
==== Front
Proc Natl Acad Sci U S A
Proc Natl Acad Sci U S A
PNAS
Proceedings of the National Academy of Sciences of the United States of America
0027-8424
1091-6490
National Academy of Sciences

38277439
202315871
10.1073/pnas.2315871121
research-articleResearch ArticlechemChemistry410
Physical Sciences
Chemistry
Design principles of heterointerfacial redox chemistry for highly reversible lithium metal anode
Lin Liang a 1
Li Jiantao jiantao.li@anl.gov
b 1 2 https://orcid.org/0000-0003-2277-849X

Zhang Yinggan a
Zheng Hongfei c
Huang Youzhang a
Zhang Chengkun a
Sa Baisheng d https://orcid.org/0000-0002-9455-7795

Wang Laisen a
Lin Jie a
Peng Dong-Liang dlpeng@xmu.edu.cn
a 2
Lu Jun junzoelu@zju.edu.cn
c e 2 https://orcid.org/0000-0003-0858-8577

Amine Khalil amine@anl.gov
b 2
Xie Qingshui xieqsh@xmu.edu.cn
a f 2 https://orcid.org/0000-0003-2105-6962

aState Key Lab for Physical Chemistry of Solid Surfaces, Fujian Key Laboratory of Surface and Interface Engineering for High Performance Materials (Xiamen University), College of Materials, Xiamen University, Xiamen 361005, China
bChemical Sciences and Engineering Division, Argonne National Laboratory, Lemont, IL 60439
cCollege of Chemical and Biological Engineering, Zhejiang University, Hangzhou 310027, China
dMultiscale Computational Materials Facility, College of Materials Science and Engineering, Fuzhou University, Fuzhou 350100, China
eQuzhou Institute of Power Battery and Grid Energy Storage, Quzhou 324003, China
fShenzhen Research Institute of Xiamen University, Shenzhen 518000, China
2To whom correspondence may be addressed. Email: jiantao.li@anl.gov, dlpeng@xmu.edu.cn, junzoelu@zju.edu.cn, amine@anl.gov, or xieqsh@xmu.edu.cn.
Edited by Peidong Yang, University of California, Berkeley, CA; received September 14, 2023; accepted December 5, 2023

1L.L. and J. Li contributed equally to this work.

26 1 2024
30 1 2024
26 7 2024
121 5 e231587112114 9 2023
05 12 2023
Copyright © 2024 the Author(s). Published by PNAS.
2024
https://creativecommons.org/licenses/by-nc-nd/4.0/ This article is distributed under Creative Commons Attribution-NonCommercial-NoDerivatives License 4.0 (CC BY-NC-ND).

Significance

The practical application of Li (lithium) metal anodes, as the candidates for next-generation Li batteries, is hindered by poor electrochemical reversibility induced by the instability of formed SEI (solid electrolyte interface) and irregular Li deposition. The heterointerface chemistry based on alloying kinetics has been designed previously to solve these issues; however, the underlying redox mechanism is still uncertain. Herein, we investigate and clarify the relationship between the redox potential of alloying heterointerfaces and the electrochemical transformation reversibility: A low redox potential favors the formation of highly binding surface after Li stripping and the maintenance of binding ability during subsequent Li plating, thus enhancing CE (Coulombic efficiency). This work guides the design of next-generation surface/interface materials for highly reversible metal batteries.

High electrochemical reversibility is required for the application of high-energy-density lithium (Li) metal batteries; however, inactive Li formation and SEI (solid electrolyte interface)-instability-induced electrolyte consumption cause low Coulombic efficiency (CE). The prior interfacial chemical designs in terms of alloying kinetics have been used to enhance the CE of Li metal anode; however, the role of its redox chemistry at heterointerfaces remains a mystery. Herein, the relationship between heterointerfacial redox chemistry and electrochemical transformation reversibility is investigated. It is demonstrated that the lower redox potential at heterointerface contributes to higher CE, and this enhancement in CE is primarily due to the regulation of redox chemistry to Li deposition behavior rather than the formation of SEI films. Low oxidation potential facilitates the formation of the surface with the highly electrochemical binding feature after Li stripping, and low reduction potential can maintain binding ability well during subsequent Li plating, both of which homogenize Li deposition and thus optimize CE. In particular, Mg hetero-metal with ultra-low redox potential enables Li metal anode with significantly improved CE (99.6%) and stable cycle life for 700 cycles at 3.0 mA cm−2. This work provides insight into the heterointerfacial design principle of next-generation negative electrodes for highly reversible metal batteries.

Li metal anode
interfacial redox chemistry
SEI formation
Li deposition behavior
coulombic efficiency
National Natural Science Foundation of China 51931006 Qingshui Xie
==== Body
pmcElectrolyte consumption and Li deactivation, yielded on surface chemistry and morphology evolution, usually induce the atrophy of electrochemical transformation reversibility, which eventually are ascribed to thermodynamic and mechanical instability of solid electrolyte interface (SEI) films (1–3). Unfortunately, as deposition/dissolution behavior negatively impacts SEI stability, previous research endeavors to enhance the SEI stability have been limited in effectiveness on cycling reversibility (1, 4–6), thereby retaining the Coulombic efficiency (CE) degradation issue (7–9).

Heterointerfacial chemistry modulation has been utilized to ameliorate deposition/dissolution behavior detrimental to the reversibility of metal batteries (10–13). The binding and interaction force with Li+ ions at the heterointerfaces, known as lithiophilicity, is regarded as the critical factor regulating Li deposition behavior (14, 15), especially at alloying-type heterointerfaces. Unfortunately, the dramatic volumetric variation of hetero-metals during electrochemical redox and the formation of the Li-containing alloy phase with weak lithiophilicity before Li nucleation is not conducive to the improvement of reversibility. The former leads to the pulverization of the hetero-metals and deposited Li, which have been broadly viewed as the root cause of CE degeneration when using complete stripping cutoff voltage (16, 17). The latter depends on the oxidation products at different cutoff stripping potentials and reduction potentials, which impact subsequent deposition behavior by changing surface interaction (18). The heterointerfacial redox process under the given cutoff voltage will affect the electrochemical transformation behavior associated with CE directly; however, these are always ignored (16). The Li/substrate redox chemistry at the heterointerface is not only involved in the SEI formation but also has a complex and comprehensive impact on the subsequent Li deposition/dissolution processes (18, 19). The relationship between the redox chemistry at heterointerfaces and the reversibility of electrochemical transformations will serve in the selection and design of heterointerfaces. Therefore, it is necessary to explore the interplay of the heterointerfacial redox chemistry with SEI formation and Li deposition behavior for maintaining the stability of the SEI film during cycling and enhancing the reversibility of Li-metal batteries.

