The accelerating retirement of electric vehicles has created a critical need for efficient and safe methods to manage end-of-life lithium-ion batteries. Hierarchical recycling, which involves sorting and repurposing batteries based on their residual health, presents a sustainable solution with significant economic and environmental potential. Conventional sorting methods largely rely on electrical parameters such as capacity and internal resistance. However, these macro-scale measurements are insufficient for identifying localized internal defects, material inhomogeneity, or incipient failure modes that are not yet reflected in overall electrical performance. Such internal inconsistencies can severely impact the safety and longevity of repurposed lithium-ion battery packs. Therefore, a non-destructive, spatially resolved evaluation technique is essential for reliable hierarchical sorting. This study investigates and implements ultrasonic bulk wave technology as a powerful tool for the internal inspection and uniformity assessment of prismatic aluminum-shell lithium-ion battery cells, establishing a foundation for their accurate classification for secondary use.
Ultrasonic testing, a well-established non-destructive evaluation (NDE) technique, has recently emerged as a promising modality for lithium-ion battery characterization. The underlying principle is that mechanical waves propagating through a multi-layered structure are sensitive to changes in the material’s elastic properties, density, and internal geometry. Within a lithium-ion battery, key parameters such as the state of charge (SOC) and state of health (SOH) are intrinsically linked to the physical state of the electrodes. During cycling, lithium-ion intercalation and de-intercalation cause expansion and contraction of active materials, altering their elastic moduli and the porosity of the electrode stack. Furthermore, degradation mechanisms like lithium plating, gas generation, and electrode delamination introduce discontinuities that significantly affect ultrasonic wave propagation. Consequently, measurable ultrasonic features, including time-of-flight (TOF), signal amplitude, attenuation, and wave velocity, can serve as effective proxies for the internal state and uniformity of a lithium-ion battery.
Previous research has demonstrated correlations between ultrasonic signals and SOC in laboratory settings. However, the practical application for sorting heterogeneous, retired lithium-ion battery cells under complex, non-ideal conditions remains less explored. This work addresses this gap by systematically studying ultrasonic bulk wave behavior under simulated real-world operational stresses and subsequently deploying a multi-element array system for the high-throughput, uniformity-based sorting of commercial prismatic cells.
1. Fundamentals of Ultrasonic Wave Interaction with Lithium-Ion Battery Structure
A typical prismatic aluminum-shell lithium-ion battery is a complex multi-layered structure. From the exterior to the interior, it generally consists of: an aluminum casing, the electrode assembly (comprising alternating layers of anode, separator, and cathode coated on metal foils), and electrolyte filling the pores. When an ultrasonic longitudinal wave (bulk wave) is transmitted through the thickness of the cell, it interacts with each layer. The propagation characteristics are governed by the acoustic impedance, \( Z \), of each material, defined as:
$$ Z = \rho v $$
where \( \rho \) is the density and \( v \) is the longitudinal wave velocity. The reflection and transmission coefficients at each interface depend on the impedance mismatch. The overall transmitted signal amplitude, \( A_t \), through a multi-layered medium can be modeled as a function of the transmission coefficients at each interface and the attenuation within each layer:
$$ A_t(f) = A_0(f) \cdot \prod_{i=1}^{N} T_i(f) \cdot \exp\left(-\sum_{j=1}^{M} \alpha_j(f) d_j\right) $$
where \( A_0(f) \) is the initial amplitude at frequency \( f \), \( T_i \) are the transmission coefficients, \( \alpha_j \) are the frequency-dependent attenuation coefficients for the \( j \)-th layer, and \( d_j \) are the layer thicknesses.
Changes within the lithium-ion battery, such as electrolyte depletion, electrode porosity change, or delamination, will alter \( \rho \), \( v \), and \( \alpha \) for the affected layers. This, in turn, modifies the final transmitted signal’s TOF and amplitude. For instance, electrode expansion during charging typically increases the overall acoustic path length and may change the wave velocity, potentially increasing TOF. Simultaneously, better contact between layers or changes in porosity might affect signal attenuation. Therefore, monitoring these ultrasonic parameters provides a direct, non-invasive window into the internal physical state of the lithium-ion battery.
2. Experimental Investigation of Ultrasonic Bulk Waves Under Complex Operating Conditions
2.1. Single-Element Pitch-Catch Experimental System
To understand the fundamental response of ultrasonic bulk waves to internal changes in a lithium-ion battery, a controlled single-element pitch-catch experiment was established. The core system consisted of a function generator, a digital oscilloscope, a programmable battery cycler, and a temperature chamber. Two piezoelectric ceramic transducers were coupled to opposite large faces of a commercial prismatic aluminum-shell lithium-ion battery (nominal capacity: 50 Ah) using ultrasonic couplant. One transducer acted as the transmitter, and the other as the receiver.
