Structural evolution and capacity degradation mechanism of LiNi0.6Mn0.2Co0.2O2 cathode materials

Journal of Power Sources 400 (2018) 539–548

Contents lists available at ScienceDirect

Journal of Power Sources journal homepage: www.elsevier.com/locate/jpowsour

Structural evolution and capacity degradation mechanism of LiNi0.6Mn0.2Co0.2O2 cathode materials

T

Yanli Ruana,b,c,∗, Xiangyun Songc, Yanbao Fuc, Chengyu Songd, Vincent Battagliac,∗∗ a

State Key Laboratory of Separation Membranes and Membrane Processes, Tianjin, PR China School of Environmental and Chemical Engineering, Tianjin Polytechnic University, Tianjin, PR China c Energy Storage and Distributed Resources Division, Lawrence Berkeley National Laboratory, Berkeley, CA, 94720, USA d National Center for Electron Microscopy, Lawrence Berkeley National Laboratory, Berkeley, CA, 94720, USA b

H I GH L IG H T S

capacity degradation of NMC 622 depends on the charge cut-off voltage. • The structural transition leads to degradation at 4.5 V or below. • Surface cracking accompanied with phase changes causes degradation at 4.8 V. • Particle • An insulating surface film also contributes to the capacity degradation at 4.8 V.

A R T I C LE I N FO

A B S T R A C T

Keywords: Structural evolution Degradation mechanism Cut-off voltage Interfacial impedance Particle cracking

LiNi0.6Mn0.2Co0.2O2 is a promising cathode material with a high capacity for Li-ion batteries. However, the rapid capacity degradation in the high-voltage cycles constrain their further applications. Accordingly, the performances of LiNi0.6Mn0.2Co0.2O2 have been systematically investigated using various microstructural characterizations as well as electrochemical analyses to explore its degradation mechanism. Our results indicate that the capacity decay of LiNi0.6Mn0.2Co0.2O2 strongly depends on the charge cut-off voltage. For the cell that is cycled at 4.2 or 4.5 V, the degradation mechanism is primarily due to transformation from layered to rock salt structure on the particle surface, increasing the charge transfer impedance. For the cell that is cycled at 4.8 V, another two reasons should be considered. The irreversible structural change in the bulk lattice of LiNi0.6Mn0.2Co0.2O2 during the high-degree delithiation process eventually disintegrates the secondary particles, resulting in the poor electrical contact between particles. Another one is that the insulating surface film which is generated on the surface of particles after cycling at 4.8 V increases the interfacial impedance of LiNi0.6Mn0.2Co0.2O2. All these factors contribute to the overall capacity degradation at high voltages.

1. Introduction High-energy density and high-power density lithium-ion batteries (LIBs) are basic necessities for extensive applications of consumer electronics and electric vehicles [1–3]. In order to meet the rapid development of the energy-demanding devices, more efforts have been made to develop electrode materials with large specific capacities at high voltages [4–6]. Compared with cathode materials of olivine- (ca. 170 mAh g−1) and spinel-structured (ca. 150 mAh g−1) materials, the layered cathode material of LiNixMnyCozO2 (NMC, x + y + z = 1) has a larger theoretical specific capacity of 280 mAh g−1 [7,8]. The NMC-

∗

based materials are essentially a solid solution of LiNiO2, LiMnO2, and LiCoO2, where nickel (Ni) provides high capacity, cobalt (Co) improves the rate capability, and manganese (Mn) increases the structural stability [9]. Ni-rich NMC compounds (x ≥ 0.5 in LiNixMnyCozO2), such as LiNi0.5Mn0.3Co0.2O2 (NMC 532), LiNi0.6Co0.2Mn0.2O2 (NMC 622), and LiNi0.8Co0.1Mn0.1O2 (NMC 811) are promising cathode materials, given that an extra capacity can be attained by charging the cells to a high voltage (≥4.5 V vs. Li/Li+) [10]. However, the intrinsic thermal instability of the Ni-rich NMC materials during the delithiation process as well as their structural instability when Ni4+ is reduced to Ni2+ always decreases their capacity [11]. Many efforts have been devoted to

Corresponding author. State Key Laboratory of Separation Membranes and Membrane Processes, Tianjin, PR China. Corresponding author. E-mail addresses: [email protected] (Y. Ruan), [email protected] (V. Battaglia).

∗∗

https://doi.org/10.1016/j.jpowsour.2018.08.056 Received 6 May 2018; Received in revised form 15 August 2018; Accepted 19 August 2018 0378-7753/ © 2018 Elsevier B.V. All rights reserved.

Journal of Power Sources 400 (2018) 539–548

Y. Ruan et al.

improvement for the NMC622 material.

