An interesting glimpse into severely overdischarged Li-ion cells

Readers may be interested in a recent study which provides an interesting glimpse into what occurs when Li-ion cells are severely overdischarged (e.g. as may occur when a cell in series is reverse charged). The study linked below examines what occurs when cells are discharged to up to -100% SOC (i.e. reverse charged at various levels up to 100% capacity)

One interesting result is that overdischarges < -12% SOC reliably cause internal short-circuits - caused by copper foil dissolution and deposition, as illustrated in the diagram below.



Below is the associated voltage profile.


Notice how the voltage rebounds in Stage II. This occurs because the anode has reached so high potential (3.5-3.6V) that it enables dissolution of the copper current collector. Then copper ions dissolve into the electrolyte, pass through the separator, and deposit onto the cathode, which increases potential as the copper ions are reduced. This process continues in Stage III. and the internal short(s) become more severe, while the voltage asymptotically approaches zero.

One implication of this is that the cell voltage alone cannot be used as a reliable indicator of how overdischarged the cell was since, e.g. above -0.5V could be either -9% or -100% SOC. All the more reason to be very careful with cells that have been severely overdischarged.

The article is freely available on nature.com's scientific reports: Mechanism of the entire overdischarge process and overdischarge-induced internal short circuit in lithium-ion batteries, by Rui Guo et al., Scientific Reports, 6:30248 · July 2016, DOI: 10.1038/srep30248

Update There is much further discussion in the parallel thread here.

Aside from the destructive discharge, they recharged these over-discharged cells at C/3 ! That is a *guaranteed* fail.

They are not following the reciprocal of the discharge knee - that is, when you drop far down into it, you come OUT of it with very little current, like C/100 or even less until the normal nominal voltage is reached. (ie, 3.2 for Lifepo4, or 3.6v for any other lithium chemistry).

If you don't, you create an ion-storm of intercalation that can't be processed that fast. Not to mention aggravating other secondary reactions.

Most good smart chargers acknowledge this and charge slowly if the battery is very discharged - and not even taken to such extreme levels like their test did.

I wonder why they forced the issue by recharging at C/3 - these guys should have known better, especially with such destructive discharge levels.

@IonicBond The C/3 recharge is certainly not a "guaranteed fail" since, as the study shows, the cells reverse-charged less than 12% showed no internal shorts or no significant capacity loss after a C/3 recharge.

The purpose of the recharge was not to attempt to reverse the damage but rather to measure it (esp. the magnitude of the internal short). As such, your comments do not apply.

Am I understanding correctly that this is similar to what occurs with batteries left at very low states of charge, but more rapid?

@iamlucky13 Copper dissolution and internal shorts can also occur when cells are left for long periods at very low voltages (usually below 2.0V). But it typically happens much more slowly than in the extreme cases in the study.

They study cells that are reverse-charged from 0-100%. Reverse charging can occur when you have cells in series and one of the cells has much lower capacity than the others, and the pack lacks appropriate protection guarding against cell undervoltage. Such cell capacity mismatch can occur if cells were not well matched during pack assembly, or if one cell degrades faster than others, e.g. if it is closer to hot components.

Generally, once the cells start diverging in capacity, the lowest capacity cell degrades faster than the others, since it ends up being exposed to more extreme voltages (which accelerate degradation). So the capacity difference does not remain constant, but continually increases. Once the difference is large enough, reverse charging can occur if there is no protection circuit guarding against it.

Anyone who has harvested enough laptop batteries has probably seen this in practice, viz. some cells will be dead but others will still have reasonable capacity. This is often because the dead cells were closer to hotter components (esp. on older Pentium laptops that ran very hot), e.g. see the photo below. But quality laptop batteries have a proper BMS that will never let the pack reach a state where reverse-charging can occur. Generally the BMS will disable the pack long before that point (and most users would stop using it before that point due to the greatly diminished capacity).



Excerpted from Battery Monitoring Basics - TI Training, by Texas Instruments.

So when the study (and charts) talk about -nn% SOC, they are describing the amount of recharging in a reversed polarity situation, ie: -10% SOC has been barely reverse charged as compared to say -40% SOC?

Timothybil said:

So when the study (and charts) talk about -nn% SOC, they are describing the amount of recharging in a reversed polarity situation, ie: -10% SOC has been barely reverse charged as compared to say -40% SOC?

Click to expand...

