Optimization of CO₂ Capture Using a New Aqueous Hybrid Solvent (MDEA-[TBPA][TFA]) with a Low Heat Capacity: Integration of COSMO-RS and RSM Approaches

Abstract

This study aims to evaluate the performance of a new hybrid solvent, comprising aqueous MDEA and tetrabutylphosphonium trifluoroacetate ([TBP][TFA]), for CO₂ capture and to optimize its CO₂ absorption efficiency. First, this study focused on predicting the thermodynamic properties of aqueous MDEAs and [TBP][TFA] and their interaction energy with CO₂ using COSMO-RS. Based on the prediction, it aligns with the principle that CO₂ solubility in the MDEA-[TBP][TFA] hybrid solvent decreases as the Henry’s Law constant increases, with the interactions primarily governed by van der Waals forces and hydrogen bonding. The aqueous MDEA-[TBP][TFA] hybrid solvent was prepared in two steps: synthesizing and blending [TBP][TFA] with aqueous MDEAs. The formation and purity of [TBP][TFA] were confirmed through NMR, FT-IR, and Karl Fischer. The heat capacity of the hybrid solvents was lower than their aqueous MDEA solutions. The performance and optimization of CO₂ capture were studied using RSM-FC-CCD design, with the optimal value obtained at 50 wt.% MDEA, 20 wt.% [TBP][TFA], 30 °C, and 30 bar (12.14 mol/kg), aligning with COSMO-RS predictions. A 26% reduction in the heat capacity was achieved with the optimal ratio (wt.%) of the hybrid solvent. These findings suggest that the aqueous MDEA-[TBP][TFA] hybrid solvent is a promising alternative for CO₂ capture, providing a high removal capacity and lower heat capacity for more efficient regeneration compared to commercial aqueous MDEA solutions.

1. Introduction

Natural gas is a hydrocarbon mixture primarily composed of light paraffins, such as methane and ethane, along with non-hydrocarbon acidic gases, like carbon dioxide (CO₂) and hydrogen sulfide (H₂S). It is considered a crucial resource due to its lower CO₂ emissions compared to coal and oil [1,2]. However, impurities, particularly CO₂ in natural gas, must be removed before it can be transported, used for heating, or processed into electricity [3]. This is necessary to enhance its quality and comply with environmental regulations. One common method for removing CO₂ from natural gas is CO₂ scrubbing, which involves passing the gas through a scrubber containing a liquid solvent that chemically reacts with and absorbs CO₂ molecules [4]. Although CO₂ scrubbing technology for removing CO₂ from natural gas has been in use for a long time, there have been recent advancements in CO₂ separation and handling methods [5]. These advancements refer to more efficient techniques or methods for separating and managing CO₂, which can reduce costs and improve the effectiveness of the process [6]. As a result, unconventional natural gas reserves (such as those with a higher CO₂ content than usual) can now be exploited more economically, making them more competitive and profitable to develop in the energy industry.

Amines are commonly used chemical solvents for removing acidic gases like CO₂ and H₂S from natural gas [7]. Common amines include monoethanolamine (MEA), diglycolamine (DGA), diethanolamine (DEA), triethanolamine (TEA), methyldiethanolamine (MDEA), 2-amino-2-methyl-1-propanol (AMP), and piperazine (PZ). MEA is widely used due to its high selectivity and absorption capacity [8,9,10,11], but it requires significant energy for regeneration [12]. DEA is less energy-intensive but has lower absorption rates and can degrade over time. MDEA is effective for removing high concentrations of acidic gases [11] but also requires high regeneration energy and has slow reaction kinetics. A major concern with using aqueous amine solutions for CO₂ removal is the high energy demand for solvent regeneration [13]. The formation of zwitterion upon the reaction between primary amines and CO₂ results in carbamate formation [14], which generates higher heat of absorption and leads to increased regeneration energy costs. Chemical absorption processes are highly dependent on heat absorption and require substantial regeneration energy, accounting for approximately 50–60% of the total regeneration cost [15]. This significant energy demand increases steam quality requirements for reboilers, thereby raising production costs [16,17]. New and improved solvents are needed to address these limitations and enhance CO₂ capture efficiency.

Ionic liquids (ILs) are emerging as promising alternatives to traditional amine solvents for CO₂ capture [18]. They offer advantages, such as thermal stability, negligible vapor pressure, and task-specific properties [19], making them more environmentally friendly and efficient. While early ILs had limitations, advancements in IL design, including the incorporation of amine moieties and superbase anions [20], have significantly improved their CO₂ absorption capacity. The choice of cation and anion plays a crucial role in determining the IL’s CO₂ absorption capacity and selectivity [21]. Phosphonium cation-based ILs offer a wide range of functional properties, allowing for the creation of tailored solvents for CO₂ capture. These advancements in IL technology hold significant potential for improving the efficiency and sustainability of CO₂ capture processes. While ILs have demonstrated a high CO₂ capacity, their viscosity can significantly increase after CO₂ absorption [22]. This increase is primarily due to the formation of strong hydrogen-bonded networks between the compounds, leading to higher cohesive energy within the solution [23]. For example, [bmim][BF₄] has a much higher viscosity (79.5 cP) compared to pure MEA (25 cP) or a 30% aqueous MEA solution (2 cP) [24]. This increased viscosity can slow down diffusion, increase the equilibrium time, and limit the rate of CO₂ absorption [25]. Studies indicate that equilibrium for ILs can take up to 48 h at 303 K [26]. Additionally, the high synthesis cost and energy requirements for recycling ILs make them unsuitable for large-scale industrial applications [27]. These drawbacks limit the economic feasibility of using ILs for CO₂ removal [26].

To address the viscosity and economic feasibility challenges of ILs, the research has explored mixing them with alkanolamines. Mixtures of alkanolamine and ionic liquids (ILs) offer significant potential for improving CO₂ capture efficiency [28,29,30]. The research has demonstrated that these mixtures can overcome limitations associated with traditional amine solvents, such as high energy consumption and viscosity. Studies have shown that mixtures of MEA and various ILs can increase CO₂ solubility [31], reduce energy consumption [31], and improve absorption rates [32]. Additionally, these mixtures tend to have lower viscosity [33], making them more practical for industrial applications. By combining the strengths of both alkanolamines and ILs, these mixtures present a promising avenue for developing more efficient and sustainable CO₂ capture technologies. Phosphonium-based ionic liquids (ILs) are promising candidates for CO₂ capture due to their lower cost, higher thermal stability, and potential for forming hybrid solvents with alkanolamines [34]. The selection of appropriate ionic liquids is crucial for identifying those that are compatible with aqueous amine solvents, aiming to achieve both a high CO₂ removal capacity and reduced heat capacity, while also lowering regeneration costs. However, conducting experiments on all phosphonium-based ILs with various anions would require considerable time and costs. Therefore, employing predictive tools, like Conductor-like Screening Model for Real Solvents (COSMO-RS), is a highly efficient and time-saving approach.

The COSMO-RS model is a robust predictive tool for screening of ionic liquids by estimating essential thermodynamic properties, including activity coefficients, solubility, Gibbs free energy, Henry’s Law constant, heat capacity, density, and viscosity [35,36]. In addition, it is applicable to solvents such as aqueous methyldiethanolamine (MDEA), thereby expanding its utility in thermodynamic predictions [37]. Utilizing the Conductor-like Screening Model for Real Solvents (COSMO-RS) allows for the efficient evaluation of potential ionic liquids for their suitability in this application. Prior to the current study, a screening of 417 tetrabutylphosphonium-based ionic liquids (ILs) was conducted using COSMO-RS, focusing on their Henry’s Law constant, heat capacity, and miscibility with water. Performance index analysis revealed that tetrabutylphosphonium trifluoroacetate ([TBP][TFA]) exhibited a low Henry’s Law constant, high water miscibility, and significantly low heat capacity, making it a promising candidate for CO₂ absorption. Additionally, [TBP][TFA] was chosen for its high CO₂ capacity and moderate cation–anion interaction. Studies indicate that phosphonium-based ILs with longer alkyl chains enhance CO₂ solubility [38]. [TBP][TFA] also provides the advantages of low heat capacity and high-water miscibility, reinforcing its potential as a candidate for CO₂ capture from natural gas. Moreover, COSMO-RS can be employed to evaluate the sigma profile, sigma potential, and interaction energy [39] between the ionic liquids and aqueous MDEA, or any other solvents, in relation to gas or liquid analytes.

The Response Surface Methodology (RSM) is a powerful statistical technique used to optimize processes by identifying the optimal levels of multiple variables [40,41]. Unlike traditional methods, like One Factor at a Time, the RSM can evaluate interactions between variables, providing a more comprehensive understanding of process behavior [42]. By fitting a polynomial equation to experimental data, the RSM generates a mathematical model that describes the process and helps identify optimal conditions for maximizing desired outcomes. RSM has been widely applied in various fields, including CO₂ capture. Researchers have used the RSM to optimize factors, such as temperature, pressure, flow rate, and partial pressure, to improve CO₂ removal efficiency using different absorbents [43]. These studies demonstrate the effectiveness of the RSM in enhancing the performance of CO₂ capture processes.

In this study, the thermodynamic properties of the aqueous methyldiethanolamine-tetrabutylphosphonium trifluoroacetate ([MDEA]-[TBP][TFA]) hybrid solvent, as well as its interaction energy with CO₂, is predicted using COSMO-RS. The performance of the [MDEA]-[TBP][TFA] hybrid solvent is evaluated in comparison to a commercial MDEA-based absorbent for CO₂ removal. The first step involves synthesizing the [TBP][TFA] ionic liquid and confirming its formation through structural and impurity analyses. Aqueous [MDEA]-[TBP][TFA] hybrid solvents are then prepared. The heat capacity of these hybrid solvents is determined using differential scanning calorimetry (DSC) to assess their heat regeneration performance. CO₂ removal using aqueous [MDEA]-[TBP][TFA] hybrid solvents is studied using a CO₂ absorption system under experimental conditions generated by the response surface methodology (RSM) based on a face-centered (FC) central composite design (CCD). RSM-FC-CCD is also used to analyze the effects of independent variables on CO₂ removal efficiency and to predict the removal efficiency based on the generated equation model. A validation study is conducted to confirm the effectiveness of the RSM-CCD model.

2. Computational and Experimental Details

2.1. Thermodynamic Properties and Interaction Energy Predictions Using COSMO-RS

In this study, COSMO-RS approaches were employed to predict the thermodynamic properties of the hybrid solvent components (i.e., aqueous MDEA and [TBP][TFA]) in relation to CO₂ and to evaluate their interaction energy with CO₂ molecules. Prior to the predictions, the optimized structures of the hybrid solvent components were sourced from the COSMOthermX database. Density functional theory (DFT) calculations were performed with the def-TZVP basis set and the B3-LYP functional. Predictions of all relevant thermodynamic properties were carried out using COSMOthermX (version 19.0.4) with BP-TZVP parameterization. In the COSMO-RS prediction process, solute molecules are first enclosed by a virtual conductor, constructed via the continuum solvation model. In this model, the solvent is treated as a continuous medium with a uniform dielectric constant, rather than individual molecules. The virtual conductor acts as an idealized environment where the solute is placed in a hypothetical conductor. The screening charge density distribution (σ-profile) is then computed for the solute using quantum chemical methods. The σ-profile describes how the virtual conductor screens the solute’s electric field, producing a specific charge distribution on the solute’s surface. This three-dimensional surface charge density distribution can be transformed into a histogram plot called the σ-profile, P(σ), as expressed by the following Equation (1) [44].

$$P\left(\mathsf{\sigma }\right)={\sum }_i{\mathcal{x}}_i{P}^{{\mathcal{x}}^i}$$

(1)

where ${\mathcal{x}}_i$ denotes as the molar fraction of the i segment, $P^{{\mathcal{x}}^i}$ refers to the σ-profile of the i segment, and i represents a specific component in the mixture. The σ-profile describes the polarity of the molecules, while the electrostatic potential derived from the σ-profile, known as the σ-potential, describes the interaction behavior and affinity between molecules in the system. The σ-potential, μ (σ), of the molecules is calculated using Equation (2) based on the calculated σ-profile using the Klamt equation [44]:

$$\mu \left(\mathsf{\sigma }\right)=-RTln\left(\int d{\mathsf{\sigma }}^{\prime }P\left(\mathsf{\sigma }\right).e^{\left(\left(-1/2\right){\alpha }^{\prime {\left(\mathsf{\sigma }+{\mathsf{\sigma }}^{\prime }\right)}^2}-\left(\left({\mu }^{\prime }\left({\mathsf{\sigma }}^{\prime }\right)\right)/RT\right)\right)}\right)$$

(2)

where α′ refers to a general interaction fitting parameter to describe the energy of the geometrically optimized structure of the molecule, and μ(σ) denotes the chemical potential.

In addition to estimating essential thermodynamic properties, COSMO-RS can also be utilized to predict the interaction energy within the studied system, specifically between aqueous MDEA and [TBP][TFA]. From the σ-profile and σ-potential, various interaction energies and hydrogen bonding properties can be predicted, including misfit interaction energy (E_MF), van der Waals interaction energy (E_vdW), hydrogen bond acceptor moment (M_HBA), and hydrogen bond energy (E_HB). The total energy (E_total) is the summation of E_HB, E_vdW, and E_MF. Mathematically, the energies are described using Equations (3)–(5) [44]:

$$E_{MF}\left(\mathsf{\sigma },{\mathsf{\sigma }}^{\prime }\right)={\alpha }_{eff}.\left({\alpha }^{\prime }/2\right){\left(\sigma +{\sigma }^{\prime }\right)}^2$$

(3)

$$E_{HB}={\alpha }_{eff}.c_{HB}min\left(0,\mathsf{\sigma }{\mathsf{\sigma }}^{\prime }+{{\mathsf{\sigma }}_{HB}}^2\right)$$

(4)

$$E_{vdW}={\alpha }_{eff}.({\tau }_{vdW}+{\tau }_{vdW}^{\prime })$$

(5)

where α_eff represents the effective contact area between two surface segments, $c_{HB}$ denotes the hydrogen bonding interaction strength coefficient, ${\sigma }_{HB}$ is the charge density threshold for hydrogen bonding, while the general fitting for the van der Waal interaction is represented by τ_vdW + τ_vdW′.

2.2. Experimental Section

2.2.1. Materials and Reagents

Tetrabutylphosphonium hydroxide solution (40 wt.% in H₂O), trifluoroacetic acid (99.9%), and N-methyldiethanolamine (MDEA, 99.5%) were acquired from Sigma Aldrich (Saint Louis, MO, USA) and used directly without further treatment. Analytical-grade dichloromethane and acetone, along with deuterated dimethyl sulfoxide (DMSO-d6) for NMR analysis, were sourced from Merck (Merck, Darmstadt, Germany). Ultrapure water (conductivity of 18.2 µS/cm) was used throughout the experiments. High-purity CO₂ (99.999%) was provided by Linde Malaysia (Petaling Jaya, Selangor, Malaysia).

2.2.2. Apparatus

The determination of CO₂ absorption was based on the isochoric saturation method [45,46] using a CO₂ absorption unit built by Dixson Engineering Sdn Bhd (Shah Alam, Selangor, Malaysia). This system includes a high-pressure equilibrium setup with an equilibrium cell and CO₂ reservoir immersed in a water bath. The water bath features a stirring unit for uniform temperature distribution and a temperature controller to maintain precise temperature conditions throughout the experiment. The system also includes a vacuum set for degassing purposes. Additionally, the system has two valves for CO₂ pressure regulation: one controls the flow from the CO₂ supply, while the other manages CO₂ entry into the reaction site within the equilibrium cell. A schematic of the experimental setup is shown in Figure 1.

CO₂ is initially stored in the reservoir, and the gas from the CO₂ supply flows through Valve A, which is regulated by a pressure gauge with a range of 0–40 bars. The CO₂ is then transferred from Valve A to Valve B, where it is held in the reservoir. The number of moles of CO₂ between Valve A and Valve B, with a known volume, referred to as initial CO₂, is calculated using Equation (6):

$$n_{{CO}_2={\displaystyle \frac{P_i{V}_i}{{Z_i}_{{CO}_2}R{T}_i}}}^i$$

(6)

where ${n^i}_{{CO}_2}$ is the initial number of moles of CO₂, $P_i$ is the initial pressure (atm), $T_i$ is the initial temperature set in Kelvin, $Z_i$ is the compressibility factor at $T_i$ and $P_i$, $V_i$ is the volume from Valve A to Valve B (V_A–_B)(L), and R is the gas constant.

During the experiment, approximately 1 g of the liquid absorbent was transferred into the equilibrium cell, which was then attached to the system, secured tightly, and sealed with white tape to prevent any ingress of water or gas. The system was degassed from Valve B to Valve C to ensure it was vacuumed before introducing CO₂ to the absorbent. Once a constant value of the desired $T_i$ and $P_i$ was achieved, CO₂ was released from Valve A to Valve B into the absorbent by opening Valve B while keeping Valve C closed. The pressure drop was monitored at 1 min intervals while maintaining a constant temperature. The system typically reached equilibrium for aqueous MDEA within 90 to 240 min, depending on the concentration. At equilibrium, the moles of CO₂ can be calculated using the following equation:

$$n_{{CO}_2}^{eq}={\displaystyle \frac{P_{eq}(V_{total}-V_s)}{Z_{{CO}_2}^{f}R{T}_{eq}}}$$

(7)

where $n^{eq}$ represents the number of moles of CO₂ remaining after the pressure reaches equilibrium, $P_{eq}$, at the equilibrium temperature, $T_{eq}$; $z^f$ is the compressibility factor at $P_{eq}$and $T_{eq}$; $V_{total}$ is the combined volume from V_A to V_C; and V_S is the volume of the sample (absorbent). The amount of CO₂ absorbed into the sample, $n_{{CO}_2}^{abs}$, can be calculated using Equation (8):

$$n_{{CO}_2}^{abs}=n_{{CO}_2}^i-n_{{CO}_2}^{eq}$$

(8)

Based on Equation (8), the removal capacity of the absorbent can be calculated by dividing the number of moles of absorbed CO₂ ($n_{{CO}_2}^{abs})$ by the mass of the absorbent, then multiplying this by the value of 100; where it is expressed in mol/kg.

2.2.3. Preparation of the Absorbent

Synthesis of the Tetrabutylphosphonium Trifluoroacetate Ionic Liquid

The tetrabutylphosphonium trifluoroacetate [TBP][TFA] ionic liquid was prepared via an acid base neutralization reaction by mixing tetrabutylphosphonium hydroxide [TBP][OH] (5.53 g, 0.02 mol) and trifluoroacetic acid (2.74 g, 0.024 mol) in an ice bath for 24 h under reflux. The resulting product was extracted with dichloromethane, and the water-soluble unreacted reagents were discarded. The structure of the IL was confirmed using proton and carbon nuclear magnetic resonance (¹H-NMR and ¹³C-NMR) spectroscopies (Bruker Avance III, 500 MHz, Bruker, Germany), along with FTIR analysis (Perkin Elmer, Frontier 01, Massachusetts, Germany) for functional group identification.

¹H-NMR of [TBP][TFA] (400 MHz, DMSO-D6, δ/ppm): δ 0.80–1.02 (t, 12 H, -P⁺-CH₂-CH₂-CH₂-CH₃), δ 1.30–1.55 (m, 16 H, -P⁺-CH₂-CH₂-CH₂-CH₃), and δ 2.13–2.27 (t, -P⁺-CH₂-CH₂-CH₂-CH₃).

¹³C-NMR of [TBP][TFA] (600 MHz, DMSO-D6, δ/ppm): δ 12.65 (s, -P⁺-CH₂-CH₂-CH₂-CH₃), δ 16.57 (s, -P⁺-CH₂-CH₂-CH₂-CH₃), δ 16.95 (s, -P⁺-CH₂-CH₂-CH₂-CH₃), δ 22.0–23.2 (q, -P⁺-CH₂-CH₂-CH₂-CH₃), δ 113.08–120.54 (q, CF₃), and δ 157.31 (s, COO⁻).

Preparation of Aqueous MDEA–[TBP][TFA] Hybrid Solvents

The preparation of aqueous MDEA-[TBP][TFA] solutions was performed using MDEA concentrations ranging from 10 to 50 wt.% and [TBP][TFA] concentrations ranging from 2 to 20 wt.%, with water as the solvent, based on the Response Surface Methodology (RSM) using a Face-Centered Central Composite Design (FC-CCD), which will be discussed in Section 2.2.4. The ratio of MDEA, [TBP][TFA], and water used in the preparation of aqueous MDEA-[TBP][TFA] solutions for subsequent CO₂ absorption is tabulated in

Table 1. Preparation of MDEA-[TBP][TFA] hybrid solvents for CO₂ absorption.

No.	MDEA (wt.%)	[TBP][TFA] (wt.%)	H2O (wt.%)
1	10	0	90
2	10	2	88
3	10	11	79
4	10	20	70
5	30	0	70
6	30	2	68
7	30	11	59
8	30	20	50
9	50	0	50
10	50	2	48
11	50	11	39
12	50	20	30

. The components were combined in a vial and thoroughly mixed to ensure the homogeneity of the hybrid solvent.

2.2.4. Design of Experiment: RSM-FC-CCD

The effect of the several independent variables, concentration of MDEA (wt.%), concentration of [TBP][TFA] (wt.%), pressure, and temperature, on the CO₂ removal capacity was studied. The removal capacity of CO₂ is an indication of the targeted response. The concentration of MDEA was set in the range of 10 to 50 wt.%, for [TBP][TFA] in the range of 2 to 20 wt.%, pressure at a range of 2–30 bar, and temperature in the range of 30–60 °C. The Face-Centered Central Composite Design (FC-CCD) Response Surface Methodology (RSM) was utilized to determine the optimal conditions of CO₂ absorption using MDEA-[TBP][TFA] hybrid solvents. It is a randomized design with less experimental trials than other methods. In this RSM-FC-CCD, each factor has three levels (low, medium, and high). Thus, the number of experiments can be calculated as 2⁴ + 2(4) + 6 = 30 (16 cubes, 8 center points in cube, and 6 axial points) following the expression 2^k + 2(k) + nc, where k and nc represent the number of factors and the number of center points in cube, respectively. The FC-CCD experimental design can be used as a tool for fitting second-order polynomial equations. The general coded model equation is represented in Equation (9):

$$y={\beta }_0+{\sum }_{i=1}^N{\beta }_i{x}_i+{\sum }_{i=1}^N{\beta }_{ii}{x_i}^2+{\sum }_{i<j}\sum {\beta }_{ij}{x}_{i}x+\epsilon$$

(9)

where $x_i$ represents the coded parameter, ${\beta }_0$ denotes the intercept, ${\beta }_i$ refers to the linear effect, ${\beta }_{ii}$ is the squared effect, the interaction between independent values is denoted by ${\beta }_{ij}$, and $\epsilon$ denotes the error. The optimum conditions were determined using the FC-CCD and based on the desirability value approach, where the desirability values closest to 1 were chosen to obtain the desired response value. The range and factors for the FC-CCD are summarized in

Table 2. Summary of the studied variables and ranges for each variable in the FC-CCD.

Coded Levels of Each Factor in FC-CCD	Factors with Actual Level
	MDEA (wt.%) (x1)	Ionic Liquid (wt.%) (x2)	Temperature (x3)	Pressure (x4)
−1	10	2	30	2
0	30	11	45	16
1	50	20	60	30

.

2.2.5. Thermophysical Characterization of MDEA-[TBP][TFA]

Density and Viscosity

The thermophysical characterization of the MDEA-[TBP][TFA] system includes measurements of density and viscosity, both critical for understanding its behavior and performance. Density (ρ) was measured using an automated SVM 3000 Anton Paar rotational Stabinger viscometer–densimeter (Anton Paar, Graz, Austria) over a temperature range of 20 to 80 °C at a pressure of 0.1 MPa. Temperature control was achieved efficiently using a Peltier element (thermoelectric cooler). The absolute uncertainty of density measurements was ±0.0005 g cm⁻³. Dynamic viscosity (η) was measured using a DHR-1 Rheometer (TA Instruments, New Castle, DE, USA), equipped with a 40 mm diameter parallel plate and 50 μm gaps, over the same temperature range of 20 to 80 °C. A small-amplitude oscillatory shear test was performed with frequencies ranging from 0.1 to 100 rad/s at a shear strain of 5%. The absolute uncertainty of viscosity was 0.0005 g cm⁻³, with a temperature uncertainty value of ±0.1 °C.

Heat Capacity Measurement

The heat capacity of MDEA-[TBP][TFA], aqueous MDEA, and the pure ionic liquid was measured using a heat flow differential scanning calorimeter (DSC) (model DSC1, Mettler Toledo, Columbus, OH, USA). Approximately 60–80 mg of each sample was weighed, placed in a 100 mg aluminum pan, and covered with a lid. The DSC measurement was baseline-corrected using a reference and validated with an indium check. The resulting heat flow curves were compared to a sapphire standard, both with corrected baselines. The heat capacity of each sample was derived from the heat flow data and was corrected for calibration discrepancies. Measurements were conducted under a nitrogen purge (purity 99.9999%) at a flow rate of 50 mL/min. The DSC measurement uncertainties were ±0.6 °C for temperature and ±2 J/g for heat flow.

3. Results and Discussion

3.1. Prediction of Thermodynamic Properties of Aqueous MDEA and [TBP][TFA] for CO₂ Absorption Using COSMO-RS

In this COSMO-RS simulation study, aqueous methyldiethanolamine (MDEA) and the ionic liquid [TBP][TFA] were evaluated as the components of a hybrid solvent for CO₂ absorption. This section utilizes COSMO-RS to predict the miscibility of [TBP][TFA] with water, the solubility of CO₂ in the components of hybrid solvents (i.e., aqueous MDEA at 10 to 50 wt.% and [TBP][TFA]), the Henry’s Law constant of CO₂ in these mixtures (aqueous MDEA and [TBP][TFA]), and the density of the components to assess their molar volumes in relation to CO₂ absorption. Aqueous MDEA has long been recognized as one of the standards in CO₂ scrubbing technology due to its effectiveness in capturing CO₂. The incorporation of the ionic liquid [TBP][TFA] into this system is expected to enhance the overall performance by achieving optimal miscibility with aqueous MDEA. It is essential for each component in the hybrid solvent system to be fully compatible and soluble with one another, as this combination significantly influences critical properties, such as viscosity, CO₂ solubility, and thermal capacity. This unique solvent system, consisting of water, [TBP][TFA], and MDEA, facilitates synergistic interactions between the ionic liquid and MDEA, thereby improving CO₂ capture efficiency and enhancing the solvent’s overall performance. Therefore, obtaining the miscibility data for ionic liquid [TBP][TFA] with water is crucial. Thus, the solubility of [TBP][TFA] was measured as log₁₀(x_solubility) at 298.15 K. The predictions indicated that the solubility (x_solubility) values for [TBP][TFA] with water was 1, demonstrating high miscibility with water [47]. These findings emphasize the compatibility of the ionic liquid with the aqueous system, particularly in conjunction with MDEA. Following this, the solubility of CO₂ in the hybrid solvent components, specifically the ionic liquid [TBP][TFA] and aqueous MDEA at concentrations of 10, 30, and 50 wt.%, was predicted at a temperature of 298.15 K. Figure 2 illustrates the solubility of CO₂ in both components at varying pressures, ranging from 2 to 20 bar of CO₂ partial pressure, predicted at 298.15 K. As the pressure increases, the predicted solubility of CO₂ in all absorbents gradually increases. The highest solubility trend is observed in [TBP][TFA], followed by aqueous MDEA, in the following order: 50 wt.% aqueous MDEA > 30 wt.% aqueous MDEA > 10 wt.% aqueous MDEA. The results indicate that, as the water weight percentage in the aqueous MDEA increases from 50 to 90 wt.%, the solubility of CO₂ is predicted to decrease significantly [48]. It has been reported that water content leads to a significant reduction in both the CO₂ loading capacity and absorption rate of the absorbent [48]. Pure [TBP][TFA], with its lower water content, allows for stronger Coulombic interactions between the CO₂ molecules and the fluorine atoms in the anion [49]. This interaction, along with the presence of fluorinated components, results in higher CO₂ solubility.

