Welcome to HeteroFAM

At the cutting edge of materials research, HeteroFAM is a joint Pacific Northwest National Laboratory (PNNL) and University of Washington (UW) project designed to revolutionize the modeling, discovery, and understanding of complex material systems. By integrating advanced computational tools, machine learning, and automation, HeteroFAM enables breakthroughs in areas such as moiré materials, correlated electron physics and emergent quantum phenomena, and magnetic materials.

Rooted in the three-pillared modeling framework, HeteroFAM bridges first principles, empirical insights, and scalable workflows to tackle the scientific challenges of the 21st century.

Three-Pillared Modeling Framework

The three-pillared modeling framework is the backbone of the HeteroFAM project, combining three core approaches to tackle the most challenging problems in materials science, from moiré materials to quantum systems. These pillars—analytic models, data-driven models, and automation—work together seamlessly to provide a comprehensive and adaptable strategy for discovery.

1. Analytic Models (Rationalism)

Rooted in fundamental physical laws, analytic models are the cornerstone of the framework, offering mechanistic insights derived from quantum mechanics, thermodynamics, and classical mechanics. These models provide a deep understanding of the intrinsic behaviors of systems and enable reliable predictions based on first principles.

While analytic models often operate at the fundamental level, they can integrate phenomenological elements to address emergent phenomena and computational challenges across scales, particularly in multi-scale frameworks.

HeteroFAM’s Contributions:

• Advancing quantum mechanical models to accurately describe electronic interactions in materials, including moiré systems and transition metal oxides.
• Integrating first-principles simulations with phenomenological multi-scale models to explore complex, emergent behaviors such as high-temperature superconductivity, bridging quantum-scale phenomena with macroscopic properties.

2. Data-Driven Models (Empiricism)

Informed by the philosophy of empiricism, data-driven models utilize machine learning and statistical techniques to extract patterns and relationships from large datasets. These models are indispensable for exploring systems where analytical models may be computationally expensive or difficult to implement.

HeteroFAM’s Contributions:

• Training machine learning models to uncover hidden correlations in experimental data, enabling rapid discovery of new materials and correlated states.
• Automating the analysis of materials properties, optimizing design spaces based on both theoretical and experimental datasets.

3. Automation (Pragmatism)

With the pragmatic goal of enabling scalable, efficient, and reproducible research, automation integrates the analytic and data-driven pillars into dynamic, adaptable workflows. Automation enhances productivity by handling repetitive tasks, optimizing parameters, and seamlessly managing multi-scale simulations.

HeteroFAM’s Contributions:

• Creating high-throughput workflows that combine analytic and data-driven models, accelerating the discovery of new materials and quantum phases.
• Employing optimization algorithms, genetic algorithms, and AI-driven co-pilots to automate decision-making and improve efficiency in the exploration of complex systems.

Together, these three pillars create a robust framework that balances fundamental understanding with practical data applications and scalable automation. This holistic approach enables HeteroFAM to solve some of the most pressing challenges in modern materials science.

Get Involved

HeteroFAM invites collaboration with researchers across disciplines. Whether you’re exploring moiré materials, advancing reaction networks, or tackling quantum magnetism, our platform and expertise are here to support your work.

Key Focus Areas

• Thrust 1: Machine Learning Workflows for Correlated and Topological Moiré Systems

  Moiré materials, formed by stacking and twisting 2D layers with mismatched lattice constants, have emerged as a platform for discovering correlated electron states and topological phases. These systems exhibit highly tunable properties, such as superconductivity, magnetism, and exotic quantum phenomena.

  HeteroFAM’s Contributions:

  • Developing machine learning workflows that analyze electronic interactions and band structures in moiré systems.
  • Automating the discovery of new correlated and topological states by integrating high-throughput data with advanced computational models.
  • Enabling rapid exploration of parameter spaces, such as twist angles, external fields, and layer alignments.

• Thrust 2: Prototype Development for Correlated Electron Physics and Emergent States

  Understanding the physics of strongly correlated electrons is critical for discovering correlated electron states and topological phases and designing materials with unique quantum properties.

  HeteroFAM’s Contributions:

  • Creating prototypes to study emergent states in correlated materials, including:
    • Complex spin configurations
    • Quantum critical points
    • Novel magnetic phases
  • Build and refine next-generation models to understand and reliably predict complex quantum behaviors from first principles.
  • Reinforce and validate machine learning and AI-driven workflows while bridging the gap between theory and experiment.

• Thrust 3: AI-Driven Expert System for Spin Configurations in Transition Metal Oxides

  Transition metal oxides are key materials for applications in energy, catalysis, and spintronics. Their complex spin configurations present both challenges and opportunities for material design.

  HeteroFAM’s Contributions:

  • Designing an AI-driven expert system to predict and optimize spin configurations in transition metal oxides.
  • Integrating first-principles calculations with data-driven approaches to uncover trends in magnetic and electronic behaviors.
  • Streamlining workflows to design materials with specific magnetic properties for quantum computing and energy storage applications.

Tools and Technologies

• Advanced Algorithms:

  HeteroFAM leverages state-of-the-art computational algorithms to enhance the precision and efficiency of its modeling workflows. This includes genetic algorithms, machine learning techniques, and optimization frameworks designed to tackle complex problems in materials science.

• Computational Platforms:

  Utilizing cutting-edge computational platforms such as quantum computing and exascale computing systems enables HeteroFAM to perform large-scale simulations and handle computationally intensive tasks effectively, driving forward the boundaries of scientific research.

• EMSL Arrows:

  EMSL Arrows acts as the workflow engine behind HeteroFAM, integrating various modeling approaches and facilitating seamless collaboration across different research teams. This web-based platform ensures that complex simulations are accessible and manageable.