Influence of simulated methanol reformate impurities (CO₂, CO, H₂O) on HT PEMFC performance

The effects of water vapor and chlorine on the Si1−xGex epitaxial growth are studied through thermodynamic analyses. Since Ge does not form chemical species with oxygen at an appreciable level, the water vapor and oxygen background level requirements for Si1−xGex epitaxial growth are mostly similar to the ones for Si growth. At large excesses of Ge gas phase sources and low system water vapor level, however, the critical temperature above which oxide-free epitaxial growth occurs decreases with increasing system pressure. Analyses of the results obtained suggest that there are two mechanisms for the removal of SiO2(s) from the surface: one is by forming volatile SiO, and the other is by forming gas phase H2O. The former one appears to dominate for most conditions studied; the latter one appears to dominate when the water vapor level of the system is very low and/or small amounts of the Si gaseous source are present. The boundary between etching and deposition of Si1−xGex films is examined for all Ge solid phase compositions. The etch/deposition boundaries for pure Si and pure Ge show opposite trends with increasing chlorine concentrations, because of the different affinities of Si and Ge to form chlorinated species. The chlorine concentration needed to reach etching increases with increasing system temperature for Si, whereas it decreases with increasing temperature for Ge. It is found that the formation of SiCl4 is favored over GeCl2 at low temperatures, while the opposite happens at higher temperatures. The etch/deposition boundary for Si1−xGex thin films is found to form a saddlelike contour because of the interaction of the different behaviors of Si and Ge. Overall, calculated results are seen to be in good agreement with available experimental results.

The importance of hydrogen production by proton exchange membrane water electrolysis (PEMWE) in the energy sector makes it worthwhile to understand the system behaviour in failure cases and possible recovery procedures. In particular, insufficient feed water supply to the anode can be caused by inhomogeneous distribution in the stack, pump failure or erroneous operation strategy. For the PEMWE, a distinct turnover point has been reported if feed water supply is reduced in isothermal conditions.[1] Further reduction of the water supply results in higher ohmic resistance due to gradual membrane dehydration and increased mass transport losses.[2] Here, we will use the term “water starvation” to describe this event. The drainage of water from the anode can be explained by oxygen evolution reaction (OER), humidification of oxygen gas and water crossover to the cathode. The last one was measured for well hydrated membranes at atmospheric pressure [3] or below typical operating temperatures [4].In the first part of this work, we present the water crossover in a PEMWE on single cell level to understand their dependence on the operating parameters such as temperature and pressure. Media were supplied and conditioned to a 5 cm2 single cell by using an in-house test bench and a potentiostate with booster. The electrochemical measurements were performed with a commercially available catalyst coated perfluorosulfonic acid membranes (loading of 0.3 mg/cm2 geo Pt/C on cathode and 1 mg/ cm2 geo Ir-Oxide on anode). The anode was kept at atmospheric pressure, whereas the cathode pressure was varied from atmospheric to 400 kPaabs at various current densities and cell temperatures of 40 – 80 °C. To determine the water crossover under several operating conditions, the water vapour was condensed to liquid phase in a heat exchanger and measured by a gravimetric method.Furthermore, the measured water crossover is used to force a water starvation in the second part of this work. After cell assembly, a conditioning procedure was performed until steady cell performance was achieved. Cell performance was evaluated using polarization curve, cyclic voltammetry (CV), electrochemical impedance spectroscopy (EIS) and water crossover. The last two are of particular interest, because they provide information about membrane humidification. The Tafel slope and mass transport overvoltage were evaluated from the polarization curves and high frequency resistance (HFR). Water starvation experiments were then carried out at a constant cell voltage by reducing the feed water supply, resulting in a dramatic decay of the current density. The analysis of Tafel slope and EIS data reveals an increase of ohmic and mass transport resistances. Furthermore, the HFR measurements were correlated with water crossover as a function of membrane hydration. We observed that the specific water crossover per current density decreases as a consequence of membrane dehydration. After this step, feed water supply was increased to an unrestricting level for membrane rehydration and performance recovery. For this recovery procedure, different operation modes were employed by varying the current densities. At constant current density, membrane hydration is influenced by the ionic current from the cathode to the anode via proton drag. EIS and water crossover measurements allow us to determine the influence of operation on the dynamics of the membrane rehydration process. Together with the polarization curves, cyclic voltammetry and HFR data, reversible and irreversible degradation processes caused by the water starvation can be identified.Altogether, we present a deeper understanding of water crossover for a wide operating parameter range and provide a guideline for recovery of the performance for PEMWE.[1] Christoph Immerz et al 2020 Meet. Abstr. MA2020-02 2456[2] C. Immerz, B. Bensmann, P. Trinke, M. Suermann, R. Hanke-Rauschenbach, J. Electrochem. Soc. 2018, 165, F1292-F1299.[3] K. Onda, T. Murakami, T. Hikosaka, M. Kobayashi, R. Notu, K. Ito, J. Electrochem. Soc. 2002, 149, A1069.[4] P. Medina, M. Santarelli, International Journal of Hydrogen Energy 2010, 35, 5173.

Hydrogen can be produced using renewable energy sources through alkaline water electrolysis (AWE). Alkaline water electrolyzers are made of inexpensive materials such as nickel perforated electrodes and the Zirfon porous diaphragm. However, one of the bottlenecks of this technology is the high internal ohmic resistance compared to other technologies, which implies that the technology is traditionally operated at low current densities. To decrease the ohmic resistance advanced AWE electrolyzers use a zero-gap configuration. However, even with such a zero-gap configuration the ohmic resistance remains relatively high.An analysis of polarization data on zero-gap alkaline electrolyzers with the commonly used Zirfon diaphragm suggests that the ohmic area resistances are significantly higher than what would be expected based on the properties of the diaphragm supplier [1]. In this study we use Electrochemical Impedance Spectroscopy (EIS) to further investigate this phenomenon [2], since EIS enables the separation of ohmic and charge transfer resistances.EIS in combination with polarization measurements enables us to determine electrodes overpotentials, diaphragm ohmic losses and bubbles resistance and to estimate key parameters such as the reversible cell potential, the Tafel slopes and the exchange current densities of hydrogen and oxygen evolution reactions. Impedance data were fitted with an equivalent electric circuit, which consist of an ohmic resistance connected in series with two components, each including a resistance and a constant phase element (CPE) in parallel. Besides, the polarization measurements were fitted to a given correlation [1] using MATLAB as shown in Figure 1.It was found that the ohmic resistance determined with EIS (~0.48 Ω.cm²) is lower than the fitted area resistance from the I-V curve (0.68 Ω.cm²), but is still higher than what would be expected based on the diaphragm properties for Zirfon. Obtained values for Tafel slopes and exchange current densities, both by EIS and I-V curves, are comparable to typical values from the literature. The resistance does not increase with current density suggesting that an increase of bubble volume with current density does not play a significant role in the resistance. Therefore, it still remains unclear what the origin is of the increased ohmic resistance.[1] M.T. de Groot, A.W. Vreman. Ohmic resistance in zero gap alkaline electrolysis with a Zirfon diaphragm. Electrochimica Acta, 2021, 369, 137684. https://doi.org/10.1016/j.electacta.2020.137684[2] Rodríguez, J.; Palmas, S.; Sánchez-Molina, M.; Amores, E.; Mais, L.; Campana, R. Simple and Precise Approach for Determination of Ohmic Contribution of Diaphragms in Alkaline Water Electrolysis. Membranes, 2019, 9, 129. https://doi.org/10.3390/membranes9100129 Figure 1

