Toward Operando Structural, Chemical, and Electrochemical Analyses of Solid-State Batteries Using Correlative Secondary Ion Mass Spectrometry Imaging

The global transition from fossil fuels to green energy underpins the need for efficient and reliable energy storage systems. Advanced analysis and characterization of battery materials is not only important to understand fundamental battery properties but also crucial for their continued development. A deep understanding of these systems is often difficult to obtain through only pre- and/or post-mortem analyses, with the full complexity of a battery being hidden in its operational state. Thus, we have developed an operando methodology to analyze solid-state batteries (SSBs) structurally as well as chemically before, during, and after cycling. The approach is based on a specially designed sample holder, which enables a variety of electrochemical experiments. Since the entire workflow is performed within a single focused ion beam scanning electron microscope equipped with an in-house developed magnetic sector secondary ion mass spectrometer, we are able to pause the cycling at any time, perform analysis, and then continue cycling. Microstructural analysis is performed via secondary electron imaging, and the chemical mapping is performed using the secondary ion mass spectrometer. In this proof-of-concept study, we were able to identify dendrites in a short-circuited symmetric cell and to chemically map dendritic structures. While this methodology focuses on SSBs, the approach can directly be adapted to different battery systems and beyond. Our technique clearly has an advantage over many alternatives for battery analysis as no transfer of samples between instruments is needed and a correlation between the microstructure, chemical composition, and electrochemical performance is obtained directly.


Sample preparation
To perform a proof of concept of the operando measurement, three different ways to divide the sample and treat the freshly exposed surface of interest were tested. As a plane sample surface is required to successfully perform SIMS analysis, we decided to divide the LLZO pellets in half. The three different preparation methods were:  Cutting the LLZO pellet with a wire saw and cleaning with 1 M HCl followed by 1 h vacuum drying at room temperature.  Cutting the LLZO pellet with a wire saw, polishing with SiC sandpaper up to #4000 and cleaning with 1M HCl followed by 1 h vacuum drying at room temperature.  Physically breaking the LLZO pellet by hand and cleaning with 1M HCl followed by 1 h vacuum drying at room temperature.
In order to get a straight edge when physically breaking the LLZO pellet, a notch was carefully carved with a cutter knife to dictate the direction of breaking.
A comparison of the three sample preparation methods has been made based on the microstructural appearance (laser profilometer), the SEM image quality and the SIMS feasibility ( Figure S1). Figure S1 a-c) show SEM images of a polished LLZO surface, an unpolished surface (cut via wire saw), and a sample surface which has been cracked by hand, respectively. The polished sample ( Figure S1 a) has the advantage that it has the least topography, hence resulting in promising SIMS results. The unpolished sample (Figure S1 b) has a very rough topography with artefacts in SIMS. The line profiles in d) reveal that the polished sample is obviously the one with least topography. The unpolished sample shows topographic differences with up to 10 μm which is very unfavourable for SIMS analysis. The cracked sample reveals more pronounced roughness than the polished sample, however it has plateaus which are large and flat enough to successfully perform SIMS analysis. Unlike the polished sample, the cracked sample has the benefit that the granular shape of LLZO remains intact and that there is no crosscontamination possible related to cutting (often lubricant or liquid used) or polishing. One drawback of the polished sample is that polishing introduces small artefacts, e.g. lines that can clearly be seen in the SEM (Figure S1 a) and in the SIMS image (Figure S1 e). One the other hand, the cracked sample also presents a minor drawback: due to locally different incidence angles of the primary beam, slightly different sputter yields are obtained, resulting in different counts at different faces of a LLZO grain (details under Results and Discussion). Nonetheless, we decided that the cracked sample has the lowest risk of cross-contamination (structurally as well as chemically) and presents very good SIMS results in the ratio of sputter yields, hence was selected to be the sample preparation of choice for this proof-of-concept study.

