Is SEI Layer growth temperature independent??

[mlnsai]

Hi Team,

I have been conducting simple cyclic testing simulations for 350 cycles at different temperatures. (15C, 30C, 45C)

Here is the cycle

    experiment = pybamm.Experiment(
        
        [
            "Charge at 1 C until 4.2 V",
            "Hold at 4.2 V until C/10",
            "Rest for 5 minutes",
            "Discharge at 1 C until 2.8 V",
            "Rest for 5 minutes",
        ]* 350
    )

Here are the options

model = pybamm.lithium_ion.DFN(options=
{"thermal": "x-lumped", 
"SEI":"solvent-diffusion limited", 
"SEI porosity change" : "true", 
"lithium plating": "reversible"
})

Interestingly, the results I have got indicates that SEI Layer growth is temperature independent. But literature suggests the temperature dependency on SEI Layer growth. Whys is this happening in PyBaMM?

Any suggestions @brosaplanella @tinosulzer ? :)

Here I have attached the result I have got. Also attached the spreadsheet for the data.
SEI_Analysis_Final.xlsx

[brosaplanella]

That's probably because the parameters you use for SEI do not depend on temperature, so the behaviour is not captured even if the temperature changes in the model.

[mlnsai]

Thanks @brosaplanella for the reply. I am using the default SEI parameters and I am not updating/calling any parameters specifically.

The model I selected is "solvent-diffusion limited" and it is the same result for "ec reaction limited" as far as temperature independency is concerned. I have further enabled "SEI porosity change" : "true" option.

It would be great if you could point out those parameters which can capture the SEI dependency on temperature. I would like to re-run incorporating this temperature impact on SEI layer growth.

[brosaplanella]

If you don't explicitly pass parameter values the model uses the default parameter values (you can get them doing model.default_parameter_values). I am not aware of any parameter sets fully capturing the temperature dependence of SEI growth, so you should double check the published literature.

[mlnsai]

Hi @brosaplanella , thanks for the suggestion.

I did pass "Chen" and "Ramadass" and "Ai2020"parameter sets as of now. These didn't capture the temperature dependency. If you have any suggestion about any parameters that can capture the temperature dependency, kindly let me know.

We are also trying from our end. I will keep you posted.

[brosaplanella]

At least Chen2020, which is the one I am familiar, uses some literature default SEI parameters which probably don't capture temperature dependence well. I might look into parameterisation of such models in the near future, will keep you posted.

[mpegis]

If its solvent-diffusion limited, I imagine a T-dependent diffusivity of EC would solve your problem. Is your diffusivity T-dependent?

[mlnsai]

Hi, @mpegis, thanks for the reply.

But "EC diffusivity [m2.s-1]": 2e-18," this is constant for all parameter sets in PyBaMM. Is there any equation that actually makes this EC diffusivity temperature dpendent?

Thanks in advance

[mpegis]

I believe you should be able to put in a T-dependent function for that parameter although I could be wrong. You could find a literature T-dependent NMR study (e.g. https://pubs.acs.org/doi/10.1021/je3003089) and start from there, although you could also just manually change it for each temperature.

I would say generally, the fact that pybamm can model T-dependent kinetics requires parameters which contain T-dependence such as exchange current density, Li ion diffusivity, conductivity/etc. So you would want to modify your SEI submodel to contain best guesses for what could be happening.

[mlnsai]

Hi @mpegis , It did worked by making the diffusivity a temperature dependent function which changes with ambient temperature.

An Arrhenius dependent function that takes ambient temperature as input worked just fine.

The temperature-dependent diffusivity property is descripted by this Arrhenius equation

Here's the change I have done in code.

If it's an EC reaction limited SEI submodel.

    def ec_diffusivity(T_amb):
        D_sj = 2e-18,
        EaD = 4000
        arrhenius = exp(EaD / constants.R * (1 / 298.15 - 1 / T_amb))
        return D_sj * arrhenius
        parameter_values.update({"EC diffusivity [m2.s-1]":ec_diffusivity(tempK)})

If it's an Solvent-diffusion limited SEI submodel.

    def sd_diffusivity(T_amb):
        D_sd = 2.5e-22
        EaD = 4000
        arrhenius = exp(EaD / constants.R * (1 / 298.15 - 1 / T_amb))
        return D_sd * arrhenius
    parameter_values.update({"Outer SEI solvent diffusivity [m2.s-1]":sd_diffusivity(tempK)})

Here's a snapshot of the output

Thanks @mpegis and @brosaplanella for timely inputs.

[Ansh-Lodhi]

I was going through the source code of SEI growth sub Model and I found this :

Apparently, SEI growth is already dependent on temperature. Can somebody confirm this?

[brosaplanella]

Yes, but it only captures the Arrhenius dependence. Plus, you need to pass a non-zero value for the activation energy, otherwise the effect is ignored.

[ganmad]

Hello,

How to introduce a temperature dependence on diffusivity locally on notebook? Recent updates are throwing an error. Please check below minimum example. Did I miss something? Thank you.

import pybamm
model = pybamm.lithium_ion.SPM({"cell geometry": "arbitrary","thermal": "lumped","SEI": "ec reaction limited","SEI film resistance":"distributed",
        "SEI porosity change": "true",})
param = pybamm.ParameterValues("OKane2022")


def ec_diffusivity(T_dim):
       
    D_sj = 2.5*10**(-22)
    EaD = 30000
    arrhenius = pybamm.exp(EaD / pybamm.constants.R * (1 / 298.15 - 1 / T_dim))
    return D_sj * arrhenius

param.update({"EC diffusivity [m2.s-1]":ec_diffusivity})

exp = pybamm.Experiment(["Discharge at 0.1C until 2.5 V"])
sim = pybamm.Simulation(model,parameter_values=param,experiment=exp,)
sim.solve()

TypeError: Cannot process parameter '<function ec_diffusivity at 0x7fc0f89e6170>'

[mlnsai]

Hi, try giving ambient temperature in kelvin (tempK) as the input to the function while calling

param.update({"EC diffusivity [m2.s-1]":ec_diffusivity (tempK)})

[ganmad]

Hi @mlnsai, Thanks for the response.

Do you mean calling a variable like below?

tempK = model.variables(["Cell temperature [K]"])  or tempK = model.variables(["Ambient temperature [K]"])

param.update({"EC diffusivity [m2.s-1]":ec_diffusivity(tempK)})

[MirAbbasAli2A]

@ganmad, could you please tell me where you got the EaD of 30,000 from?

[ganmad]

Introducing temperature dependence is not always straight forward. I think one should introduce new functions as hinted in this answer:
#3103 (comment)

Need to check whether the parameter is defined using Parameter class or FunctionParameter class. I understand EC diffusivity [m2.s-1] is defined using Parameter class, hence we can't introduce temperature dependence locally. I still have to figure out how to implement though. Let me know if I am wrong. Thanks.

[brosaplanella]

Yes, EC diffusivity currently is a constant, so not temperature dependent. Again, it should be fairly easy to fix this so feel free to try and tackle it. Let me know if you need any help.