Determination of the mechanism and extent of surface degradation in Ni-based cathode materials after repeated electrochemical cycling Open Access

Beyond powering small devices, the demands for large-scale applications of lithium ion batteries (LIBs) — including incorporation into electric vehicles¹ and smart grids² — have been increasing. It is critical to improve the figures-of-merit of electrode materials in order to meet the requirements of these large-scale applications. Because of their high energy density, Ni-based lithium transition metal oxides (LiTMO₂) with the layered structure are among the most promising cathode materials for use in next generation LIBs. However, they have disadvantages with respect to structural instability, which results in both capacity loss and safety issues. It is well known that incorporation of other metal elements can modify the properties of cathode materials, for example, Mn helps with the retention of the original layered structure during intercalation and de-intercalation of lithium ions, but it does so at the expense of capacity. Co improves the rate performance, but it is both expensive and toxic.³ Since the incorporation of each element has both advantages and disadvantages, the optimization of the composition of these elements with Ni in LiTMO₂ is of particular interest.

Changes in the structure and chemistry of electrode materials are detrimental to their performance;⁴ thus, a number of diagnostic tools have been applied to understand the degradation mechanisms. X-ray based techniques such as x-ray diffraction (XRD) and x-ray absorption spectroscopy (XAS) have provided valuable insights into the changes in crystallographic and electronic structures that occur within the bulk of Ni-based cathode materials.^3,5,6 Transmission electron microscopy has been exploited to investigate site-specific structural modifications at the nanoscale.^3,7–9 Chemical changes in Ni-based cathode materials have been explored with x-ray photoelectron spectroscopy (XPS),^10,11 energy dispersive spectroscopy (EDS),¹² and electron energy loss (EEL) spectroscopy.^8,13,14 A number of analytical tools have been utilized in order to elucidate the mechanism of degradation, but few studies have been conducted to determine the depth of structural evolution. This is despite the fact that identifying the degree of degradation inside a particle is critical to evaluate the stability of cathode materials inside lithium cells.

In this work, we take advantage of scanning transmission electron microscopy (STEM) and electron energy loss spectroscopy (EELS) to understand what causes the degradation of Ni-based cathode materials (LiNi_0.8Mn_0.1Co_0.1O₂ and LiNi_0.4Mn_0.3Co_0.3O₂ — referred as NMC811 and NMC433, respectively) and to determine how much area has been damaged after a number of electrochemical cycles. By quantifying the depth of degradation when the cathodes are exposed to the same conditions of electrochemical testing, it is possible to directly compare the structural stability of promising Ni-based cathode materials following repetitive de-intercalation and intercalation of lithium ions.

Ni-based cathode materials (Li_xNi_0.4Mn_0.3Co_0.3O₂, Li_xNi_0.8Mn_0.1Co_0.1O₂) — which are commercial products — were electrochemically cycled within a voltage range of 2.0–4.8 V at a rate of C/10 using a galvanostatic condition (i.e., constant current) in the 2032-type coin cells. Coin cells were assembled with a cathode part which included the active particles, a Li metal for the anode, a Celgard separator, and an electrolyte of 1M LiPF₆ dissolved in ethylene carbonate (EC), ethyl methyl carbonate (EMC), and dimethyl carbonate (DMC) (1:1:1 by volume). The cathode part was prepared as a mixed slurry of 80 wt.% of the active NMC particles, 10 wt.% of the conducting agent (Denka Black), and 10 wt.% of a polyvinylidene fluoride (PVDF) binder in a N-methyl pyrrolidone (NMP) solvent. The mixed slurry was cast onto the Al current collector. After 20 charge/discharge cycles, the coin cells were disassembled in the discharged state. The cathode part was immersed in a pure DMC solution to clean off any residual salts before the subsequent analysis.

X-ray diffraction (XRD) patterns were acquired with a D-max 2500/PC X-ray diffractometer (Rigaku) using Cu Kα radiation over a 2θ range of 10°–70° at a scan rate of 2°/min. The XRD patterns were used to analyze the bulk structure of the NMC433 and NMC811 particles in the mixed slurries.

For the scanning transmission electron microscopy (STEM) observations, NMC samples were acquired by abrading the slurry coated on the Al foil. Supersonic vibration was used to ensure an adequate dispersion of particles in a small vial of pure DMC before the solution was dropped onto a lacey-carbon TEM grid. The sample preparation and loading into the TEM sample holder were performed in either a dry room or an Ar-filled glove box to minimize the exposure of the sample to air and moisture. Bright field (BF) images, high-angle annular dark field (HAADF) images, and electron energy loss (EEL) spectra were acquired with a Tecnai G² F20 (FEI) TEM at a relatively low accelerating voltage of 120 kV to minimize the electron beam effect and a Quantum 963 GIF system (Gatan). The background of the spectra was subtracted by using the power-law method embedded within the TEM Imaging and Analysis (TIA) software (FEI). The energy resolution, which was determined from the zero-loss peak, was approximately 1.0 eV.

