Production of Aromatic Compounds by Catalytic Depolymerization of Technical and Downstream Biorefinery Lignins

biomolecules Article Production of Aromatic Compounds by Catalytic Depolymerization of Technical and Downstream Biorefinery Lignins Alfonso Cornejo 1,* , Fernando Bimbela 1,* , Rui Moreira 2, Karina Hablich 1, Íñigo García-Yoldi 1, Maitane Maisterra 1, António Portugal 2, Luis M. Gandía1and Víctor Martínez-Merino 1 1Institute for Advanced Materials and Mathematics (InaMat2) and Department of Sciences, Universidad Pública de Navarra, E31006 Pamplona, Spain; [email protected] (K.H.); [email protected] (Í.G.-Y.); [email protected] (M.M.); [email protected] (L.M.G.); [email protected] (V.M.-M.) 2CIEPQPF, FCTUC, Department of Chemical Engineering, University of Coimbra, Rua Sílvio Lima, Pólo II—Pinhal de Marrocos, 3030-790 Coimbra, Portugal; r[email protected] (R.M.); [email protected] (A.P.) *Correspondence: [email protected] (A.C.); [email protected] (F.B.); Tel.: +34-948-166304 (A.C.); +34-948-166259 (F.B.) Received: 31 July 2020; Accepted: 11 September 2020; Published: 18 September 2020   Abstract: Lignocellulosic materials are promising alternatives to non-renewable fossil sources when producing aromatic compounds. Lignins from Populus salicaceae.Pinus radiata and Pinus pinaster from industrial wastes and biorefinery effluents were isolated and characterized. Lignin was depolymerized using homogenous (NaOH) and heterogeneous (Ni-, Cuor Ni-Cu-hydrotalcites) base catalysis and catalytic hydrogenolysis using Ru/C. When homogeneous base catalyzed depolymerization (BCD) and Ru/C hydrogenolysis were combined on poplar lignin, the aromatics amount was ca. 11 wt.%. Monomer distributions changed depending on the feedstock and the reaction conditions. Aqueous NaOH produced cleavage of the alkyl side chain that was preserved when using modified hydrotalcite catalysts or Ru/C-catalyzed hydrogenolysis in ethanol. Depolymerization using hydrotalcite catalysts in ethanol produced monomers bearing carbonyl groups on the alkyl side chain. The analysis of the reaction mixtures was done by size exclusion chromatography (SEC) and diffusion ordered nuclear magnetic resonance spectroscopy (DOSY NMR). 31 P NMR and heteronuclear single quantum coherence spectroscopy (HSQC) were also used in this study. The content in poly-(hydroxy)-aromatic ethers in the reaction mixtures decreased upon thermal treatments in ethanol. It was concluded that thermo-solvolysis is key in lignin depolymerization, and that the synergistic effect of Ni and Cu provided monomers with oxidized alkyl side chains. Keywords: lignin; depolymerization; homogeneous and heterogeneous catalysis; bio-based aromatic compounds 1. Introduction Lignin is a very complex natural amorphous polymer, mainly composed of three different phenylpropanoid units: p-coumaryl alcohol, sinapyl alcohol and coniferyl alcohol [ 1 , 2 ]. In spite of the relatively simple structure of these three monomeric units, their layout is highly complex, conforming an intricate tridimensional structure of distinct and chemically different motifs and natural alternative monomers as a consequence of their random radical polymerization [ 3 ]. These moieties and distinctive bonds are most responsible for the unique structures and properties of lignins rather than the three main Biomolecules 2020,10, 1338; doi:10.3390/biom10091338 www.mdpi.com/journal/biomolecules Biomolecules 2020,10, 1338 2 of 30 phenolic building blocks, and they are often the key to elucidate lignin morphology, physicochemical properties and reactivity. The structure and properties of native lignin are altered upon lignin isolation, as can be seen in Table 1, which presents different methodologies for lignin isolation. Great amounts of lignin-rich streams are produced, foremost in the pulp and paper industry by Kraft pulping [ 4 ]. More recently, large-scale processes for producing biofuels, e.g., second-generation cellulosic bioethanol, have generated large amounts of downstream biorefinery effluents [ 5 , 6 ]. Due to the inherent recalcitrance of lignins, their main industrial use is as low-grade fuel to produce energy [ 7 ], whilst allowing for the recovery of Kraft cooking reagents [ 8 ]. However, the large amounts of lignin produced in biorefineries generate surplus stocks that often exceed the energy demands of those plants [ 5 ]. The structure and properties of both Kraft and downstream biorefinery lignins are very different to those of the native lignins. Other alternatives for obtaining lignins include different organosolv (Org) methods which use organic solvents, either pure or in aqueous solutions [ 9 , 10 ], sometimes combined with other reactants, as in the soda-organosolv method (Seol) [ 11 ]. Due to the mild conditions used, the structure of organosolv lignins is very similar to their native structure. These isolation methods yield lignins often referred to as technical lignins. Biomolecules 2020,10, 1338 3 of 30 Table 1. Comparison of various lignin isolation