The Influence of Hydrogen Bond Donors on the CO2 Absorption Mechanism by the Bio-Phenol-Based Deep Eutectic Solvents

Recently, deep eutectic solvents (DESs), a new type of solvent, have been studied widely for CO2 capture. In this work, the anion-functionalized deep eutectic solvents composed of phenol-based ionic liquids (ILs) and hydrogen bond donors (HBDs) ethylene glycol (EG) or 4-methylimidazole (4CH3-Im) were synthesized for CO2 capture. The phenol-based ILs used in this study were prepared from bio-derived phenols carvacrol (Car) and thymol (Thy). The CO2 absorption capacities of the DESs were determined. The absorption mechanisms by the DESs were also studied using nuclear magnetic resonance (NMR), Fourier transform infrared (FTIR), and mass spectroscopy. Interestingly, the results indicated that CO2 reacted with both the phenolic anions and EG, generating the phenol-based carbonates and the EG-based carbonates, when CO2 interacted with the DESs formed by the ILs and EG. However, CO2 only reacted with the phenolic anions when the DESs formed by the ILs and 4CH3-Im. The results indicated that the HBDs impacted greatly on the CO2 absorption mechanism, suggesting the mechanism can be tuned by changing the HBDs, and the different reaction pathways may be due to the steric hinderance differences of the functional groups of the HBDs.


Introduction
The climate crisis has been one of the most serious issues threatening the environment, animals, and humankind, which is regarded as the feedback of global warming. Greenhouse gases generated from human activity are the major cause of climate change, which mainly include fluorinated gases, methane (CH 4), carbon dioxide (CO 2 ), and nitrous oxide (N 2 O) [1]. Among these gases, CO 2 is the main contributor, mainly released from the combustion process of fossil fuels (e.g., crude oil and coal) in the industrial sectors, such as power plants, cement kilns, and ammonia production [2]. With the rapid development of the global economy in recent decades, resulting in the huge combustion of fossil fuels, the atmospheric CO 2 concentration has reached more than 410 ppm in 2020 [3]. In order to reduce and control the CO 2 emission into the atmosphere, many technologies have been developed and used in industry, of which chemical absorption processes based on amine aqueous solutions are the most used methods because of the relatively high capture efficiency [4,5]. However, the amine-based absorption technology suffers inherent shortcomings, including the high energy cost of the regeneration processes, serious equipment corrosions, and the high volatility of the solvents [6].
In recent decades, ionic liquids (ILs) have been widely investigated to capture CO 2 because of their unique properties, such as high thermal stability, tunable structures, and low vapor pressure [7]. Of them, functional ILs, including amine-based ILs, azolide-based ILs, and phenol-based ILs, which could chemically capture CO 2 due to the active sites on the cations or anions, exhibited high capacities for CO 2 , making them promising candidates as chemical solvents to capture carbon [8]. However, the complex synthesis procedures and the high cost are the main obstacles limiting their practical applications. the cations or anions, exhibited high capacities for CO2, making them promising candi dates as chemical solvents to capture carbon [8]. However, the complex synthesis proce dures and the high cost are the main obstacles limiting their practical applications.
