Enhanced electrochemical performance of li-ion battery via ultrasonic-assisted inorganic-rich and thin SEI layer

Highlights

• •Ultrasound enhances the electrochemical performances of lithium-ion batteries.
• •Ultrasound changes the chemical composition of solid electrolyte interface (SEI) to be inorganic-rich.
• •Ultrasound largely detaches organic species from SEI and thus flattens the surface morphology.
• •The charge transfer is improved via Ultrasonic-assisted inorganic-rich and thin SEI layer.

Abstract

Lithium-ion batteries (LIBs) stand as a compelling solution to energy source transition in various applications such as the vehicle industry due to their energy and power density. However, the impact of mechanical factors on them remains understudied. Of particular interest is the effect of vibration, an inherent characteristic of vehicles, on battery performance. Ultrasound has been reported to improve mass transfer and surface cleaning, yet its effects on LIBs are still not thoroughly investigated. This study investigates the influence of ultrasound on the solid electrolyte interphase (SEI) layer, resulting in a thin, inorganic-rich layer. The induced SEI layer alteration improves charge transfer, showing enhanced kinetics. We also reveal that ultrasound application enhances cycling stability, maintains discharge capacity at high charging rates, and facilitates inorganic-rich SEI layer creation. This novel combination of ultrasound and LIBs presents a promising pathway for achieving high-performance batteries.

1. Introduction

Amid the escalating climate crisis, largely driven by industrialization and fossil fuel consumption, the search for innovative energy alternatives has become more critical than ever. The automotive industry, as a prominent contributor to the global environmental challenge, has been actively seeking environmentally friendly alternatives to replace internal combustion engines (ICEs) [1]. Lithium-ion batteries (LIBs), with their notable energy and power density, present compelling solutions to this challenge [2], [3]. In light of the spatial limitations associated with electric vehicles (EVs), efforts to optimize the electrochemical performance of LIBs have become an active area of research [4], [5]. This includes exploring avenues such as material modifications of anodes [6], [7], cathodes [8], [9], electrolytes [10], [11], separators [12], and a comprehensive understanding of their internal mechanisms [13]. That is, the prevailing focus of current research has been on enhancing electrochemical performance through material engineering.

The implementation of batteries across various applications (e.g., electric vehicles; EVs) exposes them to a wide spectrum of environmental conditions. Surprisingly, the research concerning the influence of these conditions on battery performance remains somewhat narrow in scope. Recent studies, for example, have illustrated how temperature fluctuations can significantly impact battery performance. This is because ion migration in electrodes and electrolytes, typically governed by the Arrhenius equation, is temperature-dependent [14], [15]. At elevated temperatures, batteries demonstrate lower internal resistance, resulting in improved electrochemical performances. Contrarily, lower temperatures can lead to a discharge capacity decrease of up to 14.1 %. External mechanical parameters, such as mechanical damage, penetration, and sudden large loads, have also been identified as performance influencers, causing structural failures and short circuits in batteries [16], [17]. Despite some attempts to simulate failure scenarios [18], the reliability of these tests necessitates a more in-depth examination of the underlying mechanisms triggered by mechanical input.

Considering vehicles inherently generate vibrations, the understanding of how mechanical parameters, particularly vibration, influence their electrochemical performances is vital [19]. Current research addressing the correlation between batteries and vibration is sparse, despite the well-documented impact of vehicle vibration frequencies, especially at lower ranges, on battery performance [20], [21], [22], [23], [24]. Continuous vibration exposure has been linked to poor SEI formation, increased internal resistance, and reduced capacity [24], [25], [26]. Additionally, the chemical composition of the SEI has been shown to shift from Li-C (lithium carbide) to O-C = O (ester compounds), known for their high ionic resistance, under vibrational stress [24].

In recent years, the effect of ultrasound has been widely investigated in various fields [27]. For instance, micro-jets from ultrasound enabled cost- and energy-efficient emulsification of foods and pharmaceuticals [28], [29]. These micro-jets also allow US-assisted extracting of oils, proteins, and lipids by facilitating solvent penetration into the matrix [30]. In sono-electrochemistry, ultrasound has been proven to boost mass transfer via acoustic streaming and prevent electrode surface fouling through cavitation [31], [32], [33], [34], and thus it might largely alter the composition of the surface layer on electrodes. Given these thoughts, it is anticipated that ultrasound might be effectively utilized within LIBs. Meanwhile, the effect of ultrasound on LIBs has not been covered yet, despite the presence of a high-frequency range in EVs aligning with those used in ultrasonic applications [35].

