In this study, the chemicals and their information are shown in the Supplementary Material. All chemicals were used without further purification. MEA was used for validation of experimental methods. Deionized water was used to prepare the aqueous solutions, and the amine solutions in this study are all liquid at room temperature.

In this study, a customized gas–liquid double-circulation device made of glass was adopted to determine the VLE data of the HEPZ aqueous solution. The working principle is gas–liquid dual circulation. The two circulating pumps are used to forcefully circulate the gas and liquid parts at the same time so that the gas and liquid phases are fully contacted and the time to balance is shortened. The equilibrium temperature can be accurately measured, and partial condensation is very small. The construction of the gas–liquid double-circulation device is shown in Figure 2.

The device is made of glass. The solution is pumped into the boiling chamber from the reagent inlet at the bottom of the equipment and heated to boiling using the heating rod. The outer-side air of the boiling chamber is evacuated using a vacuum pump to vacuum for heat preservation. The boiling solution is flushed to the liquid sampling port by bubbles, and the thermometer put in the temperature-measuring cell candetermine the temperature of the boiling liquid phase in order to obtain the boiling point, which is stable after reaching boiling. The vapor enters the condensing tube and flows back to the gas sampling port after condensing; we closed the cock to hold the condensate, which is a sample of the gas phase. After opening the cock, the condensate and reagent are stirred using a magnet and continued to be heated to boil. The entire equipment is connected to an external pressure control device, which is brought to a certain negative pressure using a vacuum pump, achieving the control of the equipment pressure. Therefore, the equipment can obtain VLE data at different pressures. After the solution is boiled, the liquid and vapor phase samples are collected, and the concentration of each phase solution is measured using an Abbe refractometer. This equipment used in this study can measure the boiling point of a certain concentration of the solution and the vapor–liquid composition after boiling at a certain negative pressure or atmospheric pressure. The construction and the operation principle of the Abbe refractometer are provided in the Supplementary Material.

The main sources of error of the device are as follows: the sampling time after boiling is difficult to determine, and the accuracy of the pressure control device is limited; the fluctuation is about ± 0.2 kPa. The uncertainty of the thermometer used in this study is ±0.2°C. The color of some solutions turns yellow at high temperature. There will be a certain error in the analysis of the refractometer.

According to Dong et al. (2010), CO₂ solubility has been measured in the stainless-steel reactor. The apparatus contains four 400 cm³ stainless-steel equilibrium cells (they have an identical structure with the one shown in Dong’s research) that were designed to operate at a temperature as high as 130°C and a pressure as high as 1 MPa, a 500 cm³ stainless-steel gas container with a temperature transducer and a pressure transducer on the top, and CO₂ and N 2 tanks.

The key part of the device is the gas–liquid balance reactor, and the temperature in the gas and liquid phases was controlled using a heating jacket and then determined by two temperature transducers, whose accuracy was 0.1 K. A stirring paddle equipped in the kettle is driven by the external motor to generate magnetic force and drive the solution inside of the sealed kettle. The overall pressure was determined by a pressure transducer (JYB-KO-HAA, Kunlunhaian Co.), whose accuracy was about 0.5%. Meanwhile, the same temperature and pressure transducers were adopted by the CO₂ gas container, which is used to obtain the total amount of CO₂ that was introduced into the reactor by recording the pressure difference before and after each injection. Due to the simple structure of the gas chamber, the actual volume can be directly calibrated by the drainage method. However, the internal structure of the reactor is complicated and cannot be directly measured by the drainage method. Its actual volume is calibrated by the gas pressure difference method after measuring the volume of the CO₂ gas chamber. At the beginning of each experiment, the reactor cell is washed by the remaining air through N₂, and then, the aqueous amine solution with a volume of 100 cm³ is injected into the cell. Meanwhile, the temperature of the reactor was set at the experiment temperature. Later, CO₂ was injected to the cell, and 10 h was provided for the absorption equilibrium after every injection.

