Towards Sustainable Industrial Processes: A Preselection Method for Screening Green Solvents in the 1,3-Butadiene Extractive Distillation Process

Abstract

The transition toward sustainable industrial practices has intensified the search for green solvents. However, their true sustainability often remains unverified due to limited and fragmented evaluation criteria. This study addresses this challenge by proposing a holistic, multi-criteria framework that integrates environmental, health, and safety (EHS) considerations alongside technical and economic factors for solvent selection. The adoption of green solvents plays a crucial role in advancing the UN Sustainable Development Goals (SDGs). By implementing these alternatives, industries can reduce pollution, enhance worker safety, and minimize environmental impact, thereby contributing to multiple SDGs. This shift not only supports sustainable economic growth, but also safeguards human and ecological health. Focusing on the 1,3-butadiene extractive distillation process, this research examines the feasibility of replacing hazardous conventional solvents, such as DMF and NMP, with safer and more sustainable alternatives. A structured methodology is employed, incorporating EHS assessments, physicochemical property evaluations, and economic feasibility analyses, with decision-making guided by the Analytic Hierarchy Process. The study identifies propylene carbonate as the most promising alternative, offering high selectivity, favorable physicochemical properties, and cost-effectiveness. Additionally, limitations of traditional solvent evaluation methods, such as reliance on infinite dilution selectivity, are discussed, and process-specific assessments that better reflect industrial conditions are presented.

1. Introduction

1.1. Holistic Green Solvent Assessment

Nowadays, several studies propose and discuss alternative ‘green’ solvents for industrial applications; however, there is often a lack of comprehensive evidence to validate their true sustainability. Many of these alternatives are assessed based on limited metrics, such as low toxicity or biodegradability, without a holistic evaluation of their environmental, economic, and process performance. This oversight underscores the need for rigorous, multi-criteria assessments that incorporate environmental, health, and safety (EHS) criteria alongside analyses of energy efficiency and process feasibility. Such comprehensive evaluations are essential to ensure that these so-called ‘green’ solvents are truly superior alternatives to conventional hazardous solvents [1]. By identifying and implementing these alternatives, industries can reduce pollution, enhance worker safety, and minimize environmental impact, thereby contributing to multiple United Nations Sustainable Development Goals (UN-SDGs), and to more sustainable industrial practices.

Furthermore, many solvents claimed to be sustainable are currently only available in large quantities from fossil carbon sources [1]. The impact of solvents on both the environment and human health obliges a careful balance between their practical utility and the reduction of their potentially harmful effects. This challenge underscores the importance of developing and selecting greener solvents that can achieve the desired industrial outcomes while minimizing ecological and health-related risks, in line with the UN-SDGs.

There has been longstanding concern regarding the use of traditional solvents such as benzene, carbon disulfide, and carbon tetrachloride, due to their well-documented hazards. These substances exhibit severe flammability, neurotoxicity, and explosiveness, and are associated with chronic health effects, including teratogenicity and reproductive toxicity. Beyond their direct risks to human health, these solvents pose significant environmental threats, such as persistence in ecosystems and potential for bioaccumulation. These issues have driven the need for safer, more sustainable alternatives that minimize such risks while maintaining adequate performance in industrial processes [2,3,4].

In striving to minimize the risks associated with hazardous solvents commonly used in industry, two primary pathways have emerged: (1) the development and adoption of alternative greener solvents that are less hazardous, more sustainable, and exhibit reduced environmental impact; and (2) the implementation of innovative process intensification techniques that optimize solvent use, reduce waste, and enhance overall efficiency. These approaches are not mutually exclusive, and can often be combined to achieve more sustainable and economically viable solutions. Efforts to reduce hazardous chemical emissions are increasingly aligned with the broader goals of sustainability and environmental protection. This has spurred significant advancements in both industry and academia to develop alternative solvents and process technologies that prioritize renewable resources. However, it is important to acknowledge that not all bio-based solvents are inherently non-toxic or completely environmentally benign, and their efficacy can vary depending on the application. For example, d-limonene, a naturally derived solvent, is often regarded as having low toxicity for humans. Upon oral intake, d-limonene primarily affects the liver, with no widely recognized hazardous effects beyond unfavorable cutaneous reactions, such as skin irritation or potential allergic responses [5]. However, classifications provided by companies to the European Chemicals Agency (ECHA) for Registration, Evaluation, Authorization, and Restriction of Chemicals (REACH) compliance identify d-limonene as potentially fatal if swallowed or inhaled, and as very toxic to aquatic life, with long-lasting effects [6].

These findings underscore the necessity of carefully evaluating both the environmental and toxicological profiles of bio-based solvents to ensure their sustainability and safety in industrial applications, while also emphasizing the importance of implementing stringent exposure control measures to protect workers and minimize environmental contamination.

