Skip to content

[Bug]: Unphysical Elastic Constants for Silicon Using CHGNet via ForceFieldRelaxMaker #240

@adigasuhas

Description

@adigasuhas

Hi,

I'm trying to compute the elastic constants of silicon using the CHGNet model via ForceFieldRelaxMaker and the ElasticMaker from the atomate2 workflow.

Setup:
Structure: Silicon (mp-149)
Model: CHGNet
Workflow: ElasticMaker with ForceFieldRelaxMaker as the relaxer
Relaxation: fmax = 1e-5
No static DFT steps, fully force-field based

Issue:
The computed elastic constants are highly unphysical. Specifically, the diagonal components of the elastic tensor (C11, C22, C33) are ~ of order of -(1e6), while the off-diagonal and shear terms seem reasonable. I assumed the units are in GPa, but given the magnitude, it’s possible something else is wrong.

Questions:
Are the units of the elastic constants (from CHGNet or ForceFieldElasticMaker) in GPa, or something else (e.g., eV/ų)? What could be the reason for this unphysical value of elastic constants ?

I can share my code snippet if needed.

Thanks for your time!

Best regards,
Suhas Adiga

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type
    No fields configured for issues without a type.

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions