Free-Standing Carbon Materials for Lithium Metal Batteries

Graphical Abstract

Lithium metal batteries are promising next-generation high-energy-density anode materials, but their rapid capacity degradation is a significant limitation for commercialization. This review introduces strategies to stabilize lithium metal plating/stripping behavior and maximize energy density by using free-standing carbon materials as hosts and current collectors. Considerations for engineering the chemical properties of carbon and designing three-dimensional structures are discussed in detail.

Abstract

Lithium metal, with its high theoretical capacity and low redox potential, is the most promising next-generation high-energy-density battery anode material. However, the formation of uneven surface layers and dead lithium, significant volume changes in the electrode, and dendrite growth lead to rapid capacity degradation, low cycling stability, and safety issues, limiting the commercialization of lithium metal batteries (LMBs). As a strategy to improve the stability of LMBs, introducinga three-dimensional (3D) structure with a large surface area can accommodate lithium (Li) inside the structure and homogenize local current density. Also, as a current collector and host material, free-standing carbon materials, with the advantages of lightness, low cost, electrochemical and mechanical stability, and excellent electronic conductivity, can effectively enhance energy density and cycle performance. In this review, we first discuss the chemical properties of carbon, and then summarize recent research progress related to the 3D structuring and chemical modification of carbon materials as a Li metal host. Finally, we present perspectives on future research for the practical application of free-standing carbon materials for LMBs.

1 Introduction

Secondary batteries, which can be recharged and used continuously, have found applications ranging from small devices to various fields of energy storage systems. Recently, the market for Electric Vehicles (EVs) and their most crucial component, rechargeable secondary batteries, has experienced explosive growth due to the repercussions of climate policies. In line with this trend, there is also a significant increase in interest in lithium-ion batteries (LIBs), which dominate the secondary battery market.^1-4 Alongside the market growth, the demand for high energy density and rapidly rechargeable batteries is also increasing. However, commercial graphite anodes fail to meet the performance requirements for the next generation due to their low theoretical capacity (372 mAh g⁻¹) and poor rate performance.^{5, 6}

As an alternative to the graphite anode, a lithium metal battery (LMB) using lithium (Li) metal with high theoretical capacity (3860 mAh g⁻¹) and low electrochemical potential (standard hydrogen electrode, SHE vs. −3.04 V) as an anode material is an attractive anode system for high energy density batteries (Figure 1A).^{7, 8} Furthermore, Li metal anodes are considered essential materials for maximizing energy density in next-generation energy storage systems such as Li-sulfur and Li-air batteries.^9-12 The LMBs operate differently from conventional LIB systems, providing capacity through the plating/stripping of Li metal,¹³ in contrast to materials such as graphite, silicon which respectively provide capacity through intercalation/deintercalation¹⁴ and alloying/dealloying^{15, 16} reactions. However, several phenomena arising from these characteristics become crucial failure factors for LMBs.

Figure 1

Schematic illustration of (a) Li metal battery, (b) anode-free Li metal battery, (c) Li metal battery using free-standing carbon materials.

(1) The uneven solid electrolyte interphase (SEI) layer and the rough Li surface induce uneven charge distribution across the entire surface, leading to non-uniform nucleation and growth processes.¹⁷ Furthermore, charge density accumulates around areas such as tips on the rough Li metal surface, resulting in the concentration of Li. This can lead to the preferential nucleation and growth of Li, promoting the formation of dendrites.¹⁸ When sharp dendrites continue to grow, they may penetrate the separator, leading to internal short-circuiting upon reaching the cathode and causing safety issues.

(2) Li metal can undergo infinite volume changes during cycling processes. The stress resulting from these volume changes can lead to the destruction of the original anode structure, generation of cracks in the SEI, and acceleration of electrolyte depletion as the exposed Li metal surface reacts with the electrolyte.^19-22

(3) During the stripping process of Li, inactive dead Li is formed, which is mechanically separated from the current collector and cannot participate in further reactions due to the broken electrical contact. The dead Li accumulated on the surface during the cycle slows the Li⁺ transport dynamics together with the thickened SEI layer due to repeated formation. The limited diffusion of Li⁺ increases the non-uniformity of Li electrodeposition, consequently increasing the possibility of dendrite growth. Moreover, the significant increase in interfacial resistance caused by the dead Li layer leads to a considerable decrease in Coulombic efficiency (CE), rate performance, and cycle life, resulting in battery failure.^23-25

To address this issue, numerous studies have been conducted mainly to modify the interface of Li metal. Various approaches, including electrolyte salt modifications,^{26, 27} electrolyte additives,^28-30 and the introduction of artificial layers,^31-33 have enabled fast Li movement while forming a robust electrolyte-anode interface. While these approaches have had a beneficial effect on enhancing the cycle stability of LMBs, they have not yet provided a fundamental solution for achieving stable operation of LMBs under high current density and high capacity conditions, as the improvements are still limited to the surface. In addition, in order to provide high energy density at the full cell level, control of the negative to positive capacity (N/P) ratio is also a key factor.^{34, 35} Although the thickness of Li metal foil can be controlled through the calendaring process, and the amount of Li in the anode can be adjusted, the utilization of Li metal foil still faces limitations from a commercial standpoint due to the complexity and increased cost associated with the production processes required for thin Li metal foils.^{36, 37}

Therefore, an alternative system to the Li metal foil anode was needed, and the concept of anode-free LMBs has been proposed, gaining significant attention in recent times. In anode-free LMBs, the anode part does not contain active materials such as graphite, silicon, or Li metal. Instead, only a copper foil, serving as the current collector, is present in the anode part (Figure 1B). Li metal directly plating/striped on the Cu surface in anode-free LMBs, providing capacity without the presence of active materials. With only a thin current collector, this design enables the maximization of energy density and simplification of the electrode manufacturing process.^{38, 39} However, simply plating Li metal on the bare Cu foil can lead to uneven Li metal nucleation reactions on the surface due to the lattice mismatch between Cu and Li.⁴⁰ The non-uniform Li nucleation step and the limitation of surface area provided by the two-dimensional (2D) plane of the Cu foil result in the concentration of local current density. The increase in the local current density shortens the sand‘s time, defined as the time when the cation concentration reaches zero, and from this time, a local spatial charge is generated on the electrode surface, resulting in the growth of Li dendrite. The mentioned Sand's time can be expressed as formula 1.^{41, 42}

$$

Where τ represents the sand‘s time, D is the ambipolar diffusion coefficient, C₀ is the initial Li-ion concentration, e is the charge of an electron, J is the effective current density, and μ_a, μ_cdenote the mobility of anions and cations, respectively.

According to the sand's time theory, extending the sand‘s time by reducing the local current density is essential for stable Li plating. Introducing current collector structures with a large surface area in a three-dimensional (3D) configuration is thus a highly effective solution for the success of LMBs. Another advantage of a 3D host structure is its ability to alleviate the stress caused by the volume changes of LMBs. Without a host structure, Li metal, which undergoes a volume expansion of 4.85 μm per 1 mAh cm⁻²,⁴³ would experience significant battery degradation. Therefore, it is essential to design a structure capable of accommodating Li without internal electrode expansion. However, despite the introduction of 3D metal structures such as copper and nickel foam,^44-46 mesh copper,^{47, 48} the challenging control of thickness makes it difficult to regulate volumetric capacity. Furthermore, the high intrinsic density of Cu restricts the specific energy density of the battery.⁴⁹ When considering the overall energy density, reducing the weight of the metal current collector can lead to a significant increase in the specific energy density of batteries. Table 1 shows properties of Cu foil, Cu mesh, nickel foam, and recent research results on lightweight current collectors for LMBs (AgCuGF: Ag/Cu-coated glass-fiber fabric,⁵⁰ CMMC: Cu metal micromesh decorated with Cu_xO nanowires⁵¹).

Table 1. Summary of the properties of reported metal current collectors for LMBs.

Materials	Areal density (mg cm^−2)	Thickness (μm)	ref
Cu foil	7.5	8	[50]
Cu mesh	19.71	62	[51]
Ni foam	33.00	100	[51]
AgCuGF	3.1	20	[50]
CMMC	2.375	6.03	[51]

The lithiophobic nature of commonly used Cu materials also limits the uniform plating of Li. To overcome these limitations, there has been significant attention on the introduction of lightweight, highly electrically conductive, chemically stable carbon materials as hosts onto Cu foil.^52-55 Especially, carbon materials with a lower intrinsic density (2.26 g cm⁻³) compared to Cu (8.96 g cm⁻³) and Ni (8.90 g cm⁻³) can effectively reduce the weight of the current collector. However, when 3D structure carbon particles are simply coated onto a Cu foil using a slurry casting method with a binder and used as a host, there may be incomplete contact between all particles and the current collector, leading to possible peeling of the host material from the current collector.⁵⁶

On the other hand, when utilizing free-standing 3D carbon materials composed of networks of carbon as hosts, it can overcome the drawbacks mentioned above with the binder-based slurry casting method for carbon hosts and metal current collectors. This material can simultaneously serve as a host for Li metal and as a current collector in the battery. Moreover, having only lightweight carbon on the anode side makes it an ideal approach for achieving high energy density batteries (Figure 1C). But materials utilizing only carbon as a host still face challenges, as the intrinsic low lithophilicity of carbon results in a high nucleation overpotential and low wettability towards Li.⁵⁷

Therefore, in this review, we have provided an exploration of how carbon materials can be made free-standing. Alongside this, we have summarized existing progress related to how the chemical properties of carbon influence the behavior of LMBs and how these two characteristics can be combined for the advancement of LMBs. Finally, we have presented a comprehensive perspective for future developments (Figure 2).

Figure 2

Schematic diagram summarizing this review. Chemical properties, research progress on free-standing carbon materials categorized by building structure, and electrode design considerations for application in LMBs.

2 Chemical Properties of Carbons

As mentioned above, carbon materials have gained significant attention as strong candidates for Li metal plating. However, bare graphitic carbon materials exhibit lithiophobic characteristics due to the nonpolar structure of sp² carbon and the large lattice mismatch between graphene and Li metal.^{58, 59} These characteristics make carbon can only have weak interactions with Li, causing significant energy barriers for nucleation and a high possibility of uneven nucleation and dendrites formation on the surface. For this reason, Li and carbon form a large contact angle between them, making it difficult for Li to uniformly growth on the surface. It also restricts the application of pre-lithium infusion processes such as molten Li infusion. According to Young's equation, the contact angle of a metal nucleus is determined by the interfacial energies between Metal-Substrate (γ_MS), Metal-Electrolyte (γ_ME) and Substrate-Electrolyte Electrolyte (γ_SE), and it can be expressed as follows.⁶⁰

$$

That is, since contact angle can be reduced by reducing the interfacial energy (γ_MS) between the carbon host material and Li metal by modifying the characteristics of carbon. Therefore, to utilize carbon substrate as an LMBs host, it is considered necessary to adjust the surface characteristics of carbon substrate in a form favorable to Li plating by introducing lithophilic seed particles or adjusting the chemical properties of the carbon itself.

When assessing the suitability of a substrate material as a Li host material for LMBs, lithiophilicity, and surface diffusion barriers are mentioned as the most representative factors from a chemistry perspective. Lithiophilicity significantly contributes to the nucleation step, and surface diffusion barriers are closely associated with the subsequent deposition behavior of Li. In the typical charging process, under the electric field, solvated Li⁺ approach the SEI layer on the anode side. Subsequently, solvated Li⁺ undergoes desolvation, and only the Li⁺ pass through the SEI layer to reach the electrode surface. Afterward, Li⁺ is reduced to Li atoms on the electrode surface, aggregated through surface diffusion until reaching a stable critical size, forming nuclei, and then grow further. When reviewing the entire process of Li plating, the Li nucleation step depends on the surface chemistry of the structure. In other words, the binding energy between Li and the substrate, indicative of the interaction strength at the atomic level, serves as a crucial indicator associated with the lithiophilicity of the substrate.

$$

E_bind, E_total, E_Li, E_substrate represent the binding energy, total energy of the adsorbed Li atom and the bound substrate, energy of the Li atom, and substrate bound energy, respectively.⁶¹ Substrates capable of strong interactions with Li atoms are known to reduce Li nucleation overpotential, thus substrates with moderately high negative binding energy are preferred. Graphitic carbon typically exhibits low binding energy, so this means is that for the successful application of carbon as a host in LMBs, modification of carbon properties is required.

However, during the synthesis and adjustment of surface chemistry in carbon materials, defects are almost inevitably generated in carbon. The defects in carbon can be divided into intrinsic defects and extrinsic defects. Intrinsic defects include topology, vacancy, and edge defects, while extrinsic defects include heteroatom doping and defects arising from their binding sites.⁶² Defects in carbon typically enable stronger interactions with other reactants and act as electrochemical active sites. Thus, carbon defect engineering is of interest in various energy storage fields.⁶³ When applied to LMB, defects in carbon not only form sites with charge and spin density, enhancing the interaction capability withLi and reducing nucleation energy barriers but also lower the diffusion barriers of Li⁺ and Li atoms. However, excessive defect concentrations can disrupt the C−C sp² conjugated structure, impairing its electrochemical activity.⁶⁴ Therefore, precise defect control based on a comprehensive and deep understanding of defect chemistry is essential for the rational design of carbon materials for LMBs.

2.1 Intrinsic Defect Sites in Carbons

The topological defects refer to the distortion of the carbon lattice into one or more unconventional polygonal forms or the manifestation of dangling bonds, accompanied by the redistribution of asymmetric electrons. Vacancy defects are characterized by the absence of carbon atoms in the hexagonal carbon lattice, while edge defects result from the loss of carbon atoms in a fixed orientation, leading to high spin and high charge density.⁶⁴ Intrinsic carbon defects inevitably occur, but carbon materials can have different characteristics depending on the intrinsic carbon defect structure. The surface chemistry of the defect structure is an element that must be understood and utilized because it is directly related to the ability to interact with Li metal.

X. X. Ma et al. investigated the chemical properties of carbon materials according to various defect types based on the 2D graphene model.⁶⁵ As shown in Figures 3A–B, seven types of defects were considered, specifically Stone–Wales (SW), inversion Stone-Wales (ISW), single vacancies (SV), three double vacancies (DV1, DV2, and DV3), and quadra vacancies (QV) defects. The low formation energy of defects indicates a higher possibility of their formation (Figure 3C). Carbon atoms at the typical C6 site form three σ bonds with nearby neighboring carbon atoms and delocalized π bonds. The saturated coordination structure of these carbons results in a relatively weak electrostatic interaction between the Li atom and the C6 ring. On the other hand, the carbon atom of the perfect site can have coordination-unsaturated states or single electrons or electron pairs, which can lead to strong bonds with Li atoms. All investigated types of defects exhibit negative binding energies compared to pristine graphene (PG), indicating the potential to enhance lithiophilicity (Figure 3D). In particular, the SV defect shows the strongest interaction with Li due to increased local polarity resulting from the dangling bond. However, at high current densities, kinetic factors also play a crucial role in Li deposition behavior, so it is important to consider both binding energy and the diffusion characteristics of Li in the carbon scaffold. When the localized kinetic of Li atoms on the surface exceeds the surface diffusion kinetic, there is a tendency for Li to aggregate on the surface, leading to dendrite growth.⁶⁶ In contrast, when the Li diffusion barrier is low, initially deposited Li atoms can be uniformly plated across the entire carbon plane. The diffusion barrier is also associated with the critical current density. Below the critical current density, the surface diffusion rate is higher than the deposition rate, allowing for uniform Li plating. However, beyond this current density, Li prefers rapid vertical growth. Therefore, they presented a diagram considering both binding energy and the diffusion barrier derived critical current density (Figure 3E). SW, DV2, and QV defects, with higher binding energy with Li compared to PG, can reduce nucleation overpotential while ensuring a uniform deposition at high current densities due to their low diffusion barrier. Hence, these types of defects are preferred for Li metal plating. Meanwhile, SV defects exhibit the strongest interaction with Li, but adsorbed Li atoms can become trapped only in vacancies, and a low critical current density can lead to serious dendrite growth from a kinetic perspective.

Figure 3

(A) Schematic of the deposition of Li ions on PG and graphene with various defects. (B) Modeling of graphene with defects: SW, ISW, SV, DV1, DV2, DV3, and QV. (C) The defect formation energy of graphene and (D) binding energy between a Li atom and defective graphene. (E) The diagram according to the binding energy and the critical current density. Reproduced from⁶⁵ Copyright (2021), with the permission from Wiley-VCH. (F) Schematic illustrations of the atomistic structure of the p−CP and d−CP surfaces. (G) Plots of the Fermi levels of p−CP and d−CP during lithiation, (H) Schematic image of the SEI microstructure formed on p−CP and d−CP. (I) Configurations of Li atom clusters absorbed on perfect BLG and Li−MV (Li₃C₈).The lithiated Li and Li clusters, carbon in the upper layer, and carbon in the lower layer are shown in blue, purple, yellow, and gray colors, respectively. Reproduced from⁶⁷ Copyright (2022), with permission from Springer Nature.

One of the challenges when using carbon-based materials in battery systems is the increase in side reactions. The formed uneven SEI further promotes the uneven initial Li nucleation and dendrite growth in Li systems where surface chemistry plays a crucial role. To address this, H. Kwon et al. introduced multivacancy (MV) defect sites on the surface of carbon paper to adjust the Fermi level of carbon as shown in Figure 3F, forming a thin and uniform SEI and inducing uniform Li nucleation.⁶⁷ Experimental results show that pristine carbon paper (p−CP) forms an irregular mosaic SEI with randomly dispersed organic and inorganic particles. In contrast, defective carbon layer-coated carbon paper (d−CP) exhibits improved CE and a uniform, thin SEI. They interpreted this phenomenon through the Fermi level. The defective structure exhibits a lower Fermi level of the electrode surface than that of p−CP in all lithiation state regions (Figure 3G). The reduced Fermi level on the Carbon surface due to MV defects inhibits electron transfer to the lowest unoccupied molecular orbital (LUMO) of the electrolyte, preferring to accept electrons from Li at electron-deficient MV defect sites, thereby suppressing side reactions. Furthermore, according to quantum mechanical tunneling theory, decreasing the Fermi level increases the tunneling barrier. This allows electrical insulation to be reached at lower SEI thicknesses, mitigating electrolyte decomposition and restricting the growth of the SEI. To investigate nucleation behavior on the surface, Li atom cluster arrays on bilayer graphene (BLG) and MV defects were calculated via density functional theory (DFT). The results indicate that Li grows vertically on the p−CP surface. On the other hand, Li nucleation on d−CP involves the continuous adsorption of five Li atoms due to strong orbital hybridization between Li and MV defects, leading to their growth along the defect side. Therefore, the combination of lithiated MV defects with lithiophilic properties, along with a thin and dense SEI providing a uniform Li⁺ flux, promotes homogeneous nucleation (Figure 3H). During additional deposition processes, Li nuclei undergo lateral growth and merging with nearby nuclei, allowing for flat Li deposition on the carbon substrate (Figure 3I).

2.2 Non-Netal Heteroatom Doped Carbons

Heteroatom doping involves substitution of some carbon atoms with heteroatoms, which integrate into the sp²-centered network or defect sites of carbon, serving as an easy and effective method to modify carbon properties. The formed heteroatom-carbon bonds can modulate charge distribution and electronic properties, altering localized chemical reactivity. Depending on the characteristics of the dopant, carbon exhibits different properties such as lithiophilicity, wettability, and electronic conductivity. Introducing appropriate dopants can effectively enhance electrochemical activity.

X. Chen et al. summarized the calculated binding energy between various heteroatom-doped carbons and Li atoms (Figures 4A–B). Then, demonstrated that the lithiophilicity of doping sites is correlated with the electronegativity, local dipole, and charge transfer with Li of the dopant. Subsequently, they provided comprehensive guidelines for the design of lithiophilic doped carbon.⁶⁸ The relatively electronegative N (3.04) and O (3.44) compared to carbon (2.55) can strongly interact with the electropositive Li atom (0.98) through Lewis acid-base interactions. Doping with B (2.04), which has a lower electronegativity than carbon, can render adjacent C electronegative, leading to enhanced Li adsorption. Additionally, considering the actual effective electronegativity is crucial as the interaction capabilities between the dopant and Li may vary depending on the chemical state of the doped site. For example, The N, O, and F atoms all form σ bonds with adjacent carbon and withdraw electrons through inductive effects. Among them, N and O atoms can contribute additionally to the negative charge state due to their participation in the delocalized π system. On the other hand, the F atom forms p-π conjugation with the carbon plane, feeding electrons back to C from its filled p orbital. As a result, despite having the highest electronegativity, the F site exhibits low effective electronegativity, leading to low Li adsorption energy. Thus, the limitation of comparing different dopants solely based on electronegativity led this research group to consider additional factors. They additionally considered the local dipole defined as dipoles formed by doped atoms and adjacent carbon atoms. As a result, an implicit linear correlation between binding energy and factors associated with the sum of electronegativity and local dipole has been proposed.

