Chemical and Structural Stability of Lithium-Ion Battery Electrode Materials under Electron Beam

The investigation of chemical and structural dynamics in battery materials is essential to elucidation of structure-property relationships for rational design of advanced battery materials. Spatially resolved techniques, such as scanning/transmission electron microscopy (S/TEM), are widely applied to address this challenge. However, battery materials are susceptible to electron beam damage, complicating the data interpretation. In this study, we demonstrate that, under electron beam irradiation, the surface and bulk of battery materials undergo chemical and structural evolution equivalent to that observed during charge-discharge cycling. In a lithiated NiO nanosheet, a Li2CO3-containing surface reaction layer (SRL) was gradually decomposed during electron energy loss spectroscopy (EELS) acquisition. For cycled LiNi0.4Mn0.4Co0.18Ti0.02O2 particles, repeated electron beam irradiation induced a phase transition from an layered structure to an rock-salt structure, which is attributed to the stoichiometric lithium and oxygen removal from 3a and 6c sites, respectively. Nevertheless, it is still feasible to preserve pristine chemical environments by minimizing electron beam damage, for example, using fast electron imaging and spectroscopy. Finally, the present study provides examples of electron beam damage on lithium-ion battery materials and suggests that special attention is necessary to prevent misinterpretation of experimental results.

The investigation of chemical and structural dynamics in battery materials is essential to elucidation of structure-property relationships for rational design of advanced battery materials. Spatially resolved techniques, such as scanning/transmission electron microscopy (S/TEM), are widely applied to address this challenge. However, battery materials are susceptible to electron beam damage, complicating the data interpretation. In this study, we demonstrate that, under electron beam irradiation, the surface and bulk of battery materials undergo chemical and structural evolution equivalent to that observed during charge-discharge cycling. In a lithiated NiO nanosheet, a Li 2 CO 3 -containing surface reaction layer (SRL) was gradually decomposed during electron energy loss spectroscopy (EELS) acquisition. For cycled LiNi 0.4 Mn 0.4 Co 0.18 Ti 0.02 O 2 particles, repeated electron beam irradiation induced a phase transition from an R 3m layered structure to an Fm 3m rock-salt structure, which is attributed to the stoichiometric lithium and oxygen removal from R 3m 3a and 6c sites, respectively. Nevertheless, it is still feasible to preserve pristine chemical environments by minimizing electron beam damage, for example, using fast electron imaging and spectroscopy. Finally, the present study provides examples of electron beam damage on lithium-ion battery materials and suggests that special attention is necessary to prevent misinterpretation of experimental results.
L ithium-ion batteries are used in a variety of consumer electronics, such as mobile phones, tablets and laptops, and are now being commercialized for use in electric vehicles (EVs) and plug-in hybrid electric vehicles (PHEVs) [1][2][3][4] . In order to achieve reliable battery packs for these applications, various thrusts of research and development must be conducted simultaneously, including but not limited to the design of advanced battery materials, cell configuration, circuit optimization, and diagnostics. With respect to the chemistries of electrode materials, various analytical diagnostic tools have been performed to probe chemical and structural environments from the bulk to the surface [5][6][7][8][9] . In-depth diagnostic analysis can provide insights into structure-property relationships 10,11 as well as into failure mechanisms 12,13 . In combination with many ensemble-averaged measurements, fast electron microscopy and spectroscopy are poised to reveal chemical and structural information with extremely high spatial resolution and potentially improved temporal resolution through in situ transmission electron microscopy (TEM) visualization 14,15 . To date, a vast number of studies have been successfully implemented by virtue of the state-of-the-art imaging and spectroscopy capabilities of fast electron microscopy.
The chemical environment at the particle surface determines the way an active electrode interacts with its surrounding electrolyte and impacts the resulting cycling behaviors. Surface modifications, such as artificial solidelectrode interphases, were reported to improve cycling performance 16,17 . On the other hand, electrochemistryinduced surface reaction layers (SRLs), such as solid-electrolyte interphases at anode material particle surfaces, also play critical roles in determining the way electrode materials interact with electrolytic solutions and how lithium ions intercalate 18 . SRLs are generally composed of oligomers and inorganic compounds that are formed from the electrochemical reduction of solvent and electrolytic salt 19 . Similar to the behavior of many organic compounds under electron beams, SRLs are prone to beam damage. Structural changes may also occur as a result of charge-discharge cycling, and have been proven to impact battery performance. For example, in performed following ex-situ TEM procedures, and the electrochemical performance was successfully correlated to the phenomena observed using TEM. Lithium-containing compounds (e.g., electrode materials, SRLs) are easily degraded under electron beams due to the knock-on effect (i.e., atomic displacements by electronnuclear collisions) 22 and thermal effects. It is therefore, critical to distinguish these processes from those arising from the electrochemistry to ensure the correct interpretation of the experimental results.
