Center for Advanced Life
Cycle Engineering

calce address

CALCE  Battery  Research  Group

Current Research Projects

The strides made over the past two decades of lithium ion battery commercialization have opened the door to many future opportunities. However, the technology surrounding lithium-ion batteries is still in its infancy and has room for improvement. This improvement can be realized through either development of new battery materials or through optimization of battery performance and reliability. CALCE has its projects spread across each of the sections shown in the figure on the right. --> CALCE is majorly focused on the development of state of the art battery management systems (BMS) for single and multi-cell systems to provide the most accurate state of charge (SOC) and state of health (SOH) metrics. CALCE is working towards this goal through studies on fundamental process that degrade battery cells, battery testing methods, techniques for battery failure analysis, and advanced data processing techniques for implementation of battery prognostics and health management (PHM) solutions.

Partial State of Charge (SOC) Cycling Effects on Capacity Fade of Lithium-ion Cells

Figure 1  
Li-ion diffusion coefficient of LiFePO4 measured using current solid solution GITT and our phase transformation GITT. The phase region is marked based on the lithiation equilibrium potential-composition curve.  

In practical applications, batteries may undergo charge-discharge cycling only for partial SOC ranges as opposed to the full 0%–100% range. CALCE has been performing cycle life testing of commonly used graphite/LiCoO2 pouch cells in different SOC ranges (e.g., 0-100%, 20-80%, 20-100%) to understand the effects of different SOC ranges on battery capacity fade and to model the battery capacity fade as a function of mean SOC, ∆SOC and cycle count. The developed models can be used for battery health management in field applications as well as for accelerated testing during battery qualification.

SOC ranges during cycling affect degradation mechanisms such as SEI layer formation and crack generation in the electrode. For example, as we increase the upper limit of SOC (end of charge voltage) in cycling, the amount of lithium in the anode increases, resulting in anode lattice volume expansion and causing localized stress. Also continuous cycling with higher ∆SOC value increases the probability of crack generation in the anode due to cyclic fatigue. These cracks in the anode provide fresh sites for electrolyte reduction and SEI layer growth, causing loss of cycleable lithium and higher electronic resistance. The figure below shows the effect of lowering the ∆SOC on the battery degradation.

Ref: Saurabh Saxena, Christopher Hendricks and Michael Pecht, Cycle Life Testing and Modeling of Graphite/ LiCoO2 cells under different state of charge ranges, Journal of Power Sources, 327 (2016), pp.394-400, 2016.

Accelerated Degradation Models for C-rate Loading of Lithium-ion Batteries

Qualification testing for Li-ion batteries can be quite time consuming. To give a perspective, an example of battery cycle life testing with C-rate as one of the stress factors can be considered. The battery current is usually expressed in terms of C-rate; e.g. the battery current normalized to the rated capacity (C) of the battery. For a 1 Ah battery, a C-rate of 1C represents a 1 A current; C/2 (0.5C) rate represents a 0.5 A current. Thus, in order to test a battery at a C-rate of C/2, it will take approximately 4h to complete 1 cycle (not counting any resting period to cool the battery down). For Li-ion batteries that are expected to have cycle lives in the range of 1000 cycles, then nearly 6 months is required to test at a C-rate of C/2. Testing at a high C-rate is one approach to accelerated testing for Li-ion batteries.

To assess the reliability of a product quickly, accelerated reliability testing is conducted by increasing the loading (stress) conditions on the product. Acceleration models are used to extrapolate the testing results in accelerating stress variable and often in time as well. CALCE is currently developing accelerated degradation models which utilize both the historical degradation data and the basic physics behind the degradation of batteries. These models will be useful in reducing the time required for battery qualification to less than 200 cycles or 1 month.

Simplified electrochemical model and parameter estimation

Figure 1  
Simplified electrochemical thermal coupling model for Li-ion battery.  

As electrochemical models can accurately simulate battery behaviors with the entire scope of state of charge (SOC), they are very appealing in BMS. However, the large computational cost of solving partial differential equations limits their practical applications. On account of low computational cost, simplified electrochemical models are more suitable. A simplified electrochemical thermal coupling model with reduced and regrouped model parameters was established and a method for nondestructive parameter estimation for individual cells was developed based on excitation response analysis. The parameters were classified into two categories: inherent characteristic parameters and mechanistic parameters. Inherent characteristic parameters could be obtained by consulting manufacturers directly or measuring. Mechanistic parameters were obtained by the excitation response analysis. According to different response time of different processes in the developed model when a cell was applied with different current excitations, the corresponding parameters were then obtained by least square fit.

Figure 1  
Comparative results of measurements and simulations at different C rates corresponding to two ambient temperatures (25℃ and 40℃).  

