1. Introduction
In the context of mitigating climate change, achieving carbon neutrality is imperative, particularly considering the correlation between increased atmospheric carbon dioxide (CO
2) concentrations and the occurrence of extreme weather phenomena such as El Niño. In addition to CO
2 capture of industrial emissions, the emerging technology of carbon dioxide capture and storage (CCS), represented by direct air capture (DAC), is evolving as one of the more effective methods for reducing atmospheric CO
2. DAC technologies are characterized by the use of advanced absorbents capable of extracting CO
2 at sub-500 ppm concentrations and must exhibit operational stability across diverse atmospheric conditions [[
1], [
2], [
3], [
4], [
5], [
6]]. A significant challenge in advancing DAC technology lies in optimizing the absorption efficiency for these lower CO
2 concentrations, a range where existing CCS strategies are notably deficient. Another critical focus is on reducing the desorption temperature for adsorbents, particularly 2-aminoethanol (MEA), which conventionally requires temperatures exceeding 393 K [
3,[
7], [
8], [
9], [
10], [
11]]. Lowering this temperature threshold is vital for reducing energy demand and enhancing the economic viability of CCS. Optimal adsorbents should also facilitate straightforward, energy-efficient regeneration following CO
2 capture.
Regarding CO
2 capture technologies designed to address climate challenges, advancements have been made in solid amine-based adsorbents and CO
2 absorption systems based on phase separation [[
12], [
13], [
14], [
15], [
16], [
17], [
18], [
19], [
20], [
21], [
22], [
23], [
24], [
25], [
26], [
27], [
28]]. CO
2 absorption in liquid-phase systems has been improved by modifying the structure of amine adsorbents, with pyrrolidine-based diamines showing higher efficiency than conventional monoethanolamine (MEA), as demonstrated by Hanusch et al [
29]. Thus, there is a continued need for enhanced CO
2 uptake, especially at lower concentrations. Another innovative approach is liquid–liquid phase separation in amine–water mixtures with a lower critical solution temperature, which markedly reduces adsorbent regeneration costs [
24,
30,
31]. These systems have a higher CO
2 capacity and are more cost-effective than traditional MEA systems; however, their use is limited by the volatile and corrosive nature of the solvents involved. Improving the effectiveness of these systems under varying CO
2 flow conditions is critical. Liquid–solid phase separation could bypass equilibrium limitations in the CO
2 uptake reaction, leading to more efficient absorption. This approach is advantageous, as liquid absorbents are generally more effective than other types of absorbents in contacting dissolved CO
2, facilitating absorption from large gas streams. Liquid–solid phase separation systems using compounds such as triethylenetetramine, polyethylene glycol, bis(iminoguanidine), and potassium alaninate have been reported. Custelcean et al. [
23] developed a DAC system using aqueous amino acids in conjunction with solid bis-iminoguanidines, resulting in a synergistic system that enhances CO
2 absorption and facilitates efficient capture from ambient air. Cai et al. [
32] provided a method for capturing CO
2 from air using a trichelating iminoguanidine ligand that forms stable carbonate crystals, enabling efficient CO
2 sequestration and easy regeneration under mild conditions. Jang et al. [
33] explored a novel CO
2 capture approach by optimizing the chemical absorption–precipitation process, significantly reducing the energy required for solvent regeneration. The low solubility of the absorbent necessitates a large solvent volume, underscoring the need for the development of more efficient systems. Further research has focused on creating simple, versatile solid–liquid separation systems that are environmentally suitable for CO
2 uptake and can achieve efficient desorption at lower temperatures.
The desorption temperature and energy consumption of PCAs present significant opportunities for optimization. Here, we introduce a novel solid–liquid phase change system designed for efficient CO
2 capture, characterized by high CO
2 absorption rates, low-temperature thermal regeneration, and reduced energy consumption. Isophorone diamine (IPDA) with an amino-cyclohexyl (C
6H
11NH
2) group was selected as the phase change host because of its low energy consumption and low-temperature thermal regeneration [
34,
35]. Six ketone organic solvent molecules, comprising four acyclic ketones and two cyclic ketones, were introduced and compared as phase-change mediators in conjunction with the IPDA host to form a comprehensive PCA system. We comprehensively evaluated the performance of the PCAs in terms of CO
2 absorption, desorption, and cycling, as well as assessing energy consumption on the basis of engineering models, material characterization, density functional theory (DFT) calculations, molecular dynamics (MD) simulations, noncovalent bond interaction (NCI) studies, and technoeconomic assessment (TEA). The results revealed that the CO
2 removal efficiency of this PCA system exceeded 95% at carbon dioxide concentrations ranging from 400 to 10 000 ppm in air [[
36], [
37], [
38]] and from 50 000 to 150 000 ppm in industrial emissions. The highest CO
2 removal efficiency was achieved with the IPDA-methyl isobutyl ketone (MIK) variant. These results demonstrate a wide range of applications for CO
2 capture with this PCA system. The absorbed CO
2 was completely desorbed and released at 333 K, and the system could be reused for up to 20 absorption–desorption cycles with low decay, demonstrating the low-temperature thermal regeneration and excellent cycling stability of this PCA system. This release behavior is based on the effect of NCI on noncyclic ketone molecules, which prevents extensive hydrogen bonding in IPDA(NHCOO
−)
2, allowing smaller-scale hydrogen bonds to release CO
2 at a lower temperature. The efficiency of IPDA during repeated CO
2 adsorption–desorption remained above 95% for 20 h under DAC conditions, suggesting its stability and practicality for continuous operation. The high CO
2 capture rate (482.6 mmol·h
−1 per mole of amine) further supports the suitability of IPDA for practical applications. Additionally, the ketone-based organic solution system featuring the IPDA PCA demonstrated high flux stability without visible foaming, stability under low-temperature thermal regeneration, and recyclability, and the reaction mechanism was clearly different from that of the aqueous system. These properties make this ketone-based IPDA PCA system a promising candidate for wide-scale CO
2 capture applications, and the TEA results indicate high potential for the industrialization of ketone-based IPDA PCAs.