The present work was designed to understand the effect of electron beam damage on surface reaction layers and NMC materials during data acquisition using TEM. It is shown that the chemistry and structure can be severely altered under certain circumstances, such as low scan rates. The degradation processes resemble the lattice reconstruction and chemical evolution that are caused by chargedischarge cycling 5 . It is suggested that reducing the accelerating voltage and increasing acquisition speed (e.g., performing fast EELS) are necessary to avoid the beam damage.

Results
The identification of the chemical environment of SRLs (e.g., solidelectrolyte interphases) is essential to understanding their functionalities in the battery operation 19 . Owing to its high spatial resolution, STEM-EELS is believed to be one of the most advantageous techniques for this purpose. Carbonate groups have been shown to dominate at the surfaces of many cathode and anode materials. Figure 1 shows EELS measurements on a lithiated NiO nanosheet. In a previous study, we showed that, in NiO nanosheets, the SRLs are primarily composed of Li 2 CO 3 embedded in a complex organic matrix 6 . Indeed, the fingerprints for the 1s to p* transition of the CO 3 22 group are simultaneously observed in the C K-edge and O K-edge spectra, as shown in Figure 1b and Figure 1c, respectively. The well-defined characteristic peaks were gradually degraded after repeated EELS acquisition over the area indicated by the dashed boxed in Figure 1a, and the fine structures of the CO 3 22 group disappeared after four acquisitions. The compounds in the SRLs are usually prone to thermal decomposition 23 . Here we showed that the combined knock-on, ionization, and thermal effects of the electron beam readily broke down the surface structure leading to the degradation of the SRL. The damage can be mitigated by acquiring EELS data from a larger area while keeping the dose below the critical limit. In addition, the critical dose limit (measured in electrons/Å 2 ) is generally dependent on the dose rate (measured in electrons/Å 2 /sec). By decreasing the dose rate (i.e. reducing the incident beam current while keeping the scanning area constant), the damage rate can potentially be reduced 24 .
A focused electron beam was scanned over a LiNi 0.4 Mn 0.4 Co 0.18 Ti 0.02 O 2 (abbreviated as NMC hereafter) particle that had undergone 20 electrochemical cycles between 2.0-4.7 V vs. Li 1 /Li, ending in the discharged state. There are two dominant crystal structures found in the particle; R 3m and Fm 3m (Figure 2a). The Fm 3m rock-salt structure was generated by lattice reconstruction of R 3m (filling of R 3m 3a sites by transition metals) as a result of oxygen loss and concomitant lithium ion removal during cycling 5,21,25 . In Figure 2b and Fm 3m[1 10], respectively. During electron beam irradiation, the R 3m structure gradually converted to the Fm 3m structure, as recorded in a BF-STEM mode movie (Movie S1). Selected FFT patterns shown in Figure 2c give evidence of the conversion of the R 3m structure. Two characteristic diffraction spots, i.e., R(011) and R(00-3), are indicated in the patterns and their intensities were monitored during the irradiation process. Ultimately, after an extended irradiation period (13 min 50.5 sec, 110 th frame), these two diffraction spots lost their intensities completely. The superimposed FFT patterns are presented in Movie S2.
The crystal orientation relationship between the Fm 3m and R 3m structures is visualized by inverse FFT imaging, as shown in Figure 3. The Fm 3m (111) and R 3m (003) facets are selectively visualized by masking the corresponding diffraction spots in the FFT pattern (Figures 3a and 3b, and the insets therein). The lattice spacing of the R 3m (003) of the NMC phase is approximately twice as large as that of the Fm 3m (111) spacing of the rock-salt phase (MO, M 5 Ni, Mn, Co); i.e., there is ,5% of mismatch, so that the rock-salt phase grew epitaxially on the layered phase. The inset of Figure 3b exemplifies the epitaxial growth of Fm 3m on R 3m in real space. The inverse FFT images were generated for the R 3m/Fm 3m composite particle before (Figure 3c) and after (Figure 3d) electron beam damage. The initial R 3m/Fm 3m composite particle (Figures 3c) was completely converted to a pure Fm 3m phase after electron beam damage (Figure 3d). The phase transition process likely proceeded via transition metals moving from 3b sites to 3a sites, resulting in the collapse of the layered structure and the formation of a pure rock-salt structure.
