**Contents**



## **About the Editor**

**Hieronim Maciejewski** received his M.Sc. (1986) from Poznan University of Technology, ´ Poland and his Ph.D. (1995) and D.Sc. (2004) from Adam Mickiewicz University in Poznan, Poland. ´ Since 2014, he has been a full professor of chemistry. He is the head of the Department of Chemistry and Technology of Silicon Compounds at the Faculty of Chemistry at Adam Mickiewicz University in Poznan. Prof. Maciejewski is also the vice president of the Management Board of A. Mickiewicz ´ University Foundation and the deputy director of Poznan Science and Technology Park. He is the ´ author or co-author of over 180 publications, 90 patents, 11 licenses and 42 technologies as well as 12 book chapters and 1 book. Prof. Maciejewski has participated in many scientific and research projects and co-operated with various industrial partners. His research activity is focused on catalysis with organometallic and inorganometallic compounds (also with the use of ionic liquids). In addition, prof. Maciejewski is interested in properties and applications of inorganic polymers and hybrid materials (especially organosilicon compounds).

### *Editorial* **Ionic Liquids in Catalysis**

**Hieronim Maciejewski 1,2**


Ionic liquids play a larger and larger as well as more and more diversified role in catalysis [1–4], which is reflected by the fact that the number of literature reports on this subject increases annually by over 400 items. In the case of catalytic reactions, the ionic liquids enable us to obtain higher selectivities and yields and, first and foremost, they allow easy isolation of catalysts from the post-reaction mixture. Until now, the most popular application of ionic liquids is their use as a solvent and immobilizing agent. Their low vapor pressure and high thermal stability make them an alternative to classic organic solvents (the so-called green alternative). Nevertheless, at the beginning of this century, some critical voices were raised that pointed to the often complicated synthesis of the ionic liquids, which was also waste-generating and expensive. The shortcomings of ionic liquids also included their low biodegradability (or its absence) and sometimes their toxicity [5–7]. This is why a search has been made for the development of synthesis methods with a limited amount of waste (or waste-free ones), based first of all on the use of biodegradable and nontoxic starting materials, often of natural origin [8–11]. Due to the selection of new methods of synthesis, as well as the application of renewable and biodegradable raw materials, a larger and larger group of derivatives is considered safe and environmentally friendly, and the application trends include more and more areas. At present, one can notice a keen interest in two areas of the applications of ionic liquids. One of them is biotransformation (in a broad sense) that is used in the processing of biomass and lignocellulosic waste to obtain valuable chemical products [12–14]. The other widely developing trend is obtaining heterogeneous catalytic systems with the participation of ionic liquids, i.e., SILPC or SILCA (Supported Ionic Liquids Phase Catalysts or Supported Ionic Liquids Catalysts) [15–17] and SCILL (Supported Catalysts with Ionic Liquid Layer) [15,18–20] as well as catalysts in which ionic liquid makes a structural element [21–24]. In systems of this type, all problems observed in the case of catalysis in homogeneous systems are eliminated. First of all, there is no problem with the isolation of catalyst which is heterogeneous. Since in the case of SILPC and SCILL, the ionic liquid covers the supports with a thin layer, a small amount of the ionic liquid is needed which results in a considerable cost reduction (despite the high price of ionic liquids). Moreover, due to the thinness of the layer, the liquid viscosity does not influence the process course and mass transfer is not a problem. Such a form of catalyst can be applied in the continuous flow reactors what also influences the process economics.

The subject of papers published in this Special Issue reflects the two dominating trends of interest. The first one concerns biomass processing and the synthesis and application of ionic liquids from biorenewable resources. Parajo et al. presented an effective (one or two-step) method for the processing of eucalyptus wood towards producing furfural and levulinic acid with the use of acidic ionic liquid (C3SO3Hmim)HSO3 as a catalyst [25]. Bengoa et al. described the processing of cellulose (obtained from municipal and industrial sewage) into valuable levulinic acid, while also using a Brønsted acidic ionic liquid as a catalyst [26]. During the conversion of cellulose to levulinic acid, 5-hydroxymethylfurfural (5-HMF) is formed as a by-product. By rehydration reaction of the latter, it can be converted into levulinic acid. The process of cellulose degradation towards 5-HMF and glucose in the

**Citation:** Maciejewski, H. Ionic Liquids in Catalysis. *Catalysts* **2021**, *11*, 367. https://doi.org/10.3390/ catal11030367

Received: 8 March 2021 Accepted: 9 March 2021 Published: 11 March 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the author. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

presence of DES (deep eutectic solvents based on oxalic acid/choline chloride) and metal chlorides (CrCl3 or SnCl4) has been presented by Zhang's research group [27]. Brønsted acidic ionic liquids based on trimethylamine and sulfuric acid were employed as a solvent and catalyst in the synthesis of bis(2-ethylhexyl) terephthalate which is used as a plasticizer in the manufacture of plastics coming into contact with food. This new approach, which is characterized by the replacement of conventional acids and organometallic compounds with ionic liquids, was described by Chrobok's research group [28]. Bouquillon et al. presented the synthesis of a series of biobased ionic liquids with natural carboxylates (proline-based ionic liquids) [29]. These compounds were employed as a solvent and basic catalyst in Michael reactions.

