Highly efficient sulfonated styrene resins and optimization for the ethanolysis of furfuryl alcohol to ethyl levulinate

Abstract

Different types of sulfonated styrene resins (SSRs) were applied to the ethanolysis of furfuryl alcohol (FA) to ethyl levulinate (EL). The catalytic performance results showed that the SSR-2 catalyst prepared through sulfonation at 60 °C for 5 h demonstrated the best activity with 100% FA conversion and 97% EL selectivity under the reaction conditions of 126 °C, 2.2 h, and 0.12 g catalyst dosage, which was dramatically superior to the previous results. Notably, the SSR-2 catalyst maintained excellent stability over five cycles, which was attributed to the synergistic effect of the stable, three-dimensional crosslinked network structure of the styrene-based resins and the appropriate SO₃H groups, effectively alleviating the leaching problem of the active groups. Innovatively, Response Surface Methodology (RSM) was used to optimize the reaction conditions of FA ethanolysis over the SSR-2 catalyst and the catalyst dosage was identified as the most crucial factor for the production of EL, focusing on the effects of reaction temperature, reaction time and catalyst dosage on EL selectivity. Moreover, the predicted EL selectivity of 97.4% obtained from RSM matched well with the experimental value of 97% under the optimal reaction conditions, paving the way for catalyst engineering in EL production from FA ethanolysis.

---

1. Introduction

Over the past few decades, the challenges of sustaining the supply of conventional fossil fuels and the accompanying serious environmental concerns have sparked a surge of interest in renewable bioenergy.^1,2 Distinguished from coal, petroleum, and other conventional sources, lignocellulose containing hemicellulose, cellulose and lignin as the prime components is regarded as the most promising renewable bioresource for the production of platform chemicals due to its key merits of wide availability, relatively low cost, abundant carbon content, and eco-friendly nature.^3,4 Notably, furfural can be obtained through acid hydrolysis of renewable cellulose/hemicellulose⁵ and can be further converted into many high-value-added chemicals, such as furfuryl alcohol (FA),⁶ 2-methylfuran,⁷ 5-hydroxymethyl-furfural (5-HMF),⁸ levulinic acid (LA),⁹ ethyl levulinate (EL),¹⁰ and γ-valerolactone,¹¹ and can effectively mitigate the reliance on fossil resources, and thus will have broad market prospects (Scheme 1).

Scheme 1 Composition and conversion of renewable lignocellulose to high-value-added chemicals.

Among them, alkyl levulinates were identified as one of the top ten biorefinery candidates by the U.S. Department of Energy owing to their specific physicochemical properties,^12,13 which provoked a boom in the research of biomass platform compounds. In particular, EL is regarded as a high-value and prospective bio-based chemical. For example, the unique fragrance of EL, derived from the carbonyl and ester groups, endows it with functions in spices and food flavorings.¹⁴ In addition to that, EL can also be employed in a variety of fields, such as solvents,¹⁵ pharmaceuticals, additives,¹⁶ plasticizers,^17,18 and so on. More significantly, EL shows great potential as a bio-based diesel additive in the future, owing to its properties of high lubricity, and stability of pour point, boiling point and cloud point, which can improve the cold flow properties of biodiesel¹⁹ and significantly reduce hydrocarbons and CO emissions.²⁰

Nowadays, EL can be produced by the esterification of LA or through the ethanolysis of FA, 5-HMF or biomass.^2,21 Nevertheless, the synthesis of EL from biomass faces challenges such as longer reaction route or higher cost of isolation and purification, which severely hampers its industrial application.²² Despite the shorter reaction path from LA or 5-HMF to EL, their separation still faces significant challenges due to the low conversion efficiency.^23,24 By contrast, ethanolysis of FA to EL is considered a more promising route, mainly because it involves a simpler process and also addresses the problem of oversupply of cheap furfural.^25,26

Acidic catalysts are essential for the ethanolysis of FA to produce EL^27,28 and can be divided into homogeneous and heterogeneous catalysts. Homogeneous catalysts suffer from corrosion, pollution and difficulties in isolation and purification, which will inevitably limit their commercial applications.²⁹ By contrast, heterogeneous catalysts have attracted more attention owing to their non-corrosive nature to equipment, easier separation and regeneration. Heterogeneous catalysts have been extensively researched for the ethanolysis of FA to EL in the reported literature, as listed in Table 1, including zeolites,^30,31 metal salts,³² sulfated metal oxides,³³ and heteropoly acids,³⁴ sulfated carbon materials,^35,36 and ion-exchange resins.^37,38 It is noted that heteropoly acids, zeolites, metal salts and sulfated metal oxide catalysts face challenges such as laborious catalyst preparation and difficult regeneration, while the activity and stability of sulfated carbon catalysts still need to be further enhanced^2,35 owing to the easy shedding of the acidic groups. Among the heterogeneous acid catalysts, ion-exchange resins are preferred and have been proven to be effective in etherification^39,40 and esterification reactions involving alcohols^41,42 to produce biofuels. Besides, styrene resins have a stable structure owing to the strong bonding and the sulfonic acid groups (SO₃H groups), which are favorable for the enhancement of catalytic activity and stability.^39,40 Consequently, these resins are expected to be an efficient catalyst in the ethanolysis of FA to EL.

Table 1 Catalytic performance of FA ethanolysis to EL over different heterogeneous acid catalysts

Catalysts	Catalyst dosage (g)	Reaction conditions	C FA (%)	Y EL (%)	Ref.
NB-H-ZSM-5	0.45	120 °C, 6 h	100	64	30
Fe/USY	3	130 °C, 4 h	98	90	31
Al2(SO4)3	0.07	150 °C, 15 h	100	79	32
S-FCM	0.05	150 °C, 6 h	100	69	35
SFCB-4	0.1	150 °C, 3 h	100	81	36
SSR-2	0.1	120 °C, 2 h	100	94	This work
SSR-2	0.12	126 °C, 2.2 h	100	97	This work

---

RSM offers advantages of model visualization, high efficiency and low cost, making it an effective tool for rapidly identifying optimal conditions.⁴³ Currently, RSM is commonly employed for esterification reactions, such as the esterification of rubber seed oil,⁴⁴ the esterification of acidified oil with ethanol⁴⁵ and the conversion of oil palm biomass to EL.⁴⁶ Both Jisieike⁴⁴ and Ma⁴⁵ obtained high catalytic performance via applying RSM to evaluate the optimal molar ratio, catalyst loading and time, while Tiong⁴⁶ additionally predicted the influence of reaction temperature on performance. However, there have been few efforts devoted to employing RSM to optimize the reaction conditions for the ethanolysis of FA to EL and elucidating the reciprocal effects on the selectivity of EL. In this paper, a variety of sulfonated styrene resins (SSRs) are prepared and the reaction conditions and the reciprocal influence of the resin catalyst are investigated and optimized through RSM.