Deep eutectic solvents (DESs), exhibiting similar properties of ILs, have gained grea attentions in recent years and they are also studied as solvents for CO2 absorption [9,10] At present, DESs are mainly formed by mixing hydrogen bond donors (HBDs) with hy drogen bond acceptors (HBAs) through the intermolecular hydrogen bonding interac tions between HBDs and HBAs, which results in the formation of liquid solvents with lower melting points than the individual components in the system [11]. A lot of DES have been prepared as CO2 absorbents so far. Among them, amine-based DESs [12][13][14][15][16] anion-functionalized DESs [17,18], and superbase-based DESs [19][20][21][22] showed attractiv CO2 capacities through the reaction between CO2 and active sites in the components a ambient pressure, suggesting that adjusting the structures of the components in the DES can be an effective strategy to meet the particular demands. Moreover, the interaction between CO2 and the functionalized DESs were also reported in the literatures. In ou previous work, we found that CO2 reacted with ethylene glycol when CO2 was captured by the DESs composed of azolide-based ILs and EG, but CO2 did not react with the azolid anions in the DESs [17]. However, CO2 reacted with both the azolide anions and EG in th DESs formed by the superbase-derived ILs and EG [20]. Another different reaction path way between CO2 and DESs was reported when CO2 interacted with the DESs composed of polyamine-based ILs and EG, in which CO2 reacted with the amine groups on the cati ons and did not react with the [Im] − anion of the ILs in the solvent [18]. These result suggest that it is important to know the roles that the components of DESs take part in when DESs interact with CO2, which will be useful to the design of new DESs for efficien CO2 absorption.
In this work, we report that the CO2 absorption mechanism by DESs based on phenol derived ILs can be tuned by changing the HBDs. The DESs used in this work were pre pared by mixing phenol-based ILs with HBDs ethylene glycol (EG) or 4-methylimidazol (4CH3-Im). The phenol-based ionic liquids used in this study can be easily synthesized b the acid-base reactions between tetraethylammonium hydroxide ([Et4N][OH]) aqueou solution and the bioderived phenols, including carvacrol (Car) and thymol (Thy)

Results and Discussion
The melting point (T m ) and decomposition temperature (T d ) of the DESs used in this work were investigated, and the results were shown in Table S1. The melting points of EG-based DESs were not observed in the temperature range studied (−80-25 • C). The viscosities of the four DESs were determined at 25 • C (Table S1). The CO 2 absorption capacities of the four phenol-based DESs were measured at 1.0 atm and 25 • C (Figure 1) [Thy]:4CH 3 -Im (1:2) (1298 mPa·s). The low viscosity of EG-based DESs was beneficial to CO 2 diffusion in the solvent, which may promote the reactions between CO 2 and DESs, thus EG-based DESs exhibited a faster CO 2 absorption rate. The comparison of CO 2 capacities by DESs used in this work with other DESs previously reported was shown in Table S2.
cosities of the four DESs were determined at 25 °C (Table S1). The CO2 absorption cap ities of the four phenol-based DESs were measured at 1.0 atm and 25 °C (Figure 1 [Thy]:4CH3-Im (1:2), and [Et [Car]:4CH3-Im (1:2) could capture 0.90, 0.87, 0.90, and 0.88 mol CO2 / mol solvent, resp tively. These results indicated that these four phenol-based DESs can efficiently capt CO2. Moreover, as shown in Figure 1, CO2 was faster to reach saturation in EG-based D than in 4CH3-Im-based DESs. This phenomenon was mainly due to the influence of viscosity of the solvent.  Figure S1). As shown in Figure S1, the CO2 cap ity decreased with increasing temperature. The CO2 desorption by the DESs used in work was also investigated. The captured CO2 by DESs can be released at 80 °C.  In order to clarify the absorption mechanism, nuclear magnetic resonance (NM spectroscopy and Fourier transform infrared (FTIR) spectroscopy were used to study reaction between CO2 and the DESs. The NMR spectra of [Et4N] [Car]:EG (1:2) before after CO2 uptake were shown in Figure 2. As seen in the 1 H NMR spectra, two new pe at 3.53 (H-2) and 3.80 (H-3) ppm can be observed after CO2 absorption. In the 13 C N spectra, four new peaks at 61.4 (C-2), 66.4 (C-3), 157.7 (C-4), and 160.3 (C-j) ppm can found after absorption. Moreover, H-f and H-d shifted downfield from 6.24 and 6.02 p  Figure S1). As shown in Figure S1, the CO 2 capacity decreased with increasing temperature. The CO 2 desorption by the DESs used in this work was also investigated. The captured CO 2 by DESs can be released at 80 • C. In order to clarify the absorption mechanism, nuclear magnetic resonance (NMR) spectroscopy and Fourier transform infrared (FTIR) spectroscopy were used to study the reaction between CO 2 and the DESs. The NMR spectra of [Et 4 N] [Car]:EG (1:2) before and after CO 2 uptake were shown in Figure 2. As seen in the 1 H NMR spectra, two new peaks at 3.53 (H-2) and 3.80 (H-3) ppm can be observed after CO 2 absorption. In the 13 C NMR spectra, four new peaks at 61.4 (C-2), 66.4 (C-3), 157.7 (C-4), and 160.3 (C-j) ppm can be found after absorption. Moreover, H-f and H-d shifted downfield from 6.24 and 6.02 ppm to 6.80 and 6.48 ppm, respectively. C-a shifted upfield from 167.3 to 156.3 ppm and C-f shifted downfield from 108.8 to 113.4 ppm. It is reasonable to expect that CO 2 could react with the anion [Car] − in the solvent because of the acid-base interactions, which would produce a carbonate species, and then a new carbon signal of the carbonate carbonyl carbon would be found in the 13 C NMR spectra. However, if CO 2 only reacts with the anion, it is difficult to explain why there are four new peaks in the 13 C NMR spectra after absorption. As reported in our previous work [17], CO 2 reacted with the -OH group of EG, forming a carbonate species when it was captured by the DESs composed of azolide ILs and EG. Therefore, we think that CO 2 may also react with EG in the DESs [Et 4 N] [Car]:EG (1:2).
ing a carbonate species when it was captured by the DESs composed of azolide ILs and EG. Therefore, we think that CO2 may also react with EG in the DESs [Et4N] [Car]:EG (1:2).
In order to confirm our assumption, the 1 H-13 C 2D heteronuclear multiple bond correlation (HMBC) spectroscopy was used to further investigate the interaction between CO2 and [Et4N] [Car]:EG (1:2). The 1 H-13 C HMBC spectrum of [Et4N] [Car]:EG (1:2) after CO2 uptake was shown in Figure 3. As can be seen in Figure 3, H-2 correlated with C-3, and H-3 correlated not only with C-2 but also with C-4. The HMBC results confirmed that CO2 reacted with EG. On the basis of the HMBC results and previous results [23,24], the new peak at 157.7 (C-4) ppm was the carbonyl carbon of the carbonate formed by CO2 and EG. The new hydrogen peaks (H-2 and H-3) and the new carbon peaks (C-2 and C-3) can be ascribed to the methylene groups of the EG-based carbonate. Moreover, C-j did not correlate with H-2 or H-3, suggesting the peak at 160.3 (C-j) ppm was the carbonyl carbon of the Car-based carbonate [25]. The similar new peaks can also be found in the 1 H and 13 C NMR spectra of [Et4N] [Thy]:EG (1:2) after absorption ( Figure S3).   In order to confirm our assumption, the 1 H-13 C 2D heteronuclear multiple bond correlation (HMBC) spectroscopy was used to further investigate the interaction between CO 2 and [Et 4 N] [Car]:EG (1:2). The 1 H-13 C HMBC spectrum of [Et 4 N] [Car]:EG (1:2) after CO 2 uptake was shown in Figure 3. As can be seen in Figure 3, H-2 correlated with C-3, and H-3 correlated not only with C-2 but also with C-4. The HMBC results confirmed that CO 2 reacted with EG. On the basis of the HMBC results and previous results [23,24], the new peak at 157.7 (C-4) ppm was the carbonyl carbon of the carbonate formed by CO 2 and EG. The new hydrogen peaks (H-2 and H-3) and the new carbon peaks (C-2 and C-3) can be ascribed to the methylene groups of the EG-based carbonate. Moreover, C-j did not correlate with H-2 or H-3, suggesting the peak at 160.3 (C-j) ppm was the carbonyl carbon of the Car-based carbonate [25]. The similar new peaks can also be found in the 1 H and 13 C NMR spectra of [Et 4 N] [Thy]:EG (1:2) after absorption ( Figure S3).
The interaction between CO 2 and [Et 4 N] [Car]:EG (1:2) was also supported by the FTIR results. As shown in Figure 4, compared to the FTIR spectrum of the virgin [Et 4 N] [Car]:EG (1:2), a new peak appeared near at 1635 cm −1 , and a shoulder was observed at around 1614 cm −1 . The peak at 1635 cm −1 can be attributed to the C=O stretching band of the EG-based carbonate, [24,26] whereas the shoulder at 1614 cm −1 was the C=O stretching band of the Car-based carbonate [27]. The FTIR results suggested again that CO 2 reacted with both EG and the [Car] − anion in the [Et 4 N] [Car]:EG (1:2). Moreover, after CO 2 capture, the new band at 1422 cm −1 was assigned as the symmetric stretching band of COO − , and a shoulder near 1382 cm −1 was probably related to the doubly degenerate stretching of carbonate [28]. The peak at 1594 cm −1 attributed to aromatic ring mode shifted to 1591 cm −1 after CO 2 absorption. Similarly, the C=O stretching modes of the EG-based and Thy-based carbonates appeared at 1627 and 1612 cm −1 (Figure S4), respectively.    The interaction between CO2 and [Et4N] [Car]:EG (1:2) was also supported by FTIR results. As shown in Figure 4, compared to the FTIR spectrum of the virgin [E [Car]:EG (1:2), a new peak appeared near at 1635 cm −1 , and a shoulder was observe around 1614 cm −1 . The peak at 1635 cm −1 can be attributed to the C=O stretching ban the EG-based carbonate, [24,26] whereas the shoulder at 1614 cm −1 was the C=O stretch band of the Car-based carbonate [27]. The FTIR results suggested again that CO2 reac with both EG and the [Car] − anion in the [Et4N] [Car]:EG (1:2). Moreover, after CO2 c ture, the new band at 1422 cm −1 was assigned as the symmetric stretching band of CO and a shoulder near 1382 cm −1 was probably related to the doubly degenerate stretch of carbonate [28]. The peak at 1594 cm −1 attributed to aromatic ring mode shifted to 1 cm −1 after CO2 absorption. Similarly, the C=O stretching modes of the EG-based and T based carbonates appeared at 1627 and 1612 cm −1 (Figure S4), respectively.  On the basis of the products and previous reports, a possible reaction mechanism between CO 2 and [Et 4 N] [Car]:EG (1:2) can be proposed, which involved two steps (Scheme 2a). In the first step, there was an acid-base reaction between the anion [Car] − and EG, producing carvacrol and the anion HO-CH 2 -CH 2 -O − . The position of equilibrium for this acid-base reaction can be determined by comparing the pK a values of carvacrol (pK a = 11.02) [29] and EG (pK a = 14.22) [30], and the value of the equilibrium constant (K eq ) can be obtained using the equation below (Equations (1) and (2)): pK eq = pK a (EG) − pK a (carvacrol) = 3.20 (1) K eq = 10 −3.20 = 6.31 × 10 −4  Figures 5 and 6, respectively. As shown in Figure 5a, no new peaks can be found after absorption. The H-f and H-d obviously shifted downfield from 6.61 and 6.23 ppm to 6.86 and 6.50 ppm, respectively. However, the chemical shifts of the peaks of H-1′ (from 7.50 to 7.57 ppm) and H-3′ (from 6.72 to 6.75 ppm) on the imidazole ring changed very little. In the 13 C NMR spectra, only one new peak at 160.3 ppm (C-j) can be found after reaction, and this new carbon peak was the same as the Car-based carbonate peak (C-j) of the [Et4N] [Car]:EG (1:2) + CO2. Moreover, the C-a moved upfield obviously from 164.0 to 156.3 ppm, whereas C-f moved downfield from 110.8 to 113.0 ppm. The peak of C-1′ of 4CH3-Im shifted little from 135.6 to 134.8 ppm, and the chemical shift of the peak of C-2′ (131.2 ppm) did not move. These NMR results discussed above indicated that CO2 may only react with the anion [Car] − in the DESs [Et4N] [Car]:4CH3-Im (1:2), and the reaction between CO2 and the anion [4CH3-Im] − did not happen. Furthermore, the 1 H and 13 C NMR results ( Figure S5) 3 -Im] − . The pK eq value of this reaction can also be calculated using the pK a values of carvacrol and 4CH 3 -Im (pK a ≈ 14.52) [31]. Therefore, it is reasonable to assume that CO 2  The NMR and FTIR spectra of [Et 4 N] [Car]:4CH 3 -Im (1:2) before and after CO 2 absorption were presented in Figures 5 and 6, respectively. As shown in Figure 5a, no new peaks can be found after absorption. The H-f and H-d obviously shifted downfield from 6.61 and 6.23 ppm to 6.86 and 6.50 ppm, respectively. However, the chemical shifts of the peaks of H-1 (from 7.50 to 7.57 ppm) and H-3 (from 6.72 to 6.75 ppm) on the imidazole ring changed very little. In the 13 C NMR spectra, only one new peak at 160.3 ppm (C-j) can be found after reaction, and this new carbon peak was the same as the Car-based carbonate peak (  As seen in Figure 6, a new, sharp peak at 1614 cm −1 attributed to C=O stretching band can be observed in the presence of CO2, which is in agreement with the C=O stretching frequency around at 1614 cm −1 of the Car-based carbonate in the [Et4N] [Car]:EG (1:2) + CO2 system, suggesting strongly that CO2 reacted with the anion [Car] − . Moreover, one new peak at 1423 cm −1 due to the symmetric stretching band of COO − can be found. One new band at 1381 cm −1 , probably related to the doubly degenerate stretching of carbonate, can also be observed. The peak at 1592 cm −1 attributed to aromatic ring mode shifted to 1590 cm −1 after CO2 absorption. For the [Et4N] [Thy]:4CH3-Im (1:2) solvent, one new peak at 1611 cm −1 corresponding to the C=O stretching of the carbonate species was also found in the FTIR spectra ( Figure S6) after CO2 was loaded.
The reaction between CO2 and [Et4N] [Car]:4CH3-Im (1:2) was also studied using 1 H-13 C HMBC spectra and mass spectroscopy. In the HMBC spectra (Figure 7), there was no correlation peak between carbonyl carbon C-j and the H-1′ or H-3′on the imidazole ring, suggesting that CO2 was not bonded to the anion [4CH3-Im] − . In the mass spectrum (Figure 8), one peak at m/z 194.96 and one strong peak at m/z 216.96 can be found, corresponding to the ions [Car+CO2+H] + and [Car+CO2+Na] + , respectively. In addition, no peaks related to [4CH3-Im+CO2+H] + (m/z 127.1) or [4CH3-Im+CO2+Na] + (m/z 149.1) can be observed. The HMBC and mass spectroscopy results confirmed again that CO2 reacted with [Car] -and did not react with the anion [4CH3-Im] -when it was captured by the DESs  As seen in Figure 6, a new, sharp peak at 1614 cm −1 attributed to C=O stretching ba can be observed in the presence of CO2, which is in agreement with the C=O stretch frequency around at 1614 cm −1 of the Car-based carbonate in the [Et4N] [Car]:EG (1:2 CO2 system, suggesting strongly that CO2 reacted with the anion [Car] − . Moreover, o new peak at 1423 cm −1 due to the symmetric stretching band of COO − can be found. O new band at 1381 cm −1 , probably related to the doubly degenerate stretching of carbon can also be observed. The peak at 1592 cm −1 attributed to aromatic ring mode shifted 1590 cm −1 after CO2 absorption. For the [Et4N] [Thy]:4CH3-Im (1:2) solvent, one new p at 1611 cm −1 corresponding to the C=O stretching of the carbonate species was also fou in the FTIR spectra ( Figure S6) after CO2 was loaded.
The reaction between CO2 and [Et4N] [Car]:4CH3-Im (1:2) was also studied using 13 C HMBC spectra and mass spectroscopy. In the HMBC spectra (Figure 7), there was correlation peak between carbonyl carbon C-j and the H-1′ or H-3′on the imidazole ri suggesting that CO2 was not bonded to the anion [4CH3-Im] − . In the mass spectrum (F ure 8), one peak at m/z 194.96 and one strong peak at m/z 216.96 can be found, co sponding to the ions [Car+CO2+H] + and [Car+CO2+Na] + , respectively. In addition, peaks related to [4CH3-Im+CO2+H] + (m/z 127.1) or [4CH3-Im+CO2+Na] + (m/z 149.1) can observed. The HMBC and mass spectroscopy results confirmed again that CO2 reac As seen in Figure 6, a new, sharp peak at 1614 cm −1 attributed to C=O stretching band can be observed in the presence of CO 2 , which is in agreement with the C=O stretching frequency around at 1614 cm −1 of the Car-based carbonate in the [Et 4 N] [Car]:EG (1:2) + CO 2 system, suggesting strongly that CO 2 reacted with the anion [Car] − . Moreover, one new peak at 1423 cm −1 due to the symmetric stretching band of COO − can be found. One new band at 1381 cm −1 , probably related to the doubly degenerate stretching of carbonate, can also be observed. The peak at 1592 cm −1 attributed to aromatic ring mode shifted to 1590 cm −1 after CO 2 absorption. For the [Et 4 N] [Thy]:4CH 3 -Im (1:2) solvent, one new peak at 1611 cm −1 corresponding to the C=O stretching of the carbonate species was also found in the FTIR spectra ( Figure S6) after CO 2 was loaded.
The reaction between CO 2 and [Et 4 N] [Car]:4CH 3 -Im (1:2) was also studied using 1 H-13 C HMBC spectra and mass spectroscopy. In the HMBC spectra (Figure 7), there was no correlation peak between carbonyl carbon C-j and the H-1 or H-3 on the imidazole ring, suggesting that CO 2 was not bonded to the anion [4CH 3 -Im] − . In the mass spectrum (Figure 8)

Materials and Characterizations
Tetraethylammonium hydroxide (35% w/w) aqueous solution was purchased from Alfa Aesar (Shanghai, China). 4-Methylimidazole was purchased TCI (Shanghai, China). Ethylene glycol (99.5%) was purchased J&K Scientific Ltd (Beijing, China). Thymol (98%) absorbed by the DES can be calculated through the weight change of the tube with and without CO 2 .
In the desorption process, the glass tube was immersed in an oil bath at 80 • C. N 2 was bubbled through the solvent in the tube at a flow rate of 40 mL/min.

Conclusions
The four anion-functionalized DESs based on bio-derived phenols could efficiently capture CO 2 , up to 0.90 mol CO 2 /mol solvent. Hydrogen bond donors EG and 4CH 3 -Im impact greatly on the absorption mechanism by these DESs, and the reaction pathway between CO 2 and absorbents can be tuned by changing HBDs. CO 2 reacted with both EG and anions in the solvent when EG acted as the HBDs, forming the EG-based carbonate species and phenol-based carbonate species. Nevertheless, when 4CH 3 -Im was used as the HBDs, CO 2 only reacted with the phenolic anion in the solvent. The different reaction pathway may be due to the steric hinderance differences of the functional groups of the HBDs. We think the findings of this work will be very useful to the design of new, efficient anion-functionalized DESs for carbon capture and utilizations, which will promote the development of the DESs community.
Supplementary Materials: The following are available online, Table S1: The melting point (T m ), decomposition temperature (T d ) and viscosity of DESs, Table S2: Comparison of CO 2 capacities in this study with previously reported DESs, Figure S1: