Hexoses Biorefinery: Driving Glucose Dehydration over Sulfonic Polymer and Hybrid Acid Catalysts

Abstract

Glucose is the most abundant monosaccharide as it is the primary unit of cellulose and starch, which are the more relevant feedstocks for biorefineries. Dehydration of glucose can lead to anhydroglucoses, whose interest has been increasing due to its potential industrial use. Commercial sulfonic polymer resins and a synthesized organic–inorganic mesoporous material were taken as Brønsted acid catalysts. High hexose conversion (up to 98%) and selectivity to anhydroglucoses (~80%) could be reached, turning this process into an alternative route to carbohydrate pyrolysis that presents an energy-intensive downstream. Hexose conversion to anhydroglucoses was related to the amount of acid sites, and the removal of one molecule of water from hexoses to produce anhydroglucoses was found as the preferential dehydration route over a bare Brønsted acid catalyst in anhydrous polar aprotic solvent (DMF) at mild conditions. Product distribution changed dramatically upon catalyst deactivation with HMF and fructose emerging as relevant products. It was suggested that an additional Lewis surface is produced during the deactivation process, probably arising from the formation of insoluble high molecular weight compounds in acidic media.

1. Introduction

Lignocellulosic and starchy biomass are the major sources of renewable carbon [1,2], and therefore the saccharides released from these sources play a foremost role as feedstocks within the biorefinery concept. Glucose is the most abundant monosaccharide as it is the primary unit of cellulose and starch and as such, its conversion into biofuels, biofuel-related products and chemicals is a central issue [1].

Dehydration of glucose can lead to anhydroglucoses (1,6-anhydro-β-D-glucopyranose or levoglucosan, and 1,6-anhydro-β-D-glucofuranose) and 5-hydroxymethylfurfural (HMF), which are all highly valuable chemicals. In addition, there has been increasing interest in the industrial production of levoglucosan (LGA) due to its potential use in the synthesis of chiral polymers, surfactants, pharmaceuticals, propellants and resins [3,4]. LGA can be obtained by fast pyrolysis of biomass and it is indeed the most abundant primary product when cellulose or starch are taken as feedstocks [3,5,6,7]. However, if lignocellulose is used, a wide variety of other chemicals will also be produced, particularly furan derivatives and aromatics, due to the carbohydrate and lignin fractions in the biomass. Downstream separation and purification operations will thus be required and they may be very energy-demanding depending on the intended use of LGA, as in the manufacture of fine chemical products and medicines. Therefore, obtaining LGA should be more interesting either through pyrolysis or hydrolysis of beforehand fractionated cellulose. In the later approach, cellulose hydrolysis to glucose is followed by glucose dehydration to anhydroglucoses. While the highest yield of LGA from fast pyrolysis of cellulose achieves around 80% at 500 °C and a high heating rate [8], it steadily reaches about 70% in glucose dehydration at 140 °C [7]. In addition to requiring a much higher temperature, pyrolysis is also highly dependent on reactor configuration and the feedstock, which would explain the broad range of LGA yields (5–80%C) reported in the literature [4,7]. Increasing the yield of LGA from glucose dehydration can be the key to establishing an efficient low-energy-demanding process for its production from carbohydrates derived from agro-industrial waste and, as a consequence, help generate new markets for levoglucosan-derived products. This is particularly challenging though, considering that HMF may also be produced upon glucose dehydration. This is probably why only a few studies are reported in the literature regarding LGA production from glucose dehydration [9,10,11].

Production of HMF, on the other hand, has been much more extensively investigated over the last few years. HMF is a platform chemical taken to produce value-added products to pharmaceutical, polymers and transport fuel sectors [12]. This versatility arises from its furan ring, and an aldehyde and hydroxyl functional groups. Just like anhydroglucoses [9], dehydration of glucose to HMF is catalyzed by Brønsted acid sites. Nevertheless, the presence of Lewis acid centers is highly demanded in this case as they can first promote the isomerization of glucose to fructose, which is more efficiently converted to HMF over Brønsted sites [12,13,14,15,16,17]. However, running this reaction in water is rather challenging as HMF yields are usually very low (<10%) [18] and there are a large number of more energetically favorable side reactions, mainly condensation to humin precursors [12]. Hence, the uses of organic solvents and monophasic and biphasic solvent systems have been explored to overcome these challenges and increase HMF yields to acceptable levels. The extensive literature on these issues has been reviewed for different research groups over the last decade [12,16,17,19,20,21,22,23,24].

Controlling the removal of either one or three water molecules during glucose dehydration to drive the reaction towards anhydroglucoses or HMF is a challenge and has received much less attention. In this contribution, glucose dehydration was evaluated from this perspective. Commercial sulfonic polymer resins and a synthesized organic–inorganic mesoporous material were taken as Brønsted acid catalysts and the reaction was performed in an organic solvent. The polymeric catalysts have been successfully used for esterification [25,26], alkylation [27,28,29,30] and etherification [27,28,29,30] processes due to their strongly acidic character and open continuous pore network. However, the application of bare polymeric resins in glucose dehydration is limited as they hold only Brønsted acid sites, which do not favor the formation of HMF as glucose isomerization to fructose over Lewis acid sites is a critical step [12,15]. Furthermore, their thermal stability is restricted, generally up to 150 °C, while most bioproducts derived from glucose in acid-catalyzed reactions are produced above 170 °C. As for anhydroglucose production, some acidic resins have been shown to be active, rendering up to 70% yields [9]. Herein, in addition to the nature of the catalyst, the impacts of reaction temperature and different amounts of acid sites were also examined as well as the influence of hexose molecular structure. Stability was evaluated via consecutive runs in order to envisage the potential use of such acid catalysts in hexose biorefineries.

2. Materials and Methods

Macroporous strongly acidic sulfonic resins from DuPont (Wilmin, DE, USA) were used as polymer catalysts, namely Amberlyst 15 (A-15), Amberlyst 35 (A-35) and Amberlyst 36 (A-36).

An organic–inorganic hybrid mesoporous silica catalyst was synthesized using tetraethylorthosilicate (TEOS, Sigma-Aldrich, Saint Louis, MO, USA) as a silicon source. SBA-15 was firstly obtained by dissolving 2 g of Pluronic P123 (molecular weight = 5800 g mol⁻¹, Sigma-Aldrich) in 75 mL of HCl solution (2 mol L⁻¹) and adding 4.6 mL of TEOS. The solution was kept under stirring at 40 °C for 20 h and then transferred to a Teflon-lined stainless-steel autoclave and hydrothermally treated at 100 °C for 24 h. The obtained solid was filtered, extensively washed with water, dried overnight, and then calcined at 500 °C (1 °C min⁻¹) for 5 h and under synthetic air (50 mL min⁻¹) [31].

The propyl-sulfonic hybrid SBA-15 catalyst (SBA-15-PSO₃H) was prepared by a post-synthetic grafting procedure [32,33]. It was obtained by suspending 2 g of SBA-15 in 200 mL of HCl at 40 °C and adding 2.1 mL of 3-mercaptopropyltrimethoxysilane (MPTMS, Sigma-Aldrich) and hydrogen peroxide (6.3 mL). The suspension was kept under stirring for 7 h and then transferred to an autoclave and subjected to a hydrothermal treatment at 100 °C for 9 h. Next, it was cooled down to room temperature, filtered and washed with deionized water and ethanol and finally dried.

Porosity and acidity information of the commercial macroporous sulfonic resins (A-15, A-35 and A-36) was provided by the supplier [34,35,36]. Their morphology before and after reaction (fresh and spent samples) was scrutinized by field emission scanning electron microscopy (FE-SEM) using a Quanta FEG 450 FEI microscope (FEI, Hillsboro, OR, USA) operating with an accelerating voltage of 20 kV. Samples were spread on metal stubs covered with a carbon film before FE-SEM examination.

