Theoretical Computation-Driven Screening and Mechanism Study of Washing Oil Composite Solvents for Benzene Waste Gas Absorption

Abstract

In order to solve the problems of high volatility and insufficient absorption effect when using chemical by-product washing oil to treat benzene-containing waste gas, this study innovatively proposed a composite solvent screening method based on the solvation free energy (ΔGsol), and reasonably predicted the absorption performance of 26 solvents for benzene. Through theoretical calculation and experimental verification, tetraethylene glycol dimethyl ether (TGDE) was finally determined to be the optimal composite component of washing oil. The absorption efficiency of the composite solvent reached 96.2%, and the regeneration efficiency was stable after 12 cycles with a mass loss of only 2.4%. Quantum computing simulation revealed that the dispersion force is dominant between benzene and the solvent, and TGDE enhances the electrostatic interaction through weak hydrogen bonds. The synergistic effect of the two improves the absorption performance. This study provides theoretical and technical support for the development of efficient and renewable benzene waste gas recovery solvent systems.

1. Introduction

Benzene is a typical volatile organic compound (VOC) [1], and has been classified as a Group 1 carcinogen, drawing global attention due to its environmental and health impacts [2]. As a fundamental chemical in the petrochemical industry, benzene serves as a key precursor for derivatives such as styrene, cumene, cyclohexane, and aniline [3], and is widely used in plastics, rubber, and agrochemicals [4]. However, its high volatility leads to easy release into the environment during production, transportation, and fuel combustion [5], resulting in severe contamination of air, groundwater, and soil ecosystems [6]. Consequently, various countries have established regulations to limit excessive industrial emissions of VOCs [7].

Current commonly used technologies for benzene waste gas treatment include membrane separation [8], condensation, adsorption [9], and absorption [10,11]. Among these, solvent absorption is widely employed industrially due to its mature technology, high processing capacity, and efficient removal of aromatic hydrocarbons [12]. Solvent selection is a critical factor determining the economic feasibility and efficiency of absorption. Traditional solvents such as washing oil (a high-boiling-point hydrocarbon mixture) are mainstream choices in the coke chemical industry because they have a strong affinity for benzene and are usually recovered from rich wash oil by slightly overpressure distillation. In rare cases, vacuum distillation is used for regeneration [13,14]. Studies by K. BAY et al. [15] reported an infinite activity coefficient of 0.628 for biodiesels in absorbing the benzene series, indicating good absorption performance. A. Erto et al. [16] investigated four different copolymers for benzene removal and found that Pluronic P123 exhibited favorable Henry’s constants (59.6 atm at 293 K and 1178.1 atm at 363 K) during absorption/desorption. However, washing oil has significant limitations in practical applications. First, its high volatility leads to considerable loss: Vecer et al. [17] reported a volatility loss of 3.9 wt.% during benzene absorption, increasing solvent replenishment costs and posing secondary pollution risks. Second, the absorption efficiency of washing oil is constrained by its reliance solely on non-polar interactions, making it difficult to meet increasingly stringent environmental regulations [18]. Moreover, its regeneration performance declines with repeated cycles, undermining long-term economic viability. These drawbacks have motivated the exploration of more efficient, low-volatility solvent systems.

In recent years, novel blended absorbents have been extensively studied for treating benzene series compounds [19,20]. For instance, Zhang et al. [21] demonstrated that an ionic liquid (IL)-based solvent ([HMIM][Cl]–PEG200) achieved a benzene absorption efficiency of 87.15% from mixed gases, with quantum chemical calculations further revealing hydrogen bonding interactions with benzene molecules. Liu et al. [22] combined experiments and molecular dynamics simulations to show that supramolecular deep eutectic solvents (DES) enhance benzene absorption capacity through π–π stacking interactions. These studies indicate that compounding different components can enhance absorption via synergistic effects. However, the high synthesis costs of ILs hinder large-scale application [23,24]. Moreover, DESs and ILs often have poorer gas–liquid mass transfer efficiency and greater viscosities than traditional organic solvents. Inspired by compounding strategies, incorporating efficient, low-volatility components into washing oil systems could balance absorption performance and economic efficiency. Nevertheless, rapidly identifying suitable compound components from numerous candidates remains challenging. Traditional screening methods rely heavily on extensive experiments, which are time-consuming and costly. While some progress has been made in rapidly predicting solvent properties via computational simulations [25,26], no study has yet applied this approach to optimize washing oil compound systems.

To address this, this study proposes an innovative washing oil compounding strategy, employing computational solvation free energy (ΔGsol) calculations for the first time to screen highly efficient, low-volatility components from a broad range of candidate solvents. Dynamic absorption and cyclic regeneration experiments were conducted to validate the absorption performance of the compound system with washing oil. Furthermore, quantum chemical calculations were employed to elucidate the interaction mechanisms among the compound component, washing oil, and benzene molecules. The aim is to establish a combined theoretical and experimental strategy for solvent screening and optimization, providing theoretical guidance for the design of high-efficiency absorption systems.

2. Materials and Methods

2.1. Materials

The reagents used in the experiment are as follows: the washing oil solvent consists of biphenyl (purity ≥ 99%), naphthalene (purity ≥ 98%), 1-methylnaphthalene (purity ≥ 98%), 2-methylnaphthalene (purity ≥ 97%), dimethylnaphthalene (purity ≥ 98%), fluorene (purity ≥ 99%), dibenzofuran (purity ≥ 98%), and acenaphthylene (purity ≥ 97%). The specific content of each component is shown in Table S1. The 26 other solvents selected for the experiment are shown in Table S2. These solvents were screened based on their diverse functional groups, high boiling points to minimize volatility loss, and commercial availability for potential industrial application. All the above reagents were purchased from Shanghai Aladdin Reagent Co., Ltd., Shanghai, China.

It is important to note that a model washing oil composed of eight representative components was employed in this study, rather than real industrial washing oil. This simplification was necessary to facilitate the quantum chemical calculations based on defined molecular structures and to ensure experimental reproducibility. However, this model excludes heavy polymer components typically present in industrial oil, which implies that the actual thermal stability in industrial applications may be higher than that observed in this study.

2.2. Determination of Partition Coefficient

The partition coefficient (K) of benzene in different solvents was determined using the standard static method [27]. The headspace vials used in this study were equipped with polypropylene caps and polytetrafluoroethylene gaskets to ensure airtightness. Before testing, 8–10 g of solvent was added to the vial, followed by the injection of 10 μL of liquid benzene using a microsyringe. The vial was placed in a constant-temperature oven and maintained at 303 K for 24 h to allow equilibrium. The benzene concentration in the gas phase was subsequently measured using gas chromatography (GC2010 Plus, Shimadzu, Tokyo, Japan). The partition coefficient K was calculated according to the following formula [28]:

$$K={\displaystyle \frac{C_L}{C_G}}$$

(1)

where K is the distribution coefficient, dimensionless; and C_L and C_G are the concentrations of benzene in the liquid and gas phases, respectively, in mg/m³.

2.3. Absorption and Desorption Experiments

A packed column was employed to simulate real absorption conditions. A packed column was employed to simulate real absorption conditions using θ ring packing (4 × 4 mm) as the stationary phase. The column had a total height of 1 m, an internal diameter of 25 mm, and an effective packing height of 75 cm. The absorption temperature was maintained at 303 K. During operation, The waste gas flow rate was set at 4 L/min, with a benzene concentration of 10,000 mg/m³. The liquid-to-gas ratio throughout the absorption process was maintained at 3.0 L/m³. The detailed absorption process is illustrated in Figure 1. The absorption efficiency of the solvent system was calculated by measuring the benzene concentrations at the inlet and outlet of the packed column, using the following formula:

$$\mathsf{\eta }={\displaystyle \frac{{\mathrm{c}}_1}{{\mathrm{c}}_2}}\times 100$$

(2)

where c₁ and c₂ represent the concentrations of benzene in the gas phase (in mg/m³) at the outlet and inlet, respectively. The rich solution after absorption was regenerated using a rotary evaporator under conditions of 0.1 atm and 368 K for 1 h.

2.4. Quantum Chemical Calculations

Based on Density Functional Theory (DFT), analyzing the interaction forces between solvent molecules and benzene at the microscopic molecular level can contribute to understanding the absorption mechanism and provide a novel strategy for the theoretical screening of suitable solvents. The quantum chemistry software Gaussian 16 (version A.03) [29] was employed to perform geometry optimization and energy calculations for benzene and solvent molecules at the M062X/6-311G** level, with the ΔGsol computed using the solvent model based on density (SMD) solvation model. On the basis of the optimized structures from Gaussian, the σ-profile diagrams were generated using the Dmol³ module in Materials Studio 2017 to analyze trends in molecular surface electrostatic distribution [30]. Finally, electrostatic potential (ESP) analysis and independent gradient model (IGM) analysis [31,32] were conducted using the wavefunction analysis software Multiwfn 3.7 [33,34]. Energy decomposition [35] was carried out with the Psi4 software [36]. All results were visualized using Visual Molecular Dynamics (VMD, version 1.9.3) [37].

3. Results

3.1. Screening Method for Washing Oil Composite Solvent

ΔGsol is a thermodynamic parameter that characterizes the change in Gibbs free energy when a solute dissolves from the gas phase into a solvent. It serves as a key indicator for evaluating the absorption performance of solvents [38]. A negative value of ΔGsol indicates a spontaneous absorption process, and the larger the absolute value of the negative value, the stronger the absorption capacity.

In this study, 26 commonly used organic reagents with diverse structures were selected. The ΔGsol of benzene in these solvents was computed using Gaussian 16 W, and the K was experimentally determined using the headspace method. Both the experimental and computational results are summarized in

Table 1. ΔGsol and lnK for different solvents.

Solvents	ΔGsol/kJ·mol−1	K	lnK
Sulfolane	−19.18	2026.46	7.61
N-Methylpyrrolidone	−18.12	1329.84	7.19
Tetraethylene glycol dimethyl ether	−16.87	808.99	6.70
Dibutyl phthalate	−16.68	751.91	6.62
Triethylene glycol butyl ether	−16.37	665.09	6.50
Dibutyl adipate	−15.97	565.76	6.34
Diethylene glycol butyl ether	−15	384.91	5.95
Triethylene glycol dimethyl ether	−14.87	366.59	5.90
Triethylene glycol monomethyl ether	−14.78	351.77	5.86
Hexyl acetate	−14.26	287.09	5.66
Morpholine	−14.13	272.75	5.61
Benzyl alcohol	−14.1	269.89	5.60
Phenylethyl alcohol	−13.88	247.70	5.51
N,N-dimethylformamide	−13.7	230.03	5.44
Ethylene glycol	−13.42	205.56	5.33
Aniline	−13.21	189.65	5.25
Tetradecane	−13.1	175.53	5.17
Triethylene glycol	−13.02	169.51	5.13
1,2,4-Trimethylbenzene	−12.92	162.16	5.09
Ethylene glycol butyl ether	−12.82	160.02	5.08
Anisole	−12.78	147.66	4.99
triethanolamine	−12.58	122.88	4.81
Pentanol	−11.84	110.07	4.70
Octanol	−11.21	85.59	4.45
Octylamine	−11.15	83.77	4.43
Water	−9.23	38.94	3.66

. The relationship between ΔGsol and lnK is given by the following equation [39,40]:

$${\mathsf{\Delta }G}_{sol}=-RT\mathit{ln}K$$

(3)

where R is the ideal gas constant (8.314 J·mol⁻¹·K⁻¹) and T is the temperature in K. A linear fitting was performed between ΔGsol and lnK, and the result is shown in Figure 2.

Under the solvent systems and experimental conditions investigated in this study, a strong correlation was observed between ΔGsol and lnK, with a correlation coefficient R² = 0.9998. These results demonstrate that calculating the ΔGsol of benzene in various solvents can effectively predict and compare the absorption capacities of different solvents, thereby providing a theoretical basis for screening highly efficient absorbents, consistent with screening strategies reported in recent computational thermodynamic studies [41,42,43].