Herein, the influence of heterointerfacial redox chemistry on SEI formation and Li deposition behavior is systemically studied. The heterointerfacial redox percentage at low potential (Q'/Q) is defined for evaluating qualitatively the difference in redox chemistry, and it is found that Q'/Q is positively correlated to CE. This manifests that the dissimilarities of electrochemical behaviors at different redox potentials play a significant role in the electrochemical reversibility of metal anode. We demonstrate that the role played by the redox chemistry on SEI formation cannot account for CE differences. Actually, the effect of heterointerfacial redox potential on Li deposition behaviors is the key to realizing high reversibility. The heterointerfacial oxidation products at specific stripping cutoff potential determine the Li/substrate binding at subsequent Li plating. Additionally, the reduction reactions of most hetero-metals and Li+ occur at medium-high potentials, preceding Li nucleation, which requires overcoming the surface energy barrier to drive the Li/substrate recombination. In contrast, the reduction at a low potential is rapidly coupled to Li deposition, avoiding secondary surface combination. Especially, the reduction reaction at an ultra-low potential takes place simultaneously with metal deposition, defined as concomitant behavior, which significantly improves the mass transfer of Li deposition. Therefore, Mg hetero-metal with ultra-low redox potential enables Li-metal batteries with high CE (99.6%) and stable cycle life for 700 cycles at 3.0 mA cm−2.

Results and Discussion

Correlation of CE and Q’/Q.

Substrates (M-Cu foams, M = Mg, Ag, In, Si, Sn, Zn, Al, and Bi) with mainstream lithiophilic hetero-metal surfaces were fabricated by sputtering different metals on Cu foams using the magnetron sputtering method and then were assembled into Li|M-Cu foam half cells for evaluating CEs. And the sputtering time is controlled to ensure the near areic capacities of heterogeneous metals. Using the traditional operating protocol with a stripping cutoff voltage of 1.0 V, the CEs of different hetero-metals are similar and low-level (SI Appendix, Fig. S1A), whose values are 97.55 ± 0.22% except for Si of 98.17%. Low CEs are attributed to dramatic volumetric variation induced by the redox reactions between hetero-metals and Li+ ions (20). In fact, controlling the variable of the areic capacity of heterogeneous metals also leads to inconsistencies in other variables, such as SEI properties, surface morphologies, potential surface oxidation layer, film thickness, specific surface area, and redox chemistry of heterointerfaces. Fortunately, the similar CEs among different hetero-metals suggest that the inconsistencies in other variables almost do not lead to CE differences, so the effect of other variables on CEs is neglected at such a stripping cutoff voltage of 1.0 V.

In contrast, CEs are diverse at different heterointerfaces by restraining stripping cutoff voltage to 0.1 V (Fig. 1 and SI Appendix, Fig. S1B). In fact, the variation of stripping cutoff voltage from 1.0 to 0.1 V solely affects the heterointerfacial redox chemistry but does not alter other factors, such as film thickness and specific surface area. The proportion of the discharge/charge capacities below a particular cutoff potential to the total discharge/charge capacities, Q'/Q, semiquantitatively defining the percentage of heterointerfacial redox below a specific potential, is used to qualitatively evaluate the role of interfacial redox chemistry on CEs. The total discharge/charge capacity (Q) is the sum of discharge capacity (QD) and charge capacity (QC). The discharge/charge capacities below the specific cutoff potential (Q') are the sum of the discharge capacity below this potential (Q'D) and the charge capacity below this potential (Q'C). When the cutoff potential is set as 0.1 V, the Q'/Q values of different heterointerfaces decline from Mg, Ag, In, Si, Sn, Zn, Al, to Bi. The highest Q'/Q value (70%) of Mg suggests that 70% of redox reactions at the Mg heterointerface proceed below 0.1 V, followed by Ag (48.17%). The difference in redox potential at different heterointerfaces is also displayed by the cyclic voltammograms (CV) curves in SI Appendix, Fig. S2, wherein Mg and Ag metals have the strongest reduction peaks of Li+ ions in the low potential range from 0.01 to 0.1 V.

Fig. 1. Correlation of CEs and Q’/Q. The relationship between Q’/Q and the average CEs of Li|M-Cu foam half cells (M = Mg, Ag, In, Si, Sn, Zn, Al, and Bi) from the 2nd to the 150th cycle restraining cutoff stripping potential to 0.1 V, at 3.0 mA cm−2 and 1.0 mAh cm−2. The insets are the voltage–capacity curves of different substrates ranging from 0.01 to 1.0 V at 0.025 mA cm−2 and the part of these curves below 0.1 V. Since a decrease in CE occurs in some half cells after 150 cycles, the average CE of the first 150 cycles is taken for comparison.

Here, a phenomenological correlation between electrochemical reversibility and the heterointerfacial redox percentage at low potential is revealed. Overall, larger Q'/Q at low potential contributes to higher average CE, which indicates that the reductive decomposition of electrolytes and/or inert Li formation are suppressed in the participation of a low-potential redox reaction. Mg and Ag metals with the highest redox percentages below 0.1 V have the highest average CEs, corresponding to 99.6% and 99.1%, respectively. However, also at the cutoff stripping potential of 0.1 V, Si, Sn, Zn, Al, and Bi metals that reserve more Li sources in the form of Li-rich alloys during the activation process fail to enhance CEs, and their average CEs still remain at a low level (<98%), due to the absence of redox reactions.

Of note, the above results do not mean that high CEs can be obtained by increasing heterointerfacial redox percentage during Li plating/stripping. By controlling the stripping cutoff potential at heterointerfaces, the redox percentage can be adjusted. When the cutoff voltages of In, Si, and Al are controlled at 0.3 V or those of In, Sn, and Zn are limited at 0.5 V, their Q'/Qs are close to that of Ag (48.17%) with a cutoff voltage of 0.1 V (SI Appendix, Fig. S3). Nevertheless, the average CEs of these In, Si, Sn, Zn, and Al hetero-metals are still at a low level (less than 97.6%), far less than the high CE of 99.1% for Ag at 0.1 V. This manifests that heterointerfacial redox reactions are only a necessary but insufficient factor to obtain high CE. Actually, the participation of the heterointerfacial redox reactions at a low potential is the key prerequisite to improving the electrochemical reversibility of the Li metal anode.

The stability of the SEI film and the Li deposition behavior are the main factors affecting the CE in the Li metal anode (2). Therefore, for clarifying the effect mechanism of the low-potential redox chemistry at heterointerfaces on CEs, it is necessary to specifically demonstrate the role of the low-potential oxidation and reduction reactions at heterointerfaces in the SEI formation and the Li deposition behavior. These will be discussed in the following.

Heterointerfacial Reduction Effects on SEI Formation.

To investigate the influence of low potential redox on SEI formation, the impedance changes and the surface composition and phase structure evolution at different heterointerfaces before and after SEI formation are analyzed. Similar discharge curves of Mg and Ag above 0.1 V to Cu (Fig. 2A) prove the SEI growth before alloying, but the chemical and physical property evolution of SEI films during the low-potential reduction is still not known. Fig. 2B shows the similar semicircle diameters of different heterointerfaces in the high-frequency region of the open-circuit Nyquist curves, indicating their similar charge transfer impedances. The SEI films form after discharging to 0.01 V, following an extra middle-frequency semicircle (Fig. 2C), whose diameters represent the ionic impedance in the SEI film (RSEI). The RSEI value of the Mg substrate is the largest, indicating that the low redox potential does not promote ion diffusion in the SEI film. Additionally, as shown in Fig. 2 D and E, when discharged to 0.3 and 0.1 V, no XRD (X-ray diffraction) peak of the Li−Ag alloy phase is observed, but the X-ray photoelectron spectroscopy (XPS) peaks of F, O, N, S, and C elements corresponding to the SEI components from conventional ether-based electrolyte could be detected at the surface. Moreover, the XPS peak intensities of F, O, N, and S elements weaken and peak intensities of Ag enhance obviously at a low detection depth (about 50 nm) of XPS (SI Appendix, Fig. S4A), suggesting that the SEI film is relatively thin at present. At 0.05 V, the alloying reaction forming LiAg (ICSD No. 247126) alloy emerges. At 0.01 V, a two-phase alloy composed of Li9Ag4 (ICSD No. 58312) and LiAg (ICSD No. 605515) is formed. But Ag 3d XPS peaks cannot be detected at both 0.05 and 0.01 V although going through surface etching (SI Appendix, Fig. S4A), indicating that SEI film thickens after alloying. When charged to 0.1 V, a two-phase structure containing the co-existed LiAg alloys is formed and then continues to dealloy into LiAg single-phase alloy after charging to 0.3 V. With further charging to 0.5 V, Ag is obtained again. A similar evolution tendency of composition and phase structure is also observed in ex situ XRD patterns and ex situ XPS spectra of the Mg substrate (SI Appendix, Fig. S4 B–D). The above results evidence that thin SEI film is initially formed prior to Li−Ag or Li−Mg alloying, and then thickens significantly during low-potential heterointerfacial reduction, which is responsible for the increased RSEI of Mg after SEI formation.

Fig. 2. The relationship between reduction potential and SEI formation. (A) Voltage–capacity curves for the first discharge process in Li|M-Cu foam half cells (M = Mg, Ag, Bi, and Cu) during activation. Nyquist curves at an open circuit (B) and after discharging to 0.01 V (C). (D) Ex situ XRD patterns of Ag during the first activation. (E) Ex situ XPS spectra of Ag during the discharge process of the first activation. (F) CEs of Li|M-Cu foam half cells (M = Mg, Ag, In, Si, Sn, Zn, Al, and Bi) measured by the Li-reservoir-testing protocol (21); the inset is the corresponding voltage–time curve.

To evaluate the effect of the thicker SEI film with higher RSEI at low-redox-potential heterointerfaces on the electrochemical reversibility, CE is measured using a Li-reservoir-testing protocol for attenuating the interference of modified substrate properties (such as redox chemistry) and amplifying the influences of SEI properties in a given electrolyte (21). Because very few parts of the Li reservoir are used for plating/stripping according to the Li-reservoir-testing protocol, the polarization of Li deposition/dissolution happens at a low potential correspondingly. Thus, almost no redox reaction between hetero-metals and Li+ ions takes place at the heterointerface. So the obtained CEs, which have little correlation with the heterointerfacial redox chemistry during Li deposition and dissolution, can be used to evaluate the effect of heterointerfacial redox potential on the SEI formation. These CEs measured by Li-reservoir-testing protocol are similar, which are 99.43 ± 0.23% except for Al of 98.96% (Fig. 2F and SI Appendix, Fig. S5). This indicates that the role of the low-potential redox reaction in SEI formation cannot significantly affect the reversibility of the electrochemical reaction. Comparing the mid-cycle voltage curves in SI Appendix, Fig. S6, it is found that different from the flat potential plateau of other metals, the polarization potential of Mg metal increases slightly with Li plating or stripping due to its own low-potential redox ability. The redox reaction at Mg heterointerfaces is partly involved in Li deposition/dissolution, which can explain its highest Li-reservoir-testing CE (99.66%) among all metals (Fig. 2F and SI Appendix, Fig. S5). Therefore, although the heterointerfacial redox effect on Li deposition behavior is weakened by reserving Li, its contribution to the electrochemical reversibility is still greater than that of SEI formation.

