Recent developments in advanced anode materials for lithium-ion batteries

The rapid expansion of electric vehicles and mobile electronic devices is the main driver for the improvement of advanced high-performance lithium-ion batteries (LIBs). The electrochemical performance of LIBs depends on the specific capacity, rate performance and cycle stability of the electrode materials. In terms of the enhancement of LIB performance, the improvement of the anode material is significant compared with the cathode material. There are still some challenges in producing an industrial anode material that is superior to commercial graphite. Based on the different electrochemical reaction mechanisms of anode materials for LIBs during charge and discharge, the advantages/disadvantages and electrochemical reaction mechanisms of intercalation-, conversionand alloyingtype anode materials are summarized in detail here. The methods and strategies for improving the electrochemical performance of different types of anode materials are described in detail. Finally, challenges for the future development of LIBs are also considered. This review offers a meaningful reference for the construction and performance optimization of anode materials for LIBs.


INTRODUCTION
With the decline of oil and other traditional energy sources, the development and utilization of renewable energy sources, such as solar, wind and tidal power, have become critical problems to be solved in the new era [1][2][3] . However, these new energy supplies are unstable and cannot be used continuously, so they need to be converted into electricity before being exported [4] . Research into rechargeable batteries has therefore become especially important [5][6][7] . Traditional lead-acid, nickel-cadmium and nickel-metal hydride batteries have some disadvantages, such as short service life, low energy density and environmental pollution, which greatly limit their large-scale application [8][9][10][11] . The development of rechargeable batteries to replace these traditional batteries has been the main priority of the battery industry in recent decades [12][13][14] . Therefore, there is an urgent need to develop non-toxic and pollution-free battery electrode materials and separators.
Compared with traditional batteries, Li-ion batteries (LIBs) already dominate the development of electronic products and show superior development prospects due to their small size, lightweight, high working voltage, high energy density, good cycling performance, lack of memory effect and environmental friendliness [15][16][17] . LIBs have been widely used in smartphones, laptops and other portable electronic devices since the 20th century [18][19][20] . Currently, LIBs are used as the main power supply of electric vehicles but they still require higher energy density, lower costs and enhanced environmental performance [21][22][23] . The lithiumion shuttle between the anode and the cathode forms the basis of a LIB as the power supply, and the Li + insertion ability of the anode is the main factor that determines its performance [24][25][26][27] . In order to maintain the development of advanced anode materials, it is becoming ever more necessary to develop high-capacity anode materials to improve the performance of the overall LIB [28][29][30][31] .
Graphite has been the main LIB anode material since its commercialization by Sony in 1991 [32,33] . The reason for this is that graphite has numerous advantages, such as low cost, absolute abundance, non-toxicity and structural stability [19,[34][35][36] . However, the theoretical capacity of graphite is low, which limits its feasibility for high-power applications [37][38][39] . It is also found that the delithiation/lithiation in the cycle process cause a volume change in the material, which produces stress on the electrode and is not conducive to cycling stability [40][41][42] . In addition, the graphite electrode may react with the electrolyte at low operating voltages, resulting in lithium deposition [43][44][45] . This phenomenon not only reduces battery performance, but also poses a serious safety hazard. Therefore, the study of alternative anode materials to improve the electrochemical properties of LIBs, such as specific capacity, cycle stability and safety, is essential. After decades of continuous research, a variety of different types of anode materials have been discovered and improved. So far, according to the different reaction mechanisms, LIB anode materials can be divided into the following three types: intercalation (e.g., graphite [46,47] and lithium titanate [48,49] ), conversion (e.g., transition metal oxides [50,51] and sulfides [52] ) and alloying anode materials (e.g., silicon [53] , tin [54] , germanium [55] and antimony [56] ).
The earliest anode material considered for LIBs was lithium metal. However, lithium dendrites can be formed during the reaction process, thereby limiting its commercialization. In the review, recent advances in the different electrochemical reaction mechanisms of LIB anode materials are summarized, as summarized in Scheme 1. We provide a comprehensive report and in-depth discussion of the electrochemical reactions of anode materials, as well as the advantages and challenges of each type of anode material. The corresponding optimization strategies are also described in detail. This review represents an important reference for the construction and optimization of high-performance anode materials.

