LiF Splitting Catalyzed by Dual Metal Nanodomains for an Efficient

Jan 14, 2019 - The critical challenges for fluoride conversion cathodes lie in the absence of built-in Li source, poor capacity retention, and rate pe...
0 downloads 0 Views 2MB Size
Download PDF
Subscriber access provided by Iowa State University | Library

Article

LiF Splitting Catalyzed by Dual Metal Nanodomains for an Efficient Fluoride Conversion Cathode Yu Zhao, Kaiyuan Wei, Hailong Wu, Shiping Ma, Jian Li, Yixiu Cui, Zhaohui Dong, Yanhua Cui, and Chilin Li ACS Nano, Just Accepted Manuscript • DOI: 10.1021/acsnano.8b09453 • Publication Date (Web): 14 Jan 2019 Downloaded from http://pubs.acs.org on January 16, 2019

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 30 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Nano

LiF Splitting Catalyzed by Dual Metal Nanodomains for an Efficient Fluoride Conversion Cathode Yu Zhao†1, Kaiyuan Wei†1, Hailong Wu†, Shiping Ma†, Jian Li†, Yixiu Cui†, Zhaohui Dong§, Yanhua Cui*,†, Chilin Li*,‡ †Institute

of Electronic Engineering, China Academy of Engineering Physics, Mianyang 621000, China. Email: [email protected] ‡State Key Laboratory of High Performance Ceramics and Superfine Microstructure, Shanghai Institute of Ceramics, Chinese Academy of Sciences, Shanghai 200050, China. Email: [email protected] §Shanghai Synchrotron Radiation Facility, Shanghai Advanced Research Institute, Chinese Academy of Sciences, Shanghai 201204, China 1 These authors contribute equally to this work Abstract: The critical challenges for fluoride conversion cathodes lie in the absence of builtin Li source, poor capacity retention and rate performance. For lithiated fluorides, the reason to limit their competitiveness is rooted in the facile coarsing of insulating LiF (as built-in Li source) and its insufficient splitting kinetics during charging. Previous efforts on blending LiF nanodomains with reductive metal, metal oxide or fluoride by ball-milling method still face the problems of large overpotential and low current density. Herein we propose a strategy of dual-metal (Fe-Cu) driven LiF splitting to activate the conversion reaction of fluoride cathode. This lithiated heterostructure (LiF/Fe/Cu) with compact nanodomain contact enables a substantial charge process with considerable capacity release (300 mAh g-1) and low charge overpotential. Its reversible capacity is as high as 375-400 mAh g-1 with high energy efficiency (76%), substantial pseudocapacitance contribution (>50%) and satisfactory capacity retention (at least 200 cycles). The addition of Cu nanodomains greatly catalyzes the kinetics of Fe-Cu-F formation and decomposition compared with the redox process of Fe-F, leading to the energy and power densities exceeding 1000 Wh kg-1 and 1500 W kg-1 respectively. These results indicate that LiF-driven cathode is promising as long as its intrinsic conductive network is elegantly designed.

1 ACS Paragon Plus Environment

ACS Nano 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Keywords: LiF splitting, fluoride cathode, conversion reaction, Li-ion batteries, thin film The energy storage devices dependent on rechargeable Li-ions batteries (LIBs) are preferred for electric vehicles, portable electronics and grid-level power systems.1 Intercalation-type positive materials, containing transition metal oxides or polyanion compounds with one-, twoor three-dimensional Li+ transport paths, such as olivine (LiMPO4), layered (LiMO2) and spinel (LiM2O4) materials (M = Fe, Co, Ni, Mn et. al), provide highly electrochemical reversibility but with limited capacities based on 0.5-1 electron transfer.2-4 The Li-rich intercalation cathodes of xLi2MnO3(1-x)LiMO2 enable the offer of more than one Li+, leading to a reversible capacity higher than 200 mAh g-1. However these Li-rich phases usually suffer irreversible capacity loss due to layered-to-spinel phase transformation.5 In order to further improve the energy density of Li-based batteries, fluoride materials are pushed out as conversion-type cathodes owing to their extremely large theoretical specific capacity (e.g. 713 mAh g-1 and 2196 mAh cm-3 for FeF3) and high thermodynamic voltage (e.g. 3.55 V vs. Li+/Li for CuF2).6,7 The theoretical capacity of fluoride Mn+Fn, such as FeF2, FeF3, CoF3 or CuF2, is 2-4.2 time higher than that of intercalated cathodes,8-11 based on the lithiation to M0 and LiF composition (Mn+Fn + nLi+ + ne- = nLiF + M0, n ≥ 2).12,13 The advantages of capacity and voltage of fluorides lead to a theoretically high energy density around 900 Wh kg-1 for Li-FeF3 and Li-CuF2 cells.12 The challenges of conversion cathodes comes from the absence of Li source in pristine lattices, disappointed cycling reversibility, insufficient energy efficiency and poor reaction kinetics. Recently, a new concept of Li-free transition metal monoxides (MnO, FeO and CoO) as positive electrodes had been developed by blending with nanosized LiF, which enables a high voltage working of monoxides via the formation of surface M-O-F compound (M = Mn, Fe and Co).14 The capacities of LiF-MnO, LiF-FeO and LiF-CoO nanocomposites are as high as 235, 310 and 206 mAh g-1, respectively. Notablely, only 0.9 Li participates in the surface2 ACS Paragon Plus Environment