The requirement for simultaneous lithium and oxygen removal during the R 3m-to-Fm 3m transition implies that the transition metals are reduced during the process. The changes in the oxidation state of transition metals (TMs) were monitored by EELS and are shown in Figure 4. The transition metal L-edge EELS measures the dipole allowed transitions from metal 2p orbitals to unoccupied metal 3d orbitals, including both the 2p 3/2 (L 3 ) and 2p 1/2 (L 2 ) spinorbit final states, and indirectly probes the local hybridization states for metal-oxygen octahedral units (i.e., TMO 6 ) in NMC materials [26][27][28] . The O K-edge EELS corresponds to the transition from O1s states to unoccupied O2p states in the conduction band. Due to the TM3d-O2p hybridization, O K-edge spectra also reflect the unoccupied TM3d states 29 . Although the exact nature of the hole states (e.g., location) depend on the degree of covalency in the TMO 6 octahedral unit, it is generally agreed that, in the (Ni, Mn, Co)-O 6 octahedral unit, the intensity of the pre-edge of the O K-edge is in a positive relationship with the formal oxidation states of (Ni, Mn, Co) due to the sharing of hole states in the TMO 6 unit 30 . A continuous data acquisition mode was used to collect EELS spectra on a selected region shown in Figure 4a. As the acquisition proceeded, the electron beam gradually damaged the NMC particle. There was an incubation period from 0-40 s in the integrated pre-edge intensity before it dropped dramatically at 50 s, indicating that an accumulation of lattice interruptions by electron beam was  Figures 4d, 4e, 4g and 4h), there are noticeable peak shifts towards lower energy for Mn-L 3 and Co-L 3 edges, indicating that Mn and Co were both gradually reduced. As expected, Ni remained in the 12 oxidation state in the R 3m and Fm 3m structures. The synergistic changes in the oxidation states of oxygen and transition metals suggest that the hole states are shared by the transition metals and oxygen in the TMO 6 unit. Each spectrum was integrated for 10 sec with a dose rate of 113,000 electrons/Å 2 /sec and an accumulated dose of 1,130,000 electrons/Å 2 . It shows that the critical dose limit for NMC material is approximately 6,000,000 electrons/Å 2 , which is four orders of magnitude higher than that of lithium carbonate. Decreasing the dose rate is critical in reducing radiation damage because isolated bond breaking is healable in conductive solid samples 24,31 . Rastering the beam over an area is an efficient way to reduce the dose rate. If the beam is idle, the dose rate is the beam current normalized by the beam area. If the beam is scanned sufficiently fast, the dose is spread out over the scanning area; in this case, the dose rate is the beam current normalized by the entire scanning area. To test how scan rate affect the dose rate, we performed a parallel experiment to compare with the condition used in Figure 2. As shown in Figure 5, we preserved the beam current, the field of view, and the dwell time but reduced the pixel number by a factor of four (i.e. scanning was four times faster). When the beam damage was compared under these two conditions after similar number of frames (i.e., similar repeat of electron beam scanning), it shows that the structural damage was significant less than that in Figure 2. In the R 3m/Fm 3m/ Fd 3m composite particle (Figure 5a), the R 3m phase was preserved even after extended number of frames (compare Figures 5b and 5c). This suggests that ultrafast scanning in conjunction with drift correction is desirable for STEM imaging of radiation sensitive battery materials.

Discussion
We have shown that electron beam damage is a common phenomenon affecting battery materials studied by TEM. For example, the Li 2 CO 3 component in the SRL of a lithiated NiO nanosheet rapidly decomposed during EELS data acquisition. NMC cathode materials also degraded at low scan rates or with intensive doses. Li K-edge EELS energy onset is buried between the M-edges of transition metals in NMC cathode materials, while transition metal oxidation states are also prone to change. These intrinsic characteristics make the direct evidence of lithium removal unlikely using Li K-edge EELS during irradiation. Nevertheless, during electron beam irradiation, the formal oxidation states of Mn and Co were reduced from 14 to 12 and from 13 to 12, respectively. Therefore, correlating the present in situ BF-STEM and EELS results with previous studies 5, 21 , one can determine that the degradation process involved simultaneous lithium and oxygen removal from the 3a and 6c sites of the R 3m structure, and transition metals moving from the 3b sites to 3a sites, resulting in the collapse of the layered structure and ultimate conversion to a Fm 3m structure. The degradation process is similar to the lattice reconstruction observed in several classes of layered cathode materials, including NMC, lithium-rich Li(Li y Ni x2y Mn x Co 122x )O 2 , and lithium-rich/manganese-rich xLi 2 MnO 3 ?(1 2 x)LiMO 2 (M 5 Mn, Ni, Co, etc.) materials. It is also likely that the electron beam reduces transition metals first due to its strong reducing characteristics and then knocks out and/or evaporates lithium and oxygen from the NMC material, eventually leading to transition to an Fm 3m structure. Furthermore, due to the thermodynamic instability of charged NMC materials 5 , similar degradation is supposed to occur in the charged NMC particles. The present study suggests that, under some circumstances, the lattice reconstruction induced by electron beams can be used as a model system to study structural dynamics in electrochemical processes. Finally, special caution is recommended in studying battery materials to avoid indiscriminate sample degradation. This can be achieved by increasing the irradiation area and the scan rate.

Methods
Materials synthesis and battery cycling. The syntheses of NiO and LiNi 0.4 Mn 0.4 Co 0.18 Ti 0.02 O 2 (NMC) were performed according to previously developed protocols 5,6 . In short, NiO was synthesized using a solvothermal method aided with an alcohol pseudo-supercritical drying technique. NMC was synthesized using a co-precipitation method followed by high-temperature annealing with LiOH. 2032 Coin cells were fabricated using composites of NiO or NMC as working electrodes and lithium metal foils as counter electrodes. The NiO working electrodes were composed of 80 wt.% active material, 10 wt.% polyvinylidene fluoride (Kureha Chemical Ind. Co. Ltd) and 10 wt.% acetylene carbon black (Denka, 50% compressed) and loadings were typically 1-2 mg/cm 2 of active material. To make the electrodes, these solids were mixed into N-methyl-2-pyrrolidinone and the resulting slurry cast onto copper current collectors and dried. NMC working electrodes were prepared similarly and contained 84 wt.% active material, 8 wt.% polyvinylidene fluoride, 4 wt.% acetylene carbon black and 4 wt.% SFG-6 synthetic graphite on carbon-coated aluminum current collectors, with typical active material loadings of 6-7 mg/cm 2 . The coin cells were assembled in a helium-filled glove box using Celgard 2400 separators and 1 M LiPF 6 electrolyte in 152 w/w ethylene carbonate/dimethyl carbonate (Ferro Corporation). Battery testing was performed on a computer controlled VMP3 potentiostat/galvanostat (BioLogic). NiO and NMC electrodes were cycled at C/2 and C/20 rates, respectively. 1C was defined as fully discharging or charging an electrode in 1 h, corresponding to specific current densities of 718 mA/g and 280 mA/g for NiO and NMC materials, respectively.
Electron microscopy and spectroscopy. For electron microscopy and spectroscopy measurements, the electrode particles were scratched off and deposited onto TEM grids after the desired number of electrochemical cycles. A 200 keV and a 300 keV field-emission (scanning) transmission electron microscope (S/TEM) were used for in situ imaging and spectroscopic studies. The 200 keV microscope was operated with an imaging condition of 11 mrad, 12 pA. The 300 keV instrument was operated with 17 mrad, 36 pA. Electron energy loss spectroscopy data sets were acquired using Gatan Tridiem spectrometers. 3 512 pixels, pixel size 0.306 angstrom, pixel dwell time 6 us, frame rate 1.89 sec/frame). (a) BF-STEM image of a particle edge region with similar thickness and crystal orientation relative to that of Figure 2a. (b) FFT pattern of the composite particle before electron beam damage, where the rock-salt, spinel and layered structures are observed in the pattern. Note that the Fd 3m phase is an intermediate during the R 3m-to-Fm 3m transition in electrochemical cycles 5,21 . (c) FFT pattern of the composite particle after electron beam damage, which corresponds to the 200 th frame (6 min 18 s) in the series. The corresponding BF-STEM movie is provided in the supplemental information (Movie S3). F, R and S represent the Fm 3m, R 3m and Fd 3m phases in the FFT indices, respectively. www.nature.com/scientificreports SCIENTIFIC REPORTS | 4 : 5694 | DOI: 10.1038/srep05694