Unveiling and regulating heterogeneous Li⁺ transport in 3D scaffolds: a gradient electrodeposition strategy toward ultralight and stable Li metal anodes

Abstract

Three-dimensional (3D) current collectors are widely explored to suppress lithium dendrite growth in Li metal anodes, particularly under high-current-density operation. Nevertheless, a fundamental mechanistic understanding of Li electrodeposition within such complex 3D architectures remains markedly incomplete. Herein, using a 3D Cu mesh as a model system, we demonstrate that intrinsic heterogeneity in Li⁺ transport within 3D architectures promotes localized ion-flux amplification, triggering preferential deposition and dendritic nucleation. Guided by this insight, a gradient-descending electrodeposition is proposed to regulate ion flux distribution, enabling dense, uniform, and strongly adherent bulk-like Li deposition. To mitigate the gravimetric penalty associated with conventional metallic scaffolds, an ultralight PET@Cu 3D current collector is further engineered via polymer metallization, achieving an areal mass reduction of 59.2% relative to the Cu mesh while retaining sufficient electrical conductivity and mechanical robustness. When integrated with the gradient deposition strategy, the resulting PET@Cu@G-Li anodes paired with NMC811 cathodes exhibit high-rate capability and exceptional cycling stability, retaining 90.82% of their initial capacity after 600 cycles at 3C. This work elucidates the critical role of mass-transport heterogeneity in Li electrodeposition within 3D frameworks and offers a practical anode design paradigm that reconciles high power density with high gravimetric energy density in Li metal batteries.

A lightweight PET@Cu 3D scaffold with gradient Li deposition, achieving dense Li growth, reduced inactive mass, and stable high-rate cycling.

Introduction

Rapid advances in electric vehicles, portable electronics, and electric aviation impose increasingly stringent demands on the energy density of rechargeable batteries.^1–3 The energy density of state-of-the-art commercial Li-ion batteries, which typically employ graphite anodes, is now approaching 300 Wh kg⁻¹, close to their theoretical ceiling and insufficient to meet future requirements. Among various candidates, Li metal anodes stand out due to their unparalleled theoretical specific capacity (3861 mAh g⁻¹) and the lowest electrochemical potential (−3.04 V vs. SHE), making them a promising choice for next-generation high-energy-density batteries exceeding 500 Wh kg⁻¹.^4–6 Despite being hailed as the “holy grail” of anode materials, Li metal faces formidable challenges in practical implementation due to its intrinsically high reactivity and unstable electrodeposition behavior.^7,8 During charging and discharging of lithium metal anodes, especially under high-power (high current density) conditions, lithium ions tend to form irregular dendritic deposits.^9–12 These sharp lithium dendrites can penetrate the separator, causing internal short circuits and triggering thermal runaway or even fires and explosions. In addition, they tend to fracture into electrochemically inactive dead lithium, which continuously depletes limited electrolyte and active lithium sources, leading to sharp declines in coulombic efficiency and rapid capacity fading.¹³ Furthermore, spontaneous and vigorous side reactions between lithium metal and the electrolyte form unstable, thick, and porous solid electrolyte interphase (SEI) films, further aggravating interfacial impedance and irreversible lithium loss.^14–17 Thus, suppressing dendrite growth, stabilizing the electrode/electrolyte interface, and improving cycling reversibility are critical prerequisites for overcoming the intrinsic limitations of Li metal anodes and enabling their commercialization.^18–20

To tackle these challenges, a variety of strategies have been proposed, including electrolyte engineering^21–26 (e.g., high-concentration electrolytes, additives), construction of artificial SEI layers,^27–30 and the design of three-dimensional (3D) current collectors.^31–39 Among these, 3D conductive current collectors are particularly attractive as a potentially “two-birds-with-one-stone” solution. In principle, 3D frameworks (such as metal foams/meshes or carbon-based scaffolds) offer high specific surface area and porous architectures that can effectively reduce the local current density, thereby kinetically suppressing dendrite nucleation and growth. In parallel, 3D collectors provide ample internal free volume to accommodate Li volume changes and mitigate mechanical stress within the electrode. As a result, 3D current collectors are anticipated to enhance both cycling stability and rate capability of LMAs.⁴⁰ In recent years, substantial progress has been made in the design and modification of novel 3D current collectors based on metals and carbons. However, most studies have focused primarily on material fabrication, lithiophilic surface engineering, or structural optimization, while in-depth mechanistic understanding of Li electrodeposition within complex 3D topologies remains limited. Classical nucleation/growth theories developed for two-dimensional (2D) planar current collectors are frequently invoked to interpret Li deposition on 3D scaffolds, yet the complex pathways for Li⁺ and electron transport in 3D architectures have received insufficient attention.⁴¹ The similarities and differences between Li deposition on 3D current collectors and on 2D planar substrates are therefore not yet clearly established. Moreover, most commonly used 3D current collectors (e.g., Cu mesh, Cu foam, Ni foam, and carbon cloth) do not contribute any capacity and thus constitute inert components. These current collectors, particularly metal-based ones, typically exhibit excessively high areal densities, diluting the theoretical energy density advantages of lithium metal anodes due to their disproportionate weight fraction in the battery.⁴²

