Student or educator? Sign up with your school email and get 6 months of Pro free.
CoChat

The DOI matches its Crossref record; 38 of the 39 references it cites were found in a registry or an index; 8 of the 26 sources its claims cite were checked for support.

Identifying degradation patterns of lithium ion batteries from impedance spectroscopy using machine learning

Open access

Abstract

Forecasting the state of health and remaining useful life of Li-ion batteries is an unsolved challenge that limits technologies such as consumer electronics and electric vehicles. Here, we build an accurate battery forecasting system by combining electrochemical impedance spectroscopy (EIS)-a real-time, non-invasive and information-rich measurement that is hitherto underused in battery diagnosis-with Gaussian process machine learning. Over 20,000 EIS spectra of commercial Li-ion batteries are collected at different states of health, states of charge and temperatures-the largest dataset to our knowledge of its kind. Our Gaussian process model takes the entire spectrum as input, without further feature engineering, and automatically determines which spectral features predict degradation. Our model accurately predicts the remaining useful life, even without complete knowledge of past operating conditions of the battery. Our results demonstrate the value of EIS signals in battery management systems.

Overview

CoChat Summary

TL;DR

A Gaussian process regression model using full electrochemical impedance spectra estimates Li-ion battery capacity and remaining useful life from a single impedance measurement, trained on over 20,000 EIS spectra from commercial LiCoO2/graphite coin cells. Automatic relevance determination shows low-frequency impedance features are most predictive, and the model outperforms discharge-curve features in the reported benchmark.

Key findings

  • The EIS-Capacity and EIS-RUL GPR models accurately estimate capacity and predict RUL from the current cycle’s EIS alone; for 25 °C testing cells, the end-of-life values were 150, 120, 30, and 38 cycles, and for multi-temperature testing cells they were 150, 252, and 396 cycles.
  • ARD identified low-frequency impedance as most correlated with degradation: at 25 °C the largest weights were on the 91st and 100th features, corresponding to 17.80 and 2.16 Hz; at 35 °C and 45 °C the largest weight was on the 91st feature, 17.80 Hz.
  • A multi-temperature model estimated capacity and predicted RUL for cells cycled at 35 °C and 45 °C without the cycling temperature as an input, under the simplification that temperature is constant over cycles.
  • Under the same GPR method and train-test split, the EIS-based approach achieved lower predictive error than features extracted from the discharging curve (Supplementary Table 1).

Methods

  • Twelve commercially available 45 mAh Eunicell LR2032 Li-ion coin cells with LiCoO2/graphite chemistry were cycled in climate chambers at 25 °C (25C01–25C08), 35 °C (35C01–35C02), and 45 °C (45C01–45C02); each cycle used 1C CC–CV charge to 4.2 V and 2C CC discharge to 3 V.
  • EIS was measured every even-numbered cycle at nine charging/discharging stages over 0.02 Hz–20 kHz with 5 mA excitation after 15-min open circuit at SoC 0% and 100%; capacity loss was measured every odd cycle, and EoL was defined as capacity below 80% of initial after 30 room-temperature cycles.
  • GPR used the real and imaginary impedance at 60 frequencies as inputs; capacity models used a squared exponential covariance with ARD, RUL models used a linear covariance, and feature importance was defined as \(w_m = \exp(-\sigma_m)\).

Limitations

  • The paper states that all testing cells were charged and discharged the same way as training cells, and that estimating cells cycled at different operating charge/discharge rates requires further experiments.
  • The paper states that more realistic settings—variations in cycling temperature over time or charge/discharge rate—would require a significantly larger training set, and the multi-temperature model assumes future operating temperature is close to previous operating temperature.
  • Evident from the text: the dataset comprises only 12 small coin cells of one chemistry under controlled laboratory temperatures and constant rates, so validation on larger-format cells or real-world variable conditions is not reported.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.1038/s41467-020-15235-7

Cites

38 of 38 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Figures

Fig. 1 Estimating battery capacity. a Estimated (red curve) and measured (blue curve) capacity as a function of cycle number for the 25C05 cell. The coefficient of determination (R2) of this model is shown on the left bottom. b The measured capacity against the estimated capacity of all four testing cells cycled at 25 °C. The capacity is normalised against the starting capacity in each case. c ARD shows that the impedance at low frequency is most correlated with degradation. The pink points correspond to the 120 frequencies in the range of 0.02 Hz–20 kHz. The GPR model assigns the largest weights to the 91st and 100th features, corresponding to 17.80 and 2.16 Hz, respectively. The less relevant features have weights close to zero.
Fig. 1 Estimating battery capacity. a Estimated (red curve) and measured (blue curve) capacity as a function of cycle number for the 25C05 cell. The coefficient of determination (R2) of this model is shown on the left bottom. b The measured capacity against the estimated capacity of all four testing cells cycled at 25 °C. The capacity is normalised against the starting capacity in each case. c ARD shows that the impedance at low frequency is most correlated with degradation. The pink points correspond to the 120 frequencies in the range of 0.02 Hz–20 kHz. The GPR model assigns the largest weights to the 91st and 100th features, corresponding to 17.80 and 2.16 Hz, respectively. The less relevant features have weights close to zero.CC BY · PubMed Central
Fig. 2 Predicting remaining useful life. The predicted RUL of 25C05–25C08 testing cells a–d cycled at 25 °C (shown as green curves). The end of life (EoL) of these four testing cells is 150, 120, 30 and 38, respectively, which is defined as the cycle number when the capacity drops below its initial 80%. The testing EIS spectra are collected at the state V (15 min resting after fully charging). The shaded region indicates ±1 standard deviation. R2 is shown on the right bottom in each panel.
Fig. 2 Predicting remaining useful life. The predicted RUL of 25C05–25C08 testing cells a–d cycled at 25 °C (shown as green curves). The end of life (EoL) of these four testing cells is 150, 120, 30 and 38, respectively, which is defined as the cycle number when the capacity drops below its initial 80%. The testing EIS spectra are collected at the state V (15 min resting after fully charging). The shaded region indicates ±1 standard deviation. R2 is shown on the right bottom in each panel.CC BY · PubMed Central
Fig. 3 Capacity estimation without definite knowledge of the temperature. The curves show the estimated (red) and measured (blue) capacity for cell cycled at a 35 °C and b 45 °C; the cycling temperature is not an input to our model. The shaded area indicates ±1 standard deviation. Both EIS spectra are collected at the state V (15 min resting after fully charging). R2 is shown on the left bottom in each panel. ARD shows that the EIS-capacity GPR models at 35 °C c and 45 °C d assign the largest weight to the 91st feature with the corresponding frequency of 17.80 Hz. The irrelevant features have weights close to zero.
Fig. 3 Capacity estimation without definite knowledge of the temperature. The curves show the estimated (red) and measured (blue) capacity for cell cycled at a 35 °C and b 45 °C; the cycling temperature is not an input to our model. The shaded area indicates ±1 standard deviation. Both EIS spectra are collected at the state V (15 min resting after fully charging). R2 is shown on the left bottom in each panel. ARD shows that the EIS-capacity GPR models at 35 °C c and 45 °C d assign the largest weight to the 91st feature with the corresponding frequency of 17.80 Hz. The irrelevant features have weights close to zero.CC BY · PubMed Central
Fig. 4 Remaining useful life prediction without definite knowledge of the temperature. Our model accurately predicts the RUL of three testing cells a–c cycled at 25, 35 and 45 °C, respectively, without information on the cycling temperature. The EoL of these three testing cells is 150, 252 and 396, respectively, which is defined as the cycle number when the capacity reaches its initial 80%. The testing EIS spectra are collected at state V (15 min resting after fully charging). The shaded region indicates ±1 standard deviation. R2 of our method is shown on the right bottom in each panel.
Fig. 4 Remaining useful life prediction without definite knowledge of the temperature. Our model accurately predicts the RUL of three testing cells a–c cycled at 25, 35 and 45 °C, respectively, without information on the cycling temperature. The EoL of these three testing cells is 150, 252 and 396, respectively, which is defined as the cycle number when the capacity reaches its initial 80%. The testing EIS spectra are collected at state V (15 min resting after fully charging). The shaded region indicates ±1 standard deviation. R2 of our method is shown on the right bottom in each panel.CC BY · PubMed Central

Verification

Claims the paper makes

3 Valid · 4 Not valid · 3 Not checked · 33 Own results and interpretations

25 more claims

Create a free account to see all 43 claims and every source verdict.

See all claims — free account
  • Probably not a substantial claim. 8
  • It may state two separate claims as one. 8
  • The quote only partly states this claim. 1

Create a free account to see these claims.

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

Each contradiction between two passages listed here was confirmed by an independent check.

1 of 1

Citations

Cites

Show all 39 references

Research works printed but not linked: 1

  1. Popp, H., Einhorn, M. & Conte, F. V. Capacity decrease vs. impedance increase of lithium batteries. a comparative study. In Hybrid and Fuel Cell Electric Vehicle Symposium & Exhibition EVS26, Los Angeles, 6–9 (2012).

Cited by

See all 889

Metadata

Authors
Zhang, Y., Tang, Q., Zhang, Y., Wang, J., Stimming, U., & Lee, A. A.
Affiliations
  • University of Cambridge (GB)
  • The Faraday Institution (GB)
Corresponding author
  • Stimming, Ulrich
  • Lee, Alpha A.
Venue
Nature Communications
Published
2020-04-06
Volume
11
Issue
1
Pages
1706-1706
ISSN
2041-1723 (ISSN-L)
OpenAlex
W3010408268
Cited by
889
References
39
Citations by year
2026: 156 · 2025: 198 · 2024: 182 · 2023: 142 · 2022: 118 · 2021: 75 · 2020: 18
FWCI
32.63
Type
journal article
Language
en
Open access
Licence
Locations
  • Nature Communications (publishedVersion, cc-by)
  • PubMed Central (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • Apollo (University of Cambridge) (submittedVersion)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
  • Apollo (University of Cambridge) (submittedVersion, cc-by)
  • Apollo
Funders
RCUK | Engineering and Physical Sciences Research Council (EP/S003053/1)
Topics
  • Physical Sciences › Engineering › Automotive Engineering › Battery Modeling and Management
  • Physical Sciences › Engineering › Electrical and Electronic Engineering › Advancements in Battery Materials
  • Physical Sciences › Engineering › Safety, Risk, Reliability and Quality › Reliability and Maintenance Optimization
Keywords
lithium-ion batteries, battery degradation, electrochemical impedance spectroscopy, remaining useful life prediction, Gaussian process regression, battery management systems, electric vehicles, battery diagnostics

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

The claim verdicts this report counts are experimental: their accuracy is not yet measured.

A language model drew the claims from the paper on 2026-10-10, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-10-10 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).

Cited sources are checked with a language model’s help (gpt-6-sol).

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

How to cite

Cite CoChat for the checks on this page, and the paper for its own findings.

This page
CoChat. (2026). Citation check: Identifying degradation patterns of lithium ion batteries from impedance spectroscopy using machine learning. https://app.cochat.ai/papers/10.1038/s41467-020-15235-7
The paper
Zhang, Y., Tang, Q., Zhang, Y., Wang, J., Stimming, U., & Lee, A. A. (2020). Identifying degradation patterns of lithium ion batteries from impedance spectroscopy using machine learning. Nature Communications. https://doi.org/10.1038/s41467-020-15235-7

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.