**1. Introduction**

In recent years, the unprecedented amount of carbon dioxide (CO2) emissions, resulting mainly from fossil fuel utilization/based activities, have impacted global warming [1,2] and climate change [3–5]. Conventional amine-based CO2 mitigation techniques have been considered as effective CO2 capture methods over the past several decades, despite some serious drawbacks, such as solvent loss, corrosion, degradation and, more importantly, high regeneration energy cost [6–12]. Hence, there is need for alternative solvent systems that can effectively scrub CO2 in gaseous effluent streams, at both pre- and post- combustion processes, with minimum requirement of infrastructure retrofitting costs for existing CO2 capture process units in plants. For this purpose, various materials have been developed

over the past few decades. Porous adsorbents [13–17] as well as liquid solvents [18,19] have been considered in both academia and in the industry; however, due to low manufacturing costs and less requirements on new processing equipment, liquid systems have been considered more attractive. Novel solvent development for CO2 capture purposes has been centered on ionic liquids [20–25] and liquid polymers [26–29]. In recent years, deep eutectic solvents (DES) [30–32] obtained from ionic liquids, as well as from natural products (natural deep eutectic solvents (NADES)) [33–35], have also been considered for CO2 management, especially at low to moderate pressures. DES have important advantages over the other potential candidates as they are low-cost materials with appreciable renewability capability [36,37], low toxicity [32,38,39], have less environmental impact [40,41], and good solvent recovery ratio [42].

DES or NADES are formed by mixing the combination of hydrogen bond donors (HBD) and hydrogen bond acceptors (HBA) at various molar mixing ratios to form a low melting point eutectic mixture. Experimental studies show that DES and NADES samples yield to CO2 solubility that can be compared to current state-of-the-art methods [43]. Several studies on gas solubility in DES with ionic liquids are being used as HBA [30,44–47]. In contrast with DES, only a handful of studies use NADES for CO2 solubilities; these studies focus mainly on choline chloride as HBA mixed with glycerol/propanediol/malic acids [34] and monoethanolamine [48]. In our previous work, we studied gas absorption performances via novel NADES, which was obtained by considering choline chloride (ChCl), alanine (Al), and betaine (Be) as HBA, and lactic acid (La), malic acid (Ma), and phenylacetic acid (Paa) as HBD. We showed that these NADES perform very well for CO2 capture, especially at moderate pressures up to 50 bars, and between 298.15 K and 323.15 K isotherms [49–51]. CO2 sorption experiments showed maximum absorption performance between 3.5 and 5 mmol CO2/gr of NADES in these experiments at the highest pressures. Considering the monoethanolamine solution (MEA) CO2 capture performance as 1.8 mmol CO2/g MEA (117 mg/g) at 24 bar and 313 K [52], CO2 absorption via NADES have great potential for chemical processes. Despite these promising results, there is no systematic study on how HBA/HBD affect the gas sorption performance, what binding role HBA/HBD play during CO2 capture, which HBA/HBD have a superior effect once exposed to CO2 environment, and whether ionic liquid based HBA perform better than the amino acidic-based ones. In this work, we attempt to find answers to these questions, present new CO2 capture data for HBA and HBD, and compare their performances against their NADES mixtures. Furthermore, detailed density functional theory (DFT) calculations were carried out to explain the details of the binding energies and infer on the CO2 interaction mechanism of HBA/HBD/NADES separately.

#### **2. Materials and Methods**

The following were purchased from Sigma Aldrich: alanine (Al) with ≥98% purity (Chemical Abstracts Service (CAS) number 56-41-7) with melting point of 258 ◦C; betaine (Be) with ≥98% purity (CAS number 107-43-7) with melting point of 310 ◦C; choline chloride (ChCl) with ≥98.0% purity (CAS number 67-48-1) with melting point of 302 ◦C; DL-malic acid (Ma) with 99.0% purity (CAS number 6915-15-7); lactic acid (La) with 85% purity (CAS number 50-21-5); d-fructose (Fr) with ≥99% purity (CAS number 57-48-7); and citric acid (Ca) with ≥99.5% purity (CAS number 77-92-9). All materials with mentioned purities on the boxes were used without further treatment. Carbon dioxide (CO2) gas with purity of ≥99.99% was obtained from Praxair. In order to form NADES samples, Al, Be, ChCl, were mixed with La, Ma, Ca, and Fr with 1 to 1 (1:1) molar mixing ratios by following vigorous stirring of the mixture until a clear-homogenous solution was obtained, in a glove box, in which humidity and ambient conditions were controlled. Table 1 shows the chemical structures of the studied HBA and HBD, which also form the NADES samples, Al:La, Al:Ma, Be:La, ChCl:Ma, ChCl:La, ChCl:Fr. All of the prepared NADES samples were observed to be liquid state at room temperature and at atmospheric pressure. Structures of the HBA and HBD are provided in Table 1.