2. Methods and Experiments
2.1. Chemicals
In this study, we used high-purity chemicals to ensure the accuracy of our experiments. IPDA, acetone, 2-butanone, 3-pentanone, MIK, cyclohexanone (CYC), and isophorone, all with purities exceeding 99.0%, were obtained from Shanghai Aladdin Biochemical Technology Co., Ltd., China. MEA and acetone-D4 (C3D6O), with purities above 99.0% and 99.8%, respectively, were obtained from Cambridge Isotope Laboratories, Inc., USA. High-purity nitrogen (N2) and carbon dioxide (CO2) gases, with a minimum purity of 99.99 v%, were supplied by Weichuang Gas Co., Ltd., China. Deionized water was used in all experiments, and all chemicals were used as received without additional purification.
2.2. Experimental procedure
First, a series of ketone-based IPDA-ketone PCA systems were synthesized utilizing a vacuum preactivation method. The IPDA ketone-based PCA system, configured with the required proportions, was placed into a vacuum heating stirrer (the vacuum heating stirrer was placed in the glove box in advance to maintain a nitrogen environment). The device was sealed, pumped to the vacuum state, stirred at 150 r·min−1, and heated at 333.15 K for more than 1 h for activation. During the experiment, the vacuum gauge was monitored; if gas was generated, the vacuum pump was turned on to pump the gas out to maintain stable vacuum conditions for more than 1 h. Afterward, the samples to be tested were removed, sealed and immediately transferred. The absorption experiments were conducted in a 50 mL graduated tube, which was submerged in a stirred water bath maintained in a temperature range of 293.15 to 303.15 K and agitated at a rate of 150 r·min−1. The setup for the absorption process is shown in Scheme S2 in Appendix A. The total volume of the ketone-based IPDA PCA system was 15 mL. For CO2 removal tests at 400 to 10 000 ppm CO2, a concentration of 136.3 mmol·L−1 was used, and at 50 000 to 150 000 ppm CO2, a concentration of 1.0 mol·L−1 was used. Prepared CO2 gas was continuously introduced into the absorption reactor (graduated tube) from the mixing tank at a flow rate of 80 mL·min−1. The inlet and outlet CO2 flow rates were monitored using a soap film flow meter. After saturation absorption was reached, the IPDA-ketone-based PCA system separated into two phases: solid and liquid. In the noncyclic ketone-based solvent system, a clear distinction between the solid and liquid phases was visible, whereas in the cyclic ketone-based solvent system, the distinction was less evident, or the phases even appeared gelatinous. Carbon dioxide was concentrated in the solid phase, so only this CO2-rich solid phase needed to be regenerated. The solid-phase products were subsequently separated and transferred to an oil bath for thermal desorption tests, which were conducted at temperatures ranging from 313.15 to 333.15 K for durations of 30 to 60 min (Fig. S1 in Appendix A). The integration process of the absorbent is a key consideration for industrial application. Additionally, the characterization results for the absorption products are provided in the Text S1 in Appendix A. The thermal desorption process was modeled with the biphase process integration model in Schematic S1, the absorption and desorption units are presented in Schematic S2, and the regeneration energy consumption evaluation device is shown in Schematic S3.
Quantum calculations were conducted using Gaussian 16 software. The Becke three-parameter hybrid exchange-correlation functional (B3LYP) with the 6-311+G(d) basis set was employed for reaction calculations. Further refinement of the single-point energies for all stationary points was carried out with the following method. We used Grimme’s D3 dispersion correction to account for van der Waals interactions. For geometry optimization, the parameters were set to achieve forces below 10−5 Hartree/Bohr and energy changes less than 10−7 Hartree. The convergence threshold for energy calculations in the self-consistent-field procedure was established at less than 10−8 Hartree. Solvent effects were considered using the solvation model density (SMD) implicit solvent model, incorporating solvents such as acetone and methyl ethyl ketone. To analyze NCIs and understand the atomic interactions in the host–guest systems, we conducted reduced density gradient (RDG) analysis using Multiwfn software. These interactions were visualized using the VMD 1.9.3 program, and the results highlighted weak interactions and atomic contributions in the seven solution systems with different compositions (Table S1 in Appendix A), which were constructed in cubic simulation boxes. The molecular electrostatic potential (and Fukui function were calculated using the same computational setup.
MD simulations were conducted using the Forcite module with the COMPASS III force field in Materials Studio 2020 (USA). Van der Waals and Coulomb interactions were considered, using atom-based and Ewald methods with a 15.5 Å cutoff. The motion equations were integrated at a 1 fs time step. Following energy minimization, each system underwent a 400 ps relaxation period under periodic boundary conditions in the NPT ensemble (pressure (P) = 1 atmosphere, temperature (T) = 298.0 K), and the Nose thermostat and Berendsen barostat were used for stabilization of the temperature, potential, and total energy. After equilibrium, a 400 ps simulation in the NVT ensemble was performed for trajectory analysis, calculation of the radial distribution function (RDF), and hydrogen bond number analysis. Hydrogen bonds between solvent molecules and IPDA were examined across the seven solutions using the following geometric criteria in Materials Studio: distance |A–H| < 2.7 Å and angle D–H–A > 120 degrees.
3. Results and discussion
3.1. Assessment of the absorption and desorption properties of the phase-change absorbents
The CO
2 uptake performance of six different ketone-based IPDA PCA systems was investigated under near-atmospheric pressure conditions using two gas paths simulating air-concentrated CO
2 gas (i.e., 100 and 10 000 ppm CO
2, equilibrated with N
2). As mentioned in the introduction, only nonaqueous systems were considered for CO
2 capture in this study. The CO
2 capture efficacy of the PCA systems was analyzed by recording the outlet CO
2 concentration over time using an infrared CO
2 analyzer.
