Enzyme-assisted CO 2 absorption in aqueous amino acid ionic liquid amine blends

The influence of carbonic anhydrase (CA) on the CO 2 absorption rate and CO 2 load in aqueous blends of the amino acid ionic liquid pentaethylenehexamine prolinate (PEHAp) and methyl diethanolamine (MDEA) was investigated and compared to aqueous monoethanolamine (MEA) solutions. The aim was to identify blends with good enzyme compatibility, several fold higher absorption rates than MDEA and superior desorption potential compared to MEA. The blend of 5% PEHAp and 20% MDEA gave a solvent with approximately 5-fold higher initial absorption rate than MDEA and a 2-fold higher regeneration compared to MEA. Experiments in a small pilot absorption rig resulted in a mass transfer coefficient (KGa) of 0.48, 4.6 and 15 mol (m 3 s mol fraction) -1 for 25% MDEA, 5% PEHAp 20% MDEA and 25% MEA, respectively. CA could maintain approximately 70% of its initial activity after 2 h incubation in PEHAp MDEA blends. Integration of CA with amine-based absorption resulted in a 31.7% increase in mass of absorbed CO 2 compared to the respective non-enzymatic reaction at the optimal solvent: CA ratio and CA load. Combining novel blends and CA can offer a good compromise between capital and operating costs for conventional amine scrubbers, which could outperform MEA-based systems.


INTRODUCTION
Chemical absorption by aqueous amine systems is one of the most mature post combustion techniques applicable to large CO2 point emissions, such as the process industry. 1 Apart from scrubbing flue gases, CO2 absorption is also important for the upgrading of biogas or natural gas. 2 In a typical CO2 capture process, the CO2-rich gas enters the absorption column from the bottom and contacts the lean CO2 absorbing solvent, which enters from the top, in a countercurrent flow. Subsequently, the CO2-rich solvent is pumped through a stripping column, where the solvent is thermally regenerated, and then pumped back to the absorber for another cycle of absorption. During the regeneration process, a pure CO2 stream can be taken out at the top of the column and can be compressed for transportation and storage. 3 Conventional carbon capture techniques are considered expensive and energy intensive and, like in all processes, it is desired to minimize capital and operating costs. Naturally, the solvent has a major impact on the process.
Generally, fast absorption kinetics translates to small equipment size and a relatively low capital cost. On the other hand, faster reaction kinetics are connected to higher heat of reaction and so the regeneration temperature in the desorber must be high, leading to steam requirements that can make up to 90% of the total operational costs. 4 Other important factors for solvent development are the load capacity, CO2 specificity, corrosion properties and solvent degradation rate due to the high temperatures and the presence of SO2, NO2 and O2 in several flue gases. 5,6 Ultimately, it remains a challenging task to optimize a CO2 capture solvent for a specific application, where the end result is a compromise between the solvent's absorption and desorption properties.
Since the development of the conventional amine scrubber in the 1930s, numerous amine blends have been screened for the optimum compromise, where aqueous solutions of the primary amine 4 monoethanolamine (MEA) remains the industry standard. 1 However, the high absorption rates for MEA are linked to a high reactivity that, in turn, results in regeneration temperatures of 120-140 o C and steam consumption equivalent to between 3.24-4.20 GJ/ton. 7 Primary and secondary amines form carbamates (equations 1-2) and have relative low loading capacity at around 0.5 mol CO2 mol -1 amine. In water solutions, the carbamates can also decompose into HCO3 -(equation 3) and, in such cases, the amine may bind another CO2 molecule. 8 Tertiary amines, such as methyldiethanolamine (MDEA), produce bicarbonate (equation 4) and have a loading capacity of 1 mol CO2 mol -1 amine. Although they require lower regeneration temperatures, they suffer from significantly lower absorption rates. 9 Polyamines such as diethylenediamine (DETA), triethylenetetramine (TETA) and pentaethylenehexamine (PEHA) are known to display high loading capacities and display absorption rates faster or comparable with MEA. However, as they require similar regeneration temperatures, they still make the process highly energy intensive, which in turn promotes corrosion, solvent degradation and heat loss. Good corrosion properties have been identified by certain amines, such as the designer amine with cyclic structure 4-amino-1-propyl-piperidine. 10 R1R2NH + CO2 ↔ R1R2NH + COO -(zwitterion; reaction intermediate) (1) R1R2NH + COO -+ B ↔ R1R2NCOO -(carbamate) + BH + Where R2 is a hydrogen for primary amines and B base R1R2R3N + CO2 + H2O ↔ R1R2R3NH + + HCO3 - 5 Catalysts are an interesting avenue for CO2 capture as they can improve the absorption rates without compromising the desorption properties. 11 Carbonic anhydrase (CA) is one of the fastest enzymes known catalyzing the hydration of CO2 (equations [5][6][7][8], which is the rate-determining step of the reactive absorption of CO2 for MDEA.
The addition of CA in the solvent could increase the absorption rates by exploiting the ability of CA to convert CO2 to HCO3very quickly, and thus keeping the concentration gradient between the gas and liquid phases; that is the driving force for CO2 dissolution. 12 Bovine CA has been reported to enhance both the CO2 absorption rate and loading capacity in low concentrations of alkanolamines (5-10%) including MEA, diethanolamine (DEA), aminomethyl propanol (AMP) and MDEA. 13 However, enzyme catalysts have not been assumed to tolerate exposure to the high temperatures and the alkaline environment of amine-based capture and desorption processes.
Yet, an engineered CA from Desulfovibrio vulgaris (DvCA8.0) was shown to have exceptional properties for CO2 absorption at high temperatures in MDEA. 14,15 Comparatively, little work has been done on the integration of CA to amine systems other than MDEA, most likely due to the poor performance of CA in reactive highly alkaline amine In the current study, we propose to use a mix between an AAIL (PEHA prolinate, PEHAp) and a tertiary amine (MDEA) in order to promote enzyme stability and reach higher absorption rates than MDEA, achieving at the same time advanced desorption properties compared to MEA. In this ternary blend, the tertiary amine would partly function as a proton acceptor from the zwitterion to allow the fast primary and secondary amines to react with CO2. In addition, it promotes better desorption properties for the solvent, since tertiary amines only form bicarbonate with CO2, which upon heating, decomposes back to CO2 easier than carbamates. In order to promote high absorption rates and loading capacity, PEHA was selected as the cation component and proline as the anion of the AAIL. Proline was assumed to take a protective role for the enzyme with its hydrophobic character. Furthermore, it was reasoned that the desorption properties of proline would be superior to other amino acids, such as lysine. Proline carries only a secondary amine, which has shown moderate absorption rates compared to lysine that carries a 7 primary amine. 19 The amines MEA, MDEA and PEHA and the AAIL PEHAp are depicted in Figure 1. Structural characterization was done by NMR spectroscopy performed in D2O with a Bruker Avance 600 MHz instrument (Billerica, MA, USA) (SI Figure S1 and S2). The obtained data were further processed with the TopSpin 3.2 software. As one equivalent acid was used for the preparation, it is expected that the primary amine is protonated ( Figure 1).   The absorption rate, desorption rate and load capacities were calculated according to Luo et al. 20 The absorption and desorption rates (mol s -1 ) at any given time, QCO2, were calculated according to equation (10), where 2 and 2 the molar flow rates of N2 and CO2 into the solution respectively and 2 the molar fraction of CO2 out of the solution. 13 The absorption and desorption rates (mol L -1 s -1 ) were subsequently calculated according to equation (11), where VL the volume of the solution.
QCO2 was logged with time, t, and the accumulated moles of CO2 absorbed by the liquid, NCO2, was calculated according to equation (12), The CO2 load, N, at a particular time point could then be calculated according to equation (13) in mol L -1 or equation (14) in mol mol -1 amine Where Namine the amount of amine (mol) in solution.
The percent increase in absorbed CO2 for the enzyme catalyzed reactions, ΔCO2%, were calculated according to equation (15), Where 2, the mass of absorbed CO2 during enzyme-catalyzed reaction and 2, the mass of absorbed CO2 during the respective non-enzymatic reaction.

Scaled-up CO2 absorption on a packed bed column. Scaled-up demonstration experiments
were performed in a 1 m packed bed absorption rig CHE 626 (HFT Global Ltd, Derbyshire, UK) at 20 o C for selected solvents, at a fixed gas flow rate of 65 L min -1 , containing 8 % CO2 and 92% N2. The liquid flow rate was 0.58 L min -1 . The inlet and outlet CO2 concentration was recorded with a CO2 analyzer. The overall volumetric mass transfer coefficient, KGa, was calculated for dilute conditions (CO2 < 10%) according to equation (16), assuming ideal gas behavior in the vapor phase, which is derived from a general mass balance over a packed absorption column, Where P=total pressure in the column (atm), Pi=partial pressure of CO2 in the inlet stream (atm), Po=partial pressure of CO2 in the outlet stream (atm), A=cross-sectional area of the column (m 2 ), N= gram moles CO2 absorbed s -1 , Z=height of packing (m).

RESULTS AND DISCUSSION
Viscosity, density and pH of the AAIL amine blends. The viscosities and densities of PEHAp MDEA blends as a function of time were analyzed ( Figure 3A and 3B). In general, the viscosities and densities were water like due to the high water activity, which promotes hydration of the involved species and decreases their interactions. The data are consistent with similar reported blends. 21 The water-like behavior of the solvents can facilitate their direct use in conventional amine scrubbers. Increasing the temperature naturally weakens the intermolecular bonds, leading to increased fluidity and decreased density. The higher viscosity and density of the 25% PEHAp solution could be related to intermolecular hydrogen bonding with higher strength between water and PEHA compared to intra-and intermolecular hydrogen bonding in PEHA. 22    It can generally be observed that the initial absorption rates for 25% PEHA and 25% MEA are the highest, where PEHA displayed the highest overall absorption rate. In contrast to the other solutions, the 25% PEHA and MEA solutions also maintained a high absorption rate over a larger CO2 load (mol L -1 ) interval, which is consistent with the availability of a high number of strong primary and secondary amines per unit volume ( Figure 4A). The results confirm other reports where it was shown that 30% PEHA displayed faster absorption rates and higher loads compared to a 30% MEA solution. 22 Other aqueous polyamine solutions such as tetraethylenepentamine (TEPA) has also shown higher absorption rates compared to MEA 24 Table 1) so pH is not likely to influence the initial absorption rate much for these solvents, but rather do the available primary and secondary amine groups.
Blending 5% PEHAp and 20% MDEA increased the initial absorption rates almost 5-fold, The more modest improvements in the absorption rates with further addition of PEHAp may be related to the increasing viscosities with higher amounts of PEHAp and the fact that a larger fraction of the primary and secondary amines must act as proton acceptors themselves during carbamate formation.
On a mol L -1 basis, the maximum load was highest for the 25% PEHA solution at 2.3 mol L -1 closely followed by 25% MEA and 25% PEHAp at 2.13 and 2.1 mol L -1 respectively. The capacities for the blends ranged between 1.1 and 1.4 mol L -1 with a gradual increase as more PEHAp was added ( Figure 4A). The load for 25% MEA was about 0.54 mol CO2 mol -1 amine and consistent with other studies on aqueous MEA solutions. 25 In water solutions, the capacity is generally slightly higher than 0.5 mol CO2 mol -1 (theoretical value) as water and OHcan act as base as well. The 25% PEHA solution displayed a load of 2.3 mol CO2 mol -1 amine under the prevailing conditions reflecting that several of its amine groups react with CO2 ( Figure 4B). This value is consistent with another study under similar conditions where a load of about 2.46 mol CO2 mol -1 amine for a 30% aqueous PEHA solution was reported. 22 With two primary and four secondary amine sites, PEHA could theoretically be expected to bind 3 mol CO2 mol -1 amine assuming that no other reactions occur. The 25% PEHAp solution displayed a higher capacity (2.88 mol CO2 mol -1 amine) than 25% PEHA (2.46 mol CO2 mol -1 amine) even though one of the nitrogen atom got protonated. This is due to the fact that the prolinate anion is also contributing in the chemisorption of CO2. 26  Desorption rates versus CO2 load. Depending on whether the CO2 load is given in mol L -1 ( Figure 5A) or mol mol -1 amine ( Figure 5B), the desorption profile looks different for some solvents, particularly for MEA, which has a small molecular weight. However, the desorption rates generally dropped fast with decreasing CO2 loads and displayed a more gradual decrease as the lean loading, i.e. the remaining load after desorption, was approached. The initial high desorption rate is most likely related to the amount of bicarbonate which is highest at the maximum load and decomposes more easily to CO2 compared to the decomposition of the carbamates and their respective amines. 22 On a mol L -1 basis, the lean loading for the 25% PEHA, 25% MEA and 25% PEHAp solutions were between at 1.2-1.4 mol L -1 indicating the stability of the carbamates formed with the primary and secondary amines involved ( Figure 5A).
Despite that the 25% PEHA solution had the highest absorption rate, it displayed a slightly lower lean load compared to 25% MEA and a similar lean load as 25% PEHAp. This suggests destabilization of the inner secondary amine groups of PEHA in combination with the high initial pH, resulting also in a larger amount of bicarbonate formed compared to 25% PEHAp. Although 25% MEA should have similar amounts of bicarbonate after absorption, it cannot benefit from such destabilization of internal carbamates. On a mol mol -1 amine basis, 25% PEHAp had the highest lean load (1.75 mol mol -1 amine) followed by 25% PEHA (1.18 mol mol -1 amine) ( Figure 5B). The unexpectedly lower initial desorption rates for the 25% PEHAp solution compared to 25% PEHA may be related to non-specific interaction between the ionic character of the AAIL and CO2 and possibly a stabilization of the formed carbamates. Stabilization of carbamates from reversible ionic liquids have been reported. 28 Although 25% MEA forms stable carbamates, it had a lower lean load when expressed per mol mol -1 amine compared to 25% PEHA and 25% PEHAp, due to its small molecular weight and only one primary amine group ( Figure 5B). The blend 5% PEHAp 20% MDEA had lowest lean load, equal to 0.21 mol L -1 or 0.13 mol mol -1 amine ( Figure 5A and 5B, respectively) that increased by increasing the concentration of PEHAp in the blend. Interestingly, its initial absorption rate was higher compared to 5% PEHA 20% MDEA indicating that the potential stabilization of formed carbamates due to interaction between the ionic character of the AAIL and CO2 is not strong when MDEA is present in the blend.
As expected, the 5% PEHAp 20% MDEA blend displayed the highest desorption rate being the most promising tested blend (Figure 6). At the first 40 min, the desorbed amount of CO2 was clearly highest for 5% PEHAp 20% MDEA, thus for a given processing rate of CO2, the operational costs could potentially be the lowest using this solvent. However, if full regeneration conditions are used, i.e. higher temperature, the concentration would have to be increased to match the mol L -1 capacity for 25% MEA or 25% PEHA. The 5% PEHA (i.e without prolinate) 20% MDEA blend did not have as good desorption properties as 5% PEHAp 20% MDEA, which could be attributed to the fact that none of its primary amines are neutralized. As MDEA was replaced with more PEHAp in a blend, the desorption rates decreased and lean loadings increased. Hence, the proline part seemed to have a key role in maintaining a good absorption capacity and desorption potential. As previously mentioned, the prolinate anion contributes in the chemical absorption of CO2 resulting in good capacity. In the case of desorption, the carbamate formed with prolinate anion might be able to decompose fully and easier at 80 o C, requiring lower desorption energy and resulting in good desorption, too. The % regeneration increased as more PEHAp was replaced by MDEA, which is consistent with the lower desorption energy requirements for MDEA related to more bicarbonate formation during absorption. For the PEHAp MDEA blends, the % regeneration was roughly proportional to the fraction of MDEA present. A summary of the absorption and desorption performance of all tested amines and AAIL blends is presented in Table 2. The blend 5% PEHAp 20% MDEA 23 was selected as the most promising solvent, as it combines competitive absorption rates, 5-fold higher than 25% MDEA, (Figure 4) with the best % regeneration, 2-fold higher than MEA ( Figure 6). The superior desorption suggest that the blend should have considerable lower regeneration energy per ton CO2 absorbed compared to MEA.  25 Scaled-up CO2 absorption on a packed bed column. In order to compare the most promising blend with MDEA and MEA under more realistic conditions, the overall volumetric mass transfer coefficients, KGa, were determined using a packed bed absorption column (Table 3).
Here, the 5% PEHAp 20% MDEA blend displayed a 9.6 times higher KGa than 25% MDEA which could save capital costs compared to MDEA-based plants. Thermal and solvent stability of CA. As a first step for the integration of CA with chemical absorption, it was desired to confirm the compatibility of the enzyme with the developed solvent blend and with relative temperatures. Thus, the residual activity of CA was determined after challenge of lysate to different temperatures (25-100 o C) and to the different amines and AAIL blends (Figure 7). subjected to desorption at 80 o C. It was observed that the regeneration of solvent was not affected by the presence of enzyme in the solvent blends ( Figure 8B). This was expected as CA catalyzes the hydration of CO2, forming bicarbonate that destabilizes easily and decomposes to CO2 at low temperatures such as 80 o C. Moreover, it was confirmed that possible impurities present in the whole cell lysate did not affect the desorption efficiency of the solvent. Effect of solvent: CA ratio and CA load on the absorption. The effect of the solvent: CA ratio was investigated in the most promising AAIL amine blend, 5% PEHAp 20% MDEA. Initially, lysate of fixed stock concentration was added at different ratios in the blend ( Figure 9A). It was observed that a 10% increase in the mass of absorbed CO2 compared to the non-enzymatic reaction by increasing the lysate volume from 0.5 to 1 mL. In contrast, increasing the lysate volume from 1.5 to 3 mL caused a small reduction in the absorbed CO2, from 31.7% to 28.8%.
The lack of improvement in CO2 absorption could be attributed to the introduction of higher amount of water, particles and other cells components in the reaction mixture.
To assess the combined effect of solvent: CA ratio and CA load in the CO2 absorption, lysate of different concentrations was introduced in the blend at different ratios (100: 1.5 v/v and 100: 3.0 v/v) ( Figure 9B). It was observed that up to 0.02 g L -1 protein load, the increase in the amount of absorbed CO2 compared to the non-enzymatic reaction was not affected by the solvent: CA ratio.
For given high protein loads above 0.5 g L -1 , the introduction of higher lysate volume affected negatively the absorption performance. Thus, it was concluded that the negative effect is attributed to the increase of impurities in the reaction mixture. The optimal conditions for enzyme-assisted absorption were determined as 100: 1.5 v/v solvent CA: ratio and 1 g L -1 protein load, offering a 31.7% increase in the amount of absorbed CO2 compared to the non-enzymatic respective reaction.
30 Figure 9. Effect of solvent: CA ratio (A) and protein load (B) on the mass of absorbed CO2.
Absorption was carried out in 5% PEHAp 20% MDEA. The bars represent the standard deviation between duplicate runs.
Investigating the effect of protein load on the absorption rate at fixed solvent: CA ratio (100: 1.5 v/v), a 1.4-fold increase in the initial reaction rate was observed, compared to the non-enzymatic reaction ( Figure 10A). The enzymatic reaction was fastest during the first 25 min at optimal conditions, while all enzyme-catalyzed reactions converged to a CO2 load of approximately 1.1 mol L -1 ( Figure 10B). Except for the increase in the absorbed CO2 the benefits of employing CA include enhanced absorptions rates and reduced operation times. In our case, at optimal protein load the enzyme-assisted reaction was concluded at only 60 min compared the non-catalyzed reaction that lasted 80 min ( Figure 10A).
The apparent kinetic constant (kapp) was 2-fold higher for the CA-catalyzed reaction at optimal conditions, compared to the non-enzymatic reaction. Vinoba et al. 13 reported almost a 3-fold increase in the kapp constant adding bovine CA in 5% w/w MDEA, among other tested solvents.

31
The lower rate in our study could be attributed potentially to limited enzyme stability at significantly higher amine concentrations (25% w/w). Nevertheless, there are limited reports for CA-assisted absorption in amines other than MDEA, while little focus has been put on optimization of the absorption parameters with focus on the biocatalyst [29][30] .
In our study, the initial absorption rate of enzyme-assisted absorption performed in the ternary blend of 5% PEHAp, 20% MDEA and enzyme, was approximately 6 times higher than the one of the respective chemical absorption performed with 25% MDEA ( Figure 4A) This approach highlights the potential for application of greener and more sustainable bioprocesses for CO2 sequestration. However, it remains to be studied whether the cost to incorporate a process involving an enzyme and an amino acid is economically viable compared to the conventional one, and thus be projected on a large scale.

Author Contributions
The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript. ‡These authors contributed equally.

Funding Sources
The