INTERCALATION-TYPE ANODE MATERIALS
The reaction mechanism for intercalation-type anode materials is based on the intercalation and deintercalation of lithium ions in the crystal lattice of the host material. Such LIBs are also known as rocking-chair batteries [57][58][59] . As a layered carbon material, graphite was the first commercialized LIB anode material and is also the most well-known [60,61] . Layered LiC 6 can be formed by intercalating lithium ions. The discharge plateau of LiC 6 is below 0.2 V (vs. Li + /Li) and it has outstanding dynamic performance for the intercalation of lithium [5,62,63] . However, because of the slow diffusion rate of lithium ions, the rate performance of graphite is not ideal. The intercalated lithium potential is similar to the stripping potential of lithium metal and therefore lithium dendrites and solid electrolyte interphase (SEI) films can easily form [64,65] . In addition, the miscibility of the solution with an electrolyte makes the organic solvent and lithium insert into the graphite layer, causing the graphite to peel off gradually. These problems are not favorable in practical applications. There are two main methods to improve the electrochemical performance of graphite anodes, namely, accelerating the diffusion of lithium ions into graphite and the entire electrode and enhancing the interfacial reaction between graphite and electrolyte to form a thinner and more stable SEI layer.
Cheng et al. [61] prepared a multichannel graphite anode with channels etched into the graphite surface, which enabled the rapid entry of lithium ions into graphite particles for the rapid charging of LIBs. This structure can improve the accessibility of these ions inside the graphite and has good coulombic efficiency. Their results showed that the multichannel graphite anode exhibits excellent charge rate capacities of 83% and 73% at 6 C and 10 C, respectively, which are much better than the pristine graphite anode. In addition, multichannel graphite anodes exhibit a higher enhanced discharge rate capability than pristine graphite. In addition, it has excellent cycling stability with a capacity retention rate of 85% after 3000 cycles at 6 C without any additives.
Son et al. [63] systematically studied the thermal exfoliation of expandable graphite (EG) in order to increase its interlayer spacing distance and to determine the optimal temperature for volume expansion. As shown in Figure 1A, all the EG samples exhibit a worm-like morphology with rich open/semi-open inner pore structures. In particular, at the heat treatment temperature of 600 °C for a period of 30 min (EG30), there is a maximum shrinkage of 3.37 Å of d-spacing and a crystallite size of 20.96 nm, while maintaining a similar long range-ordered graphite layer/sheet. In addition, EG30 shows excellent performance in LIBs, with a very high average reversible specific capacity of 338 mAh·g -1 at 100 m·Ag -1 and a high rate capacity of 112 mA·h g -1 at 3 A g -1 [ Figure 1B]. The excellent battery performance of EG30 is attributed to its unique turbostratically arranged graphitic nanolayers/nanosheets, the range of chemical environments resulting from the distribution of redox-active sites in the pores at the defect sites, the minimum nanocrystal size and the possible formation of LiC72 compound. This finding could provide an inspiration for the construction and development of high-performance graphite electrodes for practical applications in LIBs.
Surface modification is an effective method to improve the rapid charging ability of graphite anode materials. Kim et al. [65] improved the rapid charging ability of graphite anode materials by modifying Al 2 O 3 on the surface of graphite. As shown in Figure 1C, the 1 wt.% Al 2 O 3 @graphite electrode retains a reversible capacity of ~337.1 mAh·g -1 at a high charge rate of 4000 mA g -1 , which corresponds to 97.2% of the capacity obtained at 100 mA g -1 . The full battery test with a LiCoO 2 cathode and Al 2 O 3 -coated graphite anode proves that the introduction of amorphous Al 2 O 3 improves the fast charging ability of the graphite anode material. This method is a practical way of enhancing the fast charging ability of graphite anode materials for highpower LIBs.
Similar to graphite, hard carbon (non-graphitized carbon) materials also belong to the intercalation type of anode material. Their layer spacing is generally greater than 3.8 Å, more than twice the diameter of Li + (1.52 Å) [31,66] . This large channel is conducive to the diffusion of Li + and rapid charge and discharge. Therefore, it provides high capacity and good rate performance. However, although hard carbon has a higher voltage platform than graphite, it has voltage hysteresis and a high initial irreversible capacity, which is why graphite has not been replaced as the mainstream anode material [67][68][69] . It is noteworthy that the transport of Li + is mainly dependent on the structure of the material. It was realized that the presence of defects accelerates the diffusion of Li + and so the subsequent emergence of carbon materials with unique structures, such as graphene [70][71][72] , nanofibers [73][74][75] and nanotubes [76][77][78] , has occurred.
Ai et al. [72] prepared N and S co-doped graphene (NS-G). The initial discharge capacity of the obtained material is 1636 mAh·g -1 . After 500 cycles, the electrode still offers a reversible capacity of 1090 mAh·g -1 . As the number of cycles increases, the capacity of the NS-G anode increases gradually; this can be ascribed to the enhancement of the sample properties of lithium ions. The difference in electronegativity and size between the doped and carbon atoms results in changes in the internal structure (the formation of topological defects) and charge density of the graphene, thereby enhancing the electrochemical performance of the material.
Wang et al. [73] synthesized P-doped mesoporous carbon with a high P content and large pore size through evaporation-induced self-assembly. Tricresyl phosphate was used as a phosphorus precursor, a phenolic solution was used as a carbon precursor, the triblock copolymer F127 was used as a soft template and  [63] . (C) Rate capabilities of Al 2 O 3 @graphite electrode [65] . (D) Schematic illustration of the synthesis of P-doped mesoporous carbons [73] . Reproduced from Refs. [63,65,73] with permission from Elsevier. EG: Expandable graphite; EG30: the thermal exfoliation process of EG maintains 30 min; PU foam: polyurethane foam; EISA: evaporation induced self-assembly. polyurethane foam was used as a sacrificial template [ Figure 1D]. The obtained P-doped mesoporous carbon with a high P content (up to 1.90%) consists of small collaterals and interconnected nanoparticles (10-20 nm), showing large adjustable mesopore dimensions (6.6-14.2 nm) and high surface areas (338-630 m 2 g -1 ). When used as anodes for LIBs, these materials have excellent electrochemical storage performance. The P-doped mesoporous carbon shows a high reversible capacity of 500 mA h g -1 after 200 cycles at 0.5 C, outstanding rate performance and cycling stability after 100 cycles at 10 A g -1 .
Furthermore, lithium titanium oxide (Li 4 Ti 5 O 12 ) is also a typical intercalation-type anode material and exhibits high safety and excellent cycling stability [79,80] . At room temperature, Li + (1/6) and Ti 4+ (5/6) ions are randomly dispersed at the 16d site of the octahedron while the O atoms occupy all 32e sites. The structure can be represented by [Li 3 ] 8a [Ti 5 Li] 16d [O 12 ] 32e [81,82] . The redox mechanism of Li 4 Ti 5 O 12 consists of two stages. In the first stage, Li 4 Ti 5 O 12 can be embedded with up to 3 moL Li + at ~1.5 V (vs. Li/Li + ) to produce Li 7 Ti 5 O 12 . In the second stage, Li 7 Ti 5 O 12 is inserted below 1 V (vs. Li/Li + ) by 2 moL Li + to obtain Li 9 Ti 5 O 12 . In addition, the reaction equation can be expressed as [83,84] : Surprisingly, the structure of Li 4 Ti 5 O 12 changes only slightly during charging and discharging. In addition, Figure 2A shows the delithiation/lithiation process of Li 4 Ti 5 O 12 at 3, 1-3 and 0.01-1 V, respectively [83] . However, due to the low conductivity of Li 4 Ti 5 O 12 (10 -13 S cm -1 ) [5] , it is difficult to transfer electrons from the Li 4 Ti 5 O 12 electrode to the external circuit, resulting in high electrode polarization during the continuous cycling process, especially at high current density, which seriously limits its practical application. Researchers have utilized many strategies to enhance the electronic conductivity of Li 4 Ti 5 O 12 anodes and enhance their electrochemical performance. Surface modification and doping can improve the ion diffusion and conductivity, which can improve the high rate performance of the material [85][86][87][88] .
Ion doping usually reduces the theoretical capacity of spinel Li 4 Ti 5 O 12 due to the decrease in active Ti or Li. Hence, the development of new Li 4 Ti 5 O 12 -based materials is imperative in the field of high-power LIBs. In addition, considering the cost of Li, it is vital to exploit low Li intercalation-type Ti-based anode materials. Lately, our group constructed a new Cr,Ti-based complex, Li 5 Cr 7 Ti 6 O 25 , with a low content of Li. According to the one-electron transfer of Ti 4+/3+ and Cr 2+/3+ ions, the theoretical capacity of Li 5 Cr 7 Ti 6 O 25 is ~320 mAh·g -1 when cycled between 3 and 0 V. According to the one-electron transfer of Ti 4+/3+ , the theoretical capacity of Li 5 Cr 7 Ti 6 O 25 is ~147 mAh·g -1 when cycled between 3 and 1 V. Therefore, Li 5 Cr 7 Ti 6 O 25 as a promising anode material has also received significant attention [89] . Mei et al. [90] prepared Li 5 Cr 7 Ti 6 O 25 @CeO 2 composite electrode materials by a simple high-temperature solid phase method and studied the effects of different coating amounts of CeO 2 on the electrochemical properties of electrode materials. The results show that Li 5 Cr 7 Ti 6 O 25 has the best electrochemical performance when the coating amount of CeO 2 is 3 wt.%. In particular, the reversible capacity is 101 mAh·g -1 when cycled between 3 and 1 V after 100 cycles at 5 C. According to the TEM image, the existence of a coating layer can be clearly seen, indicating that there is an interface between Li 5 Cr 7 Ti 6 O 25 and CeO 2 . Figure 2B shows the interface model. According to first-principles calculations, the crystal plane mismatch of the two materials is only 8%, which theoretically explains how the CeO 2 coating can effectively enhance the cycling performance of the materials. Figure 3A shows a schematic of Li 4 Ti 5 O 12 with various 0D, 1D, 2D and 3D nano/microstructures [80] . The morphology control of Li 4 Ti 5 O 12 can be utilized to reduce the diffusion distance of Li + [91,92] . In addition, the nanostructures increase the surface area that can be used to reduce interfacial charge transfer, thus substantially improving the electrochemical activity. Therefore, nanostructured electrodes generally have higher power densities than coarse powder electrodes. The addition of micrometer particles to the holes is another effective method for increasing the rate properties of electrode materials. Micron-sized porous electrodes have the advantage of being easy to assemble and are used for short diffusion pathways of lithium ions. The porous structure can increase the surface area of the electrode material, thus exposing more active sites for lithium-ion insertion into the electrode material, while the nanometer thick wall shortens the diffusion distance of lithium. The nanostructures help to reduce electrode polarization. Microparticles with  [83] . Reproduced from Ref. [83] with permission from Elsevier. (B) Interface model between Li 5 Cr 7 Ti 6 O 25 and CeO 2 [90] . Reproduced from Ref. [90] with permission from the American Chemical Society.
