SorbentBase

What it is

Sorbent performance is reported inconsistently across the water-treatment literature. The same material appears with capacity in mg/g in one paper and mmol/g in another, measured under different pH, temperature and initial concentration, sometimes as an equilibrium isotherm and sometimes as a kinetic rate. Comparing two candidate materials usually means reading both papers in full.

SorbentBase harmonizes that. Every measurement is converted to a molar basis, tagged with the conditions it was taken under, and separated strictly by metric type so an isotherm capacity is never silently compared against a kinetic constant.

Why the provenance matters more than the size

Every record carries its DOI and the verbatim sentence it was extracted from. That is the part that makes the database usable for real work: you can check any number against its source without leaving the row, and a value you cannot verify is a value you should not design around.

The extraction pipeline runs full text from OpenAlex through a language model, then a deterministic harmonization layer. Model output is treated as an estimate, not a fact, which is why the quote travels with the record.

How it was validated

A subset of extracted rate constants was checked against the NDRL reference values and agreed on 12 of 13. That is a small sample and it is reported as one. A broader per-field precision audit is in progress and the results will be published as a table rather than a claim.

Who uses it

Process engineers screening sorbents for PFAS, heavy metals and organic contaminants before committing to bench work. Researchers looking for gaps in what has already been measured. Teams training models on sorbent structure and performance, where the citation trail matters as much as the numbers.

Access

The database is free and stays free. Full CSV download and a static JSON API, no key and no account. Open data with citations is the point of the project, not a funnel into a paywall.