Herein, we are the first to report the transformative impact of ultrasound on the SEI layer, resulting in a thin and inorganic-rich layer. The ultrasound-induced inorganic-rich SEI layer not only changes the chemical composition but also reduces the SEI thickness, enhancing ion diffusion and charge transfer (Fig. 1). Our findings reveal that, compared to cells without ultrasound, those exposed to ultrasound show a narrower redox potential difference. Furthermore, ultrasound application appears to bolster the cycling stability of cells with good coulombic efficiency and discharge capacity over 200 cycles than those without ultrasound. The effect of ultrasound aids in maintaining discharge capacity at high charging rates. In addition to this, a comprehensive characterization of electrodes is conducted. An observed uniformity in electrode surface and the shift of SEI layer composition towards inorganic-rich induced by selective dissection from ultrasound due to the low mechanical stability of organic components, resulting in a thin SEI layer. The application of ultrasound, therefore, presents a possible approach for developing high-energy batteries, maintaining capacity over extended cycles, and facilitating the creation of an inorganic-rich SEI layer without any pretreatments on materials. This unexpected synergy between ultrasound and LIBs may open a novel avenue for achieving high-performance batteries.

Fig. 1.

A schematic illustration of a thin-inorganic SEI layer on the graphite surface with ultrasound.

2. Materials and methods

2.1. Materials and electrochemical measurements

Commercially available 40 mAh coin cells (CR2032, EEMB) (Fig. S1) consisting of graphite and lithium cobalt oxide (LCO) electrodes were employed for electrochemical tests. Cyclic Voltammetry (CV) scans were investigated at various scan rates of 0.1–1.0 mV/s in the voltage window of 3.1–4.2 V (ZIVE SP1). The rate capability test was studied at C-rates of 0.25, 0.5, 1, and 2C in the voltage range of 3.2–4.15 V. To evaluate cycling performance, galvanostatic charge and discharge (GCD) profiles were investigated at 1C in the voltage window of 3.2 – 4.15 V. Ragone plot was calculated from the GCD profiles of the rate capability test. The charge transfer resistance was investigated by potentiostatic electrochemical impedance spectroscopy (PEIS) in the frequency range of 15 mHz – 50 kHz with an amplitude of 10 mV. Galvanostatic Intermittent Titration Technique (GITT) was employed to investigate the effect of ultrasound on lithium-ion diffusion coefficients. The cell was first charged at 0.1C (40 mA) for 20 min followed by a 2-hour relaxation period to reach electrochemical equilibrium.

2.2. Material characterization

For the physiochemical analysis, the commercial cells were disassembled in an Ar-gas filled glove box. The obtained graphite and LCO electrodes were rinsed by fresh dimethyl carbonate (DMC) several times and dried overnight in the glove box to remove residual electrolytes before analysis. The microscopic morphology of the electrodes was characterized by a field emission scanning electron microscopy (FE-SEM) with an energy dispersive X-ray spectroscopy (EDS) (GEMINI 500, ZEISS). The cross-sectional view of graphite was investigated by focused ion beam SEM (FIB-SEM, Scios, Thermoscientific). Cu K-alpha X-ray was applied to analyze the crystallinity of electrodes after covering the electrode surface with Kapton tape (XRD, Xpert 3, Bruker). Raman spectroscopy profiles (Ramantouch, Nanophoton) were investigated after the charge and discharge cycles. Surface chemistry of the electrodes was analyzed using X-ray photoelectron spectroscopy (XPS, axis Supra, Kratos).

2.3. Ultrasound imposition on cells

A commercialized bath sonicator (Ultrasonic bath, Powersonic 510) was used to impose ultrasound. The sonicator operates at the frequency of 40 kHz and glass with 80 mL mineral oil was placed in the middle of a sonicator bath. During the electrochemical test, cells were covered with mineral oil and placed in the middle of an ultrasonic bath (Fig. S2). The temperature of oil in the ultrasonic bath was controlled at 34 ± 1 ℃. The cells with and without ultrasound imposition were denoted by Cell-US and Cell-Non, respectively.

