀₂, larger L_a, L_c and increased I_G/I_D, and more sp² hybridization carbon. Nitrogen adsorption–desorption measurements and Brunauer–Emmett–Teller (BET) surface areas showed that HC-1 had the smallest pore volume and pore size; in addition, HC-0.5 could not even be measured accurately due to the gradual closing of the micropores. The slower heating rate provides sufficient time for gas escape, microporous closure and carbon atom movement and reorganization, which promote the growth of hexagonal carbon rings and contributes to the formation of defect-less graphite flakes with extended base plane size. Superior structural features greatly reduce irreversible capacity loss, permitting the HC-0.5 to achieve the highest ICE of 86.1% and maintain stable after 100 cycles at a current density of 20 mA g⁻¹ in SIBs. This also facilitates the formation of SEI with lower impedance. The resistance of SEI (R_SEI) was 98.96 Ω and only 10.09 Ω under fast heating and slow heating rate, respectively. In a study reported by Han et al. ^[58][95], mangrove wood was heat-treated at 500 °C at high pressure (0.7 MPa) for a long period (39 days). The ICE of hard carbon obtained by this method was as high as 80% in LIBs. It was attributed to the long-time high temperature treatment that increases the structural order of the material, decreases the pore size (mainly distributed between 0.3–0.55 nm) and reduces impurity elements such as H and O (the atomic ratios of H/C and O/C are 4.88% and 3.52%, respectively).

Pores with different diameters play different roles in the electrochemical processes. Micropores in carbon materials can act as active sites for ion insertion, which can adsorb metal ions and form metal clusters, thus increasing the reversible specific capacity of carbon materials ^[59][60][96,97]. At the same time, these pores can mitigate the volume changes generated by the repeated insertion and de-insertion of ions. Nevertheless, the high activity of micropores induces ion trapping effects, aggravating the electrolyte decomposition to form SEI, which causes high irreversible capacity loss. However, when the pore size is too small and the sizes of electrolyte molecules or ions are relatively large, the electrolyte interface will form a closed protection at the opening instead of inside the pore ^[61][98]. In this situation, the irreversible capacity loss during the first cycle are greatly reduced, while the closed protection retains the functionality of the pore itself as an active storage site. In contrast, materials with large pore sizes are unable to form closed protection SEI at the opening end, resulting in a decrease in reversible capacity from 789 mAh g⁻¹ to 175 mAh g⁻¹ after the first cycle. Therefore, pore size is a crucial issue that must be considered at present to improve ICE and SEI stability. Mesopores and macropores can shorten ion transport distance ^[62][63][99,100], which contributes to the rate performance of the carbon material. Therefore, constructing a hierarchical porous structure in which the surface layer is rich in macropores or mesopores and the inner layer is rich in micropores, is a very promising approach for carbon materials. The large pores in the surface layer facilitate the diffusion of electrolyte and accelerate ion transport, which can improve the rate performance of the material. The internal micropores act as active sites for sodium storage and increase the capacity of the material ^[42][43][79,80].