Heterointerfacial Oxidation Effects on Li Deposition.

The instability of the SEI film and Li deactivation induced by irregular Li deposition are detrimental to CEs, so the role played by low potential redox in the Li deposition behavior is critical to optimize the CEs. Comparing voltage–capacity curves of different hetero-metals during Li plating/stripping processes (Fig. 3A and SI Appendix, Fig. S7A), the alloying potential plateaus of hetero-metals occur at the initial discharge stage of Mg-modified and Ag-modified cells per cycle, and the significant oxidation platforms appear at the end of Li dissolution at Mg and Ag heterointerfaces. While no redox voltage plateau is observed from the voltage–capacity curves of other metals. Additionally, the dealloying voltage platform at the Ag heterointerface significantly shifts toward low potential from the 2nd to 100th cycle, which corresponds to the slight CE decline during the first 100 cycles. Differently, both the dealloying voltage platform at the Mg heterointerface and the corresponding CE stability are well maintained within the first 100 cycles. Lower CEs are exhibited at other heterointerfaces, especially Si, Sn, Zn, Al, and Bi, which do not undergo significant dealloying reactions in the low potential range. Thus, the low-potential oxidation percentage at the heterointerface during Li stripping should be directly related to the CEs. As expected, the average CEs are strictly positively correlated with the low-potential oxidation percentage (Qc’/Qc) shown in Fig. 3B, which is ascribed to the effect of Qc’/Qc on reduction percentage and the binding force between oxidation products and Li+ ions during consequent Li plating. For the former, how much alloy is de-alloyed determines how much metal can be alloyed again. For the latter, the interaction force of oxidation products with Li+ ions impacts Li nucleation. The nucleation overpotentials of the Mg, Ag, and In alloys after experiencing alloying reactions during activation are lower than those of Si, Sn, Zn, Al, and Bi alloys (Fig. 3C and SI Appendix, Fig. S7B). Especially, the dealloying product of Li-Mg alloy exhibits a low nucleation overpotential, although the formation energy of Mg with Li is the highest among all metals (22, 23). This is because the dealloying product would be altered by controlling the stripping cutoff voltage, which affects CEs resultantly. At the cutoff voltage of 0.1 V, the subsequent Li nucleation does not depend on either metallic Mg or metallic Ag as nuclei substrates but on the dealloying products of them at this cutoff voltage. Deep dealloying of alloy interfaces (such as Li-Mg, Li-Ag, and Li-Zn) to a high cutoff voltage of 1.0 V would result in large volume change and then reduce average CEs during cycling (Fig. 3 D and E and SI Appendix, Figs. S8 and S9). For high-redox-potential metals, such as Zn, the moderate degree of dealloying at a proper stripping cutoff potential of 0.3 V will cause Li to be plated on the surface of Li-poor alloy, which has higher adsorption energy to Li than Li-rich alloy (Fig. 3F) formed at a low stripping cutoff potential of 0.1 V, and meanwhile can avoid the severe volume change at a high stripping cutoff potential of 1.0 V, finally leading to the best CE (SI Appendix, Fig. S9). In conclusion, when the cutoff voltage is set as 0.1 V, the Li is deposited on Li-poor alloys of Mg and Ag with high adsorption energy, while on Li-rich alloys of other metals (Si, Sn, Zn, Al, and Bi) with low adsorption energy, which accounts for the dissimilarity of CEs at a cutoff voltage of 0.1 V at different heterointerfaces (Fig. 3G).

Fig. 3. The role of oxidation potential during Li stripping. (A) Voltage–capacity curves in Li|M-Cu foam half cells (M = Mg, Ag, Bi, and Cu) at the 2nd, 5th, 10th, 50th, and 100th cycles. (B) The relationship between Qc’/Qc and the average CEs of Li|M-Cu foam half cells (M = Mg, Ag, In, Si, Sn, Zn, Al, and Bi) from the 2nd to the 150th cycles restraining cutoff stripping voltage to 0.1 V. (C) The nucleation overpotential at the first cycle after activation. CE versus cycle number (D) and the average CEs (E) of Mg when stripping to different cutoff potentials. (F) The adsorption energies of metal, Li-poor alloy, and Li-rich alloy of Ag (Ag, LiAg, and Li9Ag4) and Mg (Mg, LiMg3, and LiMg). (G) Schematic illustration regarding oxidation products at the heterointerface with high or low redox potentials and different cutoff voltages.

Heterointerfacial Reduction Effects on Li Deposition.

The Li deposition behaviors are distinguished at the heterointerfaces with different redox potentials (Fig. 4A). The high-redox-potential heterointerfaces represented by the Bi metals first undergo an alloying reaction above 0.75 V to form Li-rich alloy (LiRM), and then Li nucleates below 0 V driven by low adsorption energy, which heightens the Li nucleation barrier. Differently, the alloying potential plateau of low-redox-potential Ag metal heterointerface approaches Li plating potential. With the current density increasing, the alloying potential plateau coincides gradually with that of Li deposition due to electrochemical polarization (Fig. 4B). The low-potential heterointerfacial reduction is rapidly coupled to Li nucleation, avoiding secondary surface combination, defined as continuous behavior. However, when the heterointerfaces are not completely alloyed at 0.4 or 3.0 mA cm−2 with a capacity of 0.5 mAh cm−2 (Fig. 4C and SI Appendix, Fig. S10A), no Li is detected at the h-Ag heterointerface in the XRD patterns, which suggests that the Ag alloying precedes Li deposition. In contrast, the XRD peak of Li metal (PDF#150401) at 2θ = 36.18° occurs at the h-Mg heterointerface, which evidences that the heterointerfacial reduction and Li deposition proceed synchronously at the Mg heterointerface with ultra-low redox potential, defined as concomitant behavior. This phenomenon can also be detected at deposited Li after activation of Mg heterointerface (SI Appendix, Fig. S10 B and C). The finding allows the interface to maintain the unsaturated alloying state throughout the Li plating process, rather than transforming into a Li-rich alloy state before or during Li nucleation, which keeps high interaction forces with Li+ ions during the Li nucleation and the charge transfer process, and thus enhances CEs.

Fig. 4. The role of heterointerfacial reduction potential on Li deposition behavior. (A) Voltage–capacity curves in Li|M-Cu foam half cells (M = Mg, Ag, and Bi) at initial Li deposition at 0.4 mA cm−2. Insets show different Li deposition behavior at the heterointerface with high, low, or ultra-low redox potentials. (B) Voltage–capacity curves of Ag from 0.5, 1.0, 3.0, 5.0, 8.0, to 12 mA cm−2 with a deposition capacity of 1.0 mAh cm−2 and a cutoff voltage of 0.1 V. (C) XRD patterns of Li|M-Cu foam half cells (h-Mg and h-Ag) at initial Li deposition at 0.4 and 3.0 mA cm−2 with a Li deposition capacity of 0.5 mAh cm−2. The substrates with a thicker modified layer (0.8 to 0.9 mAh cm−2), named as h-Mg and h-Ag, are used to enhance the detection signals.

The differences in voltage–capacity curves of Mg, Ag, Bi, and Cu heterointerfaces at different current densities with a cutoff voltage of 1.0 V are shown in SI Appendix, Fig. S11, wherein large nucleation overpotentials are observed at the Cu and Bi heterointerfaces without redox voltage platforms due to a secondary surface combination between Li-rich alloy and Li+ ions. Differently, there are almost no nucleation overpotentials at the Ag heterointerface, and as current density increases, the clear reduction voltage platforms at the low potential overlap gradually with the Li deposition voltage platform. A similar voltage platform merging phenomenon also occurs at the Mg heterointerface. Its reduction plateau at ultra-low potential and the Li deposition plateau are indistinguishable when the current density exceeds 3.0 mA cm−2 (SI Appendix, Fig. S11A). Additionally, the dealloying plateaus of Li-Mg and Li-Ag alloys are more stable than those of Li-Bi alloys. The dealloying potential plateaus of Li-Mg alloy with a certain capacity contribution confirm that the reduction reactions have occurred although their voltage platforms cannot be identified (SI Appendix, Fig. S11B). Obviously, the polarization at a high current density promotes concomitant behavior of Mg substrate to some extent, which can account for high CEs at a high current density. Besides, at 1.0 mA cm−2, the alloying potential plateau at Ag heterointerface when stripping Li to 0.1 V (Fig. 4B) is lower than that at 1.0 V (SI Appendix, Fig. S11A). This indicates that moderate oxidation products by simply adjusting the stripping cutoff voltage can promote concomitant behavior, which further explains the huge difference in CEs at Mg heterointerface at different cutoff voltages.

Significant differences in Li deposition behaviors at Mg, Ag, and Cu heterointerfaces are reflected by the morphological differences after undergoing different deposition/dissolution and cycling. Li grains aggregate locally and then dendritic Li grows laterally on the Cu surface (Fig. 5 A–E) because no alloying/dealloying reaction with Li occurs at the Cu interface. Differently, the flat morphologies of the plated/stripped or cycled Li at Ag heterointerface are dominated by densely distributed Li grains (Fig. 5 F–J), which benefits from the continuous behavior between reduction potential and Li deposition voltage. Further surface densification of Li morphologies is observed at the Mg heterointerface during the Li deposition/dissolution process, even after cycling (Fig. 5 K–O). This confirms that concomitant behavior at ultra-low reduction potential can homogenize Li deposition.

Fig. 5. Li deposition/dissolution morphological evolution. The SEM images of Li deposition and dissolution at Cu (A–E), Ag (F–J), and Mg (K–O) heterointerfaces at 1.0 mA cm−2: After plating for 1.0 (A, F, and K) and 3.0 mAh cm−2 (B, G, and L), and continuing to strip to 0.1 V (C, H, and M), and then plating again for 1.0 mAh cm−2 (G, I, and N); After 50 cycles at 1.0 mA cm−2 and 1.0 mAh cm−2 and then plating for 1.0 mAh cm−2 (E, J, and O).

Electrochemical Performance Validation.

Half cells are assembled to validate the effects of the different Li deposition behaviors induced by heterointerfacial redox chemistry on the reversibility and cyclability of Li metal anode. Fig. 6A shows the CE versus cycle number in Li|M-Cu foam half cells (M = Mg, h-Ag, Ag, and Cu) at a fixed deposition capacity of 1.0 mAh cm−2 and varying current densities of 3.0, 5.0, and 8.0 mA cm−2. Remarkable long cycle life of Mg-modified cells is exhibited (700, 350, and 300 cycles, respectively) with high CEs. The average CEs reach 99.07% after 700 cycles at 3.0 mA cm−2 and 99.24% after 350 cycles at 5.0 mA cm−2 (Fig. 6A and SI Appendix, Fig. S12A). This cyclability is superior to those of h-Ag cells and Ag cells, while Cu cells short out quite early. The h-Ag-modified cells are more stable than Ag-modified cells due to their higher Li reservoir during activation, where the CEs of h-Ag-modified cells maintain over 98.5% within 450 cycles at 3.0 mA cm−2. For parallel comparisons, the average CEs is calculated within 150 cycles, where almost all cells at different testing condition are stable. As shown in Fig. 6B, at 3.0 mA cm−2, the average CE of Mg is the highest (99.60%) than h-Ag (99.04%) and Cu (97.95%). At larger current densities of 5.0 and 8.0 mA cm−2, average CEs of Cu cells degrade severely to 97.43% and 96.13%, respectively, and those of h-Ag also decline but still keep at 98.91% and 98.54%. Differently, the high-level CEs of 99.32% and 99.10% are exhibited in Mg-modified cells at such high current densities, evidencing its reversibility advantage of Mg. Recently reported literature prided on CE, cycle life, and current density with an areal capacity of 1.0 mAh cm−2 are summarized (Fig. 6E) for validating superior performances of Mg-modified cells with ultra-low redox potentials (24–35).

Fig. 6. Electrochemical performance of the constructed half cells with different heterointerfaces. (A) CE versus cycle number in Li|M-Cu foam half cells (M = Mg, h-Ag, Ag, and Cu) for 1.0 mAh cm−2 at different current densities with a cutoff voltage of 0.1 V. The corresponding average CEs from the 2nd to the 150th cycle (B), nucleation overpotentials (C), and average voltage polarizations during the first 100 cycles (D) of Li|M-Cu foam half cells (M = Mg, h-Ag, and Cu). (E) The comparison of CEs, cycle number, and current density with previous works reported in the literature (24–35).

The dependence of interfacial nucleation and charge transfer process on redox potentials is verified by comparing the nucleation overpotentials and polarization voltage, respectively, at 3.0, 5.0, and 8.0 mA cm−2 (Fig. 6C and SI Appendix, Fig. S12B). In general, the nucleation overpotentials of h-Ag are relatively low, as low as 7.8 mV at 3.0 mA cm−2, which is attributed to the strong binding of Li-poor alloying state after Li striping and the fast coupling of the Ag alloying reaction to the Li nucleation without secondary surface combination during Li plating. The Mg heterointerfacial alloying synchronizes with Li deposition, even Li nucleation precedes Mg alloying, and the binding of Li-poor alloy of Mg is lower than that of Ag, so nucleation overpotentials of Mg are generally higher than that of h-Ag, but much lower than that of Cu. However, the polarization voltage of Mg, calculated by the average discharging terminal voltage within the first 100 cycles, is relatively lower than those of h-Ag and Cu (Fig. 6D and SI Appendix, Fig. S12C). The Ag is completely alloyed into Li-rich alloy at the initial stage of the Li deposition so that its alloying reaction cannot participate in the charge transfer process after Li nucleation. On the contrary, due to the concomitant behavior, the Li-Mg alloying proceeds companying with charge transfer during Li plating, which maintains strong surface absorption and thus decreases the driving force of charge transfer.

Proof-of-concept anode-free full cells were constructed using LiFePO4 (LFP) as cathode and the hetero-metal-modified Cu foam after activation as the current collector for evaluating practical potentiality. The cycling performance in SI Appendix, Fig. S13 A–C shows that all active Li metal can be stripped in Cu foam|LFP cell at the cutoff voltage of 3.0 V, and the capacity retention after 100 cycles is as low as 9%. At the voltage range of 3.0 to 3.8 V, the Ag, Mg, and h-Ag modified anode-free full cells almost complete the entire oxidation process, and their retentions after 100 cycles are 23.9%, 39.3%, and 37.9%, respectively. The heterointerfacial oxidation is limited by controlling the discharge cutoff voltage at 3.2 V, which is between the dealloying potentials of Li-Ag or Li-Mg alloys and facilitates the formation of Li-poor alloy with superior binding energy toward Li while avoiding severe structural collapse induced by complete dealloying. As a result, the retentions in Ag, Mg, and h-Ag modified anode-free cells are heightened to 33.3%, 44.3%, and 42.2%, respectively. The average CE of Cu foam|LFP cell is only 97.3%, while that of Mg modified cell is the highest (98.9%) at 3.2 V, significantly higher than those of Ag and h-Ag modified anode-free full cells (SI Appendix, Fig. S13D). Therefore, the specific capacity of a Mg-modified cell is higher than that of Ag when Mg and Ag loading capacities are similar, but lower than that of h-Ag until approaching 100 cycles because of the larger reserved Li in the h-Ag interlayer during activation. As a result, high reversibility and cyclability can be achieved at the Mg heterointerface with ultra-low redox potential.

Conclusion.

In summary, we found that the CE during Li plating/stripping is significantly influenced by heterointerfacial redox percentage at low potential. The effect of heterointerfacial redox chemistry on the reversibility of electrochemical transformation is clarified. Results manifest that what accounts for CE enhancement mainly is the role of redox chemistry in Li deposition behavior rather than SEI formation. The heterointerfacial oxidation products at stripping cutoff potential alter the heterointerfacial binding to Li and the reduction percentage during subsequent Li plating. And heterointerfacial reduction potential influences the sequence between reduction reaction and Li deposition, further leading to different Li deposition behavior. The continuous behavior and the concomitant behavior between (ultra-)low-potential reduction and Li deposition avoid the secondary surface combination, further homogenizing Li deposition and then improving the reversibility and cyclability of half cells and anode-free full cells. The heterointerfacial redox potential is determined by the electrochemical reaction mechanism between hetero-metals and Li+ ions, such as solid solution reaction, which may be used to explain the redox potential-CE correlation. The correlation of underlying reaction mechanism, redox potential, and CE will be a topic worthy of further study. The relationship between heterointerfacial redox chemistry and the reversibility of electrochemical behavior identified here will provide opportunities to design next-generation surface/interface materials for highly reversible metal batteries.

Materials and Methods

Preparation of Hetero-Metal-Modified Cu Foams (M-Cu Foam, M = Mg, Ag, In, Si, Sn, Zn, Al, and Bi).

The substrates are synthesized by depositing different hetero-metals on both two sides of Cu foams using a magnetron sputtering system. Commercial 3D Cu foams are used as current collectors to avoid the occurrence of localized current density in 2D materials such as Cu foils. The hetero-metals are ejected from different metal targets with a diameter of 3 in. (Zhongnuoxincai, Beijing Technology Co., Ltd.). First, the chamber is filled with argon gas after being evacuated to a high vacuum of 6.0 × 10−4 Pa to remove gas and moisture. Then, the metal-modified layers were sputtered at different barometric pressures and different powers, where the working angle between the target surface normal and the surface of the rotating substrate is 35°. Thin modified layers with near discharge capacities were obtained by controlling mass loading (SI Appendix, Table S1) via adjusting sputtering power and sputtering time. Besides, thick Mg and Ag-modified layers are prepared and named as h-Mg and h-Ag, respectively. The XRD patterns and surface morphologies of the prepared different M-Cu foams (M = Mg, Ag, In, Si, Sn, Zn, Al, and Bi) are shown in SI Appendix, Figs. S14 and S15.

Electrochemical Testing.

Li|M-Cu foam half cells.

The half cells are constructed for evaluating CE, where the anode is Li chip (diameter of 16 mm), the cathode is M-Cu foam (diameter of 12 mm), the separator is Celgard 2500, and the electrolyte is 60 µL of LiTFSI (1 M) electrolyte in a mixed solvent of DOL and DME (1:1 by volume) with 2 wt% of LiNO3 additive. Galvanostatic tests were carried out using the Neware battery tester (CT-4008-5 V 10 mA and 5 V 50 mA). Before cycling, activation is conducted in half cells at 0.025 mA cm−2 for three cycles to form a stable SEI layer, with the same cutoff voltage as the Li dissolution process. The average CEs do not incorporate the CE in the first cycle (SEI formation process), where we focused on the CE during Li plating/stripping after SEI formation. The CE about SEI formation is obtained based on a Li-reservoir-testing protocol. CV was conducted between 3.0 and 0.01 V versus Li+/Li using an Autolab electrochemical workstation (NOVA 1.9) at a scanning rate of 0.05 mV s−1 at room temperature. Electrochemical impedance spectroscopy is conducted with a frequency (0.1 MHz to 10 mHz) and a perturbation amplitude of 5 mV.

Li-testing-reservoir protocol.

The protocol is conducted with a current density of 0.4 mA cm−2 where SEI is formed and 4 mAh cm−2 of Li is plated before fully stripped to 1.0 V. And then Li reservoir (4 mAh cm−2) is deposited on the substrates, 1/8 (0.5 mAh cm−2) of which is stripped/plated for 100 cycles. Finally, all active Li is stripped fully to 1.0 V. Li-reservoir-testing CE is obtained from the quotient of the charge capacity from the 2nd to the 102nd cycle and the corresponding total discharge capacity.

Anode-free LiFePO4 (LFP) full cell test.

The full cells contain M-Cu foam (diameter of 12 mm) as the anode, an LFP electrode (diameter of 12 mm) as the cathode, a separator (Celgard 2500), and 40 µL of electrolyte corresponding to 2 M LiFSI and 2 M LiTFSI dissolved in DOL:DME (1:1 by volume) with 2 wt% LiNO3 additive. A slurry containing LFP powders (mass loading of roughly 7.45 mg cm−2), acetylene black, and poly(vinylidene difluoride) with a weight ratio of 9:0.5:0.5 is coated onto Al foil to prepare LFP electrodes. Activation of full cells is conducted from 3.0 to 3.8 V at a rate of 0.2 C (1 C = 170 mA g−1) in the initial cycle. The discharge cutoff voltages of 3.0 and 3.2 V are used for controlling the oxidation percentage in the subsequent cycles.

Characterizations.

The morphology evolutions of Li metal at different heterointerfaces were monitored by scanning electron microscopy (SU-70) with energy-dispersive X-ray spectroscopy, and the compositions evolutions of SEI and metal substrates at heterointerfaces during SEI formation were tracked by ex situ X-ray photoelectron spectrometer (XPS, Thermo Scientific ESCALAB Xi+). The crystallographic structures of the as-prepared M-Cu foams and the Mg-modified and Ag-modified Cu foams after Li plating were investigated by Ultima IV XRD with Cu Kα radiation. Besides, the phase structure evolutions of hetero-metals were monitored by the ex situ XRD.

Calculation.

The adsorption energies for Li atoms at the surfaces of Ag, LiAg, Li9Ag4, Mg, LiMg3, and LiMg were calculated using the Vienna Ab-initio Simulation Package code. The treatments of calculation models and results were conducted on the ALKEMIE platform (36). The computational details can reference our previous work (37).

Supplementary Material

Appendix 01 (PDF)

Click here for additional data file.

The work was supported by financial support from the National Natural Science Foundation of China (Grant Nos. U22A20118, 51931006 and 52272240), the Fundamental Research Funds for the Central Universities of China (Xiamen University: No. 20720220074), Guangdong Basic and Applied Basic Research Foundation (No. 2021A1515010139), Science and Technology Projects of Innovation Laboratory for Sciences and Technologies of Energy Materials of Fujian Province (HRTP-[2022]-22) and the “Double-First Class” Foundation of Materials Intelligent Manufacturing Discipline of Xiamen University.

Author contributions

L.L., J. Li, D.-L.P., J. Lu, K.A., and Q.X. designed research; L.L., Y.Z., H.Z., Y.H., and J. Li performed research; H.Z., Y.H., C.Z., B.S., L.W., J. Lin, D.-L.P., J. Lu, K.A., and Q.X. analyzed data; and L.L., J. Li, Y.Z., and Q.X. wrote the paper.

Competing interests

The authors declare no competing interest.

Data, Materials, and Software Availability

All study data are included in the article and/or SI Appendix.

Supporting Information

This article is a PNAS Direct Submission.
==== Refs
1 Y. Liu , Self-assembled monolayers direct a LiF-rich interphase toward long-life lithium metal batteries. Science 375 , 739–745 (2022).35175797
2 S. Ko , Electrode potential influences the reversibility of lithium-metal anodes. Nat. Energy 7 , 1217–1224 (2022).
3 C. Fang , Quantifying inactive lithium in lithium metal batteries. Nature 572 , 511–515 (2019).31435056
4 X. He , The passivity of lithium electrodes in liquid electrolytes for secondary batteries. Nat. Rev. Mater. 6 , 1036–1052 (2021).
5 J. M. Fu , Superionic conducting halide frameworks enabled by interface-bonded halides. J. Am. Chem. Soc. 154 , 2183–2194 (2022).
6 Q. Wang , Leap of Li Metal anodes from coin cells to pouch cells: Challenges and progress. Electrochem. Energy Rev. 6 , 22 (2023).
7 J. Xiao , Understanding and applying coulombic efficiency in lithium metal batteries. Nat. Energy 5 , 561–568 (2020).
8 J. Zheng , Design principles for heterointerfacial alloying kinetics at metallic anodes in rechargeable batteries. Sci. Adv. 8 , eabq6321 (2022).36332032
9 C. H. Wang , Identifying soft breakdown in all-solid-state lithium battery. Joule 6 , 1770–1781 (2022).
10 L. Lin , Challenge and strategies in room temperature sodium-sulfur batteries: A comparison with lithium-sulfur batteries. Small 18 , 2107368 (2022).
11 D. Wang , Towards high-safe lithium metal anodes: suppressing lithium dendrites via tuning surface energy. Adv. Sci. 4 , 1600168 (2017).
12 Y. P. Sun , Regulated lithium plating and stripping by a nano-scale gradient inorganic-organic coating for stable lithium metal anodes. Energy Environ. Sci. 14 , 4085–4094 (2021).
13 X. F. Yang, X. Li, K. Adair, H. M. Zhang, X. L. Sun, Structural Design of Lithium-Sulfur Batteries: From Fundamental Research to Practical Application. Electrochem. Energy Rev. 1 , 239–293 (2018).
14 K. Yan , Selective deposition and stable encapsulation of lithium through heterogeneous seeded growth. Nat. Energy 1 , 1–8 (2016).
15 B. T. Heligman, A. Manthiram, Elemental foil anodes for lithium-ion batteries. ACS Energy Lett. 6 , 2666–2672 (2021).
16 S. Jin , Solid-solution-based metal alloy phase for highly reversible lithium metal anode. J. Am. Chem. Soc. 142 , 8818–8826 (2020).32310653
17 L. Lin , Adjustable mixed conductive interphase for dendrite-free lithium metal batteries. ACS Nano 16 , 13101–13110 (2022).35946592
18 F. Xu , Atomic Sn-enabled high-utilization, large-capacity, and long-life Na anode. Sci. Adv. 8 , eabm7489 (2022).35544572
19 Y. Jiang , Seamless alloying stabilizes solid-electrolyte interphase for highly reversible lithium metal anode. Cell Rep. Phys. Sci. 3 , 100785 (2022).
20 S. Tang , Stable Na plating and stripping electrochemistry promoted by in situ construction of an alloy-based sodiophilic interphase. Adv. Mater. 31 , 1807495 (2019).
21 B. D. Adams, J. Zheng, X. Ren, W. Xu, J.-G. Zhang, Accurate determination of coulombic efficiency for lithium metal anodes and lithium metal batteries. Adv. Energy Mater. 8 , 1702097 (2018).
22 L. He, Q. Sun, L. Lu, S. Adams, Understanding and preventing dendrite growth in lithium metal batteries. ACS Appl. Mater. Interfaces 13 , 34320–34331 (2021).34275274
23 S. H. Wang , Tuning wettability of molten lithium via a chemical strategy for lithium metal anodes. Nat. Commun. 10 , 4930 (2019).31666514
24 M. Q. Wang , Tailoring lithium deposition via an SEI-Functionalized membrane derived from LiF decorated layered carbon structure. Adv. Energy Mater. 9 , 1802912 (2019).
25 X. Wang , Stress-driven lithium dendrite growth mechanism and dendrite mitigation by electroplating on soft substrates. Nat. Energy 3 , 227–235 (2018).
26 W. D. Zhang, H. L. L. Zhuang, L. Fan, L. N. Gao, Y. Y. Lu, A “cation-anion regulation” synergistic anode host for dendrite-free lithium metal batteries. Sci. Adv. 4 , eaar4410 (2018).29507888
27 Y. Jeon , Argentophilic pyridinic nitrogen for embedding lithiophilic silver nanoparticles in a three-dimensional carbon scaffold for reversible lithium plating/stripping. J. Mater. Chem. A 10 , 1768–1779 (2022).
28 Y. R. Zhang , Self-exfoliated covalent organic framework nano-mesh enabled regular distribution for stable lithium metal. Energy Storage Mater. 47 , 376–385 (2022).
29 A. Jamaluddin , Fluorinated graphene as a dual-functional anode to achieve dendrite-free and high-performance lithium metal batteries. Carbon 197 , 141–151 (2022).
30 S. Y. Zhou , Efficient diffusion of superdense lithium via atomic channels for dendrite-free lithium-metal batteries. Energy Environ. Sci. 15 , 196–205 (2022).
31 C. Guo , Uniform lithiophilic layers in 3D current collectors enable ultrastable solid electrolyte interphase for high-performance lithium metal batteries. Nano Energy 96 , 107121 (2022).
32 Y. L. An , Scalable and physical synthesis of 2D silicon from bulk layered alloy for lithium-ion batteries and lithium metal batteries. ACS Nano 13 , 13690–13701 (2019).31639296
33 S. S. Chi , Lithiophilic Zn sites in porous CuZn alloy induced uniform Li nucleation and dendrite-free Li metal deposition. Nano Lett. 20 , 2724–2732 (2020).32149520
34 T. S. Wang , Regulating uniform Li plating/stripping via dual-conductive metal-organic frameworks for high-rate lithium metal batteries. Adv. Funct. Mater. 30 , 2000786 (2020).
35 Y. P. Jiang , In situ growth of CuO submicro-sheets on optimized Cu foam to induce uniform Li deposition and stripping for stable Li metal batteries. Electrochim. Acta 339 , 135941 (2020).
36 G. J. Wang , ALKEMIE: An intelligent computational platform for accelerating materials discovery and design. Comput. Mater. Sci. 186 , 110064 (2021).
37 X. Yan , Understanding the anchoring effect on Li plating with Indium Tin oxide layer functionalized hosts for Li metal anodes. Chem. Eng. J. 440 , 135827 (2022).
