The pursuit of carbon neutrality is driving a profound transformation in global energy infrastructure. The lithium-ion battery stands as a cornerstone technology for electric vehicles and grid-scale energy storage. However, its widespread deployment is critically constrained by persistent challenges related to longevity and safety. Lithium plating, a parasitic side reaction, is a primary accelerator of battery degradation and a significant safety hazard. It occurs under strenuous conditions such as low-temperature or fast charging, where excessive polarization at the graphite anode prevents the intercalation of all arriving lithium ions, forcing their reduction to metallic lithium on the anode’s surface. This plated lithium, often dendritic, is highly reactive. It consumes active lithium and electrolyte through continuous solid electrolyte interphase (SEI) growth, leading to rapid, irreversible capacity loss. More dangerously, it can penetrate the separator, inducing an internal short circuit and potentially triggering thermal runaway. Therefore, the precise and timely detection of lithium plating is paramount for developing intelligent charging protocols, extending battery life, and ensuring the operational safety of lithium-ion battery systems.
Existing detection methodologies are bifurcated into ex-situ and in-situ categories. Ex-situ techniques, such as scanning electron microscopy (SEM), X-ray diffraction (XRD), and nuclear magnetic resonance (NMR), offer definitive material characterization but require cell disassembly, rendering them unsuitable for operational battery management. In-situ methods, including reference electrode potential monitoring, high-precision coulometry, voltage relaxation analysis, and ultrasonic imaging, provide pathways for real-time, non-destructive diagnosis. Among these, Electrochemical Impedance Spectroscopy (EIS) has garnered substantial interest due to its unique ability to probe the kinetic processes within a lithium-ion battery across different time scales—from ohmic conduction and charge transfer to mass diffusion—all without invasive intervention. Since the electrode/electrolyte interface kinetics are directly altered by the onset of lithium plating, EIS serves as a powerful diagnostic tool. This article provides a comprehensive review of EIS-based techniques for lithium plating detection, categorizing them into static EIS, dynamic EIS (DEIS), and current interruption methods. We delve into their underlying principles, comparative advantages, limitations, and future application landscapes.
Fundamentals of Electrochemical Impedance Spectroscopy in Lithium-Ion Batteries
Electrochemical Impedance Spectroscopy operates on the principle of perturbing a linear(ized) electrochemical system with a small-amplitude sinusoidal signal and analyzing the system’s frequency-dependent response. For a lithium-ion battery, a sinusoidal current excitation \( I(t) = I_0 \sin(\omega t) \) is applied, resulting in a voltage response \( V(t) = V_0 \sin(\omega t + \phi) \), where \( \omega \) is the angular frequency and \( \phi \) is the phase shift. The complex impedance \( Z(\omega) \) is then calculated using a form of Ohm’s law in the complex domain:
$$ Z(\omega) = \frac{V(\omega)}{I(\omega)} = |Z| e^{j\phi} = Z’ + jZ” $$
where \( Z’ \) is the real part (resistance), \( Z” \) is the imaginary part (reactance), and \( j = \sqrt{-1} \). By sweeping the frequency \( \omega \) over a wide range (typically from millihertz to megahertz), an impedance spectrum is obtained, which encodes the kinetic signatures of various physical processes. For commercial lithium-ion batteries with low internal impedance, galvanostatic EIS (applying an AC current) is typically employed over potentiostatic EIS.
The impedance response of a lithium-ion battery is commonly interpreted using equivalent circuit models (ECMs). A fundamental model for a single electrode (e.g., a graphite anode) is the Randles circuit, which includes solution resistance \( R_\Omega \), charge transfer resistance \( R_{ct} \), double-layer capacitance \( C_{dl} \), and Warburg impedance \( Z_W \) representing diffusion. For a full cell, the model becomes more complex, often comprising series and parallel combinations representing both electrodes. The occurrence of lithium plating introduces a new, parallel Faradaic reaction pathway at the anode interface. This effectively alters the interfacial kinetics, leading to measurable changes in the parameters of the ECM, particularly those related to the anode’s charge transfer process.
Before analysis, the quality of EIS data must be validated to ensure it conforms to the conditions of linearity, causality, and stability, often assessed via Kramers-Kronig transformations. The interpretation of spectra, whether through ECM fitting or the model-free Distribution of Relaxation Times (DRT) analysis, forms the basis for detecting the subtle shifts caused by lithium plating in a lithium-ion battery.
