Stripping Electrochemistry Realized by a Hybrid Li

Apr 6, 2017 - Chemistry of Materials 2017 29 (21), 9182-9191 ... Journal of The Electrochemical Society 2018 165 (1), A6058-A6072 .... Xin-hui Xia , X...
2 downloads 0 Views 3MB Size
Article pubs.acs.org/JACS

Stable Li Plating/Stripping Electrochemistry Realized by a Hybrid Li Reservoir in Spherical Carbon Granules with 3D Conducting Skeletons Huan Ye,†,‡ Sen Xin,§ Ya-Xia Yin,†,‡ Jin-Yi Li,†,‡ Yu-Guo Guo,*,†,‡ and Li-Jun Wan*,† †

CAS Key Laboratory of Molecular Nanostructure and Nanotechnology, CAS Research/Education Center for Excellence in Molecular Sciences, Institute of Chemistry, Chinese Academy of Sciences (CAS), Beijing 100190, P. R. China ‡ School of Chemistry and Chemical Engineering, University of Chinese Academy of Sciences (CAS), Beijing 100049, P. R. China § Department of Mechanical Engineering, The University of Texas at Austin, Austin, Texas 78712, United States S Supporting Information *

ABSTRACT: Lithium metal is a promising battery anode. However, inhomogeneous mass and charge transfers across the Li/electrolyte interface result in formation of dendritic Li and “dead” Li, and an unstable solid electrolyte interphase, which incur serious problems to impede its service in rechargeable batteries. Here, we show that the above problems can be mitigated by regulating the interfacial mass/charge transfer. The key to our strategy is hybrid Li storage in onion-like, graphitized spherical C granules wired on a three-dimensional conducting skeleton, which enhances the negativity of surface charge of the C host to contribute to a uniform Li plating while also forming stable Li/C intercalation compounds to offset any irreversible Li loss during cycling. As a result, the anode shows a suppressed dendrite formation and a high Li utilization >95%, enabling a practical Li battery to strike a long lifespan of 1000 cycles at a surplus Li of merely 5%.



INTRODUCTION A rapid advancement of portable electronics, electric cars, and grids stimulates extensive studies on high-energy rechargeable batteries.1−5 Li metal delivers a high theoretical capacity of 3860 mA h g−1 and a very low reduction potential (−3.04 V vs standard hydrogen electrode),6−9 rendering a higher energy density of rechargeable Li batteries than commercial Li-ion batteries based on intercalation anodes such as graphite.10 Nevertheless, challenges regarding the use of the Li anode still exist because of an unregulated mass and charge transfer across the Li/electrolyte interface, which induces serious outcomes. First, uneven Li plating on the anode surface at the microscopic level leads to formation and preferential growth of Li dendrites, which may cause an internal short circuit of the battery and lead to safety concerns.11−14 Second, an unstable Li/electrolyte interfacial chemistry gives rise to generation of “dead” Li (Li losing its electric contact with the anode) and repeated breakdown/buildup of solid electrolyte interphase (SEI) during cycles, leading to an irreversible Li consumption far beyond the theoretical amount.15−21 Recently, researchers have found that the inhomogeneous mass transfer further leads to an almost infinite dimensional change in the Li anode during its stripping/plating process, which eventually arouses mechanical damage of the anode and leads to poor performance.22,23 Targeted at the above problems of the Li anode, numerous efforts have been made, which include use of physical protection layers and electrolyte additives of liquid electrolytes (e.g., polysufide, Cs+, and LiNO3) to suppress dendrites and © 2017 American Chemical Society

applying an artificial SEI layer (such as Li3PO4) to stabilize the Li/electrolyte interface.24−40 In addition, regulating the local electrical field (e.g., by using a 3D current collector) has also been proved effective to uniformize deposition of Li.41−44 Though certain advancements have been achieved, the root cause beneath these problems, i.e., the inhomogeneous mass and charge transfers across the Li/electrolyte interface, has rarely been studied from a scientific point of view. In the light of a practical battery, other vital factors determining the performance of the Li anode, such as the low Li utilization and continuous Li loss during the plating/stripping process, should be addressed but are often neglected. To improve the efficiency and cycle life of the Li anode, an excessive amount of Li (an average value of 300% in the current studies) has often been used to pair the cathode, which inevitably decreases the energy density of the battery.8,18,19 Hence, effective strategies to strike high utilization and low active material loss of metallic Li anode and thus guarantee a rechargeable Li battery with higher energy density must be explored. Graphitic carbon materials have been serving as the anode materials for commercial Li-ion batteries due to their suitable layered structure and high conductivity, enabling stable Li storage via Li+ (de)intercalation, yet they have limited storage capacities.45 Recently, studies showed that amorphous carbon substrates with appropriate nanoarchitecture design can also be Received: February 20, 2017 Published: April 6, 2017 5916