The excitation signal was a 5-cycle Hanning-windowed sinusoidal toneburst. A central frequency of 160 kHz was selected after preliminary analysis to ensure sufficient penetration through the battery thickness while providing good signal-to-noise ratio and time resolution. The battery was subjected to various charge-discharge cycles at rates of 0.2C, 0.5C, 1.0C, and 2.0C within a temperature-controlled environment. Ultrasonic signals and temperature data were acquired at regular intervals (e.g., every 30 seconds for ultrasound, every 60 seconds for temperature) throughout the cycles.
2.2. Results and Analysis: Signal Response to Electrical and Thermal Stress
The experiments revealed clear and consistent trends in the received signal amplitude. During low-rate (0.2C, 0.5C) charge-discharge cycles at a constant ambient temperature (25°C), the signal amplitude exhibited a symmetric evolution: it increased monotonically during charging and decreased monotonically during discharging. This pattern correlates directly with the lithiation/delithiation-induced physical changes in the electrode stack. The temperature remained relatively stable during these low-rate tests, confirming that the amplitude change was primarily driven by SOC-dependent mechanical property changes within the lithium-ion battery.
In contrast, high-rate (1.0C, 2.0C) cycling and tests involving step changes in environmental temperature introduced a more complex interaction. The signal amplitude showed a strong inverse correlation with cell temperature. As the lithium-ion battery temperature rose due to high-rate operation or external heating, the received amplitude decreased significantly. When the temperature stabilized or decreased, the amplitude recovered or increased accordingly. This demonstrates that temperature is a dominant factor influencing ultrasonic propagation, likely due to changes in the viscoelastic properties of polymeric components (separator, binder) and the coupling between layers. This finding underscores the necessity of compensating for or accounting for thermal effects when using ultrasound for SOC or SOH estimation in real applications.
2.3. Signal Analysis via Cross-Wavelet Transform
To quantitatively validate the chosen excitation frequency and the consistency of the signal’s time-frequency characteristics during charge and discharge, a cross-wavelet transform (XWT) analysis was performed on signal pairs from the 0.2C and 0.5C cycles. The XWT identifies regions in the time-frequency domain where two signals have high common power and reveals their phase relationship.
The XWT spectra for multiple charge-discharge signal pairs consistently showed a region of high common power, bounded by the 95% confidence level (indicated by a thick black contour), within the time window of 100–200 µs and at frequencies below 160 kHz. The arrows within this significant region predominantly pointed right, indicating a positive correlation between the charge and discharge signals at these scales. This high degree of coherence in the target time-frequency region confirms that the 160 kHz excitation effectively probes the consistent, SOC-dependent physical changes inside the lithium-ion battery, justifying its use for further experiments.
3. Hierarchical Sorting of Retired Batteries via Multi-Element Array Uniformity Assessment
3.1. Design of Multi-Element Array Fixture and Testing System
While single-point measurement can track global changes, it is inadequate for assessing the spatial uniformity of a large-format prismatic cell—a critical factor for reliable hierarchical sorting. Localized defects like electrode wrinkling, uneven electrolyte distribution, or delamination may not significantly affect a single-point measurement but can be disastrous in a repurposed pack. To address this, a custom multi-element array fixture was designed and fabricated.
The fixture comprised two aligned plates, each with a 5×7 grid of holes (35 positions total) to hold piezoelectric transducers. The plates were connected via adjustable springs, ensuring consistent and uniform pressure across the battery surface while maintaining alignment of opposing transmitter-receiver pairs. This setup allowed for a full matrix of pitch-catch measurements, effectively performing an ultrasonic C-scan of the battery’s internal condition.
3.2. Single-Element Screening of a Batch with Known Capacity
An initial experiment was conducted on 33 retired prismatic aluminum-shell lithium-ion battery cells with known, pre-measured electrical capacities ranging from approximately 27.6 Ah to 19.7 Ah. All cells were first discharged to a uniform cut-off voltage. A single pair of transducers was then used to perform a pitch-catch measurement on each cell. The results, summarized in the table below, showed a clear stratification of the received signal amplitude into three distinct groups that correlated well with the known capacity bands.
| Capacity Group | Capacity Range (Ah) | Typical Signal Amplitude Range (V) | Interpretation |
|---|---|---|---|
| Group 1 (Cells 1-16) | 26.46 – 27.60 | 1.5 – 2.2 | Higher health, suitable for cascade use. |
| Group 2 (Cells 17-25) | 25.03 – 25.59 | 0.3 – 0.6 | Moderate health, potential capacity “virtual labeling,” requires scrutiny. |
| Group 3 (Cells 26-33) | 19.72 – 20.77 | < 0.05 | Low health, recommended for recycling. |
This preliminary screening confirmed that ultrasonic amplitude broadly correlates with the overall health state of a lithium-ion battery. More importantly, it highlighted that Group 1 and Group 2, while having somewhat similar electrical capacities, exhibited vastly different ultrasonic responses, suggesting significant internal differences not captured by capacity alone.
3.3. Multi-Element Uniformity Analysis for Precise Sorting
To move beyond global assessment and detect localized inhomogeneity, all 33 cells were subsequently evaluated using the full 35-element array fixture. A C-scan image was generated for each cell based on the fast wave signal amplitude received at each grid point. Visual inspection of these C-scans revealed significant spatial variations in signal amplitude across the surface of many cells, particularly those in Groups 2 and 3. This confirmed that a single-point measurement is insufficient for reliable grading.
To quantify internal uniformity, a statistical metric was calculated for each cell: the variance of the signal amplitude across all measurement points. A low variance indicates homogeneous internal material properties and good structural integrity, while a high variance suggests localized defects or severe inhomogeneity. The calculated uniformity variance for all cells is presented below.
| Cell Group | Cell IDs | Uniformity Variance Range | Classification Decision |
|---|---|---|---|
| Group A (High Uniformity) | 1-11, 13-15 | 0.004 – 0.042 (Most below 0.05) | In-Service / Prime for Cascade Use. Exhibits high capacity and excellent internal uniformity. |
| Group B (Low Uniformity) | 12, 16, 17-25 | 0.154 – 0.600 | Nearing Retirement / Scrutiny Required. Even if capacity is moderate, high variance indicates internal defects (e.g., delamination, drying). Unsuitable for high-demand repurposing. |
| Group C (Moderate Uniformity, Low Signal) | 26-33 | 0.007 – 0.145 (Most below 0.1) | Retired / For Recycling. Very low global signal amplitude confirms severe degradation. Uniformity is moderate but irrelevant due to poor overall health. |
The multi-element uniformity analysis provided a far more nuanced and reliable sorting criterion than either capacity or single-point amplitude alone. It successfully identified specific cells within the higher-capacity batch (e.g., Cells 12 and 16) that exhibited high variance, flagging them as potentially faulty despite their reasonable electrical performance. This demonstrates the critical advantage of ultrasonic imaging: it can detect localized failures that pose safety risks and would lead to premature failure in a second-life lithium-ion battery pack.
4. Discussion and Technological Implications
The experimental results solidify the premise that ultrasonic bulk wave interrogation is a powerful tool for the non-destructive evaluation of prismatic lithium-ion battery cells. The technology’s sensitivity to both global state changes (SOC, overall SOH) and local material inhomogeneity makes it uniquely suited for the hierarchical sorting challenge. The multi-element array approach transforms the technique from a laboratory probe into a practical inspection system capable of high-throughput screening.
The sorting logic derived from this study can be formalized into a decision algorithm. Let \( A_{avg} \) be the average transmitted amplitude from a single-point screening, \( \sigma^2 \) be the spatial variance of amplitude from the array scan, and \( C \) be the measured capacity. A multi-parameter decision boundary can be established:
- IF \( A_{avg} > A_{threshold1} \) AND \( \sigma^2 < \sigma^{2}_{threshold1} \) → Classify as “Prime for Cascade Use.”
- IF \( A_{avg} \) is moderate BUT \( \sigma^2 > \sigma^{2}_{threshold2} \) → Classify as “Defective / For Recycling.”
- IF \( A_{avg} < A_{threshold2} \) → Classify as “Degraded / For Recycling,” regardless of \( \sigma^2 \).
Future work will focus on automating this classification process using machine learning models trained on a larger database of ultrasonic C-scans from lithium-ion battery cells with known degradation histories. Furthermore, integrating temperature compensation models based on the observed inverse relationship will enhance the robustness of measurements performed on batteries at non-standard temperatures. The ultimate goal is to develop a fully automated, inline inspection station that can rapidly assess retired lithium-ion battery cells, generating a “state-of-uniformity” report to guide optimal sorting for second-life applications or direct recycling.
5. Conclusion
This study comprehensively investigated the application of ultrasonic bulk waves for the internal characterization and hierarchical sorting of prismatic aluminum-shell lithium-ion battery cells. First, the propagation behavior of ultrasonic waves was experimentally characterized under complex charge-discharge cycles and thermal variations, confirming their sensitivity to internal physicochemical changes and identifying temperature as a key interfering factor that must be managed. Subsequently, a novel multi-element ultrasonic array system was designed and implemented. By quantifying the spatial variance of the transmitted ultrasonic signal across the battery surface, this system provided a direct measure of internal material uniformity—a critical parameter not accessible via traditional electrical tests.
The experimental results on batches of retired commercial cells demonstrated that the combined analysis of global ultrasonic amplitude and local uniformity variance enables a more accurate and reliable classification than capacity-based sorting alone. Cells were successfully categorized into groups suitable for cascade use, requiring further scrutiny, or destined for direct recycling. This ultrasonic-based approach offers a fast, non-destructive, and information-rich method for quality control in the emerging lithium-ion battery recycling and repurposing industry, laying a solid technological foundation for the safe and economical implementation of a circular economy for energy storage.