Page 2 of 30

Page 3 of 30 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Nano

controlled reversible reaction in the LiF-MnO system. It seems that the size-effect of particle is crucial to modulate the capacity performance in these nanocomposite cathodes. On the other hand, a ternary solid-solution CuyFe1-yF2 was recently designed by substituting Cu into Fe fluoride lattices in order to improve the energy efficiency and kinetics of conversion fluoride, as well as to achieve the high reversibility of Cu redox reaction.15 The nanocomposite Cu0.5Fe0.5F2 can deliver a high capacity of 543 mAh g-1 and present the lowest voltage hysteresis for conversion reaction in any metal fluoride (150 mV for Cu redox and 200 mV for Fe redox). However, Fe0.5Cu0.5F2 cannot be coupled with the present non-lithiated anode such as graphite or silicon. Therefore it is highly desired to explore the lithiated fluoride with both higher reversible capacity and energy efficiency. Since it is difficult to generate LiF based solid-solution phase, the lithiated fluoride usually exits in the mixture form consisting of LiF (as built-in Li source) and transition metal (or its low-valence oxide or fluoride counterpart) owing to phase segregation.16,17 Unfortunately, the artifactitious LiF-M-X (M = transition metals, X = conductive and binder additives) compounds did not present satisfactory performance via conventional mechanical milling and solution phase route, because of inhomogeneous spatial distribution of LiF particles and their big size.18,19 The sluggish electrochemical splitting of LiF only resulted in a low capacity of 160-230 mAh g-1 for LiF-Fe-C nanocomposites.20 A complex “chemical” ballmilling using LiH and FeF3-C precursors enabled a more intimate contact between the components in LiF-Fe-C nanocomposite, benefiting to a higher initial discharge capacity above 400 mAh g-1.21 The size-effect and high voltage hysteresis highlight the difficulty on reversible redox process between LiF and transition metals. The LiF-MO-X and LiF-MF2-X composites enabled a lowering of charge overpotential for LiF splitting as a consequence of better lattice match between LiF and MO (or MF2).14,17 However the positively charged M compromises the theoretical capacity based on fewer electron transfer.

3 ACS Paragon Plus Environment

ACS Nano 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The thin-film architecture is a facile route to achieve homogeneous distribution of lithiated fluoride compositions and the growth of nanocrystals. Without the addition of conductive and binder additives, thin film electrode is an ideal model to investigate the intrinsic electrochemical characteristics.22 Herein, we presents the growth of LiF/Fe/Cu (LFC) nanoscale thin film electrodes by pulse laser deposition (PLD) technology to study the LiFsplitting effects and relative electrochemical characteristics. The thin-film architecture ensures the homogeneous distribution and compact contact between LiF and metal clusters or nanodomains. It therefore enables the LFC nanocomposites to achieve a highly reversible capacity of 375-400 mAh g-1 with high energy efficiency (76%) and capacity retention (at least 200 cycles). The first charge process is activated by electrochemical splitting of LiF with a considerable capacity release (300 mAh g-1) and a sloped charge curve with low overpotential from 2.7 to 4 V. These features make LFC serve as a promising conversion cathode for LIBs with substantial pseudocapacitance contribution (>50%). The combination of dual metals endows LFC with a best capacity performance among LiF-driven conversion systems.14,17,19-21,23-29

Results and Discussions The X-ray diffraction (XRD) patterns of composite target (LiF/Fe/Cu mole ratio in 2:1:1), asdeposited and annealed thin films are shown in Figure 1a. All the diffraction peaks for target could be assigned to LiF (JCPDS No. 45-1460), Fe (JCPDS No. 06-0696) and Cu (JCPDS No. 04-0836) polycrystalline phases without any impurity phase. For the as-deposited LFC thin film, the LiF and Cu/Fe peaks are not evident compared with the peaks of substrate coated by Ti and Pt layers. Ti layer serves as an interlayer to reinforce the contact between Si substrate and Pt current collector especially after Ti-Pt alloying. The diffraction peaks at 40.1, 43.7, 46.5 and 67.8 belong to Pt-Ti alloys e.g. Pt3Ti (JCPDS No. 17-0064) and PtTi3 (JCPDS No. 18-0979) labelled with hollow squares. These substrate peaks become more intensive after 4 ACS Paragon Plus Environment

Page 4 of 30

Page 5 of 30 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Nano

annealing the thin film at 200 oC. Meantime, the diffraction peaks at 38.7°, 43.3°, 44.7°, 45.0° and 65.5° become more pronounced and they are assigned to LiF(111), Cu(111), Fe(110), LiF(200) and LiF(220) respectively. XRD result confirms the existence of Fe, Cu and LiF components in thin films, agreeing with those in target. The crystallinity improvement in annealed thin film is favorable for the activation of electrochemical performance as shown later. However these peaks are still broad, indicating the preferential formation of nanocrystals grown by PLD as presented in Figure 1b. This technology promotes a sufficient contact between different phase structures, and the growth of certain phase is effectively limited owing to the squeezing by surrounding heterogeneous phases or steric hindrance. As a result, the nanodomain size in annealed LFC thin film is around 1.5-2 nm as estimated from the high-resolution transmission electron microscope (HRTEM) images in Figure 1c, d and S1. Therein the color contrast of nanodomains is clearly observed in a fine scale. The corresponding lattice stripes can be discerned and assigned to the planes of Fe (110), Cu (111) and LiF (200). These lattice stripes are in good accordance with the diffraction rings in selected area electron diffraction (SAED) pattern (Figure 1e). Scanning electron microscope (SEM) images display the surface and cross-section morphologies of LFC thin film (Figure 1f and g). It appears that the film is dense and roughly smooth with a thickness of 200 nm. Its grains are compactly stacked with narrow grain boundaries, which have the enough space for facile electrolyte penetration. The spatial distribution of Fe, Cu and F elements are extremely homogeneous

in

thin

film

from

the

scanning transmission electron microscopy-

energy dispersive spectrum (STEM-EDS) mapping (Figure S2). The homogeneity of metal elements can lead to the formation of electronically conductive pathway reaching to the current collector beneath. Therefore the electron and ion fluxes are sufficient to activate the splitting of surrounding LiF nanodomains without the serious compromise of voltage hysteresis during the first charge. The F, Fe and Cu atomic ratio in LFC thin film is 53.4:25.6:21.0 from the EDS mapping, and it is very close to the molar ratio of 2:1:1 in target. 5 ACS Paragon Plus Environment

ACS Nano 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 1. (a) XRD patterns of LiF/Fe/Cu target in mole ratio of 2:1:1, as-deposited and annealed LFC thin films. (b) Schematic of the formation of LiF/Fe/Cu nanocrystals grown by PLD. (c) Histograms of the particle size distribution for annealed LFC thin film. (d) HRTEM image and (e) corresponding SAED pattern of LFC thin film. (f) Surface and (g) cross-section SEM images of LFC thin film electrode.

The galvanostatic charge-discharge performances of as-deposited and annealed LFC thin films are compared based on the electrolyte consisting of 1 M LiClO4 in ethylene carbonate (EC) and polycarbonate (PC) solvents (v/v=1:1) at a current density of 41.6 mA g-1 in a 6 ACS Paragon Plus Environment

Page 6 of 30

Page 7 of 30 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Nano

voltage window of 1.5-4.2 V (Figure S3). After annealing treatment, both the discharge capacity and voltage are significantly enhanced in the first cycle (after first charge activation). Furthermore, the staged plateaus become more pronounced with great narrowing of voltage hysteresis between charge and discharge. These results confirm a kinetic improvement of LFC thin films after annealing in view of the reinforcement of mixed conductive network inside film as a consequence of grain boundary enrichment and metallic conductivity promotion. The splitting efficiency of LiF depends on the content of surrounding metal clusters, which can compensate for the electronic insufficiency and bond with F during the initial delithiation. Figure 2a compares the galvanostatic performance of three LFC thin films with LiF/Fe/Cu molar ratio in 1:1:1 (LFC111), 2:1:1 (LFC211) and 4:1:1 (LFC411). The splitting capacity of LiF is lowest (175 mAh g-1) in LFC111, whereas it increases to more than 300 mAh g-1 for LFC211 and LFC411. The increase of LiF fraction enables a higher utilization of metal conversion to metal fluoride. However excess LiF would negatively influence the voltage hysteresis and kinetic performance owing to its insulativity. The reversible capacity after first activation is higher than that of first charge process, indicating the potentiality of prior formation of MFx caused by plasma during thin film deposition. Its amorphous feature retards its detection by XRD and TEM. LFC211 releases the highest discharge capacity up to 420 mAh g-1, corresponding to 2.70 Li transfer based on the total weight of LiF, Cu and Fe. Compared with LFC111, LFC211 and LFC411 enable the appearance of new charge plateau above 3.6 V (during the first charge) and new discharge plateau around 3 V (during the following discharge). These high-voltage plateaus are responsible for the capacity grain in LFC211 and LFC411. The cyclic voltammetry (CV) profiles in Figure S4 show the similar electrochemical tendency as the galvanostatic curves. A couple of redox peaks with cathodic one at 2.1-2.2 V and anodic one at 2.9-3.0 V is common for the three film electrodes. The strengthening of additional cathodic peak around 3 V for LFC211 and LFC411 trigger the appearance and intensification of anodic peaks at 3.25 and 3.75 V. LFC211 shows the highest 7 ACS Paragon Plus Environment

ACS Nano 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

current density of CV peaks based on the same scanning rate of 0.2 mV s-1. It implies that LiF content would determine the depth of conversion reaction.

Figure 2. (a) Charge/discharge profiles of LFC111, LFC211 and LFC411 cathodes during the first charge and following cycle in a potential window of 1.5-4.2 V. The current density is 41.6 mA g-1. (b) Charge/discharge and (c) CV profiles of LiF/Cu(1:1), LiF/Fe(1:1) and LFC211 cathodes after the first charge. (d) CV profiles of LFC211 in various potential windows of 1.5-3.0 V, 1.5-3.5 V, 1.5-3.8 V and 1.5-4.0 V with a scan rate of 0.2 mV s-1. (e) GITT measurement for LFC211 at a pulse current of 41.6 mA g-1 for duration of 1h, followed by relaxation of 5h. Inset: voltage profiles for a single step of GITT at ∼3.17 V during discharging. (f) Diffusion coefficients estimated from GITT of LFC211 as a function of reaction potential during discharge and charge.

In order to emphasize the advantage of ternary composite in LFC211, we also investigated the performance of binary composite LiF/Cu(1:1) and LiF/Fe(1:1) thin films by galvanostatic and CV measurements (Figure 2b and c). LiF/Cu electrode shows the poor galvanostatic curve profiles during cycling. Its charge capacity locates above 3.5 V and 8 ACS Paragon Plus Environment

Page 8 of 30

Page 9 of 30 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Nano

discharge capacity of 200 mAh g-1 mainly below 2 V, leading to an extremely large voltage hysteresis. This phenomenon is not strange in view of the facile coarsing of Cu grains during film deposition.30 It would cause the poor phase contact between LiF and Cu, and increase the energy barrier of LiF splitting. The facile extrusion of Cu or its dissolution from the thin film is also responsible for the poor reversibility of Cu-based redox chemistry.31,32 For LiF/Fe electrode, the reversibility, capacity value and voltage hysteresis based on Fe-based redox reaction are remarkably improved. Its discharge process displays a feature of two-staged plateaus located at 3 and 1.75 V respectively, leading to a capacity release of 350 mAh g-1. The charge voltage of LiF/Fe electrode is much lowered compared with LiF/Cu, also with corresponding two-staged plateaus around 2.75 and 3.25 V. Interestingly, the combination of equimolar Cu and Fe metals even causes a further upgrade of capacity and polarization performances as shown in LFC211. The two-staged discharge plateaus are preserved and but prolonged (e.g. doubling of the higher-voltage one) with simultaneous voltage uplift (e.g. to 2 V for the lower-voltage plateau). The charge curve is substantially overlapped with that of LiF/Fe apart from the appearance of additional plateau around 3.7 V, which is responsible for the increase of charge capacity. Their corresponding CV curves show the similar behavior as the galvanostatic curves (Figure 2c). In LFC211, the two-staged peaks around 3V and 2V (denoted as C2 and C1 respectively) during cathodic process should be ascribed to the separated reduction of Cu2+ and Fe2+ to Cu0 and Fe0 according to the mechanism proposed by Wang et al.15 The three broad peaks located at 2.75, 3.25 and 3.7 V (denoted as A1, A2 and A3 respectively) during the anodic process reflect the blurred oxidation of Fe0 to Fe2+/Fe3+ and Cu0 to Cu2+. The stable CV curves of our LiF-Cu-Fe lithiated electrode are roughly similar to those of already reported CuyFe1-yF2 solid-solution (rather than the simple superposition of the CV curves of LiF/Cu and LiF/Fe), indicating that the in-situ electrochemical synthesis of CuyFe1-yF2 single phase from LiF-Cu-Fe mixture is feasible as long as the three phases exist in fine nanoscale and their interface contacts are intimate.15 The activation of Cu2+/Cu0 couple 9 ACS Paragon Plus Environment

ACS Nano 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

benefits from the coarsing limitation under the protocol of Cu-Fe co-deposition, which likely causes the additional effects on Cu extrusion and dissolution suppression. From the CVs of LFC211 in different potential windows (Figure 2d), we can further deduce the staged redox process. When the higher voltage is cut off to 3.5 V, the Peak A3 is absent. The intentional removal of Peak A3 leads to the shrinkage of current density for both Peak C1 and C2. This phenomenon indicates that the higher-voltage reaction is still associated with the oxidation of residual Fe and Cu to higher valance. The coupled occurrence of Fe and Cu oxidation steps was also disclosed in the case of CuyFe1-yF2.15 When further cutting off the voltage to 3 V, the Peak A2 is not involved and accordingly the Peak C2 disappears. It means that Cu0 does not substantially undergo the oxidation to Cu2+ in low-voltage range. However its (Cu0) existence reinforces the electron pathway inside the electrode, and improves the kinetics of Fe reduction step as confirmed from the intensification of current density of Peak C1. Galvanostatic intermittent titration technique (GITT) was carried out to further get insight into the thermodynamic potential hysteresis and kinetic overpotential (Figure 2e). The GITT measurement was performed at a pulse current of 41.6 mA g-1 for duration of 1h, followed by a relaxation of 5h. Note that the thermodynamic voltage hysteresis in high voltage region can be as small as ~152 mV, indicating an evident kinetic modification on Cu/Cu2+ couple. This value is close to the small potential hysteresis of 148 mV at the corresponding reaction stage reported in ternary metal fluoride CuyFe1-yF2.15 In the stage of redox couple transition from Cu/Cu2+ to Fe/Fex+, the potential hysteresis value becomes larger and can increase to ~530 mV. This is caused by the asymmetric conversion pathway between lithiation and delithiation.16,33 In the stage dominated by Fe/Fex+ redox (low voltage region), the potential hysteresis becomes smaller again (~200 mV) but with voltage relaxation of larger amplitude. It explains the intrinsically inferior kinetics for Fe/Fex+ than for Cu/Cu2+ in the dual-metal system. At the quasi-equilibrium potential, the cationic diffusion coefficient could be estimated by the Equation (1):34 10 ACS Paragon Plus Environment

Page 10 of 30

Page 11 of 30 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Nano

𝐷=

4 𝑚𝑏𝑉𝑚 2 𝛥𝐸𝑠 2 𝜋𝜏( 𝑀𝑏𝑆 ) (𝛥𝐸𝑡)

（1）

therein, D is the diffusion coefficient (cm2 s-1), τ is the intermittent time (s), mb, Mb, and Vm are the mass, molecular weight and molar volume of cathode active material respectively, S is the electrode area, 𝛥𝐸𝑠 is the difference between quasi-equilibrium potentials before and after titration, and Δ𝐸𝑡 is the difference between potentials at the beginning and termination of titration step.

𝑚 𝑏𝑉 𝑚 𝑀 𝑏𝑆

could be replaced by the thickness of thin film (cm). A typical voltage

response for single titration step at 3.17 V during the discharge process is shown in the inset of Figure 2e. The diffusion coefficients of LFC211 thin film depending on reaction voltages are plotted in Figure 2f. These D values are not bad and merely 1-2 order of magnitude lower than those of typical intercalated cathodes.35 They are however much higher (by roughly 4 order of magnitude) than those for ball-milled FeF2-C, FeF3-C and FeO0.67F1.33-C powder composites.36 The upgrade of D values in thin film architecture benefits from the compact phase contact, which would mitigate the resistance of cation migration across multiphase interfaces. The D value ranges from 10-13 to 10-14 cm2 s-1 in most the voltage zone for both the discharge and charge processes. During discharge process, the D value for high-voltage Cubased reaction (closing to 10-13 cm2 s-1) is higher than that for low-voltage Fe-based reaction (dropping to 10-14 cm2 s-1 or smaller). The non-sequential oxidation of Fe and Cu during charge leads to a middle level of D values. The contributions of faradaic charge and surface pseudocapacitance are separated by CV techonology based on various scan rates (ν) from 0.1 to 1.0 mV s-1 as shown in Figure 3a.37 The peak current density increases with the increase of scan rate. The logarithm of peak current (ip) presents a linear relationship with the logarithm of scan rate in Figure 3b, which could be expressed by the Equation (2): 22 𝑖𝑝 = 𝑎𝜈𝑏 or lg𝑖𝑝 = lg𝑎 + 𝑏lg𝜈

(2) 11 ACS Paragon Plus Environment

ACS Nano 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

therein, a and b are the adjustable parameters. The current comes from capacitance-controlled surface reaction if b = 1 and from diffusion-controlled bulk reaction if b = 0.5.38 The Peak C1 denoting Fex+ reduction to Fe0 has a high fraction of faradaic charge contribution because of b=0.60, agreeing with its relatively inferior kinetic performance as shown in GITT (Figure 2e). The b-values for other redox peaks exceed 0.8, indicating a dominant contribution of pseudocapacitance effect on Cu-involved reactions, which indeed display higher kinetic activity. The total current can be distinguished into the capacitive- and diffusion-contributions 12

(𝑘1𝜈 and 𝑘2𝜈

respectively) quantitatively from the Equation (3).22

12

𝑖𝑝 = 𝑘1𝜈 + 𝑘2𝜈

12

or 𝑖𝑝 𝜈1 2 = 𝑘1𝜈

(3)

+ 𝑘2

The corresponding capacitive current as a function of potential is obtained via calculating the k1v value (Figure 3c). The percent of capacitive charge exceeds 50% in the whole electrochemical window for all the scan rate conditions and is 58% at 1 mV s-1 (Figure 3d). This fraction is lower than that of LiF-MO system (94% capacitive charge),14 indicating a sufficient activation of diffusion-driven reaction in thin film system. The percentage of capacitive contribution for each redox peak is listed in Figure 3e. For most the peaks (apart from Peak C1), the ratio is close to or more than 60%, agreeing with the high b-value.

12 ACS Paragon Plus Environment

Page 12 of 30

Page 13 of 30 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Nano

Figure 3. (a) CV profiles for LFC211 cathode at various scan rates from 0.1 to 1.0 mV s−1 after the first charge. The cathodic and anodic peaks are also labeled with A1, A2, A3, C1 and C2. (b) Logarithm-scale plots of peak current vs. scan rate and their linear fitting at the corresponding characteristic peak potentials. (c) Overall (solid line) and capacitive (shaded region) current profiles of LFC211 at the scan rate of 1 mV s-1. (d) Column graphs of ratedependent charge storage contributions from both capacitive and diffusive processes for LFC211 electrode. (e) Column graphs of peak-dependent stored charge and contribution from both capacitive and diffusive processes for LFC211 electrode.

The cyclic performance of LFC211 thin film at 41.6 mA g-1 is shown in Figure 4a. The voltage curve profile and capacity are highly reversible after the first charge activation. 88% of the discharge capacity can be reserved after 200 cycles (375 mAh g-1 from initial 425 mAh g-1) with a high coulombic efficiency of 95% (Figure 4b). This conversion system also shows a superior rate performance with reversible capacities of 300, 208 and 126 mAh g-1 at higher 166.7, 333.4 and 667 mA g-1 respectively (Figure 4c). The capacity can still approach 400 13 ACS Paragon Plus Environment

ACS Nano 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

mAh g-1 when the current density returns back to 41.6 mA g-1 after undergoing the application of larger current density. Note that LFC211 thin film shows evident advantages in terms of both the energy and power densities over already reported LiF-driven conversion systems (e.g. LiF/M ,LiF/MO and LiF/MF2) (Figure 4d).14,17,19-21,23-29 When the power density is 104 W kg-1, the energy density of LFC211 can even reach to 1053 Wh kg-1, which is much higher than those of typical insertion materials (350 W kg-1 for LiFePO4 and 550 W kg-1 for LiCoO2).39,40 Its energy densities are still maintained at 491 and 217 Wh kg-1 under the extremely high power densities of 849 and 1740 W kg-1, respectively. This rate performance is expected to be superior to the corresponding charged state material (CuyFe1-yF2) reported by Wang et. al.15 Figure 4e compares the energy efficiency values among these LiF-driven conversion systems. Our system has a quite high energy efficiency value exceeding 75% even under higher current density, closing to that characterized by easier conversion starting from FeF2 or MnO.14,17 The blending of Cu nanodomains greatly improves the energy efficiency, which is usually below 65% for most metal-LiF systems.19-21,23-29

14 ACS Paragon Plus Environment

Page 14 of 30

Page 15 of 30 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Nano

Figure 4. (a) Galvanostatic charge-discharge curves of LFC211 cathode during the first charge and following 10 cycles in the potential range of 1.5-4.2 V at 41.6 mA g-1. (b) Cycling capacity and Coulombic efficiency of LFC211 as a function of cycle number at 41.6 mA g-1. (c) Rate capability of LFC211 at 41.6, 82.3, 166.7, 333.4 and 667 mA g-1. Comparison of (d) energy-power densities and (e) energy efficiency between LFC211 and already reported LiF-driven conversion systems.

A series of X-ray photoemission spectra (XPS) profiles with different etching depth in as-deposited LFC211 thin film are shown in Figure S5. It is found that O and C elements exist near the film surface due to air exposure during sample transfer. After 80s etching, the signals of O and C contaminants have been removed. Accordingly, the surface Fe atoms are prone to be oxidated with the appearance of Fe-O peak at 711 eV for Fe 2p3/2.41 During etching, the metallic Fe peak at 706.7 eV (for Fe 2p3/2) is gradually intensified along with the weakening of Fe-O peak.42 The surface oxidation of Fe is not necessarily detrimental for the electrochemistry. It is likely favorable for the interface transport between electrode and electrolyte. Since the XPS peak positions for Cu 2p are close for both Cu-O and metallic Cu, 15 ACS Paragon Plus Environment

ACS Nano 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

we cannot dicern the natural oxidation degree of Cu near the film surface. The XPS spectra of pristine and cycled electrodes are compared in Figure 5. All the electrode surfaces were cleaned by Ar ion sputtering before measurement. Li 1s peak at 56.2 eV is evidently weakened after the first charge to 4.2 V, comfirming a substantial splitting of LiF (Figure 5a).43 The excess LiF in thin film is responsible for the residual of Li signal. The residual LiF is beneficial for the construction of desired F-rich SEI at cathode side.44,45 The LiF peak is strengthened again after the following discharge to 1.5 V, indicating a reversible re-formation of LiF via conversion reaction. In the same spectra, Fe 3p peaks show the similar evolution with the concomitant shrinkage of Fe0 peak at 52.9 eV after charge and intensification after dicharge.46 Note that a new peak corresponding to Fe-F bonding generates at 56.3 eV after charge, indicating the formation of CuyFe1-yF2 phase.47 The Fe redox process is also clearly disclosed in Fe 2p3/2 spectra (Figure 5b). Therein metallic Fe peak locates at 706.7 eV in pristine electrode.48 The charged electrode presents a FeF2-like peak at 711.3 eV with a minor residual of metallic Fe signal.49 The over-oxidation of Fe cannot be ruled out from the appearance of weak FeF3-like signal at 714.8 eV.47 The discharged electrode displays a highly reversible re-generation of metallic Fe peak with the corresponding disappearance of Fex+ peaks. The satellite feature at 718.1 eV is attributed to the shake-up process by exciting a Fe 3d electron into 4s unoccupied level.50 In F 1s spectra (Figure 5c), the peak shifts from 685.4 eV to 684.6 eV after first charge, indicating a breakage of Li-F bonding and simultaneous formation of M-F bonding during Li extraction.51 This peak can return back to the initial position after discharging owing to the re-bonding between Li and F.

16 ACS Paragon Plus Environment

Page 16 of 30

Page 17 of 30 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Nano

Figure 5. XPS spectra of (a) Li 1s and Fe 3p, (b) Fe 2p3/2 and (c) F 1s for pristine LFC211 and cycled electrodes after charge to 4.2 V and discharge to 1.5 V.

The X-ray absorption near-edge structure (XANES) spectra were carried out to investigate the chemical state of Cu element in Figure 6. The K-edge positions are located at 8979 and 8988.8 eV for standard Cu foil and CuF2 powder respectively.52 For as-deposited and discharged electrodes, the curves are highly overlapped and have the similar K-edge position corresponding to metallic Cu, agreeing with the highly redox reversibility of Cu. Upon charge, the peak around 8981 eV, corresponding to the 1s−4p transition for Cu metal, is weakened in intensity. Meanwhile, an intensity increase of the peak at 8993 eV, corresponding to the 1s−4p transition for CuF2, is observed.53 These evolutions indicate the oxidation of Cu to Cu(II) species along with the splitting of LiF, well agreeing with previous reports by Grey and Wang et al.15,52

17 ACS Paragon Plus Environment

ACS Nano 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 6. Normalized Cu K-edge XANES spectra of Cu foil, CuF2, pristine LFC211 and cycled electrodes after charge to 4.2 V and discharge to 1.5 V.

Figure 7 shows the ex-situ TEM results of fully charged and discharged electrodes. Upon charge to 4.2 V, the tetragonal rutile phase is formed (space group: P42/mnm) according to the HRTEM image and SAED pattern (Figure 7a and b).15 The diffraction rings and lattice dspacing could not be assigned to monoclinic CuF2 phase (space group: P21/n).11 Based on the XPS and XANES results, the coexistence of Fe2+ and Cu2+ species illuminates the potential formation of CuyFe1-yF2 solid-solution phase after F transfter from LiF to metallic Fe and Cu. According to CV results based on different scanning window (Figure 2d), the fluorination of Fe and Cu likely occurs simultaneously rather than sequentially. The doping of CuF2 moieties into FeF2 would not change the rutile framework of Cu–Fe–F final phase although with more or less distortion or symmetry modulation.15 After discharging to 1.5 V, the amorphization (or disordering) tendency of reductive products is more evident compared with the pristine electrode as shown from the vaguer diffraction rings and finer nanodomains in Figure 7c and d. Note that the HRTEM of pristine electrode displays a clear dark contrast corresponding to the homogeneous spatial distribution of LiF/metal nanodomains. Such a contrast becomes less evident after charge, agreeing with the formation of monolithic Cu-Fe-F phase with fewer phase interfaces. The dark contract becomes distinct again but with smaller-sized 18 ACS Paragon Plus Environment

Page 18 of 30

Page 19 of 30 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Nano

nanodomains after discharge, which is caused by the electrochemical grinding of conversion products of LiF, Cu and Fe.16,54 Their re-generation is confirmed from the lattice stripes of HRTEM and diffraction rings of SAED. Such a grinding is beneficial for the further improvement of conversion kinetics and voltage polarization, e.g. as indicated from the lower charge voltage and smaller energy barrier for LiF splitting.

Figure 7. HRTEM images of cycled LFC211 electrodes after (a) charge to 4.2 V and (c) discharge to 1.5 V, and their corresponding SAED patterns in (b) and (d) respectively.

Lithiated fluoride cathode enables a Li-ion battery coupled with Li-free anode, e.g. conventional graphite or silicon.17 However the achievement of high energy density and efficiency in lithiated fluoride is still a big challenge, owing to the difficult availability of microstructure design and fine conductive network construction. The pristine Li-free fluorides can usually be decomposed into LiF/M (M: transition metal) composite with desired nanodomain distribution, which guarantees the following recombination of LiF and M into 19 ACS Paragon Plus Environment

ACS Nano 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

MFx at least during the early cycles.33 However in the past years, the direct blending of M and LiF as composite cathode is not satisfactory in view of the loose contact between components and larger grain size than those underging electrochemical grinding. The splitting of LiF during the first charge process has to overcome a high decomposition/nucleation energy barrier, leading to huge overpotential and insufficnent capacity release.27 The discharge cutoff voltage has to be extended to the anode range in order to increase the utilization of active species. The intrinsic lattice mismatch between metal and LiF also accelerates the compromise of grain compact stacking and the expansion of volume during cycling. Although using MO or MF2 instead of M can mitigate the lattice mismatch and volume evolution, its reverible capacity (200-300 mAh g-1) is limited by single-electron transfer and is merely comparable to that of layered oxide cathodes.14,17,24-26,28,29 The addition of Cu in LiF-Cu-Fe composite not only promotes the kinetics of Fe/Fex+ multi-electron transfer, but also activates the high-voltage reversibility based on Cu/Cu2+ couple. The homogeneous Cu nanodomains could build electronically conductive pathway to ensure sufficient electron flux for LiF splitting when Fe nanodomains are oxidized to Fe-F phases in the charge process. The nanoCu moieties could also improve the phase contact between LiF and Cu, and therefore decrease the energy barrier of LiF splitting during Cu-F formation. Thin-film deposition method is a facile route to achieve homogeneous spatial distribution of deposition compositions and meantime limit their grain-size and crystallinity growth. Therefore the thin film architecture enables a more compact contact between heterostructures or nanodomains (e.g. LiF and metal clusters) with the strengthened internal wiring network. Moreover, the 2D or 3D conformal attachment of electroactive species to conductive substrate benefits to the reinforcement of external conductive networks even without the assistance of excess conductive additive. In contrast, powder materials are full of loose grain boundaries and their grains are likely linked by insulating binder. These factors would lead to the interruption of conductive networks and deactivate the conversion kinetics, especially for the intrinsically difficult splitting of 20 ACS Paragon Plus Environment

Page 20 of 30

Page 21 of 30 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Nano

insulating LiF. Therefore our nanoscale film configuration guarantees a facile splitting of LiF with low overpotential and considerable capacity (>300 mAh g-1). Other methods never achieved such a superior effect (with reversible capacity close to 400 mAh g-1 as well as energy and power densities exceeding 1000 Wh kg-1 and 1500 W kg-1 ) for LiF-based conversion cathodes (Figure 4d).14,17,19-21,23-29 Figure 8 shows the potential reaction scheme of LiF/Cu/Fe during the first charge and discharge. With the extraction of Li from LiF lattice, the released F is expected to preferentially bond with surrouding nano-Fe to form Fe-F building blocks owing to lower thermodynamic potential of Fe/Fe2+ than Cu/Cu2+. The high surface energy of Fe-F moieties can trigger the following formation and connection of Cu-F building blocks promptly, leading to the nucleation and growth of Fe-Cu-F rutile phase. This rutile network can be electrochemically decomposed during the following lithiation. The Cu-F moieties are prone to be decomposed firstly and then Fe-F ones based on the appearance of two-staged plateaus.15 Injected Li atoms would take away F owing to strong bonding of Li-F and meantime reduce Cu and Fe to metal state. The mergence of these metal atoms are limited by surrounding LiF matrix with sufficient mechanical strength, leading to the discrete disctribution of fine metal nanodomains. The further development of fluoride cathodes should particularly focus on the exploration of compatible built-in (or lattice-connected) Li source and penetrative electron/ion conductive pathways both in intra-grain zone and between grains. The latter is expected to further lower the reaction overpotential and improve the conversion energy efficiency. The building-block and defect-chemical modulation in structure is also expected to enable the achievement of faster mass-charge transport.16

