Perspectives of CO2 absorption using water lean solvents

https://doi.org/10.1016/j.seppur.2020.118193

“In our historical review (Section 2), we have pointed out how the perceived benefits from employing water-lean solvents have changed with time. Hybrid solvents were created with the intention of increasing the absorption capacity of CO₂ under high pressures. Later, as the enthusiasm for this aspect of water-lean mixtures started to subside, there came a revived interest in the promised energetic savings that these solvents might deliver. Now, it has become more and more clear that amine systems are approaching an apparent minimum of energy necessary for CO₂ absorption–desorption cycles [166], [214].

Some interesting perspectives for water-lean solvents in the future are related to their use to bypass the carbamate route and absorb CO₂ in a different manner. As discussed in Section 5.1, the alkylcarbonate route of CO₂ conversion has been experiencing a renewal in interest lately. Recent NMR spectrography analyses have identified alkylcarbonate pathways even in aqueous primary alkanolamines and well understood amines such as MDEA [42], [51], [54], [128], [131]. These pathways attract the attention of researchers especially due to their potential for generating reaction products that can be reversed more easily, i.e. with less energetical input, or at least using a cheaper source of energy by desorbing CO₂ at lower temperatures.

Water-lean solvents might have a role to play in developing alkylcarbonate-forming solvents. Eimer [50] has proposed that the alkylcarbonate pathway is more likely to happen when using organic diluents with high autoprotolysis constants and weak amines. His example was that of MDEA mixed with ethylene glycol, which has a pK_S = 15.84 at 25 °C [138]. In an interesting study, Takeshita and Kitamoto [44] demonstrated how solvent polarity can determine whether the products of the reaction between CO₂ and amines are mostly carbamic acid, undissociated zwitterion or carbamate and protonated amine.

Naturally, Takeshita and Kitamoto [44] were able to carry this study by using nonpolar organic amines instead of alkanolamines. Alkanolamines would very probably demix either instantly or upon CO₂ absorption in a solvent as nonpolar as octane, which is what they have employed. However, this serves as a motivation for understanding that water-lean solvents might be practical precisely for enabling the use of these very same amines, with their own particular CO₂ reaction routes. Other researchers who have taken the opportunity to work with nonpolar amines in water-lean solvents were Yogish [183], Dinda et al. [136], [137] and Gómez-Díaz et al. [185]. Gómez-Díaz et al. [185] were interested in using tributylamine for its high stability with regards to thermal degradation. Conversely, Dinda et al. [136], [137] were simply developing a process for the production of isocyanates and urethanes.

Alkylcarbonate formation was observed by Jing et al. [184] in mixtures of dibutylamine, water and ethanol. In their case, ethanol was essential for avoiding biphasic phenomena, since pure dibutylamine is immiscible in water. Their solvent absorbed CO₂ up to a loading of 0.82 mol CO₂∙mol amine⁻¹ while its viscosity barely increased from 1.56 mPa∙s to 1.72 mPa∙s. This is even more interesting than their reported higher regeneration efficiency. It is puzzling that this viscosity increase is so small. The authors themselves have suggested that the presence of water in the solvent might have something to do with this. Indeed, their formulations seem to suffer more with mounting viscosities upon loading the higher the ethanol-to-water ratio in the diluent.

Furthermore, there are some groups that should be addressed individually. The first is that of the University of Florence. This team has been working with water-lean solvents for almost a decade, and an important part of their research is focused on mixing hindered or tertiary amines (e.g. AMP and MDEA) with organic solvents such as ethylene glycol/ethanol, ethylene glycol/1-propanol, and DEGMME, purposefully promoting alkylcarbonate formation [6], [45], [129], [130], [224]. Conversely, alkylcarbonates were not observed in nonaqueous solvents with amines such as DGA and DEA [5]. They also have employed organic diluents to force demixing and precipitation of products upon CO₂ absorption [225], [226], [227], and studied CO₂ absorption in solvent-free amines [228] (one could call this an extreme example of water-lean solvent). In our opinion, it is the pioneering work of this group in NMR spectrography and alkylcarbonate identification in amino-organic mixtures that stands out as essential literature on water-lean solvents.

