Understanding Inhomogeneous Reactions in Li‐Ion Batteries: Operando Synchrotron X‐Ray Diffraction on Two‐Layer Electrodes

To understand inhomogeneous reactions perpendicular to the current collector in an electrode for batteries, a method combining operando synchrotron X‐ray diffraction and two‐layer electrodes with different porosities is developed. The two layers are built using two different active materials (LiNi0.80Co0.15Al0.05O2 and LiMn2O4), therefore, tracing each diffraction pattern reveals which active material is reacting during the electrochemical measurement in transmission mode. The results demonstrate that the active material close to the separator is obviously more active than that one close to the current collector in the case of low porosity electrodes. This inhomogeneity should be due to the rate‐limitation and especially to low average ionic conductivity of the electrolyte in the porous electrode because the current flows first mainly into the electrode regions close to the separator. The inhomogeneity is found to be mitigated by the adjustment of the electrode density and thus porosity. Hence, the novel operando method reveals a clear inhomogeneous reaction perpendicular to the current collector.


Results and Discussion
Figure 1 a shows the cross-section image of the typical two-layer electrode recorded by scanning electron microscopy (SEM). The two-layer structure is clearly visible in Figure 1 b, from a mapping with energy dispersive X-ray spectroscopy (EDX). The result of EDX mapping clearly confi rms the two-layer structure To understand inhomogeneous reactions perpendicular to the current collector in an electrode for batteries, a method combining operando synchrotron X-ray diffraction and two-layer electrodes with different porosities is developed. The two layers are built using two different active materials (LiNi 0.80 Co 0.15 Al 0.05 O 2 and LiMn 2 O 4 ), therefore, tracing each diffraction pattern reveals which active material is reacting during the electrochemical measurement in transmission mode. The results demonstrate that the active material close to the separator is obviously more active than that one close to the current collector in the case of low porosity electrodes. This inhomogeneity should be due to the rate-limitation and especially to low average ionic conductivity of the electrolyte in the porous electrode because the current fl ows fi rst mainly into the electrode regions close to the separator. The inhomogeneity is found to be mitigated by the adjustment of the electrode density and thus porosity. Hence, the novel operando method reveals a clear inhomogeneous reaction perpendicular to the current collector.

