Simultaneous Monitoring of Structural Changes and Phase Distribution of LiFePO4 Along the Cathode Thickness of Li Metal Polymer Battery

Probing the structural changes that electrode materials undergo during electrochemical cycling while monitoring their spatial distribution within the volume gives valuable insights on dynamic processes, i.e. side reactions and evolution of phase migration barriers, often associated to capacity and power limitation. In this work, we present an electrochemical cell to perform spatial and time resolved operando synchrotron X-ray diffraction on Lithium (Li) metal polymer batteries operating at 80 °C. A 3.2 mm diameter battery made of a Li metal anode, a LiFePO 4 based cathode, and a solid polymer electrolyte acting as separator, is placed inside a glass-based casing and cycled at a beamline. The cylindrical cell geometry with its small size enables to follow the phase transformations occurring at different states of charge and at different cathode heights. It is possible to create spatially resolved phase distribution plots and to differentiate active material structural changes occurring close to the interface with the electrolyte from those at the current collector vicinity. The results provide a direct observation of the Li diffusion in the LiFePO 4 and FePO 4 phase distribution. In addition, synchrotron X-ray diffraction computed tomography (XRD-CT) measurements were performed to obtain phase distribution maps at different heights of the battery assembly.


Introduction
Lithium (Li) based batteries lead the marked of power electronic devices, hybrid and full electric vehicles, and are a promising solution for storing and releasing the energy generated by stationary applications, such as solar panels and wind turbines 1,2 . In order to meet the demands required for the above-mentioned applications, the development of improved high-energy and high-power density electrode materials with stable reversible capacities after prolonged cycling is essential.
The design and engineering of electroactive materials requires a comprehensive understanding of the complex interplay between their structural properties and their electrochemical performance in a cathode 3 . Indeed, it is known that the occurrence of incomplete or secondary (parasitic) reactions within an electrode, either due to phase migration barriers, phase segregation or merely local limitations in the electronic and ionic transport, can severely affect the redox properties and cycle life of the active materials [4][5][6] . Thus, the sole study of their structural evolution upon charge and discharge is not sufficient to fully interpret the electrochemical performance of battery systems and operando analysis is needed to identify processes leading to capacity and power fade.
An interesting example is LiFePO4, which has been intensively studied due to its great performance as cathode of Li-based batteries 7,8 . Numerous structural studies have shown that at low de(lithiation) rates LiFePO4 (LFP) transforms to FePO4 (FP) through a first-order transition. In addition, it was recently reported that at high de(lithiation) rates the first order phase transformation is suppressed and LFP transitions to FP via solid solution reaction 9,10 . In either case, the phase boundary propagation plays an important role in the de(lithiation) kinetics. In view of this, many reports have focused in the study of the two-phase boundary evolution upon electrochemical de(lithiation) through different experimental [11][12][13] and theoretical approaches 14,15 . However, even for well-studied active materials, such as LFP, few reports have tracked the dynamics of the spatial distribution of redox reaction products within the electrode upon electrochemical cycling [16][17][18] . This information allows detecting nonuniform chemical processes and identifying local domains of unreacted material, which is crucial for the development of thick electrodes with enhanced specific energy densities and for optimizing rate-dependent battery performance.
Operando synchrotron X-ray diffraction (XRD) measurements allow to obtain structural information of the battery components in times shorter than a second, which makes it a powerful tool to track in real time phase transformations and structural changes on the anode and cathode, and relate them to the battery electrochemical performance [19][20][21][22] . Operando XRD experiments on battery systems are usually performed on modified coin cells 23 , pouch cells 24,25 and the so-called "Swagelok" cells 26,27 using a transmission geometry. In the standard configuration of these cells, the X-ray beam is parallel to the electrode stack axis and the diffracted signal contains contributions from the two electrodes, the separator, the current collector, the electrolyte and the casing. However, in the last years novel cell designs have shown the great benefit of using cylindrical cells, in which measurements can be performed in transmission mode with the X-ray beam perpendicular to the electrode stack axis [28][29][30][31][32][33] . Such configuration enables not only to separately obtain structural information of the different components, but also to collect information at different heights within an individual component. This allows to differentiate processes that occur at the interface between the electrode and the electrolyte from those taking place at the interface between the electrode and the current collector. In consequence, operando XRD experiments on cylindrical shaped cells permits to finely monitor the evolution of the redox reactions with a spatially resolved approach.
In this work, we present a simple glass-based electrochemical cell suitable for in situ/operando synchrotron XRD and XRD computed tomography (XRD-CT) studies. A 3.2 mm diameter Li metal battery comprising LFP composite cathode and a poly(ethylene oxide) based solid polymer electrolyte (SPE) was assembled within the electrochemical cell and cycled at 80 ºC. Operando synchrotron XRD measurements performed along the electrode height allowed to track in real time the (de)lithiation reactions while visualizing the advance of the reaction front and the phase distribution of LFP and FP. In addition, ex-situ XRD-CT tomography enabled to map the spatial distribution of the different phases at specific heights of the battery and to clearly distinguish the different battery components.

