Elsevier

Energy Storage Materials

Volume 42, November 2021, Pages 794-805
Energy Storage Materials

Non-invasive identification of calendar and cyclic ageing mechanisms for lithium-titanate-oxide batteries

https://doi.org/10.1016/j.ensm.2021.08.025Get rights and content

Highlights

  • Long-term calendar and cyclic ageing data of 43 NMC-LCO/LTO battery cells.

  • Accurate identification of degradation mechanisms using relative shift in IC curves.

  • Data proofed two-stage ageing mechanism with stagewise increasing ageing gradient.

  • Capacity gain of up to 2.42% visible for cells stored for cells stored below 50% SOC.

  • RNN-LSTM capacity estimation model based on IC curves with high accuracy (R2=0.98).

Abstract

Lithium-titanate-oxide (LTO) batteries are one of the most promising technologies for various types of future applications in electric mobility, stationary storage systems and hybrid applications with high-power demands due to their long cyclic stability and superior safety. This paper investigates the cyclic and calendar ageing of 43 same-typed LTO cells considering 16 different operation conditions under variation of state of charge (SOC), temperature, depth of discharge, cycle SOC range and current rate. The ageing results are presented and the relative shift in incremental capacity is analysed in order to detect degradation mechanisms, separate the influence of degradation enhancing parameters and attribute them to their origin source. Our results show that the cells exhibit a two-stage ageing mechanism with stagewise increasing degradation gradient. In the first ageing stage the anode is limiting the amount of extractable capacity while the capacity fade mainly results from cathode degradation. After a certain level of degradation is reached the cathode starts limiting the amount of extractable capacity, initiating the second ageing stage with stronger occurring capacity fading gradient. A capacity gain of up to 2.42% becomes visible for cells operated and stored in a range below 50% SOC. For these cells an extended three-stage ageing mechanism is shown to be more applicable. The degradation behaviour is then estimated using a machine learning approach based on a recurrent neural network with long short-term memory, for which the presented incremental capacity data is used as training input.

Introduction

Due to continuous global warming caused by greenhouse gas emission, emission-reducing innovations, especially in the field of battery-powered mobility, are subject to many applications and studies in recent years [1], [2], [3]. Lithium-ion batteries (LIBs) have become predominant in the field of electromobility, as they offer a high energy- and power-density, a high charge and discharge efficiency, as well as a low self-discharge rate  [4].

A wide range of cathode materials for LIBs can be found within a variety of applications. Transition metal oxides, e.g. lithium nickel-manganese-cobalt oxide (NMC), lithium nickel-cobalt-aluminum oxide (NCA), lithium manganese oxide (LMO), lithium cobalt oxide (LCO) and phosphates like lithium iron phosphate (LFP) are commonly used and offer different characteristics regarding energy- and power-density, safety, reliability, lifetime and costs. However, on the anode side graphite is found almost exclusively for most applications [5]. The low potential of graphite (0.05 V vs. Li/Li+ [6]) enables high energy densities when combined with one of the aforementioned cathode materials. Nevertheless, this same feature makes the typically used electrolytes thermodynamically unstable within the limited operating range of 0.8 V - 4.5 V vs. Li/Li+ [6]. This leads to accelerated ageing due to the necessary formation and constant increase of a passivation layer referred to as Solid Electrolyte Interphase (SEI). For graphite-based LIBs it is widely accepted, that the thickening of the SEI is the main ageing mechanism resulting in irreversible loss of lithium [7], [8], [9].

A more recent auspicious type of LIBs are lithium-titanate-oxide (LTO) cells. Despite their lower nominal voltage and energy density compared to other LIB chemistries [10], the spinel Li4Ti57O12 is a promising anode material, particularly in the field of high-power electric mobility for e.g. trains, ships and heavy-duty vehicles [11], [12], [13]. Due to the high anode potential of LTO (1.55 V vs. Li/Li+ [14]), it operates within the electrochemical stability window of most electrolytes as opposed to graphite. This results in limited or no SEI formation, which has insignificant influence on cell ageing [15], [16]. Hence, fast charging at low temperatures becomes possible without the threat of dendrite-formation [17], commonly referred to as lithium plating, making LTO anodes one of the safest and most reliable cell technologies [18]. Furthermore, LTO anodes undergo a negligible volume expansion during the intercalation process (0.1 vol% [14]) compared to graphite anodes (16 vol% [19]), resulting in better cyclic stability and low charge-discharge hysteresis [20]. Only in high-temperature ranges cells with LTO anodes show demonstrably increased ageing. The generation of gas, triggered by trace water elements and solvent reactions between the electrolyte and the LTO surface, not only leads to an accelerated cell ageing but is also accompanied by a safety risk due to the increasing internal pressure [21], [22], [23], [24].

Identifying degradation mechanisms for LIBs used in electric transportation or stationary storage applications have become an important field of research. A qualitative degradation analysis is crucial for both manufacturers and customers to evaluate the rentability of a whole system, as the battery is likely the most expensive component [25]. Hence, knowing degradation amplifying or reducing operation modes can help to improve lifetime and reduce costs of the battery system by avoiding critical states. Therefore, cell suppliers perform various ageing tests on their products in order to determine individual degradation levels for different operating modes. Nevertheless, ageing characterisation of LIBs is challenging as it usually depends on several varying external factors, especially on operation characteristics that are not completely reproducible during common ageing studies [26].

There are only a few studies available characterising LTO degradation mechanisms. This can be attributed to the high costs leading to less frequent use of these type of cells. Additionally, the superior ageing behaviour of LTO cells makes it difficult to develop comprehensive test matrices, as some operating points will not show any significant degradation even after long test duration. Hence, most available studies on LTO cells focus on analysing the ageing behaviour under severe conditions at high temperatures, high current rates (C-rates) or high depths of discharge (DOD), which will inevitably produce results that are not applicable to cells under normal operation.

Different studies have proven the high cyclic stability of LTO based cells even under severe temperatures and high C-rates [10], [27], [28], [29]. Bank et al. [30] investigated the cyclic and calendar lifetime performance at elevated temperatures under variation of C-rate and DOD. The cyclic tests performed at 40°C showed, that even after more than 10.000 equivalent full cycles (EFC) at DOD=100% and a C-rate of 5C, over 85% of the initial capacity remained. For DODs<70% the cells showed excellent cyclic stability, displaying only a slight loss in capacity of under 6%. The calendar ageing tests performed at 60°C showed no significant degradation even after 300 days of storage for cells stored between 5 -70 % SOC. At storage SOC 90% a capacity degradation of under 5% became visible.

Nemeth et al. [31] investigated the performance of two different types of LTO cells and conducted cyclic lifetime tests using high current cycling with 10C at ambient temperature of 25°C. The ageing results showed remarkable cycle lifetime behaviour for both cells. °After 20.000 EFC, respectively 13.000 EFC both cell types showed a remaining capacity of approximately 95%. An incremental capacity analysis (ICA) was performed to determine the main ageing mechanisms. The authors concluded that the anode is the main determining ageing factor, as it limits the end-of-charge voltage (EOCV) and the end-of-discharge voltage (EODV), which agrees with the findings of Farmann et.al [18] and Devie et al. [32].

Liu et al. [33] gave more quantitative statements on the occurring ageing mechanisms. They investigated the cyclic ageing performance of LTO cells at room temperature under consideration of different C-rates. The results were analysed in-depth by comparing the incremental capacity (IC) and differential voltage (DV) curve shifts after different cycle levels with regards to the occurring degradation mechanisms. The results showed that loss of active material on the positive electrode (LAMPE) was identified to be the main ageing mechanism, accounting for over 80% of cell degradation, while loss of active material on the negative electrode (LAMNE) and loss of lithium inventory (LLI) only played a minor role. This is in accordance with the findings of Hall et al. [28] and Svens et al. [34], who showed that the cathode has a dominating ageing influence. They further assumed that the negative electrode only limits the amount of extractable capacity at begin of life (BOL), while the positive electrode will become the limiting electrode later on with increasing degradation. This two-stage ageing mechanism had already been proposed by Han et al. [35].

Baure et al. [36] investigated the ageing of commercial LTO cells. The cell was simulated to predict the effect of different degradation modes on the IC curves. The analysis showed that the capacity loss in their case was solely induced by LAMNE from the anode. However, their simulation also demonstrated that different operation modes will lead to different ageing triggering events for which in some cases LAMPE will be the main degradation cause. Although LAMPE did not induce any significant capacity loss for their cells at first, they concluded it still could be of substantial essence for the degradation of these cells under specific operation conditions.