The potential application of this model is that it can be applied to estimate SOC of single cell. And the model parameters can also be used as features to assess battery health state. The remaining useful life of battery can be predicted by analyzing the variations of parameters at different aging stages based on the developed model. However, the large number of model parameters and the long period of testing time are currently the limits in terms of parameter estimation for battery pack. According to the contributions of model parameters on battery behavior simulation, the selection of key parameters can be a potential solution to reduce the number of estimated parameters, and an alternative short testing schedule can also reduce the estimation cost.

Ref: [1] Li J, Wang L, Lyu C, Wang H, Liu X. New method for parameter estimation of an electrochemical-thermal coupling model for LiCoO2 battery. Journal of Power Sources. 2016;307:220-30.
[2] Li J, Wang L, Lyu C, Wang H, Lai Q. A method for SOC estimation based on simplified mechanistic model for LiFePO4 battery. Energy. 2016;114C:1266-76.

Electrochemical Characterization of Lithium-ion Batteries Using a Three-electrode System

Figure 1  
The three-electrode cell test bench.  

Recently, Lithium-ion batteries have been used in many electronic devices including laptops, smartphones, electric vehicles (EVs) and are also being considered for military and space applications. Generally, (commercial) Li-ion batteries have two-electrode. Therefore, the battery voltage or impedance can be measured only across the negative and positive electrode. In order to understand the fundamental electrochemical characteristics and to utilize the Li-ion battery chemistry more efficiently and safely, it is of interest to study the performance of each electrode separately. A three-electrode system is able to interpret the electrochemical characteristics of the individual electrodes. A widely used commercial 18650 battery (LiFePO4/ graphite) was reconstructed into a three-electrode full cell. Based on the three-electrode cell, the voltage and impedance of not only the full cell but also the individual electrodes were monitored. Accordingly, the electrochemical behavior of the commercial cell was explained and the contribution of each electrode to the full cell was identified.

Figure 1 Figure 1
Impedances of the cathode and anode under different cycle number.

Dendrite Formation Mechanism

Figure 1  
Dendrite growth during constant charging test.  
Figure 1  
Measured voltage data during the charging test.  

Lithium-ion batteries are commonly used in daily life. Concerns regarding lithium-ion battery safety are increasing with the widespread use of these cells in various applications. Among all the reported battery incidents, lithium dendrite formation causing internal short circuits was considered as the direct or indirect reason for battery failure. Dendrites can cause short-circuits, which can lead to catastrophic failures and even fires. Lithium dendrite is a metallic microstructure that forms on the negative electrode during the charging process. This is one possible reason for the internal short-circuits of lithium ion batteries. The lithium dendrite issue can occur in the lithium ion batteries when the battery is overcharged or charged at low temperatures. The dendrite growth is influenced by the applied current density.

In order to increase Li-ion battery safety, it is necessary to conduct research on lithium dendrite formation mechanism. For this purpose, an in-situ observation method was used to detect dendrite formation at various current densities and temperatures. The relationship between the applied current density and the dendrite growth rate is the research focus in the first step of study. In order to determine the dendrite growth rate, a symmetrical lithium cell (both the positive and negative electrodes were made of lithium metal) was charged at constant current. The developed in-situ testing method can be used for identification of dendrite formation inside cells. Battery safety operation boundary conditions can be determined using this work.

An Early Degradation Detection Method for Lithium-ion Batteries

Early detection of the potential degradation in the Lithium-ion batteries can avoid catastrophic failure and reduce the maintenance cost. In the conventional battery failure detection system, the sensors can only detect the external signals of the cell, such as the current, voltage and temperature. However, the cell’s health state cannot be directly reflected from these signals. Actually, the health state of a cell is also physically coupled with the distribution and changes in the material properties, such as the density and modulus. Thus, a sensor that can sense the inherent physical changes in the cell during operation is more desirable for early detection of the degradation in the cell. Ultrasonic inspection method is used to detect internal material defects in metallic, composite materials, etc. This method has the potential to reflect the internal state of the cells, and provide early fault indications of failures, such as of internal gas formation, material deposition, and internal density changes.

We are now conducting ultrasonic test for three-type of polymer lithium-ion batteries using a set of ultrasonic hardware provided by X-wave Innovations, Inc. This hardware contains of a thin transducer which is mounted on the cell surface using petroleum jelly, a pulser-receiver (model US-Key) with DAQ capability. This hardware has small volume and low power compared with commercial bulky ultrasonic probes and equipment. Ultrasonic signals are collected after every charge-discharge cycling test of the cells. The time of flight and amplitude of the reflected echo are extracted from the ultrasonic signals. Now it can be seen that there is a strong correlation between the signal features and the cell capacity. This results preliminary verify that the ultrasonic signals can provide very useful information to indicate the health state of the cell and the ultrasonic inspection method can be used for early degradation detection of the cell.

Figure 1  
Figure 2  
Evolution of the ultrasonic signals for the battery with the increasing cycle number