a 3D structure can also ensure contact between the particles, thus increasing the stability of the electrode and inhibiting the capacity loss. The three common preparation pathways of porous Li 4 Ti 5 O 12 are summarized in Figure 3B. The first route uses a template or hole front to fabricate pores in the sample. The pore size and porosity can be easily adjusted by using various templates or pore formers. The second method is combustion synthesis. The formation of pores is mainly through the diffusion of gas in the reaction process. The process of this method is relatively simple but the controllability is poor. The third method is self-assembly synthesis. Pores arise from voids produced from a particular structure and sites left by the removal of some organic groups.
In recent years, a series of research results on Nb-based oxide materials as anode materials have shown significant advantages in the performance of batteries [93,94] . Nb-based oxides have a high potential window (> 1 V vs. Li/Li + ), similar to Li 4 Ti 5 O 12 , which effectively prevents the formation of lithium dendrites and SEI films. The high specific capacity of Nb-based oxides is due to two Li + insertion and extraction during the charge-discharge process, corresponding to Nb 5+ /Nb 4+ and Nb 4+ /Nb 3+ , respectively [95][96][97] . Nb-based oxides are a large family, mainly including the following two types: the Nb-O type represented by Nb 2 O 5 and the M-Nb-O type represented by TiNb x O 2+2.5x . Nb-O compounds include NbO (2+), Nb 2 O 3 (3+), NbO 2 (4+) and Nb 2 O 5 (5+). Nb 2 O 5 is the most common and stable, as well as one of the most studied [98] . As shown in Figure 4, Nb 2 O 5 mainly includes pseudo-hexagonal (TT-Nb 2 O 5 ), orthogonal (T-Nb 2 O 5 ) and monoclinic crystal structures (M-Nb 2 O 5 ) [99] . Although Li + deintercalation may occur in each crystal structure, there are some differences in their electrochemical behavior and performance. Notably, orthogonal T-Nb 2 O 5 has attracted increasing attention due to its pseudocapacitivity [100] . In addition, its (001) lattice spacing (3.90 Å) is about twice the diameter of Li + (1.52 Å). Such large lattice spacing accelerates the diffusion of Li + . In addition, the volume change during the insertion/extraction process of Li + is very small, ensuring a long service life [101] .  [102,103] . The theoretical capacity (mAh·g -1 ) is calculated as follows [103] :  [80] . Reproduced from Ref. [80] with permission from Elsevier.   [99] . Reproduced from Ref. [99] with permission from Elsevier.
achievements. There are many strategies for improving M-Nb-O materials, mainly involving structural engineering [104,105] , doping [106,107] and conductive phase modification [108,109] . Conductive phase modification is considered to be a direct and effective method for improving their electrical conductivity.
Carbonaceous materials are the most common modified conductive phase because of their variety, low cost and good electrical conductivity. Liu et al. [110] synthesized Ti 2 Nb 10 O 29 /C composites by a simple hightemperature solid phase method. As shown in Figures 5A and B, the size and morphology of the two samples are almost the same and the average particle size is ~1 μm. Compared with pure Ti 2 Nb 10 O 29 , Ti 2 Nb 10 O 29 /C composites have relatively high uniformity and dispersion. The discharge capacities of the composite samples at 1, 5 and 10 C were 295.5, 224.8 and 204 mAh·g -1 , respectively. The discharge capacities were 255.7, 214.3 and 194 mAh·g -1 after 100 cycles, respectively, showing the good stability of the materials. Moreover, as shown in Figure 5C, the capacities reach 180.3 and 145 mAh·g -1 when the current density increases from 20 to 30 C. This significant enhancement of the magnification performance is attributed to the improved electronic conductivity resulting from the introduction of amorphous carbon. Ashish et al. [111] prepared TiNb 2 O 7 /Graphene (TNO-TG) composite nanomaterials according to a common solvothermal method, where the TiNb 2 O 7 nanoparticles were anchored on the graphene nanosheets [ Figures 5D and E]. The TNO-TG anode enhanced the electrical conductivity and showed an extremely high rate performance. The retention rate was more than 80% at 16 C and the discharge capacity was 230 mAh·g -1 .