Static EIS Method: Post-Mortem and Reference Performance Analysis
The static EIS method involves measuring the impedance spectrum of a lithium-ion battery after it has reached a steady state at a predefined state of charge (SOC) and temperature. This typically requires a prolonged relaxation period following a charge/discharge step to allow internal gradients (concentration, potential) to dissipate. In aging studies, static EIS is frequently incorporated into Reference Performance Tests (RPTs) to track the evolution of internal resistance and infer degradation modes.
The underlying principle for plating detection lies in the cumulative impact of severe or repeated lithium plating on the battery’s impedance signature. Plating consumes electrolyte and catalyzes excessive SEI growth, leading to an increase in the ohmic resistance \( R_\Omega \). Furthermore, the metallic lithium itself can create new conductive pathways or alter the interfacial properties. Research on commercial lithium iron phosphate (LFP) cells subjected to low-temperature cycling (-22°C) showed that while the charge transfer resistance \( R_{ct} \) remained relatively stable, a noticeable increase in \( R_\Omega \) was observed post-cycling, attributed to electrolyte depletion from SEI formation driven by plated lithium reactions.
A more dynamic application of static EIS involves sequential measurements during the voltage relaxation period following a charge. After a charging step that induces plating, the cell voltage relaxes, often exhibiting a characteristic plateau as the plated lithium re-intercalates or reacts. By taking EIS measurements at intervals during this relaxation, researchers have correlated the reversible contraction of the mid-frequency impedance arc (associated with \( R_{ct} \)) with the voltage plateau, providing a direct in-situ indicator of reversible lithium stripping from the anode surface.
Table 1 summarizes the characteristic impedance changes associated with lithium plating observed via static EIS in a lithium-ion battery.
| Impedance Parameter | Physical Origin | Typical Change After Severe Plating | Primary Cause |
|---|---|---|---|
| Ohmic Resistance (\( R_\Omega \)) | Ionic resistance of electrolyte, contacts | Increase | Electrolyte depletion from parasitic SEI growth. |
| Charge Transfer Resistance (\( R_{ct} \)) | Kinetics of Li-ion insertion at anode | May decrease, increase, or stay stable | Decrease: New parallel Li plating path. Increase: Pore blocking, SEI thickening. |
| Low-Frequency Warburg Impedance | Solid-state Li diffusion in anode | Increase (Warburg slope change) | Loss of active material, pore clogging by plated Li. |
The primary advantage of the static EIS method is the high quality and reproducibility of the data, owing to the stable measurement conditions. Its primary limitations are poor temporal resolution and low sensitivity to the initial onset of plating. It is more suited for post-cycling analysis or confirming the presence of substantial, accumulated lithium plating rather than for real-time detection and prevention during operation. The requirement for long relaxation times also makes it impractical for online battery management.
Dynamic EIS (DEIS) Method: Real-Time Kinetic Interrogation
Dynamic EIS addresses the real-time limitation of static EIS by superimposing a small-amplitude AC perturbation signal onto the operational DC current (e.g., during charging). This allows for the continuous or periodic acquisition of impedance spectra while the battery is under load, enabling the direct observation of kinetic changes as they occur.
The core detection principle is based on the modification of the anode’s interfacial charge transfer process. Under normal intercalation, the Faradaic current \( i_F \) is governed by the Butler-Volmer equation for Li-ion insertion into graphite. Upon the onset of lithium plating, a second Faradaic reaction—lithium metal deposition—commences in parallel. This introduces an additional charge transfer resistance \( R_{ct, plating} \) in parallel with the intercalation charge transfer resistance \( R_{ct, inter} \). The effective charge transfer resistance \( R_{ct, eff} \) as sensed by EIS is given by:
$$ \frac{1}{R_{ct, eff}} = \frac{1}{R_{ct, inter}} + \frac{1}{R_{ct, plating}} $$
Since \( R_{ct, plating} \) for lithium metal deposition is typically low, the onset of plating causes a detectable decrease in \( R_{ct, eff} \), which manifests as a contraction or distortion of the mid-frequency semicircle in the Nyquist plot or a drop in the impedance modulus at characteristic frequencies.
Practical implementations vary. One approach uses a specialized onboard impedance analyzer to perform consecutive DEIS measurements during a Constant Current-Constant Voltage (CCCV) charge. By tracking specific feature points on the impedance spectrum—such as the high-frequency real-axis intercept (\( R_0 \)), the apex of the mid-frequency arc (\( R_{Max}, X_{Max} \)), and a low-frequency point (\( R_{Min}, X_{Min} \))—researchers have identified a simultaneous drop in \( R_{Min} – R_0 \) and \( X_{Min} \) as a reliable indicator of plating onset, later validated by post-charge voltage relaxation analysis.
To drastically reduce measurement time, advanced techniques employ multi-sine excitations, where the DC current is perturbed by a sum of sinusoids at different frequencies. The voltage response is then processed via Fourier transform to deconvolve the impedance at all frequencies simultaneously. This “single-shot” EIS, combined with state estimation algorithms like the Extended Kalman Filter (EKF) to monitor the transition frequency impedance, has been successfully demonstrated for online plating detection.
Perhaps the most insightful application of DEIS couples it with Distribution of Relaxation Times (DRT) analysis. DRT deconvolves the impedance spectrum into a distribution of polarization contributions without assuming a specific ECM. Studies on three-electrode cells have shown that during fast charging, the onset of lithium plating is marked by a distinct minimum in the DRT peak corresponding to the anode charge transfer resistance (\( R_{ct,a} \)), followed by a significant rise in the peak associated with the SEI resistance (\( R_{SEI} \)). The initial rise of the \( R_{SEI} \) peak has been proposed as a highly sensitive plating onset marker, capable of detecting plating corresponding to less than 0.6% of the graphite anode’s capacity.
Recent pioneering work has utilized in-situ DEIS combined with thickness measurement to delineate the dynamic three-stage evolution of the anode charge transfer resistance \( R_{ct,a} \) during extreme fast charging of a lithium-ion battery:
- Stage I (Linear Decline): \( R_{ct,a} \) decreases slowly and linearly, corresponding to standard Li-ion intercalation without plating. The overpotential is below the plating threshold.
- Stage II (Accelerated Decline): \( R_{ct,a} \) drops precipitously. This is linked to lithium nucleation and growth on the graphite surface, introducing the parallel low-resistance plating pathway.
- Stage III (Plateau): \( R_{ct,a} \) decrease slows and plateaus. The plating reaction dominates the interface, and massive plating leads to rapid cell thickness expansion.
This multi-signal approach provides a comprehensive physical picture of the plating process.
A significant challenge for wide-frequency DEIS in fast-charging applications is the scan time versus the rapidly changing SOC. A promising solution is single-frequency EIS targeting a specific kinetic process. Research has established that the double-layer capacitance \( C_{dl} \) at the graphite anode, derivable from impedance at the characteristic frequency of the charge transfer process, is proportional to the electrochemical active surface area (ECSA). During intercalation, the ECSA remains stable, but the onset of lithium plating creates new metallic surface area, causing a sharp increase in \( C_{dl} \). Monitoring this single-frequency-derived \( C_{dl} \) in real-time offers a fast and sensitive plating detection method.
Table 2 compares different DEIS implementation strategies for lithium plating detection in a lithium-ion battery.
| Strategy | Excitation Signal | Key Detection Parameter | Advantage | Challenge |
|---|---|---|---|---|
| Sequential Frequency Sweep | Single sine, swept frequency | Drop in |Z| at mid/low freq. points (e.g., \( R_{Min}-R_0 \)) | Provides full spectrum; Well-established. | Slow; Susceptible to SOC drift during measurement. |
| Multi-Sine/Frequency Domain Analysis | Sum of sinusoids | Trend of transition frequency impedance via EKF | Very fast measurement; Suitable for dynamic profiles. | Complex signal processing; Requires careful design of excitation. |
| DRT Analysis of DEIS | Wide-band excitation | Onset of \( R_{SEI} \) peak rise in DRT | High sensitivity; Model-free; Distinguishes \( R_{ct} \) and \( R_{SEI} \). | Computationally intensive; Requires high-quality data. |
| Single-Frequency Monitoring | Single fixed-frequency sine | Sharp increase in derived \( C_{dl} \) | Extremely fast; Simple processing; Easy to implement. | Requires prior identification of characteristic frequency; Less comprehensive. |
The advantages of DEIS are clear: high sensitivity, real-time capability, and the potential to pinpoint the exact moment of plating onset for adaptive charging control. However, its widespread adoption faces hurdles: the need for specialized, high-bandwidth measurement hardware adds cost and complexity; integrating such systems into existing Battery Management System (BMS) architectures is non-trivial; and ensuring data quality under high-rate dynamic loads remains a technical challenge.
