High-Yield and High-Efficiency Conversion of HMF to Levulinic Acid in a Green and Facile Catalytic Process by a Dual-Function Brønsted-Lewis Acid HScCl4 Catalyst.

Lignocellulosic biorefineries have received considerable attention for the purpose of producing high-value chemicals and materials. Levulinic acid (LA) is an important biomass-derived platform chemical that is produced from sugar-based biomass. Unfortunately, the catalysts reported thus far have shortcomings, such as expensive starting materials, complicated synthesis or purification operations, and a low LA yield under harsh reaction conditions. Herein, we develop a novel dual-functional catalyst, HScCl4, by combining Brønsted acid (HCl) and Lewis acid (ScCl3) sites. The as-prepared HScCl4 catalyst shows high efficiency and high selectivity for converting 5-hydroxymethylfurfural (HMF) to LA in a biphasic system consisting of methyl isobutyl ketone (MIBK) and water. The density functional theory (DFT) results show that the synergistic catalytic effect, originating from the Brønsted and Lewis acidic sites of HScCl4, significantly decreases the energy barriers of reactants and intermediates, thus facilitating the conversion of HMF to LA. Moreover, the efficient separation of LA in the water-MIBK biphasic system by extracting LA to the MIBK phase minimizes the side reactions of LA and thus the formation of humins while significantly improving the LA yield. The conversion of HMF and the selectivity for LA are 100 and 95.6% at 120 °C for 35 min, respectively. The free energy (ΔG) and activation energy (E a) of the reaction are -30 kcal mol-1 and 13.7 kJ mol-1, respectively. The developed process provides a green, sustainable, and efficient pathway to produce LA from biomass-derived HMF under mild conditions.

Introduction

Lignocellulosic biomass, one of the most abundant renewable resources, has been regarded as an attractive renewable carbon source for biofuels and bio-based platform chemicals.[1] Particularly, as one of the three major components of lignocellulosic biomass, cellulose is a macromolecular polymer existing in nature that has become a valuable raw material for the production of chemical and biomedical materials.[2−4] The effective conversion to platform chemicals such as levulinic acid (LA) has attracted much attention.[5] However, due to the water-insoluble characteristics and the robust structure of cellulose, the reaction of the conversion of cellulose to LA usually needs harsh reaction conditions, and the LA yield is low. Various studies on the synthesis of LA from cellulose, glucose, and fructose have been performed,[6−8] and different catalysts, including solid acids,[9,10] inorganic acids,[11] and ionic liquids (ILs),[12−14] have been investigated. However, these catalysts studied thus far suffer shortcomings, such as expensive starting materials, harsh experimental conditions, and complicated synthesis and purification operations. Some studies have been performed on the synthesis of LA from 5-hydroxymethylfurfural (HMF) and different catalysts (Table ), including inorganic acids (H2SO4, HCl),[15,16] solid catalysts (Fe/Hβ-zeolite),[17] and ILs ([IL-SO3H]Cl + NiSO4).[4,14,18] Although some catalysts can achieve a high yield of LA, disadvantages such as complicated synthesis, corrosion equipment, and difficulty in separation and purification exist. Compared to Brønsted or Lewis acid catalysts and other catalysts, the concept of using both Brønsted and Lewis acids in the same system improves catalytic performance for the conversion of HMF to LA.[4,18]

Table 1

Comparison of Reaction Conditions and Results of the Present Study with Those Reported in the Literature

catalyst	T/°C	t/min	conversion/%	yield/%	Ea/kJ mol–1	references
H2SO4	141	12	100	85.3	110	(15)
HCl	270	3.3	100	76.5	94.1	(16)
HCl	180	240	86.5	42.4	92	(8)
H3PO4 + CrCl3	170	120	100	69.0	60.6	(19)
Fe/Hβ-zeolite	160	240	100	87.6	61	(17)
[IL-SO3H]Cl + NiSO4	175	120	100	56.3	34.3	(14)
β-zeolite + HCl	135	480	82.2	61.7	33	(20)
HScCl4	120	35	100	95.6	13.7	this work

To date, the LA yield results reported in the literature are usually low due to the side reactions occurring in the reaction system, which is typically a single-phase (water) system, where the condensations or polymerizations of reactants/intermediates/products occur, leading to the formation of humins.[8,14] Recent studies have shown that the biphasic reaction system, whereby the reactant and product are present in different phases, is of interest for high-yield conversion of biomass to the desirable product (including HMF to LA) due to the minimization of side reactions. Thus, the biphasic system is regarded as a promising process technology for biomass conversion to high-value chemicals.[18,21]

Rare earth metal-based Lewis acid catalysts have been shown to have excellent catalytic performance for reactions such as polyaddition,[22] hydration,[23] and polymerization.[24] Compared to d-series transition metals (such as Cr, Ti, and Fe), the rare earth metal possesses an abundant number of coordination sites. For example, the coordination number of Sc can reach up to 9–12,[25] which can effectively coordinate with heteroatoms, such as O, N, and S. The high coordination numbers can help facilitate high reactivity and selectivity in many reactions.[24]

In this study, we developed a novel dual-function acid catalyst, HScCl4, by combining rare earth metal-based Lewis acid-ScCl3 and Brønsted acid-HCl. It was then applied to catalyze the conversion of HMF to LA in a biphasic system consisting of methyl isobutyl ketone (MIBK) and water. The HScCl4 catalyst shows highly efficient and highly selective conversion of HMF to LA. The synergistic catalytic effect, originating from the Brønsted and Lewis acidic sites of HScCl4, obviously decreases the energy barriers of reactants and intermediates, which facilitates the conversion of HMF to LA. Moreover, the efficient separation of the water–MIBK biphasic system by extracting LA to the MIBK phase could minimize the side reactions of LA and significantly improve the yield of LA. Thus, the formation of humins decreased.

Results and Discussion

Analysis of the Frontier Molecular Orbital for Conversion HMF to LA

We proposed a possible mechanism for the conversion of HMF to LA using the dual-function Brønsted–Lewis acid HScCl4 catalyst (Figure a) based on available information in the literature.[26−28] This process involves five steps: C=C bond hydration of HMF to form intermediate IC1 (step 1), dehydration of IC1 to generate intermediate IC2 (step 2), which are consistent with the previous findings of Li et al.,[28] C–O bond cleavage of IC2 to produce intermediate IC3 with the 1,2-dicarbonyl structure (step 3), which has been reported by Velaga et al.,[27] decarboxylation of IC3 to produce intermediate IC4 and formic acid (step 4), and hydration-negative hydrogen-transfer reaction of IC4 to form LA (step 5); the last two steps agree with the findings of Shi et al.[26]

Figure 1

(a) Proposed mechanism for the conversion of HMF to LA in the presence of the HScCl4 catalyst. (b) Orbital energy gaps Δ|LUMO–HOMO| of HMF and intermediates IC1–IC4 over different catalysts. (c) Orbital energy gaps Δ|LUMO–HOMO| of transition states TS1–TS5 over the HScCl4 catalyst. (d) Effects of the different catalysts on the conversion of HMF and yields for various products [reaction conditions: 1.0 g of HMF, 10 mL of MIBK, and 0.27 mmol of the catalyst HScCl4 (in situ synthesized by 0.27 mmol of ScCl3 and 3.0 g of 0.33 wt % HCl), T = 120 °C, t = 35 min]. (e) Relative free-energy (ΔG) profile in the presence of the HScCl4 catalyst.

Full structure analyses of the reactant (HMF), intermediates (IC1–IC4), transition states (TS1–TS5), and product (LA) in the absence or presence of catalysts were performed via the density functional theory (DFT) method to assess their highest occupied molecular orbital (HOMO) and lowest unoccupied molecular orbital (LUMO) energy levels and both energy gaps Δ|LUMO–HOMO| (Figures b and S1–S3, Table S1). As shown in Figure b, among all catalysts, the energy gap Δ|LUMO–HOMO| for HMF using the HScCl4 catalyst is the lowest (3.15 eV), which is 1.88 and 0.01 eV lower than those of the HCl and ScCl3 catalysts, respectively. Additionally, the optimized structure of HMF + HScCl4 shows that the coordination (Sc3+ ← O) of the Lewis acid site Sc3+ with O (C–O–C) and the protonation (H+ ← C=C) of Brønsted acid site H+ with a C=C bond occur simultaneously (Figure S1, HMF + HScCl4).

When either HCl or ScCl3 is used alone, only one coordination (Sc3+ ← O) or protonation (H+ ← C=C) is possible (Figure S1, HMF + HCl and HMF + ScCl3), which is consistent with observations in the literature.[4,29] Compared with the single coordination or protonation of the catalyst HCl or ScCl3, the combined actions of the dual-function HScCl4 catalyst produce a synergistic inductive effect on HMF, significantly improving the 2C positive charge of the furan ring (Figure S1, HMF + HScCl4), thus facilitating the hydration of C=C of HMF to transition state TS1, which agrees with that in the literature (Figure a, step 1).[30] Therefore, the dual-function HScCl4 catalyst has excellent activation of HMF. Furthermore, the HScCl4 catalyst also significantly decreases the energy gaps Δ|LUMO–HOMO| of intermediates IC1–IC4 (Figure b), thus facilitating the conversion of HMF to LA.

We also calculated the orbital energy levels and energy gaps of transition states TS1–TS5 in the presence of the dual-function HScCl4 catalyst (Figure c). Based on our DFT results and those in the literature,[31,32] all the energy gaps Δ|LUMO–HOMO| of transition states TS1–TS5 are less than 6.0 eV, indicating that they are activated by the HScCl4 catalyst. Additionally, all transition states have only one imaginary frequency, and their energies at 0 K are the highest energy point, indicating that the structures of the transition states are credible (Table S1, Figures S4–S8). Therefore, the dual-function HScCl4 catalyst has thermodynamically favorable transition states, which is beneficial to lowering the activation energy, thus favoring the conversion of HMF to the desired product LA.

Effect of Different Catalysts for the Selective Conversion of HMF to LA

FTIR spectra using pyridine as a probe were used to study the types of acid sites of the bifunctional catalyst HScCl4, and the results showed that three bands were observed (1442, 1489, and 1545 cm–1), which were attributed to pyridine adsorbed on Lewis acid sites, a combination of Lewis and Brønsted acid sites, and pyridine interacting with Brønsted acid sites, respectively (Table S2, Figure S9). As shown in Figure d, the LA yield is significantly affected by the catalysts used. In the presence of the Brønsted acid HCl catalyst, the LA yield is the lowest (85.1%), with the formation of 3.7% humins and 11.2% angelica-lactone (ALT). In the presence of the Lewis acid ScCl3 catalyst, the LA yield was 88.7%, with the formation of 2.9% humins and 8.4% ALT. In contrast, when the Brønsted–Lewis acid HScCl4 catalyst is used, the yield for LA is the highest (95.6%), with the formation of only 1.6% humins and 2.8% ALT, much lower than those using HCl or ScCl3 alone as the catalyst. Therefore, the order of the catalytic performance for conversion of HMF to LA is as follows: HScCl4 > ScCl3 > HCl, agreeing with those of DFT analyses (Figure c). These results support the conclusion that the dual-function HScCl4 catalyst has excellent catalytic performance for HMF conversion to LA due to the synergistic catalytic effect originating from its double-acid sites.

Additionally, when formic acid was removed by distillation from the catalytic phase, the HScCl4 catalyst was recovered and reused for the next time (Figure S10). The HScCl4 catalyst was reused seven times without any obvious decrease in reaction conversion and product yields (XHMF = 100%, YLA = 95.1%). Therefore, the HScCl4 catalyst shows excellent reusable performance in the reaction. The good reusability of the HScCl4 catalyst is attributed to its excellent thermal stability (Figure S12), which comes from the strong complex ability of Sc3+ with Cl–.

Free Energy for HMF Conversion to LA

The free-energy (ΔG) profile for HMF conversion to LA in the presence of the dual-function HScCl4 catalyst is shown in Figure e: (1) the formation of transition state TS1 from HMF by the dual-function HScCl4 catalyst has a 26.7 kcal mol–1 energy barrier, while intermediate IC1 via the hydration reaction has an overall energy barrier of −1.9 kcal mol–1 (Figure a,e, step 1). (2) The 5-hydroxyl group of intermediate IC1 is activated by the Brønsted acid site (H+), and simultaneously, the oxygen atom of the furan ring is activated from the coordination of the Lewis acid site (Sc3+). Subsequently, dehydration and intramolecular C=C transfer lead to the formation of intermediate IC2, with a 10.3 kcal mol–1 energy barrier (Figure a,e, step 2). (3) The ether cleavage of intermediate IC2 forms intermediate IC3 through transition state TS3, and the critical driving force required is provided by the HScCl4 catalyst. The energy barrier of 83.4 kcal mol–1 for TS3 is the highest among all transition states due to the high stability of the dihydrofuran ring, which is similar to the finding of Su et al. in the ring-opening reaction (Figure a,e, step 3).[33] (4) Intermediate IC3 is hydrated to form transition state TS4 with a 58.8 kcal mol–1 energy barrier, following the decarboxylation of IC3, yielding intermediate IC4 and formic acid (Figure a,e, step 4). (5) Intermediate IC4 is attacked by water to form transition state TS5, with a 45.1 kcal mol–1 energy barrier, followed by migration of a hydride, ultimately forming LA, with a 55.0 kcal mol–1 energy release, where again, the critical driving force is provided by the HScCl4 catalyst (Figure a,e, step 5).

For the overall reaction process, the highest energy barrier is the ether cleavage of intermediate IC2 (83.4 kcal mol–1). Additionally, a total of 30 kcal mol–1 apparent energy is released for the HMF conversion to LA in the presence of the dual-function HScCl4 catalyst. Yang et al. used the Bronsted acid site H+ as a catalyst for the conversion of HMF to LA, and via DFT calculations, they found that the lowest free energy was −32.7 kcal mol–1 in the three reaction pathways.[34]

Effect of the Biphasic Reaction System for Selective Conversion of HMF to LA

The advantage of the water–MIBK biphasic system is shown in Figure c, where the conversion of HMF to LA is performed in water (without MIBK) under otherwise the same conditions. HMF is completely consumed, but the LA yield is 91.3% (in contrast to the 95.6% LA yield in the biphasic system), with the formation of 3.1% humins and 5.6% ALT (both are higher than those in the biphasic system at 1.6% and 2.8%, respectively). These results indicate that the efficient separation of LA in the water–MIBK biphasic system by extracting LA to the MIBK phase (Figure a) minimizes the side reactions of LA, thus leading to the increased LA yield. A similar advantage of the biphasic system, that is, increasing the product yield and minimizing the side reactions, was also reported in other reactions, such as asymmetric sulfoxidation, aldol reaction, and ammonia synthesis.[35−37]

Figure 2

(a) Schematic diagram of the water–MIBK biphasic system. (b) Distribution coefficient of LA (KLA) in a water–MIBK biphasic system as a function of temperature (conditions: 10.0 mL of MIBK, 1.0 g of LA, 3.0 mL of water, or 3.0 g of 0.33 wt % HCl aqueous solution with 0.27 mmol of ScCl3). (c) Effects of the reaction system on the conversion of HMF to LA (reaction conditions: 1.0 g of HMF, 10 mL of MIBK, and 0.27 mmol of the catalyst HScCl4 in situ synthesized by 0.27 mmol of ScCl3 and 3.0 g of 0.33 wt % HCl, T = 120 °C, t = 35 min). (d,e) Effect of reaction temperature and time on the catalytic conversion of HMF and LA yield in the presence of the HScCl4 catalyst. (f) Arrhenius graphs for the catalytic conversion of HMF to LA in the presence of the HScCl4 catalyst.

The effective extraction of LA in the biphasic system is supported by the high distribution coefficient of LA (KLA) in the water–MIBK biphasic system (Figure b), which is enhanced by the salt effect of HScCl4.

Effect of Reaction Time and Temperature for Selective Conversion of HMF to LA

As shown in Figure d, the reaction temperature and time have significant effects on the conversion of HMF. The conversion of HMF obviously increases with increasing reaction temperature and time. The HMF conversion is only 82.2% at 100 °C for 35 min (Figure d), while it reaches 100% at 130 °C for 30 min.

The effects of different reaction temperatures and times on the LA yield are shown in Figure e. The highest LA yield increases from 78.8 to 95.6% as the temperature increases from 110 to 120 °C. At 130 °C, the LA yield reaches a maximum (92.3%) at 25 min and then sharply decreases to 88.7% at 35 min, which is due to the condensation/polymerization of HMF, the intermediates, and the conversion of LA to ALT. Similar side reactions were reported by Patil and Lund and Galletti et al. in the conversion of HMF and giant reed to LA, respectively.[38,39]

Kinetic Study on Selective Conversion of HMF to LA

The results used in this section were obtained within 10 min because during this period, the side reactions (condensations/polymerizations) were minimized. The kinetics at different temperatures is analyzed (Figures S13–S16). The activation energy (Ea), calculated from the slope of the Arrhenius plot, is 13.7 kJ mol–1 (Figure f), which is lower than the values reported in the literature (Figure ),[8,14−17,19,20] supporting the advantages of the dual-function HScCl4 catalyst. Notably, the catalyst can accelerate the reaction by stabilizing the transition states and decreasing the energy barriers. Furthermore, the obtained LA yield is very high (reaching 95.6%), even higher than the highest value reported in the literature (87.6%) at 160 °C for 240 min in the presence of Fe/Hβ-zeolite (Figure and Table ).[17] Kumar et al. reported that the reaction activation energy (Ea) is 34.3 kJ mol–1, with an LA yield of only 56.3% at 175 °C for 120 min using a composite catalytic system of Brønsted acid IL [IL-SO3H]Cl and Lewis acid NiSO4 (Table ).[14] Garces et al. reported the conversion of HMF to LA in the Brønsted–Lewis acid β-zeolite catalyst with Brønsted acid HCl as a cocatalyst, and the highest LA yield was 61.7% at 140 °C (Table ).[20]

Figure 3

Comparison of the activation energy and LA yield for the catalytic conversion of HMF to LA of the present study with those reported in the literature.

Therefore, our results show great potential for using the dual-function HScCl4 catalyst for the conversion of HMF to LA. In addition to the advantages of a high yield and low activation energy, the dual-function HScCl4 catalyst biphasic system also makes it possible to carry out the reaction under mild conditions, that is, 120 °C for 35 min.

Conclusions

A novel dual-function Brønsted–Lewis acid HScCl4 catalyst was developed for the catalytic conversion of HMF to LA. The results revealed that the HScCl4 catalyst has excellent catalytic performance for reactions in a water–MIBK biphasic system. The synergistic catalytic effect of the Brønsted and Lewis acid sites of the HScCl4 catalyst obviously decreases the energy barriers of the overall reaction, thus lowering the activation energy (Ea) and free energy (ΔG) of the reaction to 13.7 kJ mol–1 and −30 kcal mol–1, respectively. Furthermore, the water–MIBK biphasic system, enhanced by the HScCl4 salt effect, leads to efficient extraction of LA to the MIBK organic phase, thus minimizing the side reactions of LA and significantly improving the LA yield. The conversion of HMF and the selectivity for LA reach 100 and 95.6% at 120 °C after 35 min, respectively. The developed process provides a green sustainable and efficient pathway to produce platform chemicals of LA from biomass-derived HMF under mild conditions.

Experimental Section

Materials

HMF, LA, scandium chloride, MIBK, and other chemicals (analytical purity) were obtained from Sinopharm Chemical Reagent Co., Ltd. All materials were used without further purification unless otherwise explicitly stated.

Computational Method

All quantum calculations were performed with Gaussian software (Gaussian 09W Ver. 05, Gaussian, Inc.). The DFT method was adopted, and the standard 3-21G (*) basis set was used for all atoms. The geometries of catalysts and reactants were fully optimized using Berny’s analytical gradient method,[40] and geometry optimizations were performed without symmetry restraints unless mentioned otherwise. In addition, vibrational frequency analysis based on the optimized structure was performed to decide whether the geometry of the reactant, molecular complexes, or products was a minimum on the potential energy surface or a transition state. The optimized geometries of the transition states were also characterized by one imaginary frequency, which was associated with a single eigenvalue of the Hessian matrix and confirmed by an IRC calculation method.[41]

Catalytic Reactions

The reaction was carried out in a 50 mL cylindrical stainless-steel reactor system (Dalian University of Technology Machinery Technology Co., Ltd., China) with external temperature and stirring controllers. In a typical procedure, HMF (1.0 g), ScCl3 (0.27 mmol), 0.33 wt % HCl aqueous solution (3.0 g), and MIBK (10 mL) were loaded into the reactor. The reactor was sealed and purged with 0.3 MPa N2 four times and then heated to the target temperatures for a preset duration with stirring at 400 rpm. After the reaction, the reactor was removed and rapidly cooled to 70 °C in a water bath, and then, the reaction mixture was stratified. The top layer was the mixture of products and analyzed by HPLC (LC-20AD), which was equipped with UV detection and an ODS-EP C18 reversed-phase column (250 mm × 4.6 mm, Intersil). The detection conditions were as follows: mobile phase, 15 mmol L–1 acetonitrile and phosphoric acid–sodium dihydrogen phosphate buffer solution (v1/v2 = 15:85); flow rate, 1.0 mL min–1; column temperature, 30 °C. Formic acid was analyzed by HPLC equipped with an Aminex HPX-87H organic acid column and an RID-20A differential refractometer. The detection conditions were as follows: mobile phase, 5 mmol L–1 aqueous solution of sulfuric acid; flow rate, 0.55 mL min–1; column temperature, 60 °C. The lower layer was a mixture of catalyst phases, including the catalyst, water, humins, and formic acid. When humins were filtered out and formic acid was removed by distillation, the catalyst was recovered. When a Lewis acid ScCl3 catalyst was used, 2.0 g of H2O was added to the reactor. When a Brønsted acid HCl catalyst was used, H2O was not added. The conversion of HMF (XHMF) and yields for LA (YLA), ALT (YALT), and humins (YHumins) are defined in eqs –4. Here, CHMF,0 and CHMF are the initial and final concentrations of HMF, respectively, and CLA and CALT are the final concentrations of LA and ALT, respectively. In addition, there are some products with a relatively low content, and we attribute the content of less than 0.2% to humins. During the reaction process, formic acid and LA were also formed. ALT was converted from LA; thus, the yield of formic acid was equal to the sum of the LA and ALT yields. By analyzing the HPLC data, it was found that the yield of formic acid was indeed the sum of the yields of LA and ALT. Therefore, the yield of formic acid is not listed in this article.

All definitions are on a molar basis.