21 ACS Paragon Plus Environment

ACS Nano 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 8. Schematic illustration of conversion reaction mechanism for LiF/Fe/Cu electrode. With the extraction of Li from LiF lattice during the first charge process, the released F preferentially bonds with Fe to form Fe-F building blocks. The as-formed Fe-F moieties trigger the formation and attachment of Cu-F building blocks on them, leading to the nucleation and growth of Fe-Cu-F rutile phase at the end of charging. This rutile network is then electrochemically decomposed during the following lithiation. Injected Li atoms take away F to regenerate LiF nanodomains with sequential reduction of Cu and Fe to their metal states.

Conclusion In summary, a dual-metal accelerated LiF splitting is proposed to achieve a lithiated composite cathode (LiF/Fe/Cu) with superior conversion reaction performance. It enables a highly reversible capacity of 375-400 mAh g-1 with high energy efficiency (76%) and capacity retention (at least 200 cycles). The first charge process is activated by a considerable capacity release (300 mAh g-1) with low overpotential, corresponding to the electrochemical synthesis of rutile-like Fe-Cu-F solid-solution phase. The energy density of LiF/Fe/Cu can be as high as 1053 Wh kg-1 under a power density of 104 W kg-1. Its energy densities are still preserved at 491 and 217 Wh kg-1 under the extremely high power densities of 849 and 1740 W kg-1 respectively. The combination of dual metals endows LiF/Fe/Cu with a best capacity performance among LiF-driven conversion systems, benefiting from a substantial pseudocapacitance contribution (>50%) as well as thin-film architecture 22 ACS Paragon Plus Environment

Page 22 of 30

Page 23 of 30 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Nano

characterized by homogeneous distribution and compact contact of LiF and metal clusters or nanodomains.

Experimental Section Sample Preparation: The targets were prepared by mixing and pressing the LiF, Fe and Cu powders (AR, Alfa Aesar Chemical reagent Inc.) with different molar ratio at a pressure of 4 MPa cm-2. The LiF/Fe/Cu, LiF/Fe and LiF/Cu thin film electrodes were grown on Pt-Ti covered Si substrates by pulse laser deposition (PLD) technology. The substrates were sonicated in acetone, ethanol and deionized water for 15 mins, respectively. A KrF excimer laser with a wavelength of 248 nm and frequency of 5 Hz is used for the thin film growth. The energy density of laser is 2.5 J·cm-2. The thin film elelctrodes were grown at room temperature under a pressure of 1.0×10-5 Pa, and then were annealed at 200C under the high-vacuum environment. The distance between target and substrate is 505 mm. The weight of thin film was obtained based on the weight difference of sample before and after deposition via electro-balance (BP 211D, Sartorius). Characterization: The surface and cross-section morphologies of LiF/Fe/Cu thin films were observed by scanning electron microscope (SEM, S-4800, Hitachi) equipped with energy dispersive spectrum (EDS) component for element mapping characterization. Transmission electron microscopy (TEM) images and selected area electron diffraction (SAED) patterns were captured by a Libra 210 FE microscope operating at 200 kV. The structure and crystallinity of pristine samples were analyzed via grazing incidence X-ray diffraction (GIXRD) with an incidence degree of 1 in a 2-theta range of 35−75° at a scan rate of 5° min−1 by Bruker D8 advance diffractometer equipped with Cu Kα radiation (λ=1.5406 Å). The evolution of bonding situations of pristine and cycled electrodes was disclosed by X-ray photoelectron spectroscopy (XPS) on a Perkin23 ACS Paragon Plus Environment

ACS Nano 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Elmer PHI 6000C ECSA system with monochromatic Al K (1486.6eV) irradiation. For XPS fitting, the binding energy was aligned based on the reference C1s peak at 284.8 eV. A Shirley-type background was used, and the experimental data were fitted using a nonlinear least-squares fitting based on a mixed Gaussian/Lorentzian peak shape. The X-ray Absorption Fine Structure (XAFS) data for Cu K-edge were collected at room temperature in fluorescence mode using Hard X-ray Microfocus Beamline BL15U1 of the Shanghai Synchrotron Radiation Facility on two single Vortex-90EX Si drift detectors (energy resolution, ΔE/E = 1.7×10−4). During the measurement, the synchrotron was operated at an energy of 10 keV and a ring current of 200 mA. The focal spot size at the position of sample was ~7 × 2.5 μm2. The incident photon energy was calibrated using a standard Cu metal foil just prior to data collection from all samples. For ex-situ XPS, XAFS and TEM characterization, the cycled electrodes were taken out from the cells in an Ar-filled glovebox and then carefully washed by electrolyte solvent to remove the residual electrolyte salt. The washed electrodes were then dried in an Ar-filled glovebox before measurement. Electrochemical Measurements: A glass-beaker-type three-electrode system was employed in the electrochemical experiments. The as-deposited thin film serves as working electrode (area of 1 cm2) and two sheets of lithium were used as reference and counter electrodes, respectively. The electrolyte consists of 1 M LiClO4 in ethylene carbonate (EC) and polycarbonate (PC) with a volume ratio of 1:1. The cells were assembled in a glove box filled with argon (