Electrolyte Interface of LiNi0.8Co0.15Al0

Aug 7, 2017 - The unstable electrode/electrolyte interface of high-capacity LiNi0.8Co0.15Al0.05O2 (NCA) cathodes, especially at a highly delithiated s...
8 downloads 20 Views 9MB Size
Research Article www.acsami.org

Stabilizing the Electrode/Electrolyte Interface of LiNi0.8Co0.15Al0.05O2 through Tailoring Aluminum Distribution in Microspheres as LongLife, High-Rate, and Safe Cathode for Lithium-Ion Batteries Peiyu Hou,† Hongzhou Zhang,‡ Xiaolong Deng,† Xijin Xu,*,† and Lianqi Zhang*,‡ †

School of Physics and Technology, University of Jinan, Jinan 250022, China Tianjin Key Laboratory for Photoelectric Materials and Devices, School of Materials Science and Engineering, Tianjin University of Technology, Tianjin 300384, China



S Supporting Information *

ABSTRACT: The unstable electrode/electrolyte interface of high-capacity LiNi0.8Co0.15Al0.05O2 (NCA) cathodes, especially at a highly delithiated state, usually leads to the transformation of layered to spinel and/or rock-salt phases, resulting in drastic capacity fade and poor thermal stability. Herein, the Alincreased and Ni-,Co-decreased electrode surface is fabricated through tailoring element distribution in micrometer-sized spherical NCA secondary particles via coprecipitation and solid-state reactions, aimed at stabilizing the electrode/ electrolyte interface during continuous cycles. As expected, it shows much extended cycle life, 93.6% capacity retention within 100 cycles, compared with that of 78.5% for the normal NCA. It also delivers large reversible capacity of about 140 mAh g−1 even at 20 C, corresponding to energy density of around 480 Wh kg−1, which is enhanced by 45% compared to that of the normal NCA (about 330 Wh kg−1). Besides, the delayed heat emission temperature and reduced heat generation mean remarkably improved thermal stability. These foregoing improvements are ascribed to the Al-increased spherical secondary particle surface that stabilizes the electrode/electrolyte interface by protecting inner components from directly contacting with electrolyte and suppressing the side reaction on electrode surface between high oxidizing Ni4+ and electrolyte. KEYWORDS: lithium-ion batteries, nickel-rich cathode material, microspheres, electrode/electrolyte interface, electrochemical properties earth, which has severely limited its application in EVs.7−9 In past decades, much effort has been devoted to develop the ideal cathodes with large capacity, long life-span, superior safety, and low cost.10−16 Among these candidates, Ni-based layered oxides, especially LiNi0.8Co0.15Al0.05O2 (NCA), have been commercialized as cathode materials of high-energy LIBs for EVs applications due to its large reversible capacity (∼200 mAh g−1).5,17−21 Nevertheless, it usually presents severe capacity fade with continuous cycles, less-than-ideal rate properties, and poor thermal stability.18,19,22−24 To unravel the degradation mechanisms of NCA electrodes, many studies have focused on the structure evolution, especially for changes in the surface structure during the initial and the subsequent cycles. Chang et al., Stach et al., Yang et al., and others have demonstrated that the structural evolution of NCA electrode first occurs on the particle surface, the elevated operating temperature and upper cutoff voltage accelerates the surface degradation.5,19−21 It has been confirmed that the

1. INTRODUCTION Lithium-ion batteries (LIBs) as high-energy power sources have been widely applied to “3C” portable electronic equipment; now, their application is gradually extended to large-capacity and high-power systems such as electric vehicles (EVs).1−4 However, some significant issues need to be solved before the widespread usage of LIBs in EVs. These LIBs in EVs are expected to have a service life of about 10 years rather than only a few years for portable devices. The acceleration process of EVs during driving requires higher-power density for LIBs than that of “3C” mobile instruments. In addition, a large battery pack is designed to ensure long driving mileage for EVs, which leads to potential safety issue owing to the exothermic reaction between highly delithiated cathode electrode and organic electrolyte.3−5 Cathode and anode electrodes act in an extremely crucial role in LIBs. In comparison to carbon anode with high reversible capacity exceeding 300 mAh g−1, long-term cyclability, and low cost,6 the commercial layered oxide LiCoO2 cathode delivers only about 140 mAh g−1, corresponding to about 0.5 Li+ reversible deintercalation/intercalation, and shows a high cost because of the rare cobalt element on the © 2017 American Chemical Society