The long-range order of the mesoporous hybrid catalyst was assessed by small-angle X-ray diffraction (SAXRD) using a Bruker D8 Advance with DaVinci design diffractometer equipped with a Lynx-eye position sensitive detector (Karlsruhe, Germany). Diffractograms were collected with CuKα radiation within the 0.5–3° 2θ angular range at a rate of 0.01° s⁻¹. Porosity was evaluated by N₂ adsorption–desorption isotherms collected at −196 °C using Micromeritics ASAP 2020 equipment (Norcross, GA, USA). Surface area was calculated using the BET equation while the BJH method was used to determine pore volume (Vp) and diameter (ϕp) using the desorption branch. Morphology was surveyed by FE-SEM using the same microscope previously described for polymeric catalyst examination. However, the hybrid sample was spread on metal stubs covered with a carbon film and then sputter-coated with platinum before analyses. High-resolution transmission electron microscopy analyses (HR-TEM) were performed using a Jeol JEM-2100F microscope (Tokyo, Japan) with an accelerating voltage of 200 kV. Before analyses, sample powder was ultrasonically dispersed in 2-propanol and deposited on a carbon-coated copper grid (5 nm). Sulfur content was determined by X-ray fluorescence spectrometry (XRF) on a Bruker S8 Tiger spectrometer (Karlsruhe, Germany). Thermal stability of grafted sulfonic groups was studied by thermogravimetry using TA Instruments SDT Q600 equipment (New Castle, DE, USA) attached to an Ametek Dymaxion quadrupolar mass spectrometer (TG-MS) (Berwyn, PA, USA). The catalyst was heated up to 500 °C following a heating rate of 20 °C min⁻¹ under synthetic air flow. Signals at m/z = 18, 44, 48, and 64 corresponding to H₂O⁺, CO₂⁺, SO⁺ and SO₂⁺, respectively, were registered simultaneously.

Glucose dehydration was studied under mass-transfer free conditions in a mechanically agitated 300 mL stainless steel batch reactor model 4566 from Parr Instrument Company (Moline, IL, USA). N,N-dimethylformamide (DMF, 99.8%, Sigma-Aldrich cat# 227056) was used as a prime solvent for the liquid-phase reactions although pure water was also occasionally used. Following previous literature [11], 80 mL of 50 mmol L⁻¹ glucose solution and 0.8 g of catalyst were loaded in the reactor, pressurized with nitrogen at 10 bar and heated at 130 or 150 °C (5 °C min⁻¹) by a reactor-attached PID-controlled furnace. Reactions were performed for 3 h and measured after the reaction temperature was reached. Afterwards, the reactor was cooled down to room temperature with the aid of an ice bath in which it was kept for 40 min. Reagent consumption and product concentration were determined by HPLC on a Waters Alliance 2695 chromatograph (New Castle, DE, USA) using a BioRad HPX-87H ion exclusion column (300 × 7.8 mm, cat# 1250140) (Hercules, CA, USA). A H₂SO₄ aqueous solution at 5 mmol L⁻¹ was used as the mobile phase at 0.7 mL min⁻¹. The HPLC column was kept at 65 °C and the refractive index detector (Waters model 2414) (New Castle, DE, USA) at 50 °C. Multiple reaction runs (3 to 5 run replicates) were performed and experimental data presented standard deviation values of up to 2%. Besides glucose (99.5%, Sigma-Aldrich cat# G7528), fructose (99%, Sigma-Aldrich cat# F0127), mannose (99%, Sigma-Aldrich cat# M6020), levoglucosan (99%, Sigma-Aldrich cat# 316555) and maltose (99%, Sigma-Aldrich cat# M5885) were eventually used as feedstock.

Conversion of hexoses (X), selectivity to products (S) and carbon balance (CB) were calculated by Equations (1)–(3), as shown below:

$$X\left(\%\right)={\displaystyle \frac{C_0-C_t}{C_0}}\times 100$$

(1)

$$S\left(\%\right)={\displaystyle \frac{C_{prod}}{C_0-C_t}}\times 100$$

(2)

$$CB\left(\%\right)={\displaystyle \frac{n{C}_t+\sum {nC}_{prod}}{n{C}_0}}$$

(3)

where C₀ is the initial concentration of hexoses, C_t is the concentration of hexoses at time t, C_prod is the concentration of products formed and n is the number of carbon atoms in the chemical molecule. Kinetic constants were estimated considering a first-order reaction. It is worth mentioning that the carbon balance took into consideration all products identified in HPLC analyses.