The acoustic power (AP) (Eq. (1)), the acoustic intensity (AI) (Eq. (2)), the power density (PD) (Eq. (3)), the specific energy (SE) (Eq. (4)), and the energy density (ED) (Eq. (5)) were calculated by the calorimetric method [36], [37]. The specific heat capacity (C_p) of mineral oil is 1.46 J g⁻¹ ◦C⁻¹.

$$AP\left(W\right)=m{C}_p(\frac{\mathit{dT}}{\mathit{dt}})$$ (1)

where C_p is the specific heat capacity (J/g ◦C⁻¹) at constant pressure, m is the mass (g) of the sample, and dT/dt is the heating rate in function of process time (◦C/s).

$$AI\left(\frac{W}{c{m}^2}\right)=\frac{4\times AP}{\pi D^2}$$ (2)

where D is the diameter of the base of the bottle into which the acoustic energy was transferred from the sonicated water in the bath.

$$PD\left(\frac{W}{\mathit{mL}}\right)=\frac{\mathit{AP}}{V}$$ (3)

$$SE\left(\frac{J}{g}\right)=\frac{\mathit{AP}\times t}{m}$$ (4)

$$ED\left(\frac{J}{{\mathit{cm}}^3}\right)=\frac{\mathit{AP}\times t}{V}$$ (5)

where V is the sample volume (mL or cm³), t is the processing time (s), m is the sample mass (g). The calculated US parameters are shown in Table 1.

Table 1.

Calculated US parameters of imposed process using Eqs. (1), (2), (3), (4), (5).

AP (W)	AI (W cm−2)	PD (W mL−1)	SE (J/g)	ED (J cm−3)
0.2892	0.0147	0.0036	26.8056	26.0282

2.4. Mechanical simulations

The stress distribution of the SEI layer on the graphite electrode with ultrasound was simulated using Ansys software. To determine the stress response to the harmonic load on the surface layer of electrodes, a linear mechanical model was employed. Based on the FIB-SEM image, the surface layer on the electrode model was built. The material properties of the surface layer were assumed to be linear and isotropic. The Young’s modulus, Poisson’s ratio, and density of components were taken from previous works in Table 2. The frequency in the simulation was 40 kHz.

Table 2.

Mechanical properties of graphite and components in the SEI layer. Values were extracted from ref [38], [39], [40], [41].

Material	Young’s Modulus (Gpa)	Poisson’s Ratio	Density (g/cm3)
Graphite	27.6	0.2	2.23
Li2EDC	22.0	0.3	1.86
Li2CO3	36.2		2.11
LiF	58.1		2.64

2.5. Density functional theory calculations

DFT calculations were performed using the Vienna Ab Initio Simulation Package (VASP) with Perdew, Burke, and Ernzerhof (PBE) exchange–correlation functional. The generalized gradient approximation (GGA) and the projector augmented-wave (PAW) method were used with an energy cutoff of 400 eV. In this study, model systems of LiF (0 0 1) and Li₂CO₃ (0 0 1) facets were chosen as representative inorganic SEI components. Additionally, organic SEI components, namely lithium ethyl carbonate (LiOCO₂C₂H₅, LiEC), lithium methyl carbonate (LiOCO₂CH₃, LiMC), and dilithium ethylene dicarbonate ([CH₂OCO₂Li]₂, Li₂EDC), were also considered. The geometries of the inorganic and organic model systems were optimized using 3 × 3 × 1 and 1 × 1 × 1 Gamma k-point grids, respectively. Electronic self-consistent iteration and ionic relaxation were carried out with convergence criteria of 1 × 10^-5 and 1 × 10^-2 eV, respectively.

3. Results and discussion

3.1. Electrochemical performance improvement in galvanostatic charge/discharge

First, the ultrasound effects on rate capability were investigated by GCD test at a varied C-rate (Fig. 2a). The discharge capacities of Cell-US were higher than those of Cell-Non at all the investigated C-rates. In detail, the discharge capacities of Cell-US were 31.1, 27.3, 18.2, and 2.5 mAh at the C-rates of 0.25, 0.5, 1, and 2C, respectively. On the other hand, Cell-Non showed lower discharge capacities of about 27.8, 21.9, 11.7, and 0.95 mAh at the C-rates of 0.25, 0.5, 1, and 2C, respectively. It is worth noting that, at 1C, Cell-US displayed almost 60 % greater capacity than that of Cell-Non. Also, the discharge capacities of Cell-US at 0.5C were comparable to those of Cell-Non at 0.25C. This result argues that ultrasound may help to enable faster charge/discharge. During the discharge process, Cell-US revealed a gentler slope than Cell-Non in the voltage profiles (Fig. 2b). Also, the electrochemical reaction of Cell-US began at a lower potential than that of Cell-Non during the charging process (Fig. 2b). Similar results were also observed in other C-rates (Fig. S3). These may indicate that ultrasound is benign to lower both charge and discharge overpotentials, accordingly leading to higher capacity [42]. Due to the discharge of Cell-US at higher voltage and its increased capacity, Cell-US showed much enhanced energy at the same power compared to Cell-Non (Fig. S4).