Received: April 28, 2017 Accepted: August 7, 2017 Published: August 7, 2017 29643

DOI: 10.1021/acsami.7b05986 ACS Appl. Mater. Interfaces 2017, 9, 29643−29653

Research Article

ACS Applied Materials & Interfaces

CSTR at a constant rate of 4 mL h−1. During CSTR, NaOH solution (5 M) as precipitation agent and ammonia (1.0 M) as ligand, at constant pH (11.8), reaction temperature (50 °C), and stirring rate (600 rpm), were selected to synthesize micrometer-sized spherical precursors. After 25 h, these solutions in Tanks 1−3 were consumed simultaneously, achieving the target CG precursors [Ni0.8Co0.15Al0.05](OH)2. The normal [Ni0.8Co0.15Al0.05](OH)2 precursors were also prepared via a coprecipitation route. The as-prepared precursors mixed with stoichiometric ratio Li2CO3 (cationic molar ratio of Li−M = 1.03:1) and were calcined at 800 °C for 12 h under oxygen atmosphere to prepare NCA cathode. 2.2. Materials Characterization. The measurements of particle size distribution, packing density, and total chemical compositions of the as-prepared precursors were the same to our previous report.36 The calculation of energy density and SEM and XRD tests were the same as those in our previous study.37 X-ray photoelectron spectroscopy (XPS) measurements for evaluation of elemental species and the chemical states were carried out by Thermo Escalab250. To obtain the cycled NCA positive electrodes for SEM analysis, the cycled coin-type half cells at discharged state were carefully disassembled in glovebox under high-purity Ar atmosphere. The crystal structure was studied by high-resolution transmission electron microscopy (HRTEM, FEI Tecnai F20). The element content measurement on the cross section was performed by electron microprobe analysis (EMPA, Shimadzu EPMA-1600). Fullprof with WinplotR package was utilized for the Rietveld refinement of XRD data. 2.3. Electrochemical Measurements. The fabrication process of NCA electrodes was similar to that in our previous reports.38,39 Electrode disks with 0.5 cm radius and similar active oxide loading (2.5−3.0 mg cm−2) were punched out for assembling half-cells. The used counter electrode was lithium metal. We dissolved 1 M LiPF6 in EC and DMC solvent (1:1 by volume) to act as electrolyte. A LAND instrument (CT−2001A) was used to test the electrochemical properties. The differential scanning calorimetry (DSC) test was the same as that in our previous reserach40 except that it was charged 4.3 V (vs Li/Li+) at 0.1C (1C = 200 mA g−1) in this work. Electrochemical workstation (Zahner IM6ex) was utilized for electrochemical impedance spectra (EIS) measurement (frequency range, 100 kHz to 10 mHz).