The second group of papers concerns the synthesis and activity of heterogeneous catalytic systems. In the review article [30], the most important issues of heterogeneous catalysis with the participation of ionic liquids were presented. The methods of synthesis and exemplary trends in the application of SILP (SILCA), SCILL, and porous ionic liquids were described in the review. Subsequent papers concern the application of heterogeneous systems in particular types of reactions. Vucetic et al. described the use of PdCl2-containing SILCA as a catalyst for a Heck reaction [31]. The catalyst showed high stability and activity in the reaction between butylacrylate and iodobenzene and enabled its multiple use. A highly active SILP material that contained a rhodium complex was applied in the synthesis of monofunctional derivatives of disiloxanes. The above catalyst, described in the paper by Kukawka et al. [32], was characterized by high stability that enabled at least 50 catalytic runs without loss of activity. Orli ´nska et al. presented a new SCILL catalytic system with N-hydroxyphthalimide as an active component and applied it in the oxidation of ethylbenzene [33]. In the next two papers, heterogeneous catalysts containing ionic liquids as structural elements were described. Bartlewicz et al. [34] presented a rhodium complex ligated by imidazolium-substituted phosphine that showed high activity for hydrosilylation of alkynes. A series of anionic platinum complexes obtained by a simple reaction of piperidinium or pyrrolidinium ionic liquids with platinum compounds were presented by Jankowska-Wajda et al. [35]. The complexes prepared in such a way appeared to be very active catalysts for hydrosilylation of different olefins. It is worth mentioning that their high activity was maintained even after multiple use. Szyling et al. [36] described an effective immobilization of a ruthenium complex in ionic liquids and applied the obtained complex as a catalyst for the selective synthesis of (E)-alkenyl boronates via borylative coupling of olefins with vinyl boronic acid pinacol ester. Additional application of scCO2 to the extraction of the product significantly reduced catalyst leaching, due to which it could be used many times.

All the above papers showed the great importance of ionic liquids in practice and that their catalytic application is growing wider and wider; particularly, they can play various functions in catalysis. In addition to the known advantages of the application of ionic liquids, the development of advanced methods of their synthesis from biorenewable natural resources results in their biodegradable and environmentally friendly character. For this reason, one can expect further development of their application, including in different catalytic processes.

**Funding:** This research received no external funding.

**Data Availability Statement:** Data available in a publicly accessible repository.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


## *Article* **Performance of 1-(3-Sulfopropyl)-3- Methylimidazolium Hydrogen Sulfate as a Catalyst for Hardwood Upgrading into Bio-Based Platform Chemicals**

#### **Mar López, Sandra Rivas, Carlos Vila, Valentín Santos and Juan Carlos Parajó \***

Chemical Engineering Department, University of Vigo (Campus Ourense), Polytechnical Building, As Lagoas, 32004 Ourense, Spain; marlopezr@uvigo.es (M.L.); sandrarivas@uvigo.es (S.R.); cvila@uvigo.es (C.V.); vsantos@uvigo.es (V.S.)

**\*** Correspondence: jcparajo@uvigo.es; Tel.: +34-988-387-033

Received: 25 July 2020; Accepted: 12 August 2020; Published: 15 August 2020

**Abstract:** The acidic ionic liquid 1-(3-sulfopropyl)-3-methylimidazolium hydrogen sulfate ([C3SO3Hmim]HSO4) was employed as a catalyst for manufacturing polysaccharide-derived products (soluble hemicellulose-derived saccharides, furans, and/or organic acids) from *Eucalyptus globulus* wood. Operation was performed in aqueous media supplemented with [C3SO3Hmim]HSO4 and methyl isobutyl ketone, following two different processing schemes: one-pot reaction or the solubilization of hemicelluloses by hydrothermal processing followed by the separate manufacture of the target compounds from both hemicellulose-derived saccharides and cellulose. Depending on the operational conditions, the one-pot reaction could be directed to the formation of furfural (at molar conversions up to 92.6%), levulinic acid (at molar conversions up to 45.8%), or mixtures of furfural and levulinic acid (at molar conversions up to 81.3% and 44.8%, respectively). In comparison, after hydrothermal processing, the liquid phase (containing hemicellulose-derived saccharides) yielded furfural at molar conversions near 78%, whereas levulinic acid was produced from the cellulose-enriched, solid phase at molar conversions up to 49.5%.

**Keywords:** acidic ionic liquid; *Eucalyptus* wood; furfural; levulinic acid

#### **1. Introduction**

Mankind is facing key challenges related to environmental issues and sustainability, caused by the massive utilization of fossil resources, which currently provide more than 90% of our energy needs and feedstocks of the chemical industry [1]. The development of new processes based on renewable vegetal biomass (the most important source of organic carbon on earth) as a feedstock is one of the strategic ways pointed out to deal with these problems. In this field, the "lignocellulose biorefinery" concept, based on the selective separation of the major components making part of lignocellulosic biomass (LB), allows the manufacture of diverse bio-based platform chemicals, including furans and levulinic acid.