Electrochemical impedance spectroscopy (EIS) is one of the most important non-destructive diagnosis tools in lithium-ion battery (LIB) technology. EIS is routinely used for characterization of battery materials, optimization of manufacturing processes, and understanding of degradation mode. However, interpretation of EIS spectra is not a straightforward procedure. Typical EIS data of LIB have heavily overwrapped capacitive semicircles in Nyquist (complex impedance) plot because LIBs’ anodes and cathodes both involve several conduction processes. Widely used method for information separation is least-squares fitting based on discrete equivalent circuit model. However, this method requires a priori assumption on the number of capacitive processes. Destructive EIS measurement methods, such as insertion of reference electrode and symmetry cell construction, are powerful for separation of cathode/anode information. But these involve meticulous work, and battery structure or degradation state often make it difficult to obtain an accurate measurement result.Recently, distribution of relaxation times (DRT) analysis has increasingly used for EIS interpretation. This analysis treats an EIS spectrum as a continuous function of relaxation time, or time constant of infinitesimal RC equivalent circuit element. DRT analysis can distinguish several different capacitive processes without the need of prior model assumption. Application field of DRT analysis currently focuses on solid oxide fuel cell, but several LIB studies have been reported for LiFePO4 cathode system[1] and LiMn2O4-LiNixCoyMnzO2 cathode system[2]. Here we demonstrate that DRT analysis is a powerful tool for degradation analysis of LIBs with LiCoO2(LCO)-based cathode and graphite anode, and provide experimental verification of information obtained by DRT analysis.Commercial-size prototype LIBs with LCO-based cathode and graphite anode are used for this work. Charge-discharge cycling tests of the cells are performed at different temperature conditions. EIS measurements are conducted at fully charged state of every fifty cycles, with frequency range of 1 MHz to 10 mHz. For better separation of different processes, the EIS spectra are obtained at 23 deg C, 10 deg C and 0 deg C in a climate chamber. DRT is calculated by MATLAB with in-house code, which is based on Fourier transfer methods utilizing window function preprocessing[3]. To understand DRT results in physical viewpoints, several electrode analyses of disassembled cells are conducted at initial and the cycled state. Figure 1 shows Nyquist plots and the corresponding DRT plots of (a) low-temperature and (b) high-temperature cycled LIBs obtained at 10 deg C. In Nyquist plot, all spectra exhibit the shape of two depressed capacitive semicircles. At a first glance, both low-temperature cycling and high-temperature cycling appear to increase the size of low-frequency semicircles. But closer look shows that low-temperature cycling cause “merge” or heavier overwrap of two semicircles, whereas high-temperature cycling retains the separated feature. These characteristics are further clarified by DRT analysis. In each cell, we can identify four different peaks in DRT plots. The high-temperature cycled cell shows gradual increase in the lowest-frequency (1) peak, whereas the peak remains almost the same intensity in the low-temperature cycled cell and the middle-frequency (2) peak increases significantly. No change is observed for the (3) peak during the cycle test. Both cycling condition cause the increase in the highest-frequency (4) peak. Further teardown analyses verify that these DRT peaks can be attributed to different electrode degradation phenomena, which are related to cycling temperature. These results demonstrate the ability of DRT analysis as a degradation diagnostics tool for LIBs with various kinds of material systems. [1] J. P. Schmidt, T. Chrobak, M. Ender, J. Illig, D. Klotz, E. Ivers-Tiffée, J. Power Sources, 196 5342 (2011). [2] B. Stiaszny, J. Ziegler, E. Krauß, J. P. Schmidt, E. Ivers-Tiffée, J. Power Sources, 251 439 (2014).[3] H. Schichlein, A. C. Müller, M. Voigts, A. Krügel, E. Ivers-Tiffée, J. Appl. Electrochem., 32 875 (2002). Figure 1

Synergic effect investigation of carbon monoxide and other compositions on the high temperature proton exchange membrane fuel cell

Quantifying the relationship between operating conditions and individual electrode behavior is critical for advancing polymer electrolyte membrane fuel cell (PEMFC) technology. Here, a platinum wire embedded within the Nafion membrane was used as a voltage probe to quantify the polarization contributions of individual cell components under varying temperature and humidity conditions. Electrochemical impedance spectroscopy (EIS) and distribution of relaxation times (DRT) analysis, combined with dc current-voltage measurements, enabled the deconvolution of ohmic and polarization resistances of each component. The results reveal that, while temperature governs electrode kinetics predominantly, the asymmetric relative humidity significantly influences membrane hydration and interfacial transport. These trends arise from a shift in the dominant water transport mechanism: back diffusion controls membrane hydration at low current density and electro-osmotic drag dominates at high current density. Moreover, a reciprocal electrode interaction was observed, with the cathode-side DRT peaks changing their position and area in response to anode humidity changes, highlighting the coupled polarization within the PEMFC.