Herein, this work presents a systematic investigation into lithium metal electrodeposition on three-dimensional (3D) current collectors and the development of high-performance Li-free anodes. We begin by selecting a representative 3D Cu mesh as a model system and employ comprehensive morphological tracking and electrochemical analyses to unravel the underlying mechanisms of Li deposition within 3D architectures. Our findings reveal that the intrinsic non-uniformity of Li⁺ transport pathways in such structures can exacerbate local reaction polarization, triggering three distinct modes of preferential Li deposition, paradoxically leading to more severe dendrite formation under certain conditions, such as low current densities. Guided by these mechanistic insights, we design and optimize a gradient-descending current deposition protocol. By programmatically modulating deposition kinetics, this approach enables the formation of dense, uniform, and strongly adherent bulk-like Li deposits on the Cu mesh, significantly enhancing electrochemical stability. To further address the gravimetric penalty imposed by conventional metallic current collectors, we adopt a polymer metallization strategy to fabricate an ultralight PET@Cu 3D current collector, achieving an areal mass reduction of over 59.2% compared to normal Cu mesh. Integrating this lightweight scaffold with our optimized deposition protocol yields a high-performance PET@Cu@G-Li anode. When paired with a high-Ni NMC811 cathode, the resulting full cells exhibit exceptional high-rate (3C) cycling stability, retaining more than 90% of their initial capacity after 600 cycles. This study not only offers new fundamental insights into Li deposition behavior in complex 3D environments but also establishes a practical anode design paradigm that simultaneously optimizes kinetic performance and gravimetric energy density, paving the way for next-generation, high-energy, high-power lithium metal batteries.

Results and discussion

To elucidate the impact of 3D Cu mesh current collectors (with structure and surface morphology shown in Fig. S1) on the electrochemical behavior of Li metal anodes, we first conducted a systematic analysis of the Li plating process. As illustrated in Fig. 1a, during charging, Li⁺ ions in the electrolyte migrate toward the Cu mesh working electrode and are reduced to metallic Li at the Cu/electrolyte interface. We initially focused on the mass transport of Li⁺ in the electrolyte and the resulting concentration field distribution. Given that the 3D porous geometry of the Cu mesh (single-wire diameter of 25 µm, Fig. S2) and the comparable thickness of the electrolyte-wetted separator (20 µm), the variations in Li⁺ transport pathways induced by the mesh architecture, relative to the overall transport distance, constitute a non-negligible factor. Longer Li⁺ transport paths (for example, toward the backside of Cu wires further away from the separator) are associated with higher ionic transport resistance. Under galvanostatic deposition, Li⁺ preferentially migrates along paths with the lowest transport resistance, typically favoring the shortest transport distance (Fig. S3). Consequently, the Cu wire surfaces adjacent to the separator experience a higher Li⁺ flux, which in turn sustains a higher bulk Li⁺ concentration in these regions at the electrode/electrolyte interface. To validate this hypothesis, we performed finite element simulations using COMSOL Multiphysics to analyze the spatial distribution of Li⁺ concentrations during deposition on the Cu mesh electrode at various current densities (Fig. 1b and S4). The simulation results clearly reveal that, as deposition progresses, the Li⁺ concentration field exhibits pronounced nonuniformity, with higher Li⁺ concentrations at Cu filament surfaces proximal to the separator, in full agreement with our analysis. It is noteworthy that the copper mesh, as a metallic conductor, constitutes an equipotential body in its entirety, thereby enabling nearly unimpeded electron transport to all locations. Consequently, the local overpotential that drives lithium deposition is governed primarily by the local Li⁺ concentration at each specific site. On the surface of an equipotential body, even minor disparities in Li⁺ concentration between distinct positions translate directly into differences in local overpotential, which in turn induce selective deposition. By comparison, the conductive surface of commercial Cu foil is macroscopically planar, and the Li⁺ transport pathway length remains uniform across all positions, precluding any concentration variation arising from geometric structure. This characteristic distinguishes 3D current collectors fundamentally from 2D planar counterparts. It should be noted that the simulation model represents an idealized geometry rather than an exact reconstruction of the experimental collector. It neglects wire intersections, out of plane curvature, and the complete connectivity of the 3D pore network. These simplifications can affect the magnitude and precise spatial distribution of the calculated Li⁺ concentration and flux. The simulation is therefore interpreted qualitatively to illustrate geometry induced transport heterogeneity rather than to predict exact local concentrations, current densities, or deposit morphologies.

Fig. 1. Analysis of lithium deposition behavior on Cu mesh. (a) Schematic illustration of Li deposition on Cu mesh. (b) COMSOL simulation of the spatial distribution of Li⁺ concentrations during deposition on the Cu mesh at 1 mA cm⁻². (c) Schematic illustration of three modes of preferential Li deposition on Cu mesh.

The electromotive force (EMF) of the Li metal electrode during electrochemical deposition on the anode side can be described by the Nernst equation:^43,44

1

where E is the cell potential under non-standard conditions (V), is the standard electrode potential of the Li⁺/Li redox couple (V), R is the universal gas constant, T is the absolute temperature (K), F is the Faraday constant, and a_Li⁺ is the activity of Li⁺ in the bulk phase. According to the Nernst equation, a higher Li⁺ concentration (activity) corresponds to a more positive equilibrium electrode potential. Thus, under a fixed applied electrode potential, regions with higher Li⁺ concentration necessarily experience a larger local overpotential, that is, a stronger electrochemical driving force for Li deposition. As a result, Li is deposited preferentially on the Cu wire surfaces closer to the separator, where the Li⁺ concentration is highest. This Li⁺ concentration gradient, originating from the non-uniformity of ion transport paths, is one of the fundamental causes of inhomogeneous Li deposition within the Cu mesh structure.

In practice, as summarized in Fig. 1c, the preferential deposition behavior is more complex and can be categorized into three main types. Type I is substrate selectivity (Li-on-Li vs. Li-on-Cu). Li tends to continue growing on pre-deposited Li rather than nucleating on bare Cu. This is primarily because Li nucleation on the heterogeneous Cu substrate must overcome a substantially higher energy barrier than deposition on a homogeneous Li substrate.⁴⁵ In addition, the surface diffusion of Li adatoms on Cu is relatively sluggish, leading to poor initial electrical contact and mechanical adhesion between the freshly deposited Li and the current collector.⁴⁶ Type II is site selectivity, which governs where subsequent Li growth is localized. Li preferentially deposits at certain hot-spots that act as dominant growth sites. These hot spots are directly related to local Li⁺ enrichment (such as regions near the separator side identified above). Furthermore, spatial heterogeneity in the properties of the SEI, including variations in thickness, ionic conductivity, and chemical composition among different sites, also plays a crucial role in inducing site-selective deposition. Type III is directional selectivity, another key factor dictating Li metal morphology. At individual deposition sites, Li tends to exhibit one-dimensional (1D) or quasi-1D preferential growth along specific directions (e.g., the [110] crystal plane) rather than uniform isotropic 3D growth. This growth mode directly produces elongated dendritic or whisker-like Li morphologies, and remains one of the most intensively discussed challenges in Li metal anode research. This behavior is typically attributed to locally elevated current densities, heterogeneous SEI coverage, and the intrinsic anisotropy of crystal growth.^47–50 Overall, although driven by distinct yet strongly coupled forces, the three forms of preferential Li deposition coexist throughout the deposition process, with their relative contributions evolving with deposited capacity. Type I can enhance Type II by concentrating subsequent deposition on a limited number of existing nuclei. Type II can intensify Type III by directing a greater ion flux toward selected sites. The protrusions generated through Type III can then shorten local transport paths and further focus the ion flux. As deposition proceeds, Cu coverage becomes more complete, the relative differences among transport paths decrease, and the SEI becomes more developed. For the 3D Cu mesh current collector, the inherent non-uniformity in Li⁺ transport pathways imposed by its architecture can directly or indirectly intensify all of these preferential deposition tendencies.

Following the above analysis, we first investigated the evolution of Li deposition morphology on the Cu mesh at a conventional current density of 1.0 mA cm⁻² (Fig. 2a and S5a). Surprisingly, even at a very small areal capacity (e.g., 0.3 mAh cm⁻²), the Cu wire surfaces already exhibit pronounced dendritic features, suggesting that factors other than current density may dominate dendrite formation. With increasing deposition capacity, these dendritic Li structures undergo progressive coarsening. When the deposited capacity reaches 5.0 mAh cm⁻², the metallic Li essentially covers the entire Cu mesh surface, yet the resulting morphology remains highly non-uniform. At an extremely low deposition capacity of 0.1 mAh cm⁻², the initial nucleation morphology of spherical lithium metal can be clearly observed (Fig. S6). However, when the deposition capacity is slightly increased to 0.3 mAh cm⁻², obvious dendritic lithium deposition becomes observable. It is widely accepted that high current densities, particularly those approaching the Li⁺ limiting diffusion current, are primary a driver of dendritic Li growth. However, we observe pronounced dendrite formation on the Cu mesh even at a conventional areal current density of 1.0 mA cm⁻² (Fig. 2a). Given that the Cu mesh current collector, by virtue of its high surface area, effectively reduces the local current density below the critical threshold for dendrite formation, these results indicate that dendritic growth on the Cu mesh is not triggered by high-rate conditions in the classical sense. To further clarify the role of current density in governing deposition behavior, we examined the morphological evolution over a broader current-density window. As shown in Fig. 2b and S5b, at a lower current density of 0.1 mA cm⁻², more pronounced dendritic Li structures were observed. This counterintuitive result further supports the conclusion that dendrite formation on the Cu mesh is not simply a consequence of excessive current density. Tracking the deposition process at a higher current density (5 mA cm⁻², Fig. 2c and S5c) reveals that, during the initial stage (<1.0 mAh cm⁻²), such a high current density promotes relatively uniform coverage of the Cu-mesh surface by Li. In the later stage at 5 mAh cm⁻², loosely stacked spherical and particulate Li is observed, exhibiting a morphology that is fundamentally distinct from the dendritic Li formed at low current density. We further observed the deposition process at an ultrahigh current density of 25 mA cm⁻² (Fig. S7). At the early stage of deposition (corresponding to capacities of 0.3 and 0.9 mAh cm⁻²), the lithium morphology was highly similar to that obtained at 5 mA cm⁻². At a higher deposition capacity of 5 mAh cm⁻², this current density of 25 mA cm⁻² produced loose spherical lithium deposits. Galvanostatic profiles recorded at different current densities (Fig. S9) show that both the nucleation overpotential and the subsequent deposition plateau overpotential increase markedly with increasing current density. Intriguingly, the high overpotential at 5 mA cm⁻² does not lead to severe dendrite formation, whereas pronounced dendritic and highly non-uniform Li deposition occurs at the lower current density of 0.1 mA cm⁻² with a much smaller overpotential.

Fig. 2. Morphological characterization of Li deposition on Cu mesh at different states with different current density protocols. SEM images of Li deposition on Cu mesh (a) at 1.0 mA cm⁻², (b) at 0.1 mA cm⁻², (c) at 5 mA cm⁻², (d) with a gradient descending current-density protocol.

The current-density-dependent electrodeposition behavior of Li on the 3D Cu mesh current collector thus differs significantly from several reported trends on conventional two-dimensional (2D) planar current collectors. Li electrodeposition generally consists of nucleation followed by growth. According to classical nucleation theory, the nucleation rate J (s⁻¹ m⁻²) can be expressed as:

2

where A is a pre-exponential factor, k is the Boltzmann constant, T is the absolute temperature, and ΔG* is the nucleation energy barrier. For heterogeneous nucleation of Li on a Cu substrate, ΔG* is related to the overpotential by:

3

where η (V) is the overpotential, B is a constant determined by:

B ∝ (γ³)/(z²F²) 4

where γ represents interfacial energies, z is ionic valence. The nucleation rate J can therefore be rewritten as:

5

In general, the nucleation density N is positively correlated with J. Under high current density, the electrode experiences a large overpotential η, which exponentially enhances the nucleation rate J, leading to a high nucleation density and a smaller critical nucleation radius.⁵¹ As a result, at high current density, a large number of nanoscale nuclei rapidly form and cover the Cu-mesh surface. This theoretical prediction is fully consistent with our experimental observation of rapid, uniform coverage at the early stage of deposition under 5 mA cm⁻² (Fig. 2c). Regarding the dendritic behavior observed on the 3D Cu mesh at low current density, we attribute it to the markedly different interplay between nucleation and growth under low η. At low current density (low η), the nucleation rate J is reduced, leading to a lower nucleation density, while the critical nucleation radius increases slightly (though likely still in the nanometer regime). We speculate that the micron-sized spherical particles frequently reported at low current densities in the literature are more plausibly the outcome of post-nucleation growth and coarsening, rather than being dictated directly by the nucleation step itself. In our Cu-mesh system, although a relatively sparse population of nanoscale nuclei is formed at low current density, the subsequent growth is strongly governed by the non-uniform Li⁺ mass transport described in Fig. 1c. Differences in Li⁺ transport path length give rise to local Li⁺ concentration inhomogeneities, which in turn amplify the tendency toward preferential deposition (i.e., Type-II and Type-III selectivity). The deposits thus evolve into dendritic morphologies rather than uniform, isotropic spherical growth. In addition, the seemingly counterintuitive observation that Li deposition at high current density (5 mA cm⁻²) exhibits a more homogeneous macroscopic spatial distribution than at low current density (Fig. 2c) can be rationalized by considering the behavior of the SEI. Under higher current density (high overpotential) conditions, the deposition process is accompanied by more frequent SEI fracture and reconstruction events, as schematically illustrated in Fig. S10. These fracture-reconstruction events exhibit strong spatial randomness, with a relatively weak correlation to the geometry of the 3D current collector (i.e., the Li⁺ transport pathways). At the separator-proximal side (Site 1), due to preferential sustained deposition, the SEI has undergone more extensive evolution, and R_SEI is higher. At the separator-distal side (Site 2), due to more frequent re-nucleation and SEI fracture-reconstruction, the local SEI is thinner and R_SEI is lower. This spatial difference in SEI impedance can partially compensate for the spatial difference in Li⁺ bulk-phase transport impedance (where Site 1 has low R_{Li⁺ diffusion} and Site 2 has high R_{Li⁺ diffusion}), thereby tending to equalize the total deposition impedance at different positions at the macroscopic level. As a result, the spatial distribution of lithium is relatively uniform, although the local morphology remains granular and structurally loose (Fig. 2c).