In this work we investigated the ageing of 43 LTO cells with a blend NMC-LCO cathode using a comprehensive test matrix. The test matrix considered calendar and cyclic ageing measurements for a total of 16 different operation conditions with variations in SOC, temperature, DOD and C-rate. Hence, a broad spectrum of the operation range was covered to give a holistic view on the occurring degradation characteristics of this LTO cell. This allows separation and allocation of different ageing accelerating or decelerating factors to the respective causes. The results of these tests are visualized and described with regards to capacity decrease and inner resistance increase. In order to draw conclusions about initial cell characteristics and investigate long-term ageing effects without the need of disassembling the cell, non-invasive methods are used. These methods are beneficial to determine the degradation state of a battery via externally measurable parameters. The ICA is such a non-invasive technique, used to retrieve information about the electrochemical properties of a cell and derived by a full charge or discharge voltage curve. Comparing the IC curve of a full cell with the respective materials IC characteristic of a half-cell enables assigning ageing mechanisms to the corresponding electrode [37], [38]. In order to identify specific degradation causes and analyse the properties of the used electrode materials, the initial IC curve characteristics of the tested cell is evaluated. By comparing the changes of these characteristics over time at certain operation modes, the separate influence of different parameters, such as temperature, DOD and SOC, on electrode degradation is analysed. Therefore, a distinction is made between LAMPE, LAMNE and LLI. The results are used to verify the often predicted two-stage ageing mechanism from various other LTO research [31], [35], [36]. For some operation points, a gain in capacity over time became visible, which is in this work as well identifiable through an increase of certain characteristics in the IC curves and will be presented here. In such cases a three-stage ageing mechanism is shown to be more suitable, considering the observed degradation mode. One of the main goals of conducting ageing tests and characterization analysis is to estimate and predict the lifetime behaviour for future investigation in high power applications. Therefore, based on the relative shift of the investigated IC curves, a recurrent neural network (RNN) with long short-term memory (LSTM) was developed to estimate the capacity decrease. The model results were validated by using the prior mentioned available ageing data.

This paper is structured as follows: The measurement equipment and experimental procedures will be presented and described in chapter 2. In order to investigate short-term and long-term ageing effects, the initial cell characteristics are determined in chapter 3. Furthermore, the results of the calendar and cyclic ageing tests under consideration of different ageing influencing factors are presented. In chapter 4 based on the comparison of relative changes in the IC curves for both calendar and cyclic ageing, degradation causes will be identified and assigned to their original source. Based on these investigations a RNN with LSTM will be implemented and presented. Finally, a conclusion will be drawn in chapter 5.

Section snippets

Experimental

In this work the ageing behaviour of 43 cells with LTO as anode material and a blend of NMC-LCO on the cathode is investigated. As measurement equipment for all characterisation tests a Digatron MCT ME type is used. The channels offer accuracies of 0.1% over the full scale. A Digatron BNT ME with 18V/200A circuits was used for the cyclic tests, offering an accuracy of 0.1% over the full scale. The cells are stored in ovens of the type Memmert in which the chosen test temperature is adjusted.

Initial cell characteristics

In order to draw conclusions about the initial cell characteristics and later on investigate long-term ageing effects without the need of disassembling the cell, the ICA method is used. Therefore a full charge or discharge curve measured at preferably low C-rates is needed to reduce the amount of kinetic artifacts. As the initially conducted QOCV curves at C/10 were omitted after the first characterisation test, the C/3 discharge curves will be used to assess changes in the phases of the

Calendar and cyclic IC curve analysis

Fig. 6 a–c shows the IC curves for all calendar ageing measurements performed at SOC=100% with different storage temperatures 35°C, 45°C and 55°C. As mentioned, in the two-phase transformation region, the anode has a constant voltage plateau. Hence, the intensity decrease and shifts for peaks ① and ② can be solely attributed to the degradation of the NMC/LCO cathode. Considering peak ①, a rather small decrease in intensity with increasing temperature is observable, which shows no significant

Conclusion

In this paper, the cyclic and calendar ageing results of in total 43 LTO cells were presented and discussed. Our investigation showed, that cathode degradation can be tracked efficiently using LTO as reference anode due to its pronounced calendar and cyclic lifetime behaviour. The ageing causes on electrode level were analysed by considering the IC curves at different operating conditions and assigning characteristic peaks to specific degradation mechanisms. Our analysis showed that in general,

CRediT authorship contribution statement

Ahmed Chahbaz: Conceptualization, Methodology, Software, Validation, Formal analysis, Investigation, Data curation, Writing – original draft, Visualization, Writing – review & editing. Fabian Meishner: Validation, Formal analysis, Investigation, Data curation, Writing – review & editing. Weihan Li: Methodology, Software, Resources, Visualization, Writing – review & editing. Cem Ünlübayir: Validation, Investigation, Data curation, Writing – review & editing. Dirk Uwe Sauer: Validation, Writing –

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgement

This work is funded by the German Federal Ministry for Transport and Digital Infrastructure (BMVI) with the funding numbers of 03B10502B and 03B10502B2. The authors would like to thank Hantian Zhang for his work in data visualization.

References (49)

  • N. Takami et al.

    High-power and long-life lithium-ion batteries using lithium titanium oxide anode for automotive and stationary power applications

    J. Power Sources

    (2013)
  • F. Hall et al.

    Experimental investigation of the thermal and cycling behavior of a lithium titanate-based lithium-ion pouch cell

    J. Energy Storage

    (2018)
  • T. Bank et al.

    Extensive aging analysis of high-power lithium titanate oxide batteries: impact of the passive electrode effect

    J Power Sources

    (2020)
  • T. Nemeth et al.

    Lithium titanate oxide battery cells for high-power automotive applications–electro-thermal properties, aging behavior and cost considerations

    J. Energy Storage

    (2020)
  • S. Liu et al.

    Analysis of cyclic aging performance of commercial Li4Ti5O12-based batteries at room temperature

    Energy

    (2019)
  • P. Svens et al.

    Analysis of aging of commercial composite metal oxide–Li4Ti5O12 battery cells

    J. Power Sources

    (2014)
  • X. Han et al.

    A comparative study of commercial lithium ion battery cycle life in electrical vehicle: aging mechanism identification

    J. Power Sources

    (2014)
  • I. Bloom et al.

    Differential voltage analyses of high-power lithium-ion cells. 4. Cells containing nmc

    J. Power Sources

    (2010)
  • M. Ecker et al.

    Calendar and cycle life study of Li (NiMnCo) O2-based 18650 lithium-ion batteries

    J. Power Sources

    (2014)
  • G. Baure et al.

    Battery durability and reliability under electric utility grid operations: 20-year forecast under different grid applications

    J. Energy Storage

    (2020)
  • S.F. Schuster et al.

    Lithium-ion cell-to-cell variation during battery electric vehicle operation

    J. Power Sources

    (2015)
  • E. Sarasketa-Zabala et al.

    Cycle ageing analysis of a LiFePO4/graphite cell with dynamic model validations: towards realistic lifetime predictions

    J. Power Sources

    (2015)
  • M. Naumann et al.

    Analysis and modeling of calendar aging of a commercial LiFePO4/Graphite cell

    J. Energy Storage

    (2018)
  • S. Watanabe et al.

    Capacity fade of LiNi (1- x- y) CoxAlyO2 cathode for lithium-ion batteries during accelerated calendar and cycle life test. I. Comparison analysis between LiNi (1- x- y) CoxAlyO2 and LiCoO2 cathodes in cylindrical lithium-ion cells during long term storage test

    J. Power Sources

    (2014)
  • Cited by (19)

    • Model-based investigation of an uncontrolled LTO wayside energy storage system in a 750 V tram grid

      2023, Applied Energy
      Citation Excerpt :

      Another reason for focusing on short service lifetimes is that the impact of ageing mechanisms (particularly internal resistance increase) on operation can be overlooked at this stage. The OCV change of an aged cell [38] should play an subordinate role, as the basic cell limits remain during its lifetime. The number of serial and parallel cell connections determine the system design.

    View all citing articles on Scopus
    View full text