TEM sample preparation for Li-ion secondary battery using ion slicer


 The main purpose in this paper is a sample preparation of transmission electron microscopy (TEM) for the lithium-ion secondary battery in the form of micro-sized powders. To avoid artefacts of the TEM sample preparation, the use of ion slicer milling for thinning and maintaining the intrinsic structure is described. Argon-ion milling techniques have been widely examined to make the optimized specimen, which makes TEM analysis more reliable. In the past few years, the correction of spherical aberration (Cs) in scanning transmission electron microscopy (STEM) has been developing rapidly, that results in the direct observation with the atomic level resolution not only for the high acceleration voltage but also its deaccelerated voltage as well. Especially, low-kV application has been markedly increased that needs the sufficient-transparent specimen without the structural distortion during the process of the sample preparation. In this study, the sample preparation for the high-resolution STEM observation has been greatly accomplished and investigations of its crystal integrity are carried out by Cs-corrected STEM.


Introduction
Lithium-ion secondary batteries comprising a positive electrode, a negative electrode, and an electrolyte have been the most revolutionary material as a type of rechargeable battery with the high capacity for the energy density per unit area. Generally, the negative electrode mainly consists of carbon, which is a theoretical capacity of 372mAh/g in the fully lithiated state (Shao, 2020). To further increase the gravimetric and volumetric capacity, silicon has been studied as one of the most promising negative electrode materials due to its higher theoretical capacity of 3579mAh/g (Obrovac, 2004;Key, 2009;Miao). Positive electrode materials including transition metals and lithium-ionic composites containing Li, Mn, Ni, Co, and O atoms play an important role in determining the performance and its cost. For that reason, it is essential to accurately identify the structural information of electrode materials at the atomic-level.
Although many significant researches for the characterization of the lithium-ion battery have been contributed, but identifying materials for revealing the atomic-scale mechanism of the lithium-ion secondary battery has been rarely reported so far (Pender, 2020;Yamada, 2013;Miao, 2019;Nitta, 2015). The reason that I pointed out is because of the difficulty in preparing a homogeneous TEM specimen against the structural distortion. In general, lithium-ion secondary batteries are synthesized in the powder form with a size of several to tens micrometers, and it is not easy to make them as the uniform structure with a thickness of 100nm or less. If the powder has a size of 100nm or less, it can be easily dispersed in a solvent and placed on a grid to be sufficiently transmitted and observed in TEM, however, in the case of secondary batteries, due to the size restriction, it must be made thin enough by the crosssectional operation to allow electrons to penetrate through the specimen. Generally, the epoxy molding method has been used for preparing powder samples, but there is a problem in processing of having resulted from the inhomogeneous phase/chemical distribution due to the difference in the milling rate between epoxy and powder (Jiang, 2019). And the curtaining effect occurs due to thickness variation according to the powder shape, thereby, the condition of the sample in the local area could be severely deteriorated. In order to overcome these problems, the broad argon ion beam (hereinafter referred to as BIB) and the backside-ion milling by means of ion slicer were introduced (Kato, 2004). Current research focused on the practical aspects for the experimental analysis, due to the problem of beam damage caused by the high acceleration voltage in TEM analysis, the application fields of acceleration voltages of 100kV or less have greatly increased not only for secondary batteries, but also for 2D materials, semiconductor electronic circuit samples micronized to several nanometers, and so on. TEM analysis at a low acceleration voltage has a desirable thickness of 50nm or less by reason of the electron transmittance and the chromatic aberration (Sasaki, 2010). Another widely used technique to perform the TEM sample preparation called focused-ion beam (FIB) that enables the specimen to be thinned after obtaining cross-sectioned lamella by the raster scan on the bulk sample using gallium ions generated from liquid metal ion source (LMIS).
Basically, it has a scanning electron microscope-based column, so it has a great advantage of targeting the region of interest (ROI) of the sample while observing the electron image, but it was often observed that artefacts such as sample alterations like the redeposition, the amorphization, vacancies and the gallium implantation from the reaction between the heavy gallium ions and the sample (Basnar, 2003;Roediger, 2011). In this study, lithium-ion secondary battery powders were preliminarily milled in the initial thinning and then finally milled in the backside-ion milling using the ion slicer and evaluate it with a spherical aberration corrected STEM. The ion slicer uses BIB and protects the surface by the inserted shield belt on the center of the area to be milled. Thinning was performed by sequentially irradiating both sides with incident ions at a low angle and simultaneously controlling the stage to create a rocking beam. And because it is processed with mechanical polishing down to 100 micrometers-thickness (μmt), it can be easily handling and can prevent the deformation of the lattice structure caused by the mechanical polishing. Since the conventional ion milling method has been used with mechanical polishing down to 10μmt or less, consequently, artefacts were often appeared and making it difficult to observe in TEM.