Charge-discharge profiles of NMC433 and NMC811 during 20 electrochemical cycles are shown in Figures 1(a) and 1(b), respectively. There is no significant difference in capacity retention, but there is considerable difference in internal resistance, which is revealed by I⋅ΔR drop at the initiation of the discharge, between NMC433 and NMC 811. After the cycle test, changes in the bulk crystallographic structure of NMC433 and NMC811 were investigated by XRD analysis, as shown in Figure 1(c). Comparing the XRD patterns of cycled materials to that of pristine (before any electrochemical reaction) samples, new features such as appearance of a new diffraction peak, peak splitting, or peak disappearing are not observed, regardless of TM compositions and cutoff voltages. Therefore, despite the repetitive de-intercalation and intercalation of lithium ions, the core structure of the cathode materials remains as the α-NaFeO₂ layered structure (space group of R $3 ̄ m$⁠). However, there are slight shifts in the positions of the diffraction peaks. The lattice parameters of Ni-based cathode materials have changed after the electrochemical tests; thus, variances in a, c lattice parameters and unit cell volumes of cathode materials are described in Figure 1(d). We found the a, c lattice parameters decreased after cycling, and the contraction in the lattice parameter is somewhat anisotropic. For NMC433, the contraction along the c-axis is suppressed, while the decrease of the lattice constant along the c-axis is noticeable in NMC811 after cycling with a cutoff voltage of 4.8 V. We suspect the different compositions of transition metals in NMC433 and NMC811 lead to the different degree of the contribution for the charge compensation among Ni, Co, and O ions, resulting in different aspects of the lattice parameter changes between NMC433 and NMC811. Shuttling of lithium ions brings about a contraction in the unit cell volume of these Ni-based cathode materials. The degree of volume contraction increases with higher cutoff voltage in the cathode materials. However, there is no significant difference in the amount of volume contraction between NMC433 and NMC811 after 20 cycles, which may be related to little difference in charge capacity retention.

Changes in the electronic structure of NMC433 cathode materials, which were cycled for 20 times within the range of 2.0–4.8 V, are presented in Figure 2. A sub-nanometer electron probe is used to scan the particle from the edge toward the inner region with 1 nm step sizes, as shown in Figure 2(a). Changes in the EEL spectra of each element [Figure 2(b)] are tracked by defining two indices: the Δ E of the O K-edge, and L₃/L₂ edge intensity ratio of the TMs. The main features of the O K-edge in the EEL spectrum of a LiTMO₂ material with the layered structure are two distinct peaks: a pre-edge peak and a subsequent main peak. The Δ E of the O K-edge is described as the energy-loss difference between the main and pre-edge peaks at their highest intensities. As the pre-edge peak is attributed to the transition of electrons from the 1s core state to the unoccupied 2p states that are hybridized with the 3d states of the TMs,¹⁵ the movement of the pre-edge peak to a lower energy loss (higher Δ E) indicates the oxidation of the TM, as it binds with O, and vice versa. Additionally, changes in the chemical state of the TM ions are also directly related to changes in the TM L edges. The L_2,3 edge of the TMs results from electron transitions from the TM 2p states to the highly localized 3d states near the Fermi level. The splitting of the L_2,3 edge originates from the degenerate 2p states that are split into 2p_1/2 and 2p_3/2 levels due to spin-orbit coupling.¹⁶ Changes in the threshold energy of the L_2,3 edges and the intensity ratio (L₃/L₂) thus correspond to changes in the chemical state of the TM ions.^17–19 Here, we use only the L₃/L₂ ratio of each TM in the compounds in order to express the changes in oxidation states, since the intrinsic energy resolution of the electron beam (0.8–1.0 eV) may lead to misinterpretations when EELS results are analyzed based on the shift of the onset of the edges in the EEL spectra. The white-line ratio (L₃/L₂) decreases with oxidation, but increases with reduction for the Mn,¹⁹ Co,²⁰ and Ni ions;²¹ thus, white-line ratios are an appropriate indicator of valence changes of these TM ions.

Figure 2(c) summarizes the O K, Mn, Co, and Ni L_2,3 EEL spectra taken from the cycled NMC433 particle, using the Δ E and L₃/L₂ indexes of TMs described above. References of the Δ E and L₃/L₂ indices are also provided in Figure 2(d), which are acquired from the sub-surface of initially charged and discharged NMC433 particles.²² We chose the discharged state after the first electrochemical cycle as the reference rather than the pristine state. According to the charge-discharge curves [Figure 1(a)], there is a considerable irreversible capacity loss during the first charge/discharge process, which is attributed to the formation of electrochemically inactive domains.²³ Thus, the electronic structure of the initially discharged state is more suitable as a reference than that of the as-synthesized state. As an electron probe moves from the edge to inner region of the particle, modifications to the electronic structure are noticeable: the L₃/L₂ of Mn decreases and the Δ E of the O K edge increases. In other words, when compared to the reference in Figure 2(d), Mn ions are reduced at the surface, which leads to the accompanying changes in the O K-edge EEL spectra. Here, the Δ E of the O K-edge is utilized as an indicator of the depth of the degradation zone, since this value reflects the chemical status of the combined effects of the TMs inside all Ni-based layered cathode materials. We define the transition point between the damaged and undamaged areas as a point which has comparable Δ E to the reference. On occasion, Δ E values does not reach to the reference value up to the point where changes in L₃/L₂ of TMs are not significant any longer with scanning the beam toward inner particles; in those cases, the transition point is decided as a point where over 20% of increases in Δ E are observed compared to the prior point. In the case of cycled NMC433, 22.9% of increase in Δ E is observed at 11 nm, thus we determined this point as the transition point. Interestingly, the EELS signals of Co and Ni L_2,3 edges are barely detected within 5 nm from the edge, implying that an ∼5 nm Mn-rich layer has formed at the surface. It is noteworthy that compositional evolution can also occur as a result of electrochemical cycles in NMC433: previous publications only have reported that chemical evolution can occur at the surface of Li-rich cathode materials after repetitive lithium movements.^24,25

In the case of cycled NMC811, we observe a similar trend in the Δ E of the O K-edge, but the behavior of the L₃/L₂ ratios of the TMs is distinctly different from that of NMC433, as presented in Figure 3. The L₃/L₂ values of Mn are slightly lower than the reference, while the L₃/L₂ values of Ni are higher than the reference value in Figure 3(d) near the surface region. This implies that Ni ions exist in a reduced form at the surface, as opposed to the existence of reduced Mn ions in cycled NMC433. Low Δ E values near the surface also indicate that the TMs are reduced at the surface. Thus, we conclude that the Ni is the major element that is responsible for the modifications of the O K-edge of cycled NMC811. We can also specify the degradation depth of cycled NMC811 by Δ E, and find it to be ∼34 nm which is the first point having comparable Δ E to the reference.

If Mn or Ni ions at the surface are reduced, either other TMs should be oxidized or oxygen ions should leave the structure in order to fulfill the charge neutrality. As we can see from Figure 2, the low L₃/L₂ ratio of Ni implies that Ni ions are oxidized to compensate the charge imbalance during Mn reduction at the surface. In the case of NMC811 (Figure 3), the L₃/L₂ ratio of Mn of NMC811 is lower than the reference value [Figure 3(d)], implying that the valence of Mn increases after 20 cycles when compared to that after initial discharge. Consequently, the presence of complementary metal ions, which can be oxidized, is necessary to suppress the formation of surface porosity which is caused by TM reduction during electrochemical cycling.

Notably, the use of a cutoff voltage of 4.8 V is actually a very severe testing condition; at this voltage unwanted reactions may happen, such as decomposition of electrolytes.²⁶ We also performed the same STEM-EELS studies with the cathode materials that were cycled within the range of 2.0–4.3 V at a rate of C/10 for 20 times. A set of experiments with the cutoff voltage of 4.3 V (Figures S1-S3 of the supplementary material) confirm that the surface changes we saw are in fact induced by repetitive lithium shuttling, and not by unwanted reactions at high voltages.

Our previous study about cycled Ni-rich cathode materials showed that there are considerable particle-to-particle variations in the Δ E of the O K-edge.⁸ It is important to investigate the particle-to-particle variations in order to understand how electrode materials vary at the nanometer scale, and thus to better understand the parameters that lead to a stable and reliable operation of the battery system. The sub-nanometer probe beam of STEM in this work enables the particle-to-particle variations to be described as accurate depths of the degradation zones. Figure 4 presents the variations in the depth of the degradation zone among 4 different particles. It clearly shows that particle-to-particle variations are much more significant in Ni-rich NMC811 than in Mn-rich NMC433. The considerable particle-to-particle variations, as well as the structural instability of Ni-rich cathode materials, represent large obstacles to overcome them in order to take full advantage of the higher energy density of Ni-rich cathode materials. The analytical methodology using STEM-EELS technique in this study can be one excellent platform to search for the next generation of cathode materials with both higher energy density and better structural stability.

Analytical characterization with (scanning) transmission electron microscopy has been exploited to not only delineate the changes in electronic structure that occur in Ni-based cathode materials after 20 electrochemical cycles but also to quantify how much of the sample has undergone these structural changes. The reduced metal ion is either Mn or Ni, depending on the overall composition of the cathode materials. Changes in the oxidation states of Ni and Mn are complementary to each other, resulting in a balance of the charge inside the cathode materials. In this study, we developed an analytical method for accurately measuring the depth of the degraded area and have investigated the mechanism by which Ni-based cathode materials lose their original properties during electrochemical cycling. The mechanism and the degree of the surface degradation offer critical understanding and insight into the rational design of new cathode materials with high capacity and long-term stability.

See supplementary material for the charge-discharge curves and STEM-EELS analysis of NMC433 and NMC811 at the cutoff voltage of 4.3 V.

This work was supported by the Korea Institute of Science and Technology (KIST) Institutional Program (Project Nos. 2Z04670, 2E26292, and 2E26330). E.A.S. acknowledges support to the Center for Functional Nanomaterials, Brookhaven National Laboratory, which is supported by the U.S. Department of Energy, Office of Basic Energy Sciences, under Contract No. DE-SC0012704.