Figs. 1(a)–(c) show the absorption profiles of a representative PCA system, which were determined according to Eqs. (S1)–(S5) in Appendix A. Notably,
Figs. 1(a)–(d) depict the breakthrough curves corresponding to increasing CO
2 concentrations in the feed gas from 400 to 10 000 ppm. The breakthrough times for each system at 400 ppm were as follows: IPDA-acetone, 696 min; IPDA-2-butanone, 802 min; IPDA-3-pentanone, 1030 min; IPDA-MIK, 1324 min; IPDA-CYC, 946 min, with a removal rate exceeding 95% lasting only 207 min; and IPDA-isophorone, 800 min, achieving a maximum removal rate of only 67.83%. The breakthrough times and corresponding ultimate CO
2 loading volumes for each system at a concentration of 10 000 ppm were as follows: IPDA-acetone, 85 min, 0.7 mol·mol
−1; IPDA-2-butanone; 104 min, 0.9 mol·mol
−1; IPDA-3-pentanone, 89 min, 0.82 mol·mol
−1; IPDA-MIK, 122 min, 1.11 mol·mol
−1; IPDA-CYC, 88 min, 0.73 mol·mol
−1; and IPDA-isophorone, 72 min, 0.61 mol·mol
−1. Furthermore, for high-concentration CO
2 capture, the breakthrough times for the IPDA-MIK system were 100 min (150 000 ppm), 181 min (100 000 ppm), and 239 min (50 000 ppm), with all high concentrations demonstrating a CO
2 removal rate exceeding 95%. The dynamic equilibrium time for CO
2 uptake was significantly shorter for all the systems, suggesting that the increase in the overall driving force improved the uptake rate. The outlet CO
2 concentration profiles of all the ketone-based IPDA PCA systems showed an abrupt drop followed by a slow rise back to the feed concentration. This phenomenon occurred only in solutions with relatively high carbon dioxide loadings, and phase change behavior was observed in all of these solutions. In addition, the absorption capacity depended mainly on the concentration of IPDA as a reaction component in the solution. These characteristics provide useful information for the design of a suitable absorption process or contactor.
The desorption performance of adsorbents, which is closely related to regeneration energy consumption and adsorbent reusability, is another key criterion for evaluating new biphasic solvents. In this study, the desorption performance of the ketone-based medium IPDA PCA systems at 333.15 K was investigated. For comparison, systems containing IPDA with different types of ketone-based phase change solvents were also studied. The CO
2 desorption rate versus time curves for the six PCAs are shown in
Figs. 1(d) and
(e). For all adsorbents, the amount of CO
2 desorbed and the desorption rate first increased to a maximum and then gradually decreased. Notably, the desorption rates followed a fast and then slow pattern, and the maximum value was approximately 0.60 mol·min
−1·L
−1. In contrast, the IPDA solution showed slow CO
2 release, with a maximum desorption rate of only 0.07 mol·min
−1·L
−1, which was much lower than the rate for IPDA-MIK. In addition, desorption from the noncyclic ketone-based IPDA PCA system was essentially complete within 30 minutes, whereas desorption from the cyclic system took approximately 50 minutes. As shown in
Fig. 1(f), the final desorption amounts of the noncyclic ketone-based medium systems were 3.45 mol·L
−1 (acetone), 6.77 mol·L
−1 (2-butanone), 5.99 mol·L
−1 (3-pentanone), and 7.62 mol·L
−1 (MIK), and those for the cyclic ketone-based medium system were 5.77 mol·L
−1 (CYC) and 7.39 mol·L
−1 (isophorone), which shows that the noncyclic ketone-based system is not suitable as an IPDA PCA. The desorption promotion rate of the system was better than that of the cyclic ketone-based system [
35,[
39], [
40], [
41]].
3.2. Assessment of regeneration energy consumption and recyclability
Desorption performance is another key criterion for evaluating new absorbents. In this study, the desorption of the IPDA-MIK system was investigated at temperatures from 313.15 to 333.15 K. For comparison, an aqueous-phase experiment with MEA was also performed. As shown in
Fig. 2(a), the CO
2 cyclic loadings of both IPDA-MIK and MEA increased with increasing desorption temperature, indicating that high temperature favors CO
2 removal. Surprisingly, the cyclic loading of IPDA-MIK was much higher than that of MEA at the same temperature. For example, at 333.15 K, the cyclic loading of IPDA-MIK reached 1.11 mol·mol
−1, which was almost fifteen times higher than that of MEA (0.13 mol·mol
−1). This result demonstrates the excellent desorption performance of the IPDA-MIK solid–liquid PCA. The vapor–liquid equilibrium (VLE) values of the six IPDA-ketone-based PCAs were determined at 313.15 and 323.15 K. The VLE plots [
42] show that the partial pressure of CO
2 in the six IPDA-ketone-based PCAs increased with increasing CO
2 loading and temperature, as shown in Figs. S1 and S2 in Appendix A. According to Eqs. (S6)–(S10) in Appendix A, VLE data can be used to calculate the enthalpy of reaction (Δ
Habs) of CO
2 uptake in biphasic solvents with different CO
2 loadings. The relevant parameters for the energy consumption calculations are outlined in Tables S1–S3. The heat consumption of the six IPDA-ketone-based PCAs during regeneration was evaluated and compared with that of 30 wt% MEA, which is typical in industrial carbon capture applications, and the results are displayed in
Fig. 2(b).
The heat consumption during regeneration of the six IPDA-ketone-based PCAs was evaluated and compared with that of 30 wt% MEA, and the results are shown in
Fig. 2(b). The heats of regeneration of the six ketone-based IPDA PCAs were as follows: IPDA-acetone, 1.14 GJ·t
−1 CO
2; IPDA-2-butanone, 1.32 GJ·t
−1 CO
2; IPDA-3-pentanone, 1.29 GJ·t
−1 CO
2; IPDA-MIK, 0.88 GJ·t
−1 CO
2; IPDA-CYC, 1.03 GJ·t
−1 CO
2; and IPDA-isophorone, 1.11 GJ·t
−1 CO
2. These values were 64.9%–76.7% lower than that of the MEA solution (3.77 GJ·t
−1 CO
2). The heat of desorption (
Qdes) of the six IPDA-ketone-based PCAs were as follows: IPDA-acetone, 0.74 GJ·t
−1 CO
2; IPDA-2-butanone, 0.73 GJ·t
−1 CO
2; IPDA-3-pentanone, 0.72 GJ·t
−1 CO
2; IPDA-MIK, 0.65 GJ·t
−1 CO
2; IPDA-CYC, 0.74 GJ·t
−1 CO
2; and IPDA-isophorone, 0.75 GJ·t
−1 CO
2. These values were 57.63%–63.28% that of the MEA solution (1.77 GJ·t
−1 CO
2), which was due to the lower heat of reaction between CO
2 and IPDA. The heat of sensing (
Qsen) values of the six IPDA-keto-based PCAs were as follows: IPDA-acetone, 0.40 GJ·t
−1 CO
2; IPDA-2-butanone, 0.59 GJ·t
−1 CO
2; IPDA-3-pentanone, 0.57 GJ·t
−1 CO
2; IPDA-MIK, 0.23 GJ·t
−1 CO
2; IPDA-CYC, 0.29 GJ·t
−1 CO
2; and IPDA-isophorone, 0.36 GJ·t
−1 CO
2. These values were 34.44%–74.44% that of the MEA solution (0.9 GJ·t
−1 CO
2), which is attributed to the fact that ketone-based organic molecules have a much lower specific heat capacity than H
2O [[
43], [
44], [
45]]. Among the six IPDA-ketone-based PCA systems in this study, the IPDA-MIK system had the lowest energy consumption of 0.88 GJ·t
−1 CO
2; this system has advantages in terms of energy consumption compared with other PCA systems [
35,[
45], [
46], [
47], [
48], [
49], [
50], [
51], [
52]] and has great potential for energy savings, as shown in Table S4 in Appendix A. Since the binary anhydrous solid–liquid two-phase solvent can be regenerated by simply transferring the enriched phase to the vapor extraction column, the ketone-based solvent is not involved in the desorption process, and there is no latent heat (heat carried away by vapor) to consider. A comparison of the energy consumption during regeneration of the six IPDA-ketone-based PCAs with that of other absorbers is presented in Table S5 in Appendix A. In conclusion, the six new IPDA-ketone-based solid–liquid PCAs have a low calorific value for regeneration, which would considerably reduce energy consumption in industrial applications. The experimental cycling results under the optimal conditions are shown in
Fig. 2(c). The regeneration efficiency of the first fully loaded enriched phase under low-temperature thermal regeneration at 333 K was 94.06%, and the efficiency of the twentieth full-load absorption–desorption cycle exceeded 93%, which indicates that this new type of absorbent has a highly stable regeneration capability.
3.3. Product analysis and characterization
On the basis of the above analysis, the solid product formed after the CO
2 absorption reaction of the ketone-based IPDA PCA system is a white powder, and its main component is IPDA(NHCOO
−)
2. The crystal structure of the solid product was analyzed by powder X-ray diffraction (PXRD). As shown in
Fig. 3(a), the peaks at diffraction angles of 6.1°, 7.1°, 12.20°, 14.94°, 16.14°, 16.79°, 17.40°, 18.42°, 19.66°, and 22.95° for the solid products formed by the absorption of CO
2 by the four noncyclic ketone-based PCA systems correspond to IPDA(NHCOO
−)
2, which indicates that the solid products formed crystals [
35]. Crystallization of the solid product facilitates the separation of crystalline products from the solid–liquid phase change products following CO
2 absorption, thereby overcoming the challenges associated with viscous products that are difficult to separate, which is advantageous for industrial applications. In comparison, the crystalline diffraction peaks of the products formed after CO
2 absorption in the four cyclic ketone-based PCA systems were less pronounced, suggesting that the solid products do not readily crystallize, resulting in viscous products that are difficult to separate, which is not conducive to industrial applications. The attenuated total reflectance-Fourier transform infrared spectroscopy (ATR-FTIR) results in
Fig. 3(b) include a stretching vibrational peak at approximately 1570 cm
−1 attributed to the carbonyl group (C=O) in IPDA-carbamate, while the peak at approximately 1712 cm
−1 corresponds to primary amine N–H bonds associated with IPDA-carbamate, with the most significant intensity observed for the IPDA-MIK product. Additionally, the stretching vibrational peaks at approximately 2871 and 2956 cm
−1 are attributed to C–H bonds, and the peak at approximately 3360 cm
−1 is associated with O–H bonds [
53]. In the thermogravimetric analysis (TGA) results (
Fig. 3(c)), the six samples exhibited three weight loss stages with increasing temperature. In the first stage (25–60 °C), the weights of the five PCAs remained almost unchanged, except for a 27% weight loss in the IPDA-3-acetone system, which was attributed to the high volatility of acetone. In the second stage (60–110 °C), weight loss was dominated by IPDA-carbamate products, and some ketone-based organic molecules remained in the solid product during solid formation. In the third stage (above 110 °C), the weight loss was dominated by IPDA and ketone-based organic molecules. In addition, thermal decomposition of carbamate occurred, which was a distinct process from thermal desorption. The weight loss of each PCA system was as follows: IPDA-acetone (73.37%), IPDA-2-butanone (70.66%), IPDA-3-pentanone (69.60%), IPDA-MIK (77.84%), IPDA-CYC (87.66%), and IPDA-isophorone (91.07%).