surface evolution goes from the layered phase (R3m ̅ ) to the disordered spinel phase (Fd3̅m) and finally to the rock-salt phase (Fm3̅m). The surface rock-salt NiO-like layer causes the drastic rise of electrochemical impedance, which is responsible for the electrochemical behaviors fading, such as reversible capacity and rate capability, during charge−discharge cycles of NCA electrode.20−22 In situ time-resolved synchrotron X-ray diffraction and mass spectroscopy indicate that the aforementioned phase transformations are highly related to the release of both O2 and CO2 gases.20 As a result, the released O2 can accelerate thermal runaway by reacting with the organic electrolyte solvents as well as graphite anode, thereby resulting in safety issues.18,23−27 These foregoing findings clearly illustrate that the substantial capacity fade and safety issues of NCA electrodes highly depend on the instability of surface structure. Then fabricating a stable electrode/electrolyte interface for NCA electrode will be an effective means to overcome the significant hurdles before its widespread usage of LIBs in EVs. Previous reports focus on surface modification by inactive metal oxides28,29 and metal phosphates,30,31 as well as substitution via Na+, Mg2+, and F−.32−34 Unfortunately, the improved effect is less-than-ideal for the incomplete coating layer and the limitation of doping amount. Besides, the introduction of inactive components into NCA cathode generally will reduce its discharge specific capacity.28−35 Thus, the prospect of exploring and developing an innovative strategy to achieve large reversible capacity, durable cycle lifespan, and superior rate capability and thermal stability under the limitation of the constant Al-content for NCA cathode is an interesting direction that has yet to be realized. In this work, a concentration-gradient (CG-)NCA is rationally proposed, in which the Ni and Co concentrations decrease step by step while the Al content increases linearly from the interior to the surface of micrometer-sized spherical secondary particles. The Ni-increased inner component can deliver high discharge capacity, while the Al-increased outer surface is capable of providing stable electrode/electrolyte interface during cycles. The designed CG-NCA cathode is prepared via the combination of modified coprecipitation route and high-temperature solid-state reaction, and the electrochemical performances and surface evolution during cycles are studied in detail to demonstrate the feasibility of overcoming these foregoing significant challenges of high-capacity NCA cathode by stabilizing the electrode/electrolyte interface.

3. RESULTS AND DISCUSSION 3.1. Formation Mechanism of Concentration-Gradient Precursors. From Table S1, Al(OH)3 has much lower solubility product constant (Ksp) 4.57 × 10−33 than those of Co(OH)2 and Ni(OH)2,41 which indicates Al3+ and ligand NH3·H2O will immediately form sediment in aqueous solution rather than the desired complex compounds. The rapid precipitation of Al3+ easily generates undesired Ni−Co−Al layered double hydroxides (LDHs) caused by chargecompensating NO3− or SO42− ions which accompany the incorporated Al 3+ ions into the pure layered M(OH) 2 structure.42 In addition, the fast precipitation of Al3+ also leads to plenty of small Al(OH)3 crystal nuclei in the CSTR within a short time, which is harmful for the continuous growth of existing secondary particles. Thus, developing high-efficiency and stable chelating agents for Al3+ is essential for preparing micrometer-sized spherical precursors for high volumetric density NCA cathode. Recently, [(OBu)2(C5H7O4)]3− was demonstrated as a high-efficiency ligand for Al3+ (eq 1),43 which can be utilized to synthesize high packing density spherical precursors.

2. EXPERIMENTAL SECTION 2.1. Preparation of Precursors and LiNi0.8Co0.15Al0.05O2. The schematic diagram of preparing CG LiNi0.8Co0.15Al0.05O2 is depicted in Figure S1. To synthesize the aimed CG precursors for the designed NCA cathode, stoichiometric Ni(NO3)2·6H2O (232.64 g) and Co(NO3)2·6H2O (43.65 g) (cationic molar ratio of Ni−Co = 0.8:0.15) were dissolved in 0.5 L of deionized water in Tank 1 to obtain 2.0 M solution as the first starting aqueous solution. Next, 12.36 g of aluminum trisecbutoxide (Al(OBu)3) was dissolved in mixture of 5.4 mL of acetylacetone (C5H8O4) and 44.6 mL of absolute ethyl alcohol in Tank 3 as the second starting material. Additionally, 50 mL of absolute ethylalcohol was added into Tank 2 as the third starting material. At the initial stage of reaction, Ni−Co aqueous solution (Tank 1) was continuously pumped into the continuously stirred tank reactor (CSTR, capacity of 5 L) at a constant rate of 20 mL h−1. Meanwhile, the solution in Tank 3 was added into Tank 2 at a constant rate of 2 mL h−1 to achieve an increasing Al concentration in Tank 2, and the formed solution in Tank 2 was also dropped into the

Al3 + + [(OBu)2 (C5H 7O4 )]3 − ↔ Al(OBu)2 (C5H 7O4 ) (1) 29644

DOI: 10.1021/acsami.7b05986 ACS Appl. Mater. Interfaces 2017, 9, 29643−29653

Research Article

ACS Applied Materials & Interfaces

Figure 1. Growth mechanism of Al-increased and Ni-, Co-decreased micrometer-sized spherical secondary precursor particles during coprecipitation process in the CSTR.