Figure 4

(A) Modeling of heteroatom-doped carbons and prinstine graphene nanoribbons (GNR) model. (B) Summary of calculated binding energy between heteroatom-doped carbon and a Li atom. Reproduced from⁶⁸ Copyright (2019), with permission from American Association for the Advancement of Science. (C) Schematic diagram of N-doped graphene with pnN, prN, qN and qnN. (D) Binding energy of a Li atom with Cu, graphene and different functional groups of N-doped graphene. (E)The deformation charge density at a Li atom adsorption site of graphene and pyrrolic-N group. Reproduced from⁷² Copyright (2017), with the permission from Wiley-VCH. (F) Schematic representation of the Li plating process on Cu, C@Cu, and Fe_SA−N−C@Cu electrodes (η : nucleus overpotential of Li deposition). Reproduced from⁷⁷ Copyright (2019), with permission from American Chemical Society. (G) Binding energy of Li atoms on different substrates including graphene, NG, SAMn@NG, SANi@NG, SACo@NG, SAZn@NG, SACo@NG, and SAZr@NG. Reproduced from⁷⁹ Copyright (2022), with the permission from Wiley-VCH. (H) Schematic representation of the deposition and growth of Li with three types of nitrogen doping in graphite defects. Reproduced from⁸¹ Copyright (2023), with permission from American Chemical Society. (I) Schematic image for Li deposition behavior on substrates without and with metallic catalysts. (J) Energy-band plots of Co−CNF and CNF (E_Vac: vacuum level; E_f : Fermi level; E_VBM: valence-band maximum and W_f: work function). (K) Li migration pathways and barriers on bare graphite which denoted G, G+Co₁₃, and G+CoN₄. Reproduced from⁸³ Copyright (2023), with the permission from Wiley-VCH.

Among various dopants, especially N (0.74 Å) atom, with its similar atomic radius to C (0.77 Å) atom and high electronegativity, can be easily doped into carbon materials, effectively modifying their electrochemical properties. These features have garnered significant attention for N-doped carbon in various energy fields.^69-71 R. Zhang et al. investigated the influence of different types of N doping.⁷² Depending on the N doping sites within the graphene structure, it can be divided into pyridinic nitrogen (pnN), pyrrolic nitrogen (prN), quaternary nitrogen on the edge (qN), and quaternary nitrogen in the bulk phase (qnN) as displayed in Figure 4C. The additional electrons from the pyrrolic nitrogen group can provide electron pairs of p orbitals to conjugated π bonds, and pyridinic N contains an isolated electron pair. These N sites can act as Lewis bases, providing sites that strongly adsorb Li ions, which act as Lewis acids (Figures 4D–E). In contrast, qN and qnN cannot provide additional electron pairs to interact with Li atoms, resulting in weak binding energies. In addition, by effectively controlling the effective co-doping design as well as single atom doping considering electronegativity and local dipole, it can be induced to have electrochemical activity that is more advantageous to Li plating than single doping. Combination co-doping such as O−B/S/P,⁶⁸ N/S,⁷³ and N/P⁷⁴ can show superior performance when applied to LMB than single doping.

2.3 Single Atom Metal Doped Carbons

The introduction of metal atoms into carbon materials can significantly enhance the electrochemical activity of the material. Among them, Single Atom Catalysts (SACs) utilizing metal atoms as catalysts provide highly dispersed individual active sites fixed on a support. By doping a single atom metal into the carbon matrix, agglomeration of metal species can be prevented and atom utilization efficiency can be maximized. Recently, there has been an increasing effort to utilize materials where metal atoms are anchored in a Metal−N−C configuration through strong interactions with coordinating atoms possessing lone pairs of electrons, such as N, as hosts for Li metal.^{75, 76}

Adding the SA catalyst improves lithiophilicity, reduces Li metal nucleation overpotential, and enables uniform Li deposition. Figure 4F schematically illustrates the Li plating process onto the Cu, carbon, and single-atom iron in an N-doped carbon matrix (Fe_SA−N−C) based current collector. Y. Sun et al. introduced single atom Fe into an N-doped carbon matrix, utilizing Fe_SA−N−C as a lithiophilic seed. They performed MD simulations and experimental tests to demonstrate the enhanced lithiophilicity achieved through the introduction of single atoms.⁷⁷ Furthermore, T. Liu et al. explored the priority of Li deposition behavior on the carbon surface induced by N and Cu_xN doping using in-situ Raman spectroscopy.⁷⁸ As the Li plating time increased, both the intensity of the D-band and G-band gradually decreased. However, the intensity of the D-band decreased more rapidly, confirming that Li metal preferentially deposited on the defect sites in real-time.

While discussions are ongoing about various types of metals as single atom dopants, Z. Yang et al. reported six kinds of nonnoble metal atoms (Mn, Ni, Co, Zn, Cu, Zr) anchored on the nitrogen-doped 3D graphene (SAM@NG, M=metal).⁷⁹ Also, SA metal doping elements with appropriate binding energy to Li atoms were explored theoretically and experimentally (Figure 4G). Having too weak adsorption interactions for Li, such as graphene, leads to uncontrollable Li dendrite growth due to non-uniform nucleation. SAZr@NG exhibits the highest negative binding energy among candidates, which has a somewhat positive impact on the initial nucleation and deposition process. However, it induces too strong adsorption with Li atoms and side reactions between Li and nitrogen, resulting in structural damage to the Zr−N_x−C configuration and leading to a low cycle life. SAMn@NG, which has a binding energy value between graphene and SAZr@NG, experimentally demonstrated the most superior cycle performance and structural stability. SAMn@NG shows little change of bond length and bond angle between the metal atom and surrounding coordinating N atoms in the processes before and after Li deposition, functioning as a reversible site with minimal chemical structural changes during Li plating/stripping. This contributed to more stable cycling. In conclusion, it is essential to design SA-doped host materials with appropriate binding energy for Li, and a sophisticated design considering various factors such as binding energy with multiple Li atoms and chemical stability is necessary.

The potential of CoN_x doped carbon as a li metal host was suggested by H. Liu through its excellent li nucleation and deposition behavior,⁸⁰ and H. Qutaish et al. discovered that the SA Co−N₄ heteroatom induces axial direction Li growth and explored the principle in depth (Figure 4H).⁸¹ The large negative Li adsorption formation energy of pyridinic N and pyrrolic N is advantageous for nucleation, but excessive adsorption of Li atoms occurs at limited defect sites, causing Li to aggregate and undergo vertical growth. Meanwhile, Co−N₄ bonds in carbon delocalize electrons into adjacent carbon atoms due to the introduction of Co and improve the electronegativity of the surrounding carbon. Co SA doping enables charge transfer over a broader area within the carbon matrix and also improves the charge storage capacity of N atoms from Li and C atoms.⁸² As a result, it can induce axial direction Li growth by playing an active role in reacting not only with single Li atoms but also with Li clusters.

Y.-S. Feng et al. prepared a self-supported carbon fiber membrane with anchored single-atom cobalt, coupled with Co QDs (Co−CNF) and presented the effects of CoN₄ SA and Co particle doping on the interface from a different perspective (Figure 4I).⁸³ The low work function (difference between vacuum level and Fermi level) of CNF promotes electron transfer to Co, improving Li⁺ adsorption ability at the Co site and can induce electron pumping by controlling the valence electron level of Co−CNF (Figure 4J). Additionally, graphene with Co nanoparticles or CoN₄ sites shows a very low diffusion barrier when Li atoms move on the surface as demonstrated in Figure 4K. Therefore, it is explained that the Co site not only promotes Li nucleation but also significantly contributes to preventing localized accumulation of Li metal and improving uniform Li deposition by enhancing surface diffusion kinetics.

In this section, we discussed the reaction characteristics and effects related to various defects and dopants. Controlling the intrinsic defect sites of carbon is an effective approach. However, forming and precisely distributing specific defects that are favorable for Li plating/stripping still requires further research. Doping with heteroatoms is also an effective strategy that can significantly enhance the affinity with Li. However, for practical success in LMBs, it is essential to be able to control the doping elements and amounts. Furthermore, it is important to understand the interaction of Li at the bulk level, rather than solely at the atomic scale, to ensure that doping can induce uniform reactions even in high-current density and high-capacity deposition systems. Moreover, screening for effective dopants that not only reduce the nucleation barrier of Li metal but also improve surface diffusion kinetics is essential for achieving uniform Li deposition during rapid charging. Lastly, while discussions on the characteristics of carbon itself have been extensive, there is a lack of exploration into how such doping influences the structuring and architecture of carbon. Therefore, future designs should also consider the synergistic effects of carbon doping with carbon building blocks and the final structure.

3 Structure of Free-Standing Carbon Materials (Classification According to Structure of Building Block)

Various carbon materials such as carbon nanotubes (CNTs), graphene, and carbon fibers have been utilized to produce free-standing carbon materials for applications in the field of energy storage. In this section, we categorize the conducted research into building block structures of 1D, 2D, and 3D in the fabrication process of free-standing films. Additionally, we aim to explore approaches for modifying the properties of carbon and film formation to apply these building structures as hosts for LMBs.

3.1 1_Dimensional Carbons

3.1.1 CNTs (Carbon Nanotube)

CNTs have a high aspect ratio (length to diameter ratio) and possess a one-dimensional (1D) cylindrical structure, similar to rolling up a graphene sheet.^{84, 85} CNTs are lightweight and have good mechanical and chemical stability, as well as a high surface area and excellent electrical conductivity, making them suitable materials for use as current collectors. Furthermore, the entangled network structure of CNT bundles is easily assembled into free-standing films with a high surface area. This makes it straightforward to manufacture CNT films using various methods, rendering them promising materials in various energy storage applications.^{86, 87} Recently, efforts have been made to leverage the advantages of these CNT substrates for use as Li metal plating hosts. This includes adjusting the carbon surface chemistry or forming composites with other lithiophilic particles to overcome the inherent lithiophobic limitations of CNTs. In addition, efforts are being made to address the issue of the high irreversible reaction in CNTs by developing a prelithiation process for a 3D structure.

Y. Zhang coated a 4 nm Al₂O₃ layer on the CNT sponge using an atomic layer deposition (ALD) technique.⁸⁸ The conformal Al₂O₃ layer could effectively protect the surface of CNTs from deposited Li with excellent chemical stability and high mechanical strength. However, there are still limitations in long-term cycling stability, leading to the necessity of introducing lithiophilic seed for further improvement. Therefore, T. Zhu et al. intertwined single-walled carbon nanotubes (SWCNTs) with lithiophilic hollow cobalt-contain nanoparticles (hollow Co₃[Co(CN)₆]₂, h−CoCoPBAs), and formed a flexible free-standing film using a vacuum filtration method.⁸⁹ These hollow lithiophilic particles induce preferential growth of Li on the inner surface, thereby mitigating the volume expansion of the electrode (Figure 5A). In a similar approach, X. Zhang mixed Cu(OH)₂ nanowires with zeolitic imidazolate framework-67 (ZIF-67) precursors, then dispersed them with CNTs and obtained a film through vacuum filtration. Subsequently, through carbonization and phosphorization processes, a film structure consisting of carbon nanoboxes entangled with CNTs and richly containing Cu₃P/CoP nanobubbles was obtained (Figure 5B).⁹⁰ The obtained 3D hierarchical porous structure can effectively trap Li, and the conductive Cu₃P/CoP can strongly chemically bond with Li atoms. In the first lithiation process, it forms a Li₃P layer with high ion conductivity, thereby homogenizing Li⁺ flux and effectively reducing local current density.