Battery assembly
The battery assembly consisted in the superposition of a Li metal anode, a solid polymer electrolyte (SPE) layer, and a composite LiFePO4 based cathode 34 . The Li metal of typical thickness 60 ± 5 µm was stored in an Argon filled glovebox (Campus, Jacomex) with sub-ppm values of oxygen and water. The SPE is a random copolymer comprising poly(ethylene oxide) (PEO), to ensure ionic motion, and poly(propylene oxide), to provide flexibility to the resulting membrane. The copolymer was doped with lithium bis-trifluoromethanesulfonimide (LiTFSI) salt at the concentration required to ensure a molar ratio of ethylene oxide to Li salt (EO:Li) of 25. The copolymer and LiTFSI were dissolved in Dimethylformamide by stirring for several hours in a vial at 80 ºC. When the solution was clear and transparent, it was poured onto a Teflon Petri dish and dried at room temperature for 8 h. To remove any traces of solvent the Petri dish was further dried in an oven at 60 ºC for three days. The resulting membrane was peeled off the Teflon substrate and transferred to the glovebox for at least a week prior to any further manipulation. The SPE was then cut into pieces, hot pressed at 80 ºC and then pressed at 200 bar for several minutes. After pressing and cooling, 18 µm thick SPE disks were punched out.
Cathode composite electrodes using LiFePO4 active material were prepared by Blue Solutions through an extrusion process. The thickness of the cathode deposited on top of an Al current collector was about 48 µm. Inside the glovebox, a 2 mm diameter Li disk and a 3 mm diameter SPE disk were punched out from the Li foil and the SPE, respectively. Both materials were laminated at 80 ºC and 3 bars multiple times using a home-made laminating machine until the SPE was fully adherent to the Li. The electrolyte thickness was checked after the laminating process and no variation was observed. Then, a 3 mm diameter cathode disk was also punched out from the electrode foil and gently pressed onto the SPE. The Li metal polymer battery was obtained after laminating several times the three layers to ensure good adhesion between them.
After assembly, the batteries were placed inside the electrochemical cells.

Operando X-ray diffraction
Operando X-ray diffraction measurements were performed at the High Energy X-ray Diffraction Beamline ID31 at the European Synchrotron Radiation Facility (ESRF). Figure 1 shows a sketch of the experimental configuration.
The electrochemical cells were connected to a SP150 or a SP240 (BioLogic SAS) potentiostat and cycled at 80 °C between 2.75 V and 3.8 V vs. Li + /Li. The batteries were galvanostatically cycled at current density (J) of 0.06 mA.cm -2 corresponding to a charge and discharge rate of C/8. At the end of the galvanostatic charge step, a constant voltage of 3.8 V vs. Li + /Li was applied for 1 h. In between each charge and discharge step, the cells were subjected to a 18 min resting period in order to relax the concentration gradients. At each step, the charge or discharge capacity were calculated by integration of J over time (t) and converted into fraction x of Li in the LixFePO4 with 0 < x < 1.
Synchrotron XRD data was collected upon electrochemical cycling using a 70 KeV monochromatic X-ray beam. The detector to sample position was calibrated using a NIST CeO2 A custom-made heater was built to perform operando XRD measurements at 80 ºC. In order to optimize beamtime usage, the heater was designed to allow the simultaneous operation of two glass cells. Pictures of the experimental setup are displayed in Figure S1 of the electronic supporting information, and schemes and a detailed description of the heater system are presented in Figure S2.

X-ray diffraction computed tomography (XRD-CT)
The glass electrochemical cell was mounted onto a Huber goniometer head. X-ray diffraction computed tomography measurements were performed using 50 translation steps, with a step size of 60 µm for each translation, covering 0 -180 degrees angular range. In total 180 angles were probed. A light 10 % trimmed mean filter was applied to remove any outliers.
The 2D diffraction image data sets were converted to 1D powder diffraction patterns using the pyFAI software. Entire diffractograms were reconstructed for each voxel of the cell using an in-house developed software adapted from Finegan et al. 37 .

Electrochemical glass cell
Many of the reported cylindrical electrochemical cells used for in situ/operando XRD experiments on battery systems utilize polytetrafluoroethylene (PTFE) or perfluoroalkaoxy alkane (PFA) casings. However, these polymers are sensitive to radiation and damage can be observed after prolonged exposure times at high X-ray doses 38 . Furthermore, they exhibit sharp Bragg reflections within the 2θ range that is of interest for most electrode materials. To diminish the background signal and X-ray absorption of the housing, we developed an electrochemical cell using a borosilicate glass tube as main component. In the case of using electrolytes that might be reactive to glass, a thin Kapton tube can be introduced in between the glass tube and the sample to prevent direct contact between them. Swagelok fittings were used to ensure a proper sealing and airtight conditions in the interior of the cell. The electrochemical cell is designed to use simple and commercial components and can be adapted  Figure S1).
To obtain a direct comparison between the background signal of the proposed glass and that of the standard PTFE Swagelok cells, a PTFE cell of 3.3 mm inner diameter was built using Swagelok straight unions as the main body. The inner and outer parts of the tube were machined in order to ensure a smooth pathway in the interior of the cell and to diminish the thickness of the PTFE wall on the way of the beam. A picture and a side-view sketch of the cell, together with its corresponding X-ray diffraction signal are depicted in Figure 2b.
The diffraction signals of the empty glass and PTFE cells are plotted together in Figure   2b. The PTFE-based cell exhibits a strong Bragg reflection at 2θ = 2.07 ° together with several small peaks at 2θ values ranging from 3.5 ° to 6.5 ° on top of a very noticeable background signal. In the case of the glass cell, a bump of intensity is observed at 2θ = 2.49 °, which evidences the amorphous nature of the cell casing.

Operando XRD analysis during charge and discharge
In Figure 3 is depicted the X-ray diffraction pattern of the battery mounted on the glass cell and heated at 80 ºC prior to cycling, i.e. discharged. The pattern was collected using a 0. After completion of the first scan, showing the initial phase distribution of the primary components, the battery was subjected to a charge-discharge electrochemical cycle. During this process, X-ray measurements along the entire assembly were performed in order to follow the behavior of the battery. Figure 5a presents the XRD patterns collected upon charge and discharge at a position of the battery height close to the interface between the cathode and the SPE. The voltage profiles acquired upon charge and discharge area displayed in Figure 5b.
Each open circle in the charge/discharge plots corresponds to one diffractogram in Figure 5a.
A common feature present in all the depicted XRD patterns is a bump of intensity at low 2θ values. This is ascribed to the SPE, which is in its molten state and therefore presents an amorphous signal. Initially, LiFePO4 is the sole phase present and the characteristic peaks of this phase are only observed. Upon charge, the voltage profile transitions to a plateau, at about 3.45 V vs. Li + /Li, which is accompanied by a gradual disappearance of the reflections corresponding to LFP. At the same time, the reflections belonging to the FePO4 (FP) phase emerge and grow. Although the growth of the FP peaks is clearly visible, it can be observed that a portion of the LFP phase remains after completion of the charge step. A post-mortem optical inspection showed that a small portion of the cathode disk was detached from the battery during the assembly and was trapped in between the side of the stack and the glass wall. This piece was not in contact with the rest of the battery and thus remained in its initial state during the entire experiment.
During the subsequent discharge process, the reduction reaction towards the formation of LFP occurs. In this case, the plateau develops at a potential of 3.35 V vs. Li + /Li, which evidences a significant polarization compared with the charge plateau. Along the discharge plateau, the intensity of FP phase reflections starts to fade, until their complete disappearance.
At the end of the discharge, LFP is again the only phase present. It can be seen from Figure   5b that the time needed to reach the lower cut-off potential is shorter than the one required to charge the battery, which evidences a low Coulombic efficiency. It has to be noted that the goal of this study is to show the capability of the electrochemical cell rather than getting high performances through battery assembly optimization.

Phase distribution at different states of charge
The main advantage of the presented glass-tube cell is that it allows to easily probe the electrochemical reaction kinetics throughout the height of the electrodes while monitoring the advancement of the reaction front. This section illustrates the dynamics of LFP and FP phase distributions along the electrode height upon electrochemical cycling. Considering the low Crate used to cycle the battery and the short times required to probe the entire height of the cathode (less than 1 min), it is assumed that the XRD patterns recorded during the experiment are representative of the state of the active materials.
The evolution of the reaction front profile is determined by quantifying the phase percentages of LFP and FP at different cathode heights during cycling. Peak areas of reflections (020) and (200) of LFP and FP, respectively, were fitted to a Gaussian function. Each full circle in Figure 6a represents the fitted areas of LFP or FP at a specific electrode height and voltage value. It is important to mention that the selected Bragg reflections of the LFP and the FP phases have different maximum areas, since they belong to different crystallographic groups and thus cannot be compared to each other. In addition, given that the portion of the cathode disk trapped in between the side of the stack and the glass wall did not react during the charge and discharge processes, the intensities belonging to this LFP amount were easily identified and subtracted from the XRD patterns. The inset number shown within each plot (#) in Figure   6a is related to the charge and discharge voltage values depicted in Figure 6b. As indicated within the plot #11, for example, the Li layer is located at left side of the cathode, while the current collector at the right side. The LFP phase percentages corresponding to the selected data points are also plotted in Figure 6b. Since total phase volume illuminated by the X-ray beam is related to the scattered intensity by a square factor, the plotted LFP phase percentages were estimated by calculating the square root of the ratio between the peak area at that point and the initial LFP peak area.
In Figure 6a, plot #0 shows the phase distributions corresponding to the initial state of the battery cell where the only phase present is LFP. LFP is symmetrically distributed around a peak centered at an electrode height of 108 µm. Selected data sets corresponding to the charge plateau, from plot #11 to plot #23, show the gradual decrease of LFP phase and the growth of the FP phase. Plot #11 displays the distribution of FP and LFP phases corresponding to an x value in LixFePO4 of 0.58. At this stage, approximately 50 % of the initial LFP has transitioned to the FP phase (see Figure 6b). It can be noticed that the spatial distribution of the newly formed FP phase is shifted towards the interface between the electrode and the SPE, while the active material located at higher electrode heights (i.e. closer to the current collector) remains unreacted. As the reaction proceeds, the FP phase becomes more dominant. In the final stages of the charge step (plot #20), when x in LixFePO4 is close to 0.22, the percentage of FP is of approximately 76 % (see Figure 6b). The charge reaction is almost completed when x in LixFePO4 reaches 0.06 (plot #23) and only 5 % of the original amount of LFP is present. At this instance, an even distribution of the FP phase across the electrode height is evidenced, and the last residue of LFP phase is located close to the interface with the current collector. After completion of the charge process (plot #26) the only phase present is FP. This propagation of the reaction front starting at the cathode/electrolyte interface and developing until the current collector is similar to that reported by Liu et al. 39 on a Li-LFP battery comprising a liquid electrolyte at a 50% state of charge using ex situ synchrotron X-ray microdiffraction. In addition, using in situ energy-dispersive XRD on coin cell, Strobridge et al. mapped the inhomogeneity of the electrochemical reaction in LFP cathodes 40 . The authors shows that on the first cycle, the ionic dynamics are limited by the Li diffusion in the electrolyte and that the LFP oxidation occurs preferentially in areas closer to the separator than to the current collector.
A similar conclusion was also drawn by Orikasa et al. who performed ex situ two dimensional X-ray absorption spectroscopy on LFP based cathode of low porosity. 41 For Li polymer electrolyte batteries the ionic transport in the polymer electrolyte acting as separator and as cathode binder is the rate limiting process 34 . Thus, during charge the Li + ions produced during the LFP oxidation have to diffuse toward the Li metal anode through the polymer electrolyte domain inducing a need of anion species to compensate locally the excess of cationic charge.
This process is thus facilitated at the cathode/SPE interface rather than in the depth of the tortuous cathode leading to a faster overall kinetics at the electrode/electrolyte interface.
Similarly, during discharge, the reaction front will also develop from the cathode/electrolyte interface to the current collector due this time to excess of anionic species because of the Li + ion consumption during LFP reduction.

X-ray diffraction computed tomography (XRD-CT)
Ex-situ room temperature XRD-CT was used to obtain phase distribution maps of the battery assembly components at different heights. The goal is to prove the effectiveness of the glass-tube cell for different X-ray characterization techniques and to demonstrate the high quality of the reconstructed diffraction patterns.
For each voxel of the reconstructed volume, an XRD pattern representative of the battery at that specific spatial position was obtained. Standard back-projections techniques were used for the reconstruction, and the initial results were corrected to take the surroundings into account as described by Finegan et al. 37 Figure 7a displays the phase distribution maps of glass, LFP and Li of a slice corresponding to the interface between the cathode and the SPE, which is semi-crystalline at room temperature. The reconstructed XRD patterns of three voxels, located at the edges of the cathode (position #1 electrolyte/electrode interface and position #3 collector/electrode interface) and on its center (position #2), are presented in Figure 7b. The high quality exhibited by the obtained patterns allows to perform refinements using the Rietveld method. Interestingly, the reconstructed XRD pattern corresponding to position #2 exhibits the highest signals of the cathode and SPE. According to reflections at 2θ = 2.31 ° and 2θ = 3.65 ° the presence of the SPE is not homogeneous through and presents significant variations at the edges. This may be caused by the SPE being trapped in the central part of the battery during the assembly process. Figure 7a also evidences the presence of a small amount of Li at the right side of the electrode. This can be caused by a displacement of the Li foil during the preparation of the cell that resulted in part of the Li film being bent at the side of the cell.

Conclusions
A versatile design of an electrochemical cell suitable for in situ/operando synchrotron X-ray diffraction characterization of Li batteries is reported. The main body of the cell is a glass tube, which enabled to collect low-background high quality X-ray diffraction data, and the rest of the primary components are commercially available items. The cell was successfully used to perform high temperature time and spatial resolved operando synchrotron X-ray diffraction studies and ex-situ X-ray diffraction computed tomography analysis on a Li metal battery comprising a LiFePO4 based cathode and a solid polymer electrolyte, operated at 80 ºC.
Operando synchrotron XRD measurements performed along the electrode principal axis allowed to monitor in real-time the structural evolution and phase distribution of the LiFePO4 and FePO4 phases within the electrode upon electrochemical cycling. The results allowed to visualize the incipient growth of the FP (resp. LFP in discharge) phase upon charging at the interface between the cathode and the SPE. During the subsequent discharge reaction, the phase transition towards the formation of LFP is also initiated at the interface between the cathode and the SPE. In this case, the diffusion of Li + ions through the newly formed LFP was severely limited and the reaction was not fully completed. The dynamic phase distribution plots revealed that the unreacted portion of the active material was located at the vicinity of the interface between the cathode and the current collector.
The reported results show that operando spatial and time-resolved X-ray diffraction studies are essential to fully understand the behavior of electrode materials under working conditions and to get insights in their degradation and failure processes.