The porosity of the current collector is very important for addressing the volume expansion problem of metallic lithium during cycling, hence a large number of studies have used current collectors with porous structures as lithium metal anodes. These include various metal foams, dealloyed porous materials, and metal meshes. Metal foams^[25⇓-27], such as copper, nickel, and titanium foams, possess advantages like high structural stability, large surface area, and good electrical conductivity, and are widely used as current collector materials. Li et al.^[27] prepared nanoporous nickel foam with a large surface area and surface defects through a redox method (Figure 3a). The macroporous structure of this skeleton can partially accommodate volume changes, while well-distributed nanopores can effectively reduce local current density, promote uniform distribution of lithium ion flux, and regulate uniform lithium nucleation and deposition.

In addition to metal foams, graphene foams^[28], melamine foams^[29] have also been used as lithium anode hosts. For example, Wu et al.^[30] prepared a 3D conductive Ti₃C₂T_x-MXene-melamine foam (MXene-MF) (Figure 3b). Thanks to the high conductivity of MXene and the good lithiophilicity of fluorine terminations, as well as the porous, lightweight, and excellent mechanical toughness of melamine foam, the resulting MXene-MF possesses a 3D porous conductive network structure and good mechanical strength, enabling highly dense and uniform lithium deposition, with good cycling life even at a high current density of 50 mA·m⁻² and a high deposition capacity of 50 mAh·cm⁻².

Interwoven 3D current collectors are formed by the interweaving of one-dimensional materials, such as metal meshes^[31-32], freestanding frameworks composed of metal nanowires^[33-34], or three-dimensional disordered network structures interconnected by carbon nanotubes and carbon fibers^[35], etc., which can effectively alleviate the volume changes of lithium.

Zhang et al.^[31]used the method of ammonia oxidation to modify lithiophilic CuO and SnO₂onto a commercial bronze mesh, serving as a lithium anode carrier. By mechanically pressing this carrier with lithium, a composite anode (3D CSM/Li) was prepared. The bronze mesh has a robust framework, and the lithiophilic CuO and SnO₂interfaces can optimize the SEI layer and reduce overpotential, thus achieving uniform lithium plating. The heavy metal current collector correspondingly limits the mass/volume energy density of lithium metal batteries. Therefore, researchers have designed the use of high-aspect-ratio metal nanowires as the host for lithium metal anodes. Yang et al.^[36]designed a self-supporting conductive nickel nanowire mesh framework coated with Li_{2O (Ni@Li2O-NW) (Figure 4). In this, the nanostructured nickel framework serves as a conductive network, while the thin dielectric and lithium-ion conductive Li2O coating not only helps in forming a uniformly distributed internal electric field but also acts as a quasi-SEI layer and promotes lithium ion transport, thereby inducing uniform deposition of lithium ions.}

It is well known that the lithium deposition pattern is crucial in determining the cycling stability and safety of batteries. In conductive three-dimensional porous frameworks, lithium preferentially deposits at the top, leading to uneven deposition and easy formation of lithium dendrites, ultimately causing a short circuit in the battery, a phenomenon commonly referred to as the "top growth" mode. However, constructing three-dimensional current collectors with gradient structures^[37] can achieve bottom-up growth of lithium. Ai et al.^[38] grew CuO nanowires on copper foils using electrochemical methods, producing a series of 3D integrated gradient copper current collectors with varying degrees of dispersion, by adjusting the dispersity of the nanowire arrays to control their interfacial structure. The study showed that interface structures composed of both sparse and densely distributed CuO nanowire arrays were unfavorable for the nucleation and deposition of lithium, leading to rapid dendrite growth. In contrast, uniformly and appropriately dispersed nanowire arrays (M-CuO@Cu) enabled stable and uniform nucleation of lithium metal at the bottom, with lateral deposition of lithium at the bottom, thus providing an ideal bottom-up lithium growth pattern (Figure 5a). The M-CuO@Cu electrode could stably cycle for 1200 h under testing conditions of 1 mA·cm⁻², 1 mAh·cm⁻². Moreover, the assembled pouch cell had a specific capacity of 120 mAh·g⁻¹ after 70 cycles at 0.5 C, with a capacity retention rate as high as 99.4%, capable of continuously driving a 3 V electric fan, demonstrating excellent practicality and compatibility. Liu et al.^[39] used a solution combustion method to deposit ZnO with different contents onto copper foam to construct a gradient lithiophilic framework (GSZO-CF) (Figure 5b), where the content of lithiophilic Zn increased from top to bottom, inducing preferential deposition of lithium within the copper foam, and the graded lithium ion transport barrier and ionic conductivity allowed for uniform regulation of lithium ions, achieving even deposition of lithium, further suppressing the formation of lithium dendrites and dead lithium. Even under ultra-high current density of 50 mA·cm⁻² and high capacity of 10 mAh·cm⁻², the symmetric cell could still stably cycle over 800 cycles (Figure 5c). Similar to metal-based skeletons, carbon structures can also serve as the main body of gradient skeletons. To suppress the growth of lithium dendrites, Kang et al.^[40] designed a dual-gradient ultra-light carbon nanofiber skeleton (CBG) (Figure 5d). The dual gradients included a top-down ZnO gradient and a bottom-up Sn gradient prepared by magnetron sputtering. Before lithium deposition, ZnO reacted with lithium ions to form Li-Zn alloys and Li₂O, endowing the carbon framework with a higher lithium ion diffusion coefficient, accelerating the transmission of lithium ions. Meanwhile, the bottom Sn, with good electrical conductivity and strong lithiophilicity, greatly reduced the overpotential for lithium nucleation, guiding the preferential deposition of lithium at the bottom. Additionally, a carbon nanotube-interwoven polyhedron (CNIP) evolved from a ZIF-8@ZIF-67 core-shell framework also formed a lithiophilic gradient from inside out (Figure 5e)^[41]. Lithiophilic ZnO concentrated within the carbon polyhedra induced the preferential nucleation and inside-out deposition of metallic lithium. Simultaneously, the hollow carbon nanotube network, with its excellent electronic conductivity, provided dual continuous channels for lithium ions and electrons, effectively inhibiting the growth of lithium dendrites and maximizing the spatial utilization of the polyhedra.

