Extremely fast-charging lithium ion battery enabled by dual-gradient structure design.

Extremely fast-charging lithium-ion batteries are highly desirable to shorten the recharging time for electric vehicles, but it is hampered by the poor rate capability of graphite anodes. Here, we present a previously unreported particle size and electrode porosity dual-gradient structure design in the graphite anode for achieving extremely fast-charging lithium ion battery under strict electrode conditions. We develop a polymer binder-free slurry route to construct this previously unreported type particle size-porosity dual-gradient structure in the practical graphite anode showing the extremely fast-charging capability with 60% of recharge in 10 min. On the basis of dual-gradient graphite anode, we demonstrate extremely fast-charging lithium ion battery realizing 60% recharge in 6 min and high volumetric energy density of 701 Wh liter-1 at the high charging rate of 6 C.

INTRODUCTION

Electric vehicle (EV) powered by the lithium ion battery (LIB) is one of the promising zero-emission transportation tools to address air pollution and energy crisis issues (). However, much longer recharging time of the EV than the gas-refilling time of traditional fuel vehicle makes it much less competitive (). In this scenario, building up extremely fast-charging LIB system is highly desirable, and the U.S. Department of Energy has recently proposed a fast-charging goal of 10 miles min−1 for the EV (). The fast-charging capability of the LIB at the material level is limited by the sluggish reaction kinetics and the low equilibrium potential (within 100 mV versus Li+/Li) () of the graphite anode, which tend to induce the metallic lithium plating under a high charging rate (,). In the past decades, porous graphite particle (), solid electrolyte interphase modification on the graphite (), and the aligned graphite via magnetic field () have been reported to improve the rate performance of the graphite anode (–). Nevertheless, all these reported graphite anodes behaved with limited electrode parameters such as low graphite mass loading, high porosity, and low thickness, which largely sacrificed the energy density of the fabricated LIBs (). Therefore, up to now, it is still very challenging to overcome the trade-off between the energy density and fast-charging performance of the LIBs.

In practical graphite anode with required energy density (porosity < 35% and thickness > 70 μm), there is a detrimental polarization effect (, ) during the fast-charging process leading to the lithium metal plating on the surface of the electrode. The polarization effect in the graphite anode is mainly attributed to the concentration polarization of Li+ ion in the whole electrolyte (–). To acquire high mass transport of Li+ ions and reduce the concentration polarization in the graphite anode, microstructure optimization of the electrode is a promising route (, ). However, we found that the critical microstructure parameters, average particle size, and electrode porosity have only been individually considered in previously reported pseudo–two-dimensional (P2D) models to reduce the electrode polarization effect (, –). The real electrode design of particle size distribution and porosity distribution calls for microstructure-resolved models to overcome the trade-off between the energy density and fast-charging performance of the LIBs (). More than that, less experimental investigation has been carried out to verify the performance improvement related to the electrode structural design.

RESULTS

Particle-level theoretical model to optimize the thick electrode structure

On the basis of previous reports (), high porosity facilitates to improve the transportation in the thick electrode but decrease the energy density. In addition, the commercialized graphite particles usually have different sizes. Considering the real graphite anode (inhomogeneous, nonisotropic with pores, and particles with different sizes and shapes) with fixed overall porosity, we first used a particle-level theoretical model to simultaneously optimize the spatial distributions of different sized particles and electrode porosity for the fast charging in a thick electrode. We assume pores and particles in the thick electrode to be randomly distributed along the current collector plane and inhomogeneously distributed perpendicular to the current collector plane. To simplify the simulation, a representative cuboid volume containing many different graphite particles is selected in the particle-level model to represent the whole electrode (fig. S1A).