The term LB is used to name the numerous types of vegetal biomass mainly made up of polysaccharides (hemicelluloses and cellulose) and an aromatic fraction (lignin). These polymeric components are known as "structural components" of LB, and they appear together with non-structural components (extractives, ash, etc.), which are not important for the objectives of this study. Hardwoods possess a high content of structural components (typically, about 90 wt %), with a hemicellulose fraction mainly made up of heteroxylan.

*Eucalyptus* spp. is the world's most widely planted hardwood species. Its fast, uniform growth, self-pruning, and ability to coppice make it favorable for applications as timber, pulpwood, or bioenergy feedstocks; as a result, it is one of the most productive and economically viable biomass crops in the world, with expansive commercialization on all populated continents [2].

Additionally, *Eucalyptus* woods show a number of favorable features to be used as industrial feedstocks, including compositional factors (comparative high cellulose content, hemicelluloses mainly made up by acetylated glucuronoxylan), as well as high density and growth rate [3,4]. *Eucalyptus globulus* wood is one of the most important commercial *Eucalyptus* species, and it represents a basic forest resource in the Atlantic regions of the Iberian Peninsula [5,6].

The industrial utilization of LB can be achieved on the basis of the biorefinery concept, which proposes the sustainable processing of biomass into a spectrum of marketable products and energy, including biodegradable plastics, platform chemicals and other bio-based compounds, and advanced biofuels [4,7]. In this field, furfural and levulinic acid have been included among the top platform chemicals that can be manufactured based on green chemistry [7].

Most studies dealing with lignocellulose biorefineries employ the biochemical route, based on pretreatment and further hydrolysis fermentation of cellulose and/or hemicelluloses. Alternatively, processing schemes based on the acidic processing of wood to cause the hydrolysis–dehydration or hydrolysis–dehydration–rehydration of polysaccharides, provide a suitable framework for a fully chemical lignocellulose biorefinery having furfural (FF), hydroxymethylfurfural (HMF), and/or organic acids (levulinic acid, denoted LevA; and formic acid, denoted FA) as target products. For the purposes of this study, FF, LevA, and FA are of special importance, since:


*Eucalyptus globulus* can be employed as a raw material for biorefineries following diverse processing schemes, including the ones shown in Figure 1a,b.


**Figure 1.** Processing schemes leading to the acid-catalyzed manufacture of furans and organic acids from wood. (**a**) One-pot approach. (**b**) Approach based on hemicellulose separation before furfural manufacture and cellulose conversion.

Each of these processing alternatives shows advantages and disadvantages with respect to each other: one-pot methods are easy to carry out and present a simple structure, requiring little equipment for reaction, and limited manpower [16]. However, the faster kinetics of FF generation/consumption with respect to the reactions leading to the formation/consumption of HMF and/or organic acids [17] means that some FF loss (or an incomplete formation of HMF/organic acids) has to be assumed. Alternatively, processes based on hemicellulose removal and the separate conversion of hemicellulose-derived saccharides and cellulose may increase selectivity and yields, but entail more reaction and downstream stages.

The kinetics of FF and HMF/LevA/FA formation from polysaccharides in aqueous, acidic media show complex mechanisms, which include the generation of reactive intermediates and the participation of these latter (together with the target products) into a number of series and parallel reactions, leading to the formation of humin-type products, which limit the yields of the target products [18–24].

A number of approaches have been proposed to improve the production of target products, including:

• Utilization of reaction media containing an immiscible solvent able to extract the target products, avoiding the unwanted reactions taking place in aqueous media [25–27]. This approach has been followed in this study, in which methyl isobutyl ketone (MIBK), a green solvent recommended by the CHEM21 guide [28], has been employed.


Owing to their "green" character and special physicochemical properties (particularly, low volatility and stability), ILs can play a number of roles in biorefineries, including physical separation, the formulation of reaction media, and utilization as catalysts [33]. In this context, the conversion of wood polysaccharides (or low molecular weight saccharides derived from hemicelluloses) in aqueous media catalyzed by Brønsted acidic ILs, eventually in the presence of an organic solvent, has been considered in the literature. SO3H-functionalized ILs have been employed to achieve the dehydration of xylose into FF [34], whereas 1-butyl 3-methylimidazolium hydrogen sulfate ([bmim]HSO4) has been employed to convert the hemicellulose fraction of lignocellulosic biomass into hemicellulosic sugars and FF [35], and to produce FF from xylose or soluble hemicellulose-derived saccharides obtained by the hydrothermal processing of *Eucalyptus* wood [32], eventually in the presence of a cosolvent [36]. Dealing with the conversion of cellulose, 1-(3-sulfopropyl)-3-methylimidazolium hydrogen sulfate ([C3SO3Hmim]HSO4), alone or in combination with metal chlorides, has been employed for LevA production from pure cellulose [37]; whereas [C3SO3Hmim]HSO4 and 1-(3-sulfobutyl)-3-methylimidazolium hydrogen sulfate ([C4SO3Hmim]HSO4) provided good results as catalysts for the conversion of *Pinus pinaster* wood polysaccharides [38]. Sulfonated ILs were successfully employed as catalysts to achieve the conversion of pure cellulose into LevA [26].