In addition, the weight loss of the four noncyclic ketone-based systems was almost complete at approximately 190 °C, while the weight loss of the two cyclic ketone-based systems was complete at approximately 650 °C, reflecting the easy desorption and separation of the IPDA noncyclic ketone-based systems after CO
2 uptake and the difficult desorption and separation of the IPDA-cyclic ketone-based systems after CO
2 uptake.
Fig. 3(d) display the products of CO
2 absorption by the IPDA-ketone system. Panels d1-d6 correspond to IPDA-acetone, IPDA-2-butanone, IPDA-3-pentanone, IPDA-MIK, IPDA-CYC, and IPDA-isophorone. As shown in
Figs. 3(e) and
(f), liquid
13C NMR and solid
13C NMR analysis of the six IPDA-ketone PCAs and their solid products formed after CO
2 uptake was performed to analyze the structural changes in the products. In the liquid
13C NMR results, the ketone solvents correspond to acetone (C1), 2-butanone (C1–C3), 3-pentanone (C1–C3), MIK (C1–C4), CYC (C1–C4), and isophorone (C1–C5). In the solid
13C NMR results, C1–C7 clearly represent the carbon-containing structural peaks of IPDA, whereas C8 represents the carbon-containing peaks of carbamate formed by the absorption of CO
2 in the ketone-based IPDA system [
35]. The carbon-containing structure peaks of IPDA for the four noncyclic IPDA-keto-based PCA solid products are clear and consistent, whereas the corresponding peaks for IPDA-CYC show noise and blurring, and the peak for IPDA-isophorone is almost undetectable, while that of the urethane salt is clear and evident. Comparative analysis of these results demonstrates the four noncyclic IPDA-ketone-based PCAs have a single solid product, which is easy to separate, while the two cyclic IPDA-ketone PCAs are complex and difficult to separate; these findings provide reference information for the selection of ketone-based solvents.
3.4. Comprehensive analysis of the reaction mechanism
To investigate the reaction properties of various types of molecules in the ketone-based IPDA PCA systems, the ESP [[
54], [
55], [
56]] and the Fukui index were calculated by the density functional theory-frontier molecular orbitals (DFT-FMO) method according to Eqs. (S11)–(S19) in Appendix A, and the results are presented in
Figs. 4(a)–(g) and Figs. S3–S5 in Appendix A. The red and blue regions in the ESP figures represent positive and negative charges, respectively, and the green and blue regions in the f
+, f
−, and CDD diagrams represent the positive and negative orbitals of each molecule, respectively. In the ESP results (Fig. S5 in Appendix A), the positive and negative electrostatic adsorption surfaces of IPDA with each type of ketone-based molecule are as follows: IPDA (65.11 eV, −42.53–22.58 eV), acetone (55.25 eV, −36.45–18.80 eV), 2-butanone (54.42 eV, −36.27–18.15 eV), 3-pentanone (52.87 eV, −36.16–16.71 eV), MIK (53.67 eV, −35.79–17.88 eV), CYC (53.99 eV, −37.78–16.21 eV), and isophorone (58.39 eV, −41.58–16.81 eV). According to the Fukui index results, the f
+, f
− and CDD poles of the four noncyclic ketone-based molecules and two cyclic ketone-based molecules are located on the oxygen atom, as are the electrophilic and nucleophilic poles. In contrast, the f
+, f
− and CDD poles of IPDA are on nitrogen atoms, and the pro-electrode and nucleophile poles are also on oxygen atoms. The Fukui index nucleophilicity calculations for each molecule is agree with the ESP calculations. The detailed data are provided in Tables S6–S28 in Appendix A. These results indicate that in the ketone-based IPDA PCA systems, both the oxygen atom in the ketone-based medium and the nitrogen atom in IPDA are the most active sites in the reaction; moreover, these findings are agreed with the high probability of contact of these two groups predicted by the RDF in the DFT-based MD simulations.
The absorption mechanisms of the PCAs, including the main reaction pathways and free energy barrier changes, were explored by quantum chemistry calculations (Gaussian 16 software package) taking into account solvent effects. The reaction equation on the left side of
Fig. 5(a) shows that the reaction of IPDA with CO
2 can be divided into two steps, which are the reactions of the two amino groups on IPDA with CO
2 to form two amino ions. The reaction was determined from computational simulation of one pathway, one initial state (IS), one final state (FS), and six intermediate and transition states (TS), as shown in
Fig. 5(b), where the main CO
2 uptake reaction processes are from IS-1 to TS-1 and from IS-2 to TS-2. The reaction process indicates an anhydrous system, different from the reaction mechanism of IPDA in aqueous media such as ionic liquids [
35], and is consistent with the results of product characterization by NMR, as well as calculations that satisfy the ESP and Fukui indices with respect to the reaction site. The deprotonation reaction processes are FS-1 to FS-1' and FS-2 to FS-2'. As shown in
Figs. 5(c)–(h), the main process in the CO
2 absorption reaction is from IS-1 to TS-1, and the free energy barrier of the original IPDA reaction is 33.9 KJ·mol
−1. For different PCA solvents, the main process in the CO
2 absorption reaction is from IS-2 to TS-2, and the free energy barrier of the original IPDA reaction is 46.6 KJ·mol
−1. It can be seen that the free energy barriers for the CO
2 uptake reaction of IPDA in the four noncyclic ketone-based phase-change solvents (acetone, 2-butanone, 3-pentanone, and MIK) are lower than those for the original uptake reactions in the main reactions IS-1 to TS-1 and IS-2 to TS-2 of the IPDA PCAs, whereas the free energy barriers for the CO
2 uptake reaction of IPDA in the two cyclic ketone-based phase-change solvents (CYC and isophorone) are lower than those of the original uptake reactions. These findings suggest that IPDA is more likely to undergo CO
2 uptake reactions in the presence of the noncyclic ketone-based phase-change solvents than under the original conditions, whereas these reactions were less likely in the presence of the cyclic ketone-based phase-change solvents than under the original conditions. Similar results were found for the desorption reactions. In addition, it is worth noting that the overall free energy barrier with acetone as a solvent was lower than that of the original reaction, so the CO
2 desorption reaction is not spontaneous; such as, IPDA(NHCOO
−)
2 is not stabilized in the IPDA-acetone system. In contrast, in the presence of cyclic phase change solvents, both CO
2 uptake and desorption reactions show a strong blocking effect. All the calculated results are consistent with the experimentally observed phenomena.