controlling the capacity degradation of Ni-rich compounds by modifying the preparation method [10,12,13]. A comprehensive understanding on the degradation mechanism would be fundamentally conducive to the improvement of the electrochemical performances of those Ni-rich NMC materials. So far, several degradation mechanisms regarding the unstable structure of Ni-rich materials have been identified. Cation migration is the first reason that decreases the capacity of the Ni-rich NMC cathode. During the charging process, Li+ is extracted from the layered structure of NMC and transitional metal (TM) ions, especially Ni2+ (0.069 nm) with a size close to Li+ (0.076 nm) will migrate to the lithium layer and occupy the vacant Li+ sites. Such a cation migration gives rise to an irreversible phase transformation of the Ni-rich NMC from a layered structure to a spinel structure or rock-salt structure, thus leading to the capacity degradation [6,14,15]. On the other hand, the migration of TM ions at the particle surface might facilitate their dissolution into the electrolyte, which has been reported for other NMC materials in the literature [8,16,17]. Formation of an undesirable surface film is the second reason that degrades cathode materials. Under high charge voltage serious oxidative decomposition of the traditional LiPF6/carbonate electrolyte can generally occur [10,18]. The accumulation of side products generated from the decomposition lead to continuous growth of the surface film on the Ni-rich NMC particles, resulting in poor cycling performance and a low coulombic efficiency of the LIBs [19]. Moreover, after the full charge, Ni4+ in NMC is unstable and can be readily reduced to Ni2+, forming an insulating NiO on the surface of NMC due to the loss of oxygen [20]. These undesirable surface films increase the impedance and deteriorate the electrochemical performance of NMC. Mechanical stress is the third reason that decreases the capacity of NMC. The solid-state diffusion and transport of Li ions induces mechanical stresses and volume changes in electrode particles during the charge and discharge process [21], which may cause particle cracking and disintegration [22]. As a result, the previously formed cathode electrolyte interphase (CEI) will be disrupted and then be rebuilt by consuming active lithium from the electrolyte. Any loss of active lithium would enhance the layered oxides degradation [23]. All the above reasons contributing to the capacity degradation of NMC usually interact and mutually aggravate, which make the studies intractable [24]. In addition, the aforementioned reasons are very sensitive to the composition of the electrode material and strongly depend on the cycling and storage conditions of the batteries [25]. Therefore, each LIB system shows its unique fatigue features which is too complex to be comprehensively predicted by any model, and thus needs to be investigated separately [26]. For instance, the degradation mechanism involved in the NMC532 material has been systematically studied and found to be different from that of the LiNi1/3Co1/3Mn1/3O2 (NMC111) material [8,20,26], indicating that the degradation mechanisms of NMC may vary with different contents of Ni, Co and Mn. In the Ni-rich NMC materials, NMC622 is a promising material which possesses not only the high capacity, close to that of NMC811, but also the good thermal stability comparable to NMC532. However, compared with other commercialized NMC materials, investigations on the degradation mechanism of NMC622 are relatively rare [11,15]. Particularly, the origin of the accelerated degradation of cycle performance and a change of the critical fade mechanism with the cut-off voltage still need to be further investigated. In this work, the degradation mechanism of NMC 622 after a long time cycling is systematically investigated by analyzing structural changes of NMC 622 at different cut-off voltages. Various micro-structural analyses such as SEM、TEM、STEM and EELS are applied to elucidate the structural changes on the surface of the cathode material. XRD、TG-DSC and electrochemical techniques are used to evaluate changes in the bulk sample, which are related to the evolution of Ni2+/ Ni3+/Ni4+. It is expected that a solid understanding on the degradation mechanisms can be definitely shed light on the performance

2. Experimental 2.1. Material and electrode preparation Spherical, micron-sized LiNi0.6Mn0.2Co0.2O2 (NMC622) active material without any surface modification was obtained from Umicore and stored in an argon-filled glove box before use. The cathode was prepared by 92.8 wt% NMC622, 3.2 wt% carbon-black (Denka Black) and 4 wt% polyvinylidene fluoride binder (PVDF, Aldrich) binder. All the powders were mixed in N-methyl-2-pyrrolidone (NMP, 99.5%, Aldrich) using a homogenizer at 3000 rpm (Polytron PT 10–35) to form a uniform slurry. The slurry was then cast onto a Al-foil with a doctor-blade to form the cathode. After that, the cathode was dried in an antechamber to a glovebox under vacuum at 130 °C for 12 h. The loading density of the active material in cathode is about 5 mg/cm2. Porosity of the electrode is about 40%.

2.2. Characterization The structure of the NMC622 samples were examined by powder Xray diffraction (XRD; Bruker, Germany) using Cu Kα radiation (λ = 0.154 nm) at a scanning speed of 1° min−1 in the 2θ range of 10–70°. The thermal analysis for the charged cathode at different depth of charge was carried out using a TG-SDT by Q600 instruments (USA) from room temperature to 600 °C with a heating rate of 5 °C/min. The particle morphology and element distribution of the cathode material were analyzed by scanning electron microscopy (SEM; Philips XL30ESEM) and energy dispersive spectroscopy (EDS). High resolution HRTEM/STEM imaging analyses and electron energy loss spectroscopy (EELS) were carried out using a probe Cs corrected JEM-2100 F at 200 kV. EELS spectra were obtained from the outermost surface region (2 nm) with an acquisition time of 2 s. The energy resolution in EELS was about 0.9 eV, which was sufficient to investigate the fine edge structures of the O K-edge and the TM (Ni, Mn, and Co) L-edges.

2.3. Electrochemical performance characterization The electrochemical tests were performed using 2325 coin-type cells, which were composed of NMC622 cathode, lithium metal anode, Celgard 2400 separator, and standard electrolyte of 1 M LiPF6 dissolved in ethylene carbonate (EC) and diethyl carbonate (DEC) solvent (1:2 mass ratio, Daikin Industries, America). Galvanostatic charge-discharge cycling tests for LIBs using the NMC622 cathode were conducted on a Maccor (series 4000) multichannel battery test system in different voltage ranges of 2.8–4.2 V, 2.8–4.5 V, and 2.8–4.8 V. The LIBs receive the nomination according to the depth of charge during cycling, which 4.2 V-cell, 4.5 V-cell, and 4.8 V-cell represents the LIB to be cycled under the cut-off charge voltage of 4.2 V, 4.5 V and 4.8 V, respectively. The current for all the cells on charge and discharge was set at 0.1C. (1C = 160 mA g−1). After the cycling test, the cells were transferred back into the glovebox and disassembled for the ex-situ analyses. The cathodes were washed with DMC solvent for several times to remove the residual electrolyte thoroughly, and then dried under vacuum. Cyclic voltammetry (CV) tests were carried out on a Bio-Logic SA electrochemical workstation within a potential range of 2.8–4.9 V. Electrochemical impedance spectroscopy measurements were carried out using a Bio-Logic VMP3 electrochemical workstation. After all the test cells were discharged to 3.73 V and maintained at this voltage for 2 h, the impedance of the three cells were collected in the frequency range of 200 kHz to 0.01 mHz.

