Compound sustainability metrics

[aronwalsh]

During the SMACT hackathon today, the topic of compound sustainability was raised. Currently, SMACT includes a sustainability metric based on the Herfindahl–Hirschman Index (HHI) from the 2013 paper by Gaultois et al. (1), which considers factors such as the supply risk of elements. While useful, the numbers are now over a decade old and do not capture the full picture of sustainability.

We’re considering expanding this functionality to include more recent and diversified indicators, such as:

• EU Critical Raw Materials (CRM) listings
• Elemental toxicity (e.g. known environmental or health hazards)
• Radioactivity (e.g. natural isotopic hazards)
• Abundance and recyclability metrics from newer databases
• Composite sustainability flags for user-defined screening

To implement this properly, we should determine which metrics are most relevant and how best to encode them in a modular and transparent way. We should also implement them properly at the composition (not element feature) level for ease of screening.

We welcome feedback from the community.

[ml-evs]

If more up-to-date HHIs would be enough, then there's some more useful work from Taylor Sparks and co. on digitising recent USGS reports (excluding US production/resource data...) in this PRX: https://doi.org/10.1103/PRXEnergy.4.023003 (and corresponding visualisation site at https://mineral-ai.net)

I also started trying to extract this data in the pre-LLM days but didn't get very far: https://github.com/ml-evs/usgs-commodities -- would be nice if we could collaborate on an authoritative source that could supplement the values in pymatgen for e.g., cost, HHI (https://pymatgen.org/pymatgen.analysis.html).

[aronwalsh]

Thanks for the rapid response! I was sure there would have been some progress made that we weren't aware of. I'll see if any other responses arrive and come up with a plan of action.

[rlaplaza]

Purely speculative points here since I'm coming from the molecular world, but some factors that seem missing here are degradation into the environment (moisture and air-sensitivity at least, which we should be almost ready to model in silico. In any case, stability is a sustainability metric in the truest sense of the word) and resources spent in synthesis, in particular things like chlorinated solvents which are then contaminated and hard to deal with.

Those are some of the things I think about when looking at molecular sustainability, at least.