This work provides a quantitative assessment of the performance of [C3SO3Hmim]HSO4 as a catalyst for *Eucalyptus globulus* wood processing in biphasic media, following both the one-pot approach (Figure 1a) and a method based on hemicellulose solubilization and the further conversion of hemicellulose-derived products and cellulose into the target products (Figure 1b). Hydrolysis–dehydration or hydrolysis–dehydration–rehydration reactions were performed in a stirred, MW-heated reactor. The experimental results obtained under diverse operational conditions are discussed in terms of the volumetric concentrations and molar conversions of the target products (FF, LevA, FA, and acetic acid) and intermediates (HMF) derived from polysaccharides.

#### **2. Results**

#### *2.1. Composition of Eucalyptus Globulus Wood*

For the purposes of this work, the relevant data concerning *E. globulus* wood are the contents of the structural components (cellulose, hemicelluloses, and lignin), and the relative amounts of the diverse hemicellulose components. The wood lot employed in this study contained, in oven-dry mass basis, 22.1% ± 0.2% Klason lignin, 1.38 ± 0.07 acid soluble lignin, 44.4% ± 0.2% cellulose, 15.6% ± 0.2% xylan, 0.43% ± 0.06% arabinosyl units, 0.90% ± 0.06% mannosyl units, 1.02% ± 0.03% galactosyl units, and 2.95% ± 0.09% acetyl groups. This composition is typical for *Eucalyptus globulus*, as it can be seen in a recent review [4]. Among the above fractions, cellulose, galactosyl units, and mannosyl units are potential substrates for manufacturing HMF and/or LevA/FA; xylan and arabinosyl groups are potential substrates for FF manufacture; and acetyl groups yield acetic acid (AcH) upon hydrolysis.

#### *2.2. One-Pot Conversion of Eucalyptus Globulus Wood*

In order to assess the ability of [C3SO3Hmim]HSO4 as a catalyst for obtaining bio-based chemicals from *Eucalyptus globulus* wood polysaccharides, the set of experiments listed in Table 1 was performed using the MW-heated reactor. Operating at moderate catalyst loadings (0.10–0.15 g/g oven-dry wood), operation was carried out at high water-to-solid ratios (6 or 10 g/g oven-dry wood, which are markedly higher than the ones typically employed in MW reactors), keeping the relative amount of MIBK at 2 g/g aqueous phase.

The first assay (experiment 1 in Table 1) was performed under mild conditions: the reaction media, made with the highest charges of solid and catalyst considered, was heated up to 170 ◦C and then cooled immediately (isothermal reaction time, 0 min). The high solid yield (53.7 g/100 g wood) indicated that the conversion of polysaccharides into the target products was incomplete, which was a fact confirmed by the concentrations of reaction intermediates in aqueous phase (12.4 g glucose/L and 18.6 and g xylose/L). Interestingly, the conversion of xylan into xylose proceeded at 62.8% molar conversion, and the acetyl groups were almost quantitatively hydrolyzed into AcH, which appears distributed in both aqueous and organic phases), confirming that [C3SO3Hmim]HSO4 could be used for manufacturing hemicellulosic sugars. However, the concentrations of FF, HMF, FA, and LevA were low.

Based on these findings, the severity of the operational conditions was increased by raising the temperature to 180 ◦C and keeping the rest of experimental variables unchanged. As expected, the solid yield decreased (to 38.5%), the glucose concentration increased (28.0 g/L) owing to the higher cellulose conversion, and the xylose concentration dropped to 8.18 g/L owing to the increased generation of FF (which appeared concentrated in the organic phase, accounting for an overall molar conversion of 49%, including the contributions of the aqueous and organic phases). Although the LevA concentrations increased up to 0.98 and 0.34 g/L in the aqueous and organic phases, respectively, its overall molar conversion was still very poor (3.40%). When the reaction conditions were modified by performing an isothermal reaction stage of 16 min (experiment 3), the solid yield dropped almost by half, even when the catalyst concentration was reduced, and the xylose concentration decreased to 1 g/L, with enhanced FF concentrations in both phases (equivalent to an overall molar conversion of 78.3%). This finding confirmed that [C3SO3Hmim]HSO4 presents a remarkable potential for obtaining FF from xylan-containing materials. However, the concentrations of LevA were still low (equivalent to 27.0% overall molar conversion), owing to the presence in the medium of other cellulose-derived intermediates (glucose and HMF, which were obtained at 22.2% and 10.7% overall molar yields, respectively).

