Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

3D Multiphase Heterogeneous Microstructure Generation Using Conditional Latent Diffusion Models

Nirmal Baishnab, Ethan Herron, Aditya Balu, Soumik Sarkar, Adarsh Krishnamurthy, Baskar Ganapathysubramanian

TL;DR

This work presents a conditional latent diffusion framework that generates high-resolution 3D multiphase microstructures on demand while targeting specific descriptors like volume fraction and tortuosity. By coupling a Variational Autoencoder, a feature predictor, and a latent diffusion model, the approach achieves fast, controllable synthesis in the latent space and simultaneously predicts manufacturing parameters to bridge digital design and fabrication. It demonstrates strong conditional accuracy (R^2 up to 0.93 on synthetic data and 0.89–0.77 on experimental OPV morphologies), substantial sample diversity, and applicability to both two-phase and three-phase systems at volumes around $128 \times 128 \times 64$. The framework enables accelerated materials discovery by providing application-specific microstructures and actionable processing guidance, with potential extension to a broad range of material systems beyond OPVs.

Abstract

The ability to generate 3D multiphase microstructures on-demand with targeted attributes can greatly accelerate the design of advanced materials. Here, we present a conditional latent diffusion model (LDM) framework that rapidly synthesizes high-fidelity 3D multiphase microstructures tailored to user specifications. Using this approach, we generate diverse two-phase and three-phase microstructures at high resolution (volumes of $128 \times 128 \times 64$ voxels, representing $>10^6$ voxels each) within seconds, overcoming the scalability and time limitations of traditional simulation-based methods. Key design features, such as desired volume fractions and tortuosities, are incorporated as controllable inputs to guide the generative process, ensuring that the output structures meet prescribed statistical and topological targets. Moreover, the framework predicts corresponding manufacturing (processing) parameters for each generated microstructure, helping to bridge the gap between digital microstructure design and experimental fabrication. While demonstrated on organic photovoltaic (OPV) active-layer morphologies, the flexible architecture of our approach makes it readily adaptable to other material systems and microstructure datasets. By combining computational efficiency, adaptability, and experimental relevance, this framework addresses major limitations of existing methods and offers a powerful tool for accelerated materials discovery.

3D Multiphase Heterogeneous Microstructure Generation Using Conditional Latent Diffusion Models

TL;DR

This work presents a conditional latent diffusion framework that generates high-resolution 3D multiphase microstructures on demand while targeting specific descriptors like volume fraction and tortuosity. By coupling a Variational Autoencoder, a feature predictor, and a latent diffusion model, the approach achieves fast, controllable synthesis in the latent space and simultaneously predicts manufacturing parameters to bridge digital design and fabrication. It demonstrates strong conditional accuracy (R^2 up to 0.93 on synthetic data and 0.89–0.77 on experimental OPV morphologies), substantial sample diversity, and applicability to both two-phase and three-phase systems at volumes around . The framework enables accelerated materials discovery by providing application-specific microstructures and actionable processing guidance, with potential extension to a broad range of material systems beyond OPVs.

Abstract

The ability to generate 3D multiphase microstructures on-demand with targeted attributes can greatly accelerate the design of advanced materials. Here, we present a conditional latent diffusion model (LDM) framework that rapidly synthesizes high-fidelity 3D multiphase microstructures tailored to user specifications. Using this approach, we generate diverse two-phase and three-phase microstructures at high resolution (volumes of voxels, representing voxels each) within seconds, overcoming the scalability and time limitations of traditional simulation-based methods. Key design features, such as desired volume fractions and tortuosities, are incorporated as controllable inputs to guide the generative process, ensuring that the output structures meet prescribed statistical and topological targets. Moreover, the framework predicts corresponding manufacturing (processing) parameters for each generated microstructure, helping to bridge the gap between digital microstructure design and experimental fabrication. While demonstrated on organic photovoltaic (OPV) active-layer morphologies, the flexible architecture of our approach makes it readily adaptable to other material systems and microstructure datasets. By combining computational efficiency, adaptability, and experimental relevance, this framework addresses major limitations of existing methods and offers a powerful tool for accelerated materials discovery.

Paper Structure

This paper contains 18 sections, 6 equations, 12 figures.

Table of Contents

  1. Introduction
  2. Results and Discussion
  3. Sampling quality
  4. Conditional sampling
  5. Diversity and prediction of manufacturing parameters
  6. Experimental Microstructures
  7. Conclusions
  8. Methodology
  9. Training dataset
  10. Generative model architecture
  11. Variational Autoencoder
  12. Feature predictor
  13. Diffusion model
  14. Training and inference
  15. Appendix
  16. ...and 3 more sections

Figures (12)

  • Figure 1: Samples from LDMs trained on (a) two phase and (b) three phase microstructures.
  • Figure 2: Conditional microstructure generation: Sample microstructures from user inputs - (a) Predominant phase A, and (b) Predominant phase mixed. First column shows the total microstructure. Second, third and fourth columns show the thresholded versions of the phase A, phase B and mixed components, respectively.
  • Figure 3: Statistical analysis of conditional microstructure generation: Correlations between all features of interest, user inputs, and the corresponding features measured from generated microstructures.
  • Figure 4: Variety of microstructures generated by the LDM given identical user inputs. The model can also suggests the manufacturing conditions required to generate such microstructures.
  • Figure 5: Statistical analysis of conditional microstructure generation: Correlations between all features of interest, user inputs, and the corresponding features measured from generated microstructures. Unlike the synthetic dataset we observe $R^2$ is below 0.9
  • ...and 7 more figures