Pharmaceutical companies, such as Astra-Zeneca, Glaxo Smith Kline (GSK), Pfizer, and Sanofi, have developed solvent selection guides and lists of ‘good’ and ‘bad’ solvents. These guidelines are generally applicable to commonly used solvents, as they are classified based on their composition, number and type of functional groups, and molecular and physicochemical characteristics. Furthermore, various assessments are conducted to evaluate their impacts on the environment, health, and safety (EHS) [7]. It is important to understand that these solvent classification guidelines are specific to each pharmaceutical industry and are based on their unique criteria. Other industries and different processes may have different criteria, so it is essential to recognize that they can vary. This difference in solvent classification may also be significant between academia and industry, and even among different types of applications, due to their different operational methods and aspects. This distinction is crucial, because the same solvent can be categorized as either ‘green/recommended’, ‘problematic/substitution advisable’, or even ‘hazardous/substitution requested’. Byrne et al., 2016, highlight the challenges in achieving consensus on solvent classification across different solvent selection guidelines, including those from GSK, the American Chemical Society, the Green Chemistry Institute, Astra-Zeneca, Sanofi, and CHEM21 [8]. The study analyzes inconsistencies in numerical scoring, threshold definitions, and assessment criteria, which are often influenced by organizational preferences, rather than standardized regulations. The CHEM21 tool, developed through a collaborative effort, attempts to address these inconsistencies by basing its rankings on the Global Harmonized System and prioritizing the most hazardous characteristic of a solvent, rather than an averaged score. However, even within this approach, some classifications remain debatable. A notable example of these discrepancies is sulfolane. While some guides, like Sanofi’s, initially rated it favorably due to its lower toxicity compared to traditional amide solvents, later assessments, including the CHEM21 guide, revised its ranking due to emerging concerns over its potential reproductive toxicity. This evolving classification underscores the ongoing challenge of aligning solvent evaluation methods across different industries and regulatory frameworks, reinforcing the idea that solvent greenness remains a dynamic and subjective concept [9].

Also, Winterton, 2021, emphasizes that existing guidelines are not designed to guide the discovery of entirely new or unknown solvents, highlighting a gap in the current frameworks [2]. Similarly, Andraos, 2013, underscores the challenges in quantitatively balancing technological efficacy, occupational safety, and environmental impact within the context of green chemistry metrics, pointing out the inherent difficulty of achieving an ideal equilibrium [10]. Tickner et al., 2021, further address the lack of integration between green chemistry principles and the broader search for alternative solvents, stressing the importance of comprehensive evaluations that consider practical applications, technical performance, toxicological implications, and sustainability metrics when assessing proposed substitutes [11].

To sum up, achieving truly sustainable processes demands a nuanced and multifaceted approach to solvent selection and evaluation. While alternative ‘green’ solvents present promising opportunities for reducing environmental impact, their sustainability cannot be assumed without comprehensive, scenario-specific assessments. The interplay between a solvent’s inherent physicochemical properties and its operational context—such as the required temperature/heat, pressure, and process efficiency—highlights the importance of a holistic evaluation framework. Also, toxicological data are often unavailable for newly studied compounds, making it difficult to assess their environmental and health impacts. This highlights why industries often prefer well-known solvents with established safety profiles over newer, less-understood alternatives. By offering a comprehensive framework for solvent selection, this work supports several UN SDGs and fosters the transition to more sustainable industrial practices.

1.2. The 1,3-Butadiene Extractive Distillation Process

1,3-butadiene is a colorless and non-corrosive gas with a mild aromatic or gasoline-like odor, with a normal boiling point of −4.4 °C. Despite being considered a hazardous chemical due to its flammability, self-reactivity, and toxicity, it is a major product of the petrochemical industry. Its simple chemical structure, combined with its low molecular weight and high chemical reactivity, makes it a very useful building block in the synthesis of polymers and copolymers, such as polybutadiene, styrene-butadiene-rubber, styrene-butadiene latex, acrylonitrile-butadiene-styrene resins, and nitrile rubbers, among others [12].

The large-scale utilization of C₄ products began as early as the 1940s, with increasing activity in the 70s. Over time, different techniques were developed to convert less valuable C₄ into products with more added value [12,13,14,15,16,17,18,19].

Nowadays, 1,3-butadiene is preferentially isolated from the C₄ fractions obtained as a co-product of the steam cracking of naphtha and/or gas oil to yield ethylene and propylene.

Table 1. Typical C₄ fraction composition [20].

Component	Composition (wt%)
1,3-Butadiene	49.9
iso-Butene	25.1
1-Butene	9.6
trans-2-Butene	5.5
n-Butane	3.4
cis-2-Butene	3.1
iso-Butane	1.5
Vinylacetylene	1.2
1,2-Butadiene	0.3
Propyne	0.2
1-Butyne	0.2
2-Methyl-1-Butene	0.010
Propane	0.005
Propene	0.005
Propadiene	0.001

shows the typical composition of a C₄ hydrocarbon fraction [20]. The conventional distillation method faces difficulties separating 1,3-butadiene from the C₄ cuts, due to the formation of multiple azeotropes and the presence of components with similar boiling points. Over 98% of global 1,3-butadiene production relies on extractive distillation processes to overcome these difficulties [21].

Extractive distillation involves the addition of a solvent to the distillation process. This solvent, typically distinguished by its high boiling point and miscibility with the target components, is essential for conveniently increasing the value of the relative volatilities of the substances being separated, thus facilitating their separation. The careful selection of an appropriate solvent is therefore critical for effectively modifying the volatilities of substances and preventing the formation of unwanted azeotropes during the separation process [22,23].