In order to improve the production of LevA, additional assays were carried out at a higher temperature (190 ◦C) for 0, 30, and 50 min of isothermal operation (experiments 4, 5, and 6 in Table 1). The results obtained in experiment 4 (26.1% solid yield, glucose concentration corresponding to 39.7% molar yield; xylose accounting for 14.9% of the stoichiometric amount) indicated that the experiment was performed under conditions too mild for practical purposes. Better results were achieved in experiment 5, with low concentrations of glucose and xylose in the aqueous phase (indicative of increased conversions into furans and/or furan-derived products), and improved concentrations of LevA and FF (which were obtained at 43.2% and 58.3% molar yields, respectively). Additionally, under these conditions, FA was obtained at 54.9% molar conversion, and acetyl groups were totally converted into AcH. Prolonging the reaction time up to 50 min (experiment 6) led to worse results (LevA and FF were obtained at 42.4% and 49.0% molar yields). In summary, the conditions of experiment 5 were the best ones for achieving the simultaneous conversion of hemicellulose and cellulose into FF and LevA/FA.

In the acidic processing of LB, increasing the relative amount of liquid with respect to the solid substrate is known to improve the experimental results, because dilution limits the participation of reactive intermediates and products in parasitic reactions. Additionally, lower catalyst charges may result in improved selectivity. In order to assess these points, experiments 7, 8, and 9 were carried

out at 180 ◦C for 15, 30, or 45 min, respectively, in media containing 0.1 g catalyst/g oven-dry wood. The solid yields obtained in these experiments (in the range 8.1–6.9%) indicated that solubilization reactions affected both polysaccharides and lignin. The aqueous phase obtained in experiment 7 contained 16.9 g glucose/L, indicating that the severity was still not enough to achieve optimal amounts of LevA and FA (which were obtained at 13.4% and 21.2% molar conversions, respectively). In turn, these conditions were excellent for FF production, which was obtained at 89.4% molar conversion. When the reaction time was extended to 30 min (experiment 8), the FF generation decreased slightly (molar conversion, 87.4%), but the LevA production increased (molar conversion, 25.3%). Looking for an improved production of LevA/FA, the reaction was performed for 45 min (experiment 9). This modification resulted in increased FF consumption (molar yield, 80.4%), but the production of LevA and FA improved up to 43.0% and 56.3% molar conversions, respectively.

Based on the above results, experiments 10, 11, and 12 were performed to assess the possible benefits derived from operating at a higher temperature (190 ◦C) for 15, 30, or 45 min, keeping the rest of the operational variables unchanged. The limited solid yields (in the range 5.6–4.7%) confirmed that almost total wood liquefaction took place under the considered conditions. In experiment 10 (reaction time, 15 min), the most remarkable finding was the excellent conversion of precursors into FF, which was obtained at 92.6% molar yield. However, LevA and FA were obtained at 35.5% and 48.1% molar yields, respectively. The data suggested that the results could be improved by enhancing the conversion of intermediates (glucose and HMF, which were obtained at 2.64% and 13.5% molar yields, respectively) into the target products. This idea was assessed by increasing the reaction time to 30 min (conditions of experiment 11). As expected, the generation of LevA and FA increased (molar yields, 44.8% and 56.8%, respectively), at the cost of worsening the FF production (molar yield, 81.3%). However, the conditions of experiment 11 can be considered as a satisfactory compromise for the one-pot production of bio-based chemicals, because harsher conditions (as the ones employed in experiment 12, in which the reaction time was fixed in 45 min) resulted in a slightly improved LevA production (molar yield, 45.8%) but in an important FF loss (molar yield, 75.8%).


**Table 1.**Operational conditions assayed and experimental results achieved in one-pot experiments using native*E. globulus*wood as a feedstock.

ketone (MIBK)/g feedstock; T, temperature in ◦C; t, isothermal reaction time in min; SY, solid yield in g solid after treatment/g feedstock; Glc, glucose concentration in g/L, Xyc, xylose concentration in g/L; FAc formic acid concentration in g/L; AcHc, acetic acid concentration in g/L; LevAc, levulinic acid concentration in g/L; HMFc, HMF concentration in g/L; FFc, furfural concentration in g/L.

#### *Catalysts* **2020** , *10*, 937

#### *2.3. Eucalyptus Globulus Wood Conversion Based on Hemicellulose Removal*

#### 2.3.1. Hydrothermal Processing

Figure 1b shows the general idea of a biorefinery process based on hydrothermal processing (also named autohydrolysis) and further individual utilization of the liquid phase (containing soluble hemicellulose-derived saccharides) and the solid phase (containing cellulose and lignin). The autohydrolysis of *Eucalyptus globulus* wood has been considered in the literature [39], and it is considered as a green, efficient, and selective method for hemicellulose separation from cellulose and lignin. Operating under suitable conditions, the major soluble reaction products are hemicellulose-derived saccharides (mainly of oligomeric nature), which are accompanied by minor amounts of sugars and sugar-dehydration products. The saccharides present in the aqueous phase are substrates suitable for FF manufacture [32], whereas lignin and cellulose undergo little modification and remain in solid phase, enabling their separate utilization. In this work, the treatments were performed using a water-to-solid ratio of 8 kg/kg. The media were heated up to reach 200 ◦C and cooled immediately, following the temperature profiles reported in Garrote and Parajó [39]. Upon hydrothermal processing, 28.9% of the wood mass was solubilized, yielding a treated solid enriched in cellulose, and a liquid phase containing xylooligomers as major products. Table 2 lists the composition of both phases.