Introduction
Over the past decades, a considerable number of studies have been conducted on the durability and safety of lithium-ion batteries. The remaining challenge related to these issues is the fact that there are some signifi cant inhomogeneous reactions in practical Li-ion batteries. [1][2][3] An inhomogeneous reaction leads to a concentration of current fl ow into a part of the electrode which then suffers from overuse and overcharge behavior. The local overuse and overcharge in lithium-ion batteries makes the lifetime shorter and the safety lower, respectively. In an electrode there are two types of inhomogeneities, in of LiNi 0.80 Co 0.15 Al 0.05 O 2 (NCA) and LiMn 2 O 4 (LMO) based electrode. In order to have a fair comparison, four kinds of two-layer electrodes were made using different order of layers or different electrode density (which correlates with the average electrode porosity). The mass of every active material is almost same in all two-layer electrodes. Sample H1 has LMO on the separator (upper) side and NCA on the current collector (lower) side with high packing density (2.9 g cm −3 , lower porosity (≈16%)), and vice versa for the sample H2. Samples L1 and L2 have the same layer confi guration as H1 and H2 with low packing density (2.4 g cm −3 , higher porosity (≈31%)), respectively.
For the in situ X-ray diffraction measurements, "coffee-bag" cells were used. [ 11 ] In Figure 2 , a schematic drawing of the "coffee-bag" cell on the X-ray beamline is shown as well as the confi guration of the four types of two-layer electrodes. Figure 3 shows the XRD patterns of four samples, H1, H2, L1, and L2, inside the "coffee-bag" cells at open circuit voltage before the electrochemical measurements together with the ex situ XRD patterns of LMO and NCA. The two-layer electrodes show identical XRD patterns matching with LMO and NCA indexation. For reference, the electrochemical behavior of each active material was confi rmed in standard two-electrodes coin-like cells. [ 11 ] Figure 4 shows the fi rst charge curves of LMO and NCA at C/4 rate. Note that the NCA showed a characteristic overshoot at the beginning of the fi rst charge curve. Figure 5 shows the fi rst charge and discharge curves of the "coffee-bag" cells with the high-density samples, H1 and H2, operated in voltage window from 2.8 to 5.0 V at C/4 rate during the in situ XRD measurements.
The two-layer electrodes showed higher polarizations because the thickness of the electrodes was double compared with single-layer electrodes as shown in Figure 4 . As shown in Figure 3 , the starting X-ray diffraction patterns are similar; however, the electrochemical behavior was different depending on the order of the two-layer confi guration. Sample H1 showed a fl at region in the beginning of the charge curve like LMO, as shown in Figure 4 b. On the other hand, sample H2 showed a characteristic overshoot at the beginning of the charge curve like NCA, as shown in Figure 4 a. The charge curves thus already indicate a signifi cant difference, preferring the reaction of the phase close to the separator. Such behavior is directly related to inhomogeneous reactions in the electrodes in the direction perpendicular to the current collector. Due to the higher binder contents than typical electrodes in commercial Li-ion batteries, the inhomogeneity of reactions in these two-layer electrodes might be exaggerated. However, these two-layer electrodes were used as model electrodes to develop a novel method to detect the inhomogeneous reactions. Figure 6 shows in situ XRD patterns of samples H1 and H2 collected during the fi rst charge up to 5.0 V. In the diffraction angle range 2 θ = 16.2-16.8° and 29.0-30.7°, one peak from LMO and one peak from NCA can be detected in the initial stage, respectively. The 101 and 113 refl ection peaks of NCA show two-phase like reaction during charge/discharge process, therefore, the peak intensities change during charge process as shown in the Supporting Information. The changes in the peak intensities and/or positions are good indicators to detect which active material is reacting. If the reaction is homogeneous, the change of the in situ diffraction pattern should be identical for both samples but actually such a behavior is NOT observed. Analyzing the in situ XRD patterns change of samples H1 (Figure 6 a) and H2 (Figure 6 b), the peak of LMO in sample H1 showed a faster shift than that in sample H2 (red arrow in Figure 6 a). On the other hand, the in situ XRD pattern change of the NCA peak for sample H2 showed a faster intensity dropping than that for sample H1 (green arrow in Figure 6 b). The peak shifts or the intensity dropping indicate the respective reaction of the active materials, therefore, the active material close to the separator side was obviously more active (i.e., reacted faster) than that close to the current collector side. In other words, the current in the electrodes tended to "rush" into the separator side in the case of the high-density (lower porosity) electrodes. This phenomenon seems to be strange from the point of view of the redox potential of the two active materials. As shown in Figure 4 , the redox potential of NCA is lower than that of LMO, therefore the NCA should be responsible for the charging reaction at the early stage. The behavior of H1 sample seems to be abnormal. To explain this phenomenon, we should take into account the overall kinetics, i.e., the inhomogeneous reaction in the perpendicular direction to the current collector.
This inhomogeneity of high-density electrodes is probably due to the rate-limitation as a consequence of the rather low ionic conductivity of the electrolyte phase in the porous electrode. Thus, the active material at the separator side will "see" a lower total ionic resistance than that at the current collector side. This behavior of the inhomogeneous reactions as shown in Figure 6 is consistent with the electrochemical signature of the two-layer electrodes ( Figure 5 ). In sample H1, the LMO layer shows a faster reaction than the NCA layer. This is the reason why the  charge curve of sample H1 in Figure 5 showed the fl at region in the beginning like LMO and vice versa for the sample H2.
To confi rm the rate-limitation effect of the ionic conductivity, the comparison between low-and high-density electrodes is valuable. Figure 7 shows the fi rst charge and discharge curves of the "coffee-bag" cells with the low-density samples, L1 and L2, during the in situ XRD measurements. Sample L1 has the same layer order as sample H1, however, sample L1 showed no fl at region in the beginning like LMO and sample H1.    sample H1, sample L1 showed that NCA, the active material closer to the current collector side, was more active. This means that the ionic conductivity of the low-density (higher porosity) electrodes was NOT rate-limiting. Additionally, the in situ XRD patterns of samples L1 and L2 showed little differences; therefore we can conclude that the low-density electrodes showed less inhomogeneity than the high density electrodes in this particular case. In other words, the current distribution perpendicular to the current collector can be controlled by the adjustment of the electrode density (i.e., the electrode porosity). The effects of temperature, thickness, and charge/discharge current are also important factor to change the degree of the inhomogeneities as shown in the other references. [ 4,7 ] On the other hand, the inhomogeneous reaction in the perpendicular direction is due to the unbalance between the ionic and electronic conductivities, therefore, the porosity can change the origin, the ionic and electronic conductivities directly. For example, lower porosity will lead to lower ionic conductivity and higher electronic conductivity, such as H1 or H2 electrodes. Thus, the porosity is supposed to be one of the main factors of the inhomogeneous reactions as well as temperature, thickness, and charge/discharge current.

Conclusion
These studies of the inhomogeneous reactions offer a starting point for improving the electrode's stability against overcharge and overuse. To understand the durability and safety of Li-ion batteries, detailed investigations of inhomogeneous reaction during in situ/operando conditions are crucial. The inhomogeneous reactions perpendicular to the current collector will depend on many factors such as (i) confi guration of the electrodes, (ii) porosity, (iii) thickness, (iv) amount of conductive carbon, (v) current rate, and so on. Our novel in situ method revealed clearly the capability to detect inhomogeneous reactions perpendicular to the current collector. We are convinced that this study can pave a new path for in situ/operando investigations of inhomogeneous processes not only in electrodes but in many potential applications.

Experimental Section
"Two-Layer" Electrode : The two-layer electrodes were built of two selfstanding fi lms with different active materials, NCA and LMO. The selfstanding fi lms were prepared by mixing 80 wt% of the respective active material, 5 wt% of conductive carbon (Imerys SuperP), and 15 wt% of PVdF-Hexafl uoropropylene (HFP) copolymer binder (Kynar Flex 2801). The two self-standing fi lms were superimposed and then calendared together at 80 °C onto an Al foil (serving as current collector) into the appropriate packing density (2.9 g cm −3 or 2.4 g cm −3 ). The physical properties of the two-layer electrodes were summarized in Table 1 . The estimated porosities were the average values of the whole electrodes. The porosities of NCA and LMO could not be controlled independently. However, for example, each NCA (or LMO) in high-density electrodes, sample H1 and H2, should have the same porosities, because the calendaring process and the controlled thicknesses were same as shown in Table 2 . The key point of this approach was the comparison with the different orders of layer in the same confi gurations; thicknesses, each loading weights, and porosities. It allowed to make fair comparisons of the effect of the inhomogeneous reactions perpendicular to the current collector.
The molar ratio of NCA to LMO was between 0.93 and 0.97. The assembled electrodes were used as working electrodes for the in situ measurements. The structural changes of NCA and LMO during the charge/discharge process were reported by our group [ 12 ] and others. [ 9,10 ] Thus it was detected that which active material was reacting during the in situ measurement by tracing each XRD pattern.
Electrochemical In Situ Measurement : For the in situ XRD measurements, "coffee-bag" cells were used [ 11 ] as shown in Figure 2 . Li metal was used as a counter electrode, and a 25 µm monolayer polypropylene (PP) separator (Celgard 2400, Celgard, USA) and a glass fi bers separator (GF/G3 grade, Whatman, UK) were soaked in 1 M LiPF 6 in ethylene carbonate (EC)/dimethyl carbonate (DMC) electrolyte. The glass fi bers separator was needed as an electrolyte absorber. The assembled "coffee-bag" cells were loaded into an automatic sample changer. [ 13 ] Then, in situ XRD patterns were collected at the MS-powder beamline (X04SA) at Swiss Light Source (SLS) at the Paul Scherrer Institut (PSI) at room temperature. The XRD patterns were collected in transmission mode using a 17.5 keV X-ray beam. The beam spot size was 0.5 × 0.5 mm. A MYTHEN (Microstrip sYstem for Time-rEsolved experimeNts) detector system was used to collect the XRD patterns in the range of −60° to −5° and 5° to 60° in 2 θ . [ 14 ] The detector system characteristics were summarized in Table 1 . [ 13,14 ]