Figure 5

(A) Schematic illustration and SEM images of the Li nucleation and plating process on h−CoCoPBAs@SWCNT . Reproduced from⁸⁹ Copyright (2022), with permission from Elsevier. (B) Scheme of the preparation process of LiǀCu₃P/CoP@C/CNT. Reproduced from⁹⁰ Copyright (2022), with permission from Elsevier. (C) Schematic illustrating the multiple roles of 3D-structured CNT current collectors in LMBs, in comparison with bare Cu current collectors. (D) Scheme and SEM images of surface of the bare Cu, DWNT, and MWNT surfaces after Li deposition processes. Reproduced from⁹¹ Copyright (2023), with permission from Elsevier. (E) Schematic image of the molten Li infusion process using a CSMF/Li/CSMF sandwiched structure and (F) corresponding optical image. (G) SEM images of pristine CSMF LiCSMF, (H) Gibbs free energy difference between CSMFI/CSMFII and molten Li. (I) Schematic diffusion process of heat and molten Li. (J) Galvanostatic cycling of a symmetric LiCSMF electrode (orange) and bare Li foil (blue) at current density of 40 mA cm⁻² with a stripping/plating capacity of 2 mAh cm⁻². Reproduced from⁹² Copyright (2020), with the permission from Wiley-VCH.

J. Kim et al. evaluated the electrochemical properties of double-walled nanotubes (DWNTs) and multi-walled nanotubes (MWNTs) scaffolds as 3D current collectors for LMB.⁹¹ The main difference between DWNT and MWNT is the number of walls serving as insertion sites. After dispersing CNTs in the solution, a 3D interconnected CNT scaffold was obtained through vacuum filtration technique. This was then attempted as a free-standing current collector for Li metal plating. While the pore distribution of DWNT and MWNT films was similar, the DWNT film had a higher BET surface area. Therefore, it was expected that DWNT would exhibit a higher capacity in a LIB system with a discharge cut-off voltage set to 0.01 V. However, the experimental results showed that MWNT provided a higher capacity. To interpret this, the analysis of lithium-ion storage behavior revealed that MWNT had a greater contribution of Faradaic (intercalative) reactions than non-Faradaic (capacitive) reactions compared to DWNT. These results suggest that, in terms of Li storage capacity, the number of graphitic layers has a more significant impact than the specific surface area. The walls of MWNT also enable reversible storage/release of Li-ions and promote more uniform and stable Li deposition compared to DWNT. When applying CNTs as Li metal hosts, some of the Li-ions are stored between the walls at the initial stage of Li deposition, and the interface of the lithiated CNT scaffold can promote uniform Li deposition. It was found that MWNT current collectors, which can reversibly store more Li-ions in the graphitic layer, are more effective than DWNTs in preventing aggregation of Li on the surface and delaying dendrite formation (Figures 5C–D).

Z. W. Wang collected CNTs with an automatically rotating cylinder to obtain a CNT substrate and compressed it into a rod to produce an expandable CNT sponge microfilm (CSMF) with a thickness of ~10 μm and applied it as a Li host.⁹² However, these CNT films could not be wet with molten Li due to low lithophilicity, and this could be addressed by modifying the host surface. Nevertheless, it was determined that byproducts generated during this process could adversely affect the electrode. Therefore, a new method for injecting molten Li was devised. When a Li foil was placed between two CSMF layers and heated the bottom with a plate at 190 °C, in such a stacked structure, a significant temperature gradient occurred layer by layer. Due to the relatively lower temperature on the surface of the upper CMSF film, molten Li in the middle layer preferred to thermodynamically diffuse into the upper CSMF (Figures 5E–H). Moreover, the pressure gradient on the CSMF surface due to thermal expansion acts as a driving force, allowing molten Li to penetrate upward against its own gravity. As schematically illustrated in Figure 5I, Li diffused upward initially, forming nuclei along the CNT, and continued heat transfer from the core structure allowed lateral growth along the CNT network, resulting in obtaining a uniform Li-coated CSMF (LiCSMF). The symmetric cell manufactured using LiCSMF demonstrates stable cycling performance for 2000 cycles, even when operated at a high current density of 40 mA cm⁻², with capacity of 2 mAh cm⁻² (Figure 5J).

3.1.2 CNFs (Carbon Nanofibers)

Unlike CNTs, where graphene layers are wrapped around to form a perfect cylindrical structure, carbon nanofibers (CNFs) have a nanostructure with various stacked arrangements of graphene sheets, resulting in a carbon nanostructure composed of discontinuous filaments with a higher aspect ratio and the possibility to form more edge sites.^{93, 94} CNFs have the ability to form a 3D network structure, and interconnected fibers can provide excellent electron/ion pathways and space for Li metal plating. In addition, CNF films are recognized as promising free-standing hosts for LMA due to their excellent mechanical properties and flexibility. The electrospinning method is commonly adopted for synthesizing CNFs due to its advantages of mass production capability and easy controllability in terms of structure and composition.⁹⁵

J. Lang et al. used polyacrylonitrile (PAN) as a precursor to manufacture fiber films through the electrospinning method, followed by heat treatment at various temperatures to produce carbon nanofibers.⁹⁶ At low temperatures (650 °C), the amorphous carbon formed on the surface has weak interactions with molten Li, resulting in poor wettability with molten Li. On the other hand, carbon nanofibers obtained through high temperature treatment (1200 °C) (HT−CNFs) demonstrate the intercalation of Li into the ~2 nm-thick graphitic carbon layer formed on the surface when in contact with molten Li. This leads to the creation of a Li affinity layer, enhancing Li affinity and wettability, and enabling the molten Li infusion process.

Additionally, several research groups have dispersed materials on the surface of CNFs that can serve as nucleation sites for Li deposition. L. Luo et al. distributed Mo₂N particles in the CNF skeleton and applied them to LMB.⁹⁷ Li₃N and Mo are generated through a conversion reaction of Li and Mo₂N in the lithiation process, and Li₃N can improve the ion conductivity in the interphase. Also, due to the strong interaction between Mo and Li, the Li−Mo bonding is chemically favored, allowing Mo to serve as a seed for Li nucleation (Figures 6A–C). The synergistic effect of Mo and Li₃N derived from Mo₂N, which guides uniform Li deposition, and the CNF scaffold that homogenizes current density flux contributes to excellent cycle stability, allowing the symmetrical cell to operate stably for over 1500 h at 6 mA cm⁻² and 6 mAh cm⁻² (Figure 6D).

Figure 6

(A) Comparison illustration of the Li nucleation and plating process on Cu, CNF and Mo₂N@CNF. (B) XPS spectra of the lithiated Mo₂N@CNF substrate after 5 min of Ar-sputtering. (C) Relaxed state of a Li monolayer on a semi-bulk Mo slab and a graphite bilayer. (D) Long-term cycling performance of symmetric cells with different electrodes at: 3 mA cm⁻² and 6 mA cm⁻². Reproduced from⁹⁷ Copyright (2019), with the permission from Wiley-VCH. (E) Schematic diagram of the process of preparing CNF/Sn. (F) Scanning electron microscopy (SEM) images of CNF, CNF/Sn, CNF/Fe and CNF/Co. (G) The adsorption configuration of Li on CMF/Sn and adsorption energies of a Li atom with CNF/Sn, CNF/Fe, CNF/Co, Li, and CNF calculated by the DFT. Reproduced from⁹⁸ Copyright (2022), with the permission from Wiley-VCH. (H) Schematic synthesis, SEM and tansmission electron microscope (TEM) image of the bifunctional lithiophilic hierarchical Co−N−CNT−CF. (I) The adsorption energy of Li on Co@G clusters and graphene doped with different N functional groups. (J) Schematic and SEM images (top-view) of the Co−N−CNT−CF during plating at 0.5 mA cm⁻² with 0.5 mAh cm⁻² to 8 mAh cm⁻² and stripping to 8 mAh cm⁻². Reproduced from¹⁰⁰ Copyright (2023), with permission from Elsevier.

H. Zhuang added Sn, Fe, or Co precursors to the electrospinning solution along with polyacrylonitrile (PAN) to obtain nanofibers, and then subjected them to high-temperature heat treatment to carbonize organic polymers and reduce metals (Figures 6E–F).⁹⁸ The lithiophilic spherical metal particles were uniformly dispersed on the fiber surface and maintained structural stability without detaching from the fiber surface even during repeated cycling processes. The calculated adsorption energies of Li atoms with Sn, Fe, and Co were −2.81, −2.68, and −3.04 eV, respectively, which are stronger than the cohesion energy of Li (−1.90 eV) as shown in Figure 6G. This suggests that during the plating process, rather than forming bulk Li, nucleation of Li is induced preferentially on the CNF/metal surface due to the stronger adsorption, followed by additional growth along the fiber surface walls. Conversely, CNF without metal seeds was calculated to have weaker adsorption energy than the cohesion energy of Li, indicating that Li-ions may not adhere well to the fiber surface and are expected to grow between the fiber gaps as particles. The analysis of ex-situ images after electrochemical Li plating revealed images that matched the predicted results. When evaluating the cycling behavior at 1 mA cm⁻² and 1 mAh cm⁻² using a symmetric cell, CNF/Sn−Li showed no change in overpotential even after 2350 hours, while bare CNF−Li experienced a rapid degradation of the electrode after 400 hours.

W. Zeng et al. fabricated nanofibers by electrospinning a solution containing PAN and ZIF-8 nanoparticles, and by carbonization, they obtained N-doped 3D CNF embedded with Zn-carbon nanocages.⁹⁹ The N-doped carbon and Zn nanoparticles served as lithiophilic sites for Li metal nucleation inside the CNF, and the internal nanocage provided sufficient space for Li deposition, effectively reducing volume change during cycling. In addition, the large surface area of CNF alleviated the local current density, extended sand time, and suppressed dendrite growth, further contributing to uniform Li plating.