A two-step iterative design method is used to optimize the electrode structure in the simulation(detailed electrochemical simulation process in the Supplementary Materials) (–). Given the thick electrode with fixed overall porosity (35%) containing randomly distributed particles, we design the spatial distribution of the porosity. First, a gradient porosity structure with higher porosity on the top than the bottom, which is referred to single-gradient type, is constructed to compare with the electrode having randomly distributed porosity (random type) (). [The distributions of the porosity and particles size along the thickness direction are shown in fig. S1 (B to D).] The charging process at high current density of 3.75 mA cm−2 (the theoretical areal capacity of the electrode is 7.5 mAh cm−2) is simulated to investigate the charging performance of the proposed electrode structures. As shown in fig. S1E, when the electrodes were charged to the cutoff voltage (0 V versus Li+/Li), the single-gradient type exhibited higher utilization and lower overpotential. However, the improvement is not evident, and then, to further improve the charging performance, gradient distribution of different sized particles is introduced into single-gradient–type structure named dual-gradient type. Two opposite particle size distributions are constructed [ST (smaller top) dual gradient with smaller particles on the top and BT (bigger top) dual gradient with bigger particles on the top]. The simulated charging curve at 3.75 mA cm−2 indicated that the ST dual gradient could improve the fast-charging performance notable (fig. S1E). As a contrast, the BT dual-gradient type even shows worse charging performance than single-gradient type (fig. S1F). We further investigated the physical reason for the improvement. The lower overpotential mainly results from the reduced concentration polarization caused by a more uniform concentration distribution in the designed structure. As shown in Fig. 1A, when the electrodes were charged to the cutoff voltage (0 V versus Li+/Li) at the current density of 3.75 mA cm−2, the Li+ ion concentration distribution in the liquid phase showed that the other type electrodes experience very sharp Li+ ion concentration polarization from the top to the bottom, whereas the dual-gradient electrode exhibits obviously much more smoothed Li+ ion concentration distribution (fig. S1F), resulting a reduced overpotential (fig. S1G). The physical reason for the smoother Li+ ion concentration distribution under fast charging in ST dual-gradient structure is from two physical aspects. On the one hand, higher porosity on the top improve the local effective diffusion coefficient, which is favor to the Li+ diffusion from the top and hence decrease the lithium concentration on the top compared to the uniformly distributed porosity for the random type electrode. On the other hand, at the late stage of charging, the smaller particles near the top of the electrode can be easily fully lithiated and then decrease the electrochemical reaction rate on the top of the electrode; hence, consume less Li+ on the top and more Li+ can reach to the bottom. As the result, the active material utilization of random and single-gradient electrodes decreased intensively, whereas the dual-gradient electrode retained the much higher active material utilization especially at high charging current densities (Fig. 1B), which is very attractive for the fast-charging graphite anode. In the following text, the ST dual gradient is referred to dual-gradient–type structure for simplification.

Fig. 1.

Concentration polarization analysis in different electrode structures and the simulated graphite particle utilization at different current densities.

(A) Spatial distribution of the Li+ ion concentration in the liquid phase (cl,Li+) across the electrodes with different structures. (B) Overall utilization of the graphite particles in different structured electrodes at the different charging current density. The insets show the utilization distribution in each individual particle for these three structured electrodes at 1 C (1 C = 3.75 mA cm−2).

Fabrication of the graphite electrode with dual-gradient structure

Using traditional manufacture technique is hard to construct as-proposed dual-gradient structure in the graphite anode because of the high viscosity of the as-used slurry (0.15 Pa·s). Figure S2A shows the fabrication procedure of our proposed dual-gradient graphite anode. We developed a polymer binder–free slurry with a viscosity as low as 1.75 × 10−3 pa s (fig. S2B) to achieve this dual-gradient structure in the graphite anode. The copper-coated graphite (G@Cu) particles and copper nanowires (CuNWs) were first prepared via facile solution methods, and corresponding characterizations (fig. S2, C to F) confirmed the uniform morphologies and good surface coatings. The polymer binder–free slurry was made by dispersing G@Cu particles and CuNWs in the ethanol. Then, the slurry was poured into a mold to induce the sequential assembly of CuNWs and G@Cu from large size to small size during the process of ethanol evaporation due to their different sedimentation speed endowed by the low viscosity of the slurry. After the ethanol evaporation, a membrane consisted of interconnected G@Cu particles and CuNWs (fig. S2G) was formed, which was then annealed and calendered to yield a free-standing dense electrode with the size of 4 cm by 10 cm (inset in fig. S2A). The color difference between the top and bottom surfaces of the electrode indicated that more CuNWs and small G@Cu particles appeared on the top surface of the electrode, and large G@Cu particles preferred to localize at the bottom surface (fig. S3, A and B).