540

Journal of Power Sources 400 (2018) 539–548

Y. Ruan et al.

Fig. 1. Electrochemical performance and impedance of the samples reveal the capacity fading strongly depends on the cut-off voltage. Charge/discharge voltage profiles (a-c) and cycle performance (d) of NMC 622 half cells cycled under different charge cut-off voltages. The discharge curves after the first cycle (e) and 100 cycles (f). Nyquist plots of the three samples collected after 100 cycles at discharge state at 3.73 V (g). Inset of (g) is an expanded view of the impedance data in high frequency region.

charge and discharge capacities of 267 and 237 mAh g−1, respectively. However, it also exhibits serious voltage decay and capacity decay after 100 cycles (58% capacity retention from 237 to 155 mAh g−1). Fig. 1a–d clearly show that the capacity of 4.5 V-cell improves to some extent for the first 30 cycles and retained relatively consistent in the first 50 cycles. The 4.8 V-cell exhibits significant and prompt capacity and voltage decay, suggesting different degradation mechanisms may be involved in the cells cycled under various depth of charge. Fig. 1e and f displays more details about the evolution of discharge curves with cycling by moving discharge curves to the same ending point of capacity. It can be seen that all the discharge curves in Fig. 1e display voltage plateaus at 3.73 V in the first discharge, which is calculated from a plot of d2Q/dV2 vs. voltage (Fig. S2). The voltage profiles are nearly overlapped from the plateaus to the end of discharge in the middle and high capacity region (about 76 mAh g−1), reflecting the over-potential of Li+ intercalation reaction and the cell impedance [25]. The part of discharge voltages above the plateau in Fig. 1e

3. Results and discussion 3.1. High cut-off voltage cycling induced capacity degradation The electrochemical performance of the NMC622 cathode was evaluated using coin-type 2325 R lithium half-cell hardware by cycling at different cut-off voltages. Fig. 1a–c shows the charge/discharge voltage profiles during cycling. Fig. 1d shows the difference in cycle performance of the cells. Clearly, the result indicates that the cycling stability of the NMC622 electrode strongly depends on the charge cutoff voltage, which shows that the higher the cut-off voltage to be applied, the faster the cell performance decays. This trend is the same to the rate performance of cells as well (Fig. S1). For instance, 4.2 V-cell shows a high capacity retention of 98% after 100 cycles (from 161 to 158 mAh g−1), better than 4.5 V-cell, which displays a 91% capacity retention (from 197 to 178 mAh g−1) after the same number of cycles. Increasing the depth of charge, 4.8 V-cell delivers the increased initial 541

Journal of Power Sources 400 (2018) 539–548

Y. Ruan et al.

process of the layered TM oxides, deintercalation of Li+ from the lattice usually causes lattice expansion along the c direction and contraction along the a and b directions, which is reversed upon discharging process [30,31]. Such an observation appears to be more significant at higher temperatures and higher charge voltages [32]. The increased c/a values in Table 2 indicate a lattice distortion. Given all the XRD samples including the pristine and cycled cathodes were measured at a samely discharged state, the increased c/a values may suggest an irreversible change in the internal structure of NMC622. The intensity ratio of the (003) to (104) peaks, designated as R, corresponds to the degree of cation mixing. A low value of R implies a high degree of cation mixing [15,33,34]. In Fig. 2 and Table 2, the R values decrease from 1.89 to 1.69 when the charge cut-off voltage is increased from 4.2 to 4.8 V, demonstrating that the cation mixing is aggravated with increasing cutoff voltage. Previous studies also suggested that cation mixing could bring about a phase transformation from layered to spinel structure or an electrochemically inactive NiO-like phase on the surface of the electrode [25,34]. Thus, XRD results reveal the occurrences of irreversible structural changes in both surface and bulk of the NMC622 cathode after it was cycled at high charge cut-off voltages. This irreversible structural change may reduce the capability of Li+ intercalation and deintercalation in the NMC622 cathode. Fig. 3 presents morphology and composition of the cycled cathode surface, obtained by SEM and EDS analysis. There are some micro-pores on the surface of the cathode for 4.2 V-cell, which can be penetrated by the electrolyte, as seen in Fig. 3a1. Some micro-cracks can be observed in the cathode of 4.5 V-cell (Fig. 3b1). Furthermore, the fatigued cathode cycled under 4.8 V exhibits obvious intergranular cracks (Fig. 3c1). For the cathodes composed of layered TM oxides, applying secondary particles densely packed by primary particles can increase the volumetric energy density of the electrode. However the primary particles may have different crystallographic orientation and expanse or contract anisotropically during lithium intercalation or deintercalation, which results in intergranular cracks of the secondary particle [23,32,35,36]. This is more obvious for the cells cycled under higher charge cut-off voltages. High cut-off voltage can lead to significant changes in lattice volume. The non-uniform accommodation of such a volume change will generate severe local strain, which can lead to mechanical failure and intergranular cracks. The evolution of intergranular cracks may cause poor connections between the NMC622 particles, large contact resistance and even the loss of active material [23,37]. This fracturing can create fresh surfaces with the formation of additional solid electrolyte interphase (SEI) layers, and generate new sites for side reactions, consequently increasing interfacial impedance (Fig. 1g) [15,38]. It is noted that some of the primary particles in the cathode of 4.8 V-cell (Fig. 3c2 and Fig. 3c3) are exposed directly to electrolyte. Fig. 3d and e provide enlarged pictures of two different areas of the electrode cycled under 4.8 V. EDS images in Fig. 3f confirm that the contents of TM cations are comparable in the two marked regions in Fig. 3d and e. Additionally, the EDS analysis (Fig. S6) from randomly selected areas (about 10 regions) in Fig. 3d and e further show the content ratio of Mn, Co and Ni is basically in consistent with that of the original NMC622 particles (1:1:3.1), indicating the dissolution of TM ions for the 4.8 V-cell is not significant. Thus it is speculated that the main reason of the capacity degradation for NCM622 electrode cycled at high voltage may be caused by some factors else, such as the interfacial evolution, rather than the loss of TM ions. More detailed observation on the microstructure of the cycled electrodes was carried out with HRTEM, as shown in Fig. 4. Fig. 4a2 shows that the cathode of 4.2 V-cell after 100 cycles maintains the wellordered layered structure with a space group of R3m, as verified by fast Fourier-transform (FFT) pattern (Fig. 4a3). Fig. 4b illustrates the typical structure of the cathode of 4.5 V-cell. The purple elliptical region (highlighted in Fig. 4b1) exhibits a cation mixing layer of 2–3 nm in thickness, as shown in Fig. 4b3 under a high magnification. The rock-