T. Lyu forms Co−MOF (metal-organic-framework) on the surface through a reaction between a Co-containing free-standing fiber substrate and an organic ligand.¹⁰⁰ During the carbonization process, dicyanamide decomposes and serves to supply carbon and nitrogen in the MOF. High-density N−CNTs were formed under the resulting Co nano-catalysts, resulting in a hierarchical structure in which Co−N−CNTs grew on Co−CNFs (Figures 6H–I). Due to the strong adsorption energy of Co-Graphene, Li-ions are preferentially deposited on the Co@G-rich fiber surface first, and then deposited and grown on the Co−N−CNT layer with relatively low adsorption energy. When a higher amount of Li is loaded afterward, it completely covers the 3D voids between Co−CNT−CF, resulting in a smooth surface (Figure 6J). This hierarchical structure enables reversible Li plating/stripping and exhibits improved cycling stability compared to a single structure where Co is simply distributed on the surface. Table 2 shows the summary of electrochemical performance of 1D materials-based free-standing carbon materials for LMBs (LM : loading mass of substrate, CD@C : Current density@Capacity, C-r@Cy@Cr : C-rate@Cycle@Capacity retention),

Table 2. A summary of electrochemical performance of 1D material-based free-standing carbon materials for LMBs.

Materials	AD (mg cm^−2)	Electrolyte (type/μL)	Asymmetric Half cell		Symmetric cell			Full cell		ref
			CD@C	Cycle	Pre-lithium infusion method	CD@C	Cycle (hour)	Cathode materials (mg cm^−2)	C-r@Cy@Cr
Co3[Co(CN)6]2@CNT		1 M LiTFSI in DOL/DME (1 : 1, v/v)+2 wt% LiNO3	1@1	200	Electrodeposition	2@1	500	SuperP/S (1.1)	0.5 C@100@87	[89]
Cu3P/CoP@C/CNT		1 M LiTFSI in DOL/DME (1 : 1, v/v)+0.2 M LiNO3/40	1@1	100	Electrodeposition	1@1	400	LFP (3.5)	1 C@[email protected]	[90]
MWNTs	0.346	1 M LiPF6 in EC/DEC (1 : 1, v/v)	1@1	100				LCO	0.5 C@50@97	[91]
CSMF	0.5	1 M LiTFSI in DOL/DME (1 : 1, v/v)+1 wt% LiNO3			Molten Li Infusion	40@2	2000	Li2S6-CSMF (1.28)	1 C@200@≈60	[92]
Mo2N@CNF		1 M LiTFSI in DOL/DME (1 : 1, v/v)+2 wt% LiNO3	4@3	150	Electrodeposition	6@6	1500	NCM811 (8~10)	C/3@150@90	[97]
CNF/Sn	1	1 M LiTFSI in DOL/DME (1 : 1, v/v)+2 wt% LiNO3	1@1	270	Electrodeposition	2@1	1000	NCM811 (6)	1 C@[email protected]	[98]
NCNFs−Zn−CCs		1 M LiTFSI in DOL/DME (1 : 1, v/v)+1 wt% LiNO3/60	1@3	110	Electrodeposition	3@3	3200	LFP (13)	1 C@[email protected]	[99]
Co−N-CNT−CF		1 M LiTFSI in DOL/DME (1 : 1, v/v)+1 wt% LiNO3/70	0.5@1	200	Molten Li (Sym-cell), Electrodeposition (Full-cell)	3@1	1000	LFP (18.6)	0.5@[email protected]	[100]

Free-standing substrates manufactured based on high aspect ratio 1D carbon materials exhibit a very high surface-to-volume ratio, enabling the effective reduction of local current density. However, the substantial irreversible reactions derived from CNTs can lead to a decrease in CE, highlighting the need to devise approaches that can enhance the initial CE. Therefore, controlling the porosity, surface area, and interfacial properties of carbon at the materials-to-electrode level, along with the introduction of effective prelithiation processes, is a critical challenge for reducing the irreversibility of electrodes using 1D-based materials.

3.2 2_Dimensional Carbons (Graphene)

Graphene is a material characterized by a hexagonal 2D planar structure formed by chemically sp² hybridized carbon atoms. It has gained attention in various energy fields due to its advantages such as high thermal/electrical conductivity, mechanical properties, and chemical stability.^{101, 102} Graphene oxide (GO) is functionalized with oxygen, imparting a negative charge and enabling the formation of stable colloidal suspensions. This makes it advantageous in modification or processing, allowing for easier composite with other materials and fabricating graphene films. Additionally, GO possesses lithiophilic properties that allow it to absorb Li, and reduced graphene oxide (rGO), with stronger binding energy with Li compared to pure graphene due to its carboxyl and alkoxy groups, exhibits enhanced surface lithiophilicity compared to bare graphene.¹⁰³ For this reason, when applying a graphene-based material as a host for LMBs, a method of fabricating a mechanically stable and flexible film with functionalized graphene and then injecting Li into the structure is drawing attention.

Chen et al. attempted to maximize the energy density of LMB by developing an ultra-thin structure based on graphene.¹⁰⁴ First, a graphene oxide (GO) solution was blade-coated onto a substrate using a doctor blade to obtain a GO film, and through the ultrafast self-expansion and reduction (USER) reaction, the GO film was transformed into a porous, expanded, and reduced graphene oxide film (eGF). The obtained eGF was calendered through a roller to adjust the thickness to 0.3–20 μm. Afterwards, molten Li was injected between the internal channels of eGF using capillary action to design a thin Li electrode (Li@eGF) with a 3D host structure. The lithiophilic 3D rGO structure inhibits inhomogeneity of local current density, provides high mechanical strength, and the low weight ratio and high conductivity of rGO play an excellent role as a current collector. As a result, the cycle stability and energy density of the battery using Li@eGF as the anode were improved. In a pouch cell manufactured with lithium iron phosphate (LiFePO₄, LFP) as the cathode, it maintained 81 % of its capacity after 200 cycles, providing a nine-time higher cycle life than when using pure Li metal as the anode.

However, when using a simple GO film as a Li host, Li metal tends to deposit only on the upper surface of the electrode at high rates due to the long and highly tortuous migration paths, limiting the utilization of the internal space. Therefore, Ni et al. fabricated a pin template using 3D printing and penetrated GO films manufactured by continuous casting or spraying. After removing the template, a structured GO film with regularly arranged macropores can be obtained.¹⁰⁵ Such porous structures vertically aligned within the GO scaffold electrode provide a more uniform Li⁺ flux and shorter diffusion paths, offering a stable scaffold for Li plating/stripping. The enhanced internal diffusion of Li also improves its utilization efficiency (Figures 7A–D).

Figure 7

(A) Fabrication diagram and photos of the GO dense film. (B) Schematic of the holey GO/Li electrode fabrication process. (C) Optical images of the 3D printed pin template surface, SEM image of the holes obtained in the GO film and the photo of flexible Li/holey GO composite. (D) Schematic of the Li plating and stripping behavior on a pure GO matrix and a holey GO matrix during cycling. Reproduced from¹⁰⁵ Copyright (2022), with the permission from Wiley-VCH. (E) Schematic of Li plating and stripping behavior in SNGO/Li anode and GO/Li anode. (F) Simulation results of electric field distribution and Li deposition behavior of SNGO/Li anode and GO/Li anode at the initial state, after charging for 1 h, and subsequently discharging for 1 h at a current density of 1 mAh cm⁻² and Li⁺ concentration and Li deposition behavior after 25 cycles. (G) Cycling performance of SNGO/Li||SPAN pouch cell at 0.2 C with mass loading of 2.5 mg cm⁻². (H) Photo of a phone charged by the SNGO/Li||SPAN pouch cells at their 180° folding state. Reproduced from¹⁰⁶ Copyright (2022), with the permission from Wiley-VCH. (I) Fabrication and characterizations of the Li/graphene-C/Li₃N composite electrode. (J) Cross-sectional SEM images and the corresponding energy dispersive X-ray (EDX) mapping images of a Li metal electrode and a Li/graphene-C/Li₃N electrode after 50 stripping/plating cycles at 1 mAh cm⁻² and 1 mA cm⁻². Reproduced from¹⁰⁷ Copyright (2022), with permission from Elsevier.

Still, structures focused solely on mechanical design led to the formation of a low-conductivity and unstable SEI (solid-electrolyte interphase) composed of Li₂CO₃ and electrolyte-derived species during the cycling process. This can ultimately result in the formation of dead Li or Li dendrites on the surface over extended cycles. Thus, in subsequent studies, Ni et al. aimed to manufacture a free-standing GO host for more stable Li metal plating/stripping, with additional consideration of surface properties.¹⁰⁶ This was achieved by providing a large Li accommodating space through uniform microchannels, ensuring rapid Li⁺ transfer paths, and simultaneously controlling the composition of SEI for enhanced stability. Using thiourea as a precursor, nitrogen and sulfur were doped into GO with microchannels (SNGO). Then, during the molten Li injection process, Li spontaneously diffused through the GO layer and microchannels. Simultaneously, it reacted with nitrogen and sulfur atoms, resulting in the uniform formation of Li₃N and Li₂S on the surface of the GO film. Li₃N and Li₂S, with their high ionic conductivity, are recognized as preferred SEI components for metal plating (Figure 7E). These compounds homogenize Li⁺ flux and enable fast Li transport on the electrode surface. Also, these inorganic-rich SEI layer contributes to a conformal and stable interface, facilitating more uniform and reversible Li deposition/stripping behavior. Through COMSOL simulation, it was confirmed that SNGO uniformly distributes electric field and Li⁺ concentration within the structure (Figure 7F), and in a full-cell design, it can effectively alleviate internal stress caused by volume expansion. In consequence, SNGO/Li exhibited stable performance for 450 hours under conditions of 1 mA cm⁻² and 1 mAh cm⁻² in a symmetric cell, and also demonstrated stable cycling performance in a full-cell configuration paired with LiNi_0.8Co_0.1Mn_0.1O₂ (NCM 811). The flexible characteristics of the electrode were demonstrated by manufacturing Li−S pouch cells using sulfur/polyacrylonitrile (SPAN) as the cathode material. In Figure 7G, it was confirmed that a capacity retention of over 80 % was achieved after 100 cycles with a sulfur mass loading of 2.5 mg cm⁻². Even when the cell was bent by 180°, it can steadily power a mobile phone, proving its ability to reliably supply power even under external stress (Figure 7H).

