In the following paragraphs, the state of the art regarding the synthesis of ALs from C6 model compounds of different complexity will be discussed, while taking into account the most promising available data that were reached with both homogeneous and heterogeneous catalysts. For a clearer discussion, ALs will be considered separately, methyl levulinate (ML), ethyl levulinate (EL), propyl levulinate (PL), butyl levulinate (BL), pentyl levulinate (PeL), and hexyl levulinate (HL).
2.1. ML Synthesis from Model Carbohydrates
Table 1 summarizes the available data for ML synthesis from C6 model carbohydrates with inorganic mineral acids or metal salts.
Sulfuric acid has been widely employed as a homogeneous Brønsted acid catalyst for the methanolysis of C6 model feedstocks, thanks to its excellent compromise between high activity and low-cost. Higher ML molar yields were achieved, starting from fructose (runs ML_1—ML_3,
Table 1), rather than from glucose (runs ML_4—ML_5,
Table 1), due to the simpler conversion of the first feedstock, which, contrarily to the glucose, does not need of the Lewis acid catalysis. The use of sulfuric acid for the one-step alcoholysis of model cellulose was proposed for the first time by Garves [
69], who reported good ML yields, working under subcritical conditions (runs ML_6—ML_7,
Table 1). However, sulfuric acid is also an excellent catalyst for alcohol etherification, which may cause a progressive and significant consumption of the solvent/reactant. To solve this issue, etherification must be controlled and limited as much as possible, in particular by minimizing the acid concentration, a process solution that also improves the environmental sustainability and reduces the plant costs of the alcoholysis process. Peng et al. [
43] investigated this aspect more in-depth for the glucose alcoholysis at 200 °C, in the presence of two different H
2SO
4 concentrations (0.1 and 0.005 mol/L). The authors found that methanol conversion to dimethyl ether, after 2.5 h, and in the presence of 0.1 mol/L of H
2SO
4, was remarkable (about 59 mol%), whilst it strongly decreased (to about 2 mol%), if the use of diluted H
2SO
4 was preferred. More recently, the use of very dilute sulfuric acid (≤0.01 mol/L) has been rightly preferred by Li et al. [
70], reporting similar ML yields (about 50 mol%), only extending the reaction time (run ML_8,
Table 1). Besides, the authors also discussed the possible ML recovery, combining atmospheric and vacuum distillation, while using
n-dodecane for improving the separation of the heavy fraction (ML) from the light one (MeOH). Wu et al. [
71] achieved similar ML yields, always preferring a very low sulfuric acid concentration (≤0.04 mol/L), and by properly optimizing reaction temperature and time (runs ML_9—ML_10,
Table 1). Besides, the authors compared the catalytic performances of sulfuric acid with those of other common Brønsted acids, such as phosphoric, formic, and acetic acids, working at the same molar concentration (0.04 mol/L). The higher catalytic activity of the sulfuric acid over phosphoric acid was attributed the stronger acidity of the former [
76], whilst organic acids, such as formic and acetic ones, were not efficient incatalyzing this reaction, due to their unwanted esterification, occurring under the typical reaction conditions that were required for ML production [
77]. The use of MW as an efficient heating system is better exploitable if applied to the conversion of more recalcitrant substrates, such as cellulose, achieving better ML yields with respect to the conventional heating systems and reduced formation of the reaction by-products, mainly humins. In this context, the best ML yields achieved by Feng et al. [
68] and Chen et al. [
67] (runs ML_11—ML_12,
Table 1) fully support these premises, highlighting that such efficient heating allows for the use of milder reaction conditions (temperature and time) than conventional heating systems. Up to now, the highest ML yield from model cellulose was reported by Chen et al. [
67] (run ML_12,
Table 1), amounting to 70 mol%, which is the best value to consider as reference for the development of new efficient catalysts. The innovation of the catalytic system is a very hot topic and many alternatives have been already evaluated. Perrhenic acid (HReO
4) is just an example, but, certainly, this is not the simplest and cheapest catalyst to propose as an alternative to the sulfuric acid. Its good efficiency towards fructose methanolysis has been recently demonstrated by Bernardo et al. (run ML_13,
Table 1) [
72], but its real use on a greater scale is not sustainable and competitive with that of the sulfuric acid. Despite the high activity of sulfuric acid, separation and corrosion issues hinder its use, especially at high concentrations, thus the research is moving towards the use of other kinds of catalysts, in particular inorganic salts. In this context, cheap metal sulfates (Na
2SO
4, K
2SO
4, MnSO
4, CoSO
4, NiSO
4, ZnSO
4, CuSO
4, Fe
2(SO
4)
3, La
2(SO
4)
3, and Ce(SO
4)
2) have been recently tested by Sun et al. [
73] for the solvothermal conversion of glucose and fructose, evaluating different reaction media, including MeOH. As a general consideration, the metal salts provides both Lewis acid sites (due to the unsaturated metal center) and Brønsted acid ones (due to the hydrolysis/methanolysis of the salt), which catalyze the isomerization of glucose to fructose and the dehydration of fructose to ML, respectively, whilst sulfate anions generally have a positive role towards this reaction, strongly chelating with the carbohydrates, thus preventing the formation of unwanted by-products, such as humins [
73]. Among the tested metal sulfates, Fe
2(SO
4)
3 and La
2(SO
4)
3 gave the highest ML molar yields at 170 °C, in the range 25–30 mol% (runs ML_14—ML_16,
Table 1), preferring the cheaper Fe
2(SO
4)
3, which suppressed the humin formation. However, despite this catalyst seeming interesting for ML synthesis, the reported ML yields from glucose and fructose are too low, especially if compared with the traditional sulfuric acid. A possible solution to improve the catalysis in the presence of metal salts provides the additional use of modest quantities of sulfuric acid, which favors the Brønsted acid-catalyzed steps. Peng et al. [
66] investigated the catalytic effect of various metal sulfates, who confirmed a significant effect of these salts on the reaction selectivity. In particular, K
2SO
4 and Na
2SO
4 were particularly effective for stabilizing the methyl glucoside intermediate, suppressing its further transformationand minimizing the humin formation. Instead, Fe
2(SO
4)
3 and Al
2(SO
4)
3 gave the highest ML yields (runs ML_17—ML_18,
Table 1), but also favored the formation of humins and dimethyl ether by-products. However, the formation of these by-products (in particular dimethyl ether) was significant also in the sole presence of these two metal sulfates, being the conversion of methanol to dimethyl ether equal to 19 and 23 mol%, adopting Fe
2(SO
4)
3 and Al
2(SO
4)
3, respectively, highlighting the importance of further optimizing the alcoholysis reaction with these more active catalytic systems [
66]. The optimization of the alcoholysis of different C6 model substrates in the presence of Al
2(SO
4)
3 was investigated more in-depth (runs ML_19—ML_32,
Table 1) [
29,
32,
67,
74,
75]. The addition of water reduces humin/coke formation and solvent consumption (runs ML_28—ML_29,
Table 1) [
32]. The Lewis acid species [Al(OH)
x(H
2O)
y]
n+ and Brønsted acid species H
+, which were generated by in-situ hydrolysis of Al
2(SO
4)
3, were responsible for the improved catalytic performances. The amount of water in MeOH could affect the equilibrium of metal salt hydrolysis, as well as the acid density of the reaction mixture, which would affect the salt reactivity towards cellulose conversion and related product distribution. In particular, the addition of water significantly inhibited the etherification reaction of MeOH to dimethyl ether, which was unavoidable in the presence of Brønsted acidic catalysts at elevated temperatures. When no water was added, the dimethyl ether yield was approximately20 mol%, whilst it decreased below 5 mol% by adding a low amount of water (0.6 mL). Moreover, water addition improved the cellulose solvolysis, moving the reaction towards the hydrolysis of the glycosidic C-O bonds, rather than to its thermal decomposition to coke, whose yield was reduced from about 40 wt%, in the absence of water, to about 10 wt%, when the water content was over 0.6 mL. The combined use of (Al
2(SO
4)
3+ H
2O), together with MW heating as the efficient heating system, and milder reaction conditions, allowed for the improvement of the ML yield from model cellulose up to a maximum of about 70 mol% (run ML_29,
Table 1), a value that is similar to that reached with the traditional sulfuric acid. Similar results were obtained under conventional heating but employing longer reaction times (run ML_24,
Table 1) [
29], while harsher reaction conditions resulted in being detrimental for ML production (run ML_26,
Table 1). Some improvements have been recently proposed for enhancing the ML yield, always in the presence of Al
2(SO
4)
3 as the main catalytic system. Chen et al. [
74] performed a ball-milling pretreatment on the cellulose feedstock, for reducing its particle size and crystallinity index, thus improving the catalyst accessibility during the alcoholysis step. Anyway, this pre-treatment did not lead to significant improvements in ML yield, which reached the maximum value of about 65 mol%, working at 170 °C and for 45 min., under MW heating (runs ML_30—ML_32,
Table 1). Very recently, an oxidation pretreatment has been considered for improving ML production from cellulose, aimed at the conversion of its hydroxymethyl groups into carboxylic ones, thus providing the Brønsted acid sites for improving the catalysis that is related to the steps of interest [
75]. However, by adopting this interesting approach, the maximum ML yield was only 52 mol%, working at 180 °C for 3 h (run ML_32,
Table 1), highlighting that, up to now, this pre-treatment is not very functional for this purpose.