Currently, the most commonly used solvents in the 1,3-butadiene extractive distillation process are N, N-dimethylformamide (DMF) and N-methylpyrrolidone (NMP). Acetonitrile, dimethylacetamide, and furfural have also been utilized in the past [24]. Several studies have been conducted to investigate the efficacy of these solvents in 1,3-butadiene purification [20,25,26,27,28,29]. While these solvents are effective in separating 1,3-butadiene from the other C₄ hydrocarbons, it is important to recognize their potential hazards, as they can pose significant risks to both human health and the environment. Some potential risks include skin problems, liver damage, cancer, bioaccumulation, and harm to wildlife habitats and species [30,31]. Hence, there is an urgent need to discover better and safer alternatives, especially given the escalating strictness of REACH regulations. From an industrial perspective, it is crucial not only to identify the most effective solvent for 1,3-butadiene purification, but also to ensure that the chosen solvent meets established industrial process standards and can be seamlessly and readily integrated as a replacement for DMF and NMP. In their 2007 study, Fischer and colleagues from ETH Zurich (Swiss Federal Institute of Technology) raised a pivotal question: ‘What defines a green solvent?’. Their response introduced a now widely adopted two-tiered assessment framework that evaluates solvents based on environmental, health, and safety (EHS) considerations and energy demand—essentially, a streamlined life cycle analysis (LCA). This approach involves assessing the energy required for solvent production and exploring end-of-life recovery options with regard to energy demand. However, it falls short of offering a holistic evaluation of a solvent’s true ‘green’ character. A comprehensive assessment must extend beyond these two criteria to include additional critical dimensions, as mentioned before.

One of the novelties of this study is to address these gaps by introducing a multifaceted approach to solvent screening that encompasses not only EHS assessment, but also compliance with REACH regulations, process performance, and the optimal physicochemical properties necessary for targeted industrial application. Moreover, the framework emphasizes practical considerations, such as market availability and scalability to industrial levels. By considering these factors collectively, this work seeks to facilitate the adoption of less hazardous solvents that not only align with sustainability goals, but also meet the technical and economic demands of industrial applications. Unlike existing studies, which often focus on narrow and specific areas, this work represents a new approach for the broader industrial sector. This holistic perspective is essential for transitioning toward greener solvent systems without compromising process efficiency, scalability, or economic viability, offering a comprehensive solution that has not been previously achieved in published works.

2. Solvent Screening Approach

The methodology employed in this study followed a structured approach to selecting a sustainable solvent for extractive distillation, integrating environmental, technical, and economic considerations. The key steps are outlined below.

2.1. Integrating Environmental, Health, Safety, and Regulatory Criteria

The first step involved a thorough review of a database containing 70 solvents (

Table 2. Solvents under study and REACH restrictions.

Solvents	CAS Number	REACH Restriction	Solvents	CAS Number	REACH Restriction
Acetic Acid	64-19-7	No	N-Formylmorpholine	4394-85-8	No
Acetone	67-64-1	No	Furfural	98-01-01	No
Acetonitrile	75-05-8	No	Heptane	142-82-5	No
N-Amyl Acetate	628-63-7	No	Hexane	110-54-3	No
Anisole	100-66-3	No	Hexylene Glycol	107-41-5	No
Benzyl Alcohol	100-51-6	No	Isobutyl Isobutyrate	97-87-0	No
Benzyl Benzoate	120-51-4	No	Isopentyl Acetate	123-92-2	No
1-Butanol	71-36-3	No	Isopropyl Acetate	108-21-4	No
2-Butanol	78-92-2	No	Isophorone	78-59-1	No
N-Butyl Acetate	123-86-4	No	d-Limonene	5989-27-5	No
tert-Butyl Acetate	540-88-5	No	Methanol	67-56-1	Yes
sec-Butyl Acetate	105-46-4	No	β-Methoxypropionitrile	110-67-8	No
Butyl Benzoate	136-60-7	No	Methyl Acetate	79-20-9	No
γ-Butyrolactone	96-48-0	No	Methyl Cellosolve	109-86-4	Yes
ε-Caprolactone	502-44-3	No	Methyl Cyclohexane	108-87-2	No
Cyclohexane	110-82-7	Yes	Methyl Ethyl Ketone	78-93-3	No
Cyclohexanol	108-93-0	No	Methyl Isobutyl Ketone	108-10-1	No
Cyclohexanone	108-94-1	No	Methyl Oleate	112-62-9	No
Cyclopentyl Methyl Ether	5614-37-9	No	N-Methyl-2-Pyrrolidone	872-50-4	Yes
CyreneTM	53716-82-8	No	Methyltetrahydrofuran	96-47-9	No
Diacetone Alcohol	123-42-2	No	Morpholine	110-91-8	No
Diethyl Carbonate	105-58-8	No	1-Nitropropane	108-03-2	No
Diisobutyl Ketone	108-83-8	No	Oxolane	109-99-9	No
N,N-Dimethyl Acetamide	127-19-5	Yes	Pentanone	107-87-9	No
Dimethyl Carbonate	616-38-6	No	1-Propanol	71-23-8	No
Dimethyl Cyclohexane	590-66-9	Yes	2-Propanol	67-63-0	No
N,N-Dimethyl Formamide	68-12-2	Yes	N-Propyl Acetate	109-60-4	No
Dimethyl Isosorbide	5306-85-4	No	Propylene Carbonate	108-32-7	No
Dimethyl Sulfoxide	67-68-5	No	Sulfolane	126-33-0	No
1,4-Dioxane	123-91-1	Yes	Syringol	91-10-1	No
Dipropylene Glycol	2396-61-4	No	Tetrahydrofurfuryl Alcohol	97-99-4	No
EastmanTM Eeh	1559-35-9	No	Toluene	108-88-3	Yes
Ethyl Acetate	141-78-6	No	γ-Valerolactone	1679-47-6	No
Ethyl Lactate	97-64-3	No	Water	7732-18-5	No
Ethylene Carbonate	96-49-1	No	Xylene	106-42-3	No