Electrochemical experiments
The electrochemical experiments such as constant current cycling, chronopotentiometry or impedance spectroscopy were performed using a SP-150 potentiostat from BioLogic. All test measurements with commercial coin cells have been performed either outside or inside but vented FIB/SEM sample chamber to prevent contamination of the chamber due to possible outgassing. Several electrochemical tests have been performed with commercial coin cell batteries, to validate the prototype design and troubleshoot potential issues. Figure S2 a) shows an example of a chrono potentiometric discharge (with 20 mA from 3.2 V to ~1.5 V) of a RS Pro CR2032 3 V Lithium Manganese coin battery, 225 mAh (primary battery). The focus of this test was the optimisation of contact between micromanipulator and electrode. The black noisy line presents an example where the contact between the micromanipulator and the corresponding plate of the sample holder was not ideal. In this case, we contacted the electrode via the μ-tip (see SE image in Figure S1 b). To increase the contact area between micromanipulator and electrode, we decided to remove the μ-tip and instead to contact the electrode with the bigger supporting rod onto which the microscopic tip is usually attached. The red curve shows the same experiment but with an intimate contact between micromanipulator and electrode, resulting in a smooth discharge curve. . Figure S2 c) shows the experimental testing set-up outside the FIB-SEM instrument for testing commercial coin cells. Additional test measurements have been performed to validate and ensure a proper and artefact-free operation of our custom-designed operando sample holder. Figure S2 d) shows EIS test measurements which have been performed in a controlled atmosphere of a glovebox. The same Li/LLZO/Li sample has been measured in a conventional Swagelok cell and in the operando sample holder. The comparison of both shows that no considerable difference is noticeable, except that the impedance (R SE/Li ) is lower in the operando holder than in the Swagelok cell. Latter is attributed to different pressures being applied for clamping the sample.

Figure S2: a) chronopotentiometry of a RS Pro CR2032 3 V Lithium Manganese coin battery with bad contact (black) and good contact (red) between micromanipulator and electrode. b) SEM image of the micromanipulator (supporting rod and the μtip) c) experimental testing set-up outside the FIB-SEM instrument. Red electrode attached to a lifting platform imitating the sample stage and the crocodile clip of the blue cable holding the micromanipulator that contacts the other electrode. d) EIS measurements of the same Li/LLZO/Li half-cell inside a conventional Swagelok cell and inside the operando sample holder. The equivalent circuit for the fitting is shown as inset and the impedance contributions of grain boundary (R gb ) and solid electrolyte -Li interface (R SE/Li ) are exemplarily shown.
Additionally, the electrical resistance owing to the experimental setup was assessed. These tests have been performed with two different surface mount resistors (100 Ω and 200 Ω), and for each the experiment has been performed with 1, 2 and 3 resistors. Figure S3 shows how the resistors were mounted on the sample holder and the zoom-in shows an SEM image of one of the resistors. The sample holder itself only adds an average of 1.3± 0.8 Ω to the system which is indeed negligible. The measurements performed with the sample holder inside the microscope and contacted using the micromanipulator add an average of 12.8 ± 3.5 Ω to the system, also this is negligible compared to the electrochemical systems which will be analysed.

Adaptation to inert gas transfer system and microscope
A contamination free sample preparation and sample transfer is possible with an inert gas transfer system which in our case is composed of a transfer chamber on the microscope and a portable air-and vacuum-tight transfer box ( Figure S4 a, b). This box can be introduced in a glove box, and the operando sample holder (with the sample) can be locked inside the transfer box. Subsequently the transfer box can be removed from the glove box and attached to the air lock at the microscope ( Figure S4 c, d), where the argon atmosphere inside the box gets pumped before the sample can be introduced inside the microscope.   Table   Table S1: Fit parameters for impedance spectroscopy of a symmetric Li/LLZO/Li cell shown in Figure 1b) The EIS measurements (Figure 1 b) were done in potentiostatic mode in a frequency range from 1 MHz to 1 Hz with an amplitude of 5 mV. The bulk resistance is not accessible in the frequency range which was used for this experiment. However, a third semi-circle (impedance at low frequencies) of unidentified origin is needed to fit the experimental data. A detailed investigation of this is beyond the scope of this study.