In addition to the above electrode materials, Li 3 VO 4 is another competitive intercalation-type anode material because of its suitable voltage platform (~1.5 V) and high theoretical capacity (394 mAh·g -1 ). Shi et al. [112] synthesized a Li 3 VO 4 /graphene composite by a one-step in situ hydrothermal method. As shown in Figure 5F-H, the Li 3 VO 4 /graphene composite exhibits a unique hollow structure with microbox morphology, a 40 nm wall thickness and is wrapped with porous graphene nanosheets. This hollow structure not only relaxes the stress/strain of Li-ion insertion/extraction, but also increases the surface area of the material. This can offer extra space for lithium storage and reduce the effective diffusion path of lithium ions. Even at 20 C (1 C = 400 mA·g -1 ), the Li 3 VO 4 /graphene composite shows a reversible capacity of 223 mAh·g -1 . After 500 cycles at 10 C, there is no significant capacity fading. However, the formation mechanism of hollow structures has not been discussed in detail. Subsequently, Shi et al. [113] synthesized hollow Li 3 VO 4 microboxes by a fast one-step in situ hydrothermal method and comprehensively studied and discussed their formation mechanism caused by the evacuation process from the interior and the precipitation on the surface. The resulting materials revealed a unique morphology and excellent electrochemical properties. Therefore, Li 3 VO 4 is a promising anode candidate for the development of highperformance, low-cost and advanced LIBs.  [110] . (D) Schematic illustration and (E) TEM image of TiNb 2 O 7 /graphene electrodes [111] . (F-H) FESEM micrographs of LVO/G [112] . Reproduced from Refs. [110,111] with permission from Elsevier. Reproduced from Ref. [112] with permission from the American Chemical Society.

CONVERSION-TYPE ANODE MATERIALS
Conversion-type anode materials (CTAMs) mainly refer to transition metal oxides [114][115][116] , sulfides [26] , phosphides [117] and nitrogen compounds [118] (M = Co, Ni, Fe or Mn). The transformation reaction of metal oxides and sulfides with Li + is as follows [119] : Because there is no position for Li + insertion and extraction in the spatial structure of CTAMs, it is generally believed that the reaction with Li at room temperature is irreversible. CTAMs possess some advantages, such as composition diversity and high theoretical capacity. A schematic of the lithium storage mechanism of typical CTAMs is depicted in Figure 6A [120] . Wang et al. [121] obtained a NiO nanooctahedron with a unique structure by pyrolysis of hexagonal nickel nanoplate microspheres. Such a structure not only offers a large surface area for rapid diffusion of Li + between the anode and the electrolyte, but the exposed active surface can also effectively improve charge transport motion mechanics and lithium-ion diffusion. The reversible specific capacity of NiO can reach 792 mAh·g -1 after 200 cycles at 0.2 C, which fully indicates that NiO has outstanding cycle performance as an anode material for LIBs. Choi et al. [122] prepared Fe 3 O 4 -decorated hollow graphene spherical composites according to spray pyrolysis. Figure 6B exhibits a detailed schematic of the formation process of hollow Fe 3 O 4 /graphene spheres and the electrochemical reaction mechanism. The structure integrity of the Fe 3 O 4 -decorated hollow graphene spheres composites can be well maintained during the deintercalation of lithium, indicating that the hollow Fe 3 O 4 /graphene spheres have good structural stability and the cycling performance of the hollow Fe 3 O 4 /graphene spheres is improved after long cycling at high current density. Fe 3 O 4 was still uniformly dispersed on the graphene spheres after a long period of cycling without aggregation.
Luo et al. [114] used SEM, selected electron diffraction and high-resolution transmission electron microscopy to study the transformation process of Co 3 O 4 nanocrystals with a particle size of ~5 nm [ Figure 7A-C] and the reaction mechanism of Co 3 O 4 as a LIB anode. As shown in Figure 7D, lithium ions enter into the phase of Co 3 O 4 (i.e., Li x Co 3 O 4 ) in the process of lithiation, which is composed of nano-Co-Li-O clusters and is regarded as the intermediate product of the transformation reaction. In the final stage of lithiation, the Co 3 O 4 nanotube is completely transformed into a mixture of Li 2 O substrate and Co elemental clusters growing in it, with significant volume expansion and lattice collapse. This in situ observation is a demonstration of the real behavior of the material during the charging and discharging process, which can deepen our understanding of the mechanism of CTAM reaction and guide us to improve the material performance.