Alloying in three-dimensional current collectors not only reduces the activity of lithium metal, thereby reducing its reactivity with the electrolyte, but also lowers the nucleation overpotential of metallic lithium, thus effectively inhibiting the formation of lithium dendrites. Therefore, alloy-structured current collectors have been widely studied to improve the performance of lithium anodes, such as lithium-magnesium alloys, lithium-indium alloys, lithium-tin alloys^[42], and lithium-silicon alloys^[43].

Wu et al.^[44]prepared a multi-component alloy composite framework through a simple two-step preparation method, obtaining a composite alloy anode (Li-Mg-Ca) by melting, rolling, and heat treatment, which can form uniformly distributed nucleation sites, uniform lithium ion flux, and reduce local current density to inhibit the growth of lithium dendrites (Figure 6a). The Li-Mg-Ca electrode exhibits excellent mechanical properties, thermal stability, interfacial stability, and electrochemical stability. The Li-Mg-Ca alloy electrode can stably cycle for 1000 h at a current density of 2 mA·cm⁻²and an areal capacity of 4 mAh·cm⁻². The assembled practical pouch cell with an energy density of 350 Wh·kg⁻¹works well and can meet the energy demands of electronic devices. Wang et al.^[45]prepared a nano-level uniformly distributed Li-Al alloy composite anode (NLA20, where the mass content of Al is 20 %) using an in-situ lithiothermic reduction method (Figure 6b). Compared to pure lithium anodes, the in-situ formed lithium aluminum alloy is uniformly distributed in the lithium anode, providing fast ion diffusion channels, reducing the nucleation potential of lithium, inducing uniform lithium deposition, and also suppressing side reactions at high temperatures. In symmetric cell tests, the NLA20 electrode exhibited an extremely low overpotential of 5 mV and ultra-long cycle life at 60 ℃.