). The selection covered a wide range of commonly used solvents from various industries and applications, as well as developing alternatives that have recently received a lot of attention in the industry. Certain solvent categories, such as ionic liquids, deep eutectic solvents, and other emerging alternatives, were not included in this study. The primary reason was the lack of comprehensive industrial implementation and long-term performance data under real process conditions. While these solvents have attracted interest for their potential environmental benefits, challenges related to their high acquisition cost, limited industrial-scale availability, undisclosed toxicity, and recyclability remain unresolved. In particular, although ionic liquids offer promising tunable properties and low volatility, their prohibitively high production costs severely limit their competitiveness in large-scale applications at present. Their high price, combined with challenges in regeneration and handling, makes them impractical for industrial adoption, especially in cost-sensitive sectors such as petrochemical processing [32]. Therefore, this study focuses on solvents that are widely used or have demonstrated industrial feasibility, ensuring a relevant and practical assessment of green alternatives that can be readily implemented in an actual 1,3-butadiene extractive distillation process.

The initial phase involved defining acceptable toxicity and ecotoxicity thresholds to ensure compliance with international safety standards, as mandated by REACH regulations, which impose restrictions on solvent usage (Table 2). This step ensured that only solvents meeting stringent regulatory and safety criteria were considered, aligning the selection process with international standards and sustainability objectives. Thus, a set of weighted environmental, health, and safety (EHS) criteria was implemented, also including the vapor pressure of each solvent as an additional parameter. The weights assigned to each criterion were carefully chosen to reflect the primary objectives of this work (replacement of hazardous solvents in the 1,3-butadiene extractive distillation process): 50% for human health, 25% for environmental impact, 20% for vapor pressure, and 5% for safety considerations. Prioritizing human health, by assigning it the highest weight, directly reflects the core aim of this study: to replace highly hazardous solvents with safer alternatives. Higher-vapor-pressure solvents pose an increased risk of volatilization and contamination, justifying the significant weighting of vapor pressure at 20%, slightly below the environmental impact factor, with a weight of 25%. While safety was assigned the lowest weight of 5%, this decision was firmly rooted in the industrial scope of the study. In industrial environments, stringent safety protocols, robust containment systems, and significant investments in maintenance minimize safety risks associated with solvent use. Additionally, safety considerations are inherently linked to the physicochemical properties of the solvents, which will be comprehensively evaluated later in the analysis.

Despite the diverse perspectives found in the literature from various pharmaceutical laboratories and ECHA data, this classification seeks to integrate insights from a broad range of sources into a cohesive framework [7,8,9,33]. Thus, it recognizes the uncertainties and limitations in the available data, while aiming to present a clear and well-balanced analysis.

2.2. Technical Evaluation of Solvent Performance

To assess the technical viability of each solvent, the study focused on its ability to enhance 1,3-butadiene separation from the C₄ mixture in extractive distillation. The desired separation by extractive distillation can be defined in terms of the separation between 1-butene and 1,3-butadiene, which are the key components that define the separation in the first extractive distillation column. As is well known, separation by conventional distillation is only possible if the relative volatility, α_i,j, defined by Equation (1), has a value other than one [34]:

$${\alpha }_{i,j}={\displaystyle \frac{y_i/x_i}{y_j/x_j}}={\displaystyle \frac{K_i}{K_j}}={\displaystyle \frac{P_i^{sat}{\gamma }_i}{P_j^{sat}{\gamma }_j}}$$

(1)

where i and j refer to the most- and least-volatile key component, respectively; y and x are the vapor and liquid mole fractions, respectively; K is the equilibrium constant for vapor–liquid equilibrium (VLE), which defines the tendency of a component to vaporize; P^sat is the saturation pressure; and $\gamma$ is the liquid phase activity coefficient.

The calculation of the activity coefficient is crucial, as it accounts for deviations from Raoult’s law. Due to non-ideal behavior, phenomena such as phase splitting, pinch points, and azeotropes may occur, which can either facilitate or complicate the separation process [35]. As mentioned by Blass, 1989, the saturation pressure ratio in Equation (1) is minimally impacted by temperature changes; therefore, the solvent should primarily affect the relative volatility by altering the activity coefficient ratio [36]. This led to the conventional definition of the solvent selectivity, S_i,j, as the ratio of the activity coefficients of the key components in the presence of the solvent, S:

$$S_{i,j}={\left({\displaystyle \frac{{\gamma }_i}{{\gamma }_j}}\right)}_S$$

(2)

The selectivity of a solvent varies with its concentration, typically reaching its highest value at infinite dilution of the solutes. Consequently, for a robust comparison of solvent effects, selectivity is typically evaluated under infinite dilution conditions (Equation (3)). This approach is the most frequently used for extractive distillation solvent comparison [37].

$$S_{i,j}^{\infty }={\left({\displaystyle \frac{{\gamma }_i^{\infty }}{{\gamma }_j^{\infty }}}\right)}_S$$

(3)

There are three main reasons for the use of selectivity at infinite dilution: the quick and well-established procedure for the experimental measurement of activity coefficients at infinite dilution [38,39,40]; the existence of data banks for activity coefficients at infinite dilution, such as the Dortmund Data Bank [41]; and the possibility of easily predicting the activity coefficients at infinite dilution by using group contribution methods, such as UNIFAC, or quantum chemistry models, such as COSMO-RS, which are mainly useful for new solvents with no experimental data available. Clearly, there is a drawback in using Equation (3) to calculate the selectivity, since when the solute is at infinite dilution, the other key component that defines the separation is not at infinite dilution, and therefore, the value of selectivity obtained by Equation (3) may not be the most reliable value for screening solvents for extractive distillation in realistic operating conditions.