Cho et al. [115] prepared a 1D nanorod-like Fe 2 O 3 /C electrode material using a simple and general electrospinning method by the Kirkendall effect and studied the formation mechanism of the nanobubble structure in detail by tracking and observing each step of the nanobubble formation process [ Figure 8A]. According to SEM [ Figure 8B] and TEM [ Figure 8C] images, the nanofiber consists of hollow Fe 2 O 3 spheres evenly dispersed in an amorphous carbon matrix, which has the advantages of a large specific surface area, short diffusion path of lithium ions and direct and efficient electron transport channels. This enables the material to interact with Li + more effectively, thus giving it with excellent rate capability. In addition, its special bubble-like structure also allows the Fe 2 O 3 bubble void and surrounding carbon atoms to effectively regulate the mechanical stress generated in the process of charging and discharging. Hence, the cycle life of the material is improved. The Fe 2 O 3 /C nanorods exhibited excellent cycling performance when used as an anode electrode of LIBs. During the first thirty cycles, the specific capacity of Fe 2 O 3 /C nanorods has a slight decrease, but almost no decrease from the 31st to the 300th cycles, reaching a maximum of 824 mAh·g -1 at 1 A g -1 . When the measured current density gradually rises from 0.5 to 5 A g -1 and then backs to 0.5 A g -1 , the capacity of the material only slightly decreases. When the current density reaches the maximum, the specific capacity remains at 491 mAh·g -1 , which fully demonstrates the excellent structure.
However, this transformation mechanism was first intercalated to form highly electroactive M nanoparticles and a Li n X matrix surrounded by a SEI film. Due to a large amount of structural rearrangement after lithiation, the voltage lag of CATMs during discharge and charge leads to low energy density and internal  [120] . (B) Schematic of the formation of hollow Fe 3 O 4 /graphene balls and the electrochemical reaction mechanism [122] . Reproduced from Refs. [120,122] with permission from Elsevier. thermal evolution. The voltage lag is closely related to the properties of anions, so the hysteresis is most obvious for fluorides, followed by oxides, sulfides, nitrides and phosphides [120] . In addition, due to its low inherent conductivity and the powder pulverization problem in the repeated cycling process, the rate capacity is poor and the capacity decay is fast [123] . Therefore, significant progress must be made before CATMs become practical electrode materials.

ALLOYING-TYPE ANODE MATERIALS
Alloyed anode materials primarily belong to the IVA and VA groups, which include Si, Ge, Sn, Pb, P, As, Sb and Bi. The lithium storage mechanism is an alloying reaction with lithium to form a Li w M alloy [124] . The  [114] . Reproduced from Ref. [114] with permission from the American Chemical Society. SAED: Selected electron diffraction; HRTEM: high-resolution transmission electron microscopy.
corresponding alloying reaction, taking Si as an example, is as follows [125] : Due to the atoms of any alloy-type material theoretically holding 4.4 or 3 Li + [5] , they exhibit very high specific capacities. For example, silicon (4200 mAh·g -1 ) is ten times higher than graphite in theoretical capacity [126] . It has the highest theoretical capacity of any anode material (with the exception of hydrogen). However, the volume changes greatly during the lithiation reaction, resulting in the formation of a significant outward stress inside the material, which causes the electrode material to be crushed and separated from the collector in serious cases [5] . The SEI film also ruptures, exposing a new surface and requiring the continued consumption of lithium to reform the SEI film. Thus, the cycling performance is relatively low [127,128] .  [115] of bubble-nanorod-structured Fe2O3-C composite nanofibers. Reproduced from Ref. [115] with permission from the American Chemical Society.
In response to the above problems, researchers have found that for a single Si particle, there is a critical size of 150 nm, below which it does not break after lithiation [129] . Therefore, by reducing the size of Si particles, the inside stress of the electrode material can be released, which reduces the possibility of peeling off from the collector in the reaction process. The pores in porous Si also alleviate the large volume expansion. Various Si nanostructures have been designed to improve their cycling stability, such as nanoparticles (0D) [128,130] , 1D structures (nanowires [131,132] , nanorods [133,134] , and nanotubes [135] ), 2D thin films [136] and 3D porous structures [137] . In addition, the reasonable design of Si/C nanocomposites with special structures is also considered to be an effective method to reduce the capacity attenuation caused by the volume change of Si anodes [138] .
Gu et al. [139] synthesized a Si-CNF composite structure via an electrospinning method and further studied the lithiation process with in situ TEM. The Si particles do not break whether embedded or attached to the carbon nanofibers. Simultaneously, compared with the Si particles attached to the surface, the particles embedded in the CNF show delayed lithiation, which limits the rate capacity of the battery. In addition, the lithiation of particles embedded in CNF produce a high stress field, resulting in cracking of CNF. Therefore, the spatial correlation between Si nanoparticles and carbon nanocomposites is very important when designing carbon-based Si nanoparticle composites.
Ge is a common semiconductor material that belongs to the same main group as Si and has similar chemical properties. Ge has better conductivity and a faster lithium-ion diffusion rate than silicon. The surface oxide layer is thinner and Ge-based anode materials generally have higher coulomb efficiency. The volume expansion of Ge is isotropic and the anode material is subjected to uniform stress, avoiding the problem of electrode material cracking caused by stress concentration [140] . As a rare metal, Ge has a relatively high cost. During battery cycling, Ge and Li ions form an alloy, which can form a Li-rich local region, which explains the high lithium storage performance based on the Ge-based anode electrode. Similar to Si, nanoscale and composite materials are effective improvement measures to solve the inherent shortcomings of Ge.
Yan et al. [141] obtained silver-embedded 3D nanoporous Ge (Ag-np-Ge) by a melt spinning and one-step dealloying method. It was found that when Ge is used as the active material, ensuring a high theoretical capability, the porous network can provide sufficient space for volume expansion and contraction of the material. In addition, the embedded Ag nanoparticles can promote the electron transport rate. Ag-np-Ge presents a high capacity of 953 mAh·g -1 after 100 cycles at 100 mA·g -1 and an excellent reversible capacity of 522 mAh·g -1 even at 1000 mA·g -1 .
Sn metal is one of the most studied anode materials due to its excellent theoretical capacity (994 mAh·g -1 ) calculated from the final lithium product of Li 4.4 Sn [142] . However, the large ~300% volume expansion during the lithium process leads to the fracture of the anode material, electrode pulverization, electrical contact failure between the anode materials and conductive additives and an unstable SEI, which limit its commercial application. Jin et al. [143] prepared a novel 3D structured Sn anode material by a simple method. First, the nanosized SnO 2 spheres were heat-treated in a tube furnace (C 2 H 2 /Ar flow mixing at 400 ℃). After heat treatment, the nanosized SnO 2 sphere was transformed into a pure Sn bulk material (~20 μm), which consisted of Sn nanowires (diameter of ~50 nm and several microns in length). The obtained samples have a unique 3D structure with rich voids between the nanowires, which reduce the volume expansion of the Sn bulk material and ensure good electrical contact between the anode material and the conductive additive. The 3D structured Sn anode material shows a specific reversible capacity of 600 mAh·g -1 , with no significant capacity degradation at 0.2 C (compared to the 20th cycle).
The theoretical specific capacity of P is 2596 mAh·g -1 , which possesses a similar electrochemical reaction mechanism to Si [144] , as follows [145] : There are four allotropes of phosphorus, among which white phosphorus is highly toxic and volatile and is therefore not suitable to be used as an electrode material. In recent decades, there has been little research on violet phosphorus. In contrast, red and black phosphorus have good chemical stability at room temperature and atmospheric pressure, so they are often used as electrode materials [146,147] . However, P and Si share the same challenges. So far, the modification of phosphorus anode materials using carbon materials has become the main route to solving these problems [148,149] .
Liang et al. [148] synthesized a free-standing flexible P/C electrode by encapsulating phosphorous in a dualconducting network of porous multichannel carbon nanofibers and in situ carbon nanotubes (P@PMCNFs/CNTs) [ Figure 9A]. The PMCNF/CNT electrode exhibited an outstanding rate performance (601 mAh·g -1 at 3 A g -1 ) and good cycling ability (802.3 mAh·g -1 at 1 A g -1 after 500 cycles). Yan et al. [149] explored a P@rGO-ACW electrode by constructing phosphorus directly from 3D wood-derived carbon and confining it to a 3D micro-channel carbon matrix [ Figure 9B]. This structure not only buffers the volume expansion of phosphorus in the alloying process, but also shortens the transport distance of lithium ions and improves the conductivity of electrons and ions. Therefore, it has good electrochemical behavior. In addition, Sun et al. [150] proposed a novel P-TiO 2 @CNT composite material. The red phosphorus was modified through the synergistic effect of titanium dioxide and CNTs. The modification not only improved the capacity of P but also regulated the stress during the expansion process of red phosphorus and avoided structural damage.
As an important alloy-type anode material, Bi metal has a pseudo-layer structure that is similar to graphite. The volume expansion is 74% due to the formation of a Li 3 Bi alloy, which is significantly smaller than that of Li 4.4 Si (~400%) and Li 4.25 Sn (~257%) [151][152][153] . Yuan et al. [151] designed an egg-carton-like Bi/C nanocomposite structure [ Figure 10A]. From Figure 10B and C, the Bi nanoparticles with a diameter of ~20 nm are placed on micron-sized carbon sheets. The obtained Bi/C fractional complex not only inherits the high electrochemical activity of the Bi nanoparticles but also gains additional advantages from the compact micron-size two-dimensional carbon framework, such as high compaction density, medium specific surface area and strong mechanical protection.