**Table 2.** Composition of solid and liquid phases from hydrothermal processing.

#### 2.3.2. Conversion of the Solids from Hydrothermal Processing

Based on the above information, the cellulose-enriched solids coming from hydrothermal fractionation (see Figure 1b), mainly made up of cellulose (58.8%) and lignin (30.6%), were treated with water, [C3SO3Hmim]HSO4, and MIBK to yield HMF, LevA, and FA. Owing to the limited xylan and arabinan contents of the treated solids (3.20 and 0.07%, respectively), the generation of FF is not discussed in this section.

The experimental plan was established keeping in mind the results obtained in assays 7 to 12 in Table 1. Those assays were performed at 180 or 190 ◦C with a catalyst concentration (denoted CC) of 0.1 g catalyst/g oven-dry wood, for reaction times (t) up to 45 min. Table 3 lists the operational conditions assayed in this part of our study (similar to the above ones, except the longest reaction time, which was fixed in 75 or 60 min for operation at 180 or 190 ◦C, respectively). The same table includes the volumetric concentrations of the target products in the aqueous and organic phases. For clarity, Figure 2; Figure 3 show the experimental results in terms of overall molar conversions (including the amounts of the target products contained in both the aqueous and organic phases).

**Figure 2.** Overall molar conversions (including the contributions of aqueous and organic phases) achieved in experiments performed at 180 ◦C using the solids from hydrothermal processing as a substrate and [C3SO3Hmim]HSO4 as a catalyst.

**Figure 3.** Overall molar conversions (including the contributions of aqueous and organic phases) achieved in experiments performed at 190 ◦C using the solids from hydrothermal processing as a substrate and [C3SO3Hmim]HSO4 as a catalyst.

The high solid yield and glucose concentration in aqueous phase obtained in experiment 1, which was performed at 180 ◦C with no isothermal stage (see Table 3), resulted in incomplete cellulose conversion, indicating that the conditions were too mild for the purposes of this study. Implementing an isothermal stage of 15–75 min (experiments 2–6 in Table 3) substantially decreased the solid yield to 15.3–13.4 g/100 g solid and caused a steady increase in the concentrations of LevA and FA. The effects of the reaction time on the cellulose conversion into glucose, HMF, and LevA are displayed in Figure 2, where it can be seen that glucose and HMF behaved as reaction intermediates, whereas the maximum LevA molar conversion (49.3%) was reached after 75 min.



T, temperature in ◦C; t, isothermalformic acid concentrationing/L;

14

 reaction time in min; SY, solid yield in g solid after treatment/g feedstock; Glc, glucose concentration

 AcHc, acetic acid concentration

 in g/L; LevAc, levulinic acid concentration

 in g/L; HMFc, HMF concentration

 in g/L, Xyc, xylose concentration

 in g/L; FFc, furfural concentration

 in g/L; FAc

 in g/L.

The effects caused by increasing the temperature up to 190 ◦C (keeping constant the rest of operational variables) were considered in experiments 7 to 11 in Table 3. As before, high solid yields and low concentrations of intermediates and target products were obtained when no isothermal reaction stage was considered (experiment 7). Increasing the reaction time up to 60 min (experiments 8 to 11) resulted in a fast increase of the concentrations of the target products. Figure 3, which presents the results in terms of molar conversions, shows a fast consumption of intermediates and a maximum molar conversion into LevA of 49.5% (the highest value reached in this study).

#### 2.3.3. Conversion of the Liquid Phase from Hydrothermal Processing

The liquid phase from hydrothermal processing (see Figure 1b), with the composition indicated in Table 2, was supplemented with [C3SO3Hmim]HSO4 and heated in the presence of MIBK to produce FF from the suitable precursors. The optimization of FF manufacture was studied using the Response Surface Methodology, based on a factorial, incomplete, and centered Box–Behnken experimental design, in which three selected independent variables (T, t, and the volumetric catalyst concentration, denoted VCC) were assayed in duplicate at three levels. The ranges considered for the operational variables were 160–180 ◦C for temperature and 2–30 min for t; whereas VCC varied from 1.42 <sup>×</sup> 10−<sup>2</sup> up to 4.26 <sup>×</sup> 10−<sup>2</sup> g/g aqueous phase. As before, the MIBK/aqueous phase mass ratio was fixed in 2 g/g. The only dependent variable considered in this part of the study was the FF molar yield (denoted FFMY), which was measured in respect to the potential substrates in the liquid phase.

The operational conditions employed in diverse experiments and the corresponding FFMY values are listed in Table 4.


**Table 4.** Operational conditions employed in experiments of furfural (FF) production from hemicellulose-derived saccharides, and experimental results obtained for the furfural molar yield. FFMY: FF molar yield, VCC: volumetric catalyst concentration.

