https://doi.org/10.1016/j.seppur.2020.118193
“Viscosity in water-lean solvents is usually higher than that of aqueous ones. This is particularly true for water-lean solvents with low volatility. Mixtures of amine and methanol, for example, have comparatively low viscosity when unloaded, but their viscosity increases steeply upon CO2 absorption [59]. Amine and ethanol mixtures also have low viscosities when unloaded [161], which is only natural as both methanol and ethanol have lower viscosity than water. For formulations with diluents such as N-methyl-2-pyrrolidone and ethylene glycol, the unloaded mixtures have higher viscosities than those of aqueous solvents, and loading with CO2 increases them even more. This has been observed also by Yuan and Rochelle [18], Bougie et al. [162] and by Guo et al. [80]. The liquid crystal MEA-based solvent developed by Rodríguez-Fabià et al. [92] shows an unbelievable increase of viscosity with loading that ultimately ends with a phase transition towards solid crystal. Unfortunately, though there is a huge amount of published data for the viscosities of unloaded water-lean solvents, little is reported on their viscosities when loaded. As such, we are unable to find any exception to this trend in literature.
In an interesting theoretical work, Esteves et al. [163] have demonstrated how the viscosity of liquid mixtures of electrolytes have a behavior, according to the Debye-Hückel approach, that follows the following expression.lnη∝I3/2ε3/2
In the expression above, I is the ionic strength of the electrolytes and ε is the dielectric constant of the medium. Notice that the ionic strength of an amine solvent generally varies between zero and the molarity of the amine itself (when every amine molecule has been protonated), with half the molarity of the amine being a typical value in the case of carbamate formation. A solvent containing 30 %wt. monoethanolamine (MEA) will contain typically around 4.9 mols per liter of MEA, and its ionic strenght will vary from 0 mol/l (unloaded) to 2.45 mol/l (maximum carbamate formation) to 4.9 mol/l (maximum bicarbonate/carbonate formation). The ionic strengths of loaded amine solvents are undoubtedly higher than those for which the Debye-Hückel approach is valid, which account solely for long-range interactions between elecrolytes, typically below I = 1 mol/l. To this effect, Esteves et al. [163] also propose the Guggenheim correction (dependent on I2 and independent on ε) to extend the range of applicability of their equation. Nonetheless, in general terms, the more loaded with CO2, the more ions in solution a solvent will have, and therefore the higher its ionic strength. Similarly, for solvents at a fixed ionic strength (which can be loosely correlated to having the same CO2 loading), lower dielectric permittivities imply higher viscosities. As discussed previously in Section 4.1, water has the highest of all dielectric permittivities among diluents for amine solvents. It can be suggested that every possible water-lean solvent will, due to electrostatic phenomena, inevitably experience steeper viscosity increase with loading.
Fig. 12 illustrates the issue being discussed here with data for aqueous MEA and for nonaqueous 2-methoxyethanol plus MEA, both at 30 %wt. concentration of the amine and 40 °C. The data was obtained by Amundsen et al. [164] and by Guo et al. [80]. One can easily observe how drastic is the increase in viscosity through CO2 loading with water-lean solvents when compared to aqueous ones.”
“Fig. 12. Viscosity increase with loading for aqueous 30 %wt. MEA and 2-methoxyethanol + 30 %wt. MEA at 40 °C, plus demonstration of its dependency on ionic strength. Data obtained from Amundsen et al. [164], Guo et al. [80] and Wohlfarth [165].”
“We have also exemplified the relationship between viscosity, ionic strength and dielectric permittivity outlined by Esteves et al. [163] in Fig. 12 by treating the data obtained by Amundsen et al. [164] and Guo et al. [80]. The dielectric permittivities of water and 2-methoxyethanol at 40 °C (respectively 73.15 and 16) were found in Wohlfarth [165], though it must be pointed out that in deriving Fig. 12 we have employed only the ε of the pure diluents, with no considerations on the effects of the ε of MEA itself. The parallel lines were forced to be parallel by performing a simultaneous linear regression on the data for both solvents (i.e. only one angular coefficient was calculated with the double set of data). Regardless of this mathematical trickery, it does look like this approach has some depth to it. Therefore, it might be fair to suggest that evaluation of viscosity of loaded and unloaded solutions should be prioritized when developing a water-lean solvent formulation. Moreover, the Pearson correlation coefficient between (I/ ε)3/2 and ln(η) for aqueous MEA is R((I/ ε)3/2,ln(η)) = 0.9823, whereas that between I2 and ln(η) is R(I2,ln(η)) = 0.9738. These are inconclusive results for evaluating whether the Guggenheim extension is necessary to aid the Debye-Hückel approach or not, though this is not the main focus of this discussion.”