Data-driven analysis of neutron diffraction line profiles: application to plastically deformed Ta.

Area of Science:

• Materials Science
• Solid State Physics
• Crystallography

Background:

• Non-destructive evaluation of plastically deformed metals is crucial for material characterization.
• Diffraction line profile analysis (DLPA) estimates dislocation densities and validates constitutive models.
• Current methods like extended convolutional multiple whole profile (eCMWP) rely on semi-analytical models.

Purpose of the Study:

• Introduce and validate two data-driven DLPA models for extracting dislocation densities.
• Develop a method capable of distinguishing instrument broadening from dislocation content.
• Explore the impact of heterogeneous dislocation densities in polycrystals.

Main Methods:

• Generated virtual diffraction profiles using discrete dislocation dynamics and virtual diffraction models.
• Created Gaussian process regression-based surrogate models from a database of virtual profiles.
• Validated the data-driven models against experimental diffraction profiles of plastically deformed tantalum (Ta) polycrystals.

Main Results:

• The proposed data-driven DLPA models accurately predict dislocation densities, consistent with eCMWP estimates.
• The models effectively distinguish instrument broadening from dislocation-induced broadening, even at low dislocation densities.
• Demonstrated the model's capability to analyze heterogeneous dislocation distributions.

Conclusions:

• Data-driven DLPA offers a powerful, accurate alternative to traditional methods for evaluating plastically deformed metals.
• This approach enhances the understanding of dislocation behavior and material properties.
• The developed models pave the way for more precise predictions of dislocation density in complex microstructures.