2. Experimental section

2.1 Materials

FA (98%), EL (99%), ethanol (99%), 2-ethoxymethylfuran (2-EMF, 98%), n-dodecane (99%), Amberlyst-15 and Amberlite 732 were all provided by Aladdin Chemistry Co., Ltd. Hydrochloric acid and concentrated sulfuric acid were purchased from Xilong Scientific Co., Ltd. Carbonized resin and acid high temperature resistant resin were supplied by Kairui Environmental Protection Technology Co., Ltd. NaCl and NaOH were obtained from Kemiou Chemical Reagent Co., Ltd. and Guangfu Technology Development Co., Ltd., respectively.

2.2 Preparation and characterization of the catalysts

The purchased cation exchange resins were industrial products, which should be pre-treated and sulfonated to obtain the required catalyst. The detailed preparation and characterization of the SSR catalysts are presented in the SI.

2.3 Catalytic evaluation

The catalytic performance of the SSR catalysts was assessed in a stainless-steel reaction kettle. For each experiment, moderate reactants (FA and ethanol) and the catalyst were filled, and then the reaction kettle was heated to the desired temperature at a stirring speed of 900 rpm. Afterwards, the catalyst was separated by centrifugation through a high-speed centrifuge (TGL-15B, Shanghai Anting Scientific Instrument Factory) at a speed of 8000 rpm. After each run, the products were analyzed via a GC-950 instrument (Shanghai Haixin Chromatography Co., Ltd.) equipped with a PEG-20M capillary column (Beijing Zhongyi Yusheng Technology Co., Ltd.). The operating parameters and peak detection time of the GC-950 instrument are summarized in the SI. The catalytic performance could be determined according to the following equation.

Y_EL = C_FA × S_EL × 100% (4)

wherein n_FA,0 is the original moles of FA, n_ethanol,1 and n_ethanol,2 are the original and moles of ethanol, and n_FA, n_EL and n_2-EMF are the remaining moles of FA, EL and 2-EMF after the reaction.

2.4 Response surface methodology

Response Surface Methodology (RSM) is a statistical method of solving multivariate problems through a logical approach and accessing appropriate data through experiments. The optimal parameters and the functional relationship between the variables and response value were obtained by the regression equation.^47,48 The Box–Behnken design was applied for the design and analysis of the experimental data using Design Expert Version 8.0.6 software (State-Ease, Inc., USA). In this work, the independent variables included reaction temperature (X₁, ranging from 100 to 140 °C), reaction time (X₂, ranging from 1 to 3 h), and catalyst dosage (X₃, ranging from 0.05 to 0.15 g), while the response value was EL selectivity (Y). Furthermore, the experimental range of every selected variable was determined by the data range obtained from previous experiments,² as listed in Table 2.

Table 2 Selected independent variables in coded form used in Box–Behnken design

Independent variables	Symbol	Range and code levels
		−1	0	+1
Reaction temperature (°C)	X 1	100	120	140
Reaction time (h)	X 2	1	2	3
Catalyst dosage (g)	X 3	0.05	0.1	0.15

---

Regression analysis was conducted using eqn (1) to predict the optimum value.

where φ₀, φ_i, φ_ii, and φ_ij are the coefficients of regression (φ₀ is the constant term, φ_i is the linear term, φ_ii is the squared term for variable i, and φ_ij is the interaction term between variable i and j), and Y is the predicted response value. n is the total number of variables, and p is a random error.

3. Results and discussion

3.1 Characterization results

3.1.1 Thermal stability of the SSRs. The TGA curves of the four SSR catalysts are shown in Fig. 1. Obviously, the SSR-1 catalyst displayed a minor decline up to 700 °C (only 4.78%), demonstrating that the structure of the styrene resin was more stable after carbonization. By contrast, the TGA curves of the other SSR catalysts presented three weight loss processes as follows. Specifically, the first weight loss (around 12.2%) that appeared at 25–250 °C could be ascribed to the desorption of water from the resins.⁴⁹ The second one (around 42.5%), ranging from 250 to 500 °C, could be attributed to the decomposition of the SO₃H groups and other oxygen-containing functional groups,^50,51 suggesting the introduction of stable SO₃H groups to the resins. Finally, the third flat weight loss (500–700 °C) could be attributed to the degradation of the polymers in the resins.^52,53 Thus, it could be concluded that the SSR catalysts had high stability at less than 120 °C.

Fig. 1 TGA curves of the SSR catalysts.

3.1.2 Surface functional groups of the SSRs. The FT-IR spectra of the SSR catalysts are shown in Fig. 2. The absorption peaks at 3435 and 1340 cm⁻¹ were attributed to the stretching vibrations of –OH groups,⁵⁴ while the absorption peaks at around 2930 cm⁻¹ and the bands ranging from 650 to 900 cm⁻¹ were assigned to the stretching vibrations of C–H bonds.⁵⁵ In addition, the absorption peak at around 1723 cm⁻¹ corresponded to the stretching vibrations of C=O bonds,⁵⁵ and the absorption peaks at 1605 and 1472 cm⁻¹ corresponded to the stretching vibrations of C=C bonds.⁵⁶ For the SSR catalysts, the peaks at around 1213 and 1161 cm⁻¹ corresponded to the asymmetric stretching vibration of O=S=O bonds, whereas the peaks at around 1123 and 1015 cm⁻¹ were assigned to the symmetric stretching vibration of O=S=O bonds.⁵⁷ Moreover, absorption peaks at 609 cm⁻¹ and 571 cm⁻¹ also appeared for the SSR catalysts, which corresponded to the stretching vibrations of the OH bonds of the SO₃H group and C–S bonds, respectively. In addition to the above similarities, there were also differences between SSR-1 and the other catalysts. Specifically, the weaker adsorption peaks of SSR-1 at 571, 609 and around 1152 cm⁻¹ demonstrated the incorporation of a lower concentration of SO₃H groups into the carbonized resin. In conclusion, the SSR catalysts have been successfully sulfonated and the typical active sites were the SO₃H groups.^55,57

Fig. 2 FT-IR spectra of the SSR catalysts.

