Traditional cycle-life testing—which is often, by comparison, limited to the repeated cycling of cells at different temperatures—is commonly conducted at rapid charge-discharge rates to shorten the testing process. This can, however, create additional problems: a cell that behaves one way during rapid and repeated charge-discharge cycles may, for example, behave quite differently when a gradual discharge is followed by a rapid charge. As a result, a new cell chemistry could conceivably pass all required safety tests, yet fail spectacularly in the field in terms of cycle life, safety, or both.

In a paper submitted to the Journal of the Electrochemical Society, the Dalhousie researchers reviewed existing literature on lithium-ion cell cycling, noting that such studies were often problematic on one level or another, especially with regard to the assessment of degradation processes in lithium-ion cell electrodes. For example, many papers on new electrode materials for lithium-ion cells report Coulombic efficiency for no more than 100 to 200 cycles (earlier post)—a fraction of the cycling expected for battery packs designed for vehicle propulsion.

Furthermore, “none of the measurements of CE presented in the literature are accurate or precise enough to be able to distinguish the impact that additives could play in increasing cycle life from 1000 to 2000 (or more) charge-discharge cycles. Such CE measurements would have to be precise on the scale of at least ±0.01%.” The scatter, or variation, of the accuracy of CE measurements reported in much of the literature is 50 times greater than 0.01%.

To mitigate this fundamental gap in their own lab, the Dalhousie team designed and built a prototype high-precision 40-channel charger system using the most accurate current sources that could be obtained commercially. Keithley 220 current sources were used for each charger channel; the 220 model is no longer commercially available, but its level of accuracy is equivalent to that of later equipment (the Keithley 6220). Keithley 2000 and 2750 voltmeters, which are accurate to between one and five microvolts, depending on voltage range, were selected to monitor cell potential.

Each of four voltmeters was set up to scan ten cells in sequence. The current sources and voltmeters were connected to the cells using four-wire terminals at each cell under test; with 40 current sources, the charger system is capable of evaluating up to 40 cells at a time.

LabView was used to write the software which controls the current sources and scanning voltmeters. Cells are evaluated in temperature chambers, referred to as thermostats, at 30, 40, 50, and 60 ºC, with a temperature accuracy of 0.1 ºC.

Results indicate that the relative advantages or disadvantages of a given change in material or process could be quantified in as little as 20 cycles, using the Dalhousie charger array. Many observed trends had not previously been quantified in existing literature, other than assertions that a given material or process improved or degraded the efficiency of the cell.

Experimental results have demonstrated the promise of CE measurements made using a high precision charger. Some expected trends were confirmed:

1. A LiCoO₂/graphite cell had a higher CE (99.90%) than a Nexelion-type (Sn-based anode) cell (99.75%);

2. Li/Li₄/3Ti₅/3O₄, Li/graphite and Li/Sn₃₀Co₃₀C₄₀ cells had CEs of 99.93, 99.90 and 99.65% at cycle 25, respectively;

3. The CE of Li/Sn₃₀Co₃₀C₄₀ cells improved from 99.65 to 99.75% upon the addition of 10% FEC (fluoroethylene carbonate) to the electrolyte;

4. Removal of water from Li/graphite cells by electrode drying improved the CE from about 99.87% to 99.90% after 25 cycles; and

5. Lithium/nickel-manganese-cobalt (Li/NMC) cells showed CE > 1.000 and a CE which increased as the upper cutoff potential was increased.

Preliminary data collection with the Dalhousie charger array also yielded a few surprises:

1. Li/Li₄/3Ti₅/3O₄, Li/graphite and Li/Sn₃₀Co₃₀C₄₀ cells required about 25 cycles for their CEs to stabilize asymptotically; and

2. The impact of upper cutoff potential on the CE of the NMC sample studied was extremely dramatic.

The researchers conclude that more precise measurements of Coulombic efficiency could much more accurately quantify the impact of electrolyte additives, electrode materials and coatings, impurities, temperature, electrolyte salts, and other factors on the cycle life of lithium ion chemistries under test. This, they note, “could represent good news for university-based and other researchers who do not have access to facilities for the construction of production-quality Li-ion cells, but who still want to do meaningful research focused on developing solutions for automotive and energy storage batteries.”

Despite the improvements in test results they have enjoyed with the Dalhousie charger array, surpassing the accuracy of known commercially available lab equipment, the researchers point out that “...it would be useful to increase the accuracy of the CE measurements by another order of magnitude compared to our system... Hopefully, some equipment manufacturers will take up those challenges so that the needed equipment can be available in a timely manner for the research at hand.”

Team member Jeff Dahn of Dalhousie University echoed that sentiment at last month’s “Beyond Lithium” conference at the IBM Almaden Research Center in California’s Silicon Valley, noting that “if this problem can be solved anywhere, surely it can be solved here.”

Resources

• Smith et al, 2009: Precision Measurements of the Coulombic Efficiency of Lithium-ion Batteries and of Electrode Materials for Lithium-ion Batteries. Journal of the Electrochemical Society (submitted)

• Jeff Dahn’s Research Group, Department of Physics and Atmospheric Science, Dalhousie University, Halifax, Canada