BOLUDO

Code accompanying the paper A Holistic Data-Driven Approach to Synthesis Predictions of Colloidal Nanocrystal Shapes

Bayesian Optimization for nanocrystaL strUcture Design Optimization
But really, it's the work of Bojana Rankovi and Ludovic Zaza, under the supervision of Prof. Raffaella Buonsanti and Prof. Philippe Schwaller, bringing Bayesian optimization into the chemistry lab and discovering new nanocrystal morphologies! Also, we fully acknowledge the meaning of "boludo" in Argentine Spanish slang - turns out even crystals can be shaped by a couple of friendly boludos.

Overview

Figure 1: Complete workflow showing ELN integration, BO model fitting, and experimental validation

BOLUDO is a machine learning framework that revolutionizes nanocrystal synthesis by: - Predicting nanocrystal shapes from reaction conditions - Suggesting optimal reaction parameters for target shapes - Operating effectively with limited data (<200 experimental points) - Enabling discovery of new nanocrystal shapes through continuous energy scale mapping

Model Architecture

The system consists of three main components:

1. Data Processing Pipeline

   • ELN data extraction
   • Feature engineering
   • Synthesis parameter standardization

2. Machine Learning Models

   • Random Forest for interpretable predictions
   • Gaussian Process for Bayesian optimization
   • Surface energy scale mapping

3. Optimization Framework

   • Bayesian optimization for parameter space exploration
   • Multi-objective optimization capabilities
   • Uncertainty quantification

Figure 2: Visualization of parameter importance in nanocrystal shape prediction

Results

Our framework has achieved significant breakthroughs: - Successfully predicted synthesis conditions for various Cu nanocrystal shapes and vice versa - Discovered novel synthesis pathways - Achieved first-time synthesis of Cu rhombic dodecahedron shape - Demonstrated effectiveness with only 115 initial data points

Figure 1: Complete workflow showing ELN integration, BO model fitting, and experimental validation

Installation

The most recent code and data can be installed directly from GitHub with:

bash $ pip install git+https://github.com/schwallergroup/boludo.git

Citation

bibtex @article{zaza2024holistic, title={A Holistic Data-Driven Approach to Synthesis Predictions of Colloidal Nanocrystal Shapes}, author={Zaza, Ludovic and Rankovic, Bojana and Schwaller, Philippe and Buonsanti, Raffaella}, journal={Journal of the American Chemical Society}, year={2024}, doi={10.1021/jacs.4c17283} }

Acknowledgments

This work was supported by NCCR Catalysis, a National Centre of Competence in Research funded by the Swiss National Science Foundation (grant number 180544).

Contributing

Contributions, whether filing an issue, making a pull request, or forking, are appreciated. See CONTRIBUTING.md for more information on getting involved.

Attribution

License

The code in this package is licensed under the MIT License.

Cookiecutter

This package was created with @audreyfeldroy's cookiecutter package using @cthoyt's cookiecutter-snekpack template.

For Developers