Bubble size in agitated solvent/reactant mixtures used in hetergeneous catalytic hydrogenation of 2-butyne-1,4-diol

Catalytic hydrogenations reactions are frequently conducted in "dead-end" multiphase stirred reactors with the reactant dissolved either in an alcohol, or in water or a mixture of the two. In such systems, the rate of gas-liquid mass transfer, which depends on bubble size, may well be the overall rate-limiting step. However, a study of bubble sizes across the whole range of solvent compositions from entirely water to entirely organic has not been reported. Here, for the first time, a systematic investigation has been made in a 3L, closed vessel simulating a "dead-end" reactor containing 1% by volume of air which is dispersed by a Rushton turbine in water, isopropanol (IPA) and mixtures of the two, with and without 2-butyne-1,4-diol simulating a reactant. Mean specific energy dissipation rates, 4, up to 50Wkg(-1) have been used and bubbles size distributions and mean size were measured using a video-microscope-computer technique. In the single component solvents (water, sigma similar to 72mNm(-1); IPA, sigma 21mNm(-1)) though the interfacial tensions are very different, irregular, relatively large bubbles of similar sizes were observed (d(32) similar to 300 mu m in IPA, and d(32) similar to 400 mu m in water) with a wide size distribution. In the mixed aqueous/organic solvents, and especially at the lower concentrations of IPA (1%, 5%, 10%), the bubbles were spherical, much smaller (d(32) from 50 to 70 mu m) with a narrow size distribution. The addition of the reactant (0.2 M 2-butyne-1,4-diol) to the mixed solvents had little effect on the mean size, shape or distribution. However, addition to water (thus producing a mixed aqueous/organic liquid phase) led to small (d(32) similar to 50 mu m) spherical bubbles of narrow size distribution. Neither Weber number nor surface tension was suitable for correlating bubble sizes since a decreased steadily from pure water to IPA whilst bubble size passed through a minimum at around 5% IPA. For any particular fluid composition, the functionality between d32 and epsilon T was similar, i.e. d(32) alpha (epsilon T)(-0.25). The above observations are explained in terms of the polarisation of bubble surfaces in miscible mixed aqueous/organic liquids caused by preferential directional adsorption at low concentrations of the organic component with its hydrophilic part directed into the aqueous phase and its hydrophobic part into the gas phase. As a result, coalescence is heavily suppressed in the low-concentration miscible alcohol (or diol)/aqueous systems whilst strong coalescence dominates bubble sizes in water and the alcohol and at high concentrations of the latter. (c) 2006 Elsevier Ltd. All rights reserved.