Fig. 2.

Electrochemical performance of cells with and without ultrasound. (a) Rate capability of Cell-US and Cell-Non at 0.25-2C (10 mA-80 mA). (b) Galvanostatic profiles of Cell-US and Cell-Non at 0.25C and 1C. (c) The discharge capacities and coulombic efficiencies (CE) of Cell-US and Cell-Non at 1C (40 mA). (d) CV curves of Cell-US and Cell-Non at scan rates of 0.1, 0.2 and 0.3 mV/s for 200 cycles. (e) “b-value” analysis for Cell-US and Cell-Non for the anodic peaks.

The galvanostatic charge and discharge profiles were investigated at a rate of 1C for 200 cycles in the voltage window of 3.2–4.15 V (Fig. 2c). Cell-US exhibited improved cycling stability compared to Cell-Non over the cycles. The capacity retentions of Cell-US from the 2nd to 200th cycle were 99.9 %, while Cell-Non maintained 58 % of capacity at the 200th cycle. Moreover, at the 10th, 50th, 100th, and 200th cycles, the voltage profiles of Cell-US minutely changed, while those of Cell-Non became steeper over the cycles, indicating cell degradation (Fig. S5). Over the cycles, Cell-US showed lower charge potential compared to Cell-Non. This may indicate that the required overpotential to drive reactions gradually increased in Cell-Non, while ultrasound suppressed the increase of overpotential in Cell-US. Also, Cell-US showed a slightly higher Coulombic efficiency (CE) of 99.59 % compared to that of Cell-Non (97.23 %).

3.2. Redox behavior analysis in cyclic voltammetry

To further investigate the enhanced capacity and energy, the redox peaks of Cell-US and Cell-Non were measured in the voltage range of 3.1–4.2 V (Fig. 2d and S6). At a scan rate of 0.1 mV/s, the CV profiles of Cell-Non displayed redox peaks at around 3.6 and 4.04 V. At the same scan rate, Cell-US presented redox peaks at around 3.66 and 3.95 V for discharging and charging, respectively. The voltage difference between the anodic peak and cathodic peak (denoted as ΔV_redox) for Cell-US and Cell-Non were 0.29 V and 0.46 V, respectively. That is, Cell-US stores charges at a lower voltage and discharges at a higher voltage than those of Cell-Non. Those peak-to peak separation (ΔV_redox) reflects electrochemical reversibility in a view of charge transfer [43]. In other words, when electron transfer faces a significant obstacle (a state referred to as electrochemical irreversibility), electron transfer reactions occur at a slow rate. This requires applying more negative (or positive) potentials in order to see reduction (or oxidation) reactions, which results in a greater ΔV_redox [43]. Even at the increased scan rate of 0.3 mV/s, ΔV_redox of Cell-US exhibited showed smaller value of 0.61 V than 0.68 V for Cell-Non. Meanwhile, at all scan rates, Cell-US showed greater current than Cell-Non (Fig. 2d and S6). The greater current of Cell-US could be attributed to the different size of the diffusion layer (regarded as SEI) [43], [44]. When ultrasound is imposed, the acoustic streaming and cavitation effect perhaps decrease the length of the diffusion layer [45]. As a result, it might be that augmented reactions occurred with the aid of shortened diffusion layer, resulting in higher capacitance (57.395F and 45.193F for Cell-US and Cell-Non at 0.3 mV/s, respectively).

To underpin the effect of ultrasound on the electrochemical behavior of cells, CV profiles were further analyzed by the power law of i = av^b where i is the current, v is the scan rate, and b is a fitting parameter. Generally, a b-value close to 0.5 refers to a semi-infinite diffusion charge storage, while 1.0 implies a capacitive-dominant process. For the anodic peaks, b-values were 0.71 and 0.54 for Cell-US and Cell-Non, respectively (Fig. 2e). For the cathodic peaks, however, b-values were estimated to be 0.51 and 0.45 for Cell-US and Cell-Non, respectively (Fig. S7). Although the Cell-US showed slightly higher b-values in the analysis, it seems that they depend on the intercalation process with the consideration of obvious plateau in the charge–discharge curves (Fig. 2b). Indeed, even for intercalative materials, b-values can be varied, not exactly equal to 0.5. For instance, intercalation pseudocapacitive material VOPO₄ exhibited the different b-value of 0.52 and 0.90 at high scan rates and slow scan rates, respectively [46]. This variation of b-values was also seen in other intercalation pseudocapacitive materials [47], [48]. Therefore, it would be reasonable to interpret that the sluggish charge transfer kinetics have boosted rather than the capacitive charge storage mechanism for enhanced capacity.

