Unsupervised bubble calorimetry analysis: Surface tension from isothermal titration calorimetry.

Hypothesis: The injection of air into the sample cell of an isothermal titration calorimeter containing a liquid provides a rich-in-information signal, with a periodic contribution arising from the creation, growing and release of bubbles. The identification and analysis of such contributions allow the accurate determination of the surface tension of the target liquid.

Experiments: Air is introduced at a constant rate into the sample cell of the calorimeter containing either a pure liquid or a solution. The resulting calorimetric signal is analyzed by a new algorithm, which is implemented into a computational code.

Findings: The thermal power generated by our experiments is often noisy, thus hiding the periodic signal arising from the bubbles' formation and release. The new algorithm was tested with a range of different types of calorimetric raw data, some of them apparently being just noise. In all cases, the contribution of the bubbles to the signal was isolated and the corresponding period was successfully determined in an automated way. It is also shown that two reference measurements suffice to calibrate the instrument at a given temperature, regardless the injection rate, allowing the direct determination of surface tension values for the liquid contained in the sample cell.