Cheméo collects chemical and physical property data and puts it behind one search box. We have been running it since 2010. Today it covers over 100,000 molecules and 16,000 mixtures, with measured and predicted values for boiling and melting points, heat capacities, viscosity, flash points, flammability limits, enthalpies of formation and combustion, and many more.
It is run by Céondo GmbH in Germany. There are no adverts, no trackers, and no paywall on the data. You search, you get the numbers, and you can see where each one comes from.
Every value in Cheméo comes from a named source, and we keep the link back to that source for each individual data point. The main ones are:
Some sources are single publications. Some are whole databases built by others. We tag each one, so for any value we know whether it is original experimental data, a single-source transcription, or an aggregate.
Merging property data from this many sources is hard work, and the only way to do it without making a mess is to do it systematically. Nothing is a one-off.
The data goes through a fixed set of steps. We keep the raw files as close to the original as we can, each with a metadata sidecar: where it came from, the URL, the DOI, the download date, the original publication date, and which parser reads it. A dedicated parser turns each source into one common internal format. Then comes reconciliation: matching every record to the right compound and giving it a single Cheméo ID, so the boiling point from ThermoML and the boiling point from the WebBook land on the same molecule instead of two near-duplicates.
Each source also carries a confidence level. When two sources disagree, the confidence decides which one is shown first, and you can still see both. Because every value keeps its source key, you can ask Cheméo to show you only the data from one specific source.
Not every number is measured. For the properties that allow it, Cheméo also computes estimates, for example with the Joback group-contribution method. Predicted values are labelled as predictions. They are never mixed in silently with experimental data: if it is an estimate, it says so.
A free account gives you the API and Excel export.
If you use Cheméo data in your work, please cite it. The exact wording for a website, an application or a paper is on the terms and conditions page, which is short and written in plain language. And when a value came from a specific measurement, the original source matters more than we do, so cite that one as well. We keep the link to it on every compound page for exactly this reason.
The database is free to search and to use value by value, but it is not for bulk download. The content is protected by copyright and the sui generis database right, and we have had companies lift the whole thing without a word of credit. If you need the full dataset, just ask us. We want the data used, not abused.