3.1.3 Surface acidity of the SSRs. The results of the elemental analysis are displayed in Table 3. It is obvious that the sulfur content of SSR-1, SSR-2, SSR-3 and SSR-4 was 2.2%, 9.8%, 12.9% and 14.4%, respectively, confirming the incorporation of sulfur element into the four resins via the sulfonation process. Notably, the SSR-1 catalyst possessed a higher proportion of C (87.7%) and lower S (2.2%) than the other SSR catalysts. This can be rationally explained by the enhanced graphitization degree when undergoing the carbonization process, which is unfavorable to the introduction of SO₃H groups into the SSR-1 resin. Compared to the other SSR catalysts, the SSR-2 catalyst possessed a lower proportion of C (41.6%) and a higher proportion of S (9.8%), which would provide an appropriate concentration of SO₃H groups. As listed in Table 3, both the concentration of total acid sites determined by acid–base titration and the concentration of SO₃H groups measured through the sulfur content increased with increasing sulfur content, meaning that the majority of acidic sites over the SSR catalysts were available.^2,36,52

Table 3 Elemental analysis and acidity distributions of the SSR catalysts

Catalysts	C content^a (%)	O content^a (%)	S content^a (%)	SO3H group concentration^a (mmol g^−1)	Total acid site concentration^b (mmol g^−1)
a Based on elemental analysis. b Based on acid–base titration method.
SSR-1	87.71	9.30	2.17	0.68	1.30
SSR-2	41.63	44.27	9.83	3.07	4.10
SSR-3	45.11	36.69	12.92	4.04	4.45
SSR-4	45.14	35.03	14.43	4.51	5.32

---

3.2 FA ethanolysis to EL

3.2.1 Performance evaluation. The catalytic performance in the ethanolysis of FA to EL over the SSR catalysts is displayed in Fig. 3. It was concluded that the SSR catalysts presented different performances and the FA conversion was in the sequence of SSR-2, SSR-3 (100%) > SSR-1 (99.8%) > SSR-4 (85%), while the EL yield followed the order of SSR-2 (94.3%) > SSR-3 (89.7%) > SSR-1 (77.5%) > SSR-4 (30.3%). Notably, the SSR-2 catalyst presented the highest EL yield of 94.3%, which surpassed those of the other SSR catalysts.

Fig. 3 Catalytic performance of the SSR catalysts (reaction conditions: 0.4 g of FA, 0.1 g of catalyst, 9.4 g of ethanol, 120 °C, 2 h).

The SSR-2 catalyst had a high acidity concentration (4.1 mmol g⁻¹) and a moderate pore size (5.51 nm), which enabled efficient mass transfer and suppressed side reactions, thus achieving the highest EL yield (94.3%). The ultra-large pores (35.38 nm) of SSR-3 significantly enhanced diffusion, but the excessive exposure of acid sites caused side reactions, resulting in a lower EL yield (89.7%) than SSR-2. Despite the high surface area (800 m² g⁻¹) of SSR-1 enhancing reactant access, the critically low acid site concentration (1.3 mmol g⁻¹) limited reaction progression. The extreme acidity concentration (5.32 mmol g⁻¹) of SSR-4 confined in undersized pores (1.65 nm) created an “acid trap”, accelerating pore blockage and side reactions and thus decreasing the EL yield to 30.3%. These results indicate that the total acid site concentration and structure (Table S1) of the acid resin catalyst had a critical influence on the ethanolysis of FA.

3.2.2 Effect of the total acid site concentration on the selectivity or yield of EL. The SSR catalysts were synthesized via a sulfonation method with different acid concentrations, and thus exhibit a large variation in catalytic activity. In particular, SSR-2 displayed FA conversion of 100% and the highest EL selectivity of 94.3%, significantly superior to other heterogeneous catalysts (Table 1). For the purpose of elucidating the structure–activity relationship, the selectivity of EL as a function of total acid site concentration (mmol g⁻¹) for the SSR-2 catalyst was determined, as presented in Fig. 4. The EL selectivity revealed a volcanic trend with increasing total acid site concentration and peaked at 4.1 mmol g⁻¹, which was consistent with our previous study.² SSR-1 exhibited a lower EL selectivity of 77.7% attributed to the lower acid site concentration, indicating that adequate acid concentration is essential to the ethanolysis of FA to EL. Nevertheless, when the total acid site concentration exceeded 4.1 mmol g⁻¹, SSR-3 and SSR-4 showed decreased EL selectivity of 89.7% and 35.6%, which was ascribed to the FA polymerization and more side reactions in an excessively acidic environment.^30,37,58 In summary, it was deemed that high acid concentration and a moderate pore size had a crucial influence on the reaction of FA to EL.

Fig. 4 Effect of total acid site concentration on the selectivity or yield of EL over the SSR catalysts (reaction conditions: 0.4 g of FA, 0.1 g of catalyst, 9.4 g of ethanol, 120 °C, 2 h).

3.3 Conditional screening of the SSR-2 catalyst

3.3.1 Effect of the sulfonation conditions and reaction conditions. The screening of the sulfonation conditions over the SSR-2 catalyst is presented in the SI (Fig. S1). Moreover, the influence of the reaction conditions, including reaction temperature, reaction time, catalyst dosage and molar ratio of furfuryl alcohol to ethanol (n_FA/n_EtOH), on the EL selectivity of SSR-2 was probed, and the results are presented in Fig. S2.

When the reaction temperature was gradually increased from 40 to 100 °C, the FA conversion and EL selectivity grew sharply. In contrast, the FA conversion remained 100%, and an apparent increase in EL selectivity from 79.8% to 94.3% was observed when the reaction temperature was increased from 100 to 120 °C, and it then declined to 89.9% as the temperature was further increased to 140 °C. Besides, the influence of reaction time on the EL selectivity was probed, and a volcanic curve was obtained. With the extension of the reaction time from 1 to 3 h, the EL selectivity increased from 90% to 94.3% and then slightly dropped to 94.1%. This phenomenon could be explained by the higher reaction temperature being favorable for FA to produce EL,⁵⁹ while a longer reaction time at 120 °C would not bring about a marked enhancement.