Process simulators like Aspen Plus^® usually rely on fugacity calculations to determine phase equilibrium, a critical step for accurately simulating chemical processes. Fugacity, a thermodynamic property that accounts for the real behavior of fluids, is used to describe the effective tendency of a component to move towards another phase in a given mixture. This approach ensures that phase equilibrium calculations reflect the non-ideal interactions between components, particularly in complex mixtures. This capability is essential for processes involving non-ideal systems, such as those with polar solvents, azeotropes, or highly interactive components. As noted by Seider et al., 2009, the use of advanced fugacity models significantly enhances the predictive accuracy of phase equilibria in process simulations, supporting the optimization of industrial processes [42]. The calculation of fugacity and fugacity coefficients requires the use of equations of state, which are particularly appropriate to describe the non-ideality of the vapor phase. The equilibrium condition, expressed in terms of fugacity equality, can be represented by Equation (4):

$${\widehat{\varphi }}_i^V·y_i·P=x_i·{\gamma }_i·{\varphi }_i^{sat}·P_i^{sat}·{\theta }_i$$

(4)

where ${\widehat{\varphi }}_i^V$ and ${\varphi }_i^{sat}$ are the vapor fugacity coefficient in the mixture and the saturation fugacity coefficient of component i, respectively; P is the total pressure; and ${\theta }_i$ is the Poynting correction factor for component i.

Therefore, when using commercial simulators like Aspen Plus^®, where calculations take into consideration not only the non-ideality of the liquid phase, but also the non-ideality of the vapor phase, the use of the traditional definition of selectivity at infinite dilution may not be the most appropriate method for solvent screening, as the simulation results may not be consistent with the expected solvent ranking based on Equation (3). Hence, a comparison was carried out between the selectivity calculated at infinite dilution by Equation (3) [43,44,45,46], which will be called method 1; and the selectivity determined at the liquid molar fraction corresponding to the point of maximum separation, which will be called method 2, defined by Equation (5), a calculation that considers the non-ideality of the vapor phase using equations of state.

$${\left(S_{i,j}\right)}_S^{\mathit{max}.sep.}={\left({\displaystyle \frac{{\gamma }_i}{{\gamma }_j}}\right)}_S^{max.sep.}$$

(5)

In this comparison, the activity coefficients were determined using Aspen Plus^® (Version 12.1), employing the UNIFAC model to account for non-idealities in the liquid phase and the Redlich–Kwong equation of state, in order to take into consideration the non-idealities of the vapor phase. Figure 1 illustrates the discrepancies between the two methods, revealing an average relative error of approximately 4%, with deviations reaching up to 25%. More importantly, these differences led to changes in the ranking order of the solvents, with one method prioritizing certain solvents over others. Water was excluded from this evaluation due to the low solubility of hydrocarbons in water, which limits its effectiveness as a solvent. For a comprehensive comparison, DMF and NMP were included alongside the other solvents in the analysis.

Table 3. Selectivity values determined using methods 1 and 2, and the liquid molar fraction of 1-butene corresponding to the maximum separation value.

Solvent	x1-Butene	Method 1 (Equation (3))	Method 2 (Equation (5))
Acetic Acid	0.6	0.68	0.85
Anisole	0.5	1.11	1.09
Benzyl Alcohol	0.5	1.11	1.08
1-Butanol	0.6	0.93	0.90
2-Butanol	0.6	0.91	0.90
N-Butyl Acetate	0.5	1.11	1.11
sec-Butyl Acetate	0.5	1.11	1.11
Butyl Benzoate	0.5	1.03	1.07
ε-Caprolactone	0.5	1.09	1.09
Cyclohexanol	0.6	0.97	0.92
Cyclohexanone	0.5	1.14	1.13
Cyclopentyl Methyl Ether	0.6	0.99	0.97
CyreneTM	0.5	1.36	1.28
Diacetone Alcohol	0.5	1.16	1.12
Diethyl Carbonate	0.5	1.13	1.17
Diisobutyl Ketone	0.5	1.05	1.05
Dimethyl Carbonate	0.5	1.35	1.40
N,N-Dimethyl Formamide	0.5	1.37	1.32
Dimethyl Isosorbide	0.4	1.32	1.48
Dimethyl Sulfoxide	0.4	1.61	1.56
Dipropylene Glycol	0.6	1.03	0.93
Ethyl Acetate	0.5	1.20	1.18
Ethyl Lactate	0.5	1.01	1.04
N-Formylmorpholine	0.5	1.22	1.13
Isobutyl Isobutyrate	0.4	1.00	1.02
Isopentyl Acetate	0.5	1.08	1.08
Isopropyl Acetate	0.5	1.15	1.14
Methyl Ethyl Ketone	0.5	1.24	1.21
N-Methyl-2-Pyrrolidone	0.5	1.40	1.45
β-Methoxypropionitrile	0.5	1.37	1.37
Pentanone	0.5	1.19	1.14
1-Propanol	0.6	0.89	0.88
2-Propanol	0.6	0.88	0.88
N-Propyl Acetate	0.5	1.15	1.14
Propylene Carbonate	0.5	1.11	1.30
Syringol	0.5	1.25	1.30
γ-Valerolactone	0.5	1.12	1.13