**Table 1.** Structures of hydrogen bond acceptors (HBA) and hydrogen bond donors (HBD) compounds.

#### *2.1. Experimental*

CO2 sorption performance of NADES samples were reported in a previous study, in which the contactless magnetic suspension gravimetric method was used for high-pressure gas absorption-desorption measurements via Rubotherm apparatus. A magnetic suspension apparatus that is equipped with an automated syringe pump was used to conduct both low- and high-pressure gas experiments. In a typical procedure, a few ml of sample is placed in a measurement chamber and it is evacuated overnight at temperatures around 70 ◦C. Once the material preparation is complete, then solubility measurements take place in the apparatus, starting from the lower pressures toward higher pressures, with stepwise gradual pressure increments. It takes about 60 min to establish equilibrium thermal in between each pressure increment. Once the highest pressure is measured, then few pressure points are re-experimented during the desorption period, in order to observe whether hysteresis occurs during the reversed conditions. The described method was used to obtain HBA + CO2 and HBD + CO2 sorption performances in this work, and further details of the measurement technique, calibration, and overall uncertainty can be obtained elsewhere [53].

#### *2.2. Theoretical*

Initial structures for all of the considered HBA, HBD compounds, and NADES molecular clusters were built with the Avogadro program [54]; ORCA code [55] was used for DFT calculations along this study for B3LYP functional [56,57], coupled with van der Waals semi-empirical contribution from the DFT-D3 method by Grimme [58], and 6-311++G\*\* basis set. The interaction energy (ΔE) for all of the considered structures were calculated by considering counterpoise correction for Basis Set Superposition Error (BSSE) [59]. The quantum Bader's atoms-in-molecules (AIM) theory [60] was used for interpretation of intermolecular interactions, and for topological analysis, which was obtained via Multiwfn program [61]. From this analysis, specifically interaction regions that were characterized by bond critical points (BCP, (3,−1) type, according to Bader's terminology). The corresponding values of electron density, <sup>ρ</sup>, and Laplacian (∇) of electron density, <sup>∇</sup>2<sup>ρ</sup> were obtained [62]. The properties of the inferred critical points may be related with the strength of the interactions [63,64]. Likewise, quantitative bond critical points and its implications on bonding strength were further analyzed via reduced density gradient analysis (RDG) for visual representation of strength and the nature of intermolecular forces through colored isosurfaces [65].

#### **3. Results**

CO2 solubility experiments for NADES molecular structures were reproduced from previous work. In this study, experimental data on CO2 capture performance via HBA and HBD were received successfully and illustrated graphically. For comparison purposes, 298.15 K isotherm was considered for qualitative and quantitative analysis. A total of 24 pressure points were collected for each HBA or HBD during CO2 sorption experiments, in which 12 were for adsorption, and 12 for desorption measurements. Figure 1 shows the CO2 capture performance for HBA and HBD prior to mixing to form the deep eutectic solvent at 298.15 K isotherm. At a low-pressure side (*p* < 5bars), there is no distinct effect observed on the sorption performance and the data are clustered around ~0.3 mmol CO2/g sorbent. As the pressure is increased to moderate to high pressures, there is a clear separation between ChCl + CO2, Be + CO2 (~4.95 mmol CO2/g) cases in comparison to Al + CO2, La + CO2, Ma + CO2, Ca + CO2 (~3.77 mmol CO2/g) cases. However, Fr + CO2 falls in between of the trend of these two groups at high pressures yielding 4.45 mmol CO2/g performance.

**Figure 1.** CO2 capture performance for hydrogen bond acceptors and hydrogen bond donors prior to mixing to form deep eutectic solvent at 298.15 K isotherm.

Furthermore, absorption capacity of DES is provided in electronic supporting information for CO2 loading amounts in mol/mol units. Figure S1 reveals significant trends on the effect of hydroxyl groups. As the amount of hydroxyl groups in the HBD increases, the absorption of capacity of the CO2 increase. This is related to the intramolecular and intermolecular hydrogen-bonding forces between the HBA and the CO2 molecule. For La, Ma, Fr, Ca there are 4, 3, 2, 1 hydroxyl groups in the HBD structure, and the absorption capacity in Figure S1 overlaps with this argument. On the other hand, for the case of HBA, as the volume of the of the group that is attached on the N atom in the HBA structure increases, it affects the distance between center of mass and the CO2 molecule and, thus, develops a hydrogen bond easier. This leads to the hydrogen-bonding force of DES and improved solvent properties. This phenomenon is also observed in this work when the molecular structures of ChCl, Be, and Al are considered, and their CO2 sorption capacities are examined in Figure S1. The development of hydrogen bonding is discussed in further detail, from the molecular point of view, in the following section of this manuscript.