It was noted that the selectivity of EL remarkably increased from 58.2% to 94.3% and then decreased to 84.5% with an increase in catalyst dosage. It could be interpreted that a smaller catalyst dosage was not enough to facilitate the ethanolysis of FA, while excessive catalyst could lead to the formation of undesirable by-products.⁵⁹ Subsequently, the effect of n_FA/n_EtOH on the catalytic performance was screened, and the results demonstrated that the EL selectivity was negatively correlated with n_FA/n_EtOH, indicative of an adverse effect of excessive FA on the generation of EL from FA ethanolysis. As shown in Fig. S2, the EL selectivity reached the peak of 97.1% when the n_FA/n_EtOH was 0.01, yet a larger n_FA/n_EtOH ratio exhibited more practical application value. In view of the above discussion, the reaction temperature, reaction time, catalyst dosage and n_FA/n_EtOH were configured as 120 °C, 2 h, 0.1 g and 0.02, respectively.

3.3.2 Catalyst reusability. As presented in Fig. 5, it can be found that the recovered SSR-2 catalyst maintained outstanding catalytic stability in the reaction of FA to EL. Notably, SSR-2 delivered 100% conversion of FA until the fifth cycle and maintained an EL selectivity of 87.1%, superior to the majority of previous literature reports.^30,60 This could be assigned to the weight loss in the recycling process and the inevitable formation of water.^2,37 In addition, the catalyst deactivation caused by the destruction of the resin skeleton in the esterification reaction was not excluded.⁶¹ In conclusion, the remarkable stability of SSR-2 was ascribed to the stable chemical bonding of the SO₃H groups to the resin, which was consistent with the TGA and FT-IR characterizations.

Fig. 5 Cycling stability of the SSR-2 catalyst (reaction conditions: 0.4 g of FA, 0.1 g of catalyst, 9.4 g of ethanol, 120 °C, 2 h).

3.4 RSM analysis

3.4.1 Model determination. A statistical summary of each model is listed in Table 4. The model fitness was evaluated by the coefficient of determination R², and a higher R² indicated better prediction of the response value. Both the linear model and 2FI model had relatively lower R² (0.4612 and 0.4790, respectively) and adjusted R² values (0.3368 and 0.1664, respectively), and thus neither of them was adequate for the experimental data. The cubic model had a higher R² of 0.9969 and an adjusted R² of 0.9877. However, they were aliased, that is, the impact of every variable became indistinguishable. Accordingly, the quadratic model was chosen to fit the experimental data.

Table 4 Statistical summary for each model

Model	p-Value	R ^2	Adjusted R^2	Suggestion
Linear	0.0398	0.4612	0.3368	Not adequate
2FI	0.95	0.479	0.1664	Not adequate
Quadratic	<0.0001	0.9954	0.9895	Suggested
Cubic	0.6226	0.9969	0.9877	Aliased

---

3.4.2 Analysis of variance. In this work, the RSM model based on Box–Behnken design was applied to the design of the experiments, and the data are listed in Table 5. After conducting seventeen sets of experiments, the quadratic equation was used to predict the relationship between the selected independent variables and response value, as shown in eqn (7). The final model equation was summarized in terms of coded factors.

Y = +93.48 + 5.85X₁ + 7.19X₂ + 13.51X₃ − 2.48X₁X₂ − 1.92X₁X₃ − 3.3X₂X₃ − 9.39X₁² − 11.82X₂² − 16.97X₃² (7)

where Y is the EL selectivity, and X₁, X₂ and X₃ are the reaction temperature, reaction time and catalyst dosage, respectively. Interestingly, it was noted that the coefficient of catalyst dosage was the biggest, and thus, catalyst dosage (X₃) had the strongest effect on EL selectivity. Furthermore, it was shown that the next significant factor was reaction time (X₂), followed by reaction temperature (X₁).

Table 5 Experimental data and predicted values for three process variables at three levels in Box–Behnken design

Run	Experimental variables			S EL (%)
	Reaction temperature (°C)	Reaction time (h)	Catalyst dosage (g)	Experimental data	Predicted values
1	100	3	0.10	74.8	76.1
2	100	1	0.10	56.5	56.8
3	100	2	0.15	77.6	76.7
4	120	2	0.10	90.2	93.5
5	140	3	0.10	83.1	82.8
6	120	2	0.10	94.7	93.5
7	120	2	0.10	93.5	93.5
8	120	3	0.05	62.3	61.7
9	100	2	0.05	46.5	45.8
10	140	2	0.15	83.9	84.6
11	120	2	0.10	94.6	93.5
12	120	1	0.05	40.3	40.7
13	140	2	0.05	60.5	61.4
14	120	2	0.10	94.4	93.5
15	120	1	0.15	73.7	74.3
16	120	3	0.15	82.5	82.1
17	140	1	0.10	74.7	73.4

---

Analysis of variance (ANOVA) was used to assess the regression significance and lack of fit of the quadratic model, and the corresponding results are listed in Table 6. The F-value of 168.84 indicated that the model was highly significant, with only a 0.01% chance that such an F-value could occur owing to random noise. The probability of error values (p-values) being below 0.01 suggested that the model terms were extremely significant.⁶² The lack of fit F-value of 0.65 demonstrated that lack of fit was not closely associated with the pure error, and there is a 62.26% probability that a lack of fit F-value of this magnitude could occur due to random noise. In this work, the p-values for X₁, X₂, X₃, X₁X₂, X₂X₃, X₁², X₂², and X₃² were less than 0.05, suggesting that these model terms were significant.

Table 6 ANOVA results for the response surface quadratic model

Source	Sum of squares	Degrees of freedom	Mean square	F-Value	p-Value (Prob > F)	Characteristics
Model	4635.77	9	515.09	168.84	< 0.0001	Significant
X 1	273.78	1	273.78	89.74	< 0.0001	—
X 2	413.28	1	413.28	135.47	< 0.0001	—
X 3	1460.70	1	1460.70	478.80	< 0.0001	—
X 1 X 2	24.50	1	24.50	8.03	0.0253	—
X 1 X 3	14.82	1	14.82	4.86	0.0633	—
X 2 X 3	43.56	1	43.56	14.28	0.0069	—
X 1 ^2	371.25	1	371.25	121.69	< 0.0001	—
X 2 ^2	587.77	1	587.77	192.66	< 0.0001	—
X 3 ^2	1211.84	1	1211.84	397.22	< 0.0001	—
Residual	21.36	7	3.05	—	—	—
Lack of fit	7.01	3	2.34	0.65	0.6226	Not significant
Pure error	14.35	4	3.59	—	—	—
Cor total	4657.12	16	—	—	—	—

---

In addition, the adequacy and quality of the quadratic equation were statistically verified using R² and adjusted R². As listed in Table 7, the R² of 0.9954 was close to 1, indicative of the high model fitness, which could explain 99.54% of the variability. The predicted R² of 0.9711 was in accordance with the adjusted R² of 0.9895, as their variance was less than 0.2, and the precision value greater than 4 was preferable for measuring the signal-to-noise ratio. The adequate precision ratio of 39.399 signified an adequate signal, indicating that this model could be applied to guide the design space.⁶³ Meanwhile, the low coefficient of variation (C.V.% = 2.31%) indicated that the model was highly accurate and suitable for fitting the analysis. In summary, this model was suitable and can be employed to predict the relationship between EL selectivity and the three independent variables including reaction temperature, reaction time, and catalyst dosage.

Table 7 Fit statistics for the response surface quadratic model

Parameters	Value	Parameters	Value
Standard deviation	1.75	R ^2	0.9954
Mean	75.52	Adjusted R^2	0.9895
C.V.%	2.31	Predicted R^2	0.9711
Press	134.54	Adequate precision	39.3990

---

The residual should conform to a normal distribution in the absence of experimental anomalies. As can be seen in Fig. S3, the points of the normal plot of the residuals lie on or near a straight line, demonstrating that the residuals of this experiment were normally distributed, indicating that the model fitted by RSM was accurate. The points in Fig. S4 displayed a scattered and irregular plot between the residuals and the predicted values of the equation, demonstrating the validity of the regression model. The plots in Fig. S5 illustrated a satisfactory linear correlation between predicted and actual EL selectivity (evaluated from eqn (7)), meaning that the proposed model was valid and can be used for the prediction of actual experiments.

3.4.3 Reciprocal effects between independent variables on EL selectivity. In this work, contour plots and surface plots have been employed to illustrate the reciprocal effects of two independent variables on EL selectivity with the other variable at its zero level. Furthermore, the circular contour plot depicted insignificant interaction, while the elliptical contour plot described the significant interaction between the corresponding variables.⁶⁴

Fig. 6(a) depicts the interaction of reaction temperature and reaction time on EL selectivity. As seen from Fig. 6(a), increasing the reaction temperature or extending the reaction time within a certain range could lead to high selectivity of EL. However, a clear decline in the selectivity of EL was observed under higher temperature and longer time, which could be attributed to the polymerization of FA or decomposition of EL to produce undesired by-products.^2,65 The elliptical contour plot in Fig. 6(a) indicated that the interaction of reaction temperature and reaction time was significant, which was in agreement with the lower p-value (0.0253) of the X₁X₂ term (Table 6).

Fig. 6 Effect of (a) reaction temperature and reaction time; (b) reaction temperature and catalyst dosage; and (c) reaction time and catalyst dosage on EL selectivity.

The interaction of reaction temperature and catalyst dosage on EL selectivity is displayed in Fig. 6(b). It was evident that the EL selectivity slightly increased with rising reaction temperature at low catalyst dosage. However, at higher catalyst dosage, the EL selectivity was enhanced significantly. Thus, the catalyst dosage played a critical role in determining the reaction rate of FA ethanolysis to form EL. Moreover, the interaction between reaction temperature and catalyst dosage (X₁X₃) was not significant due to the observation of a lower F-value of 4.86 and a higher p-value of 0.0633 (p-value >0.05).

Fig. 6(c) presents the interaction of reaction time and catalyst dosage on EL selectivity. By keeping the reaction time at 2 h and increasing the catalyst dosage from 0.05 g to 0.1 g, the EL selectivity was remarkably improved. The results were easy to accept as the moderate increase in active sites could be favorable for the production of EL. In addition, the interaction of reaction time and catalyst dosage was confirmed to be significant due to the elliptical contour curve in Fig. 6(c), which was in agreement with the lower p-value (0.0069) of the reciprocal effect X₂X₃ term.

3.4.4 Optimization of the reaction conditions. The optimal conditions for FA to EL were captured by the software Design Expert Version 8.0.6. Typically, the independent variables, including reaction temperature, reaction time, and catalyst dosage, were ranged from low to high, while the EL selectivity was set to the maximum value. It can be found that 43 solutions were suggested, and the one with the most desirability and maximum EL selectivity was selected as the optimal solution.

In particular, catalyst dosage and reaction time had a greater influence than reaction temperature on the selectivity of EL, owing to the sequence of the linear coefficient of catalyst dosage (13.51) > reaction time (7.19) > reaction temperature (5.85). Consequently, we focused on the effect of catalyst dosage and reaction time on EL selectivity under the optimized conditions, as shown in Fig. 7. The contour plot in Fig. 7(a) illustrates that the optimal conditions were desirable due to the desirability of 1.00. The optimal conditions for promoting EL selectivity were found to be as follows: reaction temperature of 125.58 °C, reaction time of 2.2 h, and catalyst dosage of 0.12 g. To validate the model, three independent experiments were conducted in triplicate, and the results are presented in Table 8. Based on Fig. 7(b), it was predicted that an EL selectivity of 97.4% could be obtained under the above conditions, which was close to the experimental average of 97%. Therefore, the quadratic model was valid to predict the optimal reaction conditions for obtaining the maximum EL selectivity. Moreover, it cannot be denied that even a minor performance improvement can have a substantial impact on industrial applications. Particularly for fine chemicals, including EL, a product purity of 97% or higher is typically required.^14,15 Accordingly, attempts to enhance the catalytic performance by designing efficient catalysts and pursuing high-purity products are significant in large-scale industrial production, and the potential benefits should not be ignored.⁶⁶

Fig. 7 Effect of catalyst dosage and reaction time on EL selectivity under the optimized conditions, keeping the reaction temperature at its zero level: (a) contour plot; (b) 3D response surface plot.