Table 1 Capacity loss analysis of samples cycled to different voltages. Sample

4.2 V 4.5 V 4.8 V

discharge capacity (mAh/g) 1st cycle

100th cycle

161 197 237

158 178 155

Total capacity loss (mAh/g)

Capacity loss due to Impedance rise (mAh/g)

The ratio of Impedance rise

3 19 82

76-73 = 3 76-66 = 10 76-55 = 21

100% 53% 25%

indicates the inherent, bulk property of the active materials, which is highly correlated to the local structure and lithium site energies [27]. After 100 cycles, the discharge voltage plateau for 4.2 V-cell and 4.5 Vcell are 3.73 V and 3.72 V, respectively, indicating no evident change in the activity of Li-ion intercalation/deintercalation for these two cells (Fig. 1f). However, the 4.8 V-cell exhibits an inclined voltage plateau at 3.65 V, significantly lower than that in the first discharge process, which suggests the degradation in bulk structure of the active material. Compared Fig. 1e with Fig. 1f, the capacity from the plateau to the end of discharge decays from 76 mAh g−1 to 73 mAh g−1, 66 mAh g−1 and 55 mAh g−1 for the 4.2 Ve, 4.5 Ve and 4.8 V-cell, respectively after 100 cycles. The capacity loss related to the cell impedance and its proportion in the total capacity loss for the three cells is listed in Table 1. For 4.2 V-cell, all the capacity degradation after 100 cycles is attributed to the increase in cell impedance, implying that the structure of the NMC622 material can be well maintained at a low-voltage cycling. For 4.5 Ve and 4.8 V-cell, the ratio of capacity loss related to cell impedance decreases to 53% and 25%, respectively, indicating the impedance rise is no longer the major contribution to the capacity decay. The structural evolution of the NMC622 material may play an important role as well. Fig. 1g shows the Nyquist plots of the three cells discharged to the voltage plateaus at 3.73 V after 100 cycles. All of the plots are roughly composed of two depressed semicircles, corresponding to the capacitive SEI-electrolyte interface resistance (Rf) at high frequencies and the charge transfer resistance (Rct) at middle frequencies, respectively, as well as a sloped line at low frequencies, representing the Warburg impedance [14,28]. The depression feature of the semicircles is typically associated with the porous nature of the electrodes [29]. The initial impedance spectra before cycling (Fig. S3a) show little variation in different cells, indicating the good cell qualities. Given that all the three samples show the similar Rf (∼80 Ω) after the first cycle (Fig. S3b), the obvious difference on Rf value after 100 cycles (inset of Fig. 1g) may indicate a more severe interfacial evolution for the 4.8 V-cell during cycling than its counterparts. It is apparent that the Rct value of the 4.8 V-cell (230 Ω) dramatically increases compared with the samples cycled under lower cut-off voltages. The similarities in the impedance behavior of 4.2 Ve and 4.5 V-cells suggest that the two samples probably undergo similar interfacial evolutions during cycling, significantly different from that in 4.8 V-cell. 3.2. Post analysis of the cells cycled to different depth of charge Fig. 2 shows XRD patterns of the pristine cathode and the cycled cathode operating at different cut-off voltages after 100 cycles. All XRD patterns indicate similar hexagonal structures with R-3m space group and no impurities were detected. The existence of doublets for the (006)/(012) and (018)/(110) diffraction peaks characterizes the wellordered layered structure of the samples [8]. The (003) diffraction peak shows an apparent left shift from 18.87° to 18.72° for the cycled cathodes compared with that for the pristine electrode, suggesting that the high-voltage cycling leads to an increase in the lattice parameters. Table 2 shows the results of XRD data obtained by Rietveld refinement, including lattice parameters, volumes of the unit cells, and the diffraction angle of the (003) peak. It is well-known that during the charge 542

Journal of Power Sources 400 (2018) 539–548

Y. Ruan et al.

Fig. 2. XRD patterns of the pristine (a) and cycled NMC 622 electrode under cut-off voltage of 4.2 V (b), 4.5 V (c), and 4.8 V (d).

likely ascribed to an internal, structure degradation and can act as the nucleation site for crack incubation [37]. This observation agrees well with the formation of severe cracks in the cathode of 4.8 V-cell, which is also discovered in the SEM images. In addition, a thicker (4 nm) amorphous layer is continuously observed and uniformly distributed on the surface of the 4.8 V-cell (Fig. 4c), while the 4.2 Ve and 4.5 V-cells do not show obvious evidences of a continuous amorphous layer. When the cathode is charged to a higher cut-off voltage, TM cations at the high oxidation states can oxidize alkyl-carbonates to form CO2, consequently resulting in the formation of Li2CO3 by reaction of Li+ and O2− from the cathode with CO2 [28,41,42]. In addition, decomposition products of the electrolyte are adsorbed or incorporated into the surface and constitute the amorphous layer [22]. This insulating amorphous layer hinders the Li+ transport and electron transfer, which may account for the increase in Rct for the 4.8 V-cell (Fig. 1g). A correlated surface analysis using high-angle annular dark-field (HADDF)-STEM and electron energy-loss spectroscopy (EELS) in Fig. 5 was performed to further elucidate the mechanism of capacity fading and impedance rise of the cell cycled at various charge cut-off voltages. Compositional analyses by EELS is sensitive to small changes of the 3d orbital occupancy of TM, and can be used to reveal the alteration of the electronic properties of NMC622, which results from nanometer-range surface structural changes. In general, the ratio of the integrated intensities of L3 and L2, I (L3)/I (L2), and the peak shift are correlated with the oxidation state of the TM ions [15,32]. Generally, the oxidation states of Ni, Co, and Mn in the pristine NMC622 cathode materials are 2+/3+, 3+, and 4+, respectively. An increase of the I (L3)/I (L2) ratio and a blue shift correspond to a decrease in the oxidation state of the TM ions [15]. The intensity ratio of the Mn L2,3-edge increased with increasing charge cut-off voltages, but its peak shows no significant shift (Fig. 5f). Table 3 summarizes the integrated I (L3)/I (L2) ratio of a selected area of the fresh and cycled cathodes. The ratio of the 4.2 V-cell is 1.94, which is quite similar to 1.85 for the pristine cathode. On the other hand, the ratio of the 4.8 V-cell is 2.91, close to that of the 4.5 Vcell (2.76), indicating a lower oxidation state of Mn. It is known that the high average oxidation state of the Mn4+ contributes to the structural stability and effectively diminishes the J-T distortion [32]. The peak of Ni L2,3-edge in Fig. 5g shows a significant blue shift, which accompanies a slight increase of the intensity ratio, indicating a higher Ni2+ concentration on the surface of the cathode for 4.5 Ve and 4.8 V-cells. Considering the HRTEM/FFT results in Fig. 4, we could predict the electrochemically inative rock-salt NiO phase was formed on the

Table 2 Lattice parameters (a, c), unit cell volumes (V), R values (the intensity ratio of the (003) to (104) peaks) and the diffraction angle of (003) peak from refined XRD data of fresh and cycled electrodes.

Fresh 4.2 V 4.5 V 4.8 V

a(Å)

c(Å)

c/a

V (Å3)

R

2θ(003)/(°)

2.8638 2.8655 2.8686 2.8703

14.2246 14.2426 14.2729 14.3288

4.967 4.970 4.976 4.992

101.0 101.3 101.7 102.2

1.41 1.38 1.23 1.13

18.87 18.78 18.75 18.72

salt phase (Fm-3m) is present as determined by the lattice images and the selected area FFT pattern (inset of Fig. 4b3). Cation mixing resulted from migration of TM ions into vacant Li sites is the primary cause of the transformation from layered to spinel structure and rock-salt phase on the surface phase [11,15]. Given that the ionic radius of Ni2+ (0.69 Å) is similar to that of Li+ (0.76 Å), Ni2+ can more readily than Mn and Co ions moves to Li sites [32], thereby impeding Li+ diffusion and lowering cell capability [39]. However, it is interesting to find that the degree of surface disordering does not increase with the increase of the charge cut-off voltage. From extensive sample examination, the depth of the cation mixing layer in the cathode of 4.8 V-cell is not deeper than that of 4.5 V-cell. The rock-salt phase is detected in the narrow region of ca. 2–3 nm from the surface, as shown in the HRTEM images and the selected area FFT patterns of the cathode in 4.8 V-cell (Fig. 4c3). In the layered cathode materials, the structural transformation can develop through oxygen evolution when Co3+/4+ t2g or the Ni3+/4+ eg orbital substantially overlaps the O2p orbital [8,26]. The high charge cut-off voltage promotes oxygen evolution, and then promotes the formation of the rock-salt phase during cycling. It is difficult to detect the second phase in the XRD patterns due to its small amount. However, in the XRD patterns of 4.5 Ve and 4.8 V-cells, more broaden peaks can be observed compared with the pristine cathode and 4.2 Vcell (Fig. 2). Such peak broadening may be associated with the second phase formation, which leads to a non-uniform distribution of local strains due to the formation of a completely different phase [40]. Considering that the surface layer of 4.8 V-cell does not grow thicker than that of 4.5 V-cell, there is likely another factor that differentiates the electrochemical performance for various cells, even though the thin layer of the rock-salt phase (Fm-3m) can increase the impedance [41]. Some edge dislocations are formed for the cathode of 4.8-V cell by splitting two neighboring TM slabs along the (003) planes, as indicated by the white dashed oval area in Fig. 4c3. Edge dislocation is more 543

Journal of Power Sources 400 (2018) 539–548

Y. Ruan et al.

Fig. 3. SEM images of the cycled electrodes under different magnifications for 4.2 V (a), 4.5 V (b), and 4.8 V (c). SEM images of the regions indicated by a blue rectangle (d) and an orange rectangle (e) in (c2). EDS image (f) of the corresponding areas for (d) and (e). (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

surface, which attributes to the highly oxidative environment that triggers the oxygen loss from the surface of the material. This may explain the significant decrease in the average valence state of the TM cations for the 4.5 Ve and 4.8 V-cells. The second main peak in the O Kedge also shows a clear difference between the high voltage (4.5 Ve and 4.8 Ve) cells and the low voltage (4.2 Ve) cell, indicating the different cation-oxygen coordination environment with increasing cutoff voltages. At the same time, the oxygen evolution accompanies the structural change on the surface of the NMC materials [32]. Due to the

surface area of the 4.5 Ve and 4.8 V-cells. The pre-peak of the O K-edge, as indicated in Fig. 5h, originates from the hybridization of the O2p orbital and the TM3d orbital, and its intensity indicates the filling of the 3d orbital of the TM ions [43]. In Fig. 5h, the pre-peak of the O K-edge becomes weaker and shifts to a higher energy, indicating that the average valence state of the TM cations decreases with increasing charge cut-off voltages. It is reported that the significant drop in intensity of the pre-peak indicates the formation of an O2− deficient surface [43]. A deeper charge more likely results in an oxygen-deficient

