Electrochemical Impedance Spectroscopy

Portable batteries for cell phones, laptops and cameras may be rapid-tested by applying a number of load pulses while observing the relationship between voltage and current. Ohm’s Law is used to calculate the internal resistance. Comparing the readings against a table of values estimates the battery’s state-of-health.

The load pulse method described above does not work well for larger batteries and AC conductance is commonly used. An AC voltage is applied to the battery terminals that floats as a ripple on top of the battery’s DC voltage and charges and discharges the battery alternatively.

AC conductance has been incorporated into a number of hand-held testers to check batteries for vehicular and stationary batteries. To offer simple, compact and low-cost units, these testers load the battery with repetitive current pulses. Because these pulses are not voltage controlled, the thermal battery voltage1) may be surpassed. The thermal voltage threshold of a lead-acid battery is 25mV per cell. Exceeding this voltage is similar to over-driving an audio amplifier. Amplified noise and distortion is the result.

AC conductance provides accurate readings, provided the battery is fully charged, has rested or has been briefly discharged prior to taking the reading. AC conductance tends to become unreliable on low charge and sometimes fails a good battery. At other times, a faulty battery may pass as good. The sensitivity to the battery’s state-of-charge is a common complaint by users. AC conductance works best in identifying batteries with definite deficiencies.

AC conductance is non-invasive, quick and the test instruments are relatively inexpensive. There are, however, some fundamental problems. Most commercial testers use only one frequency, which is commonly below 100 Hertz. Multi-frequency systems would be more accurate but require complex data interpretation software and expensive hardware. In this paper we focus on Electrochemical Impedance Spectroscopy (EIS), a method that overcomes some of the shortcomings of AC conductance.

EIS evaluates the electrochemical characteristics of a battery by applying an AC potential at varying frequencies and measuring the current response of the electrochemical cell. The frequency may vary from about 100 micro Hertz (µHz) to 100 kilo Hertz (kHz). 100µHz is a very low frequency that takes more than two hours to complete one full cycle. In comparison, 100kHz completes 100,000 cycles in one second.

Applying various frequencies can be envisioned as going through different layers of the battery and examining its characteristics at all levels. Similar to tuning the dial on a broadcast radio, in which individual stations offer various types of music, so too does the battery provide different information at varying frequencies.

Battery resistance consists of pure Ohmic resistance, inductance and capacitance. Figure 1 illustrates the classic Randles model.

Figure 1:

Randles model of a lead acid battery. The overall battery resistance consists of pure Ohmic resistance, inductance and capacitance. There are many other models.

Capacitance is responsible for the capacitor effect; and the inductance is accountable for the so-called magnetic field, or coil effect. The voltage on a capacitor lags behind the current. On a magnetic coil, on the other hand, the current lags behind the voltage. When applying a sine wave to a battery, the reactive resistance produces a phase shift between voltage and current. This information is used to evaluate the battery.

EIS has been used for a number of years to perform in-flight analysis of satellite batteries, as well as estimating grid corrosion and water loss on aviation and stationary batteries. EIS gives the ability to observe the kinetic reaction of the electrodes and analyze changes that occur in everyday battery usage. Increases in impedance readings hint at minute intrusion of corrosion and other deficiencies. Impedance studies using the EIS methods have been carried out on lead-acid, nickel-cadmium, nickel-metal-hydride and lithium-ion batteries. Best results are obtained on a single cell.

One of the difficulties of EIS is data interpretation. It is easy to amass a large amount of data; making practical use of it is more difficult. Analyzing the information is further complicated by the fact that the readings are not universal and do not apply equally to all battery makes and types. Rather, each battery type generates its own set of signatures. Without well-defined reference readings and software to interpret the results, gathering information has little meaning for the ordinary person.

Modern technology can help by storing characteristic settings of a given battery type in the test instrument. Advanced digital signal processors are able to carry out millions of instructions per second. Software translates the data into a single reading. EIS has the potential of becoming a viable alternative to AC conductance in checking automotive, traction and stationary batteries. Noteworthy advancements are being made in this field.

Cadex is developing a battery rapid test method incorporating EIS based techniques. Trademarked Spectro™, the system injects sinusoidal signals at multiple frequencies. The signals are voltage controlled and remain below the thermal battery voltage.

Spectro™ is being tested on randomly sampled automotive batteries of various states-of-health conditions. Automotive batteries serve the purpose well because of easy availability. To demonstrate the accuracy, we have tested six typical automotive batteries (A, B, C, D, E, and F) with various state-of-health conditions. The batteries are flooded lead acid of the same model.

Prior to testing, the batteries were fully charged and the actual Cold Cranking Ampere (CCA) reading was established using standards developed under SAE J537. The batteries were then re-tested using the AC conductance and Spectro™ methods. The Spectro™ approximations were conducted using model-specific matrices.

Figure 2: Comparison readings of CCA and Spectro™ using battery-specific matrices. The blue markers compare readings with AC conductance. Spectro™ follows the CCA measurements very closely

Batteries arrive for testing in all conditions, including low state-of-charge (SoC). With AC conductance, the charge level affects the CCA readings to such a degree that the test results may become meaningless. To demonstrate SoC immunity of Spectro™, Spectro was used to estimate CCA at different charge levels. The results are shown in Figure 3.

Figure 3: CCA rapid-tests at various SoC. Spectro™ provides robust readings from 40-100% SoC. The charge level strongly affects the readings when taken with AC conductance. BD

Ideally, the line should be perfectly horizontal. Spectro™ departs only moderately within the 40-100% SoC range. In comparison, the CCA approximations using AC conductance show a strong departure from the horizontal line, caused by the charge level.

In Figure 4, we evaluate the CCA readings of test batteries F, D and C at various charge levels (0-100% charge). The high performing battery F provides solid readings from 10-100% SoC; battery D is steady from 40-100% SoC. The poorly performing battery C provides less stable readings.

Figure 4: Spectro ™ generated CCA approximations of three test batteries at various SoC levels. Battery F provides solid readings from 10-100% SoC; battery D is steady from 40-100% SoC BD

Although early test results conducted with the Spectro™ based technology demonstrate strong advantages over existing test methods, the electrical requirements and complexities are demanding. The need to inject multi-frequency sinusoidal signals at controlled levels and the ability to process reams of data will make Spectro™ more expensive than competitive units. Because of the higher price class, these new generation battery testers will initially be reserved for applications in which the AC conductance does not serve adequately and alternative methods are being sought.

The Spectro™ technologies may one day provide on-line monitoring, providing continuous state-of-charge and state-of-health readings while the battery is in service. Furthermore, Spectro™ may eliminate the need to check the battery’s integrity by applying a full discharge on a load bank. The in-depth approximation approach of Spectro™ may provide better information in seconds by non-invasive means. Work will be in progress to examine ™ the integrity of fuel cell stacks. Spectro™ may one day also detect flaws in metals.

Cadex plans to release the first commercial Spectro™ units in a hand-held tester for automotive batteries towards the end of 2003. Testers for stationary batteries are scheduled for 2004. Research is continuing to include a broad range of battery sizes and chemistries, and to reduce the test time from two minutes to about 20 seconds per battery test. Patents for Spectro™ have been applied for.

1) Batteries are non-linear systems. The equations, which govern the battery’s response becomes linear below 25mV/cell at 25°C. This voltage is called the battery thermal voltage.