Screening ionic liquids for developing advanced immobilization technology for CO2 separation

Developing immobilized-ionic liquids (ILs) sorbents is important for CO2 separation, and prior theoretically screening ILs is desirable considering the huge number of ILs. In this study, the compressibility of ILs was proposed as a new and additional index for screening ILs, and the developed predictive theoretical model, i.e., electrolyte perturbed-chain statistical associating fluid theory, was used to predict the properties for a wide variety of ILs in a wide temperature and pressure range to provide systematic data. In screening, firstly, the isothermal compressibilities of 272 ILs were predicted at pressures ranging from 1 to 6,000 bar and temperatures ranging from 298.15 to 323.15 K, and then 30 ILs were initially screened. Subsequently, the CO2 absorption capacities in these 30 ILs at temperatures from 298.15 to 323.15 K and pressures up to 50 bar were predicted, and 7 ILs were identified. In addition, the CO2 desorption enthalpies in these 7 ILs were estimated for further consideration. The performance of one of the screened ILs was verified with the data determined experimentally, evidencing that the screen is reasonable, and the consideration of IL-compressibility is essential when screening ILs for the immobilized-IL sorbents.


Introduction
Excessive CO 2 emissions have led to serious problems and received great concern (Figueres et al., 2018). According to the report from International Energy Agency, the amount of CO 2 emissions in 2020 was already around 30 Gt (Iea, 2020), and the excessive CO 2 emissions have led to environmental problems, such as global warming, glacial melting, and seawater acidification (Clark et al., 2020;Singh and Polvani, 2020;Hanna et al., 2021). To reduce CO 2 emissions, carbon capture and storage (CCS) has been proposed as one of the important options, in which CO 2 separation is often needed to capture CO 2 . Technologies have been developed for CO 2 separation, which can be divided into four categories, absorption, adsorption, membrane, and cryogenic (D'Alessandro  (Figueroa et al., 2008). Hence, developing energy-efficient and cost-effective technology for CO 2 separation is necessary, and novel absorbent development is one of the research focuses.
Ionic liquids (ILs) are molten salts with cation and anion as constituents while in the liquid state at room temperature (Zhang et al., 2017;Chen et al., 2020). ILs have the advantages of low vapor pressure, high thermal stability, and designable ability. Some ILs possess relatively high CO 2 solubility and selectivity over other gases (e.g., N 2 and CH 4 ) as well as low regeneration temperature and desorption enthalpy, making them desirable absorbents for CO 2 separation (Zhang et al., 2012;Wang et al., 2020). Many ILs have been designed and synthesized for CO 2 separation, such as pyrrolidinium-, imidazolium-, quaternary ammonium-, and quaternary phosphonium-based ILs (Blanchard et al., 2001;Anderson et al., 2007;Kilaru et al., 2007). However, their high cost and high viscosity (i.e., low CO 2 mass transfer rate) are still the current drawbacks hindering industrial applications of ILs on a large scale (Goodrich et al., 2010;Li et al., 2013).
Immobilizing ILs on porous materials is an effective strategy to overcome the above deficiencies (Yan et al., 2011). Zhang et al. found that the CO 2 absorption rate in the tetrabutylphosphonium amino acid salts immobilized on silica (SiO 2 ) was much higher than that in the bulk phase (Zhang et al., 2006). Wang et al. immobilized 1-ethyl-3-methylimidazolium amino acid on the surface of the polymethylmethacrylate microspheres. It was found that the CO 2 sorption equilibrium could be reached within 15 min (Wang et al., 2013b), which is also much faster compared to that in the bulk phase. It is widely accepted that the intensification of CO 2 absorption rate is owing to the large mass transfer area after IL immobilization (Zhang et al., 2009;Vicent-Luna et al., 2013;Wang et