Abstract. Measurements of dry air mole fractions of atmospheric greenhouse gases are used in inverse models of atmospheric tracer transport to quantify their sources and sinks. The measurements have to be calibrated to a common scale to avoid bias in the inferred fluxes. For this purpose, the World Meteorological Organization (WMO) has set requirements for the interlaboratory compatibility of atmospheric greenhouse gas (GHG) measurements. A widely used series of devices for these measurements are the GHG analyzers manufactured by Picarro, Inc. These are often operated in humid air, and the effects of water vapor are corrected for in post-processing. Here, we report on rarely detected and previously unexplained biases of the water correction method for CO2 and CH4 in the literature. They are largest at water vapor mole fractions below 0.5 % H2O, which were undersampled in previous studies, and can therefore affect measurements obtained in humid air. Setups that dry sample air using Nafion membranes may be affected as well if there are differences in residual water vapor levels between sample and calibration air. The biases are caused by a sensitivity of the pressure in the measurement cavity to water vapor. We correct these biases by modifying the water correction method from the literature. Our method relies on experiments that maintain stable water vapor levels to allow equilibration of cavity pressure. In our experiments with the commonly used droplet method, this requirement was not fulfilled. Correcting CO2 measurements proved challenging, presumably because of our humidification method. Open questions pertain to differences among analyzers and variability over time. In our experiments, the biases amounted to considerable fractions of the WMO interlaboratory compatibility goals. Since measurements of dry air mole fractions of CO2 and CH4 are also subject to other uncertainties, correcting the cavity pressure-related biases helps keep the overall accuracy of measurements obtained with Picarro GHG analyzers in humid and potentially in Nafion-dried air within the WMO goals.

Anion exchange membrane water electrolysis (AEM-WE) stands out as a promising method for hydrogen production from renewable energy sources. Unlike proton exchange membrane water electrolysis (PEM-WE), AEM-WE offers the advantage of nonprecious metal catalysts due to mild alkaline conditions, while still enabling a compact cell design and operation under differential pressure. Remarkable performances of AEM-electrolysis cells have been demonstrated in literature, achieving current densities of up to 7.68 A·cm−2 [1], surpassing the record current density of PEM-WE at 6 A·cm−2 [2], both measured at 80 °C and 2 V. However, challenges related to efficiency and stability persist and demand effective solutions to unlock the full potential of AEM-WE. In order to achieve this, cell components, particularly the membrane, catalysts, and electrode structure have to be improved and advanced measurement techniques have to be applied.Different measurement tools serve to characterize, separate, and quantify the dynamic processes inherent in AEM-WE. Electrochemical impedance spectroscopy (EIS) serves as a valuable non-destructive, in-situ diagnostic tool for electrochemical cells, enabling the differentiation of various phenomena based on their respective relaxation times. Moreover, EIS enables the quantification of loss mechanisms thereby facilitating a targeted optimization of electrolysis components and the identification of weaknesses limiting long-time stability.Despite its utility, the broad use of EIS is hindered by its complex interpretation. The widely applied approach of fitting equivalent circuit models (ECMs) aims to extract quantitative data from the semi-circular shapes observed in Nyquist plots. However, the assumption of a physico-chemical meaningful ECM is challenging without prior knowledge of the number, size, and time constants of appearing loss mechanisms. Furthermore, the potential for multiple ECMs to fit one and the same experimental result introduces ambiguity, leading to potential misinterpretations of data.An alternative approach to analyse EIS spectra involves converting the measured data into the time (τ) domain and characterize the different mechanisms based on their characteristic relaxation times. The distribution of relaxation times (DRT) analysis has already been applied across diverse electrochemical systems, spanning from solid oxide fuel cells (SO-FC) and PEM fuel cells (PEM-FC) to PEM water electrolysis and AEM fuel cells (AEM-FC). However, to the best of the authors’ knowledge, DRT analysis of AEM-WE is still absent in literature.In this study we present a thorough investigation of an AEM-WE single-cell, employing a combination of electrochemical impedance spectroscopy (EIS) and the equivalent circuit model (ECM). Notably, we demonstrate the distribution of relaxation times (DRT) analysis at AEM-WE cells for the first time, utilizing a reversible hydrogen electrode (RHE) as a reference. Through half-cell EIS measurements and subsequent DRT spectra analysis, anodic and cathodic half-cell reactions are clearly identified and quantified. The DRT analysis differentiates five loss mechanisms within the AEM-WE system, encompassing the hydrogen evolution reaction, the oxygen evolution reaction, and ionic transport losses within the catalyst layers. By systematically varying operating parameters, we successfully attribute DRT peaks to their respective physicochemical origins. These findings provide valuable insights into the electrochemical processes within the AEM-WE single-cell, significantly advancing our understanding of underlying mechanisms.In fig. 1, we investigate the degradation of AEM-WE single cells through long-term experiments. Periodic EIS measurements enable an in-operando analysis of catalyst degradation. The impedance data is analysed using ECM and DRT. Ohmic contributions resulting from electric and ionic charge transport are distinguished from polarisation resistances arising from electrochemical reactions. A catalyst degradation is observable in fig 1b & c, where the low frequency resistance (LFR) in the Nyquist plot and the DRT peak area increases over time. However, fig. 1b also reveals the reduction of the high frequency resistance (HFR) resulting in an overall performance increase and voltage decrease.[1] N. Chen, S. Y. Paek, J. Y. Lee, J. H. Park, S. Y. Lee, Y. M. Lee, High-performance anion exchange membrane water electrolyzers with a current density of 7.68 a cm −2 and a durability of 1000 hours, Energy & Environmental Science 14 (12) (2021) 6338–6348. doi:10.1039/D1EE02642A.[2] M. Braig, R. Zeis, Distribution of relaxation times analysis of electrochemical hydrogen pump impedance spectra, Journal of Power Sources 576 (2023) 233203. doi:10.1016/j.jpowsour.2023.233203. Figure 1