3.3. Impedance analysis

To further investigate the charge transfer change, electrochemical impedance spectroscopy (EIS) was carried out with and without ultrasound (Fig. 3a and 3b). On the Nyquist plot, there observed two semicircles corresponding to SEI resistance (R_SEI) and charge transfer resistance (R_ct), respectively. After 10 cycles of charge and discharge, the R_SEI of Cell-US was only 230 mΩ, while Cell-Non showed a higher value of 979 mΩ. This lower R_SEI value in Cell-US compared to Cell-Non was consistently observed throughout 200 cycles (Fig. S8). That is, the growth of the SEI layer may be suppressed with the implementation of ultrasound, resulting in low R_SEI. R_ct is regarded as the difficulty of electron transfer between electrodes and electrolytes, one of the best representing how well electrochemical kinetics happen in the system [49]. Over the cycles, the R_ct of Cell-Non represented a higher value than 400 mΩ during 200 cycles while that of Cell-US was mostly averaged at 274 mΩ (Fig. 3c). As cycles increase, electrolyte resistance of Cell-US was higher than Cell-Non, which might be due to the fragments of the SEI layer. In the previous report, longer diffusion length by a thick surface layer on electrodes requires a longer time for Li⁺ to reach the VOPO₄ surface from the bulk electrolyte and may require greater overpotential, which limits charge storage kinetics [46]. In the line of the thoughts, ultrasound may shorten the diffusion length of the Li⁺ and enhance electrochemical kinetics of cells by inducing a thin and compact SEI layer.

Fig. 3.

Electrochemical impedance spectroscopy and GITT analysis with/without ultrasound. (a) Schematic of the general circuit and Nyquist plot of batteries. (b) Nyquist plot for cells with and without ultrasound after 10 cycles. (c) Charge transfer resistance after 1st, 2nd, 3rd, 5th, 10th, 20th, 30th, 50th, 80th, 100th, 150th, 200th cycles. (d) Internal resistance and (e) diffusivity of Li⁺ of Cell-US and Cell-Non during charging from GITT profiles.

3.4. Charge transfer analysis

The internal resistance and diffusion coefficient were also investigated using the galvanostatic intermittent titration technique (GITT) (Fig. S9) [50], [51]. Internal resistance, which reflects the overall kinetics of the cells at each state of charge, was obtained by dividing the overpotential into the applied current using Eq. (6) and GITT results (Fig. S9).

$$\mathrm{I}\mathrm{n}\mathrm{t}\mathrm{e}\mathrm{r}\mathrm{n}\mathrm{a}\mathrm{l}\mathrm{r}\mathrm{e}\mathrm{s}\mathrm{i}\mathrm{s}\mathrm{t}\mathrm{a}\mathrm{n}\mathrm{c}\mathrm{e}\left[\mathrm{\Omega }\right]=\frac{\mathrm{o}\mathrm{v}\mathrm{e}\mathrm{r}\mathrm{p}\mathrm{o}\mathrm{t}\mathrm{e}\mathrm{n}\mathrm{t}\mathrm{i}\mathrm{a}\mathrm{l}}{|\mathrm{a}\mathrm{p}\mathrm{p}\mathrm{l}\mathrm{i}\mathrm{e}\mathrm{d}\mathrm{c}\mathrm{u}\mathrm{r}\mathrm{r}\mathrm{e}\mathrm{n}\mathrm{t}|}$$ (6)

The internal resistance of Cell-US presented much lower than Cell-Non in both charging and discharging, consistent with the tendency of charge transfer resistance (Fig. 3d and S10a). The diffusivity of Li⁺ was calculated by using Eq. (7) [51].

$$D=\left(\frac{4{r_p}^2}{9\pi \mathrm{\Delta }{t}_p}\right){\left(\frac{\mathrm{\Delta }{E}_s}{\mathrm{\Delta }{E}_t}\right)}^2$$ (7)

The diffusivity of Cell-US and Cell-Non presented similar tendencies in both charging and discharging (Fig. 3e and Fig. S10b). The diffusivity of Cell-US was also slightly greater than that of Cell-Non.