al., 2013a;Khan et al., 2014). However, the abnormally high CO 2 absorption capacity in the immobilized ILs was also observed. For instance, Zhang et al. noticed that the CO 2 absorption capacity in the 1butyl-3-methylimidazolium tetrafluoroborate [(C 4 mim) (BF 4 )] confined in mesoporous silica gels was improved by about 1.5 times (Zhang et al., 2010). Wu et al. observed that the CO 2 absorption capacity of 1-hexyl-3-methylimidazolium bis((trifluoromethyl)sulfonyl)imide ([C 6 mim][Tf 2 N]) was increased from 0.031 to 0.386 mol-CO 2 /mol-IL after immobilized on the surface of titanium dioxide (Wu et al., 2017). Therefore, the enhanced CO 2 absorption capacity can be another important reason to improve the CO 2 absorption performance, and immobilizing ILs on the porous materials is a promising way to promote the development of IL-based technologies for CO 2 separation.
To develop IL-immobilized sorbent for CO 2 separation, it is desirable to make a prior theoretical screening based on the properties of ILs, owing to the huge amount of ILs (up to 10 18 ) that can be potentially synthesized. Normally, CO 2 absorption capacity and selectivity as well as desorption enthalpy are considered to evaluate the CO 2 separation performance. Among them, CO 2 absorption capacity directly shows the ability of ILs to capture CO 2 , and desorption enthalpy reflects the energy usage for regeneration. Consequently, these two properties can be used to primarily evaluate the performance of ILs and are often used as the index in IL screening (Maiti, 2009;Palomar et al., 2011;Lee and Lin, 2015;Zhang et al., 2016;Taheri et al., 2021), which is valid for the technologies where the bulk ILs are used. While as discussed in the previous paragraph, when ILs are immobilized, the IL properties will change, causing a difference in CO 2 absorption capacity from its bulk. For example, based on molecular simulations, Pinilla et al. found that the density of 1,3-dimethylimidazolium chloride in a confined space is twice that in the bulk phase (Pinilla et al., 2005), Sha et al. observed a liquid-to-solid phase transition monolayer when 1,3-dimethylimidazolium chloride was confined between the graphite walls (Sha et al., 2008) and confirmed its higher melting point (Sha et al., 2009). This evidenced that when IL is immobilized, due to the asymmetric and strong interaction between the IL molecule and solid surface, the properties of ILs may be very different from its bulk phase, which need to be considered in screening immobilized ILs for CO 2 separation.
To consider the special properties of the immobilized ILs, the quantity related to the density change can be used as an additional index. Both advanced experiments and computer simulations have evidenced a higher density of the immobilized ILs compared to the bulk (Shi and Sorescu, 2010;Shi and Luebke, 2013). In particular, as reported by Gubbins et al., the molecules in a fluid or solid film adsorbed on a solid substrate experience strong compression, which is equivalent to a pressure up to several thousand bar (Gubbins et al., 2018). Based on these observations, it can be inferred that the enhanced CO 2 absorption capacity in the immobilized ILs comes from the complex interaction between the IL and substrate, which may be reflected by the density change. In other words, the compressibility of ILs may be an essential factor in determining the CO 2 absorption capacity, i.e., the higher the compressibility, the higher the potential to pressurize IL (via the interaction between IL and substrate) to increase density, and the higher the CO 2 absorption capacity due to the increased density. Therefore, besides the CO 2 absorption capacity and desorption enthalpy, the compressibility of IL at the pressure that is equivalent to the asymmetric and strong interaction with the substrate, might be an additional index in screening ILs when developing immobilized-ILs for CO 2 separation.