The dense microstructure in the obtained electrode was revealed by the cross-sectional scanning electron microscopy (SEM) image (Fig. 2A). The graphite particle size analysis based on the cross-sectional SEM image from the top to the bottom indicates that the average particle size approximately increased from 15 to 35 μm (fig. S3C). The enlarged local region SEM images overlapped with elementary mapping images (Fig. 2, B to D) further show that the CuNWs percolated with each other to form connected network among the graphite particles, and the CuNWs content decreased from the top to the bottom. These results imply that G@Cu particles and CuNWs display gradient distribution in the electrode. The mass content of electrochemically inactive copper in the obtained free-standing electrode was determined as 14.13% via the thermal gravimetric analysis (TGA; fig. S3, D and E), much lower than that in the commercial graphite electrode (30 to 40%; table S1). Meanwhile, the obtained G@Cu-CuNWs graphite anode exhibited good flexibility and higher electronic conductivity than the commercial graphite anode due to the unique line-point contact endowed by the CuNWs rather than the point-point contact between carbon black dots and graphite particles (fig. S3, F and I) ().

Fig. 2.

The microstructure characterization and the electrochemical performance evaluation of thick dual-gradient G@Cu-CuNWs anode and random graphite anode.

(A) Cross-sectional SEM image of the G@Cu-CuNWs anode with a thickness of 200 μm showing the dense structure of the obtained electrode. Scale bar, 50 μm. (B to D) Magnified cross-sectional SEM image overlapped with elementary mapping at the region from the top to the bottom along the thickness direction. Scale bars, 20 μm. Yellow color represents the copper, and red color represents the carbon. (E) Simulation of the Li+ ion concentration distribution in the dual-gradient G@Cu-CuNWs and random graphite anode based on the reconstructed 3D structure of the fabricated electrodes (50 μm by 50 μm by 150 μm). Color bar represents different concentrations of Li+ ion. The applied current density is 5 mA cm−2. (F) Calculated average utilization of the graphite particles along the thickness direction. (G) Simulated charge voltage profiles of the G@Cu-CuNWs and graphite anode. L is defined as the distance to the bottom surface of electrode.

The good line-point contact in the G@Cu-CuNWs facilitated the fabrication of thick dense electrode. For the traditional polymer binder slurry fabrication, the thick graphite anode tended to be detached from the copper foil current collector after the calendar rolling (fig. S4, A and B), in contrast to the integrated G@Cu-CuNWs thick electrode (fig. S4C). In details, with the gradually rolling process, the thickness of G@Cu-CuNWs electrode decreased from 311 to 243 and then to 198 μm (fig. S4, D to F), and the corresponding compaction density increased from 0.91 to 1.16 and to 1.43 g cm−3 (fig. S4G), respectively. Meanwhile, the particle size electrode porosity dual-gradient dense structure was formed in the interconnected network electrode. Thus, our fabricated 3D network electrode with all in one structure shows the advance in the fabrication of thick graphite anode.

Evaluation of dual-gradient structure for improving the fast-charging performance

To show the role of dual-gradient structure in reducing the electrode polarization, we evaluated the electrochemical performance of the fabricated thick G@Cu-CuNWs anode and compared it to the random structured graphite anode with the same thickness. In the half cell, although both thick G@Cu-CuNWs and graphite electrodes behaved normal discharge/charge voltage profiles with good coulombic efficiency (CE) (91% for G@Cu-CuNWs and 89.2% for random graphite anode) in the first cycle, the G@Cu-CuNWs electrode exhibited lower polarization than the graphite electrode (fig. S5A). The charge/discharge voltage profiles at the different rates (fig. S5B) also indicated the lower polarization in G@Cu-CuNWs electrode compared to the graphite electrode. The cyclic voltammetry tests (fig. S5C) revealed reduced redox peak position and enhanced peak current in the G@Cu-CuNWs electrode, further indicating the lower polarization and improved reaction kinetics in the G@Cu-CuNWs electrode. Moreover, the G@Cu-CuNWs anode exhibited much better cycling stability than that of random graphite anode (fig. S5D).

We further tested the charging rate performance of the thick G@Cu-CuNWs and random graphite electrode in the symmetric cell using the lithiated graphite as the counter electrode to eliminate the influence of the large overpotential of Li metal foil at high current density (details in Materials and Methods). The charging voltage profiles (fig. S5E) show that the G@Cu-CuNWs electrode exhibited much lower charging voltage plateau and higher state of charge (SOC) at the galvanostatic charge stage than the random graphite electrode. The rate cycling test (fig. S5F) further confirmed improved SOC of the G@Cu-CuNWs electrode at the galvanostatic stage (76 and 48% at 0.5 C and 1 C, respectively) compared to the graphite electrode (51 and 32% at 0.5 C and 1 C, respectively, C-rate means the charging current density to achieve the theoretical capacity of the electrode in 1/C hour). These results indicate that the dual-gradient structured G@Cu-CuNWs has low polarization and fast-charging capability.

To confirm the efficiency of dual-gradient structure for the fast charging, we applied the abovementioned particle-level analysis on thick G@Cu-CuNWs and random graphite anode to illustrate the mechanism of dual-gradient structure for improving the fast-charging capability. The 3D reconstruction structure of the fabricated thick electrode was obtained via laboratory-based x-ray tomography (details in Materials and Methods) (). A voxel (50 μm by 50 μm by 150 μm), including the information of particle size and particle position, was extracted from the reconstructed 3D model of the G@Cu-CuNWs and graphite electrode (fig. S6), which served as the inputted configuration in the particle level model (Fig. 2E). As shown in fig. S7A, the porosity is approximately spatially linear-graded in the G@Cu-CuNWs electrode and uniformly distributed in the graphite electrode. Meanwhile, the overall porosity of the two samples is similar (33.4% for the G@Cu-CuNWs and 34.2% for the graphite). By applying a same current density of 5 mA cm−2 on the top of both electrodes, the Li+ ion distributions (Fig. 2E and fig. S7B) from the top to the bottom in two electrodes revealed much more homogeneous Li+ ion distribution in the G@Cu-CuNWs electrode, demonstrating that the dual-gradient structure largely enhanced the Li+ ion mass transport in the electrode, alleviating the concentration polarization. Meanwhile, the G@Cu-CuNWs anode displayed much higher utilization of active materials in comparison to random graphite anode (Fig. 2F). As a result, the G@Cu-CuNWs electrode showed lower voltage polarization and higher SOC compared to that of the graphite electrode at the cutoff voltage (Fig. 2G), indicating the improved charging capability of the G@Cu-CuNWs anode.

Ultrafast-charging performance of dual-gradient structured electrode

To show the practical application of the dual-gradient structured electrode, we prepared a relatively thin G@Cu-CuNWs electrode (~70 μm) and evaluated the electrochemical performance with an areal capacity of ~3 mAh cm−2. In the symmetric cell using the lithiated graphite as the counter electrode, the G@Cu-CuNWs electrode displayed much lower charging voltage plateau and higher SOC at the galvanostatic charge stage than the random graphite electrode (Fig. 3A). The rate performance of the G@Cu-CuNWs electrode is better than that of the random graphite electrode as well (Fig. 3B). Notably, the SOC of the G@Cu-CuNWs electrode at the galvanostatic charge stage can approximately reach to 60% at 6 C, implying that the G@Cu-CuNWs electrode can achieve 60% of recharge in 10 min (Fig. 3C). In addition, the G@Cu-CuNWs electrode also exhibited higher charge capacities, lower overpotential, higher state of galvanostatic charge, and more stable cycling at 1 C than the random graphite electrode in the half cells using Li foils as the counter electrodes (fig. S8). The comparison of charging performance with the G@Cu-CuNWs electrode with the other gradient structure design (no gradient and opposite gradient) confirmed that the improved charging properties is mainly due to the novel structure rather than its physical characteristics (fig. S9).