For calculation purposes, the following normalized, dimensionless variables with variation ranges (−1, 1) were defined:

x1 (dimensionless temperature for experiment j) = 2·(Tj − 170)/(180 − 160) (1)

x2 (dimensionless time for experiment j) = 2·(tj − 16)/(30 − 2) (2)

$$\begin{aligned} \text{x}\_3 \text{ (dimensionless volumetric catalyst concentration for experiment j)} &= \text{2-(VCC}\_{\text{j}} - 2.84 \times 10^{-2} \text{)} (4.26 \times 10^{-2} - 1.42 \times 10^{-2})} \end{aligned} \tag{3}$$

The values of the FFMY data in Table 4 were correlated with the values of the correspondent normalized variables using the following second-order equation involving linear, interaction, and quadratic terms.

$$\text{FFMY} = \mathbf{a}\_0 + \mathbf{a}\_1 \cdot \mathbf{x}\_1 + \mathbf{a}\_2 \cdot \mathbf{x}\_2 + \mathbf{a}\_3 \cdot \mathbf{x}\_3 + \mathbf{a}\_{12} \times \mathbf{x}\_1 \cdot \mathbf{x}\_2 + \mathbf{a}\_{13} \cdot \mathbf{x}\_1 \cdot \mathbf{x}\_3 + \mathbf{a}\_{23} \times \mathbf{x}\_2 \cdot \mathbf{x}\_3 + \mathbf{a}\_{11} \cdot \mathbf{x}\_1^2 + \mathbf{a}\_{22} \times \mathbf{x}\_2^2 + \mathbf{a}\_{33} \times \mathbf{x}\_3^2} \tag{4}$$

Table 5 lists the values of the regression coefficients (calculated by the least squares method), their statistical significance (based on a *t*-test), the statistical significance of the model (measured using the Fst Fischer's test), and the correlation coefficient.

**Table 5.** Regression coefficients, statistical significance (based on a *t*-test), and parameters measuring the correlation and significance of the statistical model.


**Statistical parameters:** R2: 0.980; *Fst*: 87.3 (significance > 99%)

The model predictions were analyzed on the basis of the response surfaces shown in Figure 4; Figure 5. Figure 4 shows that at 170 ◦C, the FFMY was favored by operation at long reaction times with limited catalyst charges, with a clearly defined optimum zone. Figure 5 depicts a closely related situation: in assays lasting 16 min, the best results were predicted for low or intermediate catalyst loadings, with a defined optimum zone appearing near the highest temperature assayed.

**Figure 4.** Dependence of the furfural molar conversion on time and volumetric catalyst concentration (data calculated for T = 170 ◦C).

**Figure 5.** Dependence of the furfural molar conversion on temperature and volumetric catalyst concentration (data calculated for t = 16 min).

The mathematical optimization of the statistical model within the experimental domain (∂FFMY/∂xi =0; −1 ≤ xi ≤ 1; i: 1 to 3) was predicted a maximum FFMY (80.4%) for T = 171 ◦C, t = 30 min, and VCC = 1.97 <sup>×</sup> 10−<sup>2</sup> g catalyst/g aqueous phase. Additional experiments performed for model validation led to an FFMY near 78%, which was close to the experimental results obtained in experiments 3, 6, and 8, and it defined a region of the experimental domain where the experimental and fitting errors overlapped the possible differences in FFMY values.

#### 2.3.4. Comparative Analysis of Results

The literature has considered the utilization of AIL as reaction media and/or as catalysts for obtaining platform chemicals from a number of substrates, including sugars, hemicellulose-derived saccharides, purified polysacharides, LB-derived fractions, and native LB. However, comparing the results reported with the ones achieved in this study is not straightforward, owing to important differences in the types of substrates and catalysts, the diversity of operational conditions (particularly, regarding catalyst and substrate charges), and the type of target products (FF and/or HMF and/or LevA+FA).

Concerning the FF production from wood hemicellulose (with or without the simultaneous production of LevA from cellulose), Zhang and Zhao [40] treated wood in 1-butyl-3-methylimidazolium chloride catalyzed with CrCl3 to obtain FF at yields up to 31%, together with HMF. The same IL was employed by Sievers et al. [41] to obtain HMF and FF (at low mass yield) from loblolly pine using trifluoroacetic acid as a catalyst. López et al. [38] employed 1-butyl-3-methylimidazolium hydrogen sulfate as a catalyst to achieve 60.5% conversion of pine wood pentosans into FF. Under the considered operational conditions, little conversion of cellulose into LevA was achieved, which is ascribed to the limited acidity of the catalyst. [C3SO3Hmim]HSO4 and [C4SO3Hmim]HSO4, two AIL of enhanced acidity (caused by the sulfonic group attached to the cation) have been successfully employed for processing purified cellulose and native LB. A near-quantitative conversion of xylosyl and arabinosyl groups in native pine wood into FF was achieved using these catalysts [38]. Concerning the production of LevA, 39.4–55% molar conversion was reached using microcrystalline cellulose as a substrate [26], [37], or 32.8% molar conversion of precursos (hexosans) in pine wood [38]. Obtaining higher LevA yields entailed the utilization of high catalyst charges (up to 4 g/g cellulose) [29], whereas the manufacture of soluble products from LB was favored when susceptible substrates of low lignin content (such as bagasse) were employed [42].