Due to the equivalent pressure is extremely high (up to several thousand bar), it is hard to determine the compressibility of ILs by using experimental measurements, and thus model prediction can be a viable option. A lot of theoretical models have been developed to predict the properties of ILs, which can be used as theoretical tools to Frontiers in Chemistry frontiersin.org 02 screen ILs, such as Conductor-like Screening Model for Real Solvents (COSMO-RS), COSMO segment activity coefficient model (COSMO-SAC), Soave Redlich Kwong, Universal Quasi-Chemical Model, and so on (Maiti, 2009;Gonzalez-Miquel et al., 2011;Lee and Lin, 2015;Farahipour et al., 2016;Kamgar and Rahimpour, 2016;Theo et al., 2016). However, none of them can be used at high pressures. In our previous work (Ji and Adidharma, 2008;Ji and Adidharma, 2010;Shen et al., 2015;Ji and Held, 2016;Shen et al., 2018;Sun et al., 2020), electrolyte perturbed-chain statistical associating fluid theory (ePC-SAFT) has been developed with ion-specific parameters, and, in particular, the model can be used up to high pressures owing to the consideration of the dispersion between IL-cations and IL-anions. The model performance has been verified extensively (Ji and Adidharma, 2012;Ji et al., 2013;Ji et al., 2014;Shen et al., 2015;Bülow et al., 2019;Sun et al., 2020;Sun et al., 2021). All these make ePC-SAFT a powerful tool for predicting the compressibility of ILs in a wide pressure range.
In this work, for the first time, the compressibility of IL at high pressures was proposed as a new index, which was then combined with the CO 2 absorption capacity and desorption enthalpy to screen ILs for developing immobilized-ILs for CO 2 separation. The ePC-SAFT model was used to predict the properties in a wide temperature and pressure range to provide systematic data for screening ILs step by step. In addition, to verify the screening results, the CO 2 separation performance of the screened ILs was compared with the experiments.
Theory ePC-SAFT ePC-SAFT was developed by Cameretti and Sadowski (Cameretti et al., 2005), as an extension of PC-SAFT proposed by Gross and Sadowski (Gross and Sadowski, 2000;Gross and Sadowski, 2001). In ePC-SAFT, the dimensionless residual Helmholtz energy (A res ) is expressed as Where A hc and A disp are the contributions from the hard chain and dispersive terms, respectively, and the expressions can be obtained from the literature (Gross and Sadowski, 2001). The ionic term (A ion ) was represented by the Debye-Hückel theory (Gross and Sadowski, 2001) Where κ is the inverse Debye-screening length with a unit of reciprocal meter, ε τ is the relative dielectric constant of the medium, ε 0 is the dielectric constant of vacuum, x j is the mole fraction of ion j, and q j is the charge of ion j. The units of ε 0 , κ, and q j are F/m, reciprocal meter, and coulomb, respectively. The definitions of κ and χ j have been described in detail in the original ePC-SAFT (Gross and Sadowski, 2001). In 2012, ePC-SAFT was extended to predict the properties of ILs, where each IL was assumed to be fully dissociated into ILanion and IL-cation . Each IL-ion with three parameters, i.e., segment number, the segment diameter, and the segment energy, while ε τ was set to be unity for pure ILs. In particular, dispersive interaction exists between IL-cation and ILanion, which is different from the ordinary aqueous electrolyte solutions.
In modeling, the parameters of ePC-SAFT for each IL-ion were taken from the literature, which were fitted to the experimental liquid-density of pure ILs or estimated with the linear equation based on the molar weight of IL-ions (Ji and Held, 2016).
Following ePC-SAFT, the density can be obtained from the dimensionless residual Helmholtz energy numerically at different temperatures and pressures, and then other thermodynamic properties can be further derived, such as compressibility, fugacity coefficient, etc. The combination of thermodynamic properties and phase equilibria can be used to predict the gas solubility, such as CO 2 solubility in ILs, and the relevant properties, such as desorption enthalpy, can be further obtained.

Compressibility
Following ePC-SAFT, the isothermal compression coefficient (κ T ) of ILs can be estimated with Eqn. 3: Where P is the pressure in bar, T is the temperature in Kalvin, and ρ is the density of ILs obtained from Eqn. 4.

CO 2 solubility
Following our previous study , the vapor pressure of ILs is negligible, and the phase equilibrium for CO 2 in a CO 2 -IL system can be represented by the following equation: Where x CO2 is the mole fraction of CO 2 in the liquid phase, φ L CO2 and φ V CO2 are the fugacity coefficients for CO 2 in the liquid and vapor phases, respectively, and v L and v V are the molar volumes Frontiers in Chemistry frontiersin.org of liquid and vapor phases, respectively. In this work, φ L CO2 , φ V CO2 , v L , and v V were calculated with ePC-SAFT, where the parameters of CO 2 were taken from the original PC-SAFT (Gross and Sadowski, 2001).

Desorption enthalpy
In this work, the ILs that physically absorb CO 2 were considered, and thus, the CO 2 desorption enthalpy (ΔH) can be calculated with the following equation: where H CO2 (T) is the Henry's constant of CO 2 in the IL. In this work, the value of H CO2 (T) was calculated with Eqn. 7 In Eqn. 7, the fugacity coefficient (φ V CO2 ) of CO 2 in the vapor phase was calculated with Eqs. 8-10: Z Pv V RT Where, μ res CO2 (T, v V ) is the chemical potential of CO 2 , k is the Boltzmann constant (1.380649 × 10 -23 J/K), Z is the compressibility factor, and R is the gas constant [8.314J/ (mol·K)].

Results and discussion
To screen ILs for developing IL-immobilization technology, in this work, it was achieved step by step. Firstly, compressibility was used to reflect the potential in density increase for enhancing the CO 2 absorption, and then ILs were primarily screened. Subsequently, the ILs with high compressibility were further screened based on the CO 2 absorption capacity. Additionally, the desorption enthalpy of the screened ILs was predicted for further verification. Finally, the screened ILs were verified with the available experimental data.
The screening is based on the properties predicted theoretically with ePC-SAFT. In predicting compressibility, to represent the interaction between the IL molecule and solid surface, we set the pressures ranging from 1 to 6,000 bar. Generally, for a CO 2 separation process, the absorption can be from room temperature to 313.15 K. Considering the heat release during absorption, the temperature was set from 298.15 to 323.15 K. Meanwhile, according to the applicability of ePC-SAFT and practical applications, the conditions in predicting the CO 2 absorption capacity and desorption enthalpy were set to be 298.15 to 323.15 K and 1 to 50 bar. The ePC-SAFT model with the available parameters was used to predict the isothermal compression coefficient of ILs. The results at 298.15 K and 1-6,000 bar were illustrated as an example as depicted in Figure 1. The green lines in Figure 1 represent the ILs with good compressibility, while the red ones indicate the ILs with poor compressibility. As shown in Figure 1, the compressibility of ILs can be significantly different, and the highest compressibility is around 3 times of the lowest one at atmospheric pressure. The IL compressibility with the significant difference makes it essential to consider the compressibility when screening ILs for developing the IL-immobilized absorbent.
Furthermore, it was found that the isothermal compression coefficient decreases with increasing pressure, indicating that it becomes more difficult to further increase the IL density at high pressures. The same phenomena were observed at other temperatures, as depicted in Supplementary Figures S1-S5  [THTDP] + always shows higher compressibility. The compressibilities of [C n mpy] + and [C n py] + are not related to the value of n. In contrast, according to the literatures, in the bulk phase, the physical CO 2 solubility can be improved by increasing the alkyl chain length on the cation (Anthony et al., 2005). This indicates that, if the CO 2 solubility is the only index for screening ILs, the ILs with the cation of long alkyl chain length will be selected. However, when the compressibility of ILs is considered, the screening result may be different.
According to the results shown in Figure 3 and Figure 4, and Supplementary Table S1, we can find that the compressibility of the ILs with [C 2 mpyr] + (IL-cation) at 1 bar is from 0.31 to 0.68 Gpa −1 , i.e., with a change of 0.37 Gpa −1 , while the compressibility of ILs with the anion of [Tf 2 N]at 1 bar only changes from 0.46 to 0.66 Gpa −1 , i.e., the difference is only 0.20 Gpa −1 . This observation indicates that the influence of IL-anion on compressibility is more than that of ILcation. Similar results can be observed for other cations and anions. Also, for the ILs with the same anion, their compressibility changes much less compared with the ILs with the same cation    Figures S6-S14). Therefore, we can conclude that the compressibility of ILs is mainly affected by anion. According to Figure 1, the κ T value for some ILs is not sensitive to the pressure, which makes it unobvious to perform screening. In order to differ the compressibility of ILs intuitively, the values of Δκ T (Δκ T = κ T,1 bar -κ T,6000 bar ) ranging from 298.15 to 323.15 K were calculated, and then ILs were screened. As depicted in Figure 5

CO 2 capacity
After the preliminary screening based on the compressibility, ePC-SAFT was also used to predict the CO 2 absorption capacity for these 30 ILs. The results of CO 2 absorption capacity are shown in Figure 6 and Supplementary Figures S15-S19 and listed in Supplementary Table S2.
As illustrated in Figure 6, the CO 2 absorption capacity in ILs increases with increasing pressure. Most of ILs with the anion of [Tf 2 N]have better CO 2 absorption capacity than the ILs with the anion of [eFAP] -. For the ILs with the cation of [C n mim] + , the CO 2 absorption capacity increases with increasing the chain length. For example, the order of CO 2 absorption capacity is in the order of [C 12  Desorption enthalpy CO 2 desorption enthalpy affects the energy demand in the desorption unit. H 2 O is the physical absorbent for biogas upgrading (i.e., CO 2 removal), while 30 wt% monoethanolamine (MEA) is a chemical absorbent for CO 2 separation from flue gases (Kim and Svendsen, 2011;Gupta et al., 2013;Chen et al., 2020). In this section, the CO 2 desorption enthalpy of the above-screened 7 ILs was

Verification
Among the screened 7 ILs, [C 6 mim][Tf 2 N] immobilized on the surface of titanium dioxide with a particle size of 25 nm (P25) has been studied in the previous study (Wu et al., 2017), showing desirable CO 2 separation performance. As shown in Figure 8, the CO 2 absorption capacity of [C 6 mim][Tf 2 N]/ P25 is about ten times that in the bulk [C 6 mim][Tf 2 N] at 298.15 K and atmospheric pressure. Banu et al. (Banu et al., 2013) studied CO 2 in the [C 2 mim][Tf 2 N] and [C 6 mim][Tf 2 N] confined in the ceramic porous membrane at 298.15 K and atmospheric pressure. Figure 9 shows that after immobilization, the CO 2 absorption capacities in [C 2 mim][Tf 2 N] and [C 6 mim][Tf 2 N] were enhanced by about 1.6 and 2 times, respectively, and the CO 2 absorption capacity intensification in [C 6 mim][Tf 2 N] is higher than that in [C 2 mim][Tf 2 N]. According to the result in Figure 4, the compressibility of [C 6 mim][Tf 2 N] is much higher than that of [C 2 mim][Tf 2 N], which is consistent with the trend of the CO 2 absorption capacity. Therefore, it can be concluded that the index proposed in this work is reasonable, and the compressibility of ILs is an important index in screening ILs for developing IL-immobilized absorbent for CO 2 separation.
As only limited experimental results on the immobilized ILs are available, in the future, the CO 2 separation performance of IL-immobilized absorbents will be determined experimentally to systematically verify the screen method and results. In addition, the CO 2 separation performance is a combination of thermodynamics and kinetics, and some ILs can absorb CO 2 chemically. In the future, the research on the screen will be extended to include CO 2 chemical absorption and consider the kinetic contribution to the CO 2 separation performance.

Conclusion
In this study, a new additional index, i.e., the compressibility of ILs, was proposed to screen ILs for developing IL-immobilization technology for CO 2 separation. The developed ePC-SAFT model was used as a theoretical tool to predict reliable and systematic data for screening. From 272 physical ILs, 30 ILs were selected firstly based on the compressibility. Then, 7 ILs, i.e., [C 12    Frontiers in Chemistry frontiersin.org be designed to further verify this screen index, and the quantitative relationship between compressibility and CO 2 absorption capacity will be studied to further understand the mechanism of the enhanced CO 2 absorption capacity in the immobilized ILs.

Data availability statement
The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding authors. Frontiers in Chemistry frontiersin.org