Figure 2. SEM images of the (a, b) CG precursors and (c, d) CG-NCA, as well as (e) cross section of a single CG-NCA particle and (f) EMPA results on this cross section.

ratio of Al/(Ni−Co) in outside tanks and the unique growth mechanism of spherical secondary particles inside CSTR. 3.2. Material Characterization. The target CG precursors have Ni and Co contents which decrease continuously, while Al content increases gradualy from the inner center to the outer surface of particles are initially synthesized. SEM images of CG and the normal precursors [Ni0.8Co0.15Al0.05](OH)2 are presented in Figures 2a,b, S2a, and S3a−c. Both the precursors reveal micrometer-sized spherical or ellipsoidal morphology assembled by nanosized sheet-like primary grains. The particle size distribution of both precursors in Figures S2b and S3d indicates a general Gaussian distribution. The average particle size D50 is 7.61 μm for the CG precursors, slightly larger than 7.36 μm of the normal precursors. XRD results demonstrate that both precursors mainly are β-Ni(OH)2 type layered phase from Figure S4, however LDHs structure is also detected due to the rapid precipitation of Al3+.44 The cationic molar ratio of Ni−Co−Al within both precursors is measured by ICP−AES, and the corresponding data are presented in Table S2. Note that the molar ratio of Ni−Co−Al in both as-synthesized precursors correspond to the target value (0.8/0.15/0.05). The as-prepared CG and normal precursors are mixed at a stoichiometric ratio with lithium carbonate and heated to 800 °C to form NCA cathodes. SEM images of both NCA particles in Figures 2c,d and S5 confirm the spherical particles are maintained even after the high-temperature calcination process, but the primary grains of both NCA show significant discrepancy compared with that of precursors. It appears to have polyhedral shapes with a size of about 500 nm. These monodispersed microsized spherical particles result in a highpacking density of 2.12 and 2.16 g cm−3 for CG and normal

Al(OBu)2 (C5H 7O4 ) + 3H 2O ↔ Al(OH)3 ↓ + C5H8O4 + (HOBu)2

(2)

M2 + + n NH3 ↔ [M(NH3)n ]2 +

(3)

[M(NH3)n ]2 + + 2OH− ↔ M(OH)2 ↓ + n NH3

(4)

First, to achieve increasing cationic ratio of Al/(Ni−Co) in the outside tanks, the Al(OBu)2(C5H7O4) solution (Tank 3) is continuously added into Tank 2 at a constant rate of 2 mL h−1 during reaction, as revealed in Figure S1. At the same time, the Ni−Co solution in Tank 1 is also dropped into the CSTR at a constant rate of 25 mL h−1. As a result, as the reaction time increases, the Al concentration in Tank 2 will enrich gradually. The corresponding molar ratio of Al/(Ni−Co) dumped into CSTR will also be improved step by step. Second, after the increasingly improved molar ratio of Al/(Ni−Co) is dropped into the CSTR, tiny grains will be first formed, in which Al3+ only presents precipitation reaction (eq 2) and Ni2+ and Co2+ are involved in a complexing reaction (eq 3) and coprecipitation reaction (eq 4). These small grains will be consumed and grow on the formed secondary particles in favor of larger particles to minimize surface free energy via dissolution−recrystallization, as depicted in Figure 1. Then, with increasing reaction time in the CSTR, the target CG microsphere precursors are assembled by numerous tiny primary grains with Al-increased and Ni-,Co-decreased content. In brief, the formation mechanism for CG structure in precursors depend on the artificially tunable cationic molar 29645

DOI: 10.1021/acsami.7b05986 ACS Appl. Mater. Interfaces 2017, 9, 29643−29653

Research Article

ACS Applied Materials & Interfaces

Figure 3. XRD patterns and Rietveld refinements of (a) normal NCA and (b) CG-NCA cathode materials.