CO2 capture performance comparison of the HBA and HBD with respect to NADES has been studied and presented in Figure 2. It was expected for NADES (or DES) CO2 capture performance to fall in between its former HBA and HBD compounds. However, interestingly, only Al:La + CO2 showed this behavior with distinct performance separation between the HBA and HBD CO2 capture data, and fall right in the middle of these two curves (Figure 2a). For the case of ChCl:La + CO2, the NADES sorption coincide over the ChCl + CO2 curve, which is observed to be underperforming

in comparison to La + CO2 case (Figure 2c). For Al:MA + CO2 case, it has a similar trend with ChCl:La + CO2 and the NADES + CO2 trend overlaps again with the HBA trend (Figure 2d), with the exception of a slight departure of NADES towards higher performance at pressures higher than 35 bars. However, for the case of Al:La + CO2, there is a distinct segregation of HBA+CO2, HBD+CO2 and NADES+CO2 trends (Figure 2c). In this specific case, pressures above 10 bars, CO2 capture performance trend was observed as La + CO2 > NADES + CO2 > Al + CO2, or in other words HBD + CO2 > NADES + CO2 > HBA + CO2. Maximum solubility performances were obtained via the highest achieved experimental pressure at 50 bars, reported in Table 2. For all cases, except for Be:La + CO2 (Figure 2b), HBA experimental sorption performances was superior than that of HBD, which was also mentioned by D.O. Abranchesn et al. (that Be possess weak interaction with itself, but act as excellent HBA) [66]. ChCl+CO2 showed the best performance with 4.96 mmol CO2/g, whereas Al + CO2 showed the worst capture performance with 3.86 mmol CO2/g. In the case of Be:La + CO2, the NADES + CO2 profile was observed to be lower than its constituents (Be and La), which can be explained due to the negative excess volume that was created via mixing the HBA and HBD [67]. Likewise, in the case of Al:Ma + CO2, the NADES profile was observed to be higher than its HBA and HBD, which is a sign of positive excess volume when Al and Ma was mixed to form NADES.

**Figure 2.** CO2 capture performance of studied deep eutectic solvents (DES) systems and comparison to their constituents. (**a**) ChCl:La + CO2, (**b**) Be:La + CO2, (**c**) Al:La + CO2, (**d**) Al:Ma + CO2.


**Table 2.** Gas sorption data and binding energies ranking.

As clearly reported in Table 1, the CO2 capture performance of NADES is lower in comparison to sole HBA and HBD cases, for most of the studied cases. When the melting point of the studied HBA and HBD are considered, they are solid in room temperatures, and even at elevated process temperatures, at which typical pre- and post-combustion CO2 capture operations take place. However, the NADES compounds exist in liquid phase at mentioned temperatures, thus, making their processability much easier than their constituents. One of the main objectives of considering solvents for CO2 capture is to be able to utilize existing infrastructures that were built with the consideration of current-state-of-the-art capture agents (e.g., monoethanolamine-based solvents), with minimum retrofitting requirements on the equipment infrastructure.

DFT simulations were carried out for the same structures for which the experimental findings were shared above. The main purpose of the utilization DFT simulations was to obtain insights on electronic configuration of the studied compounds and infer on the behavior of the interaction sites between the HBA/HBD/NADES structures and the CO2 molecule. Macroscopic properties can be estimated through calculation intensive molecular dynamic simulations. DFT simulations were carried out for each NADES, HBA, and HBD with CO2 gas presented around them at various spatial positions. Three different spatial positions were considered for potentially high CO2 interactions sites for NADES and two positions were considered for HBA/HBD compounds, respectively. Obviously, a restricted amount of CO2 molecules around the NADES structure would not represent the overall bulk phase conditions and the entire solubility phenomena. It should be noted that DFT is considered a tool to assess the interaction nature as well as characteristics of the studied compounds in case they are exposed to CO2.

Once the DFT simulations were obtained and the interactions energies were corrected for the BSSE [59], the cases that gave the highest interaction energies were considered for further DFT analysis. In this work B3LYP with the 6-311++G\*\* theory level was selected, based on its accuracy and reasonable computational time [49,50]. The summary of interaction energies for the selected cases are provided in Table 3 in eV units. Binding energies for each HBA, HBD, and NADES with CO2 are provided in Table 4, along with comparisons of binding energies with respect to each other within each NADES group.


**Table 3.** Summary of energies of optimized structures.

**Table 4.** Optimized energies of HBA, HBD, and natural deep eutectic solvents (NADES) interacting with CO2 at various spatial positions.



**Table 4.** *Cont.*

Once the binding energy results are deeply analyzed, HBA > HBD trend is observed for CO2 interactions, which is similar to the experimental behavior CO2 sorption performance in HBA/HBD. The only exception was observed in Be + CO2 and La + CO2 comparisons, for which the binding energy is higher for HBD + CO2 than HBA + CO2 case. There is no particular correlation between the maximum sorption amount and the binding energy for the NADES + CO2 cases. On the other hand, the maximum sorption amounts and the binding energies for both HBD and HBA follows the same rank. In other words, it was observed that as the binding energy for CO2 increases, the CO2 sorption performance also increases for both HBD and HBA cases.

Furthermore, the interfacial properties of NADES/HBA/HBD and CO2 can also be studied using molecular dynamics simulations. Garcia et al. showed how CO2 establishes remarkable interactions with the bulk NADES structure through analyzing the evolution of the number of hydrogen bondings between the HBA and HBD [46]. Furthermore, strong affinity between DES and CO2 molecules were quantified by the corresponding interaction energies. Results shared by Garcia et al. showed the large interaction energies for interaction of CO2 molecules, with both the HBA and HBD, with slightly higher values between HBA and CO2 in comparison to HBD−CO2. Relatively small differences between the HBA and CO2 and HBD−CO2 suggests a mutual effect on CO2 solubility [46].

On the other hand, a study by Wang et al. showed the effect of the type and molar mixing ratio of HBA and HBD for ionic liquid-based DES through analyzing the radial distribution functions. It was reported that the cation of the HBA plays a superior role on the hydrogen bond development between the DES and CO2, yet again supporting the observed trend in this work [68].

Another molecular dynamics-based study that supports the CO2 affinity is driven by HBA of the DES was reported elsewhere [69]. Besides analyzing the hydrogen bond evolution between the CO2 molecule and HBA/HBD, they proved the superior effect of HBA on CO2 affinity via displaying the higher degree of clustering of CO2 molecules around the HBA through the spatial distribution function isosurfaces. All of these studies are in line with the arguments that are claimed in this work through DFT simulations on the superior binding effect of HBA on CO2.

The reduced density gradients (RDG) isosurfaces were used to visualize the interaction of CO2 with HBA and HBD, Figure 3. Strong van der Waals type interactions are recorded for CO2 binding at relevant interaction sites at HBA and HBD, which is a sign of reversible process for CO2 sorption since H-bonding has not been observed. Especially at ChCl, both -H and -Cl− sites showed more intensified RDG volumes between the -O site of CO2 (Figure 3c), which is also quantified with the highest binding energy with −0.137 eV (Table 2). Likewise, weakest interaction was observed with less intense isosurface between CO2 and Ma (Figure 3e), and it is evident with the binding energy reported in Table 2 for Ma + CO2.

**Figure 3.** Reduced Density Gradient (RDG) isosurfaces for HBA + CO2 and HBD + CO2 cases. (**a**) Al+ CO2, (**b**) Be + CO2, (**c**) ChCl + CO2, (**d**) Fr + CO2, (**e**) La + CO2, (**f**) Ma + CO2.

Quantum theory atom-in-a-molecule (QTAIM) analysis and RDG isosurfaces for NADES was presented for NADES + CO2 cases in Figures 4 and 5, respectively. For most of the cases, CO2 established critical binding with HBA of the NADES structure, except for the ChCl:Fr + CO2 case. In Figure 4, it is evident that -O site of CO2 is the most recurring interaction between the HBA -H site. The experimental values for the CO2 solubility in NADES can be ranked as ChCl:La > Al:La ~ ChCl:Ma > Be:La > Al:La. Whereas the binding energies that are calculated via DFT simulations for CO2 are ranked as ChCl:Fr > ChCl:La ~ ChCl:Ma > Be:La > Al:La. These two comparisons do not overlap with each other. A more logical comparison on overall solubility performance, with respect to binding energies, would be between the NADES constituent prior to the mixing, since they form excess volume pockets once they are mixed, which leads to excess CO2 capture performance.

**Figure 4.** Density functional theory (DFT) figures for each structure final optimized geometry.

**Figure 5.** Reduced density gradient (RDG) isosurfaces for HBA + CO2 and HBD + CO2 cases.