presents the selectivity values determined at the liquid molar fraction of 1-butene corresponding to the point of maximum separation, while Figure 1 provides a graphical representation of these results. Method 2 provides a more rigorous assessment of thermodynamic effects, making it the preferred approach over traditional methods for validating solvent selection in process simulations.

It should be noted that selectivity is not the only parameter used in the technical evaluation of solvent performance. Other approaches have been proposed, including the Pierotti–Deal–Derr method, the Parachor method, the Weimer–Prausnitz method, computer-aided molecular design (CAMD) techniques, and methods based on the analysis of residue curve maps or excess Gibbs energy, among others. The use of Hansen solubility parameters has also been suggested, although this method is more suitable for selecting solvents for extraction processes [47,48,49].

2.3. Evaluation of Physicochemical Properties

It is essential to assess the suitability of solvents’ physicochemical properties for the intended process. Key properties evaluated in this study include degradation temperature, flash point, melting point, auto-ignition temperature, explosive range, and vapor pressure. Consideration was given to selecting solvents with lower vapor pressure to minimize solvent losses and reduce atmospheric contamination. Additionally, the thermal stability of solvents was carefully evaluated, particularly in relation to their normal boiling points. If a solvent’s degradation temperature is lower or close to its boiling point, its viability for use in extractive distillation may be compromised. This limitation arises from the inability to effectively purify the solvent under normal pressure conditions, as it degrades before reaching its boiling point, potentially leading to an economically challenging process. For this reason, only solvents with a degradation temperature at least 50 °C above their normal boiling point were considered viable. To address this, careful assessment of a solvent’s thermal stability is essential, particularly for solvent recovery strategies that may require a vacuum environment to enable effective purification while preserving the solvent’s integrity. It is also imperative to thoroughly look at the flash and melting point of the solvents. A solvent with a low flash point presents significant safety hazards, as it may readily ignite under standard operating conditions. In this study, a minimum flash point of 40 °C was set as a safety threshold to reduce fire risks in industrial settings. On the other hand, a high melting point can lead to solvent solidification at lower temperatures, posing a particular risk during colder winter days. Such solidification can obstruct flow, potentially cause equipment damage, and result in process inefficiencies. To ensure smooth operation, solvents with a melting point below 0 °C were prioritized, as higher melting points could lead to solidification issues under standard industrial conditions.

A detailed analysis of fluid mechanical properties, particularly those affecting pump performance, has been deferred to a subsequent phase of this research. In this next phase, the actual plant operations will be simulated to thoroughly evaluate the new solvents’ performance, encompassing their influence on pump efficiency, flow rates, operational stability, and overall process dynamics. This comprehensive approach ensures that the selected solvents meet both physicochemical and operational requirements for industrial application.

2.4. Economic and Market Availability Assessment

Finally, the last step should be to ensure that the solvent can readily be purchased at a comparatively low cost for the desired industrial scale, and is readily available in a sufficient quantity, in order for it to be easy for companies to switch to the new solvent. The average solvent budgets were acquired through direct engagement with the manufacturing company and distributor firms, assisted by the CheMondis platform. All prices were estimated at industrial scale, around 100 metric tons.

2.5. Multi-Criteria Decision-Making Using AHP

To provide a structured and objective solvent selection process, a systematic multi-criteria decision-making approach was implemented. Initially, key parameters—including solvent technical performance (TP), physicochemical properties (PPs; these include decomposition temperature (DT), flash point (FP), melting point (MP), normal auto-ignition temperature (A-IP), and explosiveness range (ER)), economic feasibility (EF), and market availability and purchase cost (MAPC)—were identified as critical factors. These parameters were then quantitatively assessed and integrated using the Analytic Hierarchy Process (AHP), a well-established methodology that facilitates the weighting and prioritization of each criterion (

Table 4. AHP pairwise judgment matrix for key parameters.

	TP	DT	FP	MP	A-IP	ER	MAPC	${\mathsf{\Pi }}_{\mathit{i}}$	${\mathit{P}}_{\mathit{i}}$	${\mathit{p}}_{\mathit{i}}$
TP	1	5	7	6	8	8	1	1.3 × 104	3.89	35%
DT	1/5	1	7	6	6	6	1/4	7.6 × 101	1.86	17%
FP	1/7	1/7	1	1/2	3	3	1/6	1.5 × 10−2	0.55	5%
MP	1/6	1/6	2	1	3	3	1/2	1.0 × 10−1	0.72	6%
A-IP	1/8	1/6	1/3	1/3	1	6	1/7	1.9 × 10−3	0.41	4%
ER	1/8	1/6	1/3	1/3	1/6	1	1/7	5.5 × 10−5	0.25	2%
MAPC	1	4	6	5	7	7	1	5.9 × 103	3.46	31%
SUM	2.76	10.64	23.67	19.17	28.17	34.00	2.90	-	11.13	100%

) [50]. In Table 4, ${\mathsf{\Pi }}_i$ represents the product of the elements in each row; $P_i$ is calculated as the geometric mean of the elements in each row, and provides a normalized measure of the parameter’s relative importance based on the pairwise comparison judgments; and $p_i$ is the normalized weight of the parameter, calculated as the ratio of $P_i$ to the sum of all $P_i$ values, and it illustrates the relative preference or importance of each key parameter in the overall decision-making process.

The AHP was specifically applied to the top 10 solvent candidates that exhibited the best EHS scores, as illustrated in Figure 2. The evaluation considered the criteria detailed in the previous Section 2.2, Section 2.3 and Section 2.4, ensuring that only the most promising candidates, based on their overall EHS profile, were selected for a more in-depth comparison. This structured evaluation enabled a comprehensive comparison of potential solvents, leading to a final projection of the most suitable candidates for industrial application (see the Supplementary Materials for more details).

3. Results

The top 10 solvents identified based on the EHS criteria are shown in Figure 2, ranked on a scale of 0 to 10, with 10 being the most-preferred solvent and 0 the least-preferred. On the right side, the total score is displayed, incorporating the weighted EHS criteria outlined in Section 2.1. Propylene Carbonate (PC) emerged as the leading candidate, followed by Cyrene^TM (dihydrolevoglucosenone), Dipropylene Glycol, Water, N-Formylmorpholine (NFM), Dimethyl Sulfoxide (DMSO), Diethyl Carbonate (DEC), N-Butyl Acetate, iso-Pentyl Acetate, and Syringol. To evaluate the robustness of the solvent ranking and assess the impact of criterion weight variations, a sensitivity analysis was conducted. This analysis examined how changes in the weighting of EHS factors influenced the final ranking of the top 10 solvent candidates, as shown in Figure 3. The results indicated that while small fluctuations in individual weightings (+/−10%) led to slight positional shifts among certain candidates, Propylene Carbonate (PC) consistently remained the top-ranked solvent across all the tested scenarios. Overall, most solvents maintained their positions, with the exception of water, which exhibited greater sensitivity to weight adjustments, due to its low vapor pressure score. This finding suggests that the ranking of water is particularly dependent on the weighting assigned to vapor pressure. Despite this variation, the sensitivity analysis confirmed the robustness of the AHP-based ranking approach, while highlighting the influence of specific criteria on solvent positioning.

These solvents were subjected to a detailed assessment of their technical performance, physicochemical properties, and purchase cost and market availability. Selectivity was identified as a critical parameter for the separation of 1,3-butadiene from 1-butene. In summary, while the traditional method at infinite dilution (method 1) yielded results similar to those of method 2, relying solely on infinite dilution conditions can be misleading in identifying the optimal solvent power, as it disregards the concentration-dependent interactions present in real operating conditions. Hence, in solvent screening, selectivity should be evaluated at the point of maximum separation using thermodynamic equilibrium principles, with fugacity coefficients incorporated to enhance the reliability of the predictions. This approach ensures more realistic assessments by leveraging process simulators, such as Aspen Plus^®, which use fugacity to account for non-idealities in phase behavior.

The analysis shown in Figure 1 demonstrates the existence of selectivity values below unity, which indicate a solvent preference for 1-butene, making these solvents unsuitable for the desired separation due to the fact that this would require significant process alterations in the actual plant. As a result, those solvents were excluded from further consideration.

The physicochemical properties described earlier were evaluated to identify the 10 best solvent candidates, to ensure that the selected solvents were compatible with typical process operating conditions (see

Table 5. Best 10 solvents’ physicochemical properties.

Solvent	Decomposition Point (°C)	Flash Point (°C)	Melting Point (°C)	Normal Auto-Ignition Point (°C)	Explosiveness Range (% vol.)	Vapor Pressure at 20 °C (mbar)
N-Butyl Acetate	no data available	27	−90	415	1.7–7.6	11.2
Isopentyl acetate	>700	33	−78	360	1.0–10.0	6.0
DEC	>320	33	−43	445	1.4–11.0	11.5
Water	not applicable	41	−63	390	1.9–15.3	23.4
DMF	>350	57	−61	435	2.2–16.0	3.8
DMSO	>190	87	18	301	2.6–42.0	0.6
NMP	>350	91	−24	251	1.3–9.5	0.32
CyreneTM	>200	108	<−20	296	no data available	0.03
NFM	>400	118	20	345	1.2–8.2	0.03
PC	>200	116	−49	430	4.7–21.0	0.04
Dipropylene Glycol	not relevant	138	−20	350	2.9–12.6	<0.01
Syringol	no data available	140	55	no data available	no data available	0.00

). NFM and DMSO were excluded due to their high melting points (20 °C and 18 °C, respectively), which pose significant risks of solidification and equipment damage during colder days. Similarly, syringol was eliminated due to its limited industrial availability and high acquisition cost, rendering it economically unfeasible.

Economic feasibility, evaluated through CIF (Cost, Insurance, and Freight) values for a procurement scale of 100 metric tons, played a crucial role in the final selection. All solvents were required to be of technical grade, with a minimum purity of 99.70% by weight. Figure 4 compares the acquisition costs (CIF values) of the top solvent candidates, highlighting propylene carbonate as the most cost-effective option, priced 21% lower than DMF. While Cyrene^TM demonstrated strong environmental credentials, its significantly higher cost and limited market availability are key challenges. Nonetheless, its potential for use as a green solvent derived from renewable biomass may warrant further future investigation. Similarly, dipropylene glycol, despite its favorable EHS profile, was excluded from further analysis due to its selectivity value being close to unity, indicating that effective separation of 1,3-butadiene from 1-butene is not feasible.

After excluding the less-viable candidates, the AHP was applied to rank the five most promising solvents: PC, Cyrene^TM, DEC, N-Butyl Acetate, and iso-Pentyl Acetate. The AHP scheme provided in Figure 5 outlines a structured framework for decision-making, aimed at identifying the best replacement solvent. This process is divided into four hierarchical levels. At the top, the primary objective is clearly defined as finding the most suitable replacement solvent. Below this, the hierarchy is organized into criteria that influence the decision. These are divided into three categories: technical performance, physicochemical properties, and economic and market viability. After making judgements by comparing each subcriterion responsible for classifying the level 4 alternatives (solvent candidates), selectivity received the highest preference value at 35%, followed by market availability and purchase cost at 31%, decomposition temperature at 17%, and lower weights for the other subcriteria. From these comparisons, priority weights are calculated using mathematical methods, such as eigenvector analysis [46]. A detailed explanation of this analysis is provided in the Supplementary Materials. To ensure the reliability of these weights, the consistency of the pairwise comparisons is checked by calculating the consistency ratio (CR). If the CR indicates inconsistencies, adjustments are made. The calculated consistency ratio (CR) for the subcriteria weights was determined to be 0.086, which is below the commonly accepted threshold of 0.10. This indicates that the pairwise comparisons made during the AHP are acceptably consistent. A CR value of 0.086 suggests an 8.6% probability that the decision-maker’s judgments could have been made at random, which is within the allowable limit. Therefore, the final estimates derived from this process can be considered reliable and valid for decision-making purposes.

Ergo, the final preference scores were as follows: Propylene Carbonate (41%), Cyrene^TM (28%), Isopentyl Acetate (11%), Diethyl Carbonate (10%), and N-Butyl Acetate (10%). The superior performance of propylene carbonate is attributed to its high selectivity for 1,3-butadiene, favorable physicochemical properties, and cost-effectiveness. As outlined in the work of Bruce L.G. et al., 1989, this process provides a systematic approach to multi-criteria decision-making, balancing both qualitative judgments and quantitative analysis. It is particularly useful in contexts where decisions involve complex trade-offs among multiple criteria, as in the case of selecting a replacement solvent [50].

Hence, this work provides a rapid and pragmatic preliminary solvent selection approach for industrial applications; however, it needs subsequent validation through an experimental program to confirm the accuracy of the predicted values. This step is essential before making the decision to implement a solvent swap in an actual industrial process.

4. Conclusions

This study demonstrates the critical role of a comprehensive, multi-criteria framework in the identification and evaluation of sustainable solvents for industrial applications. By integrating EHS criteria, physicochemical properties, and economic considerations using AHP methodology, the research highlights the importance of balancing environmental benefits with operability, economic feasibility, and user judgment. Propylene carbonate emerged as the most suitable replacement solvent for the 1,3-butadiene extractive distillation process, offering an optimal combination of selectivity, cost-effectiveness, and compatibility with industrial standards. The second optimal candidate, Cyrene^TM, displayed a good selectivity value; however, its high acquisition cost and limited market availability underscore the persistence of challenges in transitioning to emerging green solvents. Beyond these top two candidates, DEC exhibited moderate selectivity, favorable EHS properties, and reasonable economic feasibility. However, its acquisition cost remains higher than DMF, indicating that a full process evaluation remains necessary to determine its overall viability. N-Butyl Acetate and iso-Pentyl Acetate exhibited low technical performance scores and unfavorable physicochemical properties, including high vapor pressure and low flash points, making them unsuitable candidates for this application.

This study also emphasizes the limitations of conventional selectivity assessments at infinite dilution, and advocates for process-specific evaluations that account for closer-to-actual operational conditions. Hence, this green solvent screening methodology has been demonstrated to be a quicker and more streamlined approach to identifying environmentally friendly solvents that are feasible for real industrial processes at a large scale and have competitive pricing. The authors recommend that the next phase of this research involve a comprehensive evaluation of the best solvents. This should include simulations under real operating conditions, coupled with a detailed economic analysis and comparison against the traditional solvent currently used in the 1,3-butadiene extractive distillation process. Additionally, conducting an LCA will provide valuable insights into the environmental impacts of these alternative solvents throughout their entire life cycle. Such an assessment will be critical in determining the feasibility and potential advantages of these alternative solvents in a commercial setting, ensuring that environmental, economic, and operational factors are all thoroughly considered.

By providing a comprehensive, sustainability-driven framework for solvent selection, this work promotes the adoption of greener alternatives in industrial applications that balance environmental benefits with technical and economic feasibility, thus contributing to a healthier environment, reduced chemical pollution, and responsible resource use, leading towards more sustainable industrial processes. However, before deciding to implement solvent replacement in an actual industrial process, it is essential to validate the predicted values through an experimental program to ensure their accuracy.