Combining three-dimensional metal current collectors with lithiophilic nanomaterials, utilizing the synergistic effect between the two, can achieve stable lithium metal anodes, thereby enhancing battery performance. Due to the poor plasticity of metal-based materials, the common three-dimensional structures of metal-based materials are mostly foam-like, mesh-like, and nanowires, Table 2 summarizes some studies on using lithiophilic metal nanostructures to modify copper foam and nickel foam current collectors, as well as their performance in symmetric cells. Firstly, using copper foam as a three-dimensional current collector, modifying its surface with lithiophilic nanomaterials such as ZnO nanosheets^[51], Cu nanowires^[52], Cu₂O nanowires^[53], CuS₂ nanowires^[54], etc., can induce uniform nucleation of lithium, achieving uniform electroplating of metallic lithium. Recently, Choi et al.^[55] grew hierarchical Cu₂₊₁O nanowires on the surface of copper foam via wet chemical methods and further modified the surface of the nanowires with an ultrathin and uniform layer of Sn/Cu₆Sn₅ composite material (SCC@H-CF) (Figure 8a), which can reduce the nucleation barrier for lithium, inducing uniform nucleation; the large surface area of Cu₂₊₁O nanowires can lower the local current density to evenly distribute surface charge, thus further achieving uniform deposition of lithium. During testing, the symmetric cell could stably cycle for over 3000 hours at a current density of 0.5 mA·cm⁻², with a voltage hysteresis maintained at 8 mV. Moreover, when matched with a LiFePO₄ cathode, the assembled full cell exhibited excellent cycling stability and rate capability. Similar to the above-mentioned lithiophilic modification strategies, our team^[56] vertically grew hierarchical Cu fibers on copper foam as a lithiophilic scaffold (HCF/CF). The secondary protrusion structure on the surface of the Cu fibers serves as charge centers and electrochemical sites, inducing uniform nucleation and growth of lithium. The excellent mechanical strength of HCF/CF ensures that the composite lithium anode has high structural stability during rapid and repetitive lithium plating/stripping processes. Compared to copper foam, the HCF/CF scaffold shows significantly improved lithium plating/stripping behavior, maintaining a high Coulombic efficiency of 98% over 200 cycles. To promote the practical application of lithium metal batteries, we will explore their use in pouch cells in future research.

In addition, using nickel foam as a three-dimensional current collector, preparing lithium-philic ZnO nanorods^[57], NiF_x nanosheets^[58], V₂O₅ nanoribbon arrays^[59], Co₃O₄ nanoarrays^[60], and NiCo₂O₄ nanorods^[61] on its surface can inhibit the growth of lithium dendrites and form a stable SEI film. Nan et al.^[62] synthesized a Ni₃S₂ layer with excellent lithium affinity (Figure 8b) on nickel foam, which can provide abundant nucleation sites to induce uniform lithium nucleation. Furthermore, Ni₃S₂ reacts in situ with metallic lithium to form Li₂S, creating a uniform SEI film that effectively regulates lithium ion flux and achieves rapid ion transport.

Modifying metal current collectors with heteroatom-doped carbon coating is also an effective strategy to enhance their lithiophilicity. Liu et al.^[63] prepared 3D nickel foam modified by N, S co-doped carbon using an interfacial polymerization method, which was used as a lithium metal host (NSC@Ni) (Figure 8c). The N, S co-doped carbon significantly enhanced the lithiophilicity of the skeleton, allowing for rapid injection of molten lithium into the skeleton to form the NSC@Ni-Li composite lithium anode. During the molten infusion process, in-situ formed Li₃N and Li₂S layers with high lithium ionic conductivity, and the strong interaction between metallic lithium and N, S co-doped carbon helped regulate the lithium ion flux and reduce the nucleation barrier of lithium, ultimately inhibiting the growth of lithium dendrites.

Negatively charged heteroatoms (such as B, N, O, S, P, etc.) have strong interactions with positively charged lithium ions, making heteroatom doping an effective strategy for enhancing the lithiophilicity of carbon materials. Zhang et al.^[64] summarized the calculated binding energies between various heteroatom-doped carbons and lithium atoms (Figure 9a). Zhang et al.^[65] investigated the application of nitrogen-doped graphene (NG) in lithium metal anodes, revealing the impact of different nitrogen functional groups on the uniform nucleation of lithium and the lithiophilic nature of nitrogen (Figure 9b). The overpotential of lithium plating on copper foil and NG electrodes was experimentally explored, with the overpotential of NG being much lower than that of copper foil, indicating that the former has higher lithiophilicity (Figure 9c). Density functional theory calculations showed that, compared to graphene and copper, pyrrolic and pyridinic nitrogens have greater binding energies with lithium, at -4.46 and -4.26 eV, respectively (Figure 9d). The increased local charge density between pyrrolic nitrogen and lithium indicates a strong interaction between lithium and N atoms (Figure 9e). Pyrrolic nitrogen groups contain an extra electron, and pyridinic nitrogen groups contain lone pair electrons. Nitrogen-containing Lewis base sites strongly adsorb Lewis acidic lithium ions through acid-base interactions, guiding the uniform distribution of lithium nuclei on the anode surface. In contrast, quaternary nitrogen, with its saturated electron orbitals, cannot provide additional electrons to adsorb lithium ions, resulting in relatively low binding energy. This provides theoretical guidance for incorporating nitrogen-containing functional groups into carbon-based materials to enhance lithiophilicity.

Yu et al.^[66] proposed the use of carbon cloth (CC@CN-Co) modified with nitrogen-doped carbon nanosheet arrays embedded with Co nanoparticles as a 3D matrix for pre-storing lithium. Nitrogen-containing functional groups have a strong interaction with lithium, which can guide uniform nucleation of lithium, and tiny Co nanoparticles provide active sites to guide uniform deposition of lithium. Therefore, this matrix exhibits characteristics of being free from lithium dendrites and stable cycling performance. Zhang et al.^[67] obtained an integrated reduced graphene oxide/carbon nanotube 3D framework rich in imine bonds (iPANI@rGO-CNTs) through the in-situ copolymerization of phytic acid and aniline. Using lithiophilic imine bonds as anchor points for strong LiPS adsorption and uniform lithium deposition, they deposited lithium on the matrix via electroplating to obtain the composite lithium metal anode Li-iPANI@rGO-CNTs, which can form a uniform electric field, thereby regulating the uniform deposition of lithium. As a result, the composite lithium anode endows the lithium-sulfur full cell with excellent rate performance and superb cycling performance. Compared to single-doping systems, dual-doping systems exhibit superior lithiophilicity. In the oxygen and boron co-doped honeycomb-like carbon skeleton (OBHcCs) electrode^[68], lithiophilic oxygen and boron functional groups can reduce the lithium nucleation barrier, promoting the uniform distribution of lithium ion flux. Additionally, this skeleton has a high specific surface area, effectively reducing local current density, thus inhibiting the growth of lithium dendrites. Therefore, the OBHcCs@Li||LiFePO₄ full cell shows a good capacity retention rate of 84.6% and a high coulombic efficiency of 99.6% after 500 cycles at 0.5 C.

The introduction of metal nanoparticles into carbon materials can enhance the lithiophilicity of the skeleton, thereby inducing uniform lithium nucleation. Commonly used nanoparticles include Au^[69-70], Ag, Zn^[71,72], Co^[73], Sn^[74], Ni^[75], Si^[76], etc. Gao et al.^[77] deposited a series of lithiophilic metal particles (CNF/Me, Me=Sn, Fe, Co) on three-dimensional self-supporting carbon nanofibers to induce lithium deposition behavior. Lithiophilic metals not only provide a large number of active sites but also enhance lithiophilicity, reducing the nucleation barrier of lithium and thus achieving uniform nucleation and deposition of lithium. Additionally, Mao et al.^[78] coated lithiophilic silver nanowires onto 3D carbon cloth (Ag NWs/CC) using a vacuum filtration method (Figure 10a). The introduced silver nanowires provided uniform lithium nucleation sites and alleviated the growth of lithium dendrites.

In addition to metal nanoparticles, doping with metal compounds, such as oxides (ZnO^[35,79], SnO₂ ^[80], MnO₂ ^[81], etc.), nitrides^[82], or sulfides^[83]is another method to improve the lithiophilicity of carbon-based materials, providing active sites for lithium metal and enhancing the lithium wettability of carbon materials. Recently, Peng et al.^[84]decorated lithiophilic Mo₂C clusters on the surface of carbon nanofibers (Mo₂C@CNF). On one hand, the uniformly dispersed lithiophilic Mo₂C clusters provide a large number of nucleation sites for lithium deposition and effectively reduce the nucleation barrier; on the other hand, transition metal elements act as catalysts, forming a stable and robust SEI layer containing LiF on the Mo₂C@CNF, which effectively reduces the formation of dead lithium and improves the Coulombic efficiency during long-term operation (Figure 10b). The symmetric cell assembled with it can stably cycle for more than 1600 h at a current density of 1 mA·cm⁻², with an overpotential of about 13 mV. Wei et al.^[85]modified the carbon framework with arrays of highly lithiophilic transition metal nitrides, including CoN, VN, and Ni₃N, and prepared composite lithium metal anodes by a thermal melting casting method (Figure 10c). The carbon skeleton and metal nitrides have excellent electronic conductivity, ensuring rapid electron transfer at the liquid-solid interface and providing a stable interface for lithium reduction. At the same time, the lithiophilic interconnected framework serves as a stable "nest" to store lithium metal and alleviate volume changes during cycling. In the symmetric cell test, due to the lattice matching between CoN and metallic lithium, the Li/CoN@CF composite anode can achieve stable cycling for more than 1000 h under high current densities of 20 mA·cm⁻²and high areal capacities of 20 mAh·cm⁻², demonstrating ultra-stable cycling performance. In addition, carbon nanofibers decorated with MoS₂nanosheets^[86], ultrafine TiN nanoparticles^[87], and Mo₂N^[88]also show excellent performance in stabilizing the lithium metal anode.