Current Interruption Method: Leveraging Transient Relaxation
The current interruption method is a clever and practical derivative of impedance techniques that seeks to extract kinetic information from the transient voltage relaxation following a brief interruption of the charge current, without requiring a full EIS analyzer.
When the charging current to a lithium-ion battery is abruptly interrupted, the cell voltage relaxes. The initial, rapid voltage jump is the instantaneous ohmic drop (\( I \cdot R_\Omega \)). The subsequent relaxation over milliseconds to seconds encompasses the decay of polarization associated with charge transfer and diffusion processes. The early part of this relaxation (within a few hundred milliseconds) is dominated by the charge transfer overpotential. By analyzing this short-term relaxation, one can extract parameters analogous to those found in EIS.
The fundamental link is through the time constant \( \tau_{ct} \) of the charge transfer process, which is related to the charge transfer resistance and double-layer capacitance: \( \tau_{ct} = R_{ct} C_{dl} \). Upon the onset of lithium plating, the introduction of the parallel plating pathway reduces the effective \( R_{ct} \), which in turn reduces the effective \( \tau_{ct} \). This can be detected by fitting the relaxation voltage curve to an appropriate model (e.g., a Mittag-Leffler function or a simple exponential) immediately after current interruption.
One implementation, known as the DC Resistance (DCR) tracking method, involves periodically interrupting the charge for a short duration (e.g., 3 seconds) at regular SOC intervals. The DCR is calculated from the immediate voltage step. Researchers have found that the DCR trend mirrors the transition frequency impedance from static EIS. During normal charging, DCR first decreases and then increases with SOC. However, the onset of lithium plating triggers an anomalous, sharp decrease in DCR in the high-SOC region, serving as a detection signal. This method has shown greater sensitivity than voltage relaxation plateau analysis at moderate temperatures.
A more refined approach uses high-speed data acquisition to capture the sub-second relaxation with high fidelity. By fitting the first ~250 ms of the relaxation voltage \( V(t) \) after interruption, the time constant \( \tau \) is extracted. Experiments using three-electrode cells have conclusively shown that a rapid drop in this extracted \( \tau \) coincides with the anode potential falling below 0 V vs. Li/Li+, the thermodynamic condition for lithium plating. The relationship can be modeled as a change in the dominant relaxation process:
$$ \Delta V(t) = A \cdot \exp\left(-\frac{t}{\tau_{inter}}\right) \quad \text{(Before Plating)} $$
$$ \Delta V(t) = A \cdot \exp\left(-\frac{t}{\tau_{eff}}\right) \quad \text{(After Plating Onset)}, \quad \text{where } \tau_{eff} < \tau_{inter} $$
where \( \tau_{inter} \) and \( \tau_{eff} \) are the time constants for pure intercalation and the mixed intercalation/plating process, respectively.
The major advantage of the current interruption method is its hardware simplicity and seamless integration potential. It requires no additional equipment beyond the BMS’s existing current control and voltage sensing capabilities, albeit with a requirement for relatively high sampling rates (kHz level). It provides a direct, real-time capable metric for plating onset. The principal trade-off is the intermittent pause in charging, which slightly extends the total charge time. The duration of the interruption must be carefully optimized: long enough to capture the relevant kinetic information but short enough to minimize the impact on the charging process. Table 3 outlines the key considerations for this method.
| Parameter | Typical Range | Influence on Detection | Trade-off |
|---|---|---|---|
| Interruption Duration (\( t_{int} \)) | 0.1 s to 5 s | Shorter \( t_{int} \) captures mainly \( R_\Omega \) and \( \tau_{ct} \); longer \( t_{int} \) may include diffusion effects. | Longer duration improves signal quality but increases total charge time. |
| Interruption Frequency | Every 0.5% to 2% SOC | Higher frequency provides finer resolution for onset detection. | More interruptions increase cumulative charge time and BMS processing load. |
| Voltage Sampling Rate (\( f_s \)) | > 1 kHz | Higher \( f_s \) allows accurate capture of the initial relaxation slope for \( \tau \) fitting. | Increases data volume and requires capable BMS analog-to-digital converters (ADCs). |
Synthesis, Comparative Analysis, and Future Perspectives
The three categories of EIS-based techniques form a complementary toolkit for lithium plating diagnosis in lithium-ion batteries at different stages of the battery’s life and under different operational constraints. A comparative summary is presented in Table 4.
| Method | Measurement Condition | Primary Detection Signal | Key Advantages | Key Limitations | Suitability |
|---|---|---|---|---|---|
| Static EIS | Steady-state (rest) | Cumulative changes in \( R_\Omega \), \( R_{ct} \), Warburg slope. | High data quality; Simple to implement off-line; Useful for aging mode analysis. | No real-time capability; Low sensitivity to onset; Requires long relaxation. | Post-cycling analysis; RPT in lab aging studies. |
| Dynamic EIS (DEIS) | Under load (in operando) | Real-time drop in \( R_{ct,eff} \), \( |Z| \) at key frequencies, or rise in \( C_{dl} \). | Real-time detection of onset; High sensitivity; Enables adaptive control. | Requires specialized hardware; Complex integration; Data quality under dynamics. | Online, advanced BMS for fast-charging; Laboratory mechanistic studies. |
| Current Interruption | Transient relaxation (micro-pause) | Decrease in DCR or extracted relaxation time constant \( \tau \). | Uses existing BMS hardware; Real-time capable; Low incremental cost. | Slightly increases charge time; Requires optimized interruption protocol. | Online BMS integration for mainstream EVs; Fleet management. |
Looking forward, the evolution of lithium plating detection in lithium-ion batteries will likely focus on two major frontiers beyond the current state-of-the-art:
1. Multi-Modal Sensor Fusion: Relying on a single signal like EIS, while powerful, may not be sufficiently robust across all real-world conditions (e.g., temperature variations, cell-to-cell inconsistencies, different aging states). The future lies in the synergistic fusion of EIS-derived parameters with other in-situ metrics. For instance:
- EIS + Expansion Metrics: Combining the drop in \( R_{ct} \) (kinetic change) with a sudden increase in cell thickness or swelling force (physical change) provides a highly robust, multi-physics confirmation of plating.
- EIS + Thermal Signatures: The onset of plating alters local heat generation rates. Correlating subtle changes in impedance with differential temperature readings could improve detection reliability.
- EIS + Acoustic Signals: Lithium plating and stripping have distinct acoustic signatures. Fusion with in-operando ultrasonic measurements could offer a spatially resolved detection capability.
The data fusion framework, potentially employing machine learning classifiers trained on multi-dimensional datasets, will be key to developing failsafe plating detection algorithms.
2. From Qualitative Detection to Quantitative Estimation: Most current methods answer the binary question: “Has plating started?” The next critical challenge is to answer: “How much lithium has plated, and at what rate?” Quantitative estimation is essential for predicting the remaining useful life and assessing the severity of a safety risk. Advanced EIS analysis, particularly physics-based modeling coupled with DRT or complex nonlinear least squares (CNLS) fitting of time-resolved spectra, may enable the deconvolution of the current contribution from the parallel plating reaction. Establishing correlations between specific impedance features (e.g., the magnitude of the \( R_{SEI} \) increase in DRT, or the shift in a low-frequency diffusion-related parameter) and the quantity of plated lithium measured via post-mortem analysis (e.g., titration) is a vital research direction. The ultimate goal is to develop models that can not only detect but also quantify the extent of lithium plating in a lithium-ion battery in real-time, enabling truly predictive health and safety management.
In conclusion, EIS-based methodologies have matured from a laboratory analysis tool into a cornerstone for advanced in-operando diagnosis of lithium plating in lithium-ion batteries. While static EIS remains valuable for post-analysis, the future of operational battery management lies with dynamic techniques like DEIS and the hardware-efficient current interruption method. Overcoming the challenges of integration, cost, and data interpretation will pave the way for their widespread adoption. By progressing towards multi-signal fusion and quantitative estimation, these techniques will be instrumental in unlocking the full potential of fast-charging lithium-ion battery technologies while guaranteeing their safety and longevity, thereby accelerating the global transition to sustainable electrification.