AIchemy’s Monthly Webinar Series – Sept 25

We are delighted to welcome you to our AIchemy Hub’s monthly webinar series.

This month’s talks:

Prof. Rafael Gomez-Bombarelli, MIT Department of Materials Science and Engineering  

Talk Title: The bittersweet lesson of scaling in AI for materials

AI has the potential to bring much-needed acceleration to the development of chemicals and materials for energy and sustainability, just like it has delivered intelligence gains in other fields. The path to success goes through scaling, perhaps exclusively through scaling. Rich Sutton’s ‘bitter lesson’ states that “general methods that leverage computation are ultimately the most effective” in AI. The corollaries are bittersweet in chemistry and materials. 

The lesson permeates through the use of AI in chemistry and materials. The tremendous success of universal ML interatomic potentials (MLIPs) as surrogate for quantum-mechanical energies and forces is unarguable. Trained on 100-million-count datasets, they reflect scaling laws similar to the ones in language or vision models, generalizing to new scientific questions and enabling simulations that were intractable a few years ago. Generative models, trained mostly on the same synthetic data, are currently being used to propose novel materials at machine speed. The discussion around the power of inductive bias (energy conservation, equivariance) and whether it is better reflected in model architecture or in training strategy is very much ongoing. The physical sciences may well be the last holdout for domain knowledge and inductive bias, or maybe they will ultimately follow the same trends as other domains. 

But MLIPs are surrogates for physics-based simulators, and arguably, materials only truly matter if they are made in the lab and then scaled up industrially and commercially. This is the truly bitter scaling lesson in materials. The promise of AI for chemistry and materials needs to be realized through very traditional, expensive and slow channels. In this talk, I will describe our group’s work both in the highly scalable fusion of simulations and machine learning, and in the “high-contact” and lower throughput effort of translating AI designs into tangible, scalable products, in areas like heterogeneous catalysis, battery materials or sustainable polymers

Yushu Han – University of Liverpool

Talk Title: Accelerated Porosity Screening Using a Multichannel Colorimetric Array

Porous materials are important for many technologies, but the measurement of porosity by gas adsorption isotherms is slow, taking around one day per sample using a single-port gas sorption analyser, even when using a ‘quick’ analysis method with relatively few data points. With the increased use of automated platforms for material generation, porosity analysis is now frequently the bottleneck in the discovery of new porous materials. Here, we present a semi-automated pre-screening strategy that uses dye adsorption to create a colorimetric array that is combined with computer vision analysis for porosity screening. By using a six-dye multichannel array and a defined porosity threshold, our method rapidly screened 50 candidate materials that spanned molecular solids, polymers, and metal-organic frameworks. The method showed a 98–100% classification accuracy compared with gas uptake measurements. While this method is more qualitative than quantitative, it is more than 30 times faster than conventional gas sorption measurements, and it has the scope to be made much faster with greater parallelization and automation. This makes this colorimetric method suitable for pre-screening arrays of materials to choose samples that merit more detailed conventional porosity analysis.

Following the presentations, there will be time for questions from the audience.