Regarding the FF production from LB-derived fractions, Peleteiro et al. [32] employed the liquid phase from *Eucalyptus globulus* wood autohydrolysis (containing xylooligosaccharides as major reaction products) as a substrate for reaction in biphasic media catalyzed with 1-butyl-3-methylimidazolium hydrogen sulfate. MIBK and toluene were employed as organic solvents, at a 4.4/1 mass ratio with respect to the aqueous phase, which was concentrated using membranes before reaction. Under the best operational conditions, the molar conversion of the suitable precursors into FF reached 68.4%. In a related study, a biorefinery solution enriched in C5 oligomers coming from the aqueous processing of agricultural LB was treated in a biphasic media (water/tetrahydrofuran) in the presence of the AIL 1-(4-sulfonylbutyl)pyridinium methanesulfonate, yielding FF at molar conversions up to 45% [34].

#### **3. Materials and Methods**

#### *3.1. Raw Material*

*Eucalyptus globulus* wood chips were collected in a local pulp mill, milled to pass an 8 mm screen, air-dried to a final moisture of 6.6%, and employed for reaction. The wood samples employed for analysis were further milled and screened following the NREL/TP-510-42620 standard method.

#### *3.2. Reaction*

In hydrothermal reactions, wood samples were mixed with water at the desired solid charge (8 kg liquid phase/kg oven-dry wood) and heated up to 200 ◦C in a 3.75 L stainless steel, stirred Parr reactor (Parr Instruments, Moline, Illinois, USA). In one-pot reactions, wood was mixed with water, [C3SO3Hmim]HSO4 (purchased from Ionic Liquids Technologies GmbH, Heilbronn, Germany) and MIBK, and reacted for the desired time at the target temperature and solid charge (see text and table headings for specific information) using a stirred MARS 6 MW reactor. Reactions involving the liquid phase from hydrothermal processing were carried out in the presence of [C3SO3Hmim]HSO4 and

MIBK using the same MW reactor, under the set of experimental conditions indicated in the Results section. All reaction conditions were assayed in duplicate experiments.

#### *3.3. Analysis*

Wood samples were assayed for polysaccharides (cellulose and hemicelluloses) and lignin by quantitative acid hydrolysis (NREL standard method/TP-510-42618). The target products in liquid phase (oligosaccharides, sugars, furans, and organic acids) coming from quantitative acid hydrolysis or reaction experiments were determined by HPLC-IR and HPLC-Diode Array Detector as per Rivas et al. [43].

#### **4. Conclusions**

The acidic ionic liquid [C3SO3Hmim]HSO4 was employed as a catalyst for *Eucalyptus globulus* wood processing in reaction media also containing water and wood (or wood-derived fractions). Operating in the presence of MIBK, two different operational modes were considered (see Figure 1): (a) one-pot reaction and (b) the separate conversion of hemicellulose-derived products and cellulose-enriched solids resulting from hydrothermal fractionation.

Following the one-pot approach, mild reaction conditions (170 ◦C, 0 min of isothermal reaction time, 0.15 g catalyst/g oven-dry wood) resulted in the hydrolysis of xylan into xylose at 62.8% molar conversion. Harsher conditions increased the formation of FF (up to 92.6% molar conversion) and levulinic acid (up to 45.8% molar conversion). Conditions of increased severity (190 ◦C, 30 min, 0.10 g catalyst/g oven-dry wood) resulted in slightly decreased FF production (81.3% molar conversion) but a better conversion of cellulose into LevA (44.8%). Concerning the separate conversion of the phases resulting from hydrothermal processing, the conversion of the cellulose-enriched solid into LevA under optimal conditions (190 ◦C, 45 min, 0.10 g catalyst/g oven-dry wood) proceeded at about 49.5% molar conversion, whereas the hemicellulose-derived saccharides present in the liquid phase yielded FF at near 78% molar conversion.

The experimental results confirm the suitability of [C3SO3Hmim]HSO4 as a versatile catalyst for the environmentally friendly production of bio-based chemicals from the polysaccharide fraction of *Eucalyptus globulus* wood.

**Author Contributions:** Conceptualization, M.L., S.R., C.V., V.S. and J.C.P.; methodology, M.L., S.R., C.V., V.S. and J.C.P.; validation, M.L.; investigation, M.L., S.R., C.V., V.S. and J.C.P.; resources, V.S., J.C.P.; data curation, M.L.; writing—original draft preparation, M.L., S.R., C.V., V.S. and J.C.P.; writing—review and editing, M.L., V.S. and J.C.P.; supervision, V.S. and J.C.P.; project administration, V.S. and J.C.P.; funding acquisition, V.S. and J.C.P. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the "Ministry of Economy and Competitiveness" of Spain (research project "Modified aqueous media for wood biorefineries", reference CTQ2017-82962-R).

**Acknowledgments:** M.L. thanks "Xunta de Galicia" and the European Union (European Social Fund—ESF) for her predoctoral grant (reference ED481A-2017/316). S.R. thanks "Ministerio de Ciencia, Innovación y Universidades" for her "Juan de la Cierva" contract IJC2018-037665-I.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Article*
