Visualizing Reaction Kinetics in High Mass-Loading Cathodes through Operando Neutron Imaging
Low mass-loading cathodes (< 5 mg/cm2 or 1.0 mAh/cm2) are commonly employed in research to investigate the material-level capabilities and usually demonstrate excellent rate performance. However, as the cathode mass-loading increases, poor kinetics emerge as a significant issue. Ion transfer, electron transfer, interface resistance, and ion diffusion in the CAM are the most common factors contributing to kinetic limitations in composite cathodes (Fig. 1a). The ion and electron transfer will further affect the reaction prioritization and uniformity of CAM on the electrode level. Despite previous works demonstrating ultra-thick and ultra-high areal capacity cathodes (> 10 mAh/cm2), this capacity was obtained at an extremely low rate (0.025C), even when utilizing a highly ion-conductive SE with a conductivity of 32 mS/cm.9 Faster cycling of high mass-loading cathodes in ASSBs without compromising rate performance is crucial for real-world applications. However, the primary bottleneck for achieving high mass-loading cathodes in ASSBs remains unclear, making efforts to investigate cathode mass-loading essential.
To unravel the reaction kinetic limitation, operando neutron imaging was carried out to visualize the reaction homogeneity within a thick cathode during charge and discharge processes (Supplementary Fig. 1). Figure. 1b illustrates the mechanism and setup of operando neutron imaging for ASSBs. The operando cell, featuring a high mass-loading cathode (33 mg/cm2 and 5.0 mAh/cm2), was positioned vertically in front of the detector, aligning the layer interfaces parallel to the neutron beam. Given the distinct neutron attenuation coefficients of the materials in the ASSB (Supplementary Table 1), the intensity of the transmitted neutron beam varies after passing through the battery. This variance facilitates easy differentiation between the cathode, SE, and anode layers based on the gray level, achieved by normalizing the neutron transmission intensity (Tr) from zero (no transmission) to one (full transmission). Darker areas correspond to lower neutron transmission, indicating that the material has a higher neutron attenuation coefficient. In the neutron image, the three layers- the In-6Li anode, SE, and the thick cathode- could be clearly recognized from top to bottom (Fig. 1b). Subsequently, the battery underwent charging and discharging processes while simultaneously collecting electrochemical impedance spectroscopy (EIS) data (Supplementary Fig. 2) and operando neutron imaging data.
To amplify the neutron transmission change of the cell during charge and discharge, the operando images were further calculated by dividing the image at the pristine state to get the transmission change ratio, Trt/Tr0 (Fig. 1c). The pseudo color was applied to present the value of Trt/Tr0 in the neutron image (Fig. 1d-f). For the area with limited transmission change (Trt/Tr0 ≈ 1), the color remains green. The colors toward red and blue indicating the decrease (Trt/Tr0 < 1) and increase (Trt/Tr0 > 1) of the neutron transmission, respectively. The pseudo-color video presenting the transmission change over the charge and discharge process is in Supplementary Movie 1. Considering the cylindrical shape of the operando cell, which leads to different transmission lengths of the neutron beam over the cross-section and further affects the neutron transmission, we narrowed the sampling region to the center area (1.2 mm) of the cell with the transmission length close to the cell diameter (4 mm) to have the most reliable data (Supplementary Fig. 3).
To better track the transmission changes over time, the average transmission change ratio of the sampling region of the operando battery is plotted in Fig. 2a as a function of time, and the corresponding charge/discharge profile is shown in Fig. 2b. Figure 2c and 2d provides detailed 2D images at specific states of charge (SoC) and depth of discharge (DoD). We first focused on the behaviors at the whole cell level. During the charging process, the colors of the cathode layer gradually changed to blue. This is because the CAM underwent a delithiation process, resulting in a decrease in the neutron attenuation coefficient (Supplementary Table 1). The Li+ flux was directed toward the anode side during charging, but there was almost no transmission change in the SE layer, proving that there was no Li+ concentration change in the SE layer, since Li+ in both CAM and SE are natural Li. A small amount of red color was observed at the interface between the SE and the In-6Li anode, caused by the self-diffusion of 6Li+ from the anode layer to the SE layer by replacing the natural Li+ in SE.28 For the In-6Li anode layer, due to the high neutron attenuation coefficients of In and 6Li, the neutron transmission remained almost zero throughout the process under the current setup, resulting in no observable transmission change.
During the discharge process, the Li+ flux was directed toward the cathode side. As the CAMs underwent lithiation, the color of the cathode layer gradually turned back from blue to green after discharge. For the SE layer, based on the data from the charging process, Li+ flux did not affect the Li+ concentration and neutron attenuation coefficient of the SE layer. Therefore, in Fig. 2a, the enlargement of the red color region in SE during discharge was contributed by 6Li+ diffusing into the SE along with the Li+ flux. Because 6Li+ replaced the natural Li in the SE, the neutron attenuation coefficient of SE significantly increased (Supplementary Table 1). A clear front of 6Li+ proceeding through the SE from the anode side to the cathode side was observed, which can also be seen in Fig. 2d as well as Supplementary Movie 2. The operando neutron imaging successfully visualized the Li+ diffusion in ASSBs from the anode to the cathode side through the SE layer during discharge based on the 6Li isotope.
The cathode layer was further magnified to investigate the delithiation uniformity of the CAM along the vertical direction (Fig. 2e). The thickness of the composite cathode layer with a mass loading of 33 mg/cm2 (5.0 mAh/cm2) is around 180 µm. At the start of the discharge, it was hard to observe the transmission change because of the limited concentration change. After 3 hours, with a capacity of around 0.4 mAh/cm2, the region close to the SE side started to turn blue. As mentioned above, this indicates the delithiation of the CAM. The cathode close to the current collector remained unchanged. As the charging time increased, a gradual movement of the reaction associated with the CAM delithiation was observed, progressing from the SE side toward the current collector side. After around 16 hours, the entire cathode turned blue with a specific charge capacity of 175 mAh/g (4.4 mAh/cm2) with the voltage reaching 3.7 V (4.3 V vs Li/Li+). A similar trend was observed for the discharge process (Supplementary Fig. 4). The CAM close to the SE side first turned back to green, indicating that the lithiation process also started from the SE side.
A relatively thinner cathode (16.7 mg/cm2) was also characterized under the operando neutron imaging (Supplementary Fig. 5). The reaction inhomogeneity is not obvious in the thin cathode (Supplementary Fig. 6). Although both the thick and thin cathode cells were cycled under the same C-rate, the thin cathode cell shows small overpotential on account of the smaller areal current density. Moreover, the color change in the SE layer due to the 6Li+ diffusion in the discharge process was also much lighter in the thin cathode, reflecting the smaller Li+ flux and corresponding with the smaller current density.
Identifying the Key Factors Limiting Ion Transport in Thick Cathodes
The reaction uniformity of the thick electrode is highly related to the reaction kinetics, especially the transport of charge carriers (both for ions and electrons). In the early studies of the composite cathode used in ASSBs, considering the decomposition of the SE when contacting the high surface area carbon, electron conductive additives (mainly carbon) were usually avoided in the composite cathode.29,30 The electron conductive pathway only relies on the contact between CAM particles. Therefore, the restriction of electron transport rises with the increase in SE content because of the contact loss between CAM particles.17 Recently, one-dimensional carbon fibers have been proven to be ideal electron-conductive additives in ASSBs due to their low surface area and good electrical conductivity. According to our test (Supplementary Fig. 7), the decomposition of SE is negligible in this composite cathode when using one-dimensional carbon fibers as the electron-conductive additive. Therefore, the electron-conductive additives were applied in our composite cathode.
Our operando neutron imaging data reveals that the delithiation/lithiation processes of CAM were nonuniform within the thick cathode. Most of the Li+ ions were first reacted in the area near the SE layer and hard to reach the CAM close to the current collector, indicating that the electrochemical reaction kinetics is primarily restricted by Li+ ion transport along the long pathway within the thick electrode, rather than by electron transfer. The ion transfer limitation led to slow kinetics, necessitating a lower current density and a longer time for Li+ ions to transfer to the area near the current collector side and react with the CAM. If a high current or C-rate is applied, the CAM within the thick cathode cannot be sufficiently reacted. Therefore, it is crucial to purposefully regulate the Li-ion transport within the thick cathode to ensure optimal rate performance.
To further diagnose the ion transport limitation within the thick cathode, electrochemical performances were examined with different cathode mass-loadings of 3, 10, and 30 mg/cm2 at 60 ℃. In-Li symmetric cells were also studied to evaluate the In-Li anode behavior and determine whether the anode contributes to the battery performance limitations. The symmetric cells maintained stable performance with limited overpotential (< 100 mV), even at high current densities up to 15.0 mA/cm2 (Supplementary Figs. 8 and 9). This result proved that the anode is not the main performance barrier in our ASSBs system. A low mass-loading (3 mg/cm2) cathode with high SE content (33.0 wt%) and sufficient carbon additives (2.0 wt% of carbon nanofibers) was tested to explore the intrinsic rate performance on the material level of NMC 811. The cell shows a specific discharge capacity of 199 mAh/g at C/10, approaching the theoretical capacity of NMC 811 (Fig. 3a). At C/2, 1C, and 2C, the capacities were 174, 160, and 141 mAh/g, individually, representing the rate capacity of our NMC 811 at the material level under the condition with minimized limitations of the ion and electron transfer from the electrode level.
We further evaluated the rate performance for our normal cathode (75.0 wt% NMC with 1.5 wt % of carbon nanofibers) with mass-loadings of 10 and 30 mg/cm2. When using a mass-loading of 10 mg/cm2 with an areal capacity of 1.5 mAh/cm2, there is no obvious decay of the rate performance, with the discharge capacities of 194, 170, 153, and 134 mAh/g at C/10, C/2, 1C, and 2C, individually (Fig. 3b). However, when the mass loading increased to 30 mg/cm2 with an areal capacity of 4.5 mAh/cm2, the rate performance significantly decreased with the discharge capacities of 152, 140, 109, 80, and 49 mAh/g at C/10, C/2, 1C, and 2C, individually (Fig. 3c), which are only 76%, 63%, 50% and 35% of them achieved in the 3mg/cm2 cell, respectively. The great rate performance of the 10 mg/cm2 cathode was due to the short ion pathway within the thin electrodes, whereas the cell with 30 mg/cm2 cathode showed much worse rate performance, suggesting the insufficient reaction of the cathode materials within the thick electrode because of sluggish ion transport along the long ion pathway.
The reaction homogeneity at the electrode level can also be characterized by the dQ/dV analysis.31 A uniform reaction of the CAM (NMC 811) on the electrode level can present a similar dQ/dV curve of the CAM shown on the material level, which has four clear peaks representing the sequential intercalation reactions of NMC 811.32 As shown in Fig. 3d, the dQ/dV curve for the 3 mg/cm2 cathode at the low rate (C/10) corresponds to the behavior of the CAM at the material level. The four peaks can be observed even at 1C. When the rate increases to 2C, the shape of the dQ/dV curve changes to a rounded rectangle. Since the cathode is thin, this behavior mainly contributes to the rate capacity of CAM on the material level. Using these dQ/dV curves as the baseline behavior of our CAM, we further analyzed the effects of the increase in mass-loading. For the cathodes with 10 and 30 mg/cm2 mass-loadings (Fig. 3e, f), both cells show four pairs of peaks indicating uniform and sequential multiphase transitions under a low rate of C/10 or C/20. With the increase in the rate, the change of the dQ/dV curves of the 10 mg/cm2 mass-loading cathode is similar to our baseline. However, for the cathode with 30 mg/cm2 mass-loading, the trend of the dQ/dV curves varies with the increase of the C-rate. Even at C/10, four pairs of peaks are difficult to observe. At C/5, there is a new peak emerging at 3.4 V (Fig. 3f), which is a new peak but not a shift of peak two because it cannot be simply corrected by the IR drop.31 The shape of the dQ/dV curve for oxidation transforms into a semi-isosceles triangle (yellow line in Fig. 3f). When the rate increases to C/2, the shape of the dQ/dV curve of the thick cathode becomes a right triangle (blue line in Fig. 3f), which is significantly different from that of the thin cathode of 3 and 10 mg/cm2. The sequential intercalation reactions of NMC 811 were completely unobservable. Consequently, the strange dQ/dV curves indicate a huge reaction inhomogeneity in the thick cathode of 30 mg/cm2, especially at high rates, which is consistent with the findings from the operando neutron imaging.