The MD modeling results for the ketone-based IPDA-based PCA systems, both before and after CO
2 absorption, are illustrated in
Figs. 6(a) and
(b), and Fig. S6 in Appendix A. Specifically, Fig. S6 presents the MD modeling results for the ketone-based IPDA-based PCAs before CO
2 absorption, including IPDA and IPDA(COO
−)
2. The RDF results in
Fig. 6(c) and statistics for the average number of hydrogen bonds in the six different ketone-based systems were calculated by MD modeling, as shown in
Fig. 6(d). In RDF calculations,
g(
r) is usually defined as the ratio of the probability density of finding another particle centered at some reference particle at a distance r from that particle to the probability density in the case of a random distribution. The RDF peaks at 2.3 and 3.1 Å assigned to N–H
g(
r) for IPDA(NHCOO
−)
2 increased (-510% and -245%), became sharper and shifted (0.5 Å) compared with the RDF N–H
g(
r) peak for IPDA after the system reacted with CO
2, which indicated that IPDA(NHCOO
−)
2 is more likely to agglomerate into IPDA-based molecular clusters than IPDA is. The RDF peak near 2.1 Å assigned to O–N
g(
r) for IPDA and IPDA(NHCOO
−)
2 was observed for the six ketone-based solvent molecules, with different relative changes: acetone (-70.9%), 2-butanone (-60.8%), 3-pentanone (-82.5%), MIK (-50%), CYC (-58%) and isophorone (-73.9%). The different increases in
g(
r) values indicated that the interactions of IPDA(NHCOO
−)
2 with various types of ketone-based molecules were differentially enhanced, with MIK showing the least enhancement and 3-pentanone the most enhancement after the system reacted with CO
2. This result reflects the assembly affinity between IPDA(NHCOO
−)
2 molecules, which means that clusters of IPDA molecules can easily form to crystallize from the MIK solvent. The average number of H-bonds determined from MD simulations also an important indicator for evaluating interactions in the system. For the absorbent systems before and after reaction with CO
2, the average number of hydrogen bonds between IPDA(NHCOO
−)
2 molecules (-270.85) was significantly greater than that between IPDA molecules (-206.65), indicating the formation of large-scale H-bond networks between IPDA(NHCOO
−)
2 molecules. Moreover, the average number of hydrogen bonds decreased to different degrees before and after CO
2 absorption for all six PCA systems, as follows: acetone (-13.92 to -9.69); 2-butanone (-11.65 to -8.77); 3-pentanone (-10.58 to -8.42); MIK (-10.38 to -7.50); CYC (-13.04 to -8.88); and isophorone (-9.73 to -9.27). The IPDA-CYC system had the greatest decrease (-4.16), and the average number of hydrogen bonds was the lowest (-7.50) for the IPDA-MIK system, indicating that IPDA(COO
−)
2 was more likely to precipitate and crystallize from the IPDA-MIK PCA system after absorbing CO
2, which was consistent with the results of the desorption experiments. Furthermore, Fig. S7 in Appendix A shows the statistics for the change in the number of transient hydrogen bonds from 0–400 ps in the MD model before and after the absorption of CO
2 by IPDA in the six different PCA systems. The number of transient hydrogen bonds corresponds to the fluctuation within a certain interval around the average number of hydrogen bonds shown in
Fig. 6(d), indicating that the MD computational model is robust with respect to the statistics for the number of hydrogen bonds formed in the PCA system. These results are in agreement with the high probability of contact of these two groups indicated by the ESP and the Fukui index results from the quantum chemistry calculations.
On the basis of the results for the six ketone-based PCA systems described above, the dipole moments of each solvent molecule and IPDA before and after CO
2 absorption were calculated, as shown in
Fig. 6(e). The dipole moment reflects the state of molecular polarity [
57]. The strength of molecular polarity corresponds to the numerical magnitude of the dipole moment. Before the carbon dioxide absorption reaction, the dipole moment of IPDA was 2.47 D, and the dipole moments of the solvent molecules in the remaining six systems were as follows: acetone (4.15 D); 2-butanone (4.01 D); 3-pentanone (3.94 D); MIK (4.23 D); CYC (4.40 D); and isophorone (6.31 D). The dipole moment of IPDA was not substantially different from those of the five ketone-based phase-change solvent molecules. Therefore, experimentally, IPDA showed high liquid–liquid compatibility with the first five solvents. Although the dipole moment of isophorone is quite different from that of IPDA, the two molecules are still compatible with each other. The dipole moment of the product formed by IPDA after CO
2 absorption was 9.15 D, which is substantially different from those of the above five phase change solvent molecules. This large polarity gap is why IPDA(NHCOO
−)
2 exhibited strong solid–liquid mutual repulsion and precipitation crystallization with the above five ketone-based phase change solvent molecules in the experimental results. After desorption, the dipole moment of IPDA(NHCOO
−)
2 decreased from 9.15 to 2.47 D for IPDA, indicating that the IPDA molecule possesses strong reversible polarity. Therefore, the PCA system can feasibly become two immiscible phases, solid and liquid, after the CO
2 absorption reaction. Furthermore, the CO
2-associated substance, IPDA(NHCOO
−)
2, forms molecular clusters that are deposited at the bottom of the PCA system because of its higher density and then triggers a reverse phase change to return to the liquid state after the desorption reaction. Fig. S8(a)–(f) in Appendix A presents optical microscopy images that demonstrate that the IPDA ethyl carbamate products in the IPDA-MIK system aggregated into clusters/agglomerates of various sizes with the injection of CO
2, suggesting that the solid–liquid phase transition was realized through a large-scale hydrogen bonding mechanism [
58,
59]. Fig. S8(g) shows that the viscosity increased over time after the carbon dioxide reaction, reaching a maximum value of 9.06 mPa·s at 29 min, then fluctuating between viscosities of 8.7 and 9.15 mPa·s, and after complete desorption, the viscosity returns to a value similar with that before the reaction.