Correct. Below is their test rig. The cells are 25Ah NCM pouch cells in metal cases. As the diagram shows, they connected 5 cells in series, with 1 fully discharged, and 4 fully charged. Then they discharged the pack at C/3 = 8.33A to different depth levels 12% < N% < 100%, which has the effect of discharging the empty cell to -N% capacity, i.e. it is reverse-charged N%.

Then they attempted C/3 recharges of the reverse charged cells. They found that cells that were not discharged past -12% (stage I, before the voltage starts rising due to copper dissolution), could be fully recharged and reused with "only minor side-effects" (confirmed also by SEM, XRD, and half-cell tests). But cells overdischarged any further had significant self-discharge due to internal short-circuits, and could not be fully recharged if overdischarged < -14.5%. Further they utilized the self-discharge rate to estimate the effective resistance of the internal short-circuit using a simple equivalent circuit model, obtaining very close results to the values observed.

Thanks for the followup Gauss163. That's exactly what I was asking.

Timothybil, I take that to mean, for example, that for -10% SoC, 200 mAh have flowed the wrong way through a 2000 Ah battery, and 40% means 800 mAh. At 10%, it sounds like there is minimal harm (perhaps some capacity loss and higher internal resistance?) from this if the battery is promptly recharged, but just a couple percent more and meaningful damage starts to occur.

I was just reading more of the paper, and in the most severe case they tested, they saw charging times more than double, and self-discharge of the fully charged battery all the way down to 2.5V occur in a single day.

From what I gather, the best way to measure the safety of a cell is to charge it up to 4.2v, let it sit for a few days, and then measure if the voltage is dropping (due to self-discharge).

Even healthy cells will drop a bit in voltage during the first few hours (maybe down to 4.16v), but then they should maintain that voltage for weeks.

What is an acceptable drop in initial voltage, and what is an acceptable self-discharge rate? e.g., Is a cell still okay if it drops quickly to 4.10v, and then drops slowly to 4.05v over a period of a month?

Most of my laptop pulls tend to settle to about 4.15v, and then sit there for a couple of weeks (maybe dropping another 0.005v). So, I think those are safe. But what are the limits?

WalkIntoTheLight said:

From what I gather, the best way to measure the safety of a cell is to charge it up to 4.2v, let it sit for a few days, and then measure if the voltage is dropping (due to self-discharge) [...]

Click to expand...

I'm not aware of any studies that yield any data on how well that heuristic may work. It will certainly yield many false positives since cells with high IR will typically have a large voltage drop after charge because consumer level chargers don't allow you to adjust the charge termination current low enough to fully charge them. The above heuristic would deem such cells unsafe even though they may be safe and useful for low current apps. But at least that errs on the safe side.

To remedy that you'd need to adjust the heuristics to account for the IR of the cell, and the (fixed) termination current of the charger, so you can distinguish between a small internal short vs. the voltage rebound of a high IR cell. If you could graph the voltage decay curve it would probably be easier to distinguish between the two since the voltage rebound has a typical exponential decay shape, but an internal short would add in a linear component, and the change in curve shape should be obvious. Also, as in the study, one can look for the effects the internal short has on the charge process.

Gauss163 said:

I'm not aware of any studies that yield any data on how well that heuristic may work. It will certainly yield many false positives since cells with high IR will typically have a large voltage drop after charge because consumer level chargers don't allow you to adjust the charge termination current low enough to fully charge them. The above heuristic would deem such cells unsafe even though they may be safe and useful for low current apps. But at least that errs on the safe side.

To remedy that you'd need to adjust the heuristics to account for the IR of the cell, and the (fixed) termination current of the charger, so you can distinguish between a small internal short vs. the voltage rebound of a high IR cell. If you could graph the voltage decay curve it would probably be easier to distinguish between the two since the voltage rebound has a typical exponential decay shape, but an internal short would add in a linear component, and the change in curve shape should be obvious. Also, as in the study, one can look for the effects the internal short has on the charge process.

Click to expand...

Yes, that is why I mentioned letting the cell sit for a few hours to let the voltage settle. 12 hours should be long enough, shouldn't it? After that, any internal shorts will show up as a continued voltage drop over days or weeks. If it remains a constant voltage, or drops very little (normal self-discharge), the cell should be safe. At least, that's what I'm wondering, and what parameters are within the healthy range for old cells.

I think I'd still be concerned about a cell that was discharged to 0V (or less!). But, if it holds a good voltage after charging, does that mean it's safe?