The partial pressure of CO₂ was obtained by determining the increase in the total pressure compared to the initial value after an injection of CO₂, and it was assumed that the partial pressure of N₂ and H₂O was constant in each experiment. Using the Peng–Robinson (PR) cubic equation, the exact quantity at the gas phase was judged by three factors, including volume, pressure, and temperature (Peng and Robinson, 1976). Then, the dissolved CO₂ concentration was expressed by CO₂ loading with mole CO₂/mole amine. The loading uncertainty was 8%, which was determined by the uncertainty of pressure, temperature, and volume, which were 0.5%, 0.1, and 0.5%, respectively. The uncertainty of CO₂ partial pressure was estimated as 2%, and the details are shown in the study by Dong et al. (2010).

MEA/H₂O/CO₂, a well-known and widely studied system, was selected to verify the experimental methods. The CO₂ solubility for the MEA solution has been measured in a vapor–liquid equalizer, as described in the study by Dong et al. (2010). In Dong et al.’s study, only the data at 313 K were verified. In order to verify the accuracy of experimental equipment in a wider temperature range, the same equipment was used to measure the CO₂ solubility for 30 wt% MEA at 313.15 and 393.15 K. The experimental results are highly correlated with Lee et al. (1976)’s data and shown in Supplementary Material, but the CO₂ pressure at a temperature of 313.15 K is slightly higher than the data from the literature, which is in line with the results of Dong et al. (2010).

To validate the gas–liquid double-circulation device for measurements, vapor–liquid equilibrium data for ethanol were measured at 293.15 K and 101.3 kPa. Then, these data were compared to those obtained from the literature. Data from this study agreed with the data from the study by Kurihara et al. (1993), as shown in Supplementary Material, validating the experimental method. The measurement accuracy of the Abbe refractometer is ± 0.0001    nD.

In a vapor–liquid equilibrium system, the activity of the components during the liquid phase is the same as the fugacity during the gas phase. For the components of amines and water, which use pure substances as the reference state, the formula of equilibrium can be expressed as y i ∅ i P = x i γ i P i s . (1)

For the component CO₂ using the infinite dilution state as the reference state, which is amine and CO₂, the equilibrium formula is as follows: y i ∅ i P = x i γ i ∗ H , (2) where y i represents the mole fraction of component i at the gas phase, ∅ i represents the fugacity coefficient of component i at the gas stage, P represents the total pressure at the system temperature, P i s represents the vapor pressure of component i at the system temperature, x i represents the mole fraction of component i at the liquid stage, γ i ∗ represents the asymmetric activity coefficient of i in water, and H i represents Henry’s constant of i in water at the system temperature and vapor pressure of water.

The dependence of the Henry constant on temperature can be expressed as ln ⁡ H i = A i + B i T ( K ) + C i ⁡ ln ⁡ T ( K ) + D i T ( K ) . (3)

The Redlich–Kwong (RK) equation of the state model was used to describe the gas phase. The Henry constant of CO₂ in water was taken from the study by Chen et al. (1979).

Chemical absorption reaction equations in the loaded HEPZ solution are as follows: 2 H 2 O ↔ H 3 O + + O H − , (4) C O 2 ( a q ) + 2 H 2 O ↔ H 3 O + + H C O 3 − , (5) H C O 3 − + H 2 O ↔ H 3 O + + C O 3 2 − , (6) H E P Z H + + H 2 O ↔ H E P Z + H 3 O + , (7) H E P Z H 2 2 + + H 2 O ↔ H E P Z H + + H 3 O + , (8) H E P Z C O O − + H 2 O ↔ H E P Z + H C O 3 − , (9) H H E P Z C O O + H 2 O ↔ H E P Z C O O − + H 3 O + . (10)

All the reaction equilibrium constants of the reactions mentioned above can be obtained from the standard-state Gibbs free energies of the equations’ chemicals. The calculation equation is as follows: K j = exp ( − Δ G j 0 R T ) , (11) where K j represents the chemical equilibrium constant of reaction j , R represents the universal gas constant, T represents the temperature, and Δ G j 0 represents the reference state Gibbs energy change of reaction j . Knowing the standard Gibbs free energy, the standard enthalpy of each component’s formation, and the heat capacity in the reference state in the reaction equilibrium equation, the equilibrium constant of each reaction can be calculated. For reactions four to six, the equilibrium constants from the previous studies were usually consistent with the concentrations based on the molality. However, in this study, the model is on the basis of the mole fraction. Therefore, the equilibrium constants were converted by the method mentioned in the study by Li et al. (2014a).