Experimental methods
LiMn2O3 powders used staring in composite preparation was supplied by Hanyang university. Figure. 1 (a) shows a method for preparing an ion slicer sample of LiMn2O4 powders that mixed with the G2 epoxy resin/hardener in a ratio of 10:1 to make the powder/epoxy blend.
This epoxy mixture was placed on the Si wafer and then the cover-glass was finally covered.
The thickness of the cover-glass affects the milling time of the argon beam reaching to the resin/powder blend, which is the region of interest, so 100μmt of the cover-glass is recommended for the appropriate proceeding during the secondary operation of backside-ion milling. For the homogeneous milling, the internal porous structure should be avoided and it is also useful to use pressure tongs or decompression devices as a method of removing air bubbles in the blend, then, sufficiently harden it on the hot plate for about 20-30 minutes at temperature of 100°C. The sample thus obtained was cut into 2.8mm x 1mm x 500μm size using a diamond saw, then, polished until the 1mm-thick sample becomes the 100μm-thickness with a mechanical polisher using the diamond paper (abrasive grit size: 30μm) as shown in fig. 1(b).
And the 100μmt-thinned sample was loaded on the holder stub of the ion slicer (JEOL, Japan) and the argon ion milling was proceeded after adjusting the shield belt to the center. A preliminary milling in a large area by BIB was performed for about 3 hours under the following conditions; the acceleration voltage and the incident beam angle were 6kV and 0.4°, respectively. After that, the specimen holder turned upside down and mounted it in order to implement the backside-ion milling at 6kV and 4° for about 1 hour during the ion beam etching by being milled from the bottom of the cover-glass to the composite section. A post-milling in ensuring that powders with relatively high strength compared to resin can be homogeneously processed was sequentially performed at 3.5° and 3° for 20 minutes, respectively. the final milling step was performed under the acceleration voltage of 1kV for 10 minutes. For the STEM observation, the processed sample was carefully mounted in a reinforcement ring with a diameter of 3mm. Finally, the sample area in thin enough for the electron penetration was evaluated through a probe aberration corrected 200kV STEM with cold field emission (JEM-ARM200CF; JEOL, Japan) with 0.3eV energy resolution. Fig.3 shows high-angle annular dark field (HAADF)-STEM images at low magnification shown as the whole of the sample made by FIB and Ion slicer, respectively. The vertical curtaining effect generated by the surface roughness was observed with vertically parallel to the ion beam direction as shown in fig. 3(a). On the contrary, the HAADF STEM image of the specimen prepared by Ion slicer using BIB does not appear the curtaining effect inside the cross-sectional area, even in the same sample having a similar shape as shown in fig. 3(b). Fig.   4 is a high-resolution HAADF-STEM image taken to observe the structural state and information at the atomic-level. Fig. 4(a) is an image taken from the surface of the powder, and the surface side of the powder was milled sufficiently so that the atomic arrangement was clearly observed, but it gradually deteriorates toward the inside. This was caused by the difference in the milling rate of resin/powder, and has long been considered as an obstacle problem in the TEM sample preparation for powder samples. On the other hand, In the case of the sample processed by the BIB technique as shown in fig. 4(b), the cross-sectional ion milling was processed cleanly and completely without any deformation or bending of the sample even in the inner region. Ion slicer has a shield belt in the center of the area to be thinned, so it not only protects the surface, but also enables that the curtaining effect does not occur because the uniformly dispersed argon ion beam hits the area to be thinned. In addition, the phenomenon caused by the milling rate difference in the powder/resin sample was significantly reduced. In order to identify the curtaining effect generated in the FIB sample, high-resolution HAADF-STEM images are obtained at high magnification. Fig. 5 sequentially shows the area where the curtaining effect occurred and the internal atomic structure was observed while magnifying from (a) to (f). It can be seen that the curtaining effect was severely occupied in the magnified image of (b) contrary to the low magnification of (a). In the high-resolution images from fig.   7(d) to (f), it can be seen that the area where the curtaining effect occurred is relatively thin by means of the dark-contrast in HAADF imaging. And the thickness deviation, and structural deformation had been occurred due to the damage caused by the gallium ion beam. In practice, this phenomenon poses problems for intrinsically interpreting the crystallographic structure.