Wan et al. also aimed to enhance the mechanical stability of the substrate and improve the interphase stability by providing a layered structure framework that can suppress the volume changes of Li metal while controlling the composition of the SEI.¹⁰⁷ In this study, GO and nitrogen-rich g−C₃N₄ were dispersed together in a solution and a dense film was obtained through vacuum filtration. Then, during the process of prelithiation through molten Li infusion, the C₃N₄ residing on the top reacted with Li, resulting in the formation of Li₃N on the surface (12Li+C₃N₄→4Li₃N+3C) (Figure 7I). The preformed Li₃N layer inhibits excessive electrolyte decomposition, allowing the formation of a dense SEI layer and enabling stable Li plating/stripping on the surface. The preformed Li₃N layer inhibits excessive electrolyte decomposition, allowing the formation of a dense SEI layer and enabling stable Li plating/stripping on the surface (Figure 7J). Accordingly, Li/graphene−C/Li₃N exhibits stable electrochemical behavior even after 1000 cycles in a symmetric cell. When tested in a full-cell configuration with LCoO₂ (LCO) as the cathode, it provides a high energy density of 694.05 Wh kg⁻¹ and a high-capacity retention of 75.91 % after 150 cycles at 0.5 C. The electrochemical performances of 2D material-based free-standing carbon materials for LMBs, including 3D cellular Ag/graphene composite (Ag−N−rGO),¹⁰⁸ rGO-based scaffold formulated with N/Zn derived from ZIF-8,¹⁰⁹ bacterial cellulose-derived CNF@rGO nanosheet (BC−CNF@rGO),¹¹⁰ (3D) MXene/graphene (MG) framework film (MG film)¹¹¹ are summarized in Table 3.

Table 3. A summary of electrochemical performance of 2D material-based free-standing carbon materials for LMBs.

Materials	AD (mg cm^−2)	Electrolyte (type/μL)	Asymmetric Half cell		Symmetric cell			Full cell		ref
			CD@C	Cycle	Pre-lithium infusion method	CD@C	Cycle (hour)	Cathode materials (mg cm^−2)	C-r@Cy@Cr
Holey GO/Li		1 M LiPF6 in EC/DEC (1 : 1, v/v)+1 % VC, 10 % FEC	–	–	Molten Li infusion	[email protected]	400	LFP (2.0–2.4)	1 C@[email protected] (include rate test)	[105]
SNGO/Li		1 M LiPF6 in EC/DEC (1 : 1, v/v)+1 % VC, 10 % FEC	1@1	150	Molten Li infusion	1@1	450	SPAN (3.35)	0.2@60@88	[106]
C3N4/GO		4 M LiFSI in DMC/30			Molten Li infusion	3@1	20	LCO (10.9)	0.5@[email protected]	[107]
Ag−N−rGO		1 M LiTFSI in DOL/DME (1 : 1, v/v)+2 wt% LiNO3	1@1	300	Electrodeposition	1@1	1900	NCM811 (5.0)	0.5@300@83,5	[108]
ZIF-8 @RGO		1 M LiTFSI in DOL/DME (1 : 1, v/v)+1 wt% LiNO3/60	0.5@ 0.5	350	Electrodeposition	2@1	140	LFP (4.0)	0.5@[email protected]	[109]
BC-CNF@rGO	0.1	1 M LiTFSI in DOL/DME (1 : 1, v/v)+2 wt% LiNO3	1@1	800	Electrodeposition	2@1	5000	LFP (1.31)	1@3000@≈67	[110]
MG film	3.45	1 M LiTFSI in DOL/DME (1 : 1, v/v)+1 wt% LiNO3/40	1@2	90	Molten Li infusion	3@1	500	LTO (2.0	1@[email protected]	[111]

Utilizing 2D graphene-based materials, characterized by excellent mechanical and electrochemical properties, is a highly promising approach for fabricating thin films intended for high-energy-density batteries. However, the low Li⁺ diffusion capability in the vertical direction of 2D-based films still requires improvement. In addition, considering interfacial properties together is a key strategy for utilizing 2D material-based free-standing substrates as hosts for LMBs.

3.3 3_Dimensional Carbons

In this section, we will discuss commercially available 3D carbon fiber materials or 3D carbon precursor materials that can be directly converted into films. Commercial carbon cloth¹¹² and carbon paper¹¹³ are practical Li host candidates due to their flexibility, excellent mechanical strength, and the feasibility of large-scale production. These materials typically have thicker fibers in the micrometer range, resulting in a lower aspect ratio compared to the 1D carbon materials mentioned before. Due to the relatively low issues with material film formation, research is predominantly focused on modifying and utilizing substrates already formed in 3D.

3.3.1 Surface Structure Engineering

When a structure such as a porous or hollow polyhedron or nanosheet combined with a lithiophilic seed is grown in situ on a 3D carbon surface, Li plating can be spatially guided along the structure, improving the stability of the electrode material. In particular, hollow carbon structures can effectively guide Li into the inside of the structure, reducing the stress associated with electrode volume changes.

Z. Zhuang et al. grew hollow nanocages with anchored lithophilic seeds of N and ZnO on the surface of carbon cloth.¹¹⁴ The designed surface structure allows for rapid Li⁺ diffusion, efficient charge transport, and suppresses the volume change of Li during cycling resulting in an excellent cycling lifespan.

Growing structures derived from nanosheets vertically on the surface of carbon fibers can induce anisotropic Li plating. W. Zhang et al. covered the surface of carbon cloth with leaf-shaped ZnCo−ZIF, and after heat treatment, obtained a leaf-shaped structure with dense N-doped CNTs entangled under Co catalysis (CC@Co−NCNTs) (Figure 8A).¹¹⁵ To verify the effects of Co particles present in the CC@Co−NCNTs skeleton, the charge density distribution was simulated to investigate the Li-ion plating process on CNTs. On the surface of bare CNTs without Co particles, the current density is concentrated at the top part. On the contrary, the introduction of Co particles in Co−CNT results in a relatively uniform current density at the top, and the current density on the side walls also increases, providing higher adsorption sites for Li-ions (Figure 8B). This can induce more uniform plating on the wall surface, filling the gap between Co−CNTs. Furthermore, in the Co−NCNT model, the surface charge of Co clusters transfers to NCNT, attracting Li-ions preferentially to NCNT sites. In the following steps, Co clusters undergo homogeneous charge repulsion with subsequent Li-ions, resulting in Li plating on the bare carbon nanotubes. The tendency of deposition on the Li surface with a negative curvature between the aligned nano-sheet structure and the aforementioned effect of dynamically distributed charges can induce anisotropic Li plating in Co−NCNT nano-sheets, thereby suppressing the growth of Li dendrites (Figure 8C).

Figure 8

(A) Illustration of the fabrication process of CC@Co−NCNTs@Li composite anode. (B) Simulation results of the current density distribution of CNTs and Co−CNTs. (C) Three-dimensional schematic of the plating process of Li⁺ on CC@Co−NCNTs. Reproduced from¹¹⁵ Copyright (2022), with permission from Elsevier. Schematic illustration of Li nucleation, plating/stripping behaviors on planar Cu foil, CC, and NiMn−LDHs NAs@CC electrodes. (E) Electrochemical Li plating-stripping curves of the NiMn−LDHs NAs@CC electrode at 0.5 mA cm⁻² for a capacity of 6 mA h cm⁻² and SEM images of NiMn−LDHs NAs@CC electrodes after being discharge to 0.01 V, plated with 4 mA h cm⁻², h, 6 mA h cm⁻², and charge to 0.5 V. Reproduced from¹¹⁷ Copyright (2022), with permission from American Chemical Society. (F) The pictures of the Co−MOF@CFC (purple) and CoP−C@CFC (black) with the excellent flexibility. (G) Binding energies of a Li atom with CFC and CoP−C@CFC at the top sites by DFT calculation. (H) Li plating/stripping process on CFC and CoP−C@CFC electrode. Reproduced from¹¹⁸ Copyright (2022), with the permission from Wiley-VCH. Schematic diagram of preparation route for CoSe₂−NC@CFC. (J) Voltage profile of Li plating/stripping process on CoSe₂−NC@CFC at 1 mA cm⁻² with capacity of 10 mAh cm⁻²and SEM images of CoSe₂−NC@CFC morphology evolution at corresponding state. Reproduced from¹¹⁹ Copyright (2022), with the permission from Wiley-VCH.

T. Xu et al. aligned graphene vertically on the carbon cloth surface and introduced lithiophilic Au NP particles onto the graphene surface, guiding stable Li plating in the structure.¹¹⁶ F.-Y. Tao et al. aligned NiMn-based layered double hydroxide with a unique lamellar structure on the surface of carbon cloth (honeycomb-like layered double hydroxide (LDH) nanosheet arrays supported on flexible carbon cloth, NiMn−LDHs NAs@CC).¹¹⁷ The spatial characteristics of vertically aligned NiMn-LDH NAs with elaborate honeycomb-like scaffold structures reduce local current density, control the direction of Li plating growth to mitigate the stress of volume changes, and improve Li⁺ diffusion properties. The Li symmetric cell utilizing NiMn−LDHs NAs@CC as the host exhibited stable behavior for over 1600 hours under conditions of 2 mA cm⁻² and 2 mAh cm⁻², demonstrating the potential of LDH materials as Li metal hosts (Figures 8D–E).

3.3.2 SEI Component Engineering

Another key to enhancing 3D Li hosts is to control structures with lithiophilic seeds while simultaneously managing the components of the SEI that inevitably occur during the Li plating process, thereby improving interfacial properties.

Therefore, H. Jiang et al. manufactured a structure in which Co−MOF nanoflakes were vertically grown on carbon cloth and then obtained carbon fiber cloth (CFC) modified with MOF-derived porous carbon sheath with embedded CoP nanoparticles (CoP−C@CFC) through the phosphating process.¹¹⁸ The optical image of CoP−C@CFC can be seen in Figure 8F. The strong binding energy between CoP and Li can improve the lithiophilicity of the carbon substrate. And the CoP nanoparticles undergo a stepwise conversion reaction during the lithiation process as follows: 1) CoP+xLi⁺+xe⁻↔Li_xCoP; 2) Li_xCoP+(3−x) Li⁺+(3−x) e⁻↔Li₃P+Co, transforming into Co nanoparticles surrounded by a Li₃P network. Li₃P and Co nanoparticles individually impart high ion conductivity and electron conductivity, resulting in the homogenization of Li⁺ flux and charge distribution at the interface (Figures 8G–H). Consequently, the Li₃P@Co ion/electron conductive interface formed in CoP−C@CFC can provide excellent lithiophilicity and Li⁺ diffusion kinetics, reducing nucleation overpotential and local current density, thereby inducing uniform Li plating without dendrite formation over extended cycling periods.

Using a similar method, J. Cao et al. grew Co−MOF nanoflakes on the surface of CFC and obtained a structure with distributed CoSe₂ nanoparticles through selenization (Figure 8I).¹¹⁹ CoSe₂ also exhibits a strong interaction with Li, which can improve lithiophilicity, and is converted into Co nanoparticles and Li₂Se through the reaction mechanism CoSe₂+4Li→Co+2Li₂Se during the initial lithiation process. As a component of the SEI, Li₂Se can enhance the surface stability of Li metal with its characteristics of high ion conductivity, mechanical resistance, and chemical stability. As a component of the SEI, Li₂Se can enhance the surface stability of Li metal with its characteristics of high ion conductivity, mechanical resistance, and chemical stability. In Figure 8J, the voltage profile and corresponding SEM images of the Li plating/stripping process of CoSe₂−NC@CFC, which can be subdivided into the stages of Li⁺ insertion, Li nucleation, Li plating, Li stripping, and Li⁺ extraction, are presented. A flat Li layer was shown when it reached 10 mAh cm⁻², and after stripping, Li was reversibly removed, showing an image similar to before plating, demonstrating excellent reversibility as a Li metal host.