The Henry’s Law constants, as shown in Figure 3, indicate how well CO₂ dissolves in different solvents at 298.15 K. The constant for [TBP][TFA] is relatively low (70.78 mol/(L·atm)), suggesting that CO₂ has good solubility in this ionic liquid, making it effective for CO₂ capture. Conversely, the high constant for 10 wt.% aqueous MDEA (1774.54 mol/(L·atm)) indicates lower solubility and reduced effectiveness for CO₂ absorption. There is a clear trend: as the Henry’s Law constant decreases, the solubility of CO₂ in the solvent increases. This trend is evident in the 30 wt.% aqueous MDEA solution (381.56 mol/(L·atm)), where the solubility improves, and continues with the 50 wt.% aqueous MDEA (197.43 mol/(L·atm)), which shows further enhanced solubility. Therefore, these findings suggest that increasing the concentration of MDEA leads to higher CO₂ solubility, consistent with the observation that decreasing Henry’s Law values correlate with increased CO₂ solubility, with [TBP][TFA] demonstrating the greatest potential for effective CO₂ capture applications.

Table 3. Density, ρ (g/mL), volume (ų), and COSMO-volume (ų) of ionic liquid [TBP][TFA] and aqueous MDEA (10, 30, and 50 wt.%) at the temperature range of 298.15–33.15 K.

T (K)	[TBP][TFA]
	Density, ρ (g/mL)	Volume (Å3)	COSMO-Volume (Å3)
298.15	1.0060	614.78	493.23
303.15	1.0008	617.98	493.23
308.15	0.9956	621.2	493.23
313.15	0.9905	624.43	493.23
318.15	0.9853	627.68	493.23
323.15	0.9802	630.94	493.23
328.15	0.9752	634.22	493.23
333.15	0.9701	637.52	493.23
T (K)	10 wt.% aqueous MDEA
	Density, ρ (g/mL)	Volume (Å3)	COSMO-Volume (Å3)
298.15	0.9328	500.76	390.89
303.15	0.9278	503.45	390.89
308.15	0.9229	506.16	390.89
313.15	0.9179	508.87	390.89
318.15	0.913	511.6	390.89
323.15	0.9082	514.34	390.89
328.15	0.9033	517.09	390.89
333.15	0.8985	519.85	390.89
T (K)	30 wt.% aqueous MDEA
	Density, ρ (g/mL)	Volume (Å3)	COSMO-Volume (Å3)
298.15	0.95998	836.5	660.63
303.15	0.95493	840.93	660.63
308.15	0.94991	845.37	660.63
313.15	0.94492	849.84	660.63
318.15	0.93996	854.32	660.63
323.15	0.93503	858.83	660.63
328.15	0.93012	863.36	660.63
333.15	0.92525	867.91	660.63
T (K)	50 wt.% aqueous MDEA
	Density, ρ (g/mL)	Volume (Å3)	COSMO-Volume (Å3)
298.15	0.97159	1172.25	930.36
303.15	0.96652	1178.4	930.36
308.15	0.96147	1184.59	930.36
313.15	0.95646	1190.8	930.36
318.15	0.95147	1197.05	930.36
323.15	0.9465	1203.32	930.36
328.15	0.94157	1209.63	930.36
333.15	0.93666	1215.97	930.36

shows the density, volume, and COSMO-volume data for [TBP][TFA] and aqueous MDEA solutions (10, 30, and 50 wt.%) over the temperature range of 298.15 to 333.15 K. These physical properties provide key understandings of the CO₂ solubility predictions based on Henry’s Law constant at 298.15 K. The COSMO-volume remains constant for all absorbents, indicating that the solvent-accessible surface area and charge distribution do not change with temperature. In contrast, the real volume (geometrical volume) increases, and the density decreases with the rising temperature due to thermal expansion, which impacts the CO₂ removal capacity.

At 298.15 K, [TBP][TFA] shows the lowest Henry’s Law constant (70.78), indicating higher CO₂ solubility. This can be linked to its higher density (1.006) and smaller geometrical volume (614.78 cm³/mol), which suggest tighter molecular packing, enhancing CO₂ interactions. In comparison, the aqueous MDEA solutions exhibit higher Henry’s Law constants (1774.54 for 10 wt.%, 381.56 for 30 wt.%, and 197.43 for 50 wt.%), indicating reduced CO₂ solubility. As the water content in these solutions increases, the density decreases significantly, leading to larger volumes (e.g., 500.76 cm³/mol for 10 wt.% aqueous MDEA), which reduces the CO₂ solubility. The trend suggests that lower densities and larger volumes, characteristic of solutions with a higher water content, correlate with higher Henry’s Law constants and reduced CO₂ solubility. Conversely, systems with a higher density, such as [TBP][TFA], demonstrate lower Henry’s Law constants, indicating a greater CO₂ absorption capacity. Despite the variation in real volume and density, the constant COSMO-volume across all absorbents shows that the intrinsic molecular surface area and charge distribution remain unaffected by temperature changes, emphasizing that CO₂ solubility is more influenced by the real volume and density changes than by the COSMO-volume.

3.2. Predictions of the Sigma Profile, Sigma Potential, and Interaction Energy Using COSMO-RS

In this study, sigma profiles are employed to understand the electrostatic interactions between water, MDEA, [TBP][TFA], and CO₂. The sigma profile (Figure 4) elucidates the trend of CO₂ solubility in aqueous MDEA and [TBP][TFA]. Peaks within the range of a ±0.0086 e/Ų charge density indicate that the molecules primarily engage in van der Waals interactions [50]. In contrast, peaks outside this range suggest that hydrogen bonding is the preferred interaction due to the polarity of the species. CO₂ predominantly resides in the non-polar region around ±0.0086 e/Ų, allowing for potential van der Waals interactions with the non-polar regions of MDEA and the tetrabutylphosphonium cations in [TBP][TFA]. The OH⁻ groups of MDEA and water fall within the >+0.0086 e/Ų region, acting as hydrogen bond acceptors that can form hydrogen bonds. Similarly, the H⁺ ions of water and the amine groups of MDEA occupy the <-0.0086 e/Ų region (hydrogen bond donor region), indicating their propensity to form hydrogen bonds with acceptors. Additionally, the fluorine atoms in the TFA⁻ anion are expected to fall within the >+0.0086 e/Ų region, suggesting a tendency to form halogen bonds with the oxygen atoms in CO₂, thus exhibiting hydrogen bond acceptor properties.

The sigma potential of aqueous MDEA, [TBP][TFA], and CO₂ is depicted in Figure 5. The plot indicates that the carbon atom of CO₂ does not possess the ability to act as a hydrogen bond acceptor. Additionally, the [TBP][TFA] system is largely neutral concerning hydrogen bonding, as neither the TBP⁺ cation nor the TFA⁻ anion displays significant hydrogen bond donor or acceptor properties. However, the TFA⁻ anion can engage in halogen bonding interactions through its fluorine atoms. In the context of CO₂ interactions, TFA⁻ may act as a weak hydrogen bond acceptor due to its electronegative oxygen atom, which can interact with the polar C=O bonds in CO₂. Furthermore, the presence of fluorine atoms in TFA⁻ enhances halogen bonding with CO₂, contributing to the overall interaction, despite the system’s largely neutral nature in terms of hydrogen bonding. In contrast, aqueous MDEA exhibits both hydrogen bond acceptor and donor characteristics.

The interaction energies of aqueous MDEA and [TBP][TFA] in relation to CO₂ absorption are presented in

Table 4. Interaction energy (E_total) of aqueous MDEA, [TBP][TFA], and aqueous MDEA-[TBP][TFA] with CO₂.

System	EMF kcal/mol	EHB kcal/mol	EvdW kcal/mol	Etotal kcal/mol
System I
[TBP][TFA]	11.46	−1.63	−20.90	−11.06
[TBP][TFA] + CO2	12.32	−1.63	−23.00	−12.31
10 wt.% aqueous MDEA	0.56	−5.78	−1.60	−6.82
10 wt.% aqueous MDEA + CO2	1.41	−5.78	−3.71	−8.07
30 wt.% aqueous MDEA	1.10	−5.65	−2.93	−7.49
30 wt.% aqueous MDEA + CO2	1.95	−5.65	−5.03	−8.74
50 wt.% aqueous MDEA	1.63	−5.53	−4.25	−8.15
50 wt.% aqueous MDEA + CO2	2.49	−5.53	−6.36	−9.40
System II
10 wt.% aqueous MDEA − [TBP][TFA] + CO2	12.88	−7.41	−24.60	−19.14
30 wt.% aqueous MDEA − [TBP][TFA] + CO2	13.41	−7.28	−25.93	−19.80
50 wt.% aqueous MDEA − [TBP][TFA] + CO2	13.95	−7.16	−27.25	−20.46

. From the data, it is evident that the interaction between the [TBP][TFA] system and CO₂ is predominantly driven by van der Waals forces, with a smaller contribution from hydrogen bonding, and minor involvement of misfit electrostatic interactions. For the aqueous MDEA systems, both hydrogen bonding and van der Waals interactions are significant, with a hydrogen bonding energy that is notably higher than that of the [TBP][TFA] system. The addition of [TBP][TFA] to the aqueous MDEA system increases the van der Waals and misfit interactions, enhancing the overall interaction energy with CO₂. The interaction energy, in increasing order, follows the trend: 10 wt.% aqueous MDEA + [TBP][TFA] < 30 wt.% aqueous MDEA + [TBP][TFA] < 50 wt.% aqueous MDEA + [TBP][TFA].

3.3. Preparation of the Novel Aqueous MDEA–[TBP][TFA] Hybrid Solvent

3.3.1. Synthesis and Characterization of [TBP][TFA] Ionic Liquid

Tetrabutylphosphonium trifluoroacetate ([TBP][TFA]) was synthesized via an acid-base neutralization reaction between tetrabutylphosphonium hydroxide and trifluoroacetic acid, as shown in Figure 6. The structure of [TBP][TFA] was verified using ¹H and ¹³C NMR, as well as attenuated total reflectance FT-IR (ATR-FT-IR) analyses. Figure 7a,b display the ¹H- and ¹³C-NMR spectra, respectively. The successful synthesis of [TBP][TFA] was confirmed by calculating the integral ratios for the protons in the alkyl group using ¹H NMR and the total carbons in ¹³C NMR. FT-IR spectroscopy further confirms the presence of characteristic functional groups in [TBP][TFA], as shown in Figure 8. In the FT-IR spectrum, the absorption band at 1112.15 cm⁻¹ corresponds to the C–F stretching mode, while the C–H stretching mode is evident in the functional group region at 2934.64 cm⁻¹. Additionally, two bands at 1194 and 1685 cm⁻¹ are associated with the C=O stretching mode, and phosphonium–carbon (P–C) stretching is observed at 1230 cm⁻¹. The purity of [TBP][TFA] was assessed using ¹H NMR and Karl Fischer titration. No additional peaks were observed in the ¹H NMR spectrum, indicating the absence of impurities in the prepared [TBP][TFA] ionic liquid. Additionally, Karl Fischer analysis revealed a water content of only 0.278%, confirming a high purity level of 99.72% for the [TBP][TFA].

3.3.2. Thermophysical Characterization of the Aqueous MDEA-[TBP][TFA] Hybrid Solvent

Figure 9 depicts the densities of (a) 10 wt.% aqueous MDEA-[TBP][TFA], (b) 30 wt.% aqueous MDEA-[TBP][TFA], and (c) 50 wt.% aqueous MDEA-[TBP][TFA] across temperatures ranging from 20 to 80 °C. As the temperature increases, the density of the hybrid solvents decreases due to thermal expansion, which is typical of liquid systems. The addition of [TBP][TFA] at 2, 11, and 20 wt.% shows a notable increase in density, with the order of densities being 10 wt.% < 30 wt.% < 50 wt.% aqueous MDEA-[TBP][TFA]. The higher the concentration of [TBP][TFA], the greater the density of the hybrid solvent, which can be attributed to the higher molecular weight and denser structure of the ionic liquid compared to water and MDEA. When compared with the predicted density trend, the experimental results align well with the COSMO-RS predictions. As the MDEA concentration increases from 10 to 50 wt.%, a corresponding increase in density is observed. Since [TBP][TFA] has both predicted and experimental density values higher than those of aqueous MDEA, adding the ionic liquid in increasing concentrations (from 2 to 20 wt.%) consistently increases the density of all systems.

Figure 10 illustrates the viscosities of (a) 10 wt.% aqueous MDEA-[TBP][TFA], (b) 30 wt.% aqueous MDEA-[TBP][TFA], and (c) 50 wt.% aqueous MDEA-[TBP][TFA] across a temperature range of 20 to 80 °C. The viscosity follows a similar trend to density, decreasing as the temperature increases. Due to the higher viscosity of [TBP][TFA] compared to MDEA, the addition of [TBP][TFA] significantly affects the overall viscosity of the hybrid solvents. The inclusion of [TBP][TFA] at 2, 11, and 20 wt.% causes a pronounced increase in viscosity, with the order being 10 wt.% < 30 wt.% < 50 wt.% aqueous MDEA-[TBP][TFA], consistent with the increasing [TBP][TFA] concentration. The increase in viscosity with the addition of [TBP][TFA] can be attributed to stronger intermolecular interactions, such as van der Waals forces and hydrogen bonding, between the ionic liquid and water or MDEA molecules [51]. This suggests that the structure and composition of the hybrid solvents lead to higher resistance to flow. As the temperature increases, the kinetic energy of the molecules also rises, weakening these intermolecular forces and leading to a decrease in viscosity [51]. The larger molecular size and structure of [TBP][TFA] contribute to this effect, as they hinder molecular movement more than MDEA alone, particularly at higher concentrations.

The heat capacity measurements for various concentrations of aqueous MDEA mixed with [TBP][TFA] at temperatures ranging from 20 to 80°C are summarized in the

Table 5. Heat capacity measurements of 10 and 50 wt.% aqueous MDEA–[TBP][TFA] hybrid solvents with selected [TBP][TFA] concentrations across a temperature range of 20 to 80 °C, including the heat capacity of [TBP][TFA].

T	10 wt.% MDEA	10 wt.% MDEA 2 wt.% IL	10 wt.% MDEA 20 wt.% IL	50 wt.% MDEA	50 wt.% MDEA 2 wt.% IL	50 wt.% MDEA 20 wt.% IL	[TBP][TFA] IL
°C	J/g °C
20	3.37	3.11	2.97	4.99	4.72	3.74	1.71
30	3.34	3.08	2.94	5.25	4.96	3.91	1.69
40	3.34	3.08	2.94	5.42	5.1	4.02	1.72
50	3.38	3.13	3.00	5.65	5.29	4.18	1.79
60	3.36	3.11	3.00	5.88	5.48	4.35	1.81
70	3.36	3.09	3.01	6.12	5.69	4.52	1.85
80	3.34	3.06	3.04	6.34	5.85	4.67	1.88

. Measurements were conducted on the lowest concentration of aqueous MDEA (10 wt.%) and the highest concentration (50 wt.%), comparing these values to those of their respective hybrid solvents and [TBP][TFA]. The data indicate a general increase in heat capacity values as the temperature rises across all samples. Notably, [TBP][TFA] exhibits the lowest heat capacity values among the studied temperatures compared to both 10 and 50 wt.% aqueous MDEA. The addition of [TBP][TFA] significantly alters the heat capacity values, resulting in lower heat capacities for the hybrid solvents compared to pure aqueous MDEA, particularly at the lower concentration (2 wt.% IL). This observation suggests that [TBP][TFA] may dilute the overall heat capacity of the mixture. For 50 wt.% MDEA, the heat capacity also shows an upward trend, increasing from 4.99 J/g °C at 20 °C to 6.34 J/g °C at 80 °C, indicating that a higher concentration of MDEA significantly contributes to the thermal properties of the solvent system. While the presence of [TBP][TFA] reduces the thermal properties of the hybrid solvent system, it also impacts the energy consumption required for the regeneration of the hybrid solvents following the CO₂ absorption process, leading to lower energy requirements for solvent recovery and enhancing their recyclability.

3.4. CO₂ Absorption Study of Aqueous MDEA– [TBP][TFA] Hybrid Solvents Using RSM

The response surface methodology (RSM) was employed to evaluate the influence of MDEA concentration (wt.%), [TBP][TFA] concentration (wt.%), pressure, and temperature on CO₂ absorption performance. Each of these independent variables was examined at three levels. CO₂ absorption was assessed in terms of the removal capacity (mol/kg). A total of 30 experiments was conducted, as shown in

Table 6. The face-centered central composite design (FC-CCD) matrix and the removal capacity of CO₂ in the aqueous MDEA–[TBP][TFA] hybrid solvents.

Factor 1	Factor 2	Factor 3	Factor 4	Response
A: MDEA	B: IL	C: T	D: P	CO2 Removal Capacity
wt.%	wt.%	°C	Bar	Actual Value (mol/kg)	Predicted Value (mol/kg)
30	11	45	16	6.61	6.88
30	11	45	16	6.92	6.88
30	11	45	30	9.66	9.78
50	20	30	30	12.12	12.14
50	20	60	2	1.50	2.12
10	2	30	30	9.26	8.84
30	11	45	16	6.98	6.88
30	20	45	16	6.23	6.38
30	11	30	16	7.54	7.9
10	20	60	2	2.18	1.71
30	2	45	16	6.92	6.87
50	2	60	30	8.88	8.86
50	2	60	2	3.87	3.71
30	11	45	2	3.61	3.6
50	2	30	2	5.43	5.47
30	11	45	16	6.97	6.88
30	11	45	16	7.42	6.88
10	11	45	16	5.40	5.62
10	20	30	30	9.52	9.45
30	11	45	16	6.70	6.88
50	20	60	30	8.69	8.45
10	2	60	30	6.61	6.79
50	20	30	2	4.51	4.1
30	11	60	16	6.13	5.89
50	11	45	16	7.68	7.57
10	20	60	30	7.02	7.18
10	2	30	2	2.81	2.82
10	20	30	2	2.04	2.25
50	2	30	30	12.11	12.04
10	2	60	2	2.33	2.49

, including 24 factorial points and six replicate runs at the design center. These replicates enabled the estimation of experimental errors and enhanced the accuracy of the analysis. The RSM suggested a quadratic model for four independent variables. Basically, the independent variables are coded as A, B, C, and D, or, respectively, as x₁, x₂, x₃, and x₄. In this study, the coded terms are A: concentration of MDEA, B: concentration of [TBP][TFA], C: temperature, and D: pressure. For this study, the quadratic model is presented in Equation (10):

$$Removalcapacity(\mathrm{m}\mathrm{o}\mathrm{l}/\mathrm{k}\mathrm{g})=1.21-0.00112AB+0.000761AD-0.000387BC+0.000232BD-0.00205CD-0.000708A^2-0.00309B^2+0.00070C^2-0.000962D^2$$

(10)

Analysis of variance (ANOVA) and the statistical parameters for CO₂ removal capacity are presented in

Table 7. ANOVA and fit statistics.

ANOVA
Source	Sum of Squares	df	Mean Square	F-Value	p-Value
Model	218.8	14	15.63	121.61	<0.0001	significant
A-MDEA	17.21	1	17.21	133.95	<0.0001
B-IL	1.08	1	1.08	8.42	0.011
C-Temperature	18.22	1	18.22	141.75	<0.0001
D-Pressure	171.53	1	171.53	1334.77	<0.0001
AB	0.6504	1	0.6504	5.06	0.0399
AC	2.05	1	2.05	15.94	0.0012
AD	0.7271	1	0.7271	5.66	0.0311
BC	0.0438	1	0.0438	0.3406	0.5681
BD	1.37	1	1.37	10.65	0.0052
CD	2.96	1	2.96	23.04	0.0002
A2	0.208	1	0.208	1.62	0.2226
B2	0.1625	1	0.1625	1.26	0.2784
C2	0.0006	1	0.0006	0.005	0.9444
D2	0.0922	1	0.0922	0.7172	0.4104
Residual	1.93	15	0.1285
Lack of Fit	1.53	10	0.1533	1.94	0.2399	not significant
Pure Error	0.3944	5	0.0789
Cor Total	220.73	29
Fit Statistics
Std. Dev.	0.3585		R2		0.9913
Mean	6.45		Adjusted R2		0.9831
C.V. %	5.55		Predicted R2		0.945
			Adeq Precision		41.9434

; the model p-value < 0.05 shows that the model is significant. From the table, the parameter of concentration of MDEA, temperature, and pressure have a p-value < 0.0001, except for the concentration of [TBP][TFA], with a p-value < 0.011. For a p-value < 0.0001, the parameters significantly affect the removal capacity of CO₂. The lack-of-fit F-value of 1.94 suggests that the lack of fit is not significant compared to the pure error, indicating a good model fit. With a 23.99% probability that this F-value can result from random noise, the non-significant lack of fit confirms that the model adequately represents the data, which is desirable for a reliable analysis. The fit statistics indicate that the model for CO₂ removal capacity performs exceptionally well. The R² value of 0.9913 demonstrates that the model explains 99.13% of the variance in the data, while the adjusted R² of 0.9831 confirms a strong fit, accounting for the number of predictors. The predicted R² of 0.9450 is in reasonable agreement with the adjusted R², showing that the model generalizes effectively and is not overfitting. Additionally, the Adeq Precision value of 41.943 far exceeds the threshold of 4, indicating a strong signal-to-noise ratio, which makes the model highly reliable for navigating the design space. With a low standard deviation of 0.358 and a C.V. % of 5.55%, the model is both precise and robust, making it suitable for analyzing and predicting CO₂ removal capacity.

In this work, Design Expert 13 was employed to generate three-dimensional (3D) response surfaces to represent the effects of independent variables on CO₂ removal capacity. These 3D response curves help visualize and understand the interactions between variables, allowing for the identification of optimal conditions for maximum CO₂ removal. The color variations in the plots indicate the degree of interaction based on CO₂ removal capacity, with more reliable results seen in the mutual influence of two independent variables. Figure 11a–f illustrates the 3D plots corresponding to the four independent variables studied and their effects on CO₂ removal capacity. Figure 11a shows that increasing the concentration of [TBP][TFA] has a minimal effect across the range of MDEA concentrations (10 to 50 wt.%). However, Figure 11b demonstrates that the interaction between temperature and pressure has a significant impact, with the highest CO₂ removal capacity occurring at the highest pressure (30 bar) and lowest temperature (30 °C). In Figure 11c, the combination of MDEA concentration and pressure shows that the highest removal capacity is achieved at 50 wt.% MDEA and 30 bar pressure. Figure 11d highlights that increasing the pressure results in the highest removal capacity across all concentrations of [TBP][TFA], with maximum removal observed at 30 bar. Similarly, Figure 11e illustrates that lower temperatures lead to a higher CO₂ removal capacity across all concentrations of [TBP][TFA]. Finally, Figure 11f shows that the highest CO₂ removal occurs at 50 wt.% MDEA and 30 °C. As shown in Table 6, the optimal conditions for CO₂ removal are achieved with 50 wt.% MDEA, 20 wt.% [TBP][TFA], 30 bar, and 30 °C, yielding a removal capacity of 12.14 mol/kg.

The quadratic model is validated by running the highest removal capacity for four replicate runs, as shown in

Table 8. Validation study for the quadratic model.

Condition	Removal Capacity (mol/kg)		Standard Deviation
	Predicted	Actual
50:20:30:30	12.14	11.45	0.48
	12.14	10.75	0.98
	12.14	10.20	1.37
	12.14	13.01	0.61
Coefficient of variation (%)			6.50

. The runs were based on the highest removal capacity set as the goal, where the variables were within the target range. Based on the highest desirability score obtained, which was 1, the removal capacity was optimized at MDEA (50 wt.%), [TBP][TFA] (20 wt.%), temperature (30 °C), and pressure (30 bars), simplified as 50:20:30:30 in Table 8. The actual average value obtained was 11.35 mol/kg, deviating from the predicted value by 0.87 mol/kg, with a coefficient of variation calculated at 6.50%. This small deviation and the coefficient of variation indicate the strong predictive capability of the model. The actual values show a good agreement with the predicted values within 95% confidence intervals, further validating the model’s reliability.

Based on the RSM experimental results for the removal capacity at the optimal conditions of 30 °C and 30 bar, the outcomes for 10 wt.% MDEA, 50 wt.% MDEA, their hybrid solvent, and the [TBP][TFA] ionic liquid are summarized in

Table 9. Removal capacity of aqueous MDEA (10 and 50 wt.%), the hybrid solvents, and [TBP][TFA] ionic liquid at optimum temperature (30 °C) and pressure (30 bar).

Absorbent	Removal Capacity (mol/kg)
10 wt.% MDEA	9.83
10 wt.% MDEA:2 wt.% IL	9.26
10 wt.% MDEA:20 wt.% IL	9.52
50 wt.% MDEA	14.64
50 wt.% MDEA:2 wt.% IL	12.11
50 wt.% MDEA:20 wt.% IL	12.10
[TBP][TFA] IL	10.62

. The experimental findings demonstrate that aqueous MDEA concentration (wt.%), along with temperature and pressure, plays a significant role in CO₂ removal. As the MDEA concentration increases from 10 to 50 wt.%, the removal capacity rises from 9.83 to 14.64 mol/kg, respectively. This trend aligns well with the solubility data predicted by COSMO-RS, where CO₂ solubility increases with higher MDEA concentrations.

3.5. Comparison of CO₂ Removal Capacity and Heat Capacity of Hybrid Solvents with Aqueous MDEA and [TBP][TFA] Ionic Liquid

The removal capacity of aqueous MDEA-[TBP][TFA] hybrid solvents and their components, i.e., aqueous MDEA and [TBP][TFA] ionic liquid, was compared with their heat capacity. This comparison was conducted at the optimal temperature and pressure obtained from the RSM optimization, which was 30 bar and 30 °C, and the resulting data are presented in Figure 12. Based on the results, 10 wt.% aqueous MDEA has a removal capacity of 9.83 mol/kg and a heat capacity of 3.34 J/°Cg. The addition of [TBP][TFA] at 2 wt.% slightly reduces the removal capacity to 9.26 mol/kg, but also lowers the heat capacity by 7%, bringing it down to 3.08 J/°Cg. A further addition of [TBP][TFA] at 20 wt.% results in a 12% reduction in the heat capacity to 2.94 J/°Cg, with a removal capacity of 9.52 mol/kg. This demonstrates that incorporating IL into the system slightly reduces the removal capacity while enhancing heat efficiency, indicating potential energy savings during regeneration. Although the CO₂ absorption capacity is slightly reduced, the removal capacity remains comparable despite the lower heat capacity.

For 50 wt.% aqueous MDEA, the highest removal capacity is achieved at 14.64 mol/kg, with a heat capacity of 5.25 J/°Cg. Adding [TBP][TFA] at 2 wt.% results in a reduction in removal capacity to 12.11 mol/kg, along with a 5% reduction in heat capacity to 4.96 J/°Cg. Further increasing the IL content to 20 wt.% leads to a 26% reduction in heat capacity (from 5.25 to 3.91 J/°Cg), while the removal capacity remains steady at 12.14 mol/kg. The results highlight that the addition of [TBP][TFA] to aqueous MDEA effectively reduces heat capacity, offering potential energy savings during regeneration, while maintaining a reasonable removal capacity for CO₂ capture.

4. Conclusions

In conclusion, the thermodynamic properties and interaction energy of the aqueous MDEA-[TBP][TFA] hybrid solvent were successfully predicted using COSMO-RS. Based on the prediction, the interaction of [TBP][TFA] with CO₂ is primarily driven by van der Waals forces, while hydrogen bonding is more pronounced in the aqueous MDEA. The analysis of Henry’s Law constant and CO₂ solubility confirms a reciprocal relationship, whereby an increase in CO₂ solubility corresponds to a decrease in the Henry’s Law constant. The [TBP][TFA] ionic liquid was effectively synthesized and characterized through NMR, FTIR, and Karl Fischer analyses. Subsequently, the aqueous MDEA-[TBP][TFA] hybrid solvent was also successfully prepared, characterized, and employed for CO₂ removal. Using the response surface methodology (RSM) with a central composite design (CCD), a quadratic model was obtained and validated, indicating that the optimal conditions for CO₂ removal were achieved at 50 wt.% MDEA, 20 wt.% [TBP][TFA], 30 °C, and 30 bar. The most influential variables affecting the CO₂ removal capacity were MDEA concentration (wt.%), pressure, and temperature. A comparative study of the heat capacity demonstrated that the presence of [TBP][TFA] effectively reduces the heat capacity of the hybrid solvents up to 26%, suggesting potential applications for lowering the regeneration costs of solvents during CO₂ capture.