The group from RITE [229], [230] and the group from RTI International [116], [231], [232] have both carried researches on proprietary water-lean solvents in the past decade. Yamamoto et al. [229] developed a mixture between a tertiary amine and an organic diluent that has high absorption capacity and can be regenerated at high pressures. Because of this, their formulations are called HPRT (High Pressure Regenerative Type) solvents. Meanwhile, Lail et al. [232] designed a mixture of a hindered amine with a solvent with low volatility which they have nicknamed NAS (Non Aqueous Solvents). Lail et al. [232] report that their solvent absorbs CO₂ through the carbamate route and that its heat of absorption at desorber temperatures is lower than at absorber temperatures. In both cases the authors refuse to disclose precise information on which chemicals they have employed. Conversely, Mobley et al. [116] and Tanthana et al. [231] developed their solvent-based on a very complex fluorinated amine (2-fluorophenethylamine) mixed into a fluorinated alcohol (2,2′,3,3′,4,4′,5,5′-octafluoropentanol). Their solvent consciously avoids the carbamate formation route, and its curious heat of absorption behavior, perhaps similar to that of the NAS of Lail et al. [232], has already been mentioned in Section 8.1. These are called HPS (Hydrophobic Solvents). The HPRT solvents, the NAS and the HPS show that, decades after the Amisol® and the Sulfinol® processes, development of water-lean solvents seem to have increased in complexity and sophistication.

Finally, the CO₂BOLs (CO₂-Binding Organic Liquids) developed by the group from the Pacific Northwest National Laboratory are certainly worth mentioning due to their very interesting properties. They were initially designed as nonpolar mixtures of alcohols and amines that, upon CO₂ absorption, become highly polarized ionic liquids [4], [233]. These amines are guanidines and amidines. Later, both diluent and reactant were synthesized into single molecules, alkanolamidines and alkanolguanidines, for the 2nd generation CO₂BOLs¹ [3], [234], [235], [236], [237]. The reaction products in both generations are always alkylcarbonates.

On the bright side, CO₂BOLs have a high capacity for CO₂ absorption (since the whole solution is the reactant itself) and can be regenerated at fairly low temperatures, 75–85 °C [3], [236]. Though this does not imply lower reboiler duties, it does dispense the requirement of high-pressure steam for solvent recovery. On the other hand, the high viscosity of these solvents, especially after CO₂ loading, renders their absorption rates lower than those of aqueous MEA even though their reactants have stronger basicity than most primary amines [237]. Zheng et al. [236] have tried to develop a CO₂BOL that achieved maximum viscosity of 20 mPa∙s after absorbing CO₂, but the minimum that they have reached experimentally was 356 mPa∙s. They then employed computational methods to try designing a molecule with lower viscosity when loaded, this time settling for a maximum of 100 mPa∙s in the rich solvent [238]. The CO₂ mass transfer rates in loaded CO₂BOLs appear to be one order of magnitude below that of aqueous MEA and aqueous piperazine [61], [239], [240], which is not bad considering the high viscosities of those nonaqueous systems.

If the idea of a solvent that becomes an ionic liquid when absorbing CO₂ sounds too far-fetched, one might consider that this process is similar to what happens when CO₂ reacts with pure amines in the absence of diluents. In this sense, the CO₂BOLs are similar to the functionalized ethylenediamines of A. Liu et al. [171] or the absorption into solvent-free alkanolamines of Barzagli et al. [228]. This has led the Pacific Northwest National Laboratory group to also work on water-free amine solvents [235], [241], [242]. They have also investigated precipitating nonaqueous amine solvents [243] and single-phase nonaqueous aminosilicones [244], making this one of the most productive research teams working on water-lean solvents at the moment of this writing.

“Fig. 17. Chemical composition of water-lean solvents then and now. On the left, the compositions of the Amisol® solvent [20] and of the Sulfinol-D® solvent [25]. On the right, the compositions of the HPS [116] and of one particularly promising CO₂BOL [236].”

“In energetic terms, their performance approaches that of conventional aqueous amines implemented with novel CO₂ capture plant configurations. The CO₂BOLs have an energetic demand of approximately 2.57 GJ∙ton CO₂⁻¹ captured [245]. The ION solvent, another water-lean proprietary mixture, achieves a performance of 2.54 GJ∙ton CO₂⁻¹ [246], and the NAS delivered 2.3 GJ∙ton CO₂⁻¹ in pilot plant testing [247]. For comparison, aqueous amines in novel configurations have been achieving 2.1–2.4 GJ∙ton CO₂⁻¹ as in the cases of the MHI process with the KS-1 [248], the Shell Cansolv process in Boundary Dam [249] and the latest pilot plant data with the advanced flash stripper [215].

Section 10 has given a quick overview on new concepts for water-lean solvents. A better reference to understand all of these new trends is the excellent review prepared by Heldebrant et al. [7] themselves. As we initially set out to study and compare hybrid solvents, it seems that some of the solvent-free mixtures discussed above verge too far away from our initial focus.

Heldebrant et al. [240] pointed out that perhaps the typical aqueous amine configuration for CO₂ capture is far from ideal for water-lean solvents. If that is true, then it is about time a process is developed that harvests the full potential of these mixtures.”