Table 1. Refined Lattice Parameters, Degree of Li/Ni Mixing, and R Factors for Both NCA Cathodes before Charging lattice parameters samples

a [Å]

c [Å]

c/a

V [Å3]

Li/Ni mixing

χ2

Rp [%]

Rwp [%]

CG-NCA normal NCA

2.8638 2.8645

14.1731 14.1688

4.9491 4.9463

100.665 100.687

0.82 1.74

1.65 2.04

7.45 8.18

10.3 11.5

Figure 4. HRTEM images of (a) normal NCA and (b) CG-NCA cathode materials; insets are the corresponding fast Fourier transform (FFT) results.

Figure 5. CVs curves of (a) normal NCA and (b) CG-NCA between 3.0 and 4.3 V (vs Li/Li+) at a scan rate of 0.1 mV s−1.

NCA cathodes, respectively. After lithiation, the compositional changes on the cross section of a single CG-NCA particle are measured by electron microprobe analysis (EMPA). The relative contents of Ni, Co, and Al are plotted as a function of distance from the center to the surface in a single NCA

particle. The cross-sectional image and EMPA spot scanning results for this cross section are displayed in Figure 2e,f. Clearly, the Ni and Co concentrations decrease gradually, while the Al content increases linealy from the center to the surface. Additionally, the arrangement between primary grains is 29646

DOI: 10.1021/acsami.7b05986 ACS Appl. Mater. Interfaces 2017, 9, 29643−29653

Research Article

ACS Applied Materials & Interfaces

Figure 6. (a) Initial charge−discharge curves, (b) cycle life of capacity and (c) energy density at a rate of 0.1C and 25 °C. (d) Cycle life of capacity and (e) 300th charge−discharge curve at a rate of 1C and 25 °C. (f) Cycle life of capacity at 50 °C for the normal NCA and the CG-NCA electrodes.

Figure 7. (a) Rate discharge capacity, (b) median-potential, (c) power density, and (d, e) charge−discharge curves for the normal NCA and the CGNCA cathodes at various rates and 25 °C. (f) Comparative rate capability of this work and the recent studies35,52−56 of NCA cathode materials for LIBs.

NCA, respectively. The slightly enlarged (003) interplanar spacing of the CG-NCA is consistent to the larger lattice parameter c from the XRD refinement results in Table 1. 3.3. Electrochemical Performances. Cyclic voltammograms (CVs) of CG and normal NCA electrodes at a scan rate of 0.1 mV s−1 are depicted in Figure 5. The CVs curves of both NCA are similar, in which three pairs of redox peaks are observed, caused by phase transformations from the pristine hexagonal to monoclinic phase (H1 to M), monoclinic to hexagonal phase (M to H2), and hexagonal to hexagonal phase (H2 to H3) during the charge−discharge process.48 The intensities of the two latter peaks weaken slightly for the CGNCA, indicating the reduction of significant lattice volume changes caused by M/H2 and H2/H3 phase transitions, which is beneficial for suppressing the micrometer-strain of spherical secondary particles.49,50 The initial and 300th cycle curves, cycle life of capacity, and energy density for CG and normal NCA electrodes between 3.0 to 4.3 V (vs Li/Li+) at 25 and 50 °C are presented in Figure 6. Obviously, both NCA electrodes deliver high discharge specific

extremely tight, and porosity is hardly observed inside the particle in Figure 2e. Even the particle surface exhibits some gaps in Figure 2d. XRD patterns and Rietveld refinements results of CG and normal NCA materials are revealed in Figure 3 and Table 1. The XRD diffraction peaks indicate both NCA materials are indexed to a well-defined hexagonal α-NaFeO2-type structure (space group R3̅m). Furthermore, the adjacent peaks of (006)/ (102) and (018)/(110) show clear split, suggesting a typical layered structure.45 For the CG-NCA sample, the lattice parameter a decreases, while the lattice parameter c value increases. Therefore, based on above results and previous investigations,46,47 it can be deduced that the doped Al3+ form solid solution with Ni3+ and Co3+ in the transition metal 3a sites and enhance the Li+ diffusivity. Additionally, it should be noted that CG-NCA has a decreased degree of Ni ions occupancy in Li ion sites compared with that of the normal NCA. From the HRTEM images in Figure 4, a continuous interference fringe spacing of the (003) crystal plane with distance 0.473 and 0.472 nm are detected for CG and normal 29647

DOI: 10.1021/acsami.7b05986 ACS Appl. Mater. Interfaces 2017, 9, 29643−29653

Research Article

ACS Applied Materials & Interfaces capacity of about 200 mAh g−1 with high initial Coulombic efficiency of ∼93% from Figure 6a. As expected, the CG-NCA electrode reveals much improved Li+ intercalation/deintercalation stability compared to that of the normal NCA at 25 °C, as shown in Figure 6b,c. After 100 cycles, a large reversible capacity of 188.6 mAh g−1 with superior capacity retention of 93.6% is maintained for the CG-NCA electrode. On the contrary, the normal NCA electrode suffers from gradual capacity decay, resulting in the low reversible capacity of 156.7 mAh g−1 with inferior capacity retention of 78.5% over the same cycle period. In addition, both NCA electrodes have a high energy density of around 760 Wh kg−1 during the initial cycle; however, the CG-NCA maintains the energy density of 718 Wh kg−1 after 100 cycles, much higher than that of 591 Wh kg−1 for the normal NCA. To further evaluate the structure stability of the CG-NCA electrode, the capacities as a function of the cycle number at a high rate of 1C and high working temperature of 50 °C are presented in Figure 6d−f, in which the durable cyclability and lowered electrochemical polarization are also achieved for the CG-NCA electrode. Thus, the fabrication of gradually increased Al content in the CG-NCA microsphere cathode is capable of stabilizing the cycling stability during continuous charge−discharge process. Air stability of both NCA cathodes is preliminarily evaluated by comparing the initial Coulombic efficiency, reversible capacity, and charge/discharge curves after storing NCA materials in air for 1 day, as revealed in Figure S6. It is confirmed that the higher reversible capacity, initial Coulombic efficiency, and lower polarization at 3.6−3.8 V (vs Li/Li+) are achieved for the CG-NCA cathode, indicating the improved air stability. High power density is also an extremely significant role for practical applications of LIBs in EVs.51 The rate capacity and power energy density of both NCA cathodes are measured as the discharge rate ranges from 0.1 to 20 C at 25 °C, while the charge rate is fixed at 0.5 C, as seen in Figure 7a−e. Both NCA electrodes deliver similar discharge specific capacity of ∼200 mAh g−1 at 0.1 C from Figure 7a. However, as the C rates increase, the CG-NCA electrode exhibits a higher rate capacity than the normal NCA due to the Al-increased electrode surface and the increased lattice parameter of c value.35 Importantly, the CG-NCA electrode shows large reversible capacity of about 140 mAh g−1 even at a high rate of 20 C, corresponding to about 70% capacity retention of 0.1 C. In addition, as seen in Figure 7b, the CG-NCA electrode also reveals higher operating median-potential as increases the discharge rates compared with that of the normal NCA. As a result, much improved power density is achieved for the CG-NCA electrode, as in Figure 7c. The CG-NCA electrode delivers an energy density of around 480 Wh kg−1 at 20 C, which is enhanced by 45% compared to that of the normal NCA sample (about 330 Wh kg−1). It is clear that the increasingly electrochemical polarization are observed for both NCA electrodes from 0.1 to 20 C in Figure 7d,e, whereas the CG-NCA electrode displays slowed polarization, corresponding to the higher discharge median potential than the normal NCA. Additionally, the rate capability of both NCA electrodes from 0.1 to 10 C charge/ discharge rates is also measured, as seen in Figure S7. The results demonstrate that the as-prepared CG-NCA cathode also exhibits the enhanced high charge/discharge rate capability compared with that of the normal NCA cathode. Notably, by contrast with these recent reported modified NCA electrodes,35,52−56 the CG-NCA electrode in this work presents better