S. Xia et al. obtained metal-free fluorinated carbon fibers (FCF) by doping carbon with F atoms through a scalable and low-cost process using PVDF as a fluorine source.¹²⁰ The as-prepared FCF, after molten Li infusion, could form a much more uniform and conformal LiF-dominant SEI layer compared to SEI derived from fluorinated electrolytes or additives. The excellent mechanical and charge transport properties of the in-situ formed conformal LiF layer minimized undesirable electrolyte depletion, inducing uniform Li nucleus formation and growth.

Y.-K. Lee mixed sodium carboxymethyl cellulose (Na−CMC) and polyacrylic acid (PAA), which are well dispersed in aqueous solvents and used as binder materials, with carbon fiber paper (CFP) in water solution.¹²¹ After heat treatment, the binder material was transformed into amorphous carbon and Na₂CO₃, resulting in the acquisition of carbon fiber paper (CFP) with amorphous carbon and Na₂CO₃ (ANCFP). The Na⁺ cations of Na₂CO₃ formed on the CF surface can preferentially attract and decompose TFSI⁻ anions during the initial lithiation process, leading to the formation of inorganic-rich SEI layer (Li₂CO₃, LiF) derived from TFSI⁻. These inorganic components reduce the interfacial resistance of the SEI and enhance mechanical strength, mitigating the growth of Li dendrites. And the lithiphilic properties of oxygen-containing amorphous carbon covering the CFP surface ensure uniform Li plating. Taking advantage of the lightweight properties of CFP, the anode mass of Li@ANCFP after Li plating is five times lower than that of Li@Cu, achieving a high energy density.

3.3.3 Electrode Design Engineering

The gradient or Janus structure design of 3D Li metal hosts is a powerful technique for inhibiting dendrite growth by inducing sequential Li plating from the desired regions within the electrode level.¹²² Even with the introduction of a 3D structure, there is a limitation as Li plating can be concentrated towards the top of the electrode, closer to the separator. Therefore, the integration of a 3D structure and a gradient (or Janus) design considering factors such as conductivity and lithiophilicity can maximize the stability of LMB.¹²³

J. Cai et al. fabricated a Janus-lithiophilic conductive textile host with a lihiophobic Ni side and a lithiophilic NiSb side.¹²⁴ For synthesis, a uniform Ni layer was deposited on carbon cloth by a polymer-assisted metal deposition method, and then one side of Ni-coated carbon cloth was converted into NiSb through thermal evaporation and heating process (Figures 9A–B). In the electrodeposition process, the tendency that Li nuclei are not formed on the upper Ni side but grow only on the bottom NiSb side was observed. When the Janus anode comes into contact with molten Li, Li rapidly and uniformly spreads on the NiSb side, depositing a uniform and thin Li layer, while dewetting occurs on the Ni side (Figure 9C). During repeated Li plating/stripping, the bottom-up growth mechanism of Li and the buffering space at the top significantly alleviated Li dendrite growth. The electrode volume remained constant, and this low volume change also stabilized the electrode interface. The symmetrical cell demonstrated stable cycling performance over 3000 cycles at a high rate of 10 mA cm⁻² and a high capacity of 10 mAh cm⁻² (75 % depth of discharge (DOD)). B. Hong et al. also designed a Janus lithiophilic structure to induce bottom-up Li growth. They utilized a straightforward process using magnetron sputtering to introduce lithiophilic Au nanoparticles onto the bottom surface of the carbon paper. As a result, Li could be deposited uniformly and continuously from the bottom of the carbon paper, enhancing the spatial utilization of the entire internal porosity of the 3D host.¹²⁵ Similarly, X. Yan et al. deposited Si onto the bottom side of a porous carbon paper using magnetron sputtering, resulting in a Gradient Si-modified carbon paper (GSCP) where the Si content gradually decreases from the bottom to the top. Si also exhibits lithiophilic properties, allowing Li to preferentially nucleate and grow from the bottom following this gradient distribution of Si.¹²⁶

Figure 9

(A) Schematic representation of the preparation of the Janus conductive textile host. (B) SEM image, elemental mapping, X-ray Diffraction (XRD) spectrum of the Janus conductive textile. (C) Schematic illustration, photograph, SEM image of the selective-wetting properties of the Janus Li-textile composite anode. Reproduced from¹²⁴ Copyright (2023), with the permission from Wiley-VCH. (D) Schematic diagram of the SiC/CC and Li@SiC/CC frameworks before and after Li metal deposition (E) SEM, energy dispersive spectroscopy (EDS)-mapping images of a SiC/CC and E-field distribution on the CC and the SiC/CC frameworks. (F) Cross-sectional SEM images of the CC and SiC/CC framework after the deposition of Li metal with the area capacities of 2, 4, 6 mAh cm⁻² at a current density of 1 mA cm⁻² and schematic illustration of Li metal deposition behavior. Reproduced from¹²⁷ Copyright (2022), with permission from Elsevier. (G) The fabrication process of the inverted Li metal anode based on carbon cloth (H) The overpotential of top-down and bottom-up deposition of Li metal. Reproduced from¹²⁹ Copyright (2022), with the permission from Wiley-VCH. (I) Schematic of the preparation process of Ni₂P@ISHCP. (J) TEM, EDX elemental mapping, SEM images of Ni@ISHCP (K) The binding energies and charge density of a Li atom adsorbed on ISHCP and Ni₂P@ISHC. (L) Li deposition mechanism on ISHCP and Ni₂P@ISHCP. Reproduced from¹³⁰ Copyright (2022), with permission from Elsevier.

The design of a gradient electrical field is also one of the effective strategies. B. Sun et al. obtained a 3D carbon host with gradient-decorated SiC whiskers in which the SiC whiskers gradually increased from the bottom to the top along the depth of the carbon cloth (Figures 9D–E).¹²⁷ SiC is semiconductive (<10⁻¹¹ S cm⁻¹), so a conductivity gradient structure can be formed based on the gradient of the SiC whisker due to the difference in conductivity compared to conductive carbon fibers (>10³ S cm⁻¹). As a result of the simulation, the formed conductivity gradient structure can homogenize the electrical field and Li⁺ flux (Figure 9E). The relatively low electrical conductivity of SiC whiskers at the upper layer hinders electron cohesion on the surface, leading Li metal plating from the bottom layer where the relatively high electrically conductive part. The plating proceeds in a bottom-up model, gradually filling the void spaces from the bottom to the top (Figure 9F). W. Cao et al. designed a Li₃N gradient structure by growing g-C₃N₄ on carbon cloth followed by a molten Li infusion process. The Li₃N gradient structure with high ion conductivity effectively stabilizes the interface. Furthermore, the synergistic effect of this Li₃N gradient structure and the 3D carbon architecture enables smooth Li deposition even at high current densities, ultimately providing excellent cycling performance. This result suggests that the gradient design of SEI could also be a promising strategy.¹²⁸

To address the drawbacks of complex synthesis processes in designing a gradient structure vertically, D. Li et al. proposed a simple concept of inverted structure Li anode to induce bottom-up growth of Li.¹²⁹ They simply electrochemically plated Li on top of carbon cloth (upright structure), then flipped it over and called it an inverted structure anode, and used it as an anode. At this time, the plating was performed at a high current density of 6 mA cm⁻² in order to plate only on one side (top area). (Figure 9G). In typical upright structures, a lower overpotential is induced at the top of the electrode where the diffusion path for Li⁺ is shorter, and it tends to preferentially deposit, leading to the growth of dendrites. In contrast, the inverted structure shows a relatively lower overpotential at the bottom due to the presence of Li at the bottom, thereby inducing Li plating from the bottom of the electrode (Figure 9H). In addition, in this case, the remaining carbon scaffold at the top can support the SEI layer and serve as a porous layer that regulates uniform Li⁺ diffusion. As a result, the inverted Li-deposited structure anode can demonstrate superior cycle stability in symmetric cells and full cell tests compared to the upright Li-deposited anode. In addition, to achieve higher energy density, thinner PET fabric was manufactured, and the successful application of the inverted anode structure concept was confirmed, expanding it to various porous architectures.

3.3.4 Others

Hollow carbon structures can effectively store Li metal internally and suppress volume change stress during plating/stripping. However, the absence of a lithiophilic seed can hinder the efficient utilization of the internal voids during plating. Additionally, forming a free-standing film by connecting hollow carbon structures in 3D was also a major issue. H. Jiang et al. used a gel mixed with starch and spherical SiO₂ particles as a template to synthesize interconnected stacked hollow carbon spheres (ISHCP) modified with uniformly dispersed Ni₂P nanoparticles (Ni₂P@ISHCP) (Figures 9I–J).¹³⁰ The formation of a 3D interconnected stacked structure of hollow carbon spheres, along with successful large-scale production of free-standing films through roll-to-roll processing, was achieved and applied to LMB hosts. The internally uniformly dispersed Ni₂P can induce nucleation at Ni₂P sites due to its strong affinity with Li (Figure 9K). During the lithiation process, the reaction Ni₂P+3Li⁺+3e⁻→Li₃P+2 nano Ni leads to the formation of a Li₃P layer on the surface, providing favorable Li⁺ diffusion kinetics. The interconnected, highly conductive hollow carbon framework, along with the uniformly dispersed lithiophilic Ni₂P seeds and the formation of a stable SEI layer, enhances the utilization of the internal space of substrate and improves the reversibility of Li plating/stripping reaction (Figure 9L).

In addition to commercial carbon cloth and carbon paper, other structures such as cotton textile,¹³¹ filter paper,¹³² and wood¹³³ can be carbonized to obtain interconnected films in three dimensions, which can also be used as hosts for Li metal. Therefore, the selection of appropriate materials with unique surface and structural characteristics among various cost-effective substrate candidates can be a promising approach for the commercialization of 3D carbon materials host for LMBs. Some representative works of 3D materials-based free-standing carbon materials for LMBs, including N, P dual-doped carbon derived from quantitative filter paper (NPCQP)¹³² and vertically aligned wood-derived carbon plate decorated with Co₄N nanoparticles host (Co₄N/WCP)¹³³ are listed in Table 4.

Table 4. A summary of electrochemical performance of 3D material-based free-standing carbon materials for LMBs.