The various species interaction mechanisms were further explored in depth by analyzing the single-molecule NCIs between IPDA and each PCA through quantum chemical calculations. A three-dimensional visualization of the results is presented in the form of an RDG scatter plot [
54]. The
λ2 parameter, as the second eigenvalue of the electron density Hessian matrix, is able to distinguish between attractive and repulsive noncovalent interactions by characterizing the curvature properties of the electron density distribution and enables the spatial localization of the interaction region and the semiquantitative assessment of the strength based on the sign(
λ2)
ρ function.
Fig. 7 and Fig. S9 in Appendix A depict the RDG scatter plots, for the six PCA single-molecule systems before and after the CO
2 absorption reaction. As shown by the isopotential surface diagrams (Figs. S9, left side) and RDG scatter plots (Figs. S9, right side) before the CO
2 absorption reaction, the NCI of the PCA single-molecule systems is driven by van der Waals forces (green region) as the primary interaction force, while other forces, such as dispersion and orientation forces, also play a driving role. The repulsive force (red region) is the secondary interaction force, which is caused by ring tension due to π–π interactions and steric hindrance effects in molecules with cyclic structures, such as IPDA, CYC, and isophorone, and the high electronegativity of the carbonyl group in the ketone-based molecules. The third interaction force is the strong attractive force (blue region), which is related mainly to hydrogen bonding. According to the isopotential surfaces (
Figs. 7, left side) and RDG scatter plots (
Figs. 7(a) – (f), right side) for the CO
2 absorption reaction, the NCIs of the PCA single-molecule systems still involve van der Waals forces (green area) as the main interaction force. The green area of the isopotential surface is increased compared with that prior to the reaction, and the green area of the RDG scatter plot is obviously increased, indicating that the van der Waals force and other interaction forces, such as the dispersion force and orientation force, all increased after CO
2 absorption. The repulsive force (red area) is still a minor interaction force, which is caused by the ring tension due to π–π interactions and steric hindrance effects in the molecules with cyclic structures, such as IPDA, CYC, and isophorone, and the high electronegativity of the carbonyl group in the ketone-based molecules. In addition, the carbonyl and deprotonated hydroxyl groups formed in the carbamates after the CO
2 absorption reaction are a source of repulsive forces (red areas). Once again, the interaction force is a strong attractive force (blue region), which is related mainly to hydrogen bonding, and the area of the blue region is reduced after the CO
2 absorption reaction except in the case of IPDA-isophorone, which indicates weakened hydrogen bonding between the single-molecule systems of the PCAs. Notably, the largest changes in van der Waals forces, repulsive forces, and strong forces were observed in the IPDA-MIK single-molecule system, which matched the experimental results [
60].
During the solid–liquid phase change of the solid–liquid absorbent, a number of uniformly distributed white flocculated particles appeared in the system during the CO
2 absorption reaction. As shown in
Fig. 8(a), as the CO
2 absorption reaction time increased, more white flocculated particles gradually appeared, filling the entire PCA system during the CO
2 absorption reaction. In the absence of external stirring and dispersion, continuous CO
2 absorption caused the clusters to descend at a very slow rate; after a certain time, a clear solid–liquid interface formed, with an upper layer consisting of a colorless and transparent solution and a lower layer consisting of a white solid. On the basis of the solid–liquid partitioning phenomenon, it can be inferred that the immiscible components with different polarities are uniformly partitioned by intermolecular noncovalent forces and that the solid products precipitate and crystallize under the force of a large-scale hydrogen bonding network. The carbon dioxide product-rich IPDA(NHCOO
−)
2 forms molecular clusters through the large-scale hydrogen bonding network, and as the amount of IPDA(NHCOO
−)
2 continues to increase, the molecular clusters gradually form solid precipitates and crystallize with increasing volume (v/v) and concentration (fraction), whereas the PCA solvent molecules move in the opposite direction, as shown in Stage I to Stage III. For larger molecular clusters, gravity (
Fg) and intermolecular noncovalent forces (
Fm) can be considered the main driving forces for solid–liquid phase separation, while buoyancy (
Fb) and viscous forces (
Fv) oppose the driving forces. In addition, the descending motion of large molecular clusters can be considered a discrete stage of the solid–liquid phase separation process, resulting in multiphase flow, which is inevitably affected by drag force (
FD), especially at relatively high descending velocities. The close relationship between drag and reactor scale is attributed to the effect of reactor scale on multiphase flow, which explains the difference in time between solid–liquid phase separation processes for different-sized separators. In contrast, the experimental results revealed that the effect of drag on multiphase flow was relatively weak [
60].
From the energy consumption point of view, the product IPDA(NHCOO
−)
2 formed after the absorption of CO
2 by the IPDA single system triggers the formation of a large-scale hydrogen bonding network, as shown in
Fig. 8(b), which accumulates into molecular clusters and forms a large solid-phase product, leading to a solid–liquid phase transition, but the excessive number of noncovalent strong attractive-hydrogen bonds leads to an inverse phase transition at a lower thermal regeneration temperature (333 K). After the addition of a phase change medium, IPDA can be uniformly dispersed in all IPDA PCA systems in the appropriate proportion, forming a small-scale hydrogen bonding network within a certain range; this process leads to the formation of small-sized solid products, which ensures that the inverse phase change process occurs at a lower thermal regeneration temperature (333 K) and effectively reduces the regeneration energy consumption of the system. To effectively realize the inverse phase change process at a lower thermal regeneration temperature (333 K), it is also necessary to balance the average number of hydrogen bonds, molecular polarization potential and other MD parameters. When selecting a PCA for the IPDA PCA system, carbonyl compounds with a high degree of electronegativity, repulsive NCIs, low polarity, and no characteristic ketone ring structure are the appropriate choice.