High temperature oxidation of refractory-grade SiC was carried out in O2 and Ar atmospheres mixed with various concentrations of water vapor, and its oxidation behavior was followed mainly by using a mass-spectrometer. The results were as follows, (1) In the dry and wet O2 atmospheres, main gaseous products in the temperature-raising process up to 1400°C were CO and CO2 formed from the free carbon. Water vapor accelerated the oxidation of free carbon. On the other hand, in the successive soaking process at 1400°C, CO2 formed from SiC was a dominant product. The evolution behavior of CO2 during the soaking process was well fitted with the parabolic law and a positive tendency was observed between the parabolic rate constant and the water vapor content.(2) In the wet Ar atmospheres, CO, CO2 and H2 were evolved due to oxidation of the free carbon by water vapor in the temperature-raising process. The oxidation behavior of SiC in the soaking process changed with the water vapor content. At high water vapor contents, the passive oxidation took place, obeying the parabolic law and the parabolic rate constant increased in proportion to the content of water vapor. At the water vapor content as low as 0.6vol% H2O, however, the initial part of the reaction proceeded according to the active oxidation mechanism obeying the linear law. White spots observed on the surface of the sample after the active oxidation were composed of SiO2 fiber, which was presumably formed by the vapor phase reaction among SiO, CO and H2O.

The effects of hydrogen dilution, carbon monoxide poisoning for a Pt–Ru anode in a proton exchange membrane fuel cell