rate properties from Figure 7f, especially at high rates of above 5 C. Ni-based layered oxide cathodes usually show inferior thermal stability in that the released oxygen especially at highly delithiated state and high temperature will accelerate thermal runaway. The thermal characteristics of both NCA electrodes can be measured and evaluated by DSC after the electrodes are initially charged to 4.3 V (vs Li/Li+) in Figure 8. Both DSC

Figure 8. DSC profiles of the delithiated both NCA electrodes after being initially charged to 4.3 V (vs Li/Li+).

profiles display the normal distribution curves, in which the normal NCA electrode presents a sharp peak pattern while the CG-NCA electrode exhibits a broadened one. In contrast with the sharp peak with drastic heat generation, the latter one corresponds to a gradual heat generation process. The normal NCA electrode exhibits a peak temperature of 238.9 °C, whereas the CG-NCA sample shows enhanced peak temperature (245.2 °C), indicating that drastic thermal runaway reaction is delayed. Furthermore, the CG-NCA electrode has heat generation of 1016.2 J g−1, notably reduced heat generation compared with that of normal NCA sample (1573.6 J g−1). These results demonstrate the remarkably improved thermal stability for the CG-NCA electrode, which is attributed to the Al−O bonding (512(4) kJ mol−1) being stronger than the Ni−O bonding (391.6(38) kJ mol−1) and Co−O bonding (368(21) kJ mol−1).57 This suggests the CGNCA with a gradually increased Al content from interior to surface of secondary particles will be a promising cathode for safe LIBs. 3.4. Surface Evolution of Secondary Particles. To explore and understand the origination of enhanced performances for the CG-NCA cathode, we first measure Ni and Co dissolution in electrolyte after various cycles to evaluate the side reaction between high oxidative transition-metal ion (Ni4+ and Co4+) and organic electrolyte. The amount of Ni and Co dissolution in electrolyte after 20, 50, and 100 cycles for both NCA electrodes is described in Figure 9. Obviously, the normal NCA electrode has much higher Ni and Co dissolution during various cycles than that of the CG-NCA. Specially, the normal NCA electrode exhibits severe Ni dissolution after 100 cycles, up to 134 ppm, which exceeds two times the Ni dissolution of the CG-NCA electrode (61 ppm) over the same cyclic period. These reduced Ni and Co dissolution indicates the slowed side reaction between particle surface and electrolyte during charge−discharge process. 29648

DOI: 10.1021/acsami.7b05986 ACS Appl. Mater. Interfaces 2017, 9, 29643−29653

Research Article

ACS Applied Materials & Interfaces

Figure 9. (a) Ni and (b) Co concentrations dissolved in electrolyte after 20, 50, and 100 cycles for the CG- and normal NCA electrodes at a rate of 0.1C and 25 °C.

Figure 10. SEM images of the cycled (a−c) CG-NCA and (d−f) normal NCA electrodes in the fully discharged state after 100 cycles at a rate of 0.1C and 25 °C.

transition metal fluoride NiF2 are detected at the surface of the cycled normal NCA electrode, whereas the major composition LiNi1−xMxO2 is observed for the cycled CG-NCA particles. We believe that the Ni-enriched SEI-like surface (>5 nm) probably forms owing to the side reactions between oxidizing Ni4+ and organic electrolyte for the cycled N-NCA; thus, Ni3+ originated from the host structure of NCA is not detected in Figure 11a. For the CG-NCA, the Al-increased surface suppresses the said side reactions and produces thin SEI-like surface (