Indeed, both the dendritic lithium and the loosely stacked particulate lithium exhibit high specific surface areas and unstable structures, which are detrimental to achieving high coulombic efficiency and long cycle life. Guided by the above mechanistic insights, we therefore propose a gradient descending current-density protocol (Fig. 2d). In the initial high-current-density stage, a large overpotential ensures a high nucleation density, enabling rapid and uniform wetting of the Cu substrate by Li metal and effectively suppressing Type-I (substrate selectivity) preferential deposition. In the subsequent medium-current-density stage, a dynamic balance is expected to be established between SEI-constrained stable growth and SEI fracture-reconstruction. A second balance arises between the spatially varying Li⁺ transport resistance in the electrolyte and the Li⁺ transport resistance across the SEI at different locations. These two balancing effects equalize the deposition activity among different sites at the macroscopic level, thereby mitigating Type-II (site selectivity) preferential deposition. In the final low-current-density stage, the accumulated Li loading significantly reduces the relative differences in Li⁺ transport path length imposed by the 3D current collector architecture. In addition, the depth dependent XPS spectra identify LiF and Li₂O within the SEI and demonstrate the evolution of these inorganic components during the gradient protocol (Fig. S11). LiF and Li₂O are chemically stable and electronically insulating phases that can restrict electron transfer to the electrolyte and suppress continued electrolyte decomposition. Although bulk LiF and Li₂O have limited intrinsic ionic conductivity, defects and internal interfaces between these phases may provide more favorable pathways for Li⁺ transfer. These inorganic components may also increase the resistance of the interphase to local deformation. Together with the established Li coverage, reduced transport heterogeneity, and low final current density, the evolved multiphase SEI can reduce localized reaction amplification and promote compact Li growth. Operating at low current density in this stage ensures that Li is deposited under a strongly constraining SEI, suppressing Type-III (directional selectivity) growth and promoting the formation of compact, bulk-like Li domains. As shown in Fig. 2d, this strategy ultimately yields a dense, bulk-like Li layer that is tightly and uniformly anchored onto the Cu wires, as further corroborated by the corresponding voltage profiles (Fig. S12). We further investigated the evolution of Li deposition morphology under alternative protocols of H-to-L (high current density to low current density) and L-to-H (low current density to high current density), with the results presented in Fig. S13. Under the H-to-L protocol, dendritic lithium structures emerge from the initially uniform lithium layer plated at high current density (Fig. S13a), demonstrating that, even in the absence of Type-I preferential deposition, preferential deposition of Types II and III can still induce nonuniform dendritic lithium formation. In contrast, under the L-to-H protocol, although subsequent high current densities promote uniform isotropic 3D growth on the Cu mesh to a certain extent, the dendritic lithium formed during the initial low current density phase persists and continues to evolve (Fig. S13b). A substantial portion of the Cu surface remains exposed, indicative of severe Type-I preferential deposition, while Type-II and Type-III preferential depositions endure, collectively exacerbating the increasingly nonuniform dendritic lithium deposition. The morphological evolution observed under the H-to-L and L-to-H protocols further highlight the advantages of the gradient descending current-density protocol. This multi-stage, stepwise current-density approach offers important design guidelines for constructing high-performance anode-free and Li-less anodes based on 3D current collectors.

We next focus on the electrochemical performance of Li-less anodes prepared on the Cu mesh using different deposition protocols. Three representative electrodes with distinct Li morphologies were selected for comparison (top-view and cross-sectional images in Fig. 2b–d and S14), including the Cu-mesh@L-Li with dendritic lithium fully covering the copper mesh surface obtained at a low current density of 0.1 mA cm⁻², the Cu-mesh@H-Li with loosely packed spherical lithium relatively uniformly adhered to the copper filament surfaces obtained at a high current density of 5 mA cm⁻², for the Cu-mesh@G-Li with dense bulk-like lithium tightly and uniformly bonded to the copper mesh surface obtained via gradient current density deposition. For Cu-mesh@L-Li and Cu-mesh@H-Li electrodes, a total capacity of 5 mAh cm⁻² was deposited by using a constant current of 0.1 mA cm⁻² and 5 mA cm⁻², respectively. And for Cu-mesh@G-Li electrode, a gradient descending current density deposition, consisting of sequential stages at 5 mA cm⁻² for 0.5 mAh cm⁻², 1 mA cm⁻² for 1 mAh cm⁻², and 0.1 mA cm⁻² for 3.5 mAh cm⁻², totaling 5 mAh cm⁻². The current densities and capacities of the gradient protocol were selected according to the distinct kinetic objective of each deposition stage. The initial stage deposits 0.5 mAh cm⁻² at 5 mA cm⁻². This current density generates a large nucleation overpotential and a high nucleation density, while the limited capacity prevents an excessive fraction of Li from being deposited as loose particles under the high current condition. The intermediate stage deposits 1 mAh cm⁻² at 1 mA cm⁻² and provides a gradual transition during which Li coverage and the SEI continue to evolve. The final stage deposits 3.5 mAh cm⁻² at 0.1 mA cm⁻². At this point, the surface has been substantially covered and the relative differences among transport pathways have decreased. The lower current density consequently favors compact growth under the constraint of the developed SEI. The total capacity is maintained at 5 mAh cm⁻² to permit direct comparison with the constant current electrodes. These parameters represent a mechanism guided protocol identified within the investigated experimental window rather than a global optimization of all possible current densities, capacities, and stage numbers.

The initial deposition morphology is of great significance for lithium metal anodes. Firstly, the initial deposition morphology affects the initial lithium stripping efficiency. A lower lithium stripping efficiency corresponds to a greater consumption of active lithium during the initial deposition process. Secondly, the initial lithium deposition morphology influences the subsequent state of SEI, thereby exerting a continuous impact on the subsequent lithium deposition. Such an impact will also be directly reflected in the cycling performance of the battery. We first evaluated the initial stripping coulombic efficiency of the three electrodes (Fig. 3a) to quantify the loss of active Li associated with the first deposition process. Due to their markedly different microscale morphologies (dendritic, particulate, and bulk-like), the three electrodes deliver different first-stripping capacities. Cu-mesh@G-Li has the highest initial stripping coulombic efficiency (98.35%), which is higher than those of the Cu-mesh@L-Li (96.70%) and the Cu-mesh@H-Li (96.02%). For the three anodes, after ten plating/stripping cycles (1 mA cm⁻², 1 mAh cm⁻²), clear divergences in active Li retention emerge. As shown in Fig. 3b, Cu-mesh@G-Li retains 97.73% of its active Li, distinctly higher than Cu-mesh@L-Li (96.03%) and Cu-mesh@H-Li (95.77%). These results indicate that differences in the initial deposition morphology not only affect the total amount of active lithium in the lithium-less anode, but also that poor electrical contacts (including Li–Cu interfaces and inter-Li-particle contacts) during repeated deposition/stripping cycles can induce partial disconnection of active lithium from the current collector, leading to dead Li and thereby accelerating capacity decay. The dense, bulk-like Li structure in Cu-mesh@G-Li and its intimate integration with the Cu mesh evidently provide superior preservation of electrical connectivity throughout cycling. Symmetric-cell tests (Fig. 3c and S15) further corroborate the electrochemical advantages of Cu-mesh@G-Li. Over a current-density range of 0.2–5 mA cm⁻², Cu-mesh@G-Li consistently exhibits the lowest polarization, demonstrating outstanding rate capability. By contrast, Cu-mesh@L-Li, whose 3D pores are clogged by dendritic Li deposits, shows the highest overpotential at high rates, indicating that its rate performance is compromised. When the symmetric battery was cycled at 2 mA cm⁻² with 1 mAh cm⁻² (Fig. 3d), Cu-mesh@G-Li maintained the most stable cycle at a low polarization (more than 300 hours). By contrast, voltage instability occurred after ∼50 h for Cu-mesh@L-Li and ∼160 h for Cu-mesh@H-Li due to an internal short circuit caused by either Li dendrite growth or depletion of active Li. The disparity in rate capability is further manifested in full cells employing LFP‖Li-Cu-mesh configurations (Fig. 3e). At a high rate of 5C, the LFP cell based on Cu-mesh@L-Li delivers a specific capacity of only 91.89 mAh g⁻¹, significantly lower than that of cells based on Cu-mesh@G-Li (114.57 mAh g⁻¹) and Cu-mesh@H-Li (113.28 mAh g⁻¹) under identical testing conditions. Consistently, the corresponding charge/discharge profiles (Fig. S16) show that Cu-mesh@G-Li exhibits the smallest polarization at 5C (215.4 mV), lower than that of Cu-mesh@H-Li (226.8 mV) and Cu-mesh@L-Li (265.1 mV).

Fig. 3. Electrochemical performance of three kinds of Cu mesh@Li anodes. (a) Voltage profiles from the initial coulombic efficiency measurements. (b) Voltage profiles of coulombic efficiency test after ten cycles. (c) Overpotential of symmetric cell tests at different current densities. (d) Voltage profiles of the symmetric cells during cycling. (e) Rate performance of Cu mesh@Li‖LFP(∼3 mg cm⁻², 0.495 mAh cm⁻²). (f) Cycle performance of Cu mesh@Li‖LFP cells (∼3 mg cm⁻², 0.495 mAh cm⁻²).

When paired with an LFP cathode with a low area loading of 3 mg cm⁻², these three Li-less anodes display markedly different long term cycling performance (Fig. 3f and S17). Cu-mesh@H-Li exhibits rapid capacity decay after around 670 cycles, which can be attributed to its loosely packed particulate architecture. On the one hand, poor interparticle electrical contact readily generates dead Li during cycling; on the other hand, the excessively high specific surface area amplifies continuous parasitic reactions with the electrolyte and the associated Li consumption. Cu-mesh@L-Li, in contrast, exhibits relatively robust cycling, retaining more than 80% of its capacity after 900 cycles. Under low cathode loading, corresponding to a low depth of discharge (DOD) on the anode side, the intrinsic interfacial-contact issues at the Li–Cu interface are not fully activated. Cycling predominantly occurs near the outer surface of the pre-existing Li deposits, enabling prolonged cycling stability. Cu-mesh@G-Li exhibits the optimal cycling performance, with a capacity retention of 80.11% after over 1000 stable cycles, while consistently delivering higher reversible capacity than Cu-mesh@L-Li throughout cycling, benefiting from the well-preserved dense structure during cycling. Post-cycling morphological observations (Fig. S18) further reveal that the 3D architectures of Cu-mesh@H-Li and Cu-mesh@G-Li are preserved even after 200 cycles; however, Cu-mesh@H-Li exhibits significant disorder due to severe interfacial reactions, consistent with the performance trends discussed above. When paired with a high-loading LFP cathode (∼15 mg cm⁻²), the Cu-mesh@L-Li exhibited the poorest performance, with the capacity retention rapidly decaying to 74.2% after merely 34 cycles (Fig. S19). Under high-loading conditions (corresponding to a higher DOD on the anode), the poorest cycling stability exhibited by Cu-mesh@L-Li can be attributed to the pronounced effect of poor Li–Cu interfacial contact, leading to extensive dead Li formation. Moreover, because the 3D framework is already clogged, subsequent Li deposition can only proceed on the external surface of the electrode, where rapid accumulation of SEI by-products causes a sharp increase in cell polarization. The intrinsic drawbacks of Cu-mesh@H-Li (loose packing and high specific surface area) are likewise exacerbated at high loading, and its capacity falls below 80% within only 70 cycles. In contrast, Cu-mesh@G-Li achieved a capacity retention of 80% after 86 cycles under these high-loading conditions. Collectively, these results demonstrate that, for Li-less anodes constructed with the same 3D Cu mesh current collector, the electrodeposition current density plays a decisive role in determining the macroscopic Li morphology and the resulting electrochemical performance. The gradient current-density strategy, which yields a dense bulk-like Li layer intimately bonded to the 3D Cu mesh, outperforms electrodes prepared at a single (either high or low) current density in both rate capability and long-term cycling stability (Fig. S20).

Although the 3D Cu mesh current collector significantly enhances the electrochemical performance of Li metal anodes, the intrinsically high density of Cu (ρ_Cu = 8.96 g cm⁻³) leaves room for further optimization from the perspective of achieving higher gravimetric energy density. To this end, we employed a polymer metallization strategy, using a low-density, highly flexible polyethylene terephthalate (PET, ρ_PET ≈ 1.38 g cm⁻³) mesh as a lightweight scaffold, and fabricated a lightweight 3D conductive current collector (PET@Cu) via an electroless Cu plating process.^52,53 We then applied the optimized gradient current deposition protocol (G-Li) to construct a lightweight 3D Li metal anode (PET@Cu@G-Li) on this PET@Cu framework, as schematically illustrated in Fig. 4a. The fabrication procedure of PET@Cu is illustrated in Fig. 4b. Specifically, PET mesh was sequentially subjected to alkaline degreasing, strong oxidant roughening, SnCl₂ sensitization, and AgNO₃ activation treatments, resulting in the formation of catalytic active Ag sites on the surface. Subsequently, in an alkaline plating bath, Ag catalyzed the reduction of Cu²⁺, depositing a uniform and dense metallic Cu layer on the PET fiber surface. The resulting PET@Cu mesh (optical image in Fig. S21) exhibits a uniform metallic luster. SEM images (Fig. 4c), together with energy-dispersive X-ray spectroscopy (EDS) mapping (Fig. S22), confirm that a dense and conformal Cu layer uniformly coats the PET fibers without visible cracks or voids. In addition, elemental analysis indicates that the surface is dominated by Cu and O, the latter arising from slight surface oxidation. Cross-sectional SEM images (Fig. S23) show that the Cu coating has a thickness of approximately 1–2 µm. X-ray diffraction (XRD) patterns (Fig. 4d) further verify that the plated layer consists predominantly of metallic Cu (cubic phase), accompanied by a minor CuO phase. From a manufacturing perspective, electroless Cu plating is attractive because this aqueous process does not require line of sight deposition or an external electrical connection to the insulating PET mesh and can therefore produce conformal Cu coatings on complex porous substrates. Its sequential immersion steps could in principle be adapted to larger batch systems or continuous web processing. Large area uniformity, bath lifetime, batch reproducibility, mechanical durability, and production yield should be evaluated in future pilot scale studies.

Fig. 4. Fabrication of PET@Cu@G-Li. (a) Schematic illustration of the preparation process of PET@Cu@G-Li. (b) Schematic illustration of the preparation process of PET@Cu by electroless Cu plating. (c) SEM image of PET@Cu. (d) XRD patterns of PET@Cu. (e) Galvanostatic profiles of Li deposition on PET@Cu with a gradient descending current-density protocol. (f) SEM image of PET@Cu@G-Li after 5 mAh cm⁻² deposition via the gradient descending current-density protocol.

Based on the successfully fabricated PET@Cu collector, we performed Li plating using the gradient current strategy. The corresponding galvanostatic deposition profile (Fig. 4e) reveals lower nucleation/deposition overpotentials and a smoother voltage plateau than those observed for conventional current collectors, indirectly highlighting the kinetic advantages of the 3D conductive PET@Cu network. The resulting PET@Cu@G-Li (5 mAh cm⁻²) anode (Fig. 4f) displays macroscopically flat and uniform Li coverage. SEM images show dense, bulk-like Li domains tightly anchored to the mesh surface, while the inherent 3D open architecture of the current collector is clearly preserved. This ideal electrode configuration (combining compact Li with open pore channels) suggests that uniform Li plating/stripping can be maintained during cycling, while efficiently buffering the inevitable volume changes associated with Li deposition and removal.

We further evaluated the key physical properties of the PET@Cu current collector and compared them with those of Cu mesh and commercial Cu foil. As shown in Fig. 5a, the areal density of PET@Cu (approximately 5.35 mg cm⁻²) is comparable to that of commercial Cu foil (approximately 6.24 mg cm⁻²) and far lower than that of Cu mesh (approximately 13.11 mg cm⁻²), reflecting its significant lightweight advantage. Four-probe testing shows that the sheet resistance of PET@Cu (approximately 24.6 mΩ m⁻¹) is close to that of Cu mesh (approximately 27.1 mΩ m⁻¹), fully meeting the conductivity requirements for current collectors. Mechanical tests (Fig. 5b) reveal that PET@Cu exhibits the best flexibility and tensile strength, surpassing both Cu mesh and Cu foil, and is capable of withstanding the mechanical stresses encountered during battery manufacturing and operation. Overall, PET@Cu demonstrates significant advantages in lightweight design and mechanical properties compared to Cu mesh while maintaining sufficient conductivity, making it an ideal 3D current collector solution that balances gravimetric energy density with structural reliability. Moreover, for Li‖PET@Cu half-cells (Fig. 5c and S24), stable cycling is maintained for over 50 cycles with coulombic efficiency above 98%, confirming its electrochemical suitability as a current collector for Li metal anodes.

Fig. 5. Performance of PET@Cu@G-Li. (a) Areal density and resistance of PET@Cu. (b) Tensile curve of PET@Cu. (c) Voltage profiles of Li‖Cu@PET. (d) Cycle performance of PET@Cu@G-Li‖NMC811 (∼4.5 mg cm⁻², 0.81 mAh cm⁻²) full battery at a 0.5C rate. (e) Rate performance of PET@Cu@G-Li ‖NMC811 (∼4.5 mg cm⁻², 0.81 mAh cm⁻²) full cells. (f) Cycle performance of PET@Cu@G-Li‖NMC811(∼4.5 mg cm⁻², 0.81 mAh cm⁻²) full battery at a 3C rate.

Finally, we systematically assessed the full-cell performance of the lightweight 3D anode (PET@Cu@G-Li), using a pre-Li-plated commercial Cu foil anode (Cu foil@Li) as a benchmark. It is worth noting that, to match the operational requirements of the high-voltage NMC811 cathode, the LHCE electrolyte with a wider electrochemical working window was adopted in all NMC811-based tests. When paired with an NMC811 cathode (Fig. 5d), the PET@Cu@G-Li electrode maintains a capacity retention of 82.8% after 1000 cycles at 0.5C, whereas the Cu foil@Li electrode undergoes rapid capacity decay after approximately 650 cycles. The corresponding charge–discharge curves at selected cycles (Fig. S25) further confirm the stable cycling of PET@Cu@G-Li and the rapid capacity decay of Cu foil@Li. Rate-capability tests further reveal that PET@Cu@G-Li consistently outperforms Cu foil@Li at all tested C-rates (Fig. 5e and S26), delivering a specific capacity of 170.71 mAh g⁻¹ at 2C. Moreover, PET@Cu@G-Li demonstrates exceptional electrochemical stability under high-rate testing at 3C. As illustrated in Fig. 5f and the corresponding charge–discharge curves in Fig. S27, the PET@Cu@G-Li electrode achieves a capacity retention of 90.82% after 600 cycles and sustains stable operation beyond 700 cycles. Throughout the cycling period, its capacities remain significantly higher than those of Cu foil@Li (the latter experiencing an abrupt capacity drop after around 530 cycles). When paired with a thicker NMC811 cathode (areal loading of approximately 15 mg cm⁻², 3.15 mAh cm⁻²), the electrode prepared via the gradient current strategy achieved approximately 84.4% capacity retention after 250 cycles (Fig. S28). In contrast, under the same testing conditions, the control electrode using a conventional Cu current collector exhibited pronounced capacity decay at around 120 cycles, showing significantly inferior cycling stability. This result clearly demonstrates that the strategy proposed in this work can effectively enhance battery performance even under high-loading conditions that more closely approximate practical applications. The lightweight PET@Cu current collector, obtained via the metallization of plastic, substantially reduces the weight fraction of the current collector. On this basis, the optimized gradient current density deposition enables uniform and tight deposition of bulk-like Li onto PET@Cu, yielding a 3D PET@Cu@G-Li anode that disperses current density, suppresses lithium dendrite growth, and provides space to accommodate lithium deposition while effectively mitigating volume expansion during deposition. Consequently, compared to planar lithium metal anodes, it exhibits superior rate performance and extended cycling stability at high rates. In addition, the cycling stability of the PET@Cu@G-Li anode was compared with that reported in recently published representative studies using 3D current collectors under comparable testing conditions, including cathode material loading, rate, and current density (Table S1). The results demonstrate that the PET@Cu@G-Li anode compares favorably with the reported systems in cycling stability and rate capability.

Conclusions

In this study, we systematically unravel the complex lithium deposition mechanisms in 3D current collectors and identify the intrinsic non-uniformity of Li⁺ transport pathways, arising from the 3D architecture, as a primary driver of local concentration polarization, which intensifies preferential deposition and promotes dendritic growth. Building on this mechanistic understanding, we develop a gradient-descending current deposition protocol that enables the formation of dense, uniform, and strongly adherent bulk-like lithium morphologies (Cu-mesh@G-Li) on Cu-mesh current collectors. This approach yields Li anodes with markedly enhanced electrochemical stability compared to those fabricated under constant-current conditions. To further address the gravimetric energy density limitations imposed by conventional metallic current collectors, we engineer an ultralightweight PET@Cu scaffold via electroless copper plating, achieving over 59.2% mass reduction relative to standard Cu mesh. Applying the same gradient-current strategy to this lightweight framework produces a high-performance PET@Cu@G-Li anode. Full-cell evaluations paired with NMC811 cathodes demonstrate its exceptional rate capability and cycling durability, maintaining 90.82% of its initial capacity after 600 cycles at a 3C rate, significantly outperforming conventional Li metal anodes. Collectively, this work establishes a new paradigm for actively regulating ion transport and deposition behavior in 3D electrodes. It also provides a practical and scalable anode design blueprint for lithium metal batteries that simultaneously achieve high specific energy, high-rate capability, and long cycle life.

Author contributions

Y. Shen, Y. Tang, H. Shao, L. Chen, J. Ge and Y. Zhang supervised the project, conceived the idea and designed the experiments. J. Wang, G. Jiang, Q. Tang, H. Zhang carried out the experimental work. J. Wang, Y. Shen, Y. Tang, H. Shao analysed the data and wrote the manuscript. All authors have given approval to the final version of the manuscript.

Conflicts of interest

There are no conflicts to declare.

Data availability

The supporting data are provided in the supplementary information (SI). Supplementary information: experimental methods, SI figures and SI note. See DOI: https://doi.org/10.1039/d6sc06232f.