diy solar

diy solar

Science based discussion on charging LiFePO4 below 32°F

eemarty

Solar Enthusiast
Joined
Oct 1, 2022
Messages
153
This topic has been discussed many times here and I'm well aware that standard practice is that you should not attempt to charge below 32°F but I found the previous discussions lacking in rigor and specifics. Actual tests are difficult to do without access to electron microscopes and lots of time to run many cycles in controlled conditions but fortunately there has been a ton of academic research on Lithium batteries in the last 10 years and lots of data is publicly available. A lot of this research is targeting the automotive sector which desires the fastest charge times and needs to operate outside in the winter in very cold climates. Tesla superchargers peak charge rate is about 3C for example. However, solar systems are generally designed for peak charge rates around 0.05C to 0.5C meaning it takes between 2hrs and 20hrs of full sun to fully charge the battery bank. The trend in the research is that cold temperatures and very high charge rates can cause problems. But what about low charge rates? There also seems to be some ongoing disagreement in the literature about how severe this problem is and whether the cells recover after returning to room temperature.

Here's the fundamental question I want to answer: What are the consequences of charging at .1C at 28°F? Is it significant? What charge rates are safe at what temperatures?

We can of course design systems to use energy to heat batteries or we can insulate better or we can just move to the tropics but I want to talk about what exactly happens when charging LiFePO4 at below 32°F. There are lots of opinions out there but I'm only interested in experimental results. Show me the DATA!

Here's a quote from this paper:
Meanwhile, charging the battery at low temperatures is likely to trigger lithium plating, which often leads to severe battery capacity fade. At −10 °C, an 11.5 Ah Liion cell was detected to have its capacity degradation rate increasing sharply when then charge current exceeds 0.25C and capacity loss can reach even 25% after 40 cycles at a charge rate of 0.5C [21]. Higher charge rate would lead to severer capacity loss. Even at 0 °C, a single charge cycle at 1C current would cause a 3.6% irreversible capacity loss of a 7.5 Ah cell [38]. Therefore, the charge rate of Li-ion battery is usually small at low temperatures in order to prevent lithium plating, which extends the charge durations dramatically

This paper tested capacity degradation at various temperatures for charging and discharging. They used 1C charge and discharge rates. The -20°C test degraded severely but quickly recovered when cycled one time at room temperature.
An additional observation based on the graph is the quite uncommon behavior at Tc = Td at -20 °C testing conditions. After each block of 25 cycles, there is a drastic decay of capacity and then a recuperation during the reference cycling (done at 25 °C).
Capture.PNG

This guys' thesis was testing lifepo4 battery that was going to be used in a satellite. He charged it at about 0.3C at -20°C:
The results showed that the battery can still operate properly at temperatures as low as -20°C. However a reduction of the capacity at low temperatures was observed: The battery's capacity at -20°C was 62% lower than the one at 20°C. The battery fully recovered its capacity after returning back to the nominal working range (20°C), and its performance barely changed thorough the whole range of temperatures studied.
He also noticed that the charging time was longer at lower temperatures:
Capture.PNG
The charge time increases as we lower the temperature due to the increasing internal resistance that lowers the charging current that goes through.
(he was charging at 3.5V per cell.
 
This paper tested capacity degradation at various temperatures for charging and discharging. They used 1C charge and discharge rates. The -20°C test degraded severely but quickly recovered when cycled one time at room temperature.

I don't agree with your characterization of the data based on this paragraph:

The behavior for cycling at (-20 °C, -20 °C) (Figure 1a) can be attributed to (i) kinetic restrictions during charging (a reduced ion diffusion, a deprived charge transfer resistance at the interface of electrode/electrolyte41, a reduced ion conductivity, a charge imbalance, etc.) and/or (ii) lithium plating when charging at low temperatures can quickly diffuse when cycling at high temperatures42. When the temperature is back to 25 °C, the ion diffusion is increased and there is an equilibration of the unbalanced state. This would lead to a capacity recovery. A similar behavior was not found in the literature. For the type of cells under investigation, this temperature combination is not recommended for a continuous cycling due to fast capacity decay, although there is some partial recovery of capacity after a certain recovery time at room temperature.

I agree 0°C isn't the line dividing a binary situation.

I've been struggling with this as it's suddenly gotten quite cold, and I'm not charging before noon on some days. Furthermore, the Victron GX device was permitting a very small trickle of current, 0.2-0.7A (0.002C max). it has now reverted to the prior behavior I witnessed when restricting charge current on my old FLA: 0.3A discharge while the MPPT powers loads as needed.

Without published low temp charging parameters from cell manufacturers, I'm hesitant to make any assumptions.
 
although there is some partial recovery of capacity after a certain recovery time at room temperature.
I was also a little confused by that paper, and I'm not sure exactly what they mean here. I'm not an expert in this field so some of the jargon and chemistry are over my head. That said, I'm confused why they say a "some partial recovery". If you look at the chart I posted (the blue dots) you can see the quick decay after less than 10 cycles, but then the 25th cycle is the "reference cycle" at room temperature and after that cycle, at cycle 25, 50, and 75, the capacity recovers to over 100% for a while until several more cycles at -20°C. I would call that a full recovery and then some. But maybe I'm missing something.

And of course the battery manufacturers are ultimately the ones who we would be going to for warranty claims and it's obviously simpler to just follow their recommendations. I wonder how much testing manufacturers do on this to come up with their recommendations? If a manufacturer could publish some charge rate guidelines for below 0°C they could get a huge competitive advantage with the DIY crowd.

I just want to learn more about the science as it develops and discuss the data as we find it. This is an emerging field in science and most of these papers are from the last 10 years. So if anyone finds more data out there please share it.
 
I just want to learn more about the science as it develops and discuss the data as we find it. This is an emerging field in science and most of these papers are from the last 10 years. So if anyone finds more data out there please share
Just curious, if you found good science to support this, would you charge your batteries in the cold against manufacturers specs?

I would not. Thousands of dollars and months of time went into building my two 8s battery packs. I’d rather spend the time instead of researching and risking, developing a heating pad.


I thought certain things could be put in during the manufacturing process to give a marginally colder charging cutoff, yttrium (Y) but can’t find a reference.
 
Here's a quote from this paper:
Does this paper's use of "li-ion" and not any of the specific chemistries (LFP, NMC or NCA) imply that their research applies to all chemistries (equally)?

>>>
Going astray:
And looking at the references from the "experts", is it just me or does it seem odd that many of the actual published research paper titles do not even get the spelling/chemical symbols correct?

Kinetic behavior of lifepo4/c cathode material for lithium-ion batteries

Current density and state of charge inhomogeneities in Li-ion battery cells with LIFePO4 as cathode material due to temperature gradients

Thermal analysis of a LiFePo4 Battery

These would not have escaped the scrutiny of even my high school chemistry teachers (~1977) and would not be considered correct. I have some published work and I'm certain it would not have been accepted with egregious errors like this.
Has education become less interested in getting the technical bits right?

</rant?>
 
It is about lithium-ion migration rate which gets more sluggish the colder the temperature.

Lithium ions become vulnerable when they get close to and enter the negative electrode graphite granules during charging. They need to quickly find a safe parking spot in the graphite lattice (intercalation) where their positive charge in shielded by electron charge carried in the graphite.

At cold temps the lithium-ion migration slows down causing the lithium-ions to be in the vulnerable region longer where they can combine with free electrons from the graphite becoming pure lithium metal. Once this happens it is over for the lithium contribution to battery function. The lithium is locked up as lithium metal with no further contribution to cell capacity.

Worse than the loss of free lithium, the lithium metal creates conductive dendrites that grow like conductive needles. They can grow in size to the point where they punch through the cell separator creating a short between positive and negative electrode.

0 degs C is not a hard line in the sand. The greater the charge current the greater the log jam of lithium-ions trying to find a safe parking spot in graphite and the longer the period of high vulnerability. It gets worse when approaching full charge where the remaining available parking spots in the graphite becomes limited.

Charging at low current and not charging above 80% capacity would allow lower temperature charging without significant damage.
 
Last edited:
It is about lithium-ion migration rate which gets more sluggish the colder the temperature.

Lithium ions become vulnerable when they get close to and enter the negative electrode graphite granules during charging. They need to quickly find a safe parking spot in the graphite lattice (intercalation) where their positive charge in shielded by electron charge carried in the graphite.

At cold temps the lithium-ion migration slows down causing the lithium-ions to be in the vulnerable region longer where they can combine with free electrons from the graphite becoming pure lithium metal. Once this happens it is over for the lithium contribution to battery function. The lithium is locked up as lithium metal with no further contribution to cell capacity.

Worse than the loss of free lithium, the lithium metal creates conductive dendrites that grow like conductive needles. They can grow in size to the point where they punch through the cell separator creating a short between positive and negative electrode.

0 degs C is not a hard line in the sand. The greater the charge current the greater the log jam of lithium-ions trying to find a safe parking spot in graphite and the longer the period of high vulnerability. It gets worse when approaching full charge where the remaining available parking spots in the graphite becomes limited.

Charging at low current and not charging above 80% capacity would allow lower temperature charging without significant damage.

You've consistently demonstrated a "substantially higher than average" understanding of the details associated with Lithium cells, materials, construction, operation, etc. As such, I have some questions:
  1. Most DIScharge specs are quoted at -20°C. Is this a recommendation, or does something "bad" happen as temperature drops?
  2. Can you speculate where a 0.1C charge might be considered safe? Just a best guess.
  3. How consistent are these qualities across the various lithium flavors, LFP, NMC/LMO, NCA, etc.?
  4. Do you have any knowledge of Panasonic produced HEV/PHEV cells used by Ford?
 
Just curious, if you found good science to support this, would you charge your batteries in the cold against manufacturers specs?
I would yes, but I certainly understand why things might be different for others. I live in a climate where we often have lows around 30F and it rarely gets below 25F. My batteries so far are all cheap Chinese 12V 100Ah batteries so I'm not expecting to have any manufacturer support anyway. You could say I'm already living on the edge by just using these batteries. ;) My peak charge rates will probably be 0.2C or below. I'm still designing this system so I'm still deciding whether I should add insulation or heat to the battery chambers, whether I need to find a charge controller that disables charging at low temperatures, etc. I understand that others have very expensive manufacturer supported systems and they have no interest whatsoever in "risking it". They may have no interest in this information. That's fine, this is not a personal attack on those people. It's just information, you can do with it whatever you like. :)
 
Does this paper's use of "li-ion" and not any of the specific chemistries (LFP, NMC or NCA) imply that their research applies to all chemistries (equally)?
That's a good point. I don't know.
Going astray:
And looking at the references from the "experts", is it just me or does it seem odd that many of the actual published research paper titles do not even get the spelling/chemical symbols correct?
Yeah I know there are many "low quality journals" as academics call them and I'm not well versed enough in this field to know which journals are high quality. So take it all with a grain of salt. I also personally have published in academic journals and I know I spent a lot of time getting things perfect. I also consider it a red flag if a paper makes mistakes in chemical symbols or units. I can understand if some of the grammar isn't perfect because most scientific research isn't done by native English speakers, but the science part I expect to be perfect.
 
I also consider it a red flag if a paper makes mistakes in chemical symbols or units. I can understand if some of the grammar isn't perfect because most scientific research isn't done by native English speakers, but the science part I expect to be perfect.
This is the second half of "don't believe everything you read on the internet". Part of figuring out what to believe involves looking at other things and your red flags are (part of) my red flags too.
 
Using just about any Li-Ion battery at lower temps is going to have significantly reduced performance.

Charging damage is a combination of temperature, charge current, and cell state of charge. Most manufacturers do not provide specific relationships on these three variables. It is a lot simpler to just say don't charge below 0 degs C which is conservatively safe.

With the exception of LTO battery, which takes its name from negative anode electrode, all lithium-ion batteries take their name based on positive cathode material used. Suffice to say LFP positive cathode is the most rugged cathode of all Li-Ion battery variants. LFP downside is lower cell voltage resulting in lower energy density.

Most all lithium-ion batteries operate in pretty much the same way. Positive cathode supplies the lithium and graphite negative anode stores the lithium-ions in the charged state. Electrolyte provides the lithium-ion relay exchange conveyor belt via the dissolved lithium salts in the electrolyte hydrocarbon solvent.

Common to all lithium-ion batteries is you do not want electrons from outside terminals to meet lithium-ions within the cell. It is the job of positive and negative electrodes to make this charge exchange without allowing electrons and lithium-ions to chemically copulate. When they do, and it does to some small extent, damage to cell is done. Throwing too many lithium-ions at negative anode too quickly with high cell current during charging will allow more forbidden interaction between electrons and lithium-ions. This is cold temp charging issue.

Each manufacture has some minor variant 'special sauce' added to electrolyte and 'pixy dust' added to electrodes. These provide targeted improvements for a particular parameter (usually at the degradation of another parameter). For example, an electrolyte additive may improve electrolyte lithium-ion mobility at low temperatures. The latest 'pixy dust' trend is to salt a bit of silicon dust to negative graphite electrode to bump up AH capacity of cell (and improperly call it a solid-state battery), or pre-lithiated salts to negative graphite electrode to reduce the manufacturing process of initial charge forming that builds the protective Solid Electrolyte Interface (SEI) layer and reduces the lithium consumed from cathode during this process which reduces yielded sellable cell capacity.

Cell damage and aging falls broadly into two categories of electrode damage and electrolyte damage. Charging at cold temp is in the negative electrode damage category. Overcharging is in the electrolyte damage category. Over-discharging eats at the copper and aluminum foil electrode current collector interface contaminating the electrodes. This is just a few examples of a very long list of possible ways to have cell damage.

One special mention is accelerated damage to SEI layer due to excessive charge and discharge currents. This eats up free lithium on subsequent recharging to rebuild damaged SEI protection. Fully charging bloats up negative electrode graphite by 11-13% in volume, stuffing it with lithium-ions resulting in some cracking of SEI protective layer grown around the graphite. The SEI layer helps keep lithium-ions separate from electrons in negative graphite electrode.

High cell current also requires high overpotential to drive the demand for lithium-ion migration. The greater overpotential within layers of the cell encourages many other damaging parasitic chemical reactions. Each layer of cell has a safe voltage potential gradient/temp range. Too much or too little overall cell voltage, too much cell current, or too high cell temperature may drive a given layer outside its safe operating range making that layer vulnerable to damaging parasitic chemical reactions.

The holy grail is a solid-state negative graphite anode replacement that allows negative electrode to hold more lithium-ions for given volume. Silicon can store five times more lithium-ions within the silicon lattice compared to graphite, but silicon eats so many lithium-ions the silicon bursts destroying the silicon structure. Result is very poor cycle life of less than 200 cycles. Solid state electrolyte is another big effort to accomplish.
 
I will follow this with interest. We've been starting our cars up here with lead acid batteries for decades but we know there are other chemistries available that are so much better.
 
I found a new paper. It took me a while to get access. Message me if you'd like the PDF. This paper built a model to predict the degradation at various temperatures, charge rates and state of charge. Then they tested cells at various conditions to fit and validate the model. They didn't talk much about the exact chemistry that they used except this, "In this work a high-energy Li-ion cell consisting of a graphite anode and NMC cathode was studied."

This is the key chart for our purposes:

Capture.PNG
This is exactly the chart I was hoping to find. It shows that according to their model, at -5°C, charge rates below about 0.3C aren't problematic and at -10°C the cutoff is more like 0.2C. I also find it interesting that just above 0°C charge rates above 0.5C would be problematic according to this model but is allowed by many battery manufacturers.

Here's an interesting quote:
For temperatures above 10°C, no harming states are identified for currents up to 1 C. Below 10°C the maximum possible non-harming current as a function of temperature can be identified. Additionally, only a small transition phase is observed between zero degradation and high degradation factors above 0.8. This
behavior has also been observed in experiment, when switching from one to another slightly higher current.

There was also a strong effect with state of charge:
Capture.PNG
This is interesting for us because usually when the sun first hits a solar panel in the morning, the batteries are not at high SOC because of battery usage since the previous sundown. Also as the sun starts hitting the panels it is also warming up the ambient air temperature and the batteries. So if you are using fairly low charge rates like 0.1C, by the time you get up to high SOC after a few hours the battery may have warmed some. Not a huge factor, but a slight tailwind that helps our solar use case.

And again they mention recovery mysteriously in the last line of the paper:
Further research is ongoing for implementing a complex Li plating side reaction. Recent internal studies have shown a reversible effect to a particular capacity fade which will also be incorporated in the model.
 

Attachments

  • Capture.PNG
    Capture.PNG
    89 KB · Views: 7
Here's another one that seems to correlate fairly well with the results from the paper in my last post.

Capture.PNG
c) normalised capacity vs. number of equivalent full cycles for cells charged at -10 °C with different current rates. d) linearly interpolated normalised capacity after 10 equivalent full cycles for cells tested at different temperatures and current rates. The crosses indicate the points of the conducted experiments that are used for interpolation. All cells shown in this figure are charged with a CV phase of 45 min, followed by a ‘defined discharge’.
 
I found a new paper. It took me a while to get access. Message me if you'd like the PDF. This paper built a model to predict the degradation at various temperatures, charge rates and state of charge. Then they tested cells at various conditions to fit and validate the model. They didn't talk much about the exact chemistry that they used except this, "In this work a high-energy Li-ion cell consisting of a graphite anode and NMC cathode was studied."

This is the key chart for our purposes:

View attachment 120815
This is exactly the chart I was hoping to find. It shows that according to their model, at -5°C, charge rates below about 0.3C aren't problematic and at -10°C the cutoff is more like 0.2C. I also find it interesting that just above 0°C charge rates above 0.5C would be problematic according to this model but is allowed by many battery manufacturers.

Here's an interesting quote:


There was also a strong effect with state of charge:
View attachment 120819
This is interesting for us because usually when the sun first hits a solar panel in the morning, the batteries are not at high SOC because of battery usage since the previous sundown. Also as the sun starts hitting the panels it is also warming up the ambient air temperature and the batteries. So if you are using fairly low charge rates like 0.1C, by the time you get up to high SOC after a few hours the battery may have warmed some. Not a huge factor, but a slight tailwind that helps our solar use case.

And again they mention recovery mysteriously in the last line of the paper:
Good info,

Question is if they tried to differentiate damage due to NMC cathode from graphite anode. Graphite negative anode damage for LFP would be similar. On many Li-Ion chemistries the cathode is the weaker link.

Charging is usually tougher on graphite anode, but most non-LFP cathode materials become structurally weak when cell is fully charged. This why most EV's don't want you to fully charge battery and folks see this info and apply it to LFP batteries which are not so vulnerable to fully charged cathode condition. (Other things, like electrolyte, are somewhat vulnerable to full charge, but the lower cell voltage of LFP gives more margin on safe electrolyte operating range.)

NMC is weaker cathode than NCA, both of which are weaker than LFP. LFP is almost bullet proof compared to other type cathodes, but everyone likes the higher cell voltage, giving greater energy density, compared to LFP.

Li Ion Cathode types.jpg

When looking at results taken from Li-Ion cells of other chemistries cathode material you have to be cautious that results may be due to cathode material differences.

Also have to consider electrode thickness. Thicker electrodes give greater AH capacity but causes problems with ion transfer access amplifying some damaging effects. Thick electrodes are not good for high peak cell current. The prismatic 'blue' cells used by most DIY'er are thick electrode design. Any chart showing degradation versus C(A)% current rate has to be scaled against electrode thickness for most damaging effects.

High peak current design electrodes are 30-50 um thick. High AH design electrodes are 100-150 um thick.
Li-Ion Graphite battery model.jpg
 
Last edited:
I also find it interesting that just above 0°C charge rates above 0.5C would be problematic according to this model but is allowed by many battery manufacturers.
Highly dependent on electrode thickness of given cell design.

For the typical thick electrode prismatic 'blue' cell used by DIY'er, should not be charging above 0.5 C(A) at any temperature.

For thick electrode cells, electrode lithium-ion local transfer rate starvation begins to happen on discharge above about 0.5 C(A). This drives up the overpotential required to move lithium ions (cell terminal voltage slump). Should avoid getting into ion starvation region as it drives up the voltage gradient across electrolyte increasing potential for damage.
LF280 overpotiential curve.png
LFP Over-potential Chart.png
 
Last edited:
This is exactly the chart I was hoping to find. It shows that according to their model, at -5°C, charge rates below about 0.3C aren't problematic and at -10°C the cutoff is more like 0.2C. I also find it interesting that just above 0°C charge rates above 0.5C would be problematic according to this model but is allowed by many battery manufacturers.
This paper's cell study is for NMC 4.0v cells, 1C discharge rate for 3600 secs to fully discharged = 1 AH cell, 65um thick graphite anode, 54um thick NMC cathode. This is a high current cell design capable of 4 to 6 C(A) maximum discharge current rate. Cells should be able to take a 3 to 5 C(A) charge rate at 25 degsC.

Discharge curves at 0 degs C ambient. Cell self-heating for 4 C(A) and 6 C(A) discharge rate make their curves have a 'funky' look (as cell warms itself up it has better performance).
NMC cell study.jpg
Principles apply, but the absolute C(A) cell currents stated will be significantly higher than thick electrode cells can tolerate.
It is a significantly different cell design than a typical 150 um thick electrode 'blue' prismatic LFP cell.
 
Last edited:
Back
Top