544

Journal of Power Sources 400 (2018) 539–548

Y. Ruan et al.

Fig. 4. TEM images of the cycled electrode at different cut-off voltage of 4.2 V (a), 4.5 V (b) and 4.8 V (c). HRTEM images (a2), (b2) and (c2) corresponding to the purple elliptical regions in (a1), (b1) and (c1), respectively. (a3) shows the FFT pattern of (a2). (b3) shows the enlarged HRTEM image of the purple elliptical region in (b1) and the corresponding selected area FFT patterns in surface area (pink) and bulk area (green). (c3) shows the enlarged HRTEM images of the purple rectangle region in (c1) and the corresponding selected area FFT patterns in surface area (pink) and bulk area (green). (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

reaction shifts as a function of the charge-discharge conditions. For the electrode charged to 4.8 V, the rapid structural changes result in a sharp weight loss at ca. 200 °C. The main oxygen release peak temperature of 256 °C for the electrodes charged to 4.2 V is higher than the temperature of 226 °C for that to 4.5 V (Fig. S7). The significant difference in the weight loss and decomposition temperature observed for the fresh and cycled electrodes originates from the internal structure degradation of the electrode. As described in previous studies [11,32,40], such a thermal decomposition process is accompanied by a structural change from hexagonal (H1) to monoclinic (M) via the migration of Ni2+ from their original sites to the Li-interlayer sites, as well as through displacement of the Li+ from octahedral to tetrahedral sites in the interlayer space. This type of phase transition involves an oxygen loss. Except for the main thermal decomposition peak below 300 °C, the mass loss also occurs in a wide temperature range from 350 to 500 °C for all samples, which is assigned to the decomposition of the PVDF binder. The difference in thermal stability of the samples results from the structure difference, which can be validated by the XRD analysis (Fig. 6b). The fresh electrode showed two doublets at (006)/(012) and (018)/(110) diffractions in the XRD pattern, indicating the well-ordered crystalline layered structure of samples [8]. By tracking the evolutions

formation of a large amount of oxygen vacancies at the outermost (2–3 nm) surface layers, the average value of the TM valence state on the sample surface becomes lower, which well matches with the EELS analysis of the samples integrated from surface to interior in previous reports [15,32]. EELS results indicate that the similarity in the structural evolution on the surface of the 4.5 Ve and 4.8 V-cells, in accordance with the TEM results, which validates that the degradation on the surface is not the main reason for the difference in capacity degradation between the two cells. The aforementioned analysis is limited to a small area on the outer surface of the composite electrode. To evaluate the bulk sample and investigate changes in the thermal stability of the NMC622 cathode, TG-DSC and XRD measurements were carried out to the pristine and the cycled cathodes. To ensure the identical lithiation states of the samples, all three cycled cells were charged to a selected voltage and maintained at that voltage for 2 h. The results presented in Fig. 6 show the pronounced dependence of structural and thermal stability on the charge cut-off voltages. All three cycled cathodes show an obvious thermal decomposition peak below 300 °C (Fig. 6a), which is assigned to the oxygen loss of the thermal decomposition of the NMC materials [11,22]. Obviously, the temperature range of this decomposition 545

Journal of Power Sources 400 (2018) 539–548

Y. Ruan et al.

Fig. 5. High angle annular dark-field (HAADF) STEM images (a-d) at a low magnification and EELS spectra (e) of the fresh electrode (purple) and electrodes which are cycled at cut-off voltage of 4.2 V (green), 4.5 V (red) and 4.8 V (blue), respectively. The scale bars in (a-d) are 10 nm and the EELS selected areas are 2 nm from the outmost surface. EELS spectra of Mn L-edge (f), Ni L-edge (g) and O K-edge (h). (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

instability and thermal decomposition. CV and differential capacity (dQ/dV) curves give further information on the redox reactions and phase transitions happened in the working potential range and how this redox reaction impacts the degradation of the cycled electrode. Fig. 7a shows the first five potential scan curves of the NMC622 cathode between 2.8 and 4.9 V with a scan rate of 0.01 mV s−1. During the first scan, the oxidation peak at approximately 3.75 V is predominantly associated with oxidation of Ni2+/ 3+ to Ni4+, on the basis that the formal oxidation state of Ni is a mixed valence of +2 and + 3. The second peak at higher potentials (∼4.71 V) is mainly attributed to the irreversible electrochemical reaction that removes Li+ and is accompanied by loss of O2 from the structure of the sample [20,44]. It is in agreement with the high-voltage plateau designated by the purple arrow in Fig. 1c, which is commonly seen during electrochemical oxidation of lithium-excess compounds under the high charge voltage [45]. In the second anodic scan, the

Table 3 Integrated L3/L2 Ratio of the Mn and Ni L-Edge on the assigned area. I (L3)/I (L2)

Mn

Ni

Fresh 4.2 V 4.5 V 4.8 V

1.77 1.94 2.76 2.91

1.87 1.90 2.31 2.46

of the peaks from 4.2 Ve, 4.5 Ve to 4.8 V-cell, it is clear that the intensities and shapes of the characteristic peaks (003), (101) and (104) gradually reduce, and the c lattice parameter gradually increases during the delithiated process. The (003) peak shifts and broadens in the highly delithiated samples (4.8 V-cell). This indicates that the internal structural changes irreversibly at higher working potentials than that in a traditional working potential [20], eventually accelerating the oxygen 546

Journal of Power Sources 400 (2018) 539–548

Y. Ruan et al.

Fig. 6. TGA (a) and XRD patterns (b) of the pristine electrode and the electrodes charged to different cut-off voltages.

reversible oxidation/reduction peak of Co3+/Co4+ at 4.64/4.57 V is attributed to the phase transition from rhombohedral O3 phase to H1-3 phase, which is similar to the phase transition of LiCoO2 at a higher voltage [44]. This process is associated with the Li+ migration from octahedral sites of the layered structure to tetrahedral sites of the cation-disordered rock-salt structure, resulting in capacity decay [32]. Fig. 7b ∼ d show the differential capacity (dQ/dV) curves derived from the charge/discharge curves in Fig. 1. It is noted that the dQ/dV peak

almost maintains the original shape, as well as the peak value in the potential range for 4.2 V-cell in Fig. 7b. While with increasing cut-off charge voltages, the second oxidation peaks, indicated by the dashed black arrow in Fig. 7c and d, corresponding to Ni3+/Ni4+, decrease with cycling and then disappear for the 4.8 V-cell (Fig. 7d). Compared the dQ/dV curves at 4.8 V in Fig. 7d with the CV curves in Fig. 7a, the phase transition related to Co3+/Co4+ shows an obvious irreversibility, as indicated by the dashed red arrow in Fig. 7d. This evolution of dQ/