Protonation reactions 7 and 8 produced new ions H E P Z H + and H E P Z H 2 2 + in solution, which are absent in the ion database of Aspen. Therefore, the standard-state thermodynamic properties of H E P Z H + and H E P Z H 2 2 + are lacked in Aspen and need to be manually adjusted to fit the p K a values from the study by Hamborg and Versteeg (2009). For H E P Z C O O − and H E P Z C O O H , the standard-state thermodynamic properties are regressed from C O 2 solubility data obtained in this study. Using the thermodynamic properties of the products and the reactants, the equilibrium constants of all reactions can be obtained.

Referring to 3.2, a series of reactions will take place in the loaded HEPZ solution, and multiple ions were produced in the process. The interaction between ions causes the liquid system to gradually deviate from the ideal state. It needs to introduce a coefficient model with accurate activity for calculations and simulation. In this study, the activity coefficients for binary interactions in the unloaded amine solution were calculated by the NRTL model, but those for the molecule–molecule binary, molecule–ion pair binary, and ion pair–ion pair binary in the loaded amine solution were calculated by the ENRTL model. The dependence of binary parameters on the temperature can be expressed as τ i j = a i j + b i j T , (12) where i j can be molecule–molecule, molecule–ion pair, ion–molecule pair, or ion pair–ion pair. Meanwhile, the binary parameter a i j is 0.3 for the molecule–molecule interaction, 0.2 for the molecular–ion pair interaction, and 0 for the ion pair–ion pair interaction, and the default values for both ion pair–ion pair and molecule–molecule binary parameters are 0. The default value of the molecule–ion pair binary parameters a i j is (8, −4) when the molecule is H₂O; otherwise, the default values are set at 8 and −15. The default values for b i j are 0.

The heat capacity data of HEPZ obtained from the study by Poozesh et al. (2013), Tagiuri (2019) were regressed to acquire the ideal gas heat capacity constants ( C p i g ), and Antoine’s constants were from the database of Aspen; the value and standard deviation are listed in Table 2. Most of the parameters’ standard deviations shown in Table 1 are far smaller than those contained therein. The heat capacity calculated by the model was in good agreement with the value from experimental data, and the average relative deviation was 0.09% for the heat capacity.

The dependence of Antoine’s constants on the temperature is expressed as A +   B T + C ⁡ ln ⁡ T + D T 2 , but for C p i g , it is A + B T +   C T 2   + D T 3 . Numbers 1–4 correspond to the letters A–D, respectively.

The data from experiments and the literature were regressed in this model, including excess enthalpy, the VLE data acquired in this study, and heat capacity (Poozesh et al., 2013; Tagiuri, 2019) for the mixed solution of HEPZ/H₂O. The p K a values were obtained by regression through manual adjustment of the standard-state properties of H E P Z H + and H E P Z H 2 + . The regressed parameters along with their standard deviations are shown in Table 3.

In this study, VLE data at 30 kPa, 40 kPa, 55 kPa, 70 kPa, and 85 kPa and the atmosphere pressure of 101 kPa were measured. Because the viscosity of the high-concentration solution of HEPZ is too large, it is not convenient for experimental measurements. The mass fraction range of HEPZ in this experiment is 0.1–0.75. Figure 3 shows the data obtained by the experimental measurement and the data calculated by the model. The upper line and points represent for the mole fraction of HEPZ at the gas phase ( y H E P Z ), and the lower ones stand for the mole fraction of HEPZ at the liquid phase ( x H E P Z ). Because the solution will turn yellow at high temperature, there will be certain errors in the analysis with a refractometer. Besides, the boiling point of HEPZ is much higher than that of H₂O, and the concentration of HEPZ in the gas phase distilled from the HEPZ solution in this study is very low, resulting in a larger measurement error. The average relative deviation of the VLE data x H E P Z   / y H E P Z of HEPZ at negative pressure is 1.66%/70.9%, 1.26%/57.7%, 0.294%/46.7%, 0.0519%/8.17%, and 0.529%/153%, respectively. The average relative deviation of the VLE data x H E P Z   / y H E P Z of HEPZ at atmosphere pressure is 0.760%/57.2%.

HEPZ has two dissociation constants pKa₁ and pKa₂, which are regressed by the manual adjustment of the standard-state properties of H E P Z H + and H E P Z H 2 + , with the standard formation Gibbs Free Energy at 298.15 K Δ f G 298.15 K ∞ , a q , the standard enthalpy of formation at 298.15 K Δ f H 298.15 K ∞ , a q , and the infinite dilution state heat capacity parameter at 298.15 K C p ∞ , a q . The non-randomness factor α was a fixed value (0.3), and the results of these properties are listed in Table 3. The average value of the relative deviation between the pKa₁ and pKa₂ data from the study by Hamborg and Versteeg (2009) and Aspen is 0.01%.

Poozesh et al. (2015) and Tagiuri (2019) measured the heat capacity values for HEPZ/H₂O at various temperatures ranging from 303 to 353 K, and the mole fraction of HEPZ ranges from 0.1 to 0.9. In Figure 4, the calculated value of the model is compared with the measured value of the literature. In the high-concentration area, Figure 5 shows that there is a good correlation between the value calculated by Aspen and the experiment values. However, in the area of low concentration, the experiment values increase as T increases, and the calculated values do not show the same trend. The average relative deviation of all points is 0.0235%. The excess enthalpy of HEPZ/H₂O calculated by Aspen is shown in Figure 5, and the average value of the relative deviation for excess enthalpy regression was 8.57%.

CO₂ solubility data were measured by the method in 2.3, and then, they were regressed to acquire the ENRTL parameters as well as the standard-state properties Δ f G 298.15 K ∞ , a q , Δ f H 298.15 K ∞ , a q , and C p ∞ , a q for H E P Z C O O − and Δ f G 298.15 K i g , Δ f H 298.15 K i g , and C p i g for H E P Z C O O H . The non-randomness factor α was a fixed value (0.2). The results of the regression are shown in Table 4. The experimental data as well as the calculation results obtained by the model are shown in Figure 6. The experimental data were calculated by the model, and the mean relative deviation of the regression for 5 wt% HEPZ in Figure 6A is 13.35%, and it is 9.79% for 15 wt% HEPZ in Figure 6B and 14.98% for 30 wt% HEPZ in Figure 6C.

Circulating capacity is an important property to characterize the properties of amines, and there are two ways to calculate cyclic capacity. When considering a CO₂ removal rate of 90% in the absorber, one way (way 1) is defining lean loading as the CO₂ loading when the partial pressure of CO₂ is 1 kPa at a temperature of 313.15 K, and the rich loading of CO₂ partial pressure is 10 kPa. Meanwhile, cyclic capacity represents the difference between the rich and lean loadings with the unit of g of CO₂/kg of the solvent.

However, in the actual operation of the absorption tower, the absorption tower is not at a constant temperature, and at the bottom of the tower, the rich loading is determined by the equilibrium partial pressure of CO₂ in the flue gas as well as the temperature of the liquid. Also, the lean loading is defined by the desorption tower and not by the equilibrium partial pressure of CO₂ in the top gas of the absorption tower. The other way (way 2) to calculate cyclic capacity is defining lean loading as the C O 2 loading when the partial pressure of C O 2 is 15  k P a at 393.15  K and rich loading as 15    k P a of C O 2 partial pressure at 313.15 K . All results are shown in Table 5.

Figure 7 shows the speciation data for 5 wt%, 15 wt%, and 30 wt% HEPZ solutions at a temperature of 313.15 K forecast by the model. For 5 wt% and 15 wt% HEPZ solutions, when the loading is 0–0.5, most of the CO₂ absorbed is converted into C O 3 2 − and H E P Z C O O − , and the other is converted into H C O 3 − ; most HEPZ is converted into H E P Z H + . At the loading of 0.5–1, the main reactants are CO₂, HEPZ, and H E P Z H + , and some H E P Z C O O − and C O 3 2 − are also consumed to generate H C O 3 − and H H E P Z C O O . For the 30 wt% HEPZ solutions, when the loading is lower than 0.3, HEPZ is consumed and converted to H E P Z H + , H E P Z C O O − , C O 3 2 − , and H H E P Z C O O , and the important products are H E P Z H + and H E P Z C O O − . At a loading of 0.3–0.7, H E P Z C O O − becomes a reactant which is converted to H H E P Z C O O . At a greater loading, the proportion of C O 3 2 − and H E P Z H + continues to increase, and the proportion of H E P Z and H E P Z C O O − decreases, showing the CO₂ absorbed mainly converted to C O 3 2 − . It is because the solution is more alkaline in this loading range, which is consistent with the theoretical analysis. As the concentration of HEPZ increases, the H C O 3 − produced by the reaction gradually decreases.

With the activity coefficient as well as mole fraction of species in the 30 wt% HEPZ solutions at different temperatures calculated by the Aspen model, the equilibrium constants of the reactions in the system can be obtained as a function of temperature as follows: K x j = ∏ i ( x i γ i ) v i j , (13) where x represents the K x j on the basis of the mole fraction, j represents reaction numbers 1–7, γ i is the activity coefficient of component i , x i represents the mole fraction of component i , and v i j represents the stoichiometric coefficient of component i in reaction j . The parameters of the ln ⁡ K x j − T curve calculated by Aspen through regression are shown in Table 6.

The overall differential heat of reaction Δ H r ,   o v e r a l l is the sum of the heat contributions Δ H r ,   j obtained from each of the seven reactions occurring in the solution, which are mentioned in 3.1 , and in this study, there is one more reaction that was considered to denote the dissolution equilibrium of C O 2 from the gas phase to the liquid phase: C O 2 ( g a s ) ↔ C O 2 ( a q ) . (14) Heat of Absorption Meanwhile, the heat of CO₂ physical dissolution Δ H d i s s o l u t i o n also contributes to Δ H r ,   o v e r a l l , given by Δ H r ,   o v e r a l l = ∑ j = 1 7 Δ ξ j Δ H r ,   j + Δ H d i s s o l u t i o n , (15) where Δ ξ j represents the reaction degree of the key component of reaction j when absorbing 1 mole CO₂, which can be calculated by the incremental change of the key component when 1 mole CO₂ is absorbed. Δ ξ 1 = Δ n O H − Δ n C O 2 ∗ , (16) Δ ξ 2 = 1 − Δ n C O 2 Δ n C O 2 ∗ , (17) Δ ξ 3 = Δ n C O 3 2 − Δ n C O 2 ∗ , (18) Δ ξ 4 = − Δ n H E P Z H + + Δ n H E P Z H 2 + Δ n C O 2 ∗ , (19) Δ ξ 5 = − Δ n H E P Z H 2 + Δ n C O 2 ∗ , (20) Δ ξ 6 = − Δ n H E P Z C O O − + Δ n H H E P Z C O O Δ n C O 2 ∗ , (21) Δ ξ 7 = − Δ n H H E P Z C O O Δ n C O 2 ∗ , (22) where Δ n i represents the changes of the number of moles of component i and Δ n C O 2 ∗ represents the total moles of CO₂ that were absorbed to improve the content of CO₂ from one loading to a higher one. The Van’t Hoff equation can be adopted to calculate the heat contributions Δ H r ,   j from each of the seven reactions, given as follows: Δ H r ,   j = R T 2 ( ∂ ln ⁡ K x j ∂ T ) p = R ( − C j + D j T + B j T 2 ) , (23) where the equation parameters of ln ⁡ K x j are listed in Table 6, and the equation for Δ H d i s s o l u t i o n is Δ H d i s s o l u t i o n = R ( 8477.71 − 21.9574 T + 0.00578075 T 2 ) . (24)

As this study did not take into account the species interaction, only the heat of physical dissolution and the chemical reaction heat were considered, but the excess enthalpy was not. Δ H r ,   o v e r a l l is calculated for analyzing the contribution of reactions to the heat of absorption, and the values of Δ H r ,   j at a temperature of 313.15 K are 52.53, 3.63, 10.58, 27.49, 21.71, −47.20, and 48.9 0   k J / m o l , and j   equals 1–7. The physical heat of CO₂ dissolution Δ H d i s s o l u t i o n is −18  k J / m o l at a temperature of 313.15 K. The total differential chemical reaction heat as well as the heat released by the seven reactions and the heat from CO₂ dissolution in 30 wt% HEPZ solution are shown in Figure 8. This part aims to carry out a discussion on the contribution of reactions to the total absorption heat, and Eq 17 only calculates the heat of physical dissolution, the chemical reaction heat, and the excess enthalpy, so the results are lower than the actual heat of absorption.