Table 8 Optimum conditions and validation test for obtaining maximum EL selectivity

Parameters	Reaction temperature (°C)	Reaction time (h)	Catalyst dosage (g)	S EL (%)
Predicted	125.6	2.2	0.12	97.4
Experimental	126.0	2.2	0.12	97.0

---

4. Conclusions

Four efficient and stable SSR catalysts were prepared from inexpensive resins by the facile sulfonation method. Among them, the SSR-2 catalyst with high acid site concentration and moderate pore size displayed the highest EL selectivity (97%) and excellent stability until the fifth cycle, which was superior to the majority of previously reported examples. Furthermore, RSM was adopted to further optimize the reaction conditions, and the catalyst dosage was confirmed to be the most significant factor in the EL selectivity, due to the sequence of the linear coefficient of catalyst dosage (13.51) > reaction time (7.19) > reaction temperature (5.85). The optimized EL selectivity (97.4%) obtained using the quadratic model by RSM was surprisingly close to the experimental value (97%). Furthermore, even a minor promotion of the EL selectivity from 94.3% to 97% will effectively cut the costs in separation and purification, as well as reduce energy consumption in processes including distillation and crystallization. This work provides a highly efficient and stable sulfonated resin catalyst to achieve a high-selectivity EL product and demonstrates the validity of RSM for optimizing conditions of FA ethanolysis.

Author contributions

Yanhong Quan: formal analysis, investigation, writing-review & editing, funding acquisition. Yesu Zhang: data curation, methodology, software, writing-original draft. Jun Ren: writing-review & editing, supervision, funding acquisition.

Conflicts of interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Data availability

Supplementary information is available. See DOI: https://doi.org/10.1039/D5RE00296F.

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Acknowledgements

This work was financially supported by a grant from the National Natural Science Foundation of China (22108191, 22179091, 22478277), China Postdoctoral Science Foundation (2022M712335), Natural Science Foundation of Shanxi Province (202403021221063) and Shanxi Provincial Central Government-Guided Local Science and Technology Development Funding Project (YDZJSX2024D026).

