Editors' Choice—Investigation of the Dynamic Evolution of the Cathode-Electrolyte Interphase Using Scanning Electrochemical Microscopy

All chemicals were purchased and used commercially. Propylene carbonate (PC, anhydrous, 99%) was purchased from Energy Chemical. Ethylene carbonate (EC, anhydrous, 98%) was purchased from Macklin. Lithium iron phosphate (LiFePO₄, >98%, D90 ∼ 3.556 μm, polycrystalline carbon coated, ∼ 1.5%) was purchased from Shenzhen Kejing Star Technology Co. Ethyl viologen diperchlorate (EV(ClO₄)₂, 98%) was purchased from Aladdin. Lithium hexafluorophosphate (LiPF₆, 99.9%) was purchased from DoDoChem. 12.5 μm radius Au wire (≥99.99%) was purchased from Zhongnuo Advanced Materials Technology Co. Pyrrole (99%) was purchased from TCI Chemicals. Acetone (AR) was purchased from ChinaSun Specialty Products Co., Ltd. Polyvinylidene fluoride (PVDF) and tetrabutylammonium hexafluorophosphate (TBAPF₆, 98%) were purchased from Aladdin. N-Methylpyrrolidone (NMP, 99.9%) was purchased from Sigma-Aldrich. Ketjen Black (ECP-600JD) was purchased from TIMCAL. Diethyl carbonate (DEC, 99%) was purchased from Energy Chemical. Acetonitrile (99.5%) was purchased from was purchased from TCI Chemicals.

Substrate fabrication and characterization

Electrochemical measurements

SECM feedback imaging

Approach curve

The complicated process of CEI formation can be influenced by a number of variables, including electrolyte, operating voltage, additive, and inherent properties of the cathode material, etc.^13,19 So, we first want to confirm how the properties of the cathode material itself influence dynamic evolution of CEI. An electric spray pen was used to fabricate LFP particles onto the aluminum foil to fabricate the patterned LFP electrode (Fig. 1a). As the average diameter or radius of the LFP particles is approximately 1.3 μm, it is imperative that the patterned LFPs have sufficient thickness and coverage for the SECM probe to obtain the necessary electrochemical information.^43,44 The accuracy of the total signal collected decreases with thinner LFP layers and less surface coverage (Figs. S1b, S1c). Finally, a moderate thickness has been achieved. As shown in Figs. 1b, 1c and 1d, the patterned LFP presents an array of 200 μm array with 500 μm center separation, and an average pattern height close to 8 μm in diameter.

Zoom In Zoom Out Reset image size

In order to visualize the in situ formation of CEI, we utilized the SECM imaging technique to track the LFP surface activity change while cycling. In this experiment, a 25 μm diameter Au UME was selected to monitor the change in redox couple EV²⁺/EV⁺ feedback charge transfer kinetics on patterned LFP substrate (Fig. 2a).^31,45 During the experiment, Au UME was held at 2.4 V vs E_Li⁺_/Li to reduce EV²⁺ to EV⁺, a diffusion-limited steady-state current (i_T) on the probe. This steady-state current will increase or decrease as the probe approaches the substrate surface in response to variations in the local diffusion capacity of EV⁺ and the local regeneration rate of EV²⁺ caused by differences in substrate activity and topography.³⁷ Since the original LFP sample surface is covered by 1.5% polycrystalline carbon, this ensures the LFP substrate to oxidize EV⁺ back to EV²⁺. On the contrary, when the probe approaches to the surface of the aluminum current collector (surface covered by native oxide, Al₂O₃), or the passivated LFP particle after CEI formation, a negative feedback current with deduced i_T is generated. To simplify the analysis, the measured tip current i_T, is normalized to i_T/i_{T, ∞}. The changes of i_T/i_{T, ∞} reflect the electron transfer and interphase formation across the LFP surface.

Zoom In Zoom Out Reset image size

We then examined the potential-dependent reactivity of the patterned LFP electrode to monitor the development of the CEI formation using SECM feedback mode. The substrate, working and counter electrodes used in this experiment were the patterned LFP electrode, a 25 μm diameter Au UME and a Pt wire, respectively. The SECM experiment was performed inside an Ar filled glovebox, using an electrolyte solution of 1:1 (v/v) EC/PC containing 8 mM EV(ClO₄)₂ and 0.1 M LiPF₆. SECM images of the patterned LFP region were measured at different substrate voltage biases (3.1 ∼ 4.0 V vs E_Li⁺_/Li) to study the dynamic evolution of the CEI (see Figs. 2b, 2c for representative potentials, and Fig. S4 ∼S7, for full data). From the SECM image, we can tell the distinct difference between circular positive feedback LFP pattern and negative feedback aluminum foil region. During the first cycle, as the voltage gradually rose to 4.0 V and returned back to 3.1 V, the feedback current began to gradually decrease, indicating that electrolyte decomposition was slowly occurring on the surface of the LFP particles. As the cycling continued, the feedback current continuously changed from positive feedback to negative feedback, suggesting the accumulation of decomposition products and the formation of an uneven CEI layer that progressively covered LFP surfaces (Fig. 2b). The changes in the normalized current at the surface of the LFP depicted the formation process of an inhomogeneous CEI during the initial charging and discharging steps which agrees with previous reports.^46,47 At the second cycle, the normalized current of the LFP surface turned into negative feedback current completely (Fig. 2c). This phenomenon illustrates how the CEI evolves upon cycling.

To further evaluate the evolution of CEI at different voltages during charging and discharging in more detail, we proceeded to quantify the dynamic electron-transfer kinetics constant k_f at different voltages. Following each SECM image, repetitive approach curves were obtained with progressive substrate potential (E_sub) biases, and we quantified the kinetic parameter k_f during CEI generation via mathematical fitting (Fig. 3). It is clear that during the first cycle, the approach curves trends did not change significantly and the LFP surface maintained positive feedback (Fig. 3a). However, by the second charge/discharge cycle, the passivation becomes more recognizable (Fig. 3b). It is important to note that this slow passivation is not a uniform process, as we have also observed evidence of earlier change at other sites (Fig. S8). Therefore, we repeated the experiment over 5 positions, and summarized the trend of average k_f with respect to different E_sub. As shown in Fig. 3c, the average k_f remains stable around 1.4 × 10⁻⁴ m s⁻¹ upon first cycle charge-discharge with a total duration of 19 h, but rapidly loose half activity once step into second cycle (3.3 V vs E_Li⁺_/Li).

Zoom In Zoom Out Reset image size

Considering that the CEI is not generated uniformly on the electrode surface during the actual evolution process, apart from single site approach curves, we further divide the LFP SECM pattern into three parts (i_T/i_{T, ∞} > 1 (I), 0.9 < i_T/i_{T, ∞} < 1 (II), i_T/i_{T, ∞} < 0.9 (III)) based on the normalized current value (Fig. 3d), to better assess the evolution of CEI throughout the whole electrode. Region Ⅰ represents the area with positive feedback, probably caused by higher local activity of LFP or closer tip-substrate distance due to raise topography at pattern center. Region Ⅱ shows the area with mixed feedback behavior at the edge of the pattern. Region Ⅲ covers most of the exposed aluminum current collector region with negative feedback response. We selected a strategic location to observe the changes in the feedback current to illustrate the evolving process of CEI (Fig. S9a). During the charge-discharge process, the proportion of region Ⅰ will gradually decrease and diminish at 3.4 V vs E_Li⁺_/Li of second charge cycle, which agrees with the previous calculated k_f trend. This indicates the active LFP surface gradually passivated by the formation of CEI, and eventually turned into compromised reactivity of region Ⅱ. In contrast, the stability of commercial aluminum foil is demonstrated by the fact that the aluminum current obtained (Region III) does not change during cycling (Fig. 3e). As shown in Figs. 3f, 3g, the k_f of the four sites (sample 2 ∼ 5) decreased from ∼ 1.2 × 10⁻⁴ m s⁻¹ to roughly 5 × 10⁻⁵ m s⁻¹ after cycling. As a comparison, we discover the k_f of the aluminum foil only suffers a slight activity increase which reveals the trivial electrolyte decomposition on the surface of the current collector (Fig. S9b). In general, these results show that CEI evolution is continuously proceed in both processes of charging and discharging. During this process, the CEI will gradually accumulate and cycle into a thicker state.

In addition to the SECM results mentioned above, the presence of CEI was also confirmed by Raman spectroscopy and XPS analysis. The Raman spectrum of the pristine and cycled electrode is shown in Fig. 4a. For pristine LFP sample, we can clearly locate its lattice vibrational response (below 400 cm⁻¹), antisymmetric stretching mode of PO₄³⁻ (400 ∼ 700 cm⁻¹), and intermolecular stretching motions of PO₄³⁻ (900 ∼ 1100 cm⁻¹).⁴⁸ After cycling, these characteristic responses of LFP almost disappear, leaving only broad D (1340 cm⁻¹) and G (1600 cm⁻¹) bands of the polycrystalline carbon component. This result indicates that the surface sensitive Raman interaction of LFP is blocked by the development of CEI.

Zoom In Zoom Out Reset image size

The chemical composition of the CEI was further investigated using XPS. The C 1 s, O 1 s, F 1 s of the XPS spectra demonstrate that both inorganic and organic species were observed by comparing the pristine and cycled LFP substrates. When comparing the C 1 s spectra of cycled LFP samples, we observe an enhanced C-O-C peak at 286 eV and the appearance of a new peak at 291 eV, which is thought to correspond to Li₂CO₃ or ROCO₂Li species (Fig. 4b). It is easy to find the corresponding intensified C=O peak at 533 eV in O 1 s spectra (Fig. 4c).

Except for the coexistence of common elements in both cycled and pristine substrates, the appearance of LiF (685.5 eV), Li_xPO_yF_z (687.5 eV) bands on F 1 s spectra of cycled LFP proved the reaction between LiPF₆ electrolyte and LFP (Fig. 4d). Li_xPF_y, the product of the LiPF₆ decomposition reaction, was found at 136.5 eV in P 2p spectra, 57 eV in Li 1 s spectra as well (Fig. S10). The distribution of organic and inorganic components we observed is consistent with previous research under similar conditions.^49–51

Considering that the actual working electrode is typically a composite mixture of active electrode material, conductive additive and binder. The behavior of composited cathodes may differ from the CEI generation mechanism seen within the active LFP particles alone. Here, we also prepared the composited LFP electrode and repeated the above SECM experiment in the same system. The SECM images and approach curves upon composited LFP cycling are shown in Figs. 5a, 5b and S11 ∼ S14. The boundary between the composited LFP and the aluminum foil was selected as the region of interest. The location of selected points has been shown in Fig. S15a. Unlike the patterned LFP electrode, there is no apparent CEI passivation generated in the SECM images after two full cycles. The results of the successive approach curves also monitored the k_f changes in situ. In contrast to the pattern LFP, the composited LFP revealed constant positive feedback behavior with the heterogeneous k_f fitting maintained at ∼ 1.2 × 10⁻⁴ m s⁻¹ (Figs. S15b, S15c). However, the XPS results of the composite electrode also detected substances similar to CEI (Fig. S16). Therefore, the composited LFP with mixed components demonstrated resistance towards CEI formation induced surface passivation, but it doesn't completely stop the electrolyte from slowly and continuously degrading.

Zoom In Zoom Out Reset image size

Combining the experimental results of patterned native LFP and composited LFP, we find that the conductive additive and binder are important factors in CEI growth. Thus, we propose that the conductivity at electrode-electrolyte interface can be a critical factor that controls the dynamic evolution of CEI, especially at normal operating voltages. For the patterned LFP electrode with weak electron transfer capability, the electrolyte is adsorbed on the surface of the active material and slowly decomposes upon charging. Due to the weak contact between the cathode material and current collector (native oxide densely packed aluminum foil), electrons cannot be efficiently transferred between LFP and current collector in a timely manner. This leads to accumulated local potential distortions, progressive electrolyte decomposition and continuous formation of CEI (Fig. 5c). For the composited LFP, good electrical conductivity promotes the rapid charge transfer, thus weaken the uncontrolled potential bias induced CEI accumulation. Our finding can explain why thin and uneven CEI formation is observed in real battery system after long-term cycling, when volume expansion, material inactivation and decomposition cause distortions in local conductivity.

In this work, the dynamic evolution of CEI formation on patterned LFP and composited LFP is investigated using in situ SECM. Combining SECM-based electrochemical measurements, the surface kinetics and the local passivation behavior during the CEI formation for pristine LFP sample were systematically demonstrated, with clear evidence of CEI formation during second cycle and a distinct k_f drop. On the composited LFP, however, the same results cannot be obtained under the same conditions. Therefore, we propose a point that the CEI generated can be attributed to the limitation of electron transfer leading to the decomposition of electrolytes within the operating voltage range. Our results have shown that the LFP itself is unable to transfer electrons in a timely manner throughout the cycles due to the lack of conductive additives, causing the adsorbed electrolyte on the LFP surface to oxidize. The decomposition products generate an unevenly thick passivation covering the surface of the LFP, which eventually accumulates to form CEI layer. In composited LFP, conductive additives reduce electrolyte decomposition, facilitate electron transfer and prevent CEI formation. These results not only provide a new insight into the fundamental understanding of the dynamic evolution of CEI, but also provide a potential strategy for constructing CEI with good performance to achieve cathode protection under long cycles. Additionally, it provides new ideas for SECM to be applied in the field of lithium-ion batteries, especially for the research on the cathode materials side.

This work was supported by the National Natural Science Foundation of China (22204115), the Natural Science Foundation of Jiangsu Province (BK20220485), the Suzhou Municipal Science and Technology Bureau (ZXL2022494) and the start-up research grant for a distinguished professor at Soochow University (J.H.).