DOI: 10.1021/jacs.7b01763 J. Am. Chem. Soc. 2017, 139, 5916−5922

Article

Journal of the American Chemical Society

The high-temperature test was performed at 55 °C at a rate of 1.0 mA cm−2 for 4 and 8 mA h cm−2 Li. The average diameter of the BNF and the CMN for Li deposition was 10 mm. Cycling tests were carried out by first depositing 2, 4, and 8 mA h cm−2 of Li onto the BNF and the CMN at various current rates, followed by Li stripping up to 1.0 V at the same rate. Li|Li symmetric cells were assembled with Li metal used as the working and counter electrodes to evaluate the cycling stability and cycle life of the Li anodes on the BNF and the CMN electrode. Full Cell Tests. To test the Li−CMN anode in a full cell, LiFePO4 (Sanxin Industrial) was used as a cathode material. The LiFePO4 cathode was prepared by mixing the active material LiFePO4, super carbon, and polyvinlidene fluoride at a weight ratio of 80:10:10 in Nmethylpyrrolidone to form a slurry. The slurry was pasted on aluminum foil and dried at 80 °C for 12 h. The electrode was punched into circular disks with diameters of 10 mm with LiFePO4 loading mass of approximately 8 mg cm−2, corresponding to the average electrode thickness of 121.6 μm. The areal capacity of LiFePO4 is of 1.25 mA h cm−2. The full Li−metal cells are assembled with a LiFePO4 cathode delithiated all Li+ out, and the Li−CMN anode deposited the required amount of Li. Specifically, in half cells, LiFePO4 electrode of 1.25 mA h cm−2 and the CMN electrode paired with Li foil were assembled and cycled for five cycles to maximize the utilization and to stabilize the Coulombic efficiency of both electrodes, respectively. After Li was completely extracted from the LiFePO4, the delithiated lithium iron phosphate, that is, iron phosphate (FePO4) electrode, was disassembled in the glovebox and served as the cathode of a full cell. Meanwhile, the CMN anode with predeposited corresponding Li was regarded as the anode of the full cell (wherein the subtle capacity contribution of lithiated carbon spheres is considered even though the lithiation of carbon spheres is only less than 10% of the whole areal capacity). When the capacity ratio of LiFePO4 versus the capacity of Li is 1:1 (CLiFePO4/CLi = 1:1), 1.25 mA h cm−2 of Li was deposited on the CMN. When the capacity ratio of LiFePO4 versus the capacity of Li is 1:1.05 (CLiFePO4/CLi = 1:1.05), (1.25 + (1.25 × 5%) = 1.31) mA h cm−2 of Li was deposited on the CMN. Then we coupled the Li− CMN electrode as an anode with the delithiated lithium iron phosphate cathode and assembled a new full cell abbreviated as CMN| LiFePO4. Electrochemical impedance spectra (EIS) measurement was conducted on the Autolab with frequency range from 100 kHz to 0.1 Hz.

used for Li plating/stripping, but a longer cycle life is required to make a practical full battery.23,24 To efficiently utilize the structural merits of carbon materials as Li host while improving the efficiency and lifespan of plating/stripping cycles, here we propose, for the first time, to regulate the Li/electrolyte interfacial transfer by plating Li into spherical C granules wired onto a 3D conducting skeleton. The spherical carbon consists of highly graphitized carbon layers arranged to form an onionlike structure with nanogaps in between. During the Li plating process, Li ions are preferably intercalated into the graphite layers to form a Li/C compound and then plated in the nanogaps. The curved surface of spherical C makes the delocalized π electrons of graphitic C atoms become partially localized, which increases the negativity of the surface charge of C spheres. After intercalation of Li, an electron deviation from Li to C further elevates the negativity of C atoms. The result is a stronger binding between surface C and Li+ in the electrolyte, which uniformizes the Li+ flux and improves the wettability of Li on spherical C and finally contributes to a stable Li deposition free of dendrites or hostless Li. Moreover, the Li stored in the Li/C compound acts as a backup source to offset any irreversible Li loss during cycling. The resultant Li anode shows a high Li utilization (>95%) and keeps a stable plating/ stripping performance for 500 cycles. Based on the Li anode, we further build a Li metal full battery by coupling the anode with a lithium iron phosphate (LiFePO4) cathode, which demonstrates its practicality by achieving an ultralong lifespan of 1000 cycles at a Li surplus as low as 5%.



EXPERIMENTAL SECTION

Preparation of CMN. Nickel (Ni) foams (Shang Hai Zhong Wei New Materials Co., Ltd.) with an average pore size of 150 μm were regarded as templates for the preparation of CMNs, following the preparation process of graphitic foam. The Ni foams were first washed by deionized water and alcohol alternately to remove surface impurities. Then the dried Ni foams were heated to 900 °C for 30 min under H2/Ar (5% H2 in volume) to remove a thin surface oxide layer. Next, the mixture of acetylene (C2H2) and Ar at a volume ratio of 10:1 was introduced into the quartz tube for 2−10 min to prepare CMN. The BNF was punched out into circular disks with a diameter of 10 mm as the 3D current collectors for Li anode. Structure Characterizations. SEM (6701F, operating at 10 kV), slice transmission electron microscope (TEM) (Tecnai F20), were carried out to visualize the morphological characteristics, sizes, and microstructures of the CMNs. Cycled coin cells were dissembled to obtain a Li anode for characterizations of ex situ SEM, TEM, XRD, and XPS. Before testing, the disassembled electrodes were rinsed with DOL solvent to remove residual electrolytes and Li salts and then dried in the glovebox. EDX elemental analysis (Tecnai83 F20) was employed to detect the elemental compositions of the as-obtained sample and Li anode. XRD was conducted using a Philips PW3710 with filtered Cu Kα radiation (Rigaku D/max-2500, l = 1.5405 Å) to characterize the structure of the Li anode on the CMN. An ESCALab220i-XL electron spectrometer (VG Scientific) with 300 W Al Kα radiation was performed to conduct XPS and analyze the elemental and valence. Electrochemical Measurements. Electrochemical measurements were performed using CR2032-type coin cells of a two-electrode configuration. Coin cells were assembled by using metallic Li foil as the counter electrode, a Celgard separator and BNF, and CMN as the working electrode. Lithium bis(trifluoromethanesulfonyl)imide (LiTFSI) in a cosolvent of 1,3-dioxolane (DOL) and 1,2-dimethoxyethane (DME) with lithium nitrate (1 wt %) was used as the electrolyte. To standardize the testing, 80 μL of electrolyte was used in each coin cell. The current density for the Li metal plating/stripping was set to 0.5, 1.0, 2.0, and 4.0 mA cm−2 using An Arbin BT2000 system at 25 °C.



RESULTS AND DISCUSSION Preparation for the CMN. The spherical C granules were prepared on a nickel foam via chemical vapor deposition (CVD) to form a carbon modified Ni foam (CMN). The preparation procedures follow those for graphene and carbon nanotubes46,47 and are shown in Figure 1a. Ni foams, which

Figure 1. Structural and morphological characterization results of the CMN. (a) Schematic illustration showing the preparation process of the CMN. (b) Low-magnification SEM image of the CMN. (c) Crosssectional SEM image of spherical C and (d) sliced TEM image of a spherical C, which show the onion-like structure. (e) Cross-sectional SEM image of the CMN. 5917

DOI: 10.1021/jacs.7b01763 J. Am. Chem. Soc. 2017, 139, 5916−5922

Article

Journal of the American Chemical Society served as the 3D scaffold, were heated to 900 °C for 30 min under a mixed atmosphere of Ar and H2 to remove any thin oxide layer on the surface. Then acetylene (C2H2) was introduced to induce the formation of spherical C granules on the surface of Ni foam. Within several minutes, the Ni foam became coarse and was completely covered by C spheres (Figure 1b). From the cross-sectional scanning electron microscopic (SEM) image (Figure 1c) and the transmission electron microscopic (TEM) image (Figure 1d) one can observe that the C spheres display an average diameter of around 1−2 μm and an onion-like structure consisting of graphite nanosheets in a concentric circle arrangement, with numerous nanogaps in between. Upon increasing the time of CVD treatment, the thickness of the spherical C layers on the Ni foam also increases. To minimize the capacity contribution by lithiated C spheres indicated by the red-shift of the G-band and blue-shift of the D-band in Raman spectroscopy (Figure S1a) and the sloping voltage profile (Figure S1b),48,49 the areal loading of spherical C on the Ni foam was controlled around 3−4 mg cm−2, which corresponded to a C layer thickness of 3 μm (Figure 1e). Li Metal Deposition Behavior. The Li deposition behaviors on both the bare Ni foam (BNF) and the CMN were first investigated. For the BNF, an accelerating growth of Li dendrite is observed as the current density increases from 0.5 to 2 mA cm−2 (Figure S2). Figure S3 collects the morphological evolution of Li deposition on the BNF at different stages. On the BNF, micron-sized Li particles are observed at the beginning of the plating process (Figure S3b). After that, the formed Li particles serve as nucleation sites to induce growth of long tubular-shaped Li dendrites (Figure S3c). Further increasing the areal capacity of Li plating leads to significant growth of Li and form numerous, randomly oriented Li dendrites to finally fill the pores (Figure S3d−f). Such a plating behavior demonstrates that the BNF cannot suppress the dendrite growth. The formed Li dendrites inevitably react with the electrolyte, which prompts the consumption of Li metal and electrolyte and brings a lowered Coulombic efficiency during the Li plating/stripping. The Li deposition onto the CMN shows an entirely different morphological evolution. When Li is plated with a constant areal capacity of 2 mA h cm−2 at different current densities (0.5, 1, and 2 mA cm−2), no dendritic Li forms and only regular petal-like Li gradually appears on the surface of spherical C (Figure S4). This result suggests that a large number of curved graphite sheets enable the relatively homogeneous Li+ flux distribution among the carbon granules, which improves the wettability of Li on C and thus contributes to a uniform Li deposition during the plating process. Ex situ SEM characterization was further employed to reveal the morphological evolution of Li plating/stripping on the CMN at different areal capacities (Figure 2). Compared with the pristine CMN at the open circuit voltage (Figure 2a), no obvious dendrite formation is observed at a capacity of 0.8 mA h cm−2, with only a few petal-like Li appearing on the outside of the spheric C (Figure 2b and Figure S5a). This result suggests that the Li deposition preferably occurs in the nanogaps between graphite sheets. As the Li plating capacity increases to 2 mA h cm−2, an almost complete coverage of petal-like Li is observed on the surface of spheric C (Figure 2c and Figure S5b). With further increasing Li plating capacity to 4 mA h cm−2 on the CMN, the petal-like Li layer on the C surface is significantly thickened, and the whole surface of spheric C is covered by the

Figure 2. Morphological evolution of Li plating/stripping on the CMN, including (a) pristine CMN and after plating, (b) 0.8 mA h cm−2, (c) 2 mA h cm−2, (d) 4 mA h cm−2 of Li into the CMN and after stripping, (e) 1 mA h cm−2, (f) 2 mA h cm−2, (g) 3 mA h cm−2, (h) 4 mA h cm−2 (that is, charged back to 1.0 V) of Li from the CMN. Li plating/stripping states (a−h) are marked in the (i) galvanostatic discharge/charge voltage profile of a Li|Li−CMN half cell obtained at 2 mA cm−2.