References

1. D. Londoño-Pulgarin, G. Cardona-Montoya, J. C. Restrepo and F. Muñoz-Leiva, Fossil or bioenergy? Global fuel market trends, Renewable Sustainable Energy Rev., 2021, 143, 110905.
2. Y. Zhang, Y. Quan and J. Ren, Influence of the surface SO₃H groups on the performance of activated carbon catalyst for ethanolysis of furfuryl alcohol to ethyl levulinate, Mol. Catal., 2024, 565, 114363.
3. B. M. Upton and A. M. Kasko, Strategies for the conversion of lignin to high-value polymeric materials: Review and perspective, Chem. Rev., 2016, 116, 2275–2306.
4. S. Wang, A. Cheng, F. Liu, J. Zhang, T. Xia, X. Zeng, W. Fan and Y. Zhang, Catalytic conversion network for lignocellulosic biomass valorization: a panoramic view, Ind. Chem. Mater., 2023, 1, 188–206.
5. J. Shen, R. Gao, Y. C. He and C. Ma, Efficient synthesis of furfural from waste biomasses by sulfonated crab shell-based solid acid in a sustainable approach, Ind. Crops Prod., 2023, 202, 116989.
6. G. Brufani, M. Campos, C. Lenzi, L. A. S. Nascimento, E. Paone, R. Mazzoni, L. Vaccaro, A. Piazzi and F. Mauriello, Selective transfer hydrogenation of furfural under continuous flow conditions promoted by the recoverable homogeneous Shvo's catalyst, React. Chem. Eng., 2025, 10, 1227–1232.
7. K. Zhou, J. Chen, Y. Cheng, Z. Chen, S. Kang, Z. Cai, Y. Xu and J. Wei, Enhanced catalytic transfer hydrogenation of biomass-based furfural into 2-methylfuran over multifunctional Cu-Re bimetallic catalysts, ACS Sustainable Chem. Eng., 2020, 8, 16624–16636.
8. S. Borgayary, D. Brahma, R. R. Wary, A. Baishya, S. L. Rokhum and P. Kalita, Agro-waste derived sulfonated biochar material as a sustainable heterogeneous catalyst for conversion of fructose to 5-hydroxymethylfurfural, J. Taiwan Inst. Chem. Eng., 2024, 162, 105632.
9. Y. Tian, F. Zhang, J. Wang, L. Cao and Q. Han, A review on solid acid catalysis for sustainable production of levulinic acid and levulinate esters from biomass derivatives, Bioresour. Technol., 2021, 342, 125977.
10. H. Yuan, C. Li, R. Shan, J. Zhang and Y. Chen, Antibiotic residue derived solid acids for ethanolysis of furfuryl alcohol into ethyl levulinate, React. Chem. Eng., 2023, 8, 2738–2745.
11. H. Li, Z. Fang and S. Yang, Direct conversion of sugars and ethyl levulinate into γ-Valerolactone with superparamagnetic acid–base bifunctional ZrFeOx nanocatalysts, ACS Sustainable Chem. Eng., 2016, 4(1), 236–246.
12. J. He, H. Li, Y. Xu and S. Yang, Dual acidic mesoporous KIT silicates enable one-pot production of γ-valerolactone from biomass derivatives via cascade reactions, Renewable Energy, 2020, 146, 359–370.
13. J. J. Bozell and G. R. Petersen, Technology development for the production of biobased products from biorefinery carbohydrates—the US Department of Energy's “top 10” revisited, Green Chem., 2010, 12, 539–554.
14. A. M. Api, D. Belsito, D. Botelho, M. Bruze, G. A. Burton, J. Buschmann, M. L. Dagli, M. Date, W. Dekant, C. Deodhar, M. Francis, A. D. Fryer, L. Jones, K. Joshi, S. La Cava, A. Lapczynski, D. C. Liebler, D. O'Brien, A. Patel, T. M. Penning, G. Ritacco, J. Romine, N. Sadekar, D. Salvito, T. W. Schultz, I. G. Sipes, G. Sullivan, Y. Thakkar, Y. Tokura and S. Tsang, RIFM fragrance ingredient safety assessment, ethyl levulinate, CAS Registry Number 539-88-8, Food Chem. Toxicol., 2019, 127, S48–S54.
15. A. Démolis, N. Essayem and F. Rataboul, Synthesis and applications of alkyl levulinates, ACS Sustainable Chem. Eng., 2014, 2, 1338–1352.
16. M. Li, J. Wei, G. Yan, H. Liu, X. Tang, Y. Sun, X. Zeng, T. Lei and L. Lin, Cascade conversion of furfural to fuel bioadditive ethyl levulinate over bifunctional zirconium-based catalysts, Renewable Energy, 2020, 147, 916–923.
17. G. Rodrigues Gomes, M. C. Breitkreitz and J. C. Pastre, Ash influence on the ethyl levulinate production from sugarcane molasses mediated by taurine hydrogen sulfate, ACS Sustainable Chem. Eng., 2021, 9, 13438–13449.
18. L. Yan, Q. Yao and Y. Fu, Conversion of levulinic acid and alkyl levulinates into biofuels and high-value chemicals, Green Chem., 2017, 19, 5527–5547.
19. H. Joshi, B. R. Moser, J. Toler, W. F. Smith and T. Walker, Ethyl levulinate: A potential bio-based diluent for biodiesel which improves cold flow properties, Biomass Bioenergy, 2011, 35, 3262–3266.
20. T. Lei, Z. Wang, X. Chang, L. Lin, X. Yan, Y. Sun, X. Shi, X. He and J. Zhu, Performance and emission characteristics of a diesel engine running on optimized ethyl levulinate-biodiesel-diesel blends, Energy, 2016, 95, 29–40.
21. E. Ahmad, M. I. Alam, K. K. Pant and M. A. Haider, Catalytic and mechanistic insights into the production of ethyl levulinate from biorenewable feedstocks, Green Chem., 2016, 18, 4804–4823.
22. J. Shan, Q. Wang, H. Hao and H. Guo, Critical review on the synthesis of levulinate esters from biomass-based feedstocks and their application, Ind. Eng. Chem. Res., 2023, 62, 17135–17147.
23. F. D. Pileidis and M. M. Titirici, Levulinic acid biorefineries: New challenges for efficient utilization of biomass, ChemSusChem, 2016, 9, 562–582.
24. A. F. Peixoto, R. Ramos, M. M. Moreira, O. S. G. P. Soares, L. S. Ribeiro, M. F. R. Pereira, C. Delerue-Matos and C. Freire, Production of ethyl levulinate fuel bioadditive from 5-hydroxymethylfurfural over sulfonic acid functionalized biochar catalysts, Fuel, 2021, 303, 121227.
25. A. S. Mamman, J. M. Lee, Y. C. Kim, I. T. Hwang, N. J. Park, Y. K. Hwang, J. S. Chang and J. S. Hwang, Furfural: Hemicellulose/xylose- derived biochemical, Biofuels, Bioprod. Biorefin., 2008, 2, 438–454.
26. X. Li, P. Jia and T. Wang, Furfural: A promising platform compound for sustainable production of C₄ and C₅ chemicals, ACS Catal., 2016, 6, 7621–7640.
27. G. Zhao, L. Hu, Y. Sun, X. Zeng and L. Lin, Conversion of biomass-derived furfuryl alcohol into ethyl levulinate catalyzed by solid acid in ethanol, BioResources, 2014, 9, 2634–2644.
28. J. P. Lange, W. D. van de Graaf and R. J. Haan, Conversion of furfuryl alcohol into ethyl levulinate using solid acid catalysts, ChemSusChem, 2009, 2, 437–441.
29. G. M. González Maldonado, R. S. Assary, J. A. Dumesic and L. A. Curtiss, Acid-catalyzed conversion of furfuryl alcohol to ethyl levulinate in liquid ethanol, Energy Environ. Sci., 2012, 5, 8990–8997.
30. Y. Li, S. Wang, S. Fan, B. Ali, X. Lan and T. Wang, Nano-H-ZSM-5 with short b-axis channels as a highly efficient catalyst for the synthesis of ethyl levulinate from furfuryl alcohol, ACS Sustainable Chem. Eng., 2022, 10, 3808–3816.
31. X. Kong, X. Zhang, C. Han, C. Li, L. Yu and J. Liu, Ethanolysis of biomass based furfuryl alcohol to ethyl levulinate over Fe modified USY catalyst, Mol. Catal., 2017, 443, 186–192.
32. Y. B. Huang, T. Yang, M. C. Zhou, H. Pan and Y. Fu, Microwave-assisted alcoholysis of furfural alcohol into alkyl levulinates catalyzed by metal salts, Green Chem., 2016, 18, 1516–1523.