Catalyst stability was evaluated by carrying out subsequent reaction runs. After each run, the catalyst was recovered by filtration, washed and used again with a new glucose solution under the same reaction conditions.

3. Results and Discussion

3.1. Characterization of Heterogeneous Catalysts

Three different commercial crosslinked polystyrene-co-divinylbenzene macroreticular-type sulfonic acid resins were used as catalysts. A-15 is a monosulfonated resin with a high crosslinking degree, and A-35 and A-36 are over-sulfonated with high and medium crosslinking degrees, respectively. It means that while A-15 holds one sulfonic (-SO₃H) acid group per styrene/DVB unit, both A-35 and A-36 are sulfonated at a higher level with more than one -SO₃H group per benzene ring [37]. Their total acidity and typical physical properties, as provided by the supplier, are summarized in

Table 1. Surface area (S_BET), average pore diameter (ϕp), total pore volume (Vp) and concentration of acid sites of commercial polymer and synthesized hybrid catalysts.

Catalyst	SBET (m2 g−1)	ϕp (Å)	Vp (cm3 g−1)	Acidity (mmol H+ g−1)
A-15	53	300	0.40	4.7
A-35	50	300	0.35	5.2
A-36	33	240	0.20	5.4
SBA-15-PSO3H	579	46	2.96	0.9

. Despite its lower surface area, pore volume and diameter, A-36 exhibits the highest acidity.

The morphology of commercial polymer catalysts was examined by FE-SEM and the images confirmed they were submicron spheres with smooth surfaces and narrow size distribution (Figure 1a–c).

As for the synthesized organic–inorganic SBA-15-PSO₃H hybrid catalyst, the small-angle X-ray diffraction pattern (Figure 2a) confirmed the typical mesoporous-ordered two-dimensional hexagonal (P6mm) organization of SBA-15 materials, exhibiting three peaks at a low Bragg angle related to the (100), (110) and (200) indexed planes [38]. Consistently, porosity analysis disclosed a type IVa isotherm with H1-type hysteresis (Figure 2b) as expected for the 1D cylindrical large-pore channels of an SBA-15 [38]. Porosity data are also shown in Table 1 and they demonstrate that the hybrid catalyst is a high surface area material (579 m² g⁻¹). This value is lower than those reported previously for a bare SBA-15 [39] and this decrease is a consequence of the surface post-synthetic functionalization with organosulfonic groups, which restricts the pore network. Nevertheless, it must be mentioned that porosity is line with other similar hybrid SBA-15 materials reported in the literature [32,39,40]. FE-SEM imaging showed that hybrid mesoporous catalysts present long rod-like aggregates (Figure 2c) while HR-TEM examination displayed the typical formation of well-organized honeycomb-like hexagonal arrangement of the cylindrical pore network (Figure 2d).

The acidity of SBA-15-SO₃H was estimated by sulfur loading as determined by XFR (Table 1) as it is a direct assessment of Brønsted acid sites. The hybrid catalyst exhibited a much lower concentration of acid sites, which are also weaker than those bounded to the polymer resins [41]. On the other hand, they are more thermally stable as unveiled by the thermogravimetric analysis coupled to a mass spectrometer (TG-MS) presented in Figure 3.

Three distinct events were observed on the thermogravimetric curves as distinguished by the temperature (Figure 3, curves a and b). The first one up to 120 °C is related to a minor mass loss (<4%) and is associated with the presence of water in the catalyst. The second and third events were recorded within 250–400 °C and 450–600 °C. It was found, with the aid of an on-line mass spectrometer, that these mass losses led to the release of both CO₂ and SO₂ (Figure 3, curves c and d). Therefore, it is conceivable to conclude that the organogroups grafted on the SBA-15 surface are burned-off at this temperature range as previously reported elsewhere [39]. These findings thus reveal that the SBA-15-PSO₃H hybrid catalyst is also sensitive to temperature and can only be used up to 250 °C [39,42]. Nevertheless, it offers a much wider operating temperature range than the commercial polymer catalysts, whose structure is generally compromised from 120 to 150 °C as reported by the supplier [34,35,36].

3.2. Catalytic Performance and Kinetics

All polymer and organic–inorganic hybrid acid catalysts were initially screened for glucose dehydration in anhydrous DMF (0.06% H₂O as determined by Karl Fischer titration). The reaction data depicted in Figure 4 clearly show the crucial role played by the acid catalysts, especially compared to the negligible conversion of glucose achieved in the control experiment ran without any catalyst. Furthermore, this control experiment evidences that the temperature used (130 °C) is not enough to promote undesired thermal homogenous reactions. It is indeed in line with previous literature reporting thermal dehydration of glucose typically only above 300 °C [43].

Differences in the overall activity as per catalyst mass (Figure 4) are probably related to the differences in their acidity. Despite this, it is notable that they all promoted the preferential formation of anhydroglucoses (sum of 1,6-anhydro-β-D-glucopyranose and 1,6-anhydro-β-D-glucofuranose), reaching selectivity and yield up to around 80% and 60%, respectively, over A-15. Only minor formations of HMF, fructose and formic acid (FA) were detected alongside due to side reactions. Notwithstanding, it is noted that high carbon balances were obtained (>80%) as calculated from the HPLC analyses.

New experiments were performed at the same concentration of acid sites by adjusting the catalyst mass in each run considering their acidity (Table 1) to determine the kinetic constants and examine selectivity to anhydroglucoses (AG) at the same conversion level (30–40%). These data are summarized in

Table 2. Kinetic data and selectivity to anhydroglucoses (S_AG) for glucose dehydration in anhydrous DMF at 130 °C over commercial polymer and synthesized hybrid catalysts. Concentration of acid sites = 0.73 mmol H⁺ g⁻¹.

Catalyst	k (h−1)	SAG (%)	Carbon Balance (%)
A-15	0.26	76	90
A-35	0.21	78	95
A-36	0.21	70	90
SBA-15-PSO3H	0.14	48	82

.

A very similar kinetic performance was achieved over all sulfonic polymer catalysts, indicating that differences in pore volume and surface area are not important, most likely due to their macroporous texture. The acid strength also seems to lack importance as they are different, with A-35 and A-36 holding stronger acid sites than A-15 [37]. Such a difference in acid strength is a consequence of the over-sulfonation of A-35 and A-36 resulting in the activation of sulfonic acid groups as those resins hold more than one acid group on some benzene rings in the polymer network [37]. The lower activity of SBA-15-PSO₃H, however, would suggest that much weaker acidity would lead to less active catalysts. All in all, it is concluded that glucose conversion is related to the amount of acid sites and the reaction follows the same mechanism over all the studied polymer catalysts whereby one molecule of water is released from glucose leading to anhydroglucoses. No other dehydration product or chemicals derived from further acid-catalyzed reactions were identified. Catalyst amount can thus be taken to drive conversion and yield of anhydroglucoses in glucose dehydration in anhydrous reaction medium and, therefore, reaction process configuration may take a prime role in a potential technological application.

The impact of reaction temperature was assessed using only A-35 and A-36, and up to 150 °C due to the thermal stability restrictions of these polymer catalysts [34,35,36], or SBA-15-PSO₃H hybrid catalyst at higher temperatures (190 °C) as it was found to be much more thermally resistant (Figure 3). Raising the temperature led to an expected increase in glucose conversion, reaching over 85% (

Table 3. Catalytic tests carried out with different substrates and solvents. Conversion (X), selectivity to anhydroglucoses (S_AG), glucose (S_glucose), fructose (S_fructose), mannose (S_mannose), HMF (S_HMF) and formic acid (S_FA), and carbon balance for reaction runs at 150 °C over A-36 catalysts.

Entry	Catalyst	Hexose	Solvent	X (%)	SAG * (%)	Sglucose (%)	Sfructose (%)	Smannose (%)	SHMF (%)	SFA (%)	Carbon Balance (%)
1	No catalyst	Glucose	DMF	10	0	-	39	0	0	0	94
2	A-35	Glucose	DMF	85	56	-	0	0	6	4	75
3	A-36	Glucose	DMF	90	58	-	0	0	6	6	73
4	A-36	Fructose	DMF	98	0	0	-	0	82	3	86
5	No catalyst	Fructose	DMF	20	0	13	-	0	0	12	85
6	A-36	LGA	DMF	19	19 **	0	0	0	0	19	85
7	A-36	LGA	water	97	-	91	0	2	4	3	99
8	A-36	Glucose	water	13	33	-	0	0	0	0	93
9	A-36	Mannose	DMF	94	44 ***	0	0	-	4	4	55
10	No catalyst	Mannose	DMF	13	0	13	0	-	0	0	88
11	A-36	Maltose	DMF	84	33	7	0	0	5	2	55

* anhydroglucose (AG) is the sum of 1,6-anhydro-β-D-glucopyranose (LGA) and 1,6-anhydro-β-D-glucofuranose; ** selectivity to only 1,6-anhydro-β-D-glucofuranose as LGA (1,6-anhydro-β-D-glucopyranose) was used as substrate; *** 1,6-anhydro-β-D-mannopyranose.

, entries 2 and 3). The role played by the acid catalysts is highlighted once again at this higher temperature when comparing their performances with the control experiment carried out without any catalyst, whose conversion reached only 10% rendering fructose as the main product (Table 3, entry 1). However, such an increase in temperature caused a negative impact on product distribution as anhydroglucose’s selectivity dropped to around 57% over either A-35 or A-36 (Table 3, entries 2 and 3). Indeed, the formation of humins and acid- and thermal-catalyzed condensation, degradation and decomposition reactions are known to take place upon saccharide conversion at higher temperatures [44,45,46]. The occurrence of these reactions would explain the lower carbon balance accomplished in these runs. Besides anhydroglucoses and formic acid, HMF is now observed among the dehydration products at minor levels. Its formation is in line with the thermal isomerization of glucose to fructose, as evidenced by the control experiment (Table 3, entry 1), which is known to easily dehydrate to HMF over Brønsted acid sites [12,15]. Indeed, when fructose was used as a substrate in DMF, nearly total conversion was reached within 3 h rendering 82% selectivity to HMF (Table 3, entry 4). The control experiment without a catalyst (Table 3, entry 5) indicated that only isomerization is promoted by reaction temperature. Any contribution to HMF formation from cascade dehydration of anhydroglucoses in entries 2 and 3 could be disregarded as LGA conversion is rather low and would only form formic acid when used as a substrate under the same reaction conditions (Table 3, entry 6). This low reactivity might be related to the use of anhydrous DMF as a polar aprotic solvent in close agreement with some previous report on its negative impact on LGA conversion over A-70 [47]. Solvents have been reckoned to chemically interact with sugars intervening in their conversion to bioproducts [47]. The clear differences in the conversion levels of glucose and fructose (≥90%), and LGA (19%) over A-36 in DMF experimentally ratify it, highlighting the importance of the molecular structure of these sugars in such interaction as expected. Running the reaction in water allows almost all LGA to be converted to glucose (Table 3, entry 7), indicating that the acid-catalyzed hydrolysis reaction is predominant under those conditions and the reaction does not go further to other dehydration products. It was indeed substantiated when the glucose reaction was carried out in water (Table 3, entry 8).

Finally, it is worth mentioning that levoglucosenone (LGO) was not identified in any run, which is in line with other studies at which LGO was only produced from LGA at 170 °C when H₂SO₄ was used as a homogeneous acid catalyst in THF [48,49] or over A-70 in DMSO [47], or at an even higher temperature (210 °C) over acid-functionalized silicas in THF and THF/water mixtures [50]. These conditions are much more severe than those applied in the present study. SBA-15-PSO₃H was then used to explore harsher reaction conditions. Nevertheless, undesired degradation/polymerization products were mainly formed when the reaction was carried out at 190 °C, resulting in a strong dark suspension and making it difficult to undoubtfully verify and quantify the formation of LGO and other dehydration products. It is notable that very low carbon balances (as low as 25%) are indeed reported for catalytic dehydration of LGA to LGO and high reaction temperatures are usually demanded [51], demonstrating the unsuitability of producing this dehydration product straight from LGA.

The difference in product distribution must be noted when glucose (Table 3, entry 3) or fructose (Table 3, entry 4) are used as feedstocks. Fructose is not dehydrated to anhydroglucoses, demonstrating a different mechanism prompted by the differences in the aldose and ketose molecular structures. To explore this mechanism, mannose, a glucose epimer, was also used as an aldose substrate (Table 3, entry 9) to investigate the formation of the corresponding 1,6-anhydro-hexopyranose. A high conversion level (94%) was reached and, similar to glucose, 1,6-anhydro-β-D-mannopyranose was the main dehydration product over A-36, while only epimerization to glucose was promoted by reaction temperature without a catalyst (Table 3, entry 10). These data collected in DMF expand the previous conclusion drawn for aqueous solution that glucose and mannose react similarly [52,53]. The much lower carbon balance, however, indicates that degradation of mannose to humins is favored, in line with former kinetic studies reporting that the activation energy for degradation is lower than for mannose dehydration reaction [53].

To exploit a more complex feedstock and explore the possibility of processing a broader range of waste biomass in biorefineries, particularly starchy biomass, dehydration of maltose was also evaluated. Maltose is a disaccharide at which two glucose molecules are linked by an α(1 → 4) glycosidic bond and can be taken as the key structural moiety of starch. Its dehydration in DMF over A-36 achieved high conversion (84%) and anhydroglucoses were the major products (Table 3, entry 11), showing that acid-catalyzed cascade reactions of maltose hydrolysis and glucose dehydration were efficiently promoted.

It should be mentioned that all those reactions were performed in anhydrous DMF, a high boiling solvent (153 °C) widely used in organic synthesis and chemical reactions, including a previous study on glucose dehydration to anhydroglucoses [9]. As solvent recovery and reuse are key issues for technological applications, the performance of the A-15 acid catalyst was also evaluated in THF, which has a much lower boiling point (66 °C) and is also broadly used in saccharide conversion reactions. Since the solubility of glucose in THF is very low, making it impossible to maintain the same initial concentration of glucose in the feed solution in this study, a THF/water mixture was used and, consequently, the impact of water on the reactions carried out in DMF was also verified (

Table 4. Glucose dehydration carried out in different solvents. Conversion of glucose (X), selectivity to anhydroglucoses (S_AG), fructose (S_fructose), HMF (S_HMF) and formic acid (S_FA), and carbon balance for reaction runs at 130 °C over A-15 catalyst.

Entry	Solvent	X (%)	SAG * (%)	Sfructose (%)	SHMF (%)	SFA (%)	Carbon Balance (%)
1	THF + 20%vol.H2O	15	35	0	21	11	95
2	DMF	78	75	2	5	5	90
3	DMF + 2.5%vol.H2O	58	13	13	3	10	62

* anhydroglucose (AG) is the sum of 1,6-anhydro-β-D-glucopyranose (LGA) and 1,6-anhydro-β-D-glucofuranose.

).

Catalyst activity in THF/water was very low (Table 4, entry 1) compared to that achieved in DMF under the same reaction conditions (Table 4, entry 2). Moreover, despite the formation of anhydroglucoses, glucose dehydration to HMF and degradation to formic acid were also favored. The impact of water in activity was clear in DMF as well, since glucose conversion dropped even when a much lower amount of water (2.5%vol.) was added into the reactor (Table 4, entry 3). Addition of water to DMF led to a marked decrease in the formation of anhydroglucoses as could be anticipated.

Catalyst loading and reaction time were also explored and the results for reactions carried out at 130 °C are summarized in

Table 5. Catalytic tests carried out with different catalysts, catalyst mass and reaction time. Conversion of glucose (X), selectivity to anhydroglucoses (S_AG), fructose (S_fructose), HMF (S_HMF) and formic acid (S_FA), and carbon balance for reaction runs at 130 °C.

Entry	Catalyst	Mass (g)	Time (h)	X (%)	SAG * (%)	Sfructose (%)	SHMF (%)	SFA (%)	Carbon Balance (%)
1	A-15	0.16	3	36	63	5	5	0	87
2	A-15	0.80	3	78	75	2	5	5	90
3	A-35	0.14	3	31	66	5	7	0	89
4	A-35	0.80	3	63	78	3	5	7	95
5	A-36	0.14	3	31	66	6	7	0	89
6	A-36	0.80	3	63	70	3	5	7	90
7	A-15	0.80	6	85	72	2	5	4	84

* anhydroglucose (AG) is the sum of 1,6-anhydro-β-D-glucopyranose (LGA) and 1,6-anhydro-β-D-glucofuranose.

. As expected, an increase in the amount of catalyst led to an increase in glucose conversion with a less pronounced increase in anhydroglucose selectivity irrespective of the polymer acid catalyst (Table 5, entries 1–6). Likewise, higher conversion was reached by performing the reaction for a longer time (Table 5, entry 7). As a consequence, higher yields in anhydroglucoses could be achieved with either approach, leaving room to optimize the hexose/catalyst mass ratio and operation timeframe, and thus establish a more economically feasible process.

These findings can support the development of new processes to convert hexoses to levoglucosan from lignocellulosic and starchy waste biomass, allowing an alternative route to pyrolysis that presents an energy-intensive downstream separation step. In this sense, catalyst reusability is an important asset and it was thus evaluated herein by running consecutive runs. Considering the similar behavior of polymer catalysts (Table 2), A-15 was used in this assessment and reaction runs were carried out at a lower temperature (130 °C) as it was enough to render a high anhydroglucose yield. Figure 5 shows that the catalyst undergoes progressive deactivation as glucose conversion drops from 78% to 25% within three reaction cycles. Product distribution also changes dramatically as anhydroglucose selectivity gradually decreases while fructose and HMF formation increases, reaching the highest selectivity levels in this study (10% and 20%).

It is an odd pattern since the formation of fructose and HMF would require Lewis acidity to promote glucose isomerization and then dehydration. It was shown, however, that the input of thermal energy (through reaction temperature) was not sufficient to significantly promote this isomerization reaction step. Since the catalyst integrity was preserved, as examined by FE-SEM (Figure 1d,e), these results lead to the suggestion that the catalyst deactivation process generates an additional Lewis surface, probably arising from the formation of insoluble high-molecular-weight compounds in acidic media, as reported elsewhere [47]. Further studies on the acid chemical nature and possible isomerization activity of those insoluble carbonaceous solids may lead to some breakthroughs in biomass conversion.

4. Conclusions

The dehydration of different hexoses with distinct molecular configurations, both aldoses (glucose and mannose) and ketose (fructose), was studied over commercial acid resins and a hybrid organic–inorganic mesoporous acid catalyst. High hexose conversion (up to 98%) and selectivity to anhydroglucoses (~80%) could be reached and such levels can turn this process into an alternative route to carbohydrate pyrolysis that presents an energy-intensive downstream separation step. It was concluded that hexose conversion to 1,6-anhydrohexoses is related to the amount of acid sites on macroporous and mesoporous catalysts with no accessibility constraints. Other potential dehydration products, such as HMF and levoglucosenone, were rarely obtained up to 150 °C, demonstrating that the release of one molecule of water from glucose is the preferential dehydration route over a bare Brønsted acid catalyst in anhydrous polar aprotic solvent (DMF) at mild conditions. Glucose and mannose were found to react similarly, leading to 1,6-anhydro-β-D-glucoses and 1,6-anhydro-β-D-mannopyranose, respectively, while fructose rendered HMF as the major product. This contribution can support the development of innovative process for biomass conversion, expanding it to starchy waste biomass, as acid-catalyzed cascade reactions of maltose hydrolysis and glucose dehydration cascade were also efficiently promoted.