Sulfonic acids or/and sulfonate salts or resins have been investigated for the ML production from C6 model carbohydrates, and the best available data are summarized in
Table 2.
p-Toluenesulfonic acid (PTSA) has been proposed for a long time as a reference Brønsted acid catalyst for studying the cellulose alcoholysis (runs ML_33—ML_35,
Table 2), as an effective alternative to the already discussed sulfuric acid, achieving similar maximum ML yield (about 50 mol%). Although single catalysts have been widely used for this reaction, achieving good results, recent catalytic studies are rather oriented towards the appropriate tuning of the Brønsted–Lewis acidity, generally achieved in the presence of binary catalysts. Therefore, as previously observed, Lewis acids have been used to improve the ML yield, both alone or in combination with small amounts of sulfonic acids, as Brønsted acids. In this context, In(OTf)
3 has been proposed by Tominaga et al. [
27] as an efficient Lewis acid for the cellulose alcoholysis, both alone and in combination with PTSA or 2-naphthalenesulfonic acid (2-NSA), as active Brønsted acids (runs ML_36—ML_38,
Table 2). The sole In(OTf)
3 showed catalytic performances that were similar to those of PTSA, whilst the ML yield increased up to 70–75mol%, when In(OTf)
3 and PTSA or 2-NSA were used in combination. The adopted catalytic system resulted in being stable and recyclable, being recovered as a residue after the distillation of the solvent/products. Similar promising results were obtained by Nemoto et al. [
78] with simpler feedstocks, such as glucose, mannose, or galactose, highlighting a similar reactivity of these carbohydrates (runs ML_39—ML_41,
Table 2). In a more recent work of Tominaga et al. [
79], binary catalytic systems that were composed of aluminum compounds (Al(OEt)
3, Al(acac)
3, or Al(OH)
3) and organic sulfonic acids (PTSA or 2-NSA), were found to be particularly efficient for direct ML synthesis from microcrystalline cellulose (runs ML_42—ML_44,
Table 2), achieving maximum ML yields that were similar to those of the best triflate-based catalytic systems (runs ML_37—ML_38,
Table 2). Amberlyst-15 and Nafion NR50 have been mainlyconsidered as catalysts of reference, for justifying the development of new heterogeneous catalysts, despite the available data beinglimited to the conversion soluble C6 carbohydrates (runs ML_45—ML_49,
Table 2). A novel and efficient sulfonic acid-grafted ethylenediamine-functionalized mesoporous polydivinylbenzene (PD-En-SO
3H) heterogeneous catalyst was synthesized by Pan et al. [
81]. PD-En-SO
3H showed excellent catalytic performances for the conversion of fructose to ML, being more active than commercial Amberlyst-15 and Nafion NR50 (compare run ML_50 with runs ML_46 and ML_49,
Table 2). Moreover, this new catalyst could be reused for four times, maintaining its high catalytic activityunaltered. A series of sulfonic acid-functionalized carbon materials, including poly(
p-styrenesulfonic acid)-grafted carbon nanotubes (PSSA-g-CNT), poly(
p-styrenesulfonic acid)-grafted carbon nanofibers (PSSA-g-CNF), benzenesulfonic acid-grafted CMK-5 (BSA-g-CMK-5), and benzenesulfonic acid-grafted carbon nanotubes (BSA-g-CNT), have been applied for fructose conversion to ML (runs ML_51—ML_54,
Table 2) [
80]. The catalytic activities of these sulfonic acid-functionalized carbon materials, applied to the conversion of fructose to ML, follow the order of their acid strength and PSSA-g-CNT exhibited the highest acid density and the best catalytic performances to ML (maximum yield of 69 mol%). The authors claimed the high thermal stability and ease of recovery of these catalysts. Sulfonated hyperbranched poly(aryleneoxindole)s (SHPAOs) were also used for the conversion of simple monosaccharides and inulin to ML, with high yields up to 79 mol% (runs ML_55—ML_57,
Table 2) [
84]. Becauseofthe soluble character of the hyperbranched catalyst in the alcoholic solvent, it was easily separated from the solid humins, and recovered from the solution over a commercial low molecular weight cut-off filter. Moreover, the recovered catalyst showed in a recycle run a comparable catalytic activity (per catalyst weight) and product selectivity. To exploit more sustainable carbon bio-materials, an amorphous carbon-based catalyst was prepared by the sulfonation of the bio-char obtained from fast pyrolysis (N
2atm; 550 °C) of
jatrophacurcas de-oiled seed cake and tested for the cellulose methanolysis, reaching a maximum ML yield of 30 mol% (runs ML_58—ML_59,
Table 2) [
85]. The authors demonstrated that this functionalized carbon catalyst was stable for five cycles with a slight loss in catalytic activity.
Heteropolycompounds, or heteropolyoxometalates (POMs), consist of metal oxide clusters of early transition metals [
86]. These could be heteropolyacids (HPAs) or their salts, containing one heteroatom (X = P(V), As(V), Si(IV), and B(III)) and addenda atoms (M = W(VI), Mo(VI), and V(V)). The main heteropolyanion that is used in the field of biomass conversion is the Keggin structure (XM
12O
40n-), but catalysts with Wells–Dawson structure (X
2M
18O
62m-) have also found applications [
87]. These catalysts are widely used in acid-catalyzed and oxidation reactions, with economic and green benefits, but generally exhibit poor efficiency in their bulk form, due to the limited number of exposed active sites [
88]. HPAs are usually soluble in aqueous and organic solvents, making their separation/regeneration from the reaction mixturedifficult, and, for this reason, many attempts for their heterogenization have been made by immobilization on support or by the formation of an insoluble ionic material [
86]. The use of these catalysts has also been proposed for the ML production, and the most interesting available data are summarized in
Table 3.