3.5. TEA of ketone-based IPDA PCAs
Although the total energy consumption of ketone-based media, specifically IPDA-based PCAs for CO
2 capture, is significantly lower than that of the second-generation MEA process commonly employed in industrial applications, careful consideration of costs and environmental impacts is essential for industrial scale-up. As shown in Table S29 in Appendix A, the TEA of the IPDA-MIK system is estimated to be approximately 72.05 CNY per tonne CO
2 (CNY·t
−1CO
2), which is substantially lower than the TEA of the MEA chemical absorption system (233 CNY·t
−1CO
2) that is widely utilized in industry [
61,
62]. This difference underscores the considerable potential of the ketone-based dielectric PCA system for industrial carbon capture applications. Furthermore, as illustrated in the TEA Sankey diagram and major subcost (
Figs. 9(a) and
(b)), raw material consumption constitutes the largest portion of the total cost, accounting for approximately 89.38%, followed by costs associated with heat pump operation, which represent approximately 7.36%. It is anticipated that developments in technological productivity and equipment efficiency will significantly reduce this portion of the overall cost.
3.6. Limitations and motivation for future work
Solid–liquid PCA systems for CO2 capture exhibit significant potential for enhancing efficiency. However, several challenges impede their scalability and commercial viability. A primary limitation is the substantial energy requirement associated with solid–liquid separation, particularly in systems characterized by a high solid content or elevated solvent viscosity. This energy demand not only escalates operational costs but also complicates the overall efficiency of the system, particularly as the system is scaled up. Furthermore, the risk of clogging and associated maintenance issues constitute an additional limitation. Solid CO2-bound particles can obstruct fluid flow, leading to system downtime and increased maintenance expenses due to equipment wear and tear. The abrasiveness of these solids further deteriorates critical components, thereby hindering continuous operation. Additionally, the properties of the solvent introduce further design complexities; while high-viscosity solvents are effective for CO2 capture, they impede solid–liquid separation and necessitate greater pumping energy. Over time, solvent degradation, especially in the presence of solids, diminishes long-term efficiency and heightens the necessity for regeneration or replacement, ultimately contributing to increased costs.
The transition from pilot-scale to industrial-scale systems presents significant challenges, as larger systems tend to have greater issues related to inefficiencies in separation and the risk of clogging. These challenges necessitate the use of more robust equipment, which in turn increases both capital and operational costs. To address these limitations, future research should prioritize the development of advanced solvents with optimized viscosity and stability to reduce energy consumption and facilitate solid–liquid separation. Additionally, enhancing the design of three-phase contactors and investigating alternative solid–liquid separation technologies, such as magnetic separation or filtration methods, may further improve efficiency and decrease energy requirements. The integration of heat recovery systems and the optimization of process flows could also reduce energy costs and contribute to streamlined operations. Finally, conducting research focused on scalability and economic modeling is crucial for establishing the commercial viability of this technology at an industrial scale. Concentrating on these areas can significantly enhance the sustainability and cost-effectiveness of CO2 solid–liquid phase change systems, thereby increasing their feasibility for large-scale application.
4. Conclusions
In this study, we developed a novel CO2 capture system based on phase separation between a ketone–liquid amine (IPDA)-based PCA and the solid carbamic acid IPDA(NHCOO−)2, which is formed during CO2 absorption. This system demonstrated effective CO2 capture across a wide range of concentrations, from 400 ppm in ambient air to 150 000 ppm in coal-fired industrial emissions. At an initial concentration of 400 ppm, the system achieves a CO2 capture efficiency of over 95% for more than 13 h, with a molar ratio of IPDA to CO2 of 1.11 mol·mol−1, indicating its potential for DAC. The IPDA-ketone PCA system begins CO2 desorption at ≥ 303 K and fully releases CO2 under N2 flow at 333 K, enabling at least five cycles of CO2 capture and release without degradation. During desorption, the partially dissolved IPDA-derived carbamic acid effectively releases CO2 under low-temperature conditions (303-333 K). Quantum chemical calculations and MD simulations revealed NCIs between IPDA and ketone-based solvent molecules, which modulate H bonding within IPDA(NHCOO−)2, allowing for a low-temperature (333 K, 60 °C) regeneration and crystallization process that differs fundamentally from that in aqueous systems. On the basis of the TEA, the estimated cost of CO2 capture for this system is approximately 72.05 CNY·t−1CO2, which is significantly lower than that of conventional carbon capture (233 CNY·t−1CO2), underscoring its industrial potential. Collectively, these results present valuable insights for selecting novel CO2 PCAs and advancing DAC applications with reduced energy demand.
CRediT authorship contribution statement
Qingrui Zeng: Writing – review & editing, Writing – original draft, Visualization, Validation, Supervision, Software, Methodology, Investigation, Data curation. Ziang Jia: Writing – review & editing, Data curation. Yingyang Song: Project administration, Investigation, Data curation. Yiwen Fan: Visualization, Supervision, Methodology, Investigation. Xu Liu: Supervision, Software, Resources, Methodology, Investigation, Formal analysis. Jinping Cheng: Visualization, Validation, Supervision, Methodology, Funding acquisition, Formal analysis, Conceptualization.
Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgment
This work was supported by Key Research and Development Projects of Shanghai Science and Technology Commission (20dz1204004), Shanghai Science and Technology Innovation Action Plan (22dz1208800), the Key research and development projects of Shanghai Municipal Bureau of Ecology and Environment (202306).
PDF