3.5. Morphology and surface chemistry investigation of cycled anodes

To determine the effect of ultrasound on the surfaces of the anode and cathode, Cell-US and Cell-Non were disassembled after the cycling tests. Surprisingly, a significant difference in the surface was observed between the anode of Cell-US (denoted as US-anode) and Cell-Non (denoted as Non-anode). The graphite anode of Cell-US exhibited a smooth surface without bumps, while the anode of Cell-Non was covered with small granules (Fig. 4a, 4b, and S11). The presence of those granules may indicate an inhomogeneous SEI layer deposited during cycling tests [52]. In contrast, the imposition of ultrasound may continuously break up the non-uniform SEI layer, resulting in a flat and smooth surface. To further investigate the chemical composition of the electrodes’ surface, elemental mapping was conducted for the anode of Cell-US and Cell-Non (Fig. 4c and 4d). EDS of the US-anode presented higher intensity of F/O ratio in US-anode compared to that in Non-anode (2.08 and 0.92, respectively) (Fig. S12). The results suggest that ultrasound induces surface layers with different elemental compositions. To further investigate the different elemental composition of the anodes, US-anode and Non-anode were investigated using XPS (Fig. 4e-g). High resolution C 1 s spectrums of US-anode and Non-anode were fitted with five curves: sp²-C (284.3 ± 0.1 eV), PVDF/alkyl(285.0 ± 0.1 eV), C-O (286.5 ± 0.2 eV), C = O (288.1 ± 0.1 eV), and CO₃^– (290.0 ± 0.1 eV) [53], [54]. US-anode presented slightly lower fractions of alkyl and C = O and much lower fraction of C-O than Non-anode. It is considered that the decreased species on the surface of the US-anode might be related to the breaking effects of ultrasound. It is typically considered that organic species are present mainly in the inner layer, while inorganic species exist in the outer layer [55]. Regarding inorganic species (e.g., LiF, Li₂CO₃) that have higher Young’s modulus than organic species, with the implementation of ultrasound, fragile organic components (e.g., Li₂EDC, LiEC, LiMC) were preferentially destroyed, resulting in a different chemical composition of the surface layer [38]. However, the inner layer, which has a different material composition compared to the outer layer, was able to sustain its structure (e.g., LiF, Li₂CO₃). High resolution O 1 s and F 1 s spectrums were also fitted by curves of C = O (531.6 ± 0.1 eV), C-O (533 ± 0.1 eV), LiF (684.6 ± 0.1 eV), and Phosphate (686.3 ± 0.1 eV) (Fig. S13) [53], [54]. Fractions related to alkyl carbonates were clearly diminished, while the peak related to Li₂CO₃ was slightly decreased in the O 1 s spectrum. In addition, the fraction of LiF in the F 1 s spectrum sustained its intensity after the application of ultrasound.

Fig. 4.

Post-mortem analysis of cycled anodes from Graphite||LCO batteries. Cells were cycled 200 times with and without ultrasound. SEM images of (a, b) anode of Cell-US (US-anode) and Cell-Non (Non-anode), and EDS analysis of (c) US-anode and (d) Non-anode. (e) Wide-scan survey XPS of US-anode and Non-anode. C1s spectrums of (h) US-anode and (i) Non-anode. (h) Reduced-scale SEM images of anodes of US-anode and Non-anode. (i) XRD patterns and (j) Raman spectrum of US-anode and Non-anode.

To analyze the structural differences between the US-anode and Non-anode, microscopic images and crystallographic properties of anodes were examined. There were no perceptible differences in the particle size between the samples, implying ultrasound cannot break graphite per se (Fig. 4h). To confirm the crystal structures of US-anode and Non-anode, XRD and Raman spectra were examined (Fig. 4i and 4j). Both US-anode and Non-anode displayed a peak around 26.426, 44.462, and 54.512 corresponding to crystal planes of (0 0 2), (1 0 1), and (0 0 4), respectively. In Raman spectra, both samples show very similar profiles, indicating that ultrasound does not change the crystal structure of the anodes.

3.6. Mechanical simulations on the structural changes of the SEI

The SEI structures of US-anode and Non-anode are schematically illustrated in Fig. 5a. After 200 cycles without ultrasound, a thick and rough surface was observed, perhaps deteriorating the internal resistances of the electrodes. However, the surface of electrodes with ultrasound displayed a flattened surface in cross-sectional FIB-SEM images (Fig. 5b and 5c). To investigate the feasibility of flattening under an external vibration energy source, a finite element model was developed. To simulate the different surface morphology development, irregular shapes of layers were first designed based on the FIB-SEM images of cycled graphite (Fig. 5b, 5d, and Fig. S14). The surface structures were modeled by two homogeneous layers: the inorganic inner layer and the organic outer layer. Previous research by Brett et al. indicated that the inner layer of the SEI is primarily composed of inorganic species, while organic species from solvent reduction are dominant in the outer layer [56].

Fig. 5.

(a) Schematic illustration of the surface of the graphite with and without ultrasound after 200 cycles. FIB-SEM images of the (b) Non-anode and (c) US-anode. (d) The simulation model of two homogeneous SEI layers in harmonic acoustics. Contours of von mises stress (σ_von-mises) of (e) LiF-Li₂EDC and (f) Li₂CO₃-Li₂EDC model within ultrasound.

The stress states of the surface layer were investigated by introducing harmonic acoustics to the designed structure. The Von-mises stress (σ_von-mises) on the surface was examined to understand the effect of stress on the SEI (Fig. 5e and 5f). The elements of the inner and outer layers were differentiated due to their different compositions. Specifically, the inner layer was defined as Li₂CO₃ and LiF, while the outer layer was defined as Li₂EDC. It is known that material starts to yield when the σ_von-mises exceeds its Young’s modulus. However, Young’s modulus of Li₂EDC is 22.0 GPa, which is much lower than that of Li₂CO₃ and LiF (36.2 and 58.0 GPa, respectively), indicating that the outer layer is more susceptible to breaking. To determine whether yielding occurs in the SEI layer with ultrasound, the σ_von-mises was divided by Young’s modulus of Li₂EDC (E_Li2EDC). The model of LiF and Li₂EDC showed that σ_von-mises / E_Li2EDC > 1 in most of the outer layer region, while the inner layer region was barely affected by ultrasound (Fig. 5e). Additionally, the model of Li₂CO₃ and Li₂EDC displayed similar results to that of LiF and Li₂EDC (Fig. 5f). Although the yield point of Li₂CO₃ is in the range of the contour plot, it does not exist in the inner layer. The simulation results further confirm the decreased C-O bonding in C 1s, the noticeable peak of CO₃^– in C 1s, and the presence of LiF in F 1s.

To summarize, during cycling processes, electrolytes are typically decomposed, and solid products are formed on the surface of the electrode [41]. This solid product is referred to as the SEI, which covers the electrode surface. The SEI mainly consists of both inorganic and organic species. Inorganic species are mostly found in the inner space of the SEI while organic at the outer space, showing a bilayer heterogeneous structure. These inorganic and organic species have different mechanical properties; Inorganic species generally exhibit higher mechanical properties. When ultrasound is imposed, it will selectively break the weak parts of the SEI layer (predominantly organic species), finally leaving the inorganic-rich and thin SEI on the electrode surface. It is worth mentioning that the simulation presented in this work only considered the stress distribution and thus it should be carefully applied to other cases. Nevertheless, the results indicate that ultrasound may influence the surface chemistry with high stress.

3.7. DFT calculations of ion adsorption

To investigate how the surface chemical composition change by ultrasound affects electrochemical performances, DFT calculations were performed. The interactions between LiPF₆ salt present in the electrolyte and the constituents of the SEI layer (LiF, Li₂CO₃, LiEC, LiMC, and Li₂EDC) were calculated in order to observe the electrochemical reactions. LiPF₆ adsorption energies were calculated using the following Eq. (8),

$$E_{\mathit{ad}}\left(LiP{F}_6\right)=E\left(SEI+LiP{F}_6\right)-E\left(SEI\right)-E\left(LiP{F}_6\right)$$ (8)

where $E\left(SEI+LiP{F}_6\right)$, $E\left(SEI\right)$, and $E\left(LiP{F}_6\right)$ represent the energy of adsorbed state, constituents of the SEI layer, and LiPF₆ salt. In Fig. 6a, on the surfaces of inorganic LiF and Li₂CO₃, the adsorption energies were found to be −1.046 and −0.812 eV, respectively, indicating moderate adsorption abilities. However, on the organic constituents LiEC, LiMC, and Li₂EDC, the adsorption energies were −1.794, −1.809 and −2.707 eV, respectively, demonstrating significantly strong adsorption capabilities. Such strong adsorption can lead to additional electrolyte decomposition, causing the formation of an uneven SEI layer and hindering the diffusion of Li ions towards the anode surface. To investigate the electron transfer that influence the LiPF₆ adsorption energy, the charge density difference was calculated (Fig. 6b-f). Li in LiPF₆ forms strong bonding with anions such as oxygen and fluorine. However, in the case of inorganic components, the anions already form strong ionic bonds with the bulk region, resulting in fewer electron transfers between Li and the surface anion. In the case of organic SEI layers, excessive electron transfer due to the presence of additional solvent or salt can lead to their decomposition, resulting in the formation of an uneven SEI layer. This confirms that the formation of an ultrasonic-assisted inorganic-rich thin SEI layer contributes to the improvement of electrochemical performance consistent with our experimental measurements.

Fig. 6.

(a) LiPF₆ adsorption energy on different SEI component. The charge density difference between LiPF₆ and SEI component for: (b) LiF (0 0 1), (c) Li₂CO₃ (0 0 1), (d) LiEC, (e) LiMC, and (f) Li₂EDC. (Isosurface level: 0.001 e/ų) The blue, green, purple, brown, red and white spheres represent fluorine, lithium, phosphorus, carbon, oxygen, and hydrogen atoms, respectively. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

3.8. Post-mortem analysis of cathodes

Meanwhile, as the SEI is formed at the surface of the anode, the cathode electrolyte interphase (CEI) is generated at the surface of the cathode [57]. However, no major differences were observed between the cathode of Cell-US (US-cathode) and Cell-Non (Non-cathode) (Fig. S15). Unlike the flattened surface of the anodes in the US-anode, the US-cathode displayed similar SEM images compared to Non-cathode (Fig. S15a and S15b). Additionally, EDS results presented a similar elemental composition of the surface of the cathodes (Fig. S15c, d, and S16). These indicate that ultrasound cannot change the surface chemistry of cathodes, called CEI. Previous studies reported that the outer part of the CEI reminds the inner part of the SEI, which can prevent breakages of CEI from ultrasound with the high mechanical stability of LiF [58]. Cross-sectional FIB-SEM images underpin the results above (Fig. S17).

To further investigate the elemental composition of the cathodes, US-cathode and Non-cathode were analyzed using XPS (Fig. S15e-g). The high resolution C 1 s spectrums of US-cathode and Non-cathode include five curves: C-C/C-H (284.3 ± 0.1 eV), PVDF/C-O (285.6 ± 0.4 eV), C = O/O-C-O (286.9 ± 0.1 eV), C = O-O (288.1 ± 0.1 eV), and CO₃^– (289.6 ± 0.1 eV) [59]. As can be seen, C1s spectrums of the US-cathode sustained its chemical composition after the implementation of ultrasound. Furthermore, microscopic images were examined to investigate the structural differences between the US-cathode and Non-cathode. The particle size of LCO was the same in both enlarged images (Fig. S15h), indicating that ultrasound cannot break LCO, just as graphite. These similarities between US-cathode and Non-cathode were also observed in crystal structure analysis using XRD and Raman spectra (Fig. S15i and S15j). Both samples presented peaks at 18.931, 37.388, 38.405, 39.058, and 45.22, corresponding to crystal planes of (0 0 3), (1 0 1), (0 0 6), (1 0 2), and (1 0 4), respectively. In Raman spectra, differences between US-cathode and Non-cathode were imperceptible.

4. Conclusion

In summary, we demonstrated the thin and inorganic-rich SEI layer through the application of ultrasound on LIBs for high-performance batteries. The differences in mechanical properties between SEI components contributed to the formation of a thin and inorganic-rich SEI layer. With mechanical simulations, we have shown that organic species tend to break preferentially, primarily present at the outer layer of the SEI, and possess low mechanical properties. This led to a flattened surface of the anodes in Cell-US and a reduction in the thickness of the SEI layer. The thin and inorganic-rich SEI layer reduced the diffusion length of Li⁺, facilitating Li⁺ ion diffusion with low adsorption capabilities of inorganic species. Therefore, Cell-US exhibited a significantly reduced R_ct of 274 mΩ and R_SEI of 230 mΩ on the average of 200 cycles, lower than Cell-Non with values of 442 mΩ and 979 mΩ. Owing to the enhanced charge transfer kinetics from the low resistance of Cell-US, excellent cycling stability with 87.4 % capacity retention (vs. 58.1 % of Cell-Non) was observed after 200 cycles at 1C.

This work provides a new perspective on thinning the SEI layer with a high composition of inorganic components and high energy batteries without any pretreatment. That is, ultrasound could be utilized to exploit surplus energy of batteries system. Meanwhile, these findings may be helpful to build a design guideline for LIB housings in EVs (e.g., module pack and mount-lack design) rather than employing the battery-lack-case which blocks almost all of vibrations to avoid uncontrolled issues [60].

CRediT authorship contribution statement

Sunghyun Jie: Conceptualization, Writing – original draft. Joonhee Kang: Conceptualization, Data curation, Writing – original draft. Seunghun Baek: Data curation, Writing – review & editing. Byeongyong Lee: Conceptualization, Formal analysis, Supervision, Writing – review & editing.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

The following are the Supplementary data to this article: