1. Introduction

Lithium-ion batteries (LIBs) are essential power sources in a wide range of products in our modern society. Portable devices ranging from mobile phones to laptops, from electric vehicles (EV) to large-scale energy storage systems (ESS) are powered by LIBs [¹-³]. Considerable research has been devoted to efforts to enhance their energy density, power density, life span, and total capacity. Among them, various attempts to improve the performance of the separator, which is an essential component of a robust LIB, have been made by numerous researchers. These separators provide chemical stability and mechanical strength, but they are typically too soft and will melt at high temperatures [⁴,⁵]. The difference in the polarities of the separator and electrolyte surface also has a negative impact on LIBs [⁶,⁷].

Recently, several efforts have attempted to address the weaknesses of the standard separator. Micro-nano particles prepared as inert ceramic oxides (Al₂O₃, SiO₂, and TiO₂) have been coated on the separator to increase ionic conductivity, and reduce the safety problems caused by internal electrical short circuits [⁸-¹¹]. The ceramic-coated separator provides micropore channels for lithium-ion migration, acts as a wall between the anode and the cathode, and prevents electrical shorts from the anode [⁴,⁵]. For these reasons, ceramic-coated separators are currently used in all EV batteries. Polyimide (PI) has also been used as a separator because of its thermal stability, high tensile strength, and excellent electrolyte wettability [¹²-¹⁴]. However, it is rarely used in applications, because of the high cost and complex manufacturing process of PI membranes.

In this study, we examined the differences of commercial polypropylene (PP), ceramic (Boehmite) coated PP (C-PP), and polyimide (PI) separators from an electrochemical perspective. For the battery model, the half-cell using lithium nickel manganese cobalt oxides (NMC), which is the most used in the current industry, and the full-cell using graphite as the opposite electrode were adopted [¹⁵]. We conducted rate capability tests at rates from 0.1 to 5 C, focusing on the differences in each C rate section. Also, a high C rate cycle stability test was conducted to determine which separator membrane could maintain good performance, even at fast charge and discharge rates, which is essential to meet the demand for fast charging in mobile devices and electric vehicles.

In addition, electrochemical impedance spectroscopy (EIS) tests were carried out for each separator. EIS is a powerful tool for determining charge-transfer resistance and solid electrolyte interphase (SEI) layer resistance which has high resistance similar with nonconductor [¹⁶]. After the cycle tests, we also disassembled the cell and checked both sides of each separator, and the surface of the NMC electrode. Finally, we tested the three separators in an NMC//graphite full-cell system to study each separator’s performance from a commercial perspective.

2. Experimental Procedure

3. Results and Discussion

First, we attempted to determine the surface structure of each separator by SEM analysis. Fig 1 shows the surface SEM image of the PP separator, which has a fiber structure with very small porosity. Fig 1(b) shows the surface of the C-PP separator exhibiting the porous ceramic coating layer composed of 1~10 um boehmite particles. This layer prevents the shrinkage problem that occurs at high temperatures of 100 to 200 °C [¹⁷]. The PI separator has a porous structure like a sponge, with 1~10 um sized pores, as shown in Fig 1(c). The high porosity of the PI separator allows the electrolyte to transfer easily, and this is the main way its reduces internal resistance (R_S).

Electrochemical tests were carried out to examine the difference in battery performance with the three separators. As shown in Fig 2, the discharge capacities of the three separators were 192 mAh g^-1, 190 mAh g^-1, 181 mAh g^-1. There was no significant difference between the PP cell and the C-PP cell in the first few cycles. As the cycles increased, the capacity retention characteristics and stability in the C-PP cell were excellent. Even at a high current density at the 5 C rate, the capacity of the C-PP cell was the highest, 112 mAh g^-1. Notably, the PP cell had a sharp drop in capacity when compared to the 2 C rate, which was the previous current density.

For the next 10 cycles at the 0.1 C rate, the C-PP cell showed the best cycle stability performance. The capacity of the C-PP cell decreased from 188 mAh g^-1 to 183 mAh g^-1, only a 2.3% drop. This fading was only half that of the PP cell (200 mAh g^-1 to 190 mAh g^-1, almost a 5% decrease) and the PI cell (180 mAh g^-1 to 174 mAh g^-1, almost 3.6%). These results suggest that the ceramic layer in the C-PP separator influences the stabilization of the lithium ions and intermediate ions [¹⁸]. By adding the ceramic coated layer to the separator, interfacial impedance was reduced and lithiumion migration was promoted across the coating layer [¹⁷]. Also, ceramic layers can prevent electrolyte leakage during repeated cycles [¹⁹]. The PI separator has many micro-pores, and not only lithium-ion but other intermediates can also migrate through these pores. This phenomenon is the main reason for the capacity fading observed with PI cells.

Based on these results, we established the cycle stability performance of each of the three cells at the 5 C rate, as shown in Fig 3. Since the difference in capacity and stability was large at the 5 C rate, cycle measurements were performed at the corresponding current density. As shown in Fig 3(a), the irreversible capacity and initial capacity in the PP cell and the C-PP cell were similar at 174 mAh g^-1 and 181 mAh g^-1, respectively, and only the PI cell showed a relatively low capacity of 138 mAh g^-1. However, after 200 cycles, the C-PP cell maintained a capacity of 115 mAh g^-1. The PP cell continued to decrease to a capacity of 80 mAh g^-1 and the PI cell declined to 64 mAh g^-1.

Fig 3(b) shows the charge-discharge voltages of the three cells vs. Li/Li⁺. The capacitive slope of the C-PP cell was higher than the other two cells between 3.4 V and 4.0 V. And the PI cell hardly showed any capacitive slope in the section of the 350^th charge-discharge graph.

To examine this result more precisely, we analyzed the differential capacity vs. cell voltage graph for each cell after the 350^th cycle, as shown in Fig 3(c). The y value of this graph represents the plateau of each graph, and the higher the value, the longer the plateau region. In the C-PP cell, a redox peak exists near 4.0 V and 4.15 V. Each peak represents the phase transition of NMC from monoclinic (M) to hexagonal (H2) and then from hexagonal (H2) to hexagonal (H3). In the oxidation peak, a large sharp peak was observed near 3.6 V. This peak can be explained as the phase transition of the NMC from a hexagonal to a monoclinic (H1 → M) lattice [²⁰]. The main reason for capacity fading near the discharge region is rapid volume contraction during the structural transformation from H2 to H3 phase [²¹]. Using the C-PP separator, the peaks were less polarized, less shifted, and maintained the capacity reaction.

To better understand the results of the cycle test, we used EIS analysis. As shown in Fig 3(d), the resistance of the high-frequency region is related to the SEI layer of the electrode (R_SEI) [²²]. In low-frequency region, the resistance represents the charge-transfer resistance of the two electrode-electrolyte interfaces (R_CT) [²²]. By using the Z-fit program with the EIS model of the NMC half-cell, we calculated the resistances of the solution, SEI layer, and charge-transfer for each cell, as listed in Table 1.

In Fig 3(d), the resistance of the C-PP cell’s SEI layer was determined to be the lowest of the three samples. The result indicates the intermediates of several phase of NMC were less than other separators [²³]. The ceramic inert oxides coated on the surface of the polymer separator led to an increase in ionic conductivity [⁵], and that’s why the C-PP separator had the lowest impedance value. Overall, despite the extremely fast 5C rate condition, our experiments demonstrated remarkable results in capacity stability, compared to previous papers using ceramic-coated separators [⁸,²⁴-²⁶].

4. Conclusion

In this study, we report the differences of PP, C-PP and PI separator surfaces after cycle testing. The electrochemical properties of these separators were also observed in half-cell and full-cell tests. In the rate capability test, the C-PP separator showed stable and high capacity characteristics at a high 5 C rate. Because of the ceramic layer on the PP membrane, lithium-ion migration was stabilized and the phase transition of NMC was well controlled. Also, a less resistive solid electrolyte interphase was the critical factor in the C-PP separator performance.

To examine the surface of each separator, we disassembled the cycled cells and observed the surfaces of both sides of the separator and electrode surface. Normal PP and PI separators were blocked by electrolyte intermediates, known as the SSEI layer, and this phenomenon made it difficult for lithium-ions to effectively migrate by reducing the pore size of each separator. However, with the C-PP, the coated ceramic layer worked as a barrier, so the internal fiber structure was maintained, and this played a significant role in ionic conductivity. These ceramic layers can effectively limit SSEI layer formation on the PP separator from, reducing the effect of the stacked intermediates on the separator structure and pores. In addition, based on the EDS results of the NMC electrode, the C-PP separator helped prevent manganese and nickel element elution to the anode through the electrolyte and separator. Finally, the results of the NMC//Graphite full-cell confirmed that cycle stability was excellent when the C-PP separator was used. Our work can provide guidelines for the life management of EV batteries and separator development issues.

As shown in Fig 3(d), the resistance of the C-PP cell’s SEI layer was the lowest among the three samples. The result suggests the intermediates of several phase of NMC were less than the other separators [²³]. Coating the surface of polymer separators with the inert ceramic oxides leads to an increase in ionic conductivity [⁵], and that’s why the C-PP separator’s had the lowest impedance value. Overall, despite the extremely fast 5C rate condition, our experiments achieved remarkable capacity stability compared to previous papers using ceramic-coated separators [⁸,²⁴-²⁶].