Introduction Carbon monoxide (CO) in hydrogen fuel is known to degrade a polymer electrolyte fuel cell (PEFC) performance [1-5]. The maximum allowable concentration of CO is 0.2 ppm, according to the quality standards for hydrogen fuel (ISO14687-2) in fuel cell vehicles (FCVs) [6]. Normally, a single-cell evaluation system is a hydrogen one-way pass system, which allows fuel unused by the cell to be released into the atmosphere. However, FCVs usually have a hydrogen circulation system, which returns the fuel from the anode outlet to the inlet. A comparison between the hydrogen one-way pass system and hydrogen circulation system has not been reported with respect to the effect of CO on hydrogen fuel. Our previous study showed that CO is not accumulated in hydrogen fuel when using a hydrogen circulation system [2]. However, the effect of CO on the actual FCV condition has not yet been fully understood because the experimental conditions in our previous study (CO concentration was 4.8 ppm and anode platinum loading was 0.4 mg cm−2) were far from the actual FCV conditions. In this study, using a hydrogen circulation system, the effect of CO on the PEFC’s performance degradation is investigated and is compared to that using a hydrogen one-way pass system under the conditions of low platinum loading at the anode and a low CO concentration. Experimental A JARI standard single cell (25 cm2 electrode area), whose temperature was controlled by a heating medium, and commercially available membrane electrode assemblies (MEA) were used for the single-cell tests. The platinum loading on the anode and the cathode were 0.1 and 0.4 mg cm−2, respectively. The electrolyte membrane thickness was 15 μm. The cell temperature was 60 ºC, the anode was non-humidified, and the cathode dew point was 40 ºC. The single-cell operation tests were conducted using both a conventional hydrogen one-way pass system and the hydrogen circulation system (Fig. 1). The cell was operated at a constant current density of 1000 mA cm−2. The anode gas was hydrogen mixed with CO (up to 1.0 ppm), and the cathode gas was a mixture of N2 (79%) and O2 (21%). The stoichiometry of the fuel and the oxidant was 2 and 2.5, respectively. The CO and CO2 in the purge gas in the hydrogen circulation system were sampled at a 1% purge rate and measured by gas chromatography coupled with pulsed discharge helium ionization detector. Results and Discussion Figure 2 shows, for comparison, the voltage change by adding CO at a steady state in the hydrogen one-way pass system and in the hydrogen circulation system and the CO concentration in the hydrogen circulation system. The voltage change in the hydrogen circulation system is smaller than that in the hydrogen one-way pass system. At 0.2 ppm CO concentration in the hydrogen circulation system, the voltage change induced by adding CO is about 1/10 of that incurred when the hydrogen one-way pass system is used. The CO concentration in the purge gas in the hydrogen circulation system is lower than that in the supplied CO/H2 mixture. Most of the CO in the hydrogen fuel is oxidized to CO2 and gets accumulated in the hydrogen circulation system when a low concentration of CO is supplied to the cell. When the hydrogen circulation system is used, the exhaust gas is sent to the anode inlet again. Consequently, the CO concentration at the anode inlet decreases because the gas in the cylinder, which contains CO, is diluted by the gas from the hydrogen circulation system. In addition, our previous study of the hydrogen one-way pass system revealed that a large excess of O2, compared with CO, existed in the anode [3]. O2 in the hydrogen circulation system is probably circulated and consumed by CO oxidation in the anode electrocatalyst via a non-electrochemical reaction [4]. Consequently, the voltage change in the hydrogen circulation system is smaller than that in the hydrogen one-way pass system. Acknowledgements This work was supported by the New Energy and Industrial Technology Development Organization (NEDO).

High-precision identification of polarization processes of distribution of relaxation times by polarization curve model for proton exchange membrane fuel cell

Electrochemical Impedance Spectroscopy (EIS) is a non-destructive method that provides crucial insights regarding the battery processes. EIS is mainly done by applying a small amplitude AC with a 10 mHz to 1 MHz frequency range. One can determine the electrolyte's ohmic resistance, the Solid Electrolyte Interface (SEI) capacitance, electron transfer rate and the diffusion mechanisms at separate frequency regions from each other.One disadvantage of EIS is the ambiguity of the analyses of the data. The equivalent circuit models used are not unique, and the correspondence between the various elements in the equivalent circuits to electrochemical phenomena are suspect. A new approach on EIS analysis is sought after in order to improve the interpretation. For example, distribution of relaxation times (DRT) was proposed by Zhang et al. in 2015 .Using physics-based models in order to model EIS has been published by Murbach and Schwartz in 2018. In their work, they created a physics-based simulation which uses a pseudo two-dimensional battery model. Their main aim was to present a new approach to EIS data analysis by using a physics-based model rather than equivalent circuit models. They created a data set with 38.800 separate spectra and match the experimental data by using least squares matching approach with the previously computed data. Though this work improved the understanding, multiple calculated data sets matched the experiments with similar errors.Ultimately, in order to achieve an EIS analysis method that is directly related to the fundamental parameters regarding the electrochemistry, degeneracies have to be removed via temperature dependence and the non-linearities as extra dependent variables. This way, the ambiguities and the suspicious parameter assigning can be fixed.We will be reporting on T-dependent linear and nonlinear simulations on 18650 NMC/Graphite battery. All the simulations were done on Python by using a physics-based model PyBaMM (Sulzer, Valentin, et al. "Python battery mathematical modelling (PyBaMM)." Journal of Open Research Software 9.1 (2021).). Impedance data was obtained via simulating voltage as a function of current within 100 µHz to 10 kHz frequency range. Temperature range was set to -25°C to 65°C with 10°C increments. The linear and non-linear impedance is analyzed as a function of temperature. Our results will be compared with experimental temperature dependent linear/nonlinear spectra in order to obtain proper fundamental parameters

Lithium-ion batteries are extensively utilized in modern applications due to their high energy density, long cycle life, and efficiency. With the increasing demand for sustainable energy storage solutions, accurately estimating the State of Health (SOH) is essential to address challenges related to battery degradation and secondary life management. Electrochemical Impedance Spectroscopy (EIS) is a widely used diagnostic tool for evaluating battery performance due to its simplicity and cost-effectiveness. However, EIS often struggles to decouple overlapping electrochemical processes. The Distribution of Relaxation Times (DRT) method has emerged as a powerful alternative, enabling the isolation of key processes, such as ohmic resistance, SEI resistance, charge transfer resistance, and diffusion, thereby providing deeper insights into battery aging mechanisms. This paper presents a novel approach for estimating the State of Health (SOH) of batteries by leveraging DRT parameters across multiple State of Charge (SOC) levels. This study incorporates data from three lithium-ion batteries, each with distinct initial capacities, introducing variability that reflects the natural differences observed in real-world battery performance. By employing a Long Short-Term Memory (LSTM)-based machine learning model, the proposed framework demonstrates a superior accuracy in SOH prediction compared to traditional EIS-based methods. The results highlight the sensitivity of DRT parameters to SOH degradation and validate their effectiveness as reliable indicators for battery health. This research underscores the potential of combining a DRT analysis with AI-driven models to advance scalable, precise, and interpretable battery diagnostics.

The demand for advanced energy storage systems, such as Li-ion batteries, has grown significantly in recent years. However, these batteries pose substantial safety risks if their structural integrity is compromised. Current methods are insufficient for accurately assessing battery safety for continued use in dynamic applications, such as electric vehicles (EVs), or for repurposing in stationary applications, like grid storage. Most prior research has focused on detecting imminent short circuits, but identifying pre-short circuit conditions has proven challenging, as compromised cells often exhibit negligible differences in voltage or capacity compared to intact ones.This study introduces a novel approach to assessing damage in Li-ion batteries by analyzing their dynamic electrical responses. Specifically, we utilize electrochemical impedance spectroscopy (EIS) and characterize the batteries' time constants and polarizations through an innovative formulation of the distribution of relaxation times (DRT). The DRT method converts EIS frequency-domain data into the time-constant domain using the Fredholm integral of the first kind. A key output of our algorithm is the identification of the frequency range and processes most affected by specific types of damage, enabling faster battery characterization and diagnosis.After establishing the mathematical formulation, we evaluated the changes in the DRT results of batteries via experimental studies. We collected EIS data from eight Li-ion cylindrical cells, comprising four cells with NMC811 positive electrodes and four LFP cells (LiFePO4 positive electrode). The EIS data were collected across various states of charge (SOC) and temperatures ranging from -20 °C to +40 °C (Figure 1). The controlled damage was induced via punch indentation to one cell in each chemistry and stopped before the onset of internal short circuits. EIS measurements were taken at multiple levels of indentation.Our DRT formulation establishes new criteria to identify time constants associated with key internal processes using the experiments. We employed passive electrical circuit components—inductors, resistors, capacitors, and series resistor-capacitor/inductor circuits—as basis functions for the DRT model. A novel set of criteria was introduced to optimize the transformation to the time-constant domain and address the ill-posed nature of the DRT problem. The resulting DRT outputs feature multiple peaks (Figure 2), each corresponding to specific electrochemical processes. We analyzed these peaks to determine their associated time constants, polarizations, and their evolution under varying SOC and temperature conditions.Our findings on the mechanically damaged cells reveal a significant 36% reduction in high-frequency time constant characteristics at indentations greater than 5mm, compared to less than 2.5% variation in undamaged control cells. We note that during these experiments, the cell voltage remained stable despite substantial deformation, underscoring the inadequacy of voltage measurements alone in detecting compromised batteries.The proposed non-invasive method provides a rapid and reliable solution for assessing EV battery safety following road collisions or drone impact landings. It enables the detection of hazardous mechanical damage within seconds, offering a transformative approach to ensuring the safety and longevity of Li-ion batteries. Figure 1