As shown in Table 1, six solvents—sulfolane (SUL), N-methylpyrrolidone (NMP), tetraethylene glycol dimethyl ether (TGDE), dibutyl phthalate (DBP), triethylene glycol butyl ether (TGBE), and dibutyl adipate (DBA)—exhibited significantly higher K values for benzene compared to other solvents. Specifically, their K values (SUL: 2026.46, NMP: 1329.84, TGDE: 808.99, DBP: 751.91, TGBE: 665.09, DBA: 565.76) were considerably higher than that of diethylene glycol butyl ether (K = 384.91). More importantly, compared to water (K = 38.94), which is commonly used in industrial applications, the K values of these six solvents were higher by factors of 50.74, 33.30, 20.26, 18.83, 16.65, and 14.17, respectively. The corresponding ΔGsol values (ranging from −19.18 to −15.97 kJ·mol⁻¹) indicate strong affinity between these solvents and benzene molecules. In contrast, water has a ΔGsol of only −9.23 kJ·mol⁻¹, indicating very low absorption capacity, a limitation frequently noted in conventional aqueous absorption systems due to the hydrophobic nature of benzene [44,45]. Similarly, other highly polar protic solvents, such as ethylene glycol (K = 205.56), pentanol (K = 110.07), and octanol (K = 85.59), showed relatively low K values, due to their insufficient lipophilicity and significant polarity mismatch with benzene, suggesting poor absorption performance that fails to meet the requirements for high efficiency.

To further evaluate solvent volatility, the saturated vapor pressure data of the six candidate solvents were obtained. The results showed that at 298 K, the saturated vapor pressure of NMP was 0.345 mmHg [46], which is significantly higher than those of the other solvents—for example, 0.0019 mmHg for TGDE at 298 K [47], 0.0062 mmHg for SUL at 300.6 K [48], and 2.01 × 10⁻⁵ mmHg for DBP at 298 K [49]. Since the saturated vapor pressure of a solvent is positively correlated with its volatility, a high vapor pressure can lead to significant solvent loss due to evaporation. Therefore, to minimize solvent loss in both experimental and industrial applications, NMP was excluded from the candidate list.

Consequently, DBP, SUL, TGDE, DBA, and TGBE were selected as compound components for washing oil for further theoretical analysis and experimental investigation.

3.2. Absorption–Desorption Experiment of Washing Oil Composite Solvent

Based on the screening results, dynamic absorption experiments of benzene were conducted in a counter-current packed column absorption setup to validate the absorption performance and optimize the washing oil composite system. To ensure data reliability, all dynamic absorption and regeneration experiments were performed in triplicate. The results reported in this section are expressed as mean values ± standard deviation, as indicated by the error bars in the figures. The experimental results are shown in Figure 3. Compared to the washing oil solvent, the pure composite component absorbents exhibited stronger benzene absorption capacities. Specifically, the absorption efficiencies of DBA, DBP, and TGDE were 2.1%, 2.6%, and 3.0% higher than that of washing oil, respectively, demonstrating significant advantages. In contrast, TGBE, with a relatively lower partition coefficient (K = 665.09), showed only marginal improvement in absorption performance, resulting in a negligible difference compared to washing oil. Meanwhile, the enhancement in absorption performance for SUL was also limited, which may be attributed to its high viscosity (10.07 mPa·s at 298 K) hindering mass transfer during absorption. TGDE stood out due to the highest improvement in absorption efficiency (3.0%). Moreover, TGDE possesses relatively low viscosity [50], which helps reduce flow resistance during solvent absorption and enhances mass transfer efficiency. It also exhibits excellent thermal and chemical stability. Therefore, considering both absorption performance and physical properties, TGDE was selected as the composite solvent.

To investigate the effect of TGDE compounding ratio on absorption performance, TGDE was compounded with washing oil at mass ratios of 5 wt.%, 10 wt.%, 15 wt.%, 20 wt.%, 25 wt.%, 30 wt.%, and 35 wt.%. The absorption efficiency results are shown in Figure 4. As the proportion of TGDE increased, the absorption efficiency continued to improve, indicating that TGDE significantly enhanced the benzene absorption capacity of the system by strengthening intermolecular interactions. However, when the TGDE ratio exceeded 25 wt.%, the incremental gain in absorption efficiency per 5 wt.% increase was only 0.2%, demonstrating a marginal effect. Considering the cost-effectiveness for industrial applications, a TGDE compounding ratio of 25 wt.% was identified as the optimal choice to balance high absorption efficiency and economic feasibility.

From both environmental and economic perspectives, the regenerability of solvents is a critical indicator for assessing their industrial applicability. Therefore, in this study, the benzene-rich solution obtained after absorption was regenerated via vacuum distillation (368 K, 0.1 atm, 1 h). It should be noted that these laboratory conditions are milder than typical industrial practices (413–443 K under slight excess pressure); thus, while the current results demonstrate feasibility, the long-term stability under harsher industrial regeneration conditions requires further validation. Multiple absorption–desorption cycles were conducted to evaluate cycling stability. Multiple absorption–desorption cycles were conducted to evaluate cycling stability. The cycling performance of washing oil and the TGDE composite solvent (25 wt.%) is shown in Figure 5. The results indicate that the absorption efficiency of pure washing oil decreased to 90.7% after 9 cycles, while the TGDE composite solvent maintained an absorption efficiency of 93.5% after 9 cycles and remained stable at 93.4% even after 12 cycles, demonstrating excellent cycling stability. Furthermore, mass loss analysis revealed that the mass loss of the TGDE composite solvent after 12 cycles was only 2.4%, significantly lower than the 3.4% loss observed for pure washing oil after 9 cycles. These results suggest that the molecular interactions between TGDE and washing oil not only enhance absorption efficiency but also effectively reduce solvent loss by improving thermal stability and reducing volatility.

3.3. Quantum Chemical Calculation Analysis

3.3.1. ESP and σ-Profile Analysis

Molecular surface electrostatic potential (ESP) analysis and σ-profile function analysis were employed to reveal the characteristics of molecular surface charge distribution and polarity features, respectively. As shown in Figure S1, benzene and the components of washing oil solvent exhibit similar ESP distribution patterns; the negative potential regions (blue) are primarily concentrated at the center of the benzene ring, while the positive potential regions (red) are distributed around the hydrogen atoms, demonstrating a typical charge distribution mode of non-polar aromatic molecules. The σ-profile analysis (Figure 6b) further supports this conclusion. The peaks of both benzene and washing oil components are located within the non-polar region (−0.0082 e/Ų < σ < 0.0082 e/Ų), with no observable peaks in the hydrogen bond donor (HBD, σ < −0.0082 e/Ų) or hydrogen bond acceptor (HBA, σ > 0.0082 e/Ų) regions [51]. In contrast, the ESP distribution of the composite solvents (Figure S2) shows that the negative potential regions are mainly concentrated around oxygen atoms. This suggests the presence of electrostatic attraction between the hydrogen atoms of benzene molecules and the oxygen atoms in the composite solvent molecules. Meanwhile, ESP analysis indicates that the positive and negative charge distributions on the molecular surfaces of DBA and TGDE are more uniform compared to those of SUL and DBP, indicating relatively weaker overall polarity.

The σ-profile analysis of the composite solvents (Figure 6a) reveals their significant polar characteristics. All composite components exhibit high-intensity peaks in the HBA region, indicating strong hydrogen bond acceptor capacity, while only a weak peak is observed in the HBD region of TGBE. In addition, a distinct high-intensity sharp peak is also present in the non-polar region for all composite components, with DBA and TGDE showing the highest peak values. This indicates that the composite solvents possess both polar and non-polar interaction sites. According to the similarity–intermiscibility theory, the interactions between washing oil components and benzene molecules are primarily non-polar. In contrast, the composite solvents, with their dual polar and non-polar characteristics, form multiple types of interactions with benzene molecules. This suggests that, compared to the single non-polar interaction mode dominant in washing oil, the composite solvents can more effectively promote the absorption of benzene molecules—consistent with the previously observed enhanced benzene absorption capacity of the composite solvents in absorption experiments.

3.3.2. IGM Analysis

Electrostatic potential (ESP) and σ-profile analysis provided insights into the potential binding sites between benzene and solvent molecules, along with a qualitative assessment of the solvents’ polarity. To further quantitatively characterize the types and relative strengths of weak interactions between benzene and the components of both washing oil and composite solvents, the IGM analysis was employed (further details are provided in Figure S3).

The IGM results indicate that the weak interactions between benzene and washing oil components primarily arise from π–π stacking between benzene rings, which is visualized as light green regions representing van der Waals interactions in Figure 7. In contrast to the washing oil components, π–π interactions between benzene and the composite solvent components are largely absent, except in the case of DBP. Instead, distinct weak interaction regions are identified between hydrogen atoms of benzene and oxygen atoms in the composite solvents. For instance, specific C–H···O hydrogen bonds are observed between benzene and SUL, DBA, and DBP. In the case of TGBE, which contains a hydroxyl group (see Figure S4 for more details), an O–H···π hydrogen bond is formed. Similarly, TGDE, which contains only ether groups, exhibits a C–H···π hydrogen bond interaction with benzene [52,53].

3.3.3. SAPT Energy Analysis

To further evaluate the contribution of different interactions to absorption, the binding energy (ΔΕ_bind) between benzene and each solvent molecule was computed and decomposed into the following four components: electrostatic energy (Ε_ele), induction energy (Ε_ind), dispersion energy (Ε_dis) and exchange energy (Ε_exc). Among these, Ε_ele, Ε_ind and Ε_dis are attractive and favor absorption. The decomposed binding energies for each solvent are summarized in Table S3. Figure 8 illustrates the percentage contribution of each energy component to the total attractive energy (E = E_ele + E_ind + E_dis).

The SAPT results reveal that for both washing oil and composite solvent components, dispersion energy constitutes the dominant contribution to the binding energy, followed by electrostatic energy, while induction energy accounts for only about 10%. Furthermore, it is observed that the electrostatic contribution is slightly higher in composite solvents than in washing oil components, indicating stronger electrostatic interactions between the composite solvents and benzene. This finding is consistent with the ESP analysis. However, the enhancement in electrostatic interactions is modest. Although hydrogen bonding exists between the composite solvents and benzene, the strength of these interactions is relatively low, and dispersion forces still play a significant role even in these weak hydrogen-bonded systems [54]. Thus, the overall increase in electrostatic contribution remains limited.

4. Discussion

The robust linear correlation between ΔGsol and lnK validates solvation free energy as an effective descriptor for screening benzene absorbents. The superior performance of the selected solvents (e.g., TGDE, SUL, DBP) over highly polar protic solvents (e.g., water, alcohols) aligns with the “like dissolves like” principle, attributed to their favorable lipophilicity and electron density distributions. Mechanistically, SAPT and IGM analyses demonstrate that while dispersion forces dominate the washing oil system, TGDE enhances absorption via synergistic dispersion and weak electrostatic interactions, specifically C–H···O hydrogen bonds.

Two primary limitations of this study should be noted. First, a simplified model washing oil was employed to ensure theoretical precision and experimental reproducibility; however, the exclusion of heavy polymeric components typical of industrial fluids may underestimate thermal stability. Second, solvent regeneration was conducted under mild laboratory vacuum conditions, which differ from the harsher high-temperature, slight excess pressure environment of industrial coking. Consequently, while the TGDE composite system exhibits significant potential, its long-term stability and separation efficiency under industrial conditions warrant further pilot-scale validation. Future research will address these engineering challenges through comparative studies using real industrial washing oil and petroleum-based analogs.

5. Conclusions

In this study, a combined theoretical and experimental strategy was successfully established to screen and optimize high-efficiency absorbents for benzene recovery. The results demonstrate that solvation free energy is a reliable descriptor for solvent screening. The optimized TGDE composite solvent exhibited a benzene absorption efficiency of 96.2%, significantly outperforming conventional washing oil. Mechanistic studies indicate that the interactions between washing oil and benzene are dominated by dispersion forces, while the composite components introduce additional weak electrostatic interactions, such as C–H···O hydrogen bonding, which act synergistically to improve absorption performance. The dual polar and non-polar character of the composite solvents enables superior absorption compared to washing oil alone. Finally, future research will focus on a comparative investigation of 1-methylnaphthalene, isoquinoline, quinoline and model components of petroleum-based wash oil to further elucidate how these individual constituents and different oil sources affect the overall properties and stability of the absorbent system. By integrating theoretical calculations with experimental validation, this study proposes a strategy for designing efficient and recyclable VOC absorption systems, providing a solid theoretical and technical foundation for the industrial treatment of benzene-containing waste gas.