https://doi.org/10.1016/j.seppur.2020.118193
“In principle, there are little reasons to suspect that the mechanism of reaction in water-lean solvents containing the usual primary and secondary alkanolamines will be any different than that in aqueous solutions. The formation of a carbamate species between the amine and the CO2 molecule is often the most thermodynamically favorable among the possible reactions that can happen given the set of chemicals available in either aqueous or water-lean systems. This is so self-evident that early studies on water-lean solvents did not consider an alternative mechanism of absorption before proceeding to calculate kinetic data [37], [38], [39], [40], [41]. And they were most certainly right in doing so, as recent spectrography publications [42] have shown that carbamate is the predominant reaction product at moderate CO2 loadings (α < 0.5 mol CO2∙mol MEA−1) and when using strong bases (pKa > 8), conditions typically encountered in those early studies. Under different conditions, Kortunov et al. [42] show that carbamic acid or even undissociated zwitterion might be present upon CO2 absorption. Similarly, Masuda et al. [43] carried an extensive NMR research on carbamic acid formation when working with mixtures of arylamines and organic solvents. The experimental findings from these authors confirm, and greatly expand, what had been proposed by Takeshita and Kitamoto [44] regarding the effects of solvent polarity and amine basicity on the products of reaction between CO2 and amines. However, under typical commercial conditions and employing typical amines, it is safe to assume that carbamate formation is the main mechanism of reaction.
At the same time, other authors have found NMR evidence of alkylcarbonate formations when employing water-lean solvents with hindered and tertiary amines [45], [46], [47], [48], [49]. Incidentally, alkylcarbonate formation has long been proposed to explain the reactive absorption of CO2 in nonaqueous tertiary amines [50], though not experimentally verified until very recently. Behrens et al. [51] observed substantial alkylcarbonate formation between CO2 and MDEA in aqueous solutions via NMR analyses, making it likely that a similar occurrence should happen in nonaqueous media. It is important to differentiate these two observations though. While Barzagli et al. [45] and S. Chen et al. [46] report carbonate formation between the diluent (e.g. ethylene glycol, ethanol) and CO2, what is shown by Behrens et al. [51] is the formation of a carbonate with MDEA itself. Conversely, both carbonate formation between alcohol and CO2 and between amine and CO2 have been shown to take place in the experiments carried out by Skylogianni et al. [47] and by Wanderley et al. [49].
These reaction mechanisms were schematically represented on Fig. 2. In the last reaction, one should notice that the nucleophile is the deprotonated hydroxyl group of the amine. For this reason, Jørgensen and Faurholt [52] had assumed that this reaction cannot take place unless the environment is strongly basic. There are alternative interpretations of this mechanism, some of them involving intramolecular proton transfer [53] or direct reaction with bicarbonate [54]. We will discuss the alkylcarbonate formation pathway a bit more in Section 10. For now, suffice it to say that most interpretations on water-lean solvents rely solely on the carbamate route, which is after all the predominant one for most low-pressure applications.”
“Fig. 2. Possible mechanisms of reaction for CO2 absorption by an alkanolamine, exemplified with the case of AMP.”
”
Beyond these general observations, there are also exceptions in the behavior of some particular solvents, which possibly taking part in the reactions themselves. For example, N-methyl-2-pyrrolidone is suspected of reacting in a similar way as a tertiary amine does [55], slightly increasing the absorption rates of CO2. Shannon and Bara [56] observed a similar phenomenon with N-alkylimidazoles. Meanwhile, ethylene glycol and other polyalcohols possibly undergo some sort of reaction as well, perhaps even the alkylcarbonate one mentioned above [26], [57]. All of these are examples of reactions that would increase CO2 absorption. On the other end of the spectrum, propylene carbonate has been observed to react with amines a number of times [22], [32], [34], and a reaction between ethylenediamine and ethylene glycol renders the solvent prepared with both these chemicals virtually unreactive [58]. Another series of self-destructive side-reactions has been reported previously in Wanderley et al. [59] involving esters and ketones.
Finally, there is one particular issue that must be addressed when mentioning the experimental research on reaction mechanisms in water-lean solvents. Most properties analyzed in this review obey, loosely speaking, a very predictable behavior when one transitions from aqueous solvents to semiaqueous solvents and finally to nonaqueous solvents. This will be further explored in the following sections, where Fig. 4, for example, shows that the equilibrium reaction between CO2 and amine follows a smooth trend with the increase of water mass fraction in water-lean solvents containing N-methyl-2-pyrrolidone. That is to say, the effect of trace amounts of water present in the solvent during equilibrium experiments and mass transfer rate experiments is not expected to deliver unpredictable results. And yet, this is not necessarily the case with speciation experiments, where very little amounts of water are sufficient to enable the formation of bicarbonate. For example, let us consider the case of 50 %wt. N-methyldiethanolamine (MDEA) in ethylene glycol. Those are equivalent to approximately 0.420 mol of MDEA per 100 g of solvent. For the bicarbonate formation mechanism to be enabled in a stoichiometric basis, one requires one mol of water for each mol of MDEA, which means that 7.55 g of water in 100 g of solution are sufficient to enable every molecule of MDEA to be converted to bicarbonate and protonated MDEA upon CO2 absorption. This is particularly troubling in the case of ethylene glycol, which is a known hygroscopic substance liable to fixate the moisture of the air. Thus, relatively small amounts of water are theoretically enough to shift the speciation behavior of a water-lean solvent, and speciation experiments should be carried out with a particular attention to detail.”
“Fig. 3. VLE data obtained by Dugas and Rochelle [61] (aqueous MEA) and Yuan and Rochelle [18] (MEA-NMP blends) for solvents containing different NMP-water proportions and 7 molal MEA at 40 °C. The proportions of NMP and water are given in mass basis.”
“Fig. 4 Equilibrium constants adapted from data by Dugas and Rochelle [61] and Yuan and Rochelle [18] for 7 mol MEA∙kg solvent−1 at 40 °C shown against water/diluent mass fractions and the dielectric permittivities of the diluent.”