Fig. 3.

The fast-charging performance evaluation of the G@Cu-CuNWs anode for practical full cells.

(A to C) Fast-charging performance tests in the symmetric cells using lithiated graphite electrodes. (A) Charge voltage versus SOC plots with increasing the charging rate from 1 to 6 C (1 C = 370 mA g−1). (B) SOC at galvanostatic potentiostatic stages at different rates versus cycle number. (C) Comparison of SOC at galvanostatic stage at various C-rate, indicating that the G@Cu-CuNWs electrode can realize approximately 60% SOC with the galvanostatic charging at 6 C. (D to F) Fast-charging performance test in the full cell coupled with the LCO cathode. (D and E) Typical charge/discharge voltage profiles of the full cells at the charging rate of 3 C (D) and 6 C (E), respectively. All the discharge processes were conducted at the constant rate of 0.2 C. (F) Required charging time for the full cell to reach a given SOC of 60 and 80%.

To achieve extremely fast-charging LIBs, we fabricated the full cell using dual-gradient G@Cu-CuNWs to couple with the LiCoO2 (LCO) cathode and tested the performance in a voltage range of 3.0 to 4.3 V. In the first activation cycle at 0.1 C, dual-gradient G@Cu-CuNWs and random graphite-based full cells both exhibited areal capacities of ~3 mAh cm−2 (fig. S10A). When the full cells were charged at 3 C (Fig. 3D), the cell using dual-gradient G@Cu-CuNWs anode displayed much lower charging overpotential and achieved the charge capacity of 1.92 mAh cm−2 at the galvanostatic charge state, much higher than the cell using the random graphite anode (0.63 mAh cm−2). At the charging rate of 6 C (Fig. 3E), the cell using dual-gradient G@Cu-CuNWs anode can achieve 0.82 mAh cm−2 charge capacity at the galvanostatic charge stage in contrast to only 0.18 mAh cm−2 achieved in the cell using random graphite anode. The dual-gradient G@Cu-CuNWs anode also showed better cycling stability at the charging rate of 1 and 3 C than the random graphite anode in the full cells (fig. S10, B and C). We further correlated the SOC to the charging time (fig. S10, D and E). Owing to the reduced polarization effect originated from as-constructed dual-gradient structure, the full cell using dual-gradient G@Cu-CuNWs anode can shorten the recharging time very much in comparison to the cell using random graphite anode (Fig. 3F). The G@Cu-CuNWs anode can achieve 60 and 80% recharge in 9.6 and 15.1 min at 3 C in comparison to 14.2 and 24.1 min of graphite anode. At the 6 C charging, 60 and 80% recharge can be even done by the dual-gradient G@Cu-CuNWs anode in 5.6 and 11.4 min, much shorter than the random graphite anode (12.4 and 23.8 min). To exclude the influence of CuNWs on the performance of the electrode, we also tested the electrochemical performance of the random structured G@Cu-CuNWs anode. The results (fig. S11) further confirmed that the obtained fast-charging capability is mainly contributed by the dual-gradient structure.

We compared the rate performance of our fabricated dual-gradient G@Cu-CuNWs anode to previously reported graphite anodes (Fig. 4A and details in table S2) (, , , ). With the increase of charging rate, the areal capacity of our dual-gradient G@Cu-CuNWs anode displayed much slower decay process in comparison to previously reported graphite anodes. Even at the rate of 6 C, the dual-gradient G@Cu-CuNWs anode can still retain the areal capacity of 1.99 mAh cm−2, indicating its extremely fast-charging capability. The fabricated dense dual-gradient electrode structure also endowed high volumetric energy density of the full cell at the high rate. As shown in Fig. 4B (detailed calculation in table S3), a 700 Wh liter−1 of energy density can still be obtained at the 6 C charge rate in the full cell using dual-gradient G@Cu-CuNWs anode in comparison to 550 Wh liter−1 of the cell using random graphite anode. We demonstrated that the dual-gradient structured graphite anode can overcome the trade-off between fast-charging capability and high-energy density of LIBs.

Fig. 4.

The comparison of charging rate performance and energy density of dual-gradient G@Cu-CuNWs with previously reported graphite anodes.

(A) Charging rate performance comparison of dual-gradient G@Cu-CuNWs anode and previously reported graphite anodes. References: Commercial graphite (), a-Si nanolayer and edge-plane–activated graphite (), magnetically aligned graphite (), graphite with staged porosity (), and KOH-etched graphite (). (B) Comparison of volumetric energy densities of full cells using different graphite anodes at different charge rates. The full cell with G@Cu-CuNWs delivered a higher volumetric energy density of 701 Wh liter−1 than that of random graphite (550 Wh liter−1) under an applied high charging rate of 6 C.

To explore the possibility of application in large pouch cell, we test the mechanical properties of G@Cu-CuNWs. Before the test, the electrode was soaked in the electrolyte for a few minutes. The tested Young’s modulus of G@Cu-CuNWs (215.2 MP) is approximately equal to that of graphite electrode without Cu foil (189.6 MP) (fig. S12). This could limit the assemble and test process in coin cell. However, the mechanical is far less than the graphite with Cu foil current collector. In addition, an anode tab welding could not be obtained due to the lack of current collector, which also would limit the practical application of G@Cu-CuNWs. We hold opinion that developing an electrode without tab in the pouch cell will be an effective strategy for the application of the G@CuNWs electrode in the future.

DISCUSSION

In summary, we reported a particle-level theoretical model to simultaneously design the distribution of particle size and electrode porosity for the fast-charging of the graphite anode. We designed the dual-gradient electrode, which has smaller particles and more porosity on the top and bigger particles and less porosity on the bottom. It can further enhance mass transport, substantially improve the rate performance of the electrode, and thus retain the much higher active material utilization than that in the structure without such gradient especially at high charging current densities. According to the theoretical analysis, we developed a polymer binder–free slurry technique to fabricate a particle size electrode porosity dual-gradient G@Cu-CuNWs anode to experimentally verify the dual-gradient structure relation to the fast-charging performance. The constructed dual-gradient structure can facilitate the Li+ ion mass transport in the liquid electrolyte phase leading to the fast-charging capability in both the thick G@Cu-CuNWs anode (≥9 mAh cm−2) and the industrial level G@Cu-CuNWs anode (~3 mAh cm−2). The full cells using G@Cu-CuNWs anodes exhibited both high volumetric energy density and fast-charging capability. Importantly, the full cell can achieve 60% recharge in 5.6 min and 80% recharge in 11.4 min at 6 C. Our proposed dual-gradient structured electrode design will open up new opportunity to overcome the trade-off between the energy density and fast-charging of lithium ion batteries.

MATERIALS AND METHODS

Fabrication of dual-gradient G@Cu-CuNWs anode

Synthesis of the CuNWs and G@Cu particles

CuNWs were synthesized by a chemical reduction of Cu2+ in a solution reaction according to previously reported method (). Mixed aqueous solution of NaOH (670 ml, 15 M) [Sinopharm Chemical Reagent Co. Ltd. (SCRC), AR], Cu(NO3)2 (20 ml, 0.2 M) (SCRC, AR), ethylenediamine (10 ml) (SCRC, AR), and hydrazine (342 μl, 85 weight %) (SCRC, AR) was heated at 80°C under stirring at 200 rounds per minute (rpm) for 80 min in a 500 ml of round-bottom flask. The CuNWs were collected and washed by deionized water and ethanol each three times by centrifugation at 4500 rpm for 5 min. The CuNWs were stored in 20 ml of ethanol in a refrigerator (4°C) to minimize the oxidation. The concentration of the obtained CuNWs suspension approached to 16 mg ml−1. G@Cu particles were synthesized by copper electroless deposition on the graphite particles (SAG-H, BTR Co. Ltd). Before the copper electroless deposition, the surface of the graphite particles was modified with Pd catalyst. First, the graphite particles were oxidized by chromic acid aqueous solution of potassium dichromate (0.38 M; SCRC, AR) and sulfuric acid (4.5 M; SCRC, 85%) for 30 min. After the removal of residual acid solution by filtration, 3 g of oxidized graphite particles was added into 50 ml of SnCl2 (0.2 M; SCRC, AR) hydrochloric acid (0.1 M; SCRC, ~30%) aqueous solution under stirring for 30 min at 35°C. Then, the obtained graphite powder was washed by deionized water three times by centrifugation at 8000 rpm for 3 min to remove the excess Sn2+. Next, the Sn2+-modified graphite powder was added into 200 ml of PdCl2 (20 mg; SCRC, AR) and hydrochloric acid (0.15 M; SCRC, ~30%) aqueous solution under stirring for the chemical reduction of Pd2+ on the graphite particle surface. The obtained Pd-modified powder was washed by deionized water three times by centrifugation at 8000 rpm for 3 min and dried at 80°C for the following copper electroless plating. The Cu2+ plating solution was prepared by adding CuSO4·5H2O (0.446 g; SCRC, AR), NaKC4H4O6·4H2O (0.533 g; SCRC, AR), and Na2(EDTA) (0.533 g; Alfa Aesar, AR) in 50 ml of deionized water. Two grams of Pd-modified graphite powder was dispersed in the plating solution to form a homogeneous suspension by vigorous magnetic stirring. Subsequently, 50 ml of NaOH (0.4 g; SCRC, AR) aqueous solution with 1.5 ml of formaldehyde solution (SCRC, AR) was added, and the mixture was stirred for 1 hour to guarantee the copper reduction deposition. The obtained G@Cu powder was collected, washed by deionized water, and dried in a vacuum oven at 60°C overnight for the following use.

Fabrication of dual-gradient G@Cu-CuNWs anode

A total of 400 mg of as-obtained G@Cu powder and 70.6 mg of CuNWs (4.41 ml) were mixed and diluted to 23 ml using ethanol. Then, the G@Cu and CuNWs ethanol suspension was poured into a 4 cm by 10 cm by 10 cm stainless steel mold and let the ethanol evaporate under ambient conditions. During the ethanol evaporation process, the G@Cu particles and CuNWs sequentially assembled with each other together induced by their different sedimentation rates. The obtained composite membrane was then annealed under the H2/Ar (5%/95%) flow at 450°C for 4 hours at a heating rate of 2°C min−1. The annealed membrane was calendered (MSK-HRP-01, MTI) to form a dense gradient G@Cu-CuNWs anode with the density of 1.4 to 1.5 g cm−3. The thick G@Cu-CuNWs electrode with the graphite mass loading of ~30 mg cm−2 was fabricated by simply increasing the amount of G@Cu and CuNWs ethanol suspension to 69 ml while keeping the other conditions the same. The control random structure graphite and G@Cu-CuNWs anode were prepared by a traditional slurry-coating method. The graphite powder, Super P (MTI), and polymer binder (BA-288C, BO&BS) dispersed in the deionized water with a mass ratio of 94:4:2 were ground by the ball mill and then casted on the Cu foil by the doctor blade. The slurry was dried in a vacuum oven at 80°C for 12 hours and then calendered to a similar compact density with the G@Cu-CuNWs anode (1.4 to 1.5 g cm−3). In the random structured G@Cu-CuNWs electrode, the mass ratio of graphite powder, CuNWs, and polymer binder was 85:12:3.

Characterization of the CuNWs, G@Cu, and G@Cu-CuNWs

The morphologies of CuNWs, G@Cu, and G@Cu-CuNWs were observed by the SEM (JEOL-6700F). The electrode samples for the cross-sectional SEM observation were prepared using argon (Ar) ion beam at 10 keV in ion beam sampling (EMTIC 3X, Leica Company) to cut out the flat surface. To protect the cross-sectional up-surface damage from the Ar ion beam, Pt was predeposited on the surface of the electrode via magnetron sputtering (Kurt J. Lesker Lab18). The SEM images of the obtained cross-sectional surfaces were taken in a SE mode at 5 keV (SU8220, Hitachi). The elementary mapping was collected by energy disperse spectroscopy (Oxford Aztec X-Max 80) . Powder x-ray diffraction patterns (Philips X’Pert PRO SUPER) demonstrated that the Cu nanoparticles were coated on the graphite particles. The TGA measurements were performed by TG Q5000 IR to reveal the content of Cu on the G@Cu particles. The weight loss of TGA was recorded from 298 to 1073 K in heating steps of 5 K min−1 under a constant air flow. The electronic conductivity of the electrode was tested by a M-3 Mini type four-probe tester. The thickness of the electrode was measured on the cross-sectional SEM image, and the corresponding compact density was calculated via the formula ρcompact=Ma/S × L, where Ma is the mass of the graphite, S is the bottom surface area of the electrode, and L is the thickness of the electrode.

Electrochemical performance evaluation of graphite anodes

All electrochemical tests were conducted in the stainless-steel 2032 coin cells on Bio-Logic multichannel test system (VMP-3, Germany). The electrolyte was 1.0 M LiPF6 dissolved in 1:1:1 (v/v/v) ethylene carbonate/ethyl methyl carbonate/dimethyl carbonate (Duoduo Reagent, battery grade) with 2% vinylene carbonate as the additive. The microporous polypropylene separator (20 μm, Celgard) was used. All cells were assembled in an argon-filled glove box [H2O < 0.5 part per million (ppm) and O2 < 0.5 ppm]. The tests were performed at a room temperature around 25°C.

Half-cell test

In a half cell, the graphite electrode was coupled with Li metal foil (Tianqi Lithium, 99.95%) as the counter electrode. The cyclic voltammetry test was conducted with a sweep rate of 0.1 mV s−1 in a voltage rage of 0 to 1.5 V. The electrochemical impedance spectra were collected with a frequency range of 100 kHz to 0.1 Hz. To form stable solid electrolyte interface in the graphite electrodes, in the first cycle, all the cells were discharged to 5 mV at a rate of 0.05 C followed a 5-min static duration and were discharged to 5 mV repeatedly at a low-current density of 50 μA. The cells were then charged to 1 V at 0.1 C. The rate performance of graphite electrodes was investigated by a constant current and constant voltage (CC-CV) mode. The cells were discharged at a constant current to a cutoff potential of 5 mV (versus Li+/Li) followed a subsequent potentiostatic stabilization at 5 mV until the specific current dropped to below 0.02 C. The current densities in the constant current process progressively increased from 0.2 to 6 C. The ensuing charge was performed at 0.2 C. Four consecutive cycles were performed for each C-rate. The CC-CV charge mode makes the cells reproducible condition at the end of each cycle.

Symmetrical cell test

To get rid of the Li metal rate limitation, the performance of graphite electrode was tested in a symmetrical cell with lithiated graphite (LiG) as the counter electrode. To fabricate the LiG, the graphite anode was electrochemically lithiated in a coin cell and then was disassembled and washed by dimethyl carbonate solvent in the glovebox. The electrochemical lithium insertion performance of graphite electrode was evaluated by a typical CC-CV charging procedure. The cutoff voltage of constant current process was 1 V, and the limit current in constant voltage step was 0.02 C. The current densities in constant current process progressively increased from 0.2 to 6 C to determine the Li insertion rate capability. The discharge rate is constant at 0.2 C, and the cutoff voltage is set as −1 V.

Full-cell test

The full cells were assembled with LiCoO2 (LCO, BTR) as the cathodes. The voltage range in the tests is 2.5 to 4.3 V. The LCO electrode was fabricated with the active material, Super P, and polyvinylidene fluoride binder (Solef) in a mass ratio of 90.5.5. Two mass loading levels of the cathodes were ~20 and ~60 mg cm−2. The negative/positive ratio is set to 1.05.1. We tested the cycling performance via CC-CV charge and CC discharge process. The fast-charging performance was tested in a similar procedure with that in the symmetrical cell apart from the different voltage range.