33. M. S. Tiwari, A. B. Gawade and G. D. Yadav, Magnetically separable sulfated zirconia as highly active acidic catalysts for selective synthesis of ethyl levulinate from furfuryl alcohol, Green Chem., 2017, 19, 963–976.
34. Y. Wang, J. Shi, X. Chen, M. Chen, J. Wang and J. Yao, Ethyl levulinate production from cellulose conversion in ethanol medium over high-efficiency heteropoly acids, Fuel, 2022, 324, 124642.
35. H. Guo, Y. Hirosaki, X. Qi and S. R. Lee, Synthesis of ethyl levulinate over amino-sulfonated functional carbon materials, Renewable Energy, 2020, 157, 951–958.
36. H. Gong, J. Zhang, Q. Li, L. Wei, Y. Hu, Y. Rui and X. Shi, Catalytic conversion of levulinic acid or furfural alcohol into ethyl levulinate using a sulfonic acid-functionalized coffee biochar, Fuel, 2023, 352, 129059.
37. R. Bringué, E. Ramírez, M. Iborra, J. Tejero and F. Cunill, Esterification of furfuryl alcohol to butyl levulinate over ion-exchange resins, Fuel, 2019, 257, 116010.
38. M. A. Tejero, E. Ramírez, C. Fité, J. Tejero and F. Cunill, Erratum: Esterification of levulinic acid with butanol over ion-exchange resins, Appl. Catal., A, 2016, 517, 56–66.
39. V. Trombettoni, D. Lanari, P. Prinsen, R. Luque, A. Marrocchi and L. Vaccaro, Recent advances in sulfonated resin catalysts for efficient biodiesel and bio-derived additives production, Prog. Energy Combust. Sci., 2018, 65, 136–162.
40. E. Ramírez, R. Bringué, C. Fité, M. Iborra, J. Tejero and F. Cunill, Role of ion-exchange resins as catalyst in the reaction-network of transformation of biomass into biofuels, J. Chem. Technol. Biotechnol., 2017, 92, 2775–2786.
41. J. Zhu, W. Jiang, Z. Yuan, J. Lu and J. Ding, Taguchi method for optimizing the cation exchange resin catalyzed esterification of oleic acid with ethanol and adsorption kinetics study, Process Saf. Environ. Prot., 2024, 182, 989–998.
42. V. Russo, C. Rossano, E. Salucci, R. Tesser, T. Salmi and M. Di Serio, Intraparticle diffusion model to determine the intrinsic kinetics of ethyl levulinate synthesis promoted by Amberlyst-15, Chem. Eng. Sci., 2020, 228, 115974.
43. A. Endut, S. H. Y. S. Abdullah, N. H. M. Hanapi, S. H. A. Hamid, F. Lananan, M. K. A. Kamarudin, R. Umar, H. Juahir and H. Khatoon, Optimization of biodiesel production by solid acid catalyst derived from coconut shell via response surface methodology, Int. Biodeterior. Biodegrad., 2017, 124, 250–257.
44. C. F. Jisieike, N. B. Ishola, L. M. Latinwo and E. Betiku, Crude rubber seed oil esterification using a solid catalyst: Optimization by hybrid adaptive neuro-fuzzy inference system and response surface methodology, Energy, 2023, 263, 125734.
45. L. Ma, Y. Han, K. Sun, J. Lu and J. Ding, Kinetic and thermodynamic studies of the esterification of acidified oil catalyzed by sulfonated cation exchange resin, J. Energy Chem., 2015, 24, 456–462.
46. Y. W. Tiong, C. L. Yap, S. Gan and W. S. P. Yap, Optimisation studies on the conversion of oil palm biomass to levulinic acid and ethyl levulinate via indium trichloride-ionic liquids: A response surface methodology approach, Ind. Crops Prod., 2019, 128, 221–234.
47. G. T. Jeong, Valorization of dairy by-products: Optimized synthesis of 5-hydroxymethylfurfural and levulinic acid from lactose and whey, J. Environ. Chem. Eng., 2022, 10, 108413.
48. G. T. Jeong, Valorization of lipid-extracted spent coffee grounds: The synthesis of levulinic acid by thermochemical process, Ind. Crops Prod., 2022, 187, 115410.
49. H. Gong, J. Zhang, Q. Li, L. Wei, Y. Hu, Y. Rui and X. Shi, Catalytic conversion of levulinic acid or furfural alcohol into ethyl levulinate using a sulfonic acid-functionalized coffee biochar, Fuel, 2023, 352, 129059.
50. K. Ravi and A. V. Biradar, Highly active and scalable SO₃H functionalized carbon catalyst synthesized from bagasse for transformation of bio-based platform chemicals into fuel precursors and its in-depth characterization studies, Fuel, 2022, 321, 124008.
51. F. Liu, X. Meng, Y. Zhang, L. Ren, F. Nawaz and F. S. Xiao, Efficient and stable solid acid catalysts synthesized from sulfonation of swelling mesoporous polydivinylbenzenes, J. Catal., 2010, 271, 52–58.
52. Q. Shi, H. Guo, C. Chen, B. Hou, L. Jia and D. Li, An efficient Brønsted acidic polymer resin for the carbonylation of formaldehyde to glycolic acid, React. Kinet., Mech. Catal., 2020, 130, 1027–1042.
53. N. Bothe, F. Döscher, J. Klein and H. Widdecke, Thermal stability of sulphonated styrene-divinylbenzene resins, Polymer, 1979, 20, 850–854.
54. X. Wang, D. Wang, Y. Guo, C. Yang, X. Liu, A. Iqbal, W. Liu, W. Qin, D. Yan and H. Guo, Fluorescent glutathione probe based on MnO2-phenol formaldehyde resin nanocomposite, Biosens. Bioelectron., 2016, 77, 299–305.
55. L. J. Konwar, A. Samikannu, P. Mäki-Arvela, D. Boström and J. P. Mikkola, Lignosulfonate-based macro/mesoporous solid protonic acids for acetalization of glycerol to bio-additives, Appl. Catal., B, 2018, 220, 314–323.
56. W. Qiao, S. Li, G. Guo, S. Han, S. Ren and Y. Ma, Synthesis and characterization of phenol-formaldehyde resin using enzymatic hydrolysis lignin, J. Ind. Eng. Chem., 2015, 21, 1417–1422.
57. Y. Jiang, X. Li, H. Zhao and Z. Hou, Esterification of glycerol with acetic acid over SO₃H-functionalized phenolic resin, Fuel, 2019, 255, 115842.
58. A. O. C. Iroegbu and S. S. Ray, On the chemistry of furfuryl alcohol polymerization: A review, J. Polym. Sci., 2024, 62, 1044–1060.
59. H. Zhang, S. Wang, H. Zhang, L. Cui and F. Cao, P-doped carbon as an efficient metal-free catalyst for alcoholysis of furfural alcohol to ethyl levulinate, Fuel, 2022, 324, 124024.
60. Y. Wang, D. Zhao, K. S. Triantafyllidis, W. Ouyang, R. Luque and C. Len, Microwave-assisted catalytic upgrading of bio-based furfuryl alcohol to alkyl levulinate over commercial non-metal activated carbon, Mol. Catal., 2020, 480, 110630.
61. X. Shao, L. Su, J. Zhang, Z. Tian, N. Zhang, Y. Wang, H. Wang, X. Cui, X. Hou and T. Deng, Green production of phthalic anhydride from biobased furan and maleic anhydride by an acid resin catalyst, ACS Sustainable Chem. Eng., 2021, 9, 14385–14394.
62. L. Tan, T. Liu, Y. Zhang, N. Xue and P. Hu, Enhance selective separation of vanadium and iron from acid-leaching solution with sulfamic acid modified resin, Chem. Eng. J., 2023, 457, 141286.
63. K. Y. Nandiwale, A. M. Pande and V. V. Bokade, One step synthesis of ethyl levulinate biofuel by ethanolysis of renewable furfuryl alcohol over hierarchical zeolite catalyst, RSC Adv., 2015, 5, 79224–79231.
64. P. Yin, L. Chen, Z. Wang, R. Qu, X. Liu, Q. Xu and S. Ren, Biodiesel production from esterification of oleic acid over aminophosphonic acid resin D418, Fuel, 2012, 102, 499–505.
65. M. Hronec, K. Fulajtárová and T. Soták, Kinetics of high temperature conversion of furfuryl alcohol in water, J. Ind. Eng. Chem., 2014, 20, 650–655.
66. K. Wu, H. Li, X. Li, W. Han, C. Guo and X. Gao, Inter-integration reactive distillation with vapor permeation for ethyllevulinate production: Modeling, process analysis and design, Chem. Eng. Sci., 2021, 245, 116962.

---