18 April 2016 Supercritical CO2 mystery solved.

Supercritical CO2 (scCO2) is a fluid form of carbon dioxide that appears at high pressures (>73 bar) and high-ish temperatures (>32°C). It is said to be a "green" solvent because we have plenty of it and if you can dissolve something in scCO2 then retrieve the substance by simply lowering the pressure so that the CO2 becomes a gas. A classic example is decaffeination of coffee - instead of a solvent such as methylene chloride, the coffee experiences only harmless CO2 (and water).

But there's a problem. scCO2 isn't a great solvent. It's a bit like hexane - fine for hydrocarbons but not much good for interesting molecules such as caffeine. To fix this problem, it is common to add a small % of a real solvent such as water or ethanol, to provide a significant boost to solubility. After extraction, the CO2 is released and it's then reasonably easy to remove the remaining small amount of real solvent. These extra molecules are commonly called "entrainers". The problem with entrainers is that there have been at least three proposals for how they work, with no one being at all clear which (if any) was the correct idea. This, in turn, made it difficult to know what levers to pull to optimise the solubility via rational choice of entrainer.

We now know the true mechanism for the action of entrainers. My colleague, Dr Seishi Shimizu from U. York and I have just published a paper in the Journal of Physical Chemistry B called How Entrainers Enhance Solubility in Supercritical Carbon Dioxide. {DOI:10.1021/acs.jpcb.6b01380}. Seishi is a world expert in a type of statistical thermodynamics called Kirkwood-Buff (KB) and, using published data from the scCO2 literature, we could prove via KB theory that the mechanism was not due to density changes in the CO2 nor was it due to the formation of clumps of entrainer within the CO2. Instead we could show that it was due to favourable interactions between the entrainer and the solute.

Those who know me may wonder what I could contribute to the paper - after all, I'm not a world-class statistical thermodynamicist. Well, here's the problem. KB theory isn't actually all that hard. But it's not the most exciting thing to describe or to understand, and the steps from experimental data to insight into solubility mechanisms aren't user friendly. So a paper which simply presented the KB theory would not have a chance to make an impact on the user community. And as I've stressed in other blogs, I am fed up with academic work that could be useful but is unusable in practice because the academics have not provided the basic tools to allow non-experts to benefit from their insights. So we wrote an scCO2 app that brings the KB theory of scCO2 to life and which provides 16 datasets from the literature, each of which shows that the conclusions Seishi and I have arrived at are sound. There's a bonus. It turns out that the experimental data generally available don't provide enough information to nail all the variables in the system (too few equations for too many unknowns). So we have to make educated guesses about the values of these variables. Formulators might worry that we're fudging things and entering values that suit our theory rather than values which favour other theories. Although the paper provides convincing proofs that we're not fudging the issues, the user can choose to enter whatever values they think are correct for those variables which cannot be calculated with precision. The app quickly shows that there are no plausible values which undermine our conclusions.

So what we're doing with this paper is attempting a new form of academic publication. One that is intellectually high class (the paper underwent detailed scrutiny by the reviewers who suggested some helpful tweaks) but is also practically useful via the app. Even better, the app is public domain and the code can be downloaded and checked by those who have other views. For those who want to understand the inner workings of KB calculations, the code provides a fast-track education method.

Not surprisingly, Seishi and I have plans for more academic+app papers. The idea is so good, surely we're not the only ones attempting to do it. If we are, then it's time that academia woke up to the fact that we're in 2016. With the awesome power of modern browsers with HTML5 and Javascript, there is no reason apart from inertia for others not to be doing the same.