Results and discussion
To avoid the curtaining effect, a sample was processed using BIB and high-resolution HAADF-STEM images and annular bright field (ABF)-STEM as shown in fig. 6(a) and fig. 6(b) were obtained with the beam direction [311]. Based on the results, the curtaining effect was significantly reduced, but the surface damage and the contamination were observed. In particular, it can be seen that the atomic arrangement was not clearly observed in the ABF image including the phase contrast. To solve this problem, the backside-ion milling was performed as a method of reducing the surface damage and the contamination. Hence, the conventional and backside-ion milling were compared and evaluated with high-resolution HAADF-STEM as shown in fig. 7. The backside-ion milling in which the structure was more clearly observed than the conventional ion milling. It can be seen that the surface was processed thinner as well as the difference in the surface damage and the contamination. Further, to observe the light element atomic column, HAADF and ABF STEM were simultaneously obtained to analyze the structure. Fig. 8(a) and (b) are high-resolution HAADF-and ABF-STEM images taken with the beam direction of [110], and the atomic modeling for the LiMn2O4 crystal structure is shown in fig. 8(c). Due to the Z-contrast, which is the scattering scales as approximately a Z 2 dependence for common scattering of the HAADF image, the distribution of Mn, which is as heavy element, can be easily recognized, but Li and O, which are light elements, are not observed (Pennycook, 1989). Therefore, when observing ABF image with relatively low atomic number dependence with the electron channeling effect, not only light elements but also heavy elements can be observed (Findlay, 2016). By directly comparing these two images, the column sites of Li and O were understood and it can be confirmed that they corresponded to the atomic model of the LiMn2O3 structure viewed from the spinel crystallographic orientation.

Conclusions
Through this study, the TEM sample preparation of the secondary battery powder which was usually hard to handle was carried out using the backside-ion milling technique. It has been confirmed that this technique facilitates TEM specimen having thin and uniform. For the precise analysis, it is important to prepare the specimen containing the intrinsic state of materials. Furthermore, the ultra-thin TEM sample preparation technique has been gradually essential for various applications like the low-acceleration voltage analysis, electron energyloss spectroscopy (EELS) analysis for the 0-500eV spectrum region containing vibrational (phonon) losses, etc. Figure 1 Schematic diagrams of the sample preparation showing a procedure for LiMn2O4 powder and epoxy mixture (a) and a conventional method for cutting before ion milling (  High angle annular dark eld STEM images of LiMn2O4 powder which are milled by the focused ion beam (a) and the ion slicer (b), respectively.

Figure 4
High resolution HAADF STEM images showing the edge surface of the LiMn2O4 powder prepared by the FIB milling (a) and the inner area prepared by the IS milling (b).   High resolution HAADF STEM images of the backside ion milled (left) and the conventional ion milled Figure 8 High resolution STEM image of the IS milled specimen shows the atomic arrangement in HAADF image (a) and light elements (Li, O) are clearly seen in ABF image (b). A schematic representation of LiMn2O4 crystal structure Figure 9 High resolution HAADF STEM image (a) and annular bright eld image (b).