While the electrochemical results and operando neutron imaging provide valuable insights into the reaction inhomogeneity and performance limitations in thick cathodes, it is essential to understand the underlying factors that contribute to these issues. In the conventional discussion, ionic tortuosity (τ) is considered a key parameter of the kinetics of the cathode, which controls Li+ diffusion and transportation in the electrodes.21,33 Tortuosity is defined as the fraction of the shortest pathway through a structure (Δl) and the Euclidean distance between the starting and end points of that pathway (Δx) (Fig. 3g). τ could be calculated using the equation (S1). The effective ion conductivity (σi,eff) in the cathode was usually measured by AC impedance-based techniques or polarization-interrupt method (DC method) with electron blocking electrolyte (Fig. 3g) and further simulated based on the transmission line model.18 In this case, the Faraday reaction of the CAM was ignored. For the thin cathode, these methods are acceptable. However, in the thick cathode, the Li+ flux will be dramatically different from the current collector side to the SE side because of the accumulation of Li+ flux generated from the Faraday reaction of the CAM with the increase of the thickness, as shown in Fig. 3h. The CAM not only acts as a resistor element affecting the ion and electron transfer but also provides or consumes ions and electrons through the Faraday reaction. Only using tortuosity to evaluate the ion transfer is insufficient to explain the degradation of the rate performance in the thick cathode. The effective ion conductivity cannot adequately represent the resistance of ion transfer under the dramatically different Li+ flux across the electrode thickness. A large Li+ flux accumulation at the SE side without a corresponding ion transfer channel provision becomes a serious impediment to Li+ transport. Different from the liquid electrolyte, there is no Li+ concertation gradient in the SE. Therefore, appropriate allocation of SE gradient with Li+ flux over the thick cathode will effectively improve the rate performance.
Gradient Design in Ultra-high Mass-loading Cathode for Enhancing Rate Performance
Inspired by the findings from the operando neutron imaging and the above-mentioned point of Li+ flux generated from the Faraday reaction, we designed a three-layer cathode with gradient ion transport channels for realizing homogenous electrochemical reaction and improved rate performance in the thick cathode (Fig. 4a and 4b). The overall CAM content in the three-layer cathode was kept at 75.0 wt% which is the same as the non-gradient thick cathode. From the SE layer side to the current collector side, layer A is the fastest Li+ transfer layer with 65.0 wt.% of CAM, 33.0 wt.% of SE, and 2.0 wt% of carbon nanofibers. The highest content of SE is designed to bear the largest Li+ flux close to the SE layer. More carbon additives were also used to prevent the electron isolation of CAM due to the high SE content. Layer B is composed of 75.0 wt.% of CAM, 23.5 wt.% of SE, and 1.5 wt.% of carbon nanofibers, the same as the original composite cathode. Layer C is a high energy density layer with the highest CAM content (85.0 wt.%) and lowest SE content (14.0 wt.%). A compare group with a reversed sequence of the three layers was also prepared, as well as the control group with only one composition of 75.0 wt.% of CAM, 23.5 wt.% of SE, and 1.5 wt.% of carbon nanofibers.
The rate capabilities of the three different cathodes were compared to evaluate the superiority of our gradient design (Fig. 4c). The results show that the rate performance of the cell with the three-layer cathode outperforms that of the cell without a gradient design, even when tested under a very low current rate of C/20. The superiority of the gradient cathode is further exaggerated when the current rate increased from C/20 to high rates of C/10, C/5, C/2, 1C, and 2C (9.0 mA/cm2) with 112%, 119%, 120%, 135%, 155%, and 171% of the capacities obtained from the conventional cathode, individually. To further verify that the designed gradient ion transport channels can benefit the rate performance for the thick cathode, a reversed gradient cathode was examined, and it shows a much worse rate performance than that of the non-gradient cathode. Although the initial discharge capacity of the reversed three-layer cathode is very close to the original single-layer cathode at the low rate (C/20), its capacity decays dramatically as the rate increases. There is almost no capacity obtained when the rate is higher than C/2.
We further compared the charge/discharge profiles of these three groups to dig into the fundamentals of the three-layer design. Under the lowest rate we tested (C/20) with a current density of 0.225 mA/cm2 (Fig. 4d), the three-layer cathode exhibited a noticeably smaller overpotential and larger capacity. As the C-rate increased, the advantage became more pronounced (Fig. 4e, f). The dQ/dV analysis can provide more detailed insights into kinetics. As shown in Fig. 4g, the dQ/dV curves at C/20 of the three-layer cathode and the conventional cathode are very similar, with four pairs of peaks corresponding to the sequential intercalation reactions of NMC 811. However, for the reversed three-layer cathode, in addition to the shift of the peaks due to the larger overpotential, one pair of peaks at high voltage (around 3.6 V vs In-Li/Li+ or 4.2 V vs Li/Li+) is missing due to the sluggish kinetics. Specifically, the low SE content of the top layer creates significant resistance for the Li+ transfer into or out of the cathode, resulting in a rapid voltage rise to the cut-off voltage without the last phase transition (H2 → H3) of NMC 811. When the rate increases to C/10 (Fig. 4h), the cell with the three-layer cathode still clearly shows the four pairs of peaks for the sequential intercalation reactions, indicating a homogeneity reaction throughout the cathode thickness. However, for the conventional and reversed cathodes, the four pairs of peaks become blurred, wider, or even disappear caused by reaction inhomogeneity in the electrode level during the charge and discharge process.31 Overall, the three-layer cathode demonstrates better rate performance with more homogenous delithiation and lithiation reactions across the entire cathode. Although the three-layer cathode and reversed groups have the same tortuosity, they show dramatically different performances. The result clearly proved that the appropriate allocation of SE content to match the Li+ flux over the whole thick cathode is an effective way to improve homogenous electrochemical reactions and the rate performance of the thick cathode.
To further explore the performance of our three-layer design cathode, we introduced another CAM, LiCoO2 (LCO). LCO shows much better C-rate performance on the material level (Supplementary Fig. 10). In the dQ/dV curves for the low mass-loading (3 mg/cm2) LCO cell, there is almost no shift of the reaction peaks even increasing to 2C (Supplementary Fig. 11). Therefore, LCO cathode can better exhibit the improvement of the ion transfer on the electrode level. The thick LCO cathode (30 mg/cm2) with the three-layer design shows almost no capacity decay even when increasing the rate to 2.5C (8.44 mA/cm2) with the areal capacity over 3.0 mAh/cm2 (Fig. 5a and 5b). In contrast, dramatic decay of the rate performance was observed for the ASSB with the reversed three-layer LCO cathode (Fig. 5c). Since there is no rate limitation from the material level, and the interface between SE and CAM is identical, the remarkable difference between two cells primarily relies on the different ion transport kinetics on the electrode level, which further highlights the significance of aligning the arrangement of the SE content with Li+ flux to the ion transport in the thick cathode.
The ultrahigh mass-loading cathodes (100 mg/cm2) with the theoretical capacity of 15.0 and 11.25 mAh/cm2 for NMC 811 and LCO as CAM were also studied (Fig. 5d-i). The ultrahigh mass-loading NMC 811 cathode with the three-layer design exhibited around 189 and 170 mAh/g specific capacity for the first charge and discharge, equivalent to 14.25 and 12.75 mAh/cm2 areal capacity under the current density of 0.38 mA/cm2. Even under the current densities of 1.5 and 3.0 mA/cm2, the cell can still obtain the capacities of 9.9 and 7.9 mAh/cm2, which are 1.5 and 2.5 times better than the conventional cell, respectively. The ultrahigh mass loading LCO cathode with the three-layer design exhibited even better rate performance, achieving an areal capacity of 10.4 mAh/cm2 at the current density of 2.25 mA/cm2. All of them show obvious improvement rate performance compared with the traditional one-layer cathode further, proving the importance of aligning the SE arrangement with the Li+ flux on the thick electrode.
With the increase in current density over 10 mA/cm2 for the 30 mg/cm2 cathode or over 5 mA/cm2 for the 100 mg/cm2 cathode, all cells met an unnormal failure (Supplementary Figs. 12 and 13). The capacity suddenly decreases within a few cycles. For example, the cell with 30 mg/cm2 LCO cathode cycled stably at 2.5 C with the current density of 8.44 mA/cm2 (Supplementary Fig. 14). However, when the current increased to 10.13 mA/cm2, the overpotential suddenly increased even over the cutoff voltage. The constant current (CC) charge period disappeared, and the capacity was obtained by the constant voltage (CV) charge process. This phenomenon cannot be simply explained as the ohmic resistance because it is not linearly related to the current density if we compare it with the increase of the overpotential from 2C (6.75 mA/cm2) to 2.5C (8.44 mA/cm2) (Supplementary Fig. 15). The unnormal fail is also not due to the rate performance of the CAM since both materials met the same issue under the similar current density. Therefore, the issue is still related to the ion transfer in the thick cathode, and there should be a critical current density for the thick cathode based on the SE content and distribution. Since the Li+ flux gradually increased with the thickness, but our cells only provide a three-level gradient, there are still mismatches of SE component and Li+ flux on the smaller scale. A smoother and more delicate arrangement of the SE will further increase the critical current density and benefit the rate performance of the thick cathode.
In summary, this work successfully visualized the lithium reaction gradients in an all-solid-state battery with a high mass-loading (33 mg/cm2) NMC811 cathode using operando neutron imaging. The results confirmed the inhomogeneous lithiation of CAM in the thick cathode, with a lithiation gradient from the solid electrolyte (SE) layer side to the current collector side. The electrochemical evaluations of the ASSBs with different cathode mass-loadings of 3, 10, and 30 mg/cm2 further validated the inhomogeneous lithiation of CAM in the thick cathode, especially at high rates. Based on the study, ion transfer was identified as the key limitation causing kinetic issues in the thick cathode. We pioneered the concept of "Li+ flux" and its effect on ion transfer in the thick cathode of all-solid-state batteries. Due to the Faraday reaction of the cathode active materials, which consume or generate Li+ flux, the Li+ flux across the ion conductor (SE), catholyte, in the cathode accumulate in terms of the thickness of the cathode. The mismatch between the Li+ flux and ion transfer channel causes a huge obstacle for the Li+ transport in the thick cathode.
To address the ion transfer limitation arising from the significant variation in Li+ flux from the SE layer to the current collector side in the thick cathode, a tailored arrangement of the catholyte in the composite cathodes was designed and studied, resulting in significantly improved rate performances (171% of the capacity obtained in the conventional cathode at the current density of 9.0 mA/cm2 for 30 mg/cm2 NMC 811 cathode). The effectiveness of this gradient design was further demonstrated in ultrahigh mass-loading cathodes (100 mg/cm2), which achieved areal capacities of 10.4 mAh/cm2 at a current density of 2.25 mA/cm2 with LCO cathode and 9.9 mAh/cm2 at the current density of 1.50 mA/cm2 for NMC cathode. We also observed a critical current density threshold in the thick electrodes, beyond which an abnormal capacity drop occurs attributed to the mismatch between the catholyte and Li+ flux at smaller length scales. This work highlights the importance of understanding and optimizing ion transport in high mass-loading cathodes for the development of high-performance all-solid-state batteries. The insights gained from operando neutron imaging and the demonstrated effectiveness of the gradient design provide valuable inspiration for future advancements in high mass-loading all-solid-state battery fast charge technology.