Materials	AD (mg cm^−2)	Electrolyte (type/μL)	Asymmetric Half cell		Symmetric cell			Full cell		ref
			CD@C	Cycle	Pre-lithium infusion method	CD@C	Cycle (hour)	Cathode materials (mg cm^−2)	C-r@Cy@Cr
CC@Co−NCNTs	11.95	1 M LiPF6 in EC/DMC (1 : 1, v/v)+5 wt% FEC			Molten Li	5@5	1400	LFP (4.0)	1 C@[email protected]	[115]
NiMn−LDHs NAs@CC		1 M LiTFSI+0.5 M LiNO3 in DOL/DME (1 : 1, v/v)/100	4@2	70	Electrodeposition	2@2	1600	NCM111 (1.2–1.5)	1 C@200@77	[117]
CoP−C@CFC		1 M LiTFSI in DOL/DME (1 : 1, v/v)+2 wt% LiNO3	2@2	600	Electrodeposition	2@1	550	LFP (3.5–5.0)	1 C@[email protected]	[118]
ANCFP	1.0 ±0.2	1 M LiTFSI in DOL/DME (1 : 1, v/v)+2 wt% LiNO3/60	1@1	150	Molten Li (Sym-cell), Electrodeposition (Full-cell)	2@1	300	LFP (14.0)	1 C@300@85	[121]
Ni,NiSb Janus anode		1 M LiTFSI in DOL/DME (1 : 1, v/v)+2 wt% LiNO3/50			Molten Li infusion	10@10	3000	LCO (1.8 mAh cm^−2l)	1 C@600@85	[124]
SiC/CC		1 M LiTFSI in DOL/DME (1 : 1, v/v)+2 wt% LiNO3/60	2@1	100	Electrodeposition	2@3	1000	LFP (4.5)	0.5 C@120@80	[127]
Ni2P@ISHCP		1 M LiTFSI in DOL/DME (1 : 1, v/v)+2 wt% LiNO3	2@1	250	Electrodeposition	2@1	600	LFP (3.5)	1@[email protected]	[130]
NPCQP		1 M LiTFSI in DOL/DME (1 : 1, v/v)+1 wt% LiNO3	1@2	200	Electrodeposition	0.5@1	1000	LFP	0.2@[email protected]	[132]
Co4N/WCP		1 M LiPF6 in EC/DEC (1 : 1 v,v)			Electrodeposition	2@2	2000	S@Co4N /WCP (2)	1 C@500@≈85.5	133

Through the design of the surface structure of three-dimensional carbon architectures and the control of surface layers, specifically the SEI components, Li can be stably accommodated within the three-dimensional structure even at high current densities. However, the thick thickness of commercial free-standing carbon materials like carbon cloth remains a limiting factor for achieving high energy density. Therefore, it is critical to explore precursor materials for thin 3D carbon films that can lower costs while maintaining excellent mechanical properties.

In summary, this section discusses the methods for manufacturing free-standing carbon substrates and how the characteristics of carbon materials were modified for application in LMBs. In all 1D, 2D, and 3D carbon-based free-standing carbon material hosts, engineering both the carbon itself and the interfacial properties was a key strategy. Furthermore, strategies are needed to fully utilize the 3D structure and achieve bottom growth to enhance stability. The gradient structure of electronic and ion conductivity, lithiophilicity, and SEI components enables the direct guidance of nucleation and growth sites for Li. Ultimately, by integrating these factors, it will be possible to control the process of Li nucleation and growth more precisely.

4 Conclusion and Perspectives

LMBs are considered ideal battery systems for high energy density applications due to the high theoretical capacity of Li and its low negative electrochemical potential. However, despite these advantages, the uneven nucleation and growth of Li dendrites during the growth process, along with the formation of dead Li, pose safety concerns and lead to low cycling performance, becoming significant obstacles for practical applications. Among various strategies, utilizing free-standing carbon materials as a Li metal host is a promising option to address these issues. 1) The inherent excellent conductivity of carbon, along with a superior three-dimensional conductive network, provides fast electron conduction pathways, 2) and the lightweight characteristics of the porous carbon framework can maximize the specific energy density. 3) The large surface area provided by the aligned or interconnected three-dimensional structures can reduce local current density and provide a uniform Li⁺ flux, inducing uniform Li nucleation and growth, ultimately suppressing dendrite formation. 4) Lastly, the wide spaces within the structure alleviate the stress of volume changes caused by the growth of Li metal and can stably accommodate Li metal.

This review focuses on recent research on the application of free-standing carbon materials as hosts and current collectors for LMBs, and the recent progress can be summarized as follows:

(1) Manufacturing of free-standing carbon substrate: film formation using materials such as CNT, CNF, and graphene as building structures; direct utilization of commercial carbon cloth and carbon paper; carbonization of biomass or various 3D structural materials can be used to obtain free-standing carbon materials.

(2) Defect control of carbon: Engineering the chemical properties of carbon in a form favorable to Li plating, with intrinsic defect or heteroatom doping, including single atom metal is a crucial method to induce uniform plating at the surface.

(3) Therefore, there is a need for techniques that not only enhance lithiophilicity but also provide structural channels capable of effectively accommodating Li, while controlling the SEI layer. An approach involving introducing materials on the surface that can preferentially attract certain components favored in the SEI layer or that can control the composition of the SEI by undergoing a conversion reaction with Li is gaining attention. The introduction of a material that can conformally form an SEI layer with high ionic conductivity and mechanically robust composition on the carbon surface can effectively improve interfacial stability.

(4) Gradient or Janus structure electrode design: One limitation of three-dimensional structures is that deposition can be induced only at the top of the electrode, failing to effectively utilize the entire space. Designing frameworks at the electrode level with effective heterogeneous structures such as Janus or gradient can induce bottom growth of Li, enabling efficient utilization of internal space and enhancing both energy density and stability.

To design high energy density full-cell, it is crucial to consider compatibility with other parameters, and there are still unresolved challenges to practically apply free-standing carbon materials as the anode of LMBs. Several key issues to be addressed in the future and additional factors to consider when designing free-standing carbon materials are as follows.

(1) The use of high surface area and defects of carbon frameworks can lead to high irreversible reactions, resulting in decreased CE, which can negatively impact the cycle life and overall energy density. This could be a significant obstacle to commercialization. Therefore, an effective prelithiation process for carbon needs to be established. The method used in previous studies, manufacturing cells followed by electroplating and disassembly, results in a complex process and increased costs. The molten Li infusion method also has limitations in terms of quantification of infused Li quantity, scalability, and universality. Therefore, for considering the DOD and N/P ratio, it is essential to establish a sophisticated and scalable pre-lithium infusion method that can control the amount of Li.

(2) Carbon with excessively large surface area can undergo excessive irreversible reactions. To prevent additional side reactions, the SEI layer can be pre-formed during pre-processing steps such as Li infusion, or artificial layers can be introduced.

(3) Unlike copper-based substrates, an additional Li insertion reaction occurs in carbon materials, but there is a lack of understanding of the effects of this lithiation process. However, many studies utilizing carbon as an LMB host focused on only improving the lithiophilicity of carbon. Therefore, comprehensive carbon chemistry engineering is required, taking into account lithiophilicity, carbon surface diffusion kinetics, bulk Li⁺ diffusion, and interfacial layers, after understanding the nucleation and growth processes of Li on carbon materials through in-depth experimental and theoretical analysis.

(4) It is important to design a substrate that considers practical utilization and fabrication process. The produced film should be flexible, and the process should be simple, cost-effective, and capable of large-scale production.

(5) The uncontrolled thick thickness of the 3D framework can impede volumetric energy density, so a technology that can control the thickness of the substrate according to the required capacity is needed. At this point, methods to effectively utilize the space within the 3D structure should also be considered.

(6) Testing should be conducted in a practical environment, considering the high loading mass of commercial cathode material; in other words, test conditions that utilize the high areal capacity of the Li anode.

(7) Extensive analysis has been conducted on the electrochemical behavior of free-standing carbon materials as hosts for Li metal. However, many studies have been limited to laboratory-scale coin cell tests. To evaluate commercial viability, analysis at the pouch cell level is necessary. Differences between pouch cells and coin cells, such as cell components, and scale can lead to gaps in pressure and current density distribution, areal resistance of electrodes, electrolyte content, cell design, and other factors. Recently, there has been a recognition of this importance, and evaluations of electrochemical performance at the pouch cell level are also being provided. But still, most of these evaluations are limited to cycle test results. Therefore, conducting an in-depth analysis of operational principles in pouch cells and designing free-standing carbon materials suitable for pouch cells, including multi-stack cell configurations is essential.

In an ideal scenario where these requirements are met, free-standing carbon materials have the potential to maximize energy density and improve cycle performance, making the development of these materials highly promising. Therefore, in-depth research from multiple perspectives on free-standing carbon hosts is expected to offer significant promise in the realization of practical LMBs for next-generation energy storage devices.

Acknowledgments

This research was supported by ‘regional innovation mega project’ program through the Korea Innovation Foundation funded by Ministry of Science and ICT (Project Number: 2023-DD-UP-0032).

Conflict of Interests

The authors declare no conflict of interest.

Open Research

Data Availability Statement

Data sharing is not applicable to this article as no new data were generated or analyzed in this study.

Biographical Information

Hongjung Kim is an M.S. & Ph. D. integrated student in the Department of Chemical Engineering at Pohang University of Science and Technology (POSTECH). He performed his B. S. degree (2023) in the School of Chemical Engineering at Yeungnam University. The main themes of his research in NEDL group have been synthesis of Metal-Organic-Frameworks, carbon-based materials for the applications of these materials in Metal-ion and Li metal batteries.

Biographical Information

Yeonguk Son is an Assistant Professor at department of chemical engineering in Changwon National University. He received his B. S. and Ph. D. degree in School of Energy and Chemical Engineering at UNIST. He worked as a Research Associated in Department of Engineering at University of Cambridge (2018–2021). His research interests are (i) development of electrode materials for high energy storage, (ii) advanced analysis techniques for electrode materials, and (iii) electrode and cell designs for high energy lithium-ion batteries.

Biographical Information

Changshin Jo is Assistant Professor of the Graduate Institute of Ferrous & Eco Materials Technology and the Department of Chemical Engineering at Pohang University of Science and Technology (POSTECH). He performed his B. S. degree (2010) and Ph. D. degree (2016) on synthesis of electrode materials in the Department of Chemical Engineering (POSTECH). He joined the University of Cambridge as Research Associate and Marie Skłodowska-Curie Fellow (2017–2020) in Department of Engineering. He worked two years as Assistant Professor in Chemical Engineering at Chung-Ang University (2020–2022). The main themes of his research have been synthesis of functional inorganic materials for energy storage applications.