Fig. 7. CV curves of the sample in the first five cycles in 2.8–4.9 V with a scan rate of 0.01 mV s-1 (a). dQ/dV curves of the cell cycled at various cut-off voltages with the increased cycle numbers (b-d). 547

Journal of Power Sources 400 (2018) 539–548

Y. Ruan et al.

dV peaks during repeated electrochemical cycles in Fig. 7b ∼ d suggests that with the increased charge voltage, excess Li+ extracted from the structure not only results in the internal irreversible structure change, like LiCoO2 operated at high voltage [20], but also amplifies the internal strain within a grain, which may causes the dislocation as observed in the TEM image (Fig. 4c3). In addition, the disappeared Ni3+/ Ni4+ peak in Fig. 7d suggests that the oxidation state of Ni reduces after 100 cycles, which coincides with the surface analysis by EELS (Fig. 5g). Consequently, the internal irreversible structure degradation, as well as the surface reconstruction, accounts for the capacity degradation for 4.8 V-cell. Furthermore, formation of the solid electrolyte interphase (SEI) film and the partial decomposition of some organic compositions from the surface film may lead to the impedance increase and capacity decay at the high charge voltage.

13104–13109. [3] F. Wu, N. Li, Y. Su, H. Shou, L. Bao, W. Yang, L. Zhang, R. An, S. Chen, Adv. Mater. 25 (2013) 3722–3726. [4] S. Kalluri, M. Yoon, M. Jo, S. Park, S. Myeong, J. Kim, S.X. Dou, Z.P. Guo, J. Cho, Adv. Energy Mater. 7 (2017) 1602010. [5] R.E. Ruther, A.F. Callender, H. Zhou, S.K. Martha, J. Nanda, J. Electrochem. Soc. 162 (2015) A98–A102. [6] J.M. Zheng, S.J. Myeong, W.R. Cho, P.F. Yan, J. Xiao, C.M. Wang, J. Cho, J.G. Zhang, Adv. Energy Mater. 7 (2017) 1601284. [7] J. Li, X. Wang, J. Zhao, J. Chen, T. Jia, C. Cao, J. Power Sources 307 (2016) 731–737. [8] S.-K. Jung, H. Gwon, J. Hong, K.-Y. Park, D.-H. Seo, H. Kim, J. Hyun, W. Yang, K. Kang, Adv. Energy Mater. 4 (2014) 1300787. [9] H.-J. Noh, S. Youn, C.S. Yoon, Y.-K. Sun, J. Power Sources 233 (2013) 121–130. [10] H.B. Rong, M.Q. Xu, Y.M. Zhu, B.Y. Xie, H.B. Lin, Y.H. Liao, L.D. Xing, W.S. Li, J. Power Sources 332 (2016) 312–321. [11] S.M. Bak, E.Y. Hu, Y.N. Zhou, X.Q. Yu, S.D. Senanayake, S.J. Cho, K.B. Kim, K.Y. Chung, X.Q. Yang, K.W. Nam, ACS Appl. Mater. Interfaces 6 (2014) 22594–22601. [12] P. Dong, D. Wang, Y. Yao, X. Li, Y. Zhang, J. Ru, T. Ren, J. Power Sources 344 (2017) 111–118. [13] C. Chen, T. Tao, W. Qi, H. Zeng, Y. Wu, B. Liang, Y. Yao, S. Lu, Y. Chen, J. Alloy. Comp. 709 (2017) 708–716. [14] L.W. Liang, F. Jiang, Y.B. Cao, G.R. Hu, K. Du, Z.D. Peng, J. Power Sources 328 (2016) 422–432. [15] N.Y. Kim, T. Yim, J.H. Song, J.S. Yu, Z. Lee, J. Power Sources 307 (2016) 641–648. [16] W. Choi, A. Manthiram, J. Electrochem. Soc. 153 (2006) A1760–A1764. [17] D.R. Gallus, R. Schmitz, R. Wagner, B. Hoffmann, S. Nowak, I. Cekic-Laskovic, R.W. Schmitz, M. Winter, Electrochim. Acta 134 (2014) 393–398. [18] Z.C. Zhang, L.B. Hu, H.M. Wu, W. Weng, M. Koh, P.C. Redfern, L.A. Curtiss, K. Amine, Energy Environ. Sci. 6 (2013) 1806–1810. [19] Y. Watanabe, S.I. Kinoshita, S. Wada, K. Hoshino, H. Morimoto, S.I. Tobishima, J. Power Sources 179 (2008) 770–779. [20] J. Shu, R. Ma, L. Shao, M. Shui, K. Wu, M. Lao, D. Wang, N. Long, Y. Ren, J. Power Sources 245 (2014) 7–18. [21] J.G. Xu, R.D. Deshpande, J. Pan, Y.T. Cheng, V.S. Battaglia, J. Electrochem. Soc. 162 (2015) A2026–A2035. [22] M. Boerner, F. Horsthemke, F. Kollmer, S. Haseloff, A. Friesen, M. Winter, F.M. Schappacher, J. Power Sources 335 (2016) 45–55. [23] M. Lang, M.S.D. Darma, K. Kleiner, L. Riekehr, L. Mereacre, M.A. Perez, V. Liebau, H. Ehrenberg, J. Power Sources 326 (2016) 397–409. [24] J. Vetter, P. Novak, M.R. Wagner, C. Veit, K.C. Moller, J.O. Besenhard, M. Winter, M. Wohlfahrt-Mehrens, C. Vogler, A. Hammouche, J. Power Sources 147 (2005) 269–281. [25] F. Lin, I.M. Markus, D. Nordlund, T.C. Weng, M.D. Asta, H.L. Xin, M.M. Doeff, Nat. Commun. 5 (2014) 3529. [26] J. Choi, A. Manthiram, J. Electrochem. Soc. 152 (2005) A1714–A1718. [27] J.R. Croy, J.S. Park, Y. Shin, B.T. Yonemoto, M. Balasubramanian, B.R. Long, Y. Ren, M.M. Thackeray, J. Power Sources 334 (2016) 213–220. [28] Y. Cho, P. Oh, J. Cho, Nano Lett. 13 (2013) 1145–1152. [29] J. Schmitt, A. Maheshwari, M. Heck, S. Lux, M. Vetter, J. Power Sources 353 (2017) 183–194. [30] O. Dolotko, A. Senyshyn, M.J. Mühlbauer, K. Nikolowski, H. Ehrenberg, J. Power Sources 255 (2014) 197–203. [31] Y.N. Zhou, J. Ma, E.Y. Hu, X.Q. Yu, L. Gu, K.W. Nam, L.Q. Chen, Z.X. Wang, X.Q. Yang, Nat. Commun. 5 (2014) 5381. [32] H. Kim, M.G. Kim, H.Y. Jeong, H. Nam, J. Cho, Nano Lett. 15 (2015) 2111–2119. [33] K.M. Shaju, G.V. Subba Rao, B.V.R. Chowdari, Electrochim. Acta 48 (2002) 145–151. [34] S.W. Lee, H. Kim, M.S. Kim, H.C. Youn, K. Kang, B.W. Cho, K.C. Roh, K.B. Kim, J. Power Sources 315 (2016) 261–268. [35] E.J. Lee, Z.H. Chen, H.J. Noh, S.C. Nam, S. Kang, D.H. Kim, K. Amine, Y.K. Sun, Nano Lett. 14 (2014) 4873–4880. [36] D.J. Miller, C. Proff, J.G. Wen, D.P. Abraham, J. Bareño, Adv. Energy Mater. 3 (2013) 1098–1103. [37] P.F. Yan, J.M. Zheng, M. Gu, J. Xiao, J.G. Zhang, C.M. Wang, Nat. Commun. 8 (2017) 14101. [38] G. Bucci, T. Swamy, S. Bishop, B.W. Sheldon, Y.-M. Chiang, W.C. Carter, J. Electrochem. Soc. 164 (2017) A645–A654. [39] S.P. Sheu, I.C. Shih, C.Y. Yao, J.M. Chen, W.M. Hurng, J. Power Sources 68 (1997) 558–560. [40] Y. Cho, S. Lee, Y. Lee, T. Hong, J. Cho, Adv. Energy Mater. 1 (2011) 821–828. [41] F. Schipper, E.M. Erickson, C. Erk, J.Y. Shin, F.F. Chesneau, D. Aurbach, J. Electrochem. Soc. 164 (2017) A6220–A6228. [42] O. Haik, N. Leifer, Z. Samuk-Fromovich, E. Zinigrad, B. Markovsky, L. Larush, Y. Goffer, G. Goobes, D. Aurbach, J. Electrochem. Soc. 157 (2010) A1099–A1107. [43] M. Gu, A. Genc, I. Belharouak, D. Wang, K. Amine, S. Thevuthasan, D.R. Baer, J.G. Zhang, N.D. Browning, J. Liu, C. Wang, Chem. Mater. 25 (2013) 2319–2326. [44] H. Xia, L. Lu, Y.S. Meng, G. Ceder, J. Electrochem. Soc. 154 (2007) A337–A342. [45] M.M. Thackeray, C. Wolverton, E.D. Isaacs, Energy Environ. Sci. 5 (2012) 7854.

4. Conclusion The cycling stability of the NMC622 cathode material strongly depends on the depth of charge. The cells cycled at the low charge cut-off voltage of 4.2 V show no obvious capacity decay, while the cell cycled at 4.8 V displays a rapid capacity decay. Post mortem analysis of the NMC622 electrode reveals that two different mechanisms account for the performance degradation of the cells. When the cathode is cycled under a mild charge condition (4.5 V or below), the structure change occurred on the surface of the material increase the charge transfer impedance and cause the capacity decay with cycling. However, when the charge cut-off voltage exceeds the critical value (4.8 V), the internal structure degradation during the highly deliatiated process is mainly attributed to the prompt capacity decay. Furthermore, the insulating surface film incurs the irreversible electrode side reaction, enhances the electrode interfacial impedance, and thereby lowers the capacity. In addition, the formation of intergranular cracks and loss of active lithium also aggravate the capacity loss. Determining an appropriate charge cut-off voltage is a trade-off between gaining more capacity from NMC622 and avoiding fast fatigue of the layered oxides. Therefore, in order to maximize the capacities of NMC material in the high voltage region, it is vital to stabilize the structure of NMC622 particles to alleviate the internal strain and optimize the composition of electrolyte to control the formation and growth of the insulating surface film. Acknowledgement This work was supported by the U.S. Department of Energy (DOE) under Contract No. DE-AC02-05CH11231 (Project No. ES232), National Natural Science Foundation of China (No. 21403153) and the Science and Technology Plans of Tianjin (No. 15PTSYJC00230, 16JCTPJC45200). We acknowledge support of the National Center for Electron Microscopy, Lawrence Berkeley Lab, which is supported by the U.S. Department of Energy under Contract No. DE-AC02-05CH11231. We also gratefully acknowledged the financial support from China Scholarship Council (No. 201508120017). Appendix A. Supplementary data Supplementary data related to this article can be found at https:// doi.org/10.1016/j.jpowsour.2018.08.056. References [1] A. Manthiram, J.C. Knight, S.-T. Myung, S.-M. Oh, Y.-K. Sun, Adv. Energy Mater. 6 (2016) 1501010. [2] H.Z. Zhang, Q.Q. Qiao, G.R. Li, S.H. Ye, X.P. Gao, J. Mater. Chem. 22 (2012)

548