In addition, Bi-based transition metal oxides, as potential anode materials for LIBs, have not been extensively studied. Like all other alloy-type anode materials, volume expansion occur during the repeated reaction, which eventually crushes the active material and separates it from the collector, thus reducing the cycling stability [152,153] . Liang et al. [153] developed a Bi 2 O 2.33 /rGO composite material as an anode material for LIBs. The material has large contact area and unique flexibility. After 600 cycles at 10 C, the capacity was still 346 mAh·g -1 . Deng et al. [152] prepared a Bi 2 O 3 @rGO nanocomposite by a solvothermal method and introduced chemical bonds [ Figure 10D]. The capacity of the anode material is 79% after 100 times of continuous charge and discharge at 0.1 C. Even at 10 C, the capacity can reach 270 mAh·g -1 . The application of the three typical anode material types in the field of LIBs is summarized in Table 1.

CONCLUSIONS AND PERSPECTIVE
In this work, intercalation, conversion and alloying anode materials with different electrochemical reaction mechanisms for LIBs are reviewed. The advantages and problems of each type of anode material and the corresponding optimization strategies are described in detail. Intercalation anode materials, such as Li 4 Ti 5 O 12 , usually have the disadvantages of low electronic and ionic conductivity, which can be improved by constructing nanostructures or by compounding them with other materials with high conductivity. Conversion and alloying anode materials have the advantages of high specific capacity and high energy density. However, their cycling performance is poor. Their electrochemical performance can be improved by designing their morphology (e.g., nanospheres and nanowires) or by modifying them with graphene, carbon foam and other materials with stable structures and high electrical conductivity.
In general, alloying anode materials represented by Si and Sn have become the most attractive anode materials because of their high capacity, but their large volume variation during cycling is the main factor  Figure 9. (A) Schematic illustration of the preparation process of P@PMCNF/CNT composites [148] and (B) P@rGO-ACW electrode [149] . Reproduced from Refs. [148,149] with permission from Elsevier.  [151] . (D) Synthetic processes of Bi 2 O 3 @rGO nanocomposite [152] . Reproduced from Refs. [151,152] with permission from the American Chemical Society.
that impedes their large-scale application. To solve this problem, the most successful method at present is to conduct nanoscale design and future research should focus on the following aspects. The first is a quantitative understanding of nanoscale design, such as the size dependence of nanostructure properties and the development of effective methods for encapsulating nanostructured materials into electrodes. Second are the factors affecting the first coulombic efficiency, such as the formation and properties of the SEI film. Third, in terms of structural design, it is necessary to further develop advanced in situ and non-in situ characterization techniques, combined with first-principles calculations, to reveal the microstructural changes occurring in the process of lithiation/delithiation and understand the inherent electrochemical mechanism and structural advantages. This will allow the Li storage performance to be further optimized.
Similar to alloying anode materials, conversion anode materials also have the problems of easy pulverization, instability of SEI film and large volume change during cycling. To improve the cycling stability, it is necessary to design nanomaterials to realize the mutual conversion of multiple solid phases. In addition, another challenge of conversion-type anode materials is the large voltage hysteresis (~1 V) between charge and discharge. An effective method to solve the problem is to select appropriate electrode design and electrolyte type, such as coating a conductive polymer or lithium-ion conductive solid electrolyte layer on the surface of the anode material.
Anode materials cannot be blindly pursued for high capacity. The synergistic effect of cathode and anode can maximize battery performance. Researchers must design LIB electrodes for overall battery structural stability and high performance and do not need to be limited to current commercial cathode materials. Cathode and anode materials need to be developed together to withstand higher operating voltages and capacities.
Safety is a major consideration in anode design for commercial batteries. Actual battery conditions are often more complex than laboratory test conditions. The structural stability and safety of the battery cannot be ignored. Simultaneously, the selection of materials should avoid toxic and harmful substances. The use of environmentally friendly compounds and composites should be preferred, not only to protect the users but also to reduce the threat to the environment, which is good for recycling.

Authors' contributions
Conceptualization, data curation, writing -original draft: Chang H Conceptualization, data curation, writing -editing: Wu YR Data curation: Han X Writing -review and editing, funding acquisition, supervision: Yi TF

Availability of data and materials
Not applicable.

Financial support and sponsorship
National Natural Science Foundation of China (No. U1960107), the "333" Talent Project of Hebei Province (No. A202005018), and the Fundamental Research Funds for the Central Universities (No. N2123001).

Conflicts of interest
All authors declared that there are no conflicts of interest.

Ethical approval and consent to participate
Not applicable.

Consent for publication
Not applicable.