In recent years, HPAs have attracted great interest as homogeneous acid catalysts, due to their strong Brønsted acidity, good structural mobility, and marked multi-functionality. Among these, the Keggin-type tungstosilicious and tungstophosphoric HPAs (H
4SiW
12O
40 and H
3PW
12O
40) have been proposed as efficient homogeneous catalysts for the alcoholysis of C6 model carbohydrates. In particular, H
3PW
12O
40 and H
6P
2W
18O
62 showed good catalytic performances for ML synthesis, starting from different model feedstocks, including cellulose (runs ML_60—ML_62,
Table 3), whilst H
4SiW
12O
40 was selective for the formation of methyl glucoside intermediate and dimethyl ether by-product, rather than for ML synthesis, leading to the maximum methyl glucoside yield of 57 mol%, and the corresponding conversion of methanol to dimethyl ether of 28 mol% (run ML_63,
Table 3) [
91]. HPAs were modified by partially substituting the proton with larger cations, such as Cu
II, Zn
II, Cr
III, Fe
III, Sn
IV, Ti
IV, and Zr
IV, in order to improve the Lewis acidity of the HPAs, which helps the isomerization step from glucose to fructose, and to limit their solubility in alcohols, which complicates their work-up procedures.The best ML yields (in the range 50–60 mol%) were achieved in the presence of the H
5PW
11TiO
40 catalyst, adopting feedstocks of different complexity, including cellulose (runs ML_64—ML_68,
Table 3) [
92]. In these cases, the addition of efficient Lewis acid sites, together with the appropriate balance between Brønsted and Lewis acidity, was responsible for the promising performances of the catalyst. Contrarily, too strong Lewis acid sites promoted C-C bond cleavage, so that, for example, Fe-based HPWs gave lower ML yields than Ti-based HPWs. However, in another work of Liu et al. [
89], the Brønsted–Lewis acidity of Fe-based HPWs catalysts was properly tuned in favor of the methanolysis pathway, thus allowing for a complete conversion of simple carbohydrates, reaching good ML yields, working at 130 °C for 2 h, and ensuring satisfactory recycling tests (runs ML_69—ML_72,
Table 3) [
89]. Anyway, the use of this catalyst for cellulose conversion gave an unsatisfactory ML yield (run ML_73,
Table 3), which limits its application to simpler model feedstocks. Good performances were also reported for Sn(II) exchanged HPA Keggin Sn
2SiW
12O
40, leading to the maximum ML yield of about 57 mol%, starting from fructose, working at 150 °C for 3 h (run ML_74,
Table 3) [
93]. Instead, Zhang et al. [
94] investigated the performances of the AlPW
12O
40 catalyst, demonstrating its effectiveness for the MW-assisted ML synthesis (ML yields in the range 45–70 mol%), starting from various carbohydrates, working at 160 °C for 30 min. (runs ML_75—ML_78,
Table 3). AlPW
12O
40 acts as a bulk-type catalyst in a pseudo-liquid system, improving the accessibility of the active catalyst sites. The absorption of high-polar MeOH favors this liquefaction, increasing the distance between the heteropolyanion-based species, while increasing the proton mobility [
94]. Among the other possible proposals of metal-exchanged HPAs, the catalytic performances of Cs
xH
3-xPW
12O
40 catalysts were tested for one-pot dissolution of microcrystalline cellulose and subsequent conversion to ML, in the presence of supercritical MeOH [
95]. The authors identified Cs
2.5H
0.5PW
12O
40 as the best performing catalyst for ML production. In the absence of a catalyst, under supercritical conditions and for short reaction times (300 °C/10 MPa/1 min.), the cellulose dissolution was successful, but the next conversion to ML was inefficient. On the other hand, the addition of Cs
xH
3-xPW
12O
40 activated the methanolysis pathway to ML (maximum yield of 20 mol%), due to its well-balanced Lewis–Brønsted acidic properties (run ML_79,
Table 3). The authors highlighted the improved accessibility of the solubilized carbohydrates to the Brønsted sites and the acid strength of Cs
2.5H
0.5PW
12O
40 catalyst. However, despite cellulose solubilization was significantly improved, the performances of all these catalysts were modest, especially if compared with those that were achieved under milder subcritical conditions. Heteropolyanion-based ionic liquid catalysts were tested for the conversion of different model compounds [
96,
97]. In particular, [TMEDAPS]
3 [PW
12O
40]
2 gave high ML yield from fructose (run ML_80,
Table 3), whilst [PyPS]
3PW
12O
40 was used also for more complex substrates (runs ML_81—ML_84,
Table 3), including cellulose, in this case reaching the maximum ML yield of about 71 mol%, working for 5 h at 150 °C. In particular, the remarkable efficiency of [PyPS]
3PW
12O
40 towards the cellulose alcoholysis was ascribed to the high acidic strength of the sulfonic-functionalized cation and to its synergistic effects with the corresponding heteropolyanion. Besides, the proper acidic strength of this catalyst favored the whole process consisting of cellulose hydrolysis, levulinic acid formation, and its esterification. Moreover, the authors demonstrated that [PyPS]
3PW
12O
40 can be effectively recovered by self-separation, through simple temperature control, also showing excellent reusability, even after ten runs. H
6P
2W
18O
62 was properly modified with surfactants, in order to prepare a micellar assembly (C
16TA)
xH
6-xP
2W
18O
62 [
90], which gave the maximum ML yield of about 58 mol%, starting from cellulose (run ML_85,
Table 3). In particular, the promising performances of the (C
16TA)H
5P
2W
18O
62 catalyst were attributed to its micellar structure, high acidic content, and good oxidizing ability. The assembly of HPAs with basic organic species is another possibility forsolidifying the homogeneous acid and modulate its acidity. Novel substituted pyridine phosphotungstates were prepared and tested for the direct conversion of fructose to ML [
98,
99]. 3-fluoropyridine phosphotungstate (3-FPyPW) displayed superior catalytic activity for the synthesis of ML from fructose (run ML_86,
Table 3), which was ascribed to its relatively higher acidity, and the maximum ML yield of 82 mol% was achieved [
98]. In another work of Fang et al. [
99], 3-phenylpyridine phosphotungstate (3-PhPyPW) hybrid catalyst displayed good catalytic performances in the upgrade of fructose to ML (yield of 71 mol%), after 8h at 140 °C, which was attributed to its relatively large pore size and high hydrophobicity (run ML_87,
Table 3). As a further improvement of HPA-based catalysis, very recently, a trifunctional polyoxometalate Cs
10.6[H
2.4GeNb
13O
41] catalyst, which included Brønsted acid sites, Lewis acid sites, and basic sites, was synthesized and used for this reaction, starting from different model compounds, under conventional and MW heating (runs ML_88—ML_95,
Table 3) [
100]. The further addition of basic sites improves the isomerization from methyl glucoside to methyl fructoside and the subsequent formation of 5-methoxymethylfurfural. This interesting approach for the production of ML from cellulose allowed achieving maximum ML yields of 53 and 55 mol%, under conventional and MW heating, respectively, and the catalyst maintained almost unaltered its activity after six recycling runs.
Zeolites are highly porous aluminosilicates, exhibiting a well-defined channel and cage-based structure, and, less commonly, a lamellar structure [
101]. The molecular dimensions of their channels and cages greatly contribute to their catalytic potential. The catalytic properties of zeolites are due, in part, to the exchangeable ions and water molecules trapped in their structure, as well as their commonly high adsorption capacity. In general, although zeolites are strong acids at high temperatures, their acidity is relatively modest at the common temperature ranges of biomass conversions [
101]. However, it is possible to functionalize the zeolites and modulate their existing electronic features, giving rise to zeolites with enhanced acidic properties, which are more suitable for this reaction. The most significant examples of zeolites applied to the ML production are reported in
Table 4.
Unmodified zeolites, such as H-β, HZSM-5, and HY, were mainly proposed as catalysts of reference, in order to demonstrate the effectiveness ofad hocsynthesized catalysts [
82,
95]. However, the low ML yields that were obtained with these zeolites (runs ML_96—ML_98,
Table 4) show that the modification of their acidic properties is fundamental to improve the ML production. Saravanamurugan et al. [
102] demonstrated that unmodified zeolite H-USY (6) can directly transform model mono-, disaccharides, and inulin polysaccharide to ML, with good yields (about 50 mol%) (runs ML_99—ML_103,
Table 4). The H-USY (6) zeolite was preferred by the authors for its high content of Lewis acid sites, which facilitated the isomerization of glucose to fructose and the further conversion to ML. Moreover, this zeolite maintained unaltered its structural integrity in the alcohol medium, and it was reused five times without significant changes in the ML yield. However, despite these valuable data, the reported ML yields from more complex feedstocks, e.g., starch and cellulose, were too low (runs ML_104—ML_105,
Table 4), indicating that a modification of the catalyst was necessary to improve the catalysis. In fact, the microporosity of the zeolites represents an important drawback for achieving an efficient diffusion of these bulky bio-molecules. In order to solve these shortcomings, some additional mesopores can be introduced into the zeolite framework, for example through nitric acid treatment, which removes the extra framework aluminum species, leading to an increase of mesoporosity and a slight decrease of acidity. Hierarchical H-USY was modified in such a way by Zhou et al. [
82], and tested for the conversion of model carbohydrates to ML (runs ML_106—ML_110,
Table 4). However, despite the good performances declared by the authors, the comparison with the other data discussed up to now shows that these ML yields for simple model feedstocks are not striking and, even more so, cellulose was not studied. Li et al. [
103] proposed the use of a zirconia-zeolite hybrid ZrY6(0.5) catalyst, showing a moderate acid-base site content (0.97 and 0.08 mmol/g), high stability and porosity (average mesopore diameter: 6.2 nm). It was demonstrated that metal content/type and acid-base bifunctionality were closely correlated with substrate conversion and ML yield, respectively. The catalytic activity for ML production was higher than that of other zeolites (runs ML_111—ML_117,
Table 4), but the issue of the low ML yield from cellulose remained unsolved (run ML_117,
Table 4). Anyway, the catalyst could be reused five times, maintaining stable conversion rates and ML yields. Similarly, Yang et al. [
104] enhanced the Lewis acidity of Al-β zeolite by the loading of tin species. Mesopores were generated by the hydrothermal treatment of Sn/Al-β zeolite, via desilication with tetraethylammonium hydroxide, in order to restructure the zeolite and enhance the porosity of Sn-Al-β, which facilitated the diffusion of the reactant and, consequently, the ML production. The dual effects of Lewis acidity and mesoporosity improved ML yield about 2.3 folds from glucose if compared to the parent zeolite. However, the best ML yield from glucose (run ML_118,
Table 4) is in agreement with those reported for other discussed zeolites; hence, also in this case, the improvement deriving from the use of this catalyst is not remarkable. Lastly, a bifunctional catalytic system that was composed of commercial homogeneous H
4SiW
12O
40 as the Brønsted HPA, and Sn-Beta zeolite as the Lewis acid, was recently designed by Zhou et al. [
105], to catalyze the direct conversion of model mono- and polysaccharides to ML (runs ML_119—ML_121,
Table 4). The strong Brønsted acidity and the solubility of HPA, together with the superiority of Sn-Beta zeolite towards the isomerization reaction, make this bifunctional catalyst system highly active for ML production, even when employing cellulose (run ML_121,
Table 4, yield to ML 62 mol%), achieving the best catalytic performances within these kinds of catalysts.
Another group of catalysts of great interest is that of montmorillonites (clays), which are aluminosilicates constituted of multiple layers of polyhedrons [
106]. Tetrahedral silicon oxide and octahedral hydrous alumina are the common building blocks of these catalysts. Clays have demonstrated remarkable catalytic activity towards many biomass conversion routes, due to their porous structure that provides a unique environment in which molecules can interact in specific ways, allowing for reactions to take place. Other remarkable advantages are high abundance, versatility, smart modulation of the textural properties, and environmental inertness.
Table 5 reports the main advances towards the ML synthesis from model compounds in the presence of these catalysts.
Various metal ion-exchanged montmorillonite catalysts were prepared, characterized, and evaluated in glucose conversion to ML (runs ML_122—ML_124,
Table 5) [
107]. Al
3+-exchanged montmorillonite gave the best ML yield (61 mol%), due to the presence of a large number of acid sites and a good balance of Brønsted and Lewis acid sites. The montmorillonite catalyst was easily recovered from the reaction mixture by filtration and reused at least five times without any loss of activity/selectivity after treatment with H
2O
2 solution in order to remove carbon species deposited on the catalyst surface. Similar results were reported in another work with tin-exchanged montmorillonite catalysts (runs ML_125—ML_130,
Table 5) [
108]. A high ML yield (60–65 mol%) was obtained when simple monosaccharides, such as glucose or fructose, were adopted as the starting substrate, whilst it significantly decreased to 19 mol%, starting from the insoluble cellulose. In a work conducted byXu et al. [
109], montmorillonites were modified by H
2SO
4 treatment, thus introducing sulfate groups for improving their acidity and, therefore, catalytic activity for ML production from different model C6 feedstocks (runs ML_131—ML_135,
Table 5). Under the optimal conditions, the conversions of fructose and glucose were complete, whilst the corresponding ML yields were 65 and 48 mol% (runs ML_131—ML_132,
Table 5). However, also in this case cellulose conversion gave only moderate ML yield (24 mol%, according to run ML_135,
Table 5), which restricted the use of this catalyst to simpler soluble carbohydrates. Lastly, silica-pillared montmorillonites functionalized by iron-modified tungstophosphoric acid were prepared by Lai et al. [
110] and tested for fructose methanolysis. The characterization demonstrated the high dispersion and Keggin structure of HPWFe in the framework of the MMTSi. An optimized ML yield of around 74 mol% was obtained at 180 °C for 1 h (run ML_136,
Table 5), and the catalyst recovered after calcination was active after five recyclings. These catalysts showed advantageous porosity, tuned Brønsted–Lewis acidity, and high thermal stability, making its use promising if compared with other catalysts that belongto this group.
It is well-known that metal oxides can be converted into more stable solid-acid catalysts by the sulfation treatment [
111]. These modified oxides show good stability in solvents, also showing enhanced Brønsted and Lewis acidities. Their Lewis acidity is attributed to the metal atoms, whilst the Brønsted one derives from the hydroxyl groups that are located on the surface of the oxides [
112]. Generally, the hydroxyl groups on the surface of unmodified metal oxides have very weak Brønsted acidity, but, after having performed a sulfation treatment, the S–O bonds on the sulfuric groups could bind strongly to the metal atoms, forming a coordination of the S=O bond with the surface hydroxyl groups on metal oxides, thus improving the Brønsted acidity. Even these catalysts have been tested for ML production from C6 model carbohydrates, and
Table 6 reports the main available data.
Acidic TiO
2 (anatase) nanoparticles gave good ML yields, starting from different model carbohydrates (runs ML_137—ML_140,
Table 6) [
113]. Remarkably, also for the challenging case of cellulose conversion, the best ML yield was good (42 mol%), but this achievement required a very long reaction time (20 h). The good catalytic activity of the nanoparticles was attributed to the
in-situ solvation on the surface and their better dispersion in the reaction medium. Additionally, in this case, the catalysts showed high recyclability, with only a minor loss of performance. Peng et al. [
44] tested many heterogeneous acids (ZSM-5(25), ZSM-5(36), NaY, H-mordenite, Zr
3(PO
4)
4, SO
42−/ZrO
2, SO
42−/TiO
2, and TiO
2), identifying SO
42−/TiO
2 as the most promising one for ML production (runs ML_141—ML_145,
Table 6). Besides, SO
42−/TiO
2 avoided dimethyl ether formation, which, instead, significantly occurred, for example, with ZSM-5(25) and ZSM-5(36) catalysts, which led to a consumption of about half of the starting methanol, in favor of dimethyl ether. The heterogeneous SO
42−/TiO
2 catalyst was easily recovered by filtration and it exhibited good catalytic activities after calcination in five cycles of reusing. The surface structure and acidity variations of the fresh and recycled SO
42−/TiO
2 catalysts after calcination were characterized by XRD and NH
3-TPD techniques. The authors reported that the catalyst crystal structure was preserved after multiple cycles, but the amount and strength of the acid sites of the catalyst gradually decreased with the increase of consecutive recycling runs, due to the progressive loss of sulfur, mainly occurring by solvation during the alcoholysis and by calcination necessary for the catalyst regeneration [
44]. However, the low ML yield achieved with cellulose (10 mol%, according to run ML_145,
Table 6) demonstrated that a further modification of this catalyst was necessary for improving the catalysis. For this purpose, high surface area and thermally robust SO
42−/ZrO
2 conformal monolayers, with tunable Lewis–Brønsted acid site densities, were grown over a mesoporous SBA-15 template, via sequential grafting and hydrolysis cycles, employing a zirconium isopropoxide precursor [
114]. The enhanced low-temperature activity of grafted SZ/SBA-15 was attributed to the presence of strong Lewis acid sites that drove the glucose isomerization to fructose. Catalyst reusability was confirmed over three consecutive runs, performing the calcination of the spent catalyst at 550 °C to remove organic deposits, thus overcoming the extended leaching problems of commercial SO
42−/ZrO
2 catalyst. However, the optimized balance between the Brønsted–Lewis acidity and density of the acid sites was not enough for improving the ML catalysis, leading only to a moderate ML yield, starting from glucose and working under mild reaction conditions (run ML_146,
Table 6). Njagi et al. proposed further modification of these catalysts [
115], who synthesized sulfated mixed-metal oxides (SO
42−/TiO
2–ZrO
2) of high acidity, mainly ascribed to the sulfate species, and new acidic sites generated from the charge excess. This catalyst was suitable for the high yield obtained in the conversion of fructose or sucrose to ML (runs ML_147 and ML_ 149,
Table 6), but only low yields (23 mol%) were obtained from run ML_148,
Table 6),also confirming that this mixed catalyst was not efficient for the isomerization of glucose to fructose. However, the authors declared a high selectivity to ML, which was attributed to the presence of large mesopores, whilst the dimethyl ether formation was reported to be negligible. The spent catalyst was reused after a calcination step for removing insoluble humans from the surface, maintaining its catalytic performances almost unaltered. Another proposal of catalyst improvement was done by Jiang et al. [
83], who reported the maximum ML yield of about 60 mol% from fructose, by combining SO
42−/ZrO
2 and Sn-Beta zeolite (run ML_150,
Table 6). ML yields for more complex feedstocks are moderate (ML_151—ML_154,
Table 6), in agreement with those that were achieved with the zeolite-based catalysts. Sn-Beta zeolite has Lewis acidity for allowing the isomerization of glucose to fructose and poor catalytic activity for the retro-aldol reaction, whilst the Lewis acid sites of SO
42−/ZrO
2 cannot catalyze the isomerization of glucose to fructose. Recyclability studies indicated that the combined catalyst could be reused five times without a significant decrease in product yield, proving its easy recovery and thermal stability during regeneration. Ding et al. synthesized niobium phosphate catalysts [
18] and tested for ML production from cellulose, showing good performances in the conversion of cellulose (maximum ML yield of 56 mol%) (run ML_155,
Table 6). This investigation showed that the mechanism and type of intermediates of cellulose alcoholysis in MeOH were different from those in water and that the high Brønsted/Lewis acid ratio of these solid catalysts is needed in order to prevent the generation of by-products, in particular methyl lactate and 1,1,2-trimethoxyethane. Additionally, the heterogeneous system WS
2 was tested for fructose conversion. Quereshi et al. synthesized this catalyst [
116] in a tubular furnace (600 °C), while using elemental tungsten and sulfur, obtaining multilayered flakes/sheets of WS
2. Only a moderate ML yield was reached (run ML_156,
Table 6) and the catalyst resulted in being very stable and showed similar activity after five consecutive runs.
2.2. EL Synthesis from Model Carbohydrates
The synthesis of EL from C6 model carbohydrates have been widely investigated, and many catalysts have been tested in order to increase the EL yield. The interest in this ester is enhanced by the possible use of bioethanol as alcohol, thus obtaining a fully bio-derived product. As for the synthesis of ML ester, among the employed catalysts, mineral acids and metal salts represent the simplest and cheapest alternatives to sulfuric acid. The most promising data for these kinds of catalysts arereported in
Table 7.
Several works have investigated the production of EL from mono- and polysaccharides, preferring the use of diluted sulfuric acid (0.002–0.3 mol/L) (runs EL_1—EL_10,
Table 7) [
69,
117,
118,
119,
120,
121]. Temperatures that are higher than 170 °C and relatively short reaction times (below 2 h) have been generally adopted to achieve good EL yields. Taking into account the different feedstocks, as expected fructose led to the highest EL yield (about 70 mol%), due to the easier conversion of this feedstock, in the presence of Brønsted acids (runs EL_1, EL_4—EL_7,
Table 7) [
117]. A binary reaction medium composed of ethanol-glycerol was proposed for the sulfuric acid-catalyzed conversion of glucoseto inhibit the humin formationviathe use of non-aqueous green solvents, but an improvement of the EL yield was not achieved (run EL_3,
Table 7) [
119]. In addition, the available data for the conversion of C6 carbohydrates to EL with H
2SO
4, confirm that cellulose is the most recalcitrant substrate, leading to the lowest EL yields (runs EL_7—EL_8,
Table 7), except when higher catalyst concentrations (0.1–0.3 mol/L) were employed (runs EL_9—EL_10,
Table 7) and the reaction conditions were properly modulated [
69,
117,
120,
121]. Bernardo et al. proposed perrhenic acid [
72] as strong Brønsted homogeneous acid for the synthesis of EL (runs EL_11—EL_14,
Table 7).However, this catalyst was more expensive than H
2SO
4 and required longer reaction time to achieve interesting yields. Analogously to the reactions that were performed with H
2SO
4, the highest EL yield of 80 mol% was obtained starting from fructose after 16 h, whilst under the same conditions the yield from glucose was only 27 mol% and the extension of the reaction time up to 72 h was necessary to reach the same yield obtained starting from fructose. This catalyst was also tested for the EL synthesis from inulin and sucrose, achieving the yields of 65 and 52 mol%, respectively, demonstrating that HReO
4 can hydrolyze the glycosidic bonds of di- and polysaccharides. As reported for methanolysis, in addition to the mineral acids, the inorganic salts, in particular sulfates, have also been successfully employed for the EL synthesis from model C6 carbohydrates. Sun et al. [
73] investigated different inorganic salts, such as Fe
2(SO
4)
3, La
2(SO
4)
3 and Ce(SO
4)
3 as the catalyst for the alcoholysis of fructose and glucose, working at 170 °C for 2 h (runs EL_15—EL_18,
Table 7). It was found that Fe
2(SO
4)
3 was the best catalyst for the glucose conversion to EL, leading to the highest EL yield of 39 mol%. Moreover, the EL yield that was obtained starting from glucose was higher than that from fructose, proving that the different chemical structures of these two sugars were affected by the chelation with Fe
2(SO
4)
3. For the cellulose alcoholysis Al
2(SO
4)
3 was largely employed. Zhou et al. [
75] reported the maximum EL yield of 45 mol%, working at 180 °C for 5 h (run EL_19,
Table 7), whilst Huang et al. [
32] achieved the best EL yield up to 70 mol%, working at the same temperature, but only prolonging the reaction for 0.9 h under MW irradiation with an EtOH-water medium (runs EL_20 and EL_21,
Table 7). This yield value from a recalcitrant substrate is remarkable and analogous to that obtained by the authors in the methanolysis of cellulose under the same reaction conditions (compare run ML_29,
Table 1, with run EL_21,
Table 2). As already discussed for the ML synthesis, the water addition is generally advantageous for this reaction, leading to an increase of the AL yield and improving the kinetics, at the same time reducing the humin formation and solvent consumption to give the dialkyl ether.
Sulfonic acids and sulfonate salts have been employed as homogeneous or heterogeneous catalysts for the ethanolysis of C6 carbohydrates.
Table 8 reports the most interesting available data.
Regarding the organic salts, Bodachivskyi et al. [
122] studied the EL synthesis from cellulose, in the presence of metal triflates as catalysts (runs EL_22—EL_26,
Table 8). In particular, the authors proved that harder Lewis acids, such as Al(OTf)
3, In(OTf)
3, Sn(OTf)
2, and Hf(OTf)
4, were able to catalyze this reaction, reaching the maximum EL yield of 32 mol%, with Al(OTf)
3. The catalytic performance is ascribed to the Brønsted acidity that is generated from the harder Lewis acids, as a consequence of the complexation of the protic solvent with the metal center, rather than hydrolysis, with the latter being responsible for the Brønsted acidity of inorganic salts. On the other hand, softer Lewis acids, such as Y(OTf)
3, AgOTf, La(OTf)
3, and Yb(OTf)
3, showed low activity towards the direct conversion of cellulose to EL, but their combination with a Brønsted acid, such as H
3PO
4, significantly increased their catalytic activity, in particular for the combined catalytic system (Y(OTf)
3 + H
3PO
4) [
122]. This synergic effect derived from the favorable complexation of soft Lewis acid and H
3PO
4 increased the Lewis acid-assisted Brønsted acidity, leading to the highest EL yield of 75 mol% starting from microcrystalline cellulose. Regarding the other sulfonated systems, both commercial and
ad-hoc synthesized heterogeneous catalysts have been employed for the production of EL from C6 carbohydrates. For example, Zhang et al. [
123] prepared a sulfonic acid resin by the condensation of styrene and divinylbenzene (PSDVB-SO
3H), which was employed for the fructose ethanolysis, obtaining the maximum EL yield of 26 mol% (run EL_27,
Table 8). However, better EL results have been reported in the presence of commercial styrene-divinylbenzene acid resins. In fact, Liu et al. [
80] used the commercial Amberlyst-15 for the fructose ethanolysis, working at 120 °C for 24 h, achieving the highest EL yield of 73 mol% (run EL_28,
Table 8). Besides, the same authors prepared several sulfonated carbonaceous materials, such as poly(
p-styrenesulfonic acid)-grafted carbon nanotubes (PSSA-g-CNT), poly(
p-styrenesulfonic acid)-grafted carbon nanofibers (PSSA-g-CNF), benzenesulfonic acid-grafted CMK-5 (BSA-g-CMK-5), and benzenesulfonic acid-grafted carbon nanotubes (BSA-g-CNT), employing them under the same reaction conditions of the Amberlyst-15 (runs EL_31—EL_34,
Table 8) [
80]. The concentration of Brønsted acid sites for the synthesized catalysts decreased, as follows: PSSA-g-CNT > PSSA-g-CNF > BSA-g-CMK-5 > BSA-g-CNT, and the same the trend was observed for EL yield, which was the highest for the PSSA-g-CNT catalyst (84 mol%) and lowest for the BSA-g-CNT one (45 mol%). Moreover, PSSA-g-CNT gave a higher EL yield than that with the commercial Amberlyst-15, underlining that this synthesized catalyst was particularly efficient. Additionally, Gu et al. [
124] employed Amberlyst-15 as a catalyst for the fructose ethanolysis at 150 °C for 3.5 h, ascertaining the EL yield of 75 mol% (run EL_29,
Table 8). These catalytic performances were compared with those of several
ad-hoc synthesized sulfonated hyper-cross-linked polymers, working under the same reaction conditions, and the authors proved that the catalyst that was obtained from 4,4′-bis(chloromethyl)-1,1′-biphenyl as the precursor, was the most active, due to the higher acid density and surface area (run EL_35,
Table 8) [
124]. This catalyst (HDS-3.6) led to the maximum EL yield of 70 mol%, analogous to that ascertained with Amberlyst-15. The synthesized catalyst was also employed for the ethanolysis of other feedstocks, such as glucose, inulin, starch, and cellulose. The EL yields achieved starting from aldose sugars were lower than those obtained starting from ketose, due to the lack of Lewis acid sites in the adopted catalyst, which are of paramount importance for the isomerization step (runs EL_36—EL_39,
Table 8) [
124]. The same conclusion was reported by Ming et al. [
121] for the EL synthesis from cellulose, employing the commercial sulfonic acid resin D008 (run EL_30,
Table 8). This catalyst led to the EL yield of only 20 mol%, which was lower than that obtained under the same reaction conditions with the mineral acid H
2SO
4 (51 mol%, according to run EL_10,
Table 7) [
121]. The lower EL yield that was reported with the commercial resin was attributed to the expected mass transfer limitations occurring between the protons of the heterogeneous catalyst and the solid cellulose. The detrimental absence of Lewis acid sites for EL production was confirmed for other synthesized sulfonic acid, such as the 5-chloro-sulfonated hyperbranched poly(aryleneoxindole) catalyst (5-Cl-SHPAO), whichwas tested for the ethanolysis of several saccharides (runs EL_40—EL_44,
Table 8) [
84]. Under the optimized reaction conditions, the highest EL yield of 68 mol% was obtained from fructose and it decreased, as follows: fructose > sucrose > glucose >cellobiose> cellulose. Zhang et al. [
125] synthesized Fe-impregnated sulfonated carbon of high surface area, pore volume and –SO
3H density, testing it for the ethanolysis of fructose, glucose, sucrose, inulin, and starch (runs EL_45—EL_50,
Table 8). Additionally, in this case, the highest EL yield was obtained starting from fructose and the EL yield progressively decreased by converting aldoses of increasing complexity. The problem of the low alkyl levulinate yields obtained from aldose carbohydrates was partially overcome by Karnjanakom et al. [
126], who prepared different sulfonated carbon doped with metal oxides, which were tested for the sucrose conversion to EL (runs EL_51—EL_52,
Table 8). These catalysts showed higher activity than the corresponding undoped sulfonated carbon, underlining that the synergy between Lewis and Brønsted acidity, derived from oxides and sulfonic groups, respectively, was fundamental for this reaction. In particular, the sulfonated carbon doped with ZrO
2 (Zn-SC) showed the highest selectivity towards EL, due to its acidity, whichresulted sufficient to promote the reaction, but not excessive to catalyze also the humin formation. The employment of THF as the reaction co-solvent improved the EL yield from 60 to 72 mol%, because this low-polar solvent prevented the next conversion of the desired EL. The authors also exploited the ultrasound technology as an alternative heating system for promoting this reaction, which was proven to be particularly efficient in reducing reaction times. Bosilji et al. [
119] carried out the glucose ethanolysis in an (ethanol-glycerol) solvent, testing a heterogeneous catalyst prepared by hydrothermal carbonization of the same substrate (glucose) itself, through the addition of sodium borate. This is an interesting example, because the adopted feedstock is also the precursor of the catalytic system, in principle making the whole process cheaper and sustainable. The synthesized catalyst had a high specific surface area and led to the maximum EL yield of 37 mol% (run EL_53,
Table 8). Moreover, the catalytic performances were comparable with those of H
2SO
4 (34 mol%, run EL_3,
Table 7).
Polyoxometalates (POMs) are another type of emerging acid catalystadopted for the synthesis of EL from C6 model carbohydrates.
Table 9 summarizes the significant results obtained with these catalysts.
POMs, including simple HPAs and their salts, have good acidic properties. Different HPAs, such as H
3PW
12O
40 (run EL_54,
Table 9) [
127], HPW
4Mo
10O
x (runs EL_55—EL_56,
Table 9) [
128], and H
4SiW
12O
40 (run EL_57,
Table 9) [
91], were used for the EL synthesis from fructose, glucose and cellulose, obtaining satisfactory yields, in particular with the first two catalysts. In fact, analogously to the ML synthesis, H
4SiW
12O
40 strongly promoted the formation of the intermediate ethyl glucoside, leading to a corresponding yield up to 59 mol%and an ethanol conversion of 16 mol% to the diethyl ether by-product.However, the main drawback of the HPA employment is their good solubility in the reaction medium that complicates their separation/recycling. As previously observed for ML, most of the studies have dealt with HPA salts, where one or more protons were substituted with larger cations. In particular, as for the ML synthesis, the PW
12O
40−3 heteropolyanion is that preferred for studying the alcoholysis of saccharides. For example, Zhao et al. [
129] substituted a proton of the Keggin-type H
3PW
12O
40 with larger monovalent cations, such as K
+ (KH
2PW
12O
40) and Ag
+ (AgH
2PW
12O
40), thus decreasing the starting Brønsted acidity and making the catalyst insoluble (runs EL_58—EL_62,
Table 9). KH
2PW
12O
40 was identified as the best catalyst, leading to a similar maximum EL yield than that achieved with AgH
2PW
12O
40, and involving a cheaper synthesis with KCl precursor, instead of AgNO
3. Moreover, the authors proved that the addition of toluene strongly increased the EL yield from fructose, from 51 mol% (with pure EtOH) to 69 mol%, which was attributed to the EL extraction into the toluene phase, which prevented the product degradation. Good EL yields were also ascertained from inulin and sucrose, whilst once again unsatisfactory EL yields were obtained from glucose and cellulose. Similarly, Srinivasa et al. [
130] exchanged the protons of H
3PW
12O
40 with titanium, thus adding Lewis acid sites, and reached a maximum EL yield of 63 and 21 mol%, starting from fructose and glucose, respectively (runs EL_63—EL_64,
Table 9). Pinheiro et al. [
93] synthesized several tin salts of H
4SiW
12O
40, which were almost insoluble in an alcohol medium, and compared their performances towards the EL synthesis, starting from different saccharides (runs EL_65—EL_67,
Table 9). The authors demonstrated that Sn
2SiW
12O
40 was the most active catalyst, leading to promising EL yields starting from fructose, sucrose, and inulin, and the proton exchange by Sn
2+ had a beneficial effect on EL selectivity. The very high yield from sucrose (78 mol%), the disaccharide of fructose and glucose, was ascribed to a contribution of glucose unit, which after has been released on the catalytic site, is directly isomerized to fructose and is then converted to EL. However, when glucose was employed as the starting feedstock, the EL yield was very low (about 5 mol%), probably due to the significant formation of by-products from this substrate [
93]. Besides, the replacement of protons of HPAs with organic species was proposed by Fang et al. [
99], who synthesized different phosphotungstic acid-based solid hybrids, through the reaction between H
3PW
12O
40 and different pyridines. The 3-phenylpyridine-phosphotungstate (3-PhPyPW) resulted to be the most active catalyst, giving the best EL yield of 30 mol%, starting from fructose (run EL_68,
Table 9). Recently, ionic liquids havealso been employed for modifying the acidity, polarity, surface properties, and solubility of POMs. A noteworthy example is provided by Chen et al. [
96], who employed the ionic liquid
N,
N,
N′,
N′-tetramethyl-
N,
N′-dipropanesulfonic acid-1,6-hexanediammonium (TMEDAPS) for preparing the corresponding POM salt, [TMEDAPS]
3 [PW
12O
40]
2, which was tested for the ethanolysis of different substrates, achieving the highest EL yield of 80 mol% from fructose (runs EL_69—EL_74,
Table 9). Analogously, Song et al. [
97] proposed the use of the 1-(3-sulfopropyl)pyridinium as ionic liquid to synthesize the [PyPS]
3PW
12O
40 catalyst, which gave the best EL yield of 57 mol%, starting from cellulose (run EL_75,
Table 9).
The performances of acid zeolite-based catalysts have been studied for the synthesis of EL from C6 carbohydrates, as reported in
Table 10.
Zeolite HY was tested for the EL synthesis from fructose (run EL_76,
Table 10) [
131] and glucose (run EL_77,
Table 10) [
132] reaching yields of 53 and 39 mol%, respectively, under the optimized reaction conditions. The catalytic performances of zeolites H-β and H-USY, having different Si/Al ratios, were compared by Saravanamurugan et al. [
102] for the EL synthesis from glucose. Zeolites H-β (Si/Al ratio = 19) and H-USY (Si/Al ratio = 6) gave EL yields of 28 and 41 mol%, respectively (runs EL_78—EL_85,
Table 10). Zeolite H-USY (6) was effectively adopted for the ethanolysis of other mono- and polysaccharides, leading to similar EL yields (35–47 mol%). Xu et al. [
118] investigated the glucose alcoholysis in the presence of the zeolite USY as the catalyst, and compared its performances with those of H
2SO
4, in both cases working at 180 °C (compare run EL_86,
Table 10 with run EL_2,
Table 7). The authors obtained similar EL yields, but the zeolite USY gave the maximum EL yield (47 mol%) after 2 h, whereas H
2SO
4 needed of shorter reaction time (0.5 h) to obtainthe maximum EL yield (45 mol%), being this difference ascribed to mass transfer limitations occurring between the solid zeolite and the liquid phase. However, zeolite USY led to a remarkable higher selectivity to EL, whereas diethyl ether formation was almost negligible. The combined use of zeolites and other co-catalysts has been proposed to better tune the bulk Brønsted-Lewis acidities. For instance, Chang et al. [
133] proved that the addition of H
2SO
4 to the zeolite USY strongly improved the EL yield from glucose, which increased from 38 to 51 mol% (run EL_87,
Table 10). USY zeolite resulted in being efficient for the conversion of glucose to 5-ethoxymethylfurfural, but the stronger Brønsted acidity of H
2SO
4 enhanced the overall alcoholysis, improving the EL yield. The zeolites were also employed in combination with POMs. For example, Mulik et al. [
134] used Sn-β zeolite with ZrH
2PW
12O
40 for the synthesis of EL from glucose (run EL_88,
Table 10), demonstrating that this synergy was fundamental to achieving the best EL yields. The POM salt contributed to the catalysis with both Brønsted and Lewis acidity, but the latter was insufficient, and this lack is offset just by the Sn-Beta zeolite. In particular, the ratio ZrH
2PW
12O
40/Sn-β of 80/20 wt/wt, corresponding to the Brønsted/Lewis ratio of 3.7, gave the best EL yield (54 mol%). Lastly, cheap H-ZSM-5 zeolite was employed as support of high surface area for anchoring the H
3PW
12O
40, thus overcoming the solubility drawback of the latter (runs EL_89—EL_92,
Table 10) [
127]. However, after the immobilization, the catalytic activity of H
3PW
12O
40 decreased, and the highest EL yield of 43 mol% was achieved, starting from fructose, a lower value than that obtained in the presence of unsupported HPA (50 mol%) (run EL_54,
Table 9) [
127], although the advantage of this immobilized catalyst lies in its good recyclability. Moreover, this catalytic system was also successfully employed by the authors for the EL synthesis from glucose, sucrose, and inulin, achieving moderate EL yields.
Metal oxides, also in combination with zeolites, resulted in exploitable catalysts for the synthesis of EL from fructose and glucose, and
Table 11 reports the most interesting results.
Among the commercial metal oxides, TiO
2 nanoparticles gave promising results in the EL synthesis from fructose affording the best EL yield of 71 mol% (run EL_93,
Table 11) [
113]. Gu et al. [
124] studied the fructose ethanolysis in the presence of commercial MCM-41 and SO
42−/ZrO
2 catalysts, comparing their catalytic performances with those of synthesized sulfonated hyper-cross-linked polymers (HDS-3.6). Both commercial MCM-41 and SO
42−/ZrO
2 catalystsled to moderate EL yields, equal to 25 and 44 mol%, respectively (runs EL_94—EL_95,
Table 11), whilst the synthesized HDS-3.6 gave the higher EL yield of 70 mol% (run EL_35,
Table 8). The lower EL yields with MCM-41 and SO
42−/ZrO
2 catalysts were ascribed to the few acid sites and low surface area-pore volume, respectively. Other authors largely investigated the use of SO
42−/ZrO
2 for the alcoholysis of carbohydrates. For instance, Peng et al. [
135] compared the catalytic activities of SO
42−/ZrO
2, SO
42−/TiO
2, SO
42−/ZrO
2 –TiO
2, and SO
42−/ZrO
2–Al
2O
3 catalysts in the EL synthesis from glucose. SO
42−/ZrO
2 and SO
42−/TiO
2 exhibited the best activities towards the EL production with 29 mol% yield, but SO
42−/TiO
2 increased the formation of undesired diethyl ether, thus SO
42−/ZrO
2 resulted in being the best catalyst system for EL production (run EL_96,
Table 11). Zhang et al. [
136] encapsulated SO
42−/ZrO
2 into a mesoporous Al
2O
3 (SO
42−/ZrO
2 @Al
2O
3), obtaining a catalyst with superacid properties. The latter showed both Brønsted and Lewis acid sites, affording the EL yield of 37 mol% (run EL_97,
Table 11), which was stable after four recycling tests. Morales et al. [
114] grafted SO
42−/ZrO
2 on SBA-15. ZrO
2 monolayers were grafted on the SBA-15 surface and then the sulfation was carried out with a solution of H
2SO
4. The complete coverage of the SBA-15 surface was achieved with two layers of SO
42−/ZrO
2 and this catalyst had the best acid density and the appropriate Brønsted/Lewis acid ratio for EL synthesis from glucose in the presence of Lewis acid sites that promote glucose isomerization to fructose. The best EL yield of 31 mol% was obtained with such a catalyst, adopting milder reaction conditions (temperature of 140 °C) (run EL_98,
Table 11), but too long reaction time was involved. Song et al. [
132] realized the EL synthesis from glucose with an ordered mesoporous sulfonic acid functionalized ZrO
2/organosilica catalyst (SO
42−/ZrO
2 -PMO-SO
3H). This catalyst was characterized by the synergistic effect of super-strong Brønsted acid sites (SO
42−/ZrO
2 and –SO
3H) and medium Lewis ones (Zr
4+), and the EL yield of 42 mol% was obtained under the optimized reaction conditions (run EL_99,
Table 11). Babaei et al. [
137] prepared an alumina-coated mesoporous silica SBA-15 catalyst (Al
2O
3/SBA-15), obtaining the maximum EL yield of 58 mol% from fructose (run EL_100,
Table 11). When the amount of alumina was increased, the total acidity concentration increased and Lewis/Brønsted ratio decreased, and a lower yield of EL was ascertained, which was ascribed to the humin formation, which is favored by an excess of Brønsted acidity. Besides, the authors proved that the specific surface area of the final catalyst was higher adopting a lower amount of alumina, which guaranteed the better dispersion of the active sites, having a positive effect on the catalytic performances. Jorge et al. [
138] prepared SBA-15, KCC-1, MCM-41, and dendritic mesoporous silica nanospheres (DMSi), and the resulting catalyst was functionalized with sulfonic acid. The sulfonated KCC-1, MCM-41, and SBA-15 exhibited moderate catalytic activity towards glucose conversion, obtaining EL yields in the range of 14–19mol%, these resulting significantly lower than that observed with the DMSi-SA catalyst (EL yield of 62 mol%) (run EL_102,
Table 11). This catalyst also gave excellent EL yields for fructose and sucrose (EL yields of 83 and 90 mol%, respectively) (runs EL_101 and EL_103,
Table 11). The better performances of DMSi-SA catalyst were attributed to its high porosity, which led to a high number of exposed functionalizablesilanol groups. The use of metal oxides in combination with other catalysts was reported by Heda et al. [
139], who employed SnO
2 with the zeolite H-USY for the synthesis of EL from glucose. Additionally, in this study, the synergy between the different components was fundamental because SnO
2 had a high amount of strong Lewis acid sites for glucose-fructose isomerization, while H-USY afforded the proper amount of Brønsted sites. The remarkable yield of 81 mol% was achieved with the catalytic system SnO
2+ H-USY (run EL_104,
Table 11). Lastly, Quereshi et al. proposedtungsten disulfide [
116] for the MW-assisted EL synthesis from fructose, but only a moderate EL yield of 23 mol% was reached (run EL_105,
Table 11).
Ionic liquids have been also proposed as efficient catalysts for the synthesis of EL from C6 carbohydrates, but only a few works reported their use for this reaction, due to their high cost, multistep synthesis, and environmental concerns.
Table 12 reports the available data.
Saravanamurugan et al. [
140] investigated the catalytic performances of several ionic liquids with acidic functionalities for the ethanolysis of fructose and sucrose (runs EL_106—EL_117,
Table 12). In particular, they synthesized imidazolium-, pyridinium-, and ammonium-based SO
3H ionic liquids, having HSO
4- as the anion, and imidazolium-based ionic liquids having [NTf
2]
—, [OMs]
— or [OTf]
— as the counter anion. Among these catalysts, [BMIm-SO
3H][NTf
2] gave the highest EL yield of 77 mol% from fructose and 40–43 mol% from sucrose, indicating that only fructose moieties were promptly converted, whilst the isomerization of glucose to fructose was more difficult. In fact, under these reaction conditions, glucose was converted to ethyl-D-glucopyranoside, which is an important non-ionic surfactant that can find applications in cosmetics and pharmaceutical formulations. Amarasekara et al. [
141] adopted the Brønsted acid ionic liquid 1-(1-propylsulfonic)-3-methylimidazolium chloride ([PSMIm][Cl]) for the synthesis of EL from cellulose in a (water-EtOH) solution. The authors found that cellulose conversion in sole EtOH was low and highlighted the key role of water co-solvent for promoting the hydrolysis of cellulose to glucose. Once the glucose was formed in the reaction medium, [PSMIm][Cl] catalyzed the dehydration of glucose to HMF, passing through the isomerization to fructose and the synthesis of EL, with a yield of 19 mol% (run EL_118,
Table 12).
2.4. BL and Longer-Chain AL (PeL and HL) Synthesis from Model Carbohydrates
Table 14 reports the most interesting data related to the synthesis of BL, PeL and HL from model C6 carbohydrates.
Cellulose butanolysis in the presence of diluted sulfuric acid (about 0.2 mol/L) led to good BL yields, approximately 40 mol%, working under the typical reaction conditions already adopted for the previous shorter-chain ALs (runs BL_1-BL_2,
Table 14) [
69,
142]. The choice of Hishikawa et al. [
143] of using much higher concentrations of sulfuric acid (2.0–3.5 mol/L)improved the BL yield up to the maximum of about 60 mol% (runs BL_3-BL_4,
Table 14), but such high acid concentrations are not advantageous for the sustainability of the process, additionally causing to corrosion problems of the equipment. Therefore, diluted mineral acids must be certainly preferred, to avoid these important drawbacks. Taking into account the diluted acid approach, the best BL yield (40 mol%, according to runs BL_1—BL_2,
Table 14) is similar to the best one obtained for PL (41 mol%, according to run PL_1,
Table 13), whilst it is significantly lower for EL (51 mol%, according to EL_10,
Table 7) and especially for ML (70 mol%, according to ML_12,
Table 1), thus highlighting that the synthesis of ALs in high yield becomes progressively more difficultas the length of the alkyl chain increases. AL yield strongly depends also on the steric hindrance of the alkyl chain, as demonstrated by Démolis et al. [
144] forthe cellulose butanolysis. In this context, at first, the authors studied the liquefaction of cellulose in butanol isomers (1-butanol, iso-butanol,
sec-butanol,
tert-butanol) in the absence of the acid catalyst, carried out under supercritical conditions (300 °C) and with different reaction times (1 and 2 h), achieving cellulose liquefaction in the range 70–85%, in order to give soluble oligomers and polymers.Despite the similar liquefaction performances of the different butanol isomers, in the case of
tert-butanol, a significant loss of alcohol (50 wt%) occurred, with the concomitant pressure increase, due to the formation of gaseous products, as a consequence of the significant dehydration of
tert-butanol to iso-butene, thus anticipating that this butanol isomer is less attractive for developing this reaction. Subsequently, the authors investigated the acid-catalyzed alcoholysis of cellulose in 1-butanol under subcritical conditions, employing very diluted H
2SO
4 (0.05 mol/L), and optimizing the reaction temperature and time. The highest BL yield was achieved after 1 h at 200 °C (about 50 mol%), but a significant formation of dibutyl ether occurred (yield of 36 mol%). A reaction time of 0.5 h was found to be the best compromise for reducing the dibutyl ether yield (20 mol%), keeping the BL yield high (50 mol%, according to run BL_5,
Table 14). These reaction conditions were considered as those of reference to compare the reactivity of the butanol isomerstowards the cellulose butanolysis. Iso-butanol and
sec-butanol isomers gave corresponding BL yields of 45 and 13 mol% (runs BL_6—BL_7,
Table 14), whilst
tert-butanol onlyled to traces of BL. Therefore, both 1-butanol and iso-butanol primary alcohols showed better performances than
sec-butanol secondary one, due to the higher steric hindrance of the latter. In addition, from the complementary perspective of the by-product formation, 1-butanol and iso-butanol primary alcohols gave different dibutyl ether yields of 20 and 2 mol%, respectively, proving the higher reactivity of the linear isomer. Instead,
tert-butanol was rapidly dehydrated to gaseous olefins, with remarkable loss of this alcohol (about 80 wt%), once again confirming the poor attractiveness of this butanol isomer towards this reaction. Based on the above discussion, iso-butanol represents an excellent alcohol to use for the butanolysis, to develop a new kind of BL-based bio-products. Instead, if
sec- or
tert- BLs are desired, the alcoholysis route is not so appropriate, whilst, at the actual state of the art, their synthesis can be more effectively realized by adopting 1-butene or iso-butene as the corresponding alkylating agents and levulinic acid as starting feedstocks, as demonstrated by Démolise t al. [
151]. As for the previous ALs, the catalytic activity of a series of cheap metal sulfates was proposed by An et al. [
145], instead of the traditional sulfuric acid. Among the investigated metal sulfates, Fe
2(SO
4)
3 resulted in being particularly efficient for BL production from simpler model compounds, achieving the maximum yield of about 60 mol%, starting from simple C6 carbohydrates, and about 30 mol%, starting from the more recalcitrant cellulose feedstock (runs BL_8—BL_12,
Table 14). However, Al
2(SO
4)
3 was also efficient for this purpose, especially in combination with Fe
2(SO
4)
3 (run BL_13,
Table 14) [
146], allowing the enhancement of the BL yield up to the maximum value of about 40 mol%, thus reaching similar performances to those with the sulfuric acid, although much longer reaction times were required in the former case. As for the previous ALs, the use of Cs-exchanged HPAs was also proposed by Démolis et al. [
144] for BL synthesis from cellulose, but achieving unsatisfactory BL molar yields (run BL_14,
Table 14). Few other catalysts have been considered, such as 3-PhPyPW or zeolite H-USY (6), but the reported BL molar yields data, already starting from simple monosaccharides, were not particularly noteworthy (runs BL_15—BL_16,
Table 14) [
99,
102]. SO
42−/ZrO
2 was properly modified by Liu et al. [
147] to SO
42−/SnO
2-ZrO
2, which was used alone, or in combination with organic/inorganic acidsor sulfates. In particular, the combination of SO
42−/SnO
2-ZrO
2 with small amounts ofH
2SO
4, (HCOOH)
2, PTSA, Fe
2(SO
4)
3 or CuSO
4,, was effective for improving this reaction (runs BL_17—BL_28,
Table 14). Lastly, the good BL yield of about 30 mol% was reported by Ma et al. [
148] for the cellulose conversion in the acidic ionic liquid [C
4H
8SO
3Hmim]HSO
4 (run BL_29,
Table 14), which was identified as a stable and water-tolerant catalyst.
On the other hand, only very few works have studied the synthesis of PeL and HL from C6 model feedstocks, due to their lower reactivity. Yamada et al. [
149] carried out the synthesis of PeL and HL starting from cellulose and working at the boiling point of the respective solvent (runs PeL_1 and HL_1,
Table 14). Very interesting yields were ascertained for both PeL and HL, equal to 69 and 60 mol%, respectively, but the concentration of H
2SO
4 was too high (2.1 mol/L), so that the process could be considered sustainable. Additionally, heterogeneous catalysts have been adopted for the production of PeL and HL. Quereshi et al. [
116] carried out the alcoholysis of fructose under MW heating, employing WS
2 as the catalyst and PeOH as the solvent/reagent, anyway achieving the PeL yield of only 5 mol% (run PeL_2,
Table 14). On the other hand, Mohammadbagheri et al. [
150] reported the good HL yield of 55 mol% starting from glucose, in the presence of Zn/dendritic fibrous nanosilica as the catalyst (run HL_2,
Table 14), which resulted in being effective for this purpose, due to the balanced Brønsted/Lewis acidity.