SEI as confirmed by the energy-dispersive X-ray pattern (Figure S6). Eventually, large, petal-like Li particles form on the surface of the C granules, leading to an increased diameter of spherical C (Figure 2d and Figure S5c). When the Li metal is gradually stripped from the CMN, the Li particles disappear (Figure 2e− h), and finally, the spherical C retains its initial morphology after a complete stripping at 1.0 V (Figure 2h). This result suggests an excellent structural stability and flexibility of the CMN in accommodating the volume change of Li during the plating/stripping process. Ex situ X-ray diffraction and highresolution TEM were employed to reveal the features of deposited Li inside the C spheres. Two typical peaks, (110) and (211), of crystalline Li were observed in the CMN after plating 2 mA h cm−2, demonstrating that Li was deposited in crystalline arrangements (Figure S7). HRTEM was employed to visualize the detailed distribution of deposited Li in carbon spheres. Clear lattice fringes of cubic Li metal and absence of bulk Li particles were observed in the graphite nanosheet, thereby demonstrating the fine dispersion of Li among the graphite nanosheets (Figure S8a). Clear lattice fringes, separated by a 2.45 Å distance, were observed in the enlarged TEM image of the dotted dash portion in Figure S8b, which corresponds to the interplanar distance of the (110) planes of the cubic metal Li. The results indicate a homogeneous deposition of nanosized Li into the gaps between adjacent graphite sheets. On the basis of the above discussion, the Li plating on CMN is summarized as a schematic diagram in Figure 3. The spherical C granules, which are self-assembled by graphite sheets, possess abundant nanochannels that allow Li + penetration and a high electronic conductivity to facilitate e− transmission. At the initial stage of plating, the granules are lithiated and a stable SEI is induced to form on their surface, which prevents penetration of solvent molecules and allows interfacial transfer of Li+. As the voltage goes down to below 0 V (vs Li+/Li), Li+ migrates through the SEI to enable a homogeneous deposition of Li into the nanogaps sandwiched by the graphite sheets. After the C spheres are filled by Li, Li metal begins to deposit on the outermost surface of the C 5918

DOI: 10.1021/jacs.7b01763 J. Am. Chem. Soc. 2017, 139, 5916−5922

Article

Journal of the American Chemical Society

remains at 75% in the subsequent cycles, which is in good agreement with our previous reports showing that Cu foam with a large pore size results in poor plating/stripping efficiency of only ∼40%.44 On the contrary, the CMN electrode exhibits a greatly improved initial Coulombic efficiency of 90% and quickly stabilizes its efficiency at >98% after several cycles and maintains it for 475 cycles (Figure S9a). These data suggest a stable SEI formation and a high reversibility of Li plating/ stripping reaction on the CMN. Being plated/stripped at a higher current density of 1.0 mA cm−2, the CMN electrode still displays a stable efficiency of 98.5% and a low voltage hysteresis of 75 mV for about 600 cycles (Figure S9b). Furthermore, the CMN electrode can still deliver a low hysteresis of 140 mV (Figure 4c). The voltage hysteresis increases upon cycling, indicating an elevated charge-transfer resistance due to unstable Li/electrolyte interface and growth of dendritic Li. After cycling for about 220 h, the voltage hysteresis suddenly drops from 170 mV to 20 mV, which could be attributed to the dendrite-induced short circuit. In contrast, the Li|Li−CMN cell displays an excellent cycling stability as evidenced by a negligible voltage fluctuation and a much lowered overpotential (∼80 mV in the initial cycles), which further stabilizes at 45 mV for more than 270 cycles (∼540 h) (Figure 4c). After cycling for over 500 h, no sign of short circuit is observed, confirming an effective suppression of dendrite growth. Tested at higher current densities of 2 and 4 mA cm−2, the Li|Li−CMN cell, with a corresponding overpotential of 60 and 80 mV (Figure S14), can still run for 100 and 30 h, respectively, whereas the Li| Li−BNF cell only displays a large voltage fluctuation and a voltage that quickly exceeds the preset value (3 V). These results suggest that a stable and homogeneous SEI forms on the CMN, which ensures enhanced charge-transfer kinetics. Electrochemical impedance spectra (EIS) were exploited to investigate the interfacial stability and SEI layer. The EIS spectra collected from the BNF electrode show one semicircle at high frequency corresponding to the charge-transfer reaction and an inclined line at low frequency that is related to diffusion of ions after the first plating process (Figure S15a). In contrast, obtained EIS spectra of the CMN electrodes consist of two semicircles at high frequencies (Figure S15b). The first semicircle observed at high frequency can be attributed to lithium-ion transfer through the surface SEI layer. The second semicircle related to the interface resistance shows negligible increment with increasing the cycling numbers, demonstrating stable SEI formation and reduction of electrolyte decomposition. In addition, the CMN electrode exhibits lower interfacial charge-transfer resistance than the BNF, which further implies good preservation of SEI layer and good electron and ionic conductivity of lithium sandwiched graphite sheets structure. The reduced charge transfer resistance is in favor of low voltage hysteresis during cycling. To analyze the composition of the SEI film formed on the CMN, ex situ XPS analysis was carried out. Obviously, after the initial Li plating,

Figure 3. Schematic diagrams showing the Li plating process on the CMN.

spheres, forming petal-like primary particles since there is no spatial limitation. As the plating process progresses, more Li petals form and eventually they cover the entire surface of the CMN without generating any dendrite. Electrochemical Performance. To evaluate the Li stripping/plating performance of the CMN electrode, symmetric cells were assembled by pairing a Li foil with the CMN electrode. A BNF electrode was used as the control. Figure 4a

Figure 4. Li plating/stripping performance of the BNF electrode and the CMN electrode. (a) Voltage profiles of the Li plating/stripping on the BNF and the CMN at a current density of 0.5 mA cm−2 and a Li plating capacity of 2 mA h cm−2. (b) Voltage hysteresis evolutions with number of plating/stripping cycles on both electrodes. (c) Voltage profiles showing the Li plating/stripping process in Li|Li−BNF and Li| Li−CMN symmetrical cells at a current density of 1 mA cm−2 and a Li plating capacity of 1 mA h cm−2.

compares the initial voltage profiles of Li plating/stripping on the two electrodes under a constant current density of 0.5 mA cm−2 and areal capacity of 2 mA h cm−2. During the initial plating/stripping cycle, the BNF electrode exhibits a large plating/stripping overpotential (difference between stripping/ plating voltages) of 110 mV, while the CMN electrode demonstrates a remarkably reduced overpotential of ca. 30 mV. The capacity contribution from the Li/C intercalation compound is 0.25 mA h cm−2 (Figure 4a). In addition, the BNF electrode shows random voltage oscillations and a continuous increase in voltage hysteresis peaking at 600 mV in less than 15 plating/stripping cycles, whereas the CMN electrode reveals a smaller and more stable voltage hysteresis of less than 30 mV for 475 cycles (Figure 4b). The smaller hysteresis of the CMN electrode is ascribed to its high electronic conductivity and rapid Li+ transfer through the SEI. The Coulombic efficiency of the BNF starts at 70% and 5919

DOI: 10.1021/jacs.7b01763 J. Am. Chem. Soc. 2017, 139, 5916−5922

Article

Journal of the American Chemical Society

cycles, the Li−BNF electrode exhibits obvious dendritic growth behavior, with the characteristic of tubular-shaped structure with a length of several micrometers (Figure S18), which might induce short circuiting and persistent consumption of electrolyte. Therefore, the fast capacity decline is still observed for cells with the BNF electrode (Figure S17c). In comparison with the BNF electrode, the cell with the CMN anode exhibits a more stable cycling performance. With a cathode/anode capacity ratio of 1:1, the discharge capacity after 400 cycles maintains at approximately 93 mA h g−1, and corresponds to a capacity retention of 64% (Figure 5d), which is superior to those of the full cells reported in literatures.24,26,44 The superior cycling performance of full cell could be attributed to the dendrites-free feature of the Li−CMN anode (Figure S19), a stable SEI layer forms on the Li−CMN anode (Figures S15 and 16), the superior reversibility of Li in CMN,24 and the Li stored in the Li/C compound acts as a backup source to offset the inevitable Li loss during the Li plating/stripping process. Considering that the irreversible Li loss for constructing SEI layer on the Li anode, controlling the extra amount of Li might delicately balance the cycling performance and energy density in a full Li metal cell. Thus, we introduce slightly excessive amount of Li into the CMN electrode to compensate the required Li for the formation of SEI layer. When the capacity ratio of LiFePO4 (CLiFePO4) versus the capacity of Li (CLi) is set to 1:1.05, which means that the Li anode has excessive 5% capacity versus the LiFePO4 electrode, corresponding to Li utilization of >95%, the full cells with the Li−CMN anode demonstrate an enhanced electrochemical performance. To verify the feasibility of the reserved active Li+ enchored with Li/C compounds in a pratical battery, we carefully analyze the subsequent GDC voltage profiles. As shown in Figure S20a, the full cell delivers a capacity of 146 mA h g−1 with a typical plateau output of 3.25 V, indicating the main deinsertion of Li+ into LiFePO4 originates from the deposited metallic Li. After 100 cycles, a slope GDC voltage profile is observed which is an indication of the graphite characteristic of anode, implying the reversible deintercalation of Li+ derives from the backup source in the subsequent cycles (Figure S20b). Owing to the optional Li source to offset the inevitable Li consumption, the long-term cycling performance of the cell is guaranteed. After 250 cycles, the assembled full cell still delivers a reversible capacity of 122 mA h g −1, corresponding to capacity retention of 81.3%, which is higher than the full cell with a capacity ratio of 1:1 (Figure 5d). After 400 cycles, the full battery maintains 72.3% of the initial reversible capacity. Even at a high rate of 1 C, the Li−CMN| LiFePO4 delivers a reversible capacity of 130 mA h g−1 and cycles with an average Coulombic efficiency of 99.6% for each cycle (Figure 5e). After 1000 cycles, the full cell remains approximately 76% of the initial reversible capacity. When the areal capacity of LiFePO4 is increased to 2.5 mA h cm−2, the full cell still demonstrates promising reversibility and capacity retention (Figure S21). Mechanism of Li Plating/Stripping on the Spherical C Host. Herein, the excellent plating/stripping performance of the Li−CMN anode originates from the unique structure of the spherical C granules. As shown in Figure 6a, on the planar graphite, every C atom has an unpaired, delocalized π electron that flows freely within the graphene plane. Consequently, the surface of the planar graphite is electrically neutral and shows a weak interaction to Li+ in the electrolyte. In that case, any surface defect on the graphite will induce an uneven Li+ flux,

the surface layer of the CMN is composed of alkyllithium and lithium fluoride (Figure S16a). The content of anions upon cycling does not show much increment after five cycles, implying the SEI layer remains stable during the initial plating (Figure S16b). The stable SEI layer is beneficial for the interface stabilization. To explore the feasibility of the Li−CMN anode in a practical battery system and the impact of Li utilization on battery performance, Li−CMN|LiFePO4 full cells with two different cathode/anode capacity ratios (CLiFePO4/CLi = 1:1, CLiFePO4/CLi = 1:1.05) were constructed and tested. The Li− BNF|LiFePO4 cells were also built for control tests. In our case, the areal capacity of the LiFePO4 cathode is 1.25 mA h cm−2 as the loading mass of LiFePO4 is approximately 8 mg cm−2, which corresponds to an average thickness of cathode layer of 121.6 μm (Figure 5a). The deposited amount of Li on the

Figure 5. Electrochemical performance of Li−CMN|LiFePO4 full cells. (a) Cross-sectional SEM image of the LiFePO4 cathode. (b−e) Electrochemical performance of Li−CMN|LiFePO4 full cells at 0.2 C, including voltage profiles of full cell with cathode/anode capacity ratio of (b) 1:1 and (c) 1:1.05, (d) cycling performances at 0.2 C of the full cells with different cathode/anode capacity ratios, and (e) long-term cycling performance of the full cell with a cathode/anode capacity ratio of 1:1.05 at 1 C.

CMN was precisely controlled according to the areal capacity of LiFePO4 (see details in the Experimental Section). Cells were first cycled at a low rate of 0.1 C (calculated based on the LiFePO4 cathode) for the initial three cycles to form a stable SEI and then at 0.2 C for the subsequent cycles. The initial five galvanostatic discharge/charge voltage profiles (GDC) with equal capacity for the LiFePO4 cathode and Li anode (CLiFePO4/ CLi = 1:1), are compared in Figure 5b and Figure S17a. The full cell with CMN electrode exhibits a higher reversible capacity (145 mA h g−1 at 0.2 C) and much better cyclic stability than that with BNF electrode (108 mA h g−1 at 0.2 C), demonstrating higher electrochemical activity and reversibility of Li in CMN than that of BNF. When the cathode/anode capacity ratio is increased to 1:1.05, no obvious capacity increment is observed for the cell with Li−CMN electrode (146 mA h g−1 at 0.2 C, Figure 5c), while a high capacity (121 mA h g−1) for the cell with a Li−BNF electrode (Figure S17b). This result indicates that the cell performance with the BNF electrode is highly dependent on the amount of Li. We thus believe that for the BNF electrode more excessive amounts of Li are required to deliver the same performance as the cell compared with the CMN electrode. Possible reason might come from the fact that the inhomogeneous mass and charge transfers across the Li/electrolyte interface induces the ceaselessly consumption of Li. After 100 charge/discharge 5920

DOI: 10.1021/jacs.7b01763 J. Am. Chem. Soc. 2017, 139, 5916−5922

Article

Journal of the American Chemical Society

stabilizing the anode electrochemistry of metal Li. The keyenabling technique is the use of Li-predeposited CMN anode with elaborately designed nanostructure consisting of curved graphite sheets arranged to form an onion-sphere-like secondary architecture. A hybrid intercalation/nanoplating storage mechanism is observed in the C host, which enhances the surface negativity of C and ensures a strong binding between C and Li+. In particular, the active Li+ preintercalated into the graphite have intentionally reserved as optional Li source to offset any Li loss during cycling. The result is a uniform Li+ flux and a stabilized Li+ mass/charge transfer through the Li/electrolyte interface, which suppresses the dendrite growth and brings improved anode performance. The Li−CMN anode shows a high Li utilization of 95% and stable plating/stripping ability, enabling a long-life rechargeable Li metal battery if coupled with a LiFePO4 cathode. We hope our work will provide a new vision for the research in high-energy Li metal battery and contribute to the development of practical rechargeable metal batteries.

Figure 6. Interpretation of science behind the improved plating/ stripping performance of the CMN, including electron/charge distribution and Li deposition on (a) the planar graphite sheet, (b) the curved graphite sheet, and (c) the Li-intercalated curved graphite sheet. (d) Li stored in the Li/C compound acts as a backup source to offset the irreversible Li loss from the nanogaps during cycling.



which further leads to preferential Li growth on the graphite to form dendrites. In the case of a curved graphite in the spherical C, the electron atmosphere around each C atom changes, which interferes the free flow of the π electron so that it will be partially localized to the C atom (Figure 6b). Hence, the surface of curved graphite is negatively charged, which enhances its binding to the Li+ and contributes to a uniform Li+ flux. After Li intercalation to form the Li/C compound, an electron deviation from Li to C occurs,50 which further elevates the negativity of the C atoms on the curved graphite surface (Figure 6c). As a result, the C host shows a stronger binding to the Li+, and a large number of curved graphite sheets enable the relatively homogeneous Li+ flux distribution among the carbon granules, which improves the wettability of Li on C and thus contributes to a uniform Li deposition during the plating process. In this way, the mass/charge transfer through the Li/ electrolyte interface is stabilized, which paves the way for the enhanced Li stripping performance (Figure S22). Second, the active Li+ from Li/C intercalation compound could act as the optional Li source to offset the inevitable Li consumption during Li plating/stripping process. As demonstrated in Figure 6d, in the initial charge−discharge cycles, the metal Li predeposited in the nanogap could be reversibly stripped back into the electrolyte, and most of the Li+ in the compound remain their intercalation state. As the cycling process proceeds, the metal Li confined in the nanogaps will gradually dissove into the electrolyte, which induces a reversible deintercalation of Li+ from the compound to offset any capacity loss. This point has been proved by the evolution in galvanostatic discharge/charge voltage profiles of the Li− CMN|LiFePO4 full cell, which shows a gradually slopped plateau upon cycling (Figure S20b).51,52 With such a backup Li source, the long-term cycling performance of the Li−CMN anode is guaranteed. Last, the intact structure of CMN host and the confinement of nanogaps effectively suppress the SEI formation by electrolyte decomposition and the nanosized Li sandwiched by highly conductive graphite sheets triggers an ultrahigh electrochemical activity and a relatively small volume change, which together favors the long-term cycling stability.

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/jacs.7b01763. Ex situ SEM, TEM, XRD, XPS, and EIS and additional electrochemical results (Figures S1−S22) (PDF)



AUTHOR INFORMATION

Corresponding Authors

*[email protected] *[email protected] ORCID

Sen Xin: 0000-0002-0546-0626 Yu-Guo Guo: 0000-0003-0322-8476 Li-Jun Wan: 0000-0002-0656-0936 Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS This work was supported by the National Key R&D Program of China (Grant No. 2016YFA0202500), the National Natural Science Foundation of China (Grant Nos. 51225204, U1301244, and 21127901), the “Strategic Priority Research Program” of the Chinese Academy of Sciences (Grant No. XDA09010300), and the Chinese Academy of Sciences. We thank Prof. Dr. John B. Goodenough for his insights on the mechanism of Li plating/stripping on the carbon host.



REFERENCES

(1) Goodenough, J. B.; Kim, Y. Chem. Mater. 2010, 22, 587. (2) Xin, S.; Gu, L.; Zhao, N.-H.; Yin, Y.-X.; Zhou, L.-J.; Guo, Y.-G.; Wan, L.-J. J. Am. Chem. Soc. 2012, 134, 18510. (3) Zhang, C. F.; Wu, H. B.; Yuan, C. Z.; Guo, Z. P.; Lou, X. W. D. Angew. Chem., Int. Ed. 2012, 51, 9592. (4) Chen, Y.; Freunberger, S. A.; Peng, Z.; Bardé, F.; Bruce, P. G. J. Am. Chem. Soc. 2012, 134, 7952. (5) Gao, T.; Noked, M.; Pearse, A. J.; Gillette, E.; Fan, X.; Zhu, Y.; Luo, C.; Suo, L.; Schroeder, M. A.; Xu, K.; Lee, S. B.; Rubloff, G. W.; Wang, C. J. Am. Chem. Soc. 2015, 137, 12388. (6) Tarascon, J.-M.; Armand, M. Nature 2001, 414, 359. (7) Bruce, P. G.; Freunberger, S. A.; Hardwick, L. J.; Tarascon, J.-M. Nat. Mater. 2012, 11, 19.



CONCLUSION In conclusion, we are the first to realize regulation of mass/ charge transfer through the Li/electrolyte interface for 5921

DOI: 10.1021/jacs.7b01763 J. Am. Chem. Soc. 2017, 139, 5916−5922

Article

Journal of the American Chemical Society (8) Xu, W.; Wang, J.; Ding, F.; Chen, X.; Nasybulin, E.; Zhang, Y.; Zhang, J.-G. Energy Environ. Sci. 2014, 7, 513. (9) Luo, W.; Gong, Y.; Zhu, Y.; Fu, K. K.; Dai, J.; Lacey, S. D.; Wang, C.; Liu, B.; Han, X.; Mo, Y.; Wachsman, E. D.; Hu, L. J. Am. Chem. Soc. 2016, 138, 12258. (10) Aurbach, D. J. Power Sources 2000, 89, 206. (11) Yan, K.; Lu, Z.; Lee, H.-W.; Xiong, F.; Hsu, P.-C.; Li, Y.; Zhao, J.; Chu, S.; Cui, Y. Nat. Energy 2016, 1, 16010. (12) Cohen, Y. S.; Cohen, Y.; Aurbach, D. J. Phys. Chem. B 2000, 104, 12282. (13) Rosso, M.; Brissot, C.; Teyssot, A.; Dollé, M.; Sannier, L.; Tarascon, J.-M.; Bouchet, R.; Lascaud, S. Electrochim. Acta 2006, 51, 5334. (14) Lu, D.; Shao, Y.; Lozano, T.; Bennett, W. D.; Graff, G. L.; Polzin, B.; Zhang, J.; Engelhard, M. H.; Saenz, N. T.; Henderson, W. A. Adv. Energy Mater. 2015, 5, 1400993. (15) Lu, Y.; Das, S. K.; Moganty, S. S.; Archer, L. A. Adv. Mater. 2012, 24, 4430. (16) Bhattacharyya, R.; Key, B.; Chen, H.; Best, A. S.; Hollenkamp, A. F.; Grey, C. P. Nat. Mater. 2010, 9, 504. (17) Harry, K. J.; Hallinan, D. T.; Parkinson, D. Y.; MacDowell, A. A.; Balsara, N. P. Nat. Mater. 2014, 13, 69. (18) Yoshimatsu, I.; Hirai, T.; Yamaki, J. J. Electrochem. Soc. 1988, 135, 2422. (19) Yin, Y. X.; Yao, H. R.; Guo, Y. G. Chin. Phys. B 2016, 25, 018801. (20) Xu, K. Chem. Rev. 2014, 114, 11503. (21) Qian, J.; Adams, B. D.; Zheng, J.; Xu, W.; Henderson, W. A.; Wang, J.; Bowden, M. E.; Xu, S.; Hu, J.; Zhang, J. G. Adv. Funct. Mater. 2016, 26, 7094. (22) Liu, Y.; Lin, D.; Liang, Z.; Zhao, J.; Yan, K.; Cui, Y. Nat. Commun. 2016, 7, 10992. (23) Lin, D.; Liu, Y.; Liang, Z.; Lee, H.-W.; Sun, J.; Wang, H.; Yan, K.; Xie, J.; Cui, Y. Nat. Nanotechnol. 2016, 11, 626. (24) Zheng, G.; Lee, S. W.; Liang, Z.; Lee, H. W.; Yan, K.; Yao, H.; Wang, H.; Li, W.; Chu, S.; Cui, Y. Nat. Nanotechnol. 2014, 9, 618. (25) Liang, Z.; Zheng, G.; Liu, C.; Liu, N.; Li, W.; Yan, K.; Yao, H.; Hsu, P. C.; Chu, S.; Cui, Y. Nano Lett. 2015, 15, 2910. (26) Yan, K.; Lee, H. W.; Gao, T.; Zheng, G.; Yao, H.; Wang, H.; Lu, Z.; Zhou, Y.; Liang, Z.; Liu, Z.; Chu, S.; Cui, Y. Nano Lett. 2014, 14, 6016. (27) Li, W.; Yao, H.; Yan, K.; Zheng, G.; Liang, Z.; Chiang, Y. M.; Cui, Y. Nat. Commun. 2015, 6, 7436. (28) Kim, J.-S.; Kim, D. W.; Jung, H. T.; Choi, J. W. Chem. Mater. 2015, 27, 2780. (29) Ding, F.; Xu, W.; Graff, G. L.; Zhang, J.; Sushko, M. L.; Chen, X.; Shao, Y.; Engelhard, M. H.; Nie, Z.; Xiao, J.; Liu, X.; Sushko, P. V.; Liu, J.; Zhang, J.-G. J. Am. Chem. Soc. 2013, 135, 4450. (30) Qian, J.; Henderson, W. A.; Xu, W.; Bhattacharya, P.; Engelhard, M.; Borodin, O.; Zhang, J.-G. Nat. Commun. 2015, 6, 6362. (31) Lane, G. H.; Best, A. S.; MacFarlane, D. R.; Forsyth, M.; Hollenkamp, A. F. Electrochim. Acta 2010, 55, 2210. (32) Lu, Y.; Korf, K.; Kambe, Y.; Tu, Z.; Archer, L. A. Angew. Chem., Int. Ed. 2014, 53, 488. (33) Tu, Z.; Nath, P.; Lu, Y.; Tikekar, M. D.; Archer, L. A. Acc. Chem. Res. 2015, 48, 2947. (34) Miao, R.; Yang, J.; Xu, Z.; Wang, J.; Nuli, Y.; Sun, L. Sci. Rep. 2016, 6, 21771. (35) Qian, J.; Xu, W.; Bhattacharya, P.; Engelhard, M.; Henderson, W. A.; Zhang, Y.; Zhang, J.-G. Nano Energy 2015, 15, 135. (36) Lu, Y.; Tu, Z.; Archer, L. A. Nat. Mater. 2014, 13, 961. (37) Huang, C.; Xiao, J.; Shao, Y.; Zheng, J.; Bennett, W. D.; Lu, D.; Saraf, L. V.; Engelhard, M.; Ji, L.; Zhang, J.; Li, X.; Graff, G. L.; Liu, J. Nat. Commun. 2014, 5, 3015. (38) Li, N. W.; Yin, Y. X.; Yang, C. P.; Guo, Y. G. Adv. Mater. 2016, 28, 1853. (39) Xu, Z.; Wang, J.; Yang, J.; Miao, X.; Chen, R.; Qian, J.; Miao, R. Angew. Chem. 2016, 128, 10528.

(40) Liu, Q. C.; Xu, J. J.; Yuan, S.; Chang, Z. W.; Xu, D.; Yin, Y. B.; Li, L.; Zhong, H. X.; Jiang, Y. S.; Yan, J. M. Adv. Mater. 2015, 27, 5241. (41) Kim, H.; Jeong, G.; Kim, Y.-U.; Kim, J.-H.; Park, C.-M.; Sohn, H.-J. Chem. Soc. Rev. 2013, 42, 9011. (42) Aurbach, D.; Zinigrad, E.; Cohen, Y.; Teller, H. Solid State Ionics 2002, 148, 405. (43) Cheng, X. B.; Hou, T. Z.; Zhang, R.; Peng, H. J.; Zhao, C. Z.; Huang, J. Q.; Zhang, Q. Adv. Mater. 2016, 28, 2888. (44) Yang, C. P.; Yin, Y. X.; Zhang, S. F.; Li, N. W.; Guo, Y. G. Nat. Commun. 2015, 6, 8058. (45) Kaskhedikar, N. A.; Maier, J. Adv. Mater. 2009, 21, 2664. (46) Lin, M.-C.; Gong, M.; Lu, B.; Wu, Y.; Wang, D.-Y.; Guan, M.; Angell, M.; Chen, C.; Yang, J.; Hwang, B.-J.; Dai, H. Nature 2015, 520, 324. (47) Chen, Z.; Ren, W.; Gao, L.; Liu, B.; Pei, S.; Cheng, H.-M. Nat. Mater. 2011, 10, 424. (48) Hardwick, L. J.; Ruch, P. W.; Hahn, M.; Scheifele, W.; Kötz, R.; Novák, P. J. Phys. Chem. Solids 2008, 69, 1232. (49) Xu, K. Chem. Rev. 2004, 104, 4303. (50) Kganyago, K.; Ngoepe, P. Phys. Rev. B: Condens. Matter Mater. Phys. 2003, 68, 205111. (51) Dahn, J. R.; Zheng, T.; Liu, Y.; Xue, J. Science 1995, 270, 590. (52) Stevens, D. A.; Dahn, J. R. J. Electrochem. Soc. 2001, 148, A803.

5922

DOI: 10.1021/jacs.7b01763 J. Am. Chem. Soc. 2017, 139, 5916−5922