methods. Adapted from reference [12]. Isolation Method Lignin Name Typical Process Characteristics Klason method [13] Klason 2% H2SO4 Extensive structure change, hardwood lignin is partly dissolved Kraft process [4] Kraft Na2S/NaOH Highly modified, partially fragmented. High S content Sulfite pulping process Lignosulfonate Extract lignin from waste liquor of the sulfate pulping process of soft wood. Highly modified, high average molecular weights, cleavage of ether linkages, loss of methoxyl groups and formation of new C–C bonds Björkman process [14,15] Milled Wood Lignin (MWL) Ball milling, then extracted by aqueous dioxane. Similar to the native structure, possible depolymerization due to extensive milling Organosolv process Organosolv [9,10] Using organic solvents to extract lignin. Mild conditions, results in more unaltered lignin, solvent could be recovered by distillation Soda ethanosolv [11] Basic reactant (NaOH, KOH . . . ) Reductive catalytic fractionation [16,17] Pd/C, Ru/C, Ni/Al2O3Phenolic monomers, low molecular weight poly(-hydroxy)-aromatic ethers Alkaline wet oxidation [18] Alkaline wet oxidation Oxidative pretreatment under alkaline conditions. Partial degradation via β-O-4 cleavage Steam explosion process [19] Steam explosion lignin High temperature steam explosion of the fibers. Require little or no chemical input, short treatment time, low energy requirement, changes of certain functional groups Mechano-catalytic process [20] Mechano-catalytic lignin Mechanocatalytic depolymerization. Sulfur-free lignin Biomolecules 2020,10, 1338 4 of 30 Given the chemical richness derived from its highly complex structure, lignin depolymerization is seen as an alternative to petrochemical industry for the production of aromatic compounds. Aromatics including monomeric phenols and other related aromatic compounds have been extensively sought after by different studies on catalytic depolymerization of lignin [ 15 , 16 ], some of which date back to 1963 [ 21 ]. Much effort has been made to develop routes for valorizing lignin into higher value-added chemicals and products, as can be deduced from the comprehensive reviews on the topic [ 2 – 4 , 7 , 12 , 22 – 35 ], many of which have been successful and the number of proven commercial applications and products from technical lignin is vast [36]. Most of the studies on catalytic technical lignin depolymerization rely on the heterogeneous catalysis for obtaining bioaromatics and other value-added compounds. In this context, Ru supported on carbon (Ru/C) has become the most widely used catalyst for the hydrogenolysis of technical lignins owing to its superior selectivity for cleaving C aryl -O bonds while maintaining a moderate activity for the hydrogenation of aromatic rings in comparison to other catalysts [ 37 ]. Ru/C is commonly used in combination with alcohols, mainly methanol and ethanol under supercritical conditions [ 28 ]. The main reasons for this is the rapid heat transfer, the high solubility of technical lignins and the reaction mechanism itself: alcohols not only can act as hydrogen donors but also as nucleophiles initiating the C-O-C bond cleavage [ 38 ]. Furthermore, ethanol also acts as a capping agent and formaldehyde scavenger through the formation of n-propanol thus preventing repolymerization reactions [ 39 , 40 ]. The combination of Ru/C catalysts and NaOH under methanol solvent medium has proven to be effective in the depolymerization of organosolv pine-derived lignin owing to the synergistic catalytic effect of NaOH and Ru/C [41]. Reductive catalytic fractionation (RCF) using Ru/C catalysts, including catalytic hydrogenolysis in the liquid phase [ 2 , 16 , 17 ], is an interesting approach to fractionate lignocellulosic residues. To this end, delignification is conducted via thermo-solvolysis readily followed by reductive stabilization of lignin. Using a combination of a Ru/C catalyst, methanol as solvent and pressurizing with hydrogen at 30 bar, delignification of hardwoods, such as birch sawdust, can be successfully attained. In this way, depolymerization of lignin fragments effectively leads to high yields of methoxyphenolic monomers, dimers and oligomers [ 16 ]. However, recent studies have revealed that, in the presence of bases such as KOH, RCF of birch wood suffers from partial repolymerization of lignin. This lowers the yield of phenolic monomers, but the selectivity to desirable C2-fragmented phenols can be favored over that to C3 phenols [ 17 ]. Therefore, adding bases to the reaction medium may hamper lignin depolymerization but enhances selectivity toward the desired products; hence, a trade-offhas to be achieved by an appropriate selection of reactions conditions and process design. Amidst the different proposed routes, biorefinery strategies that comprise catalytic depolymerization of lignin of downstream biorefinery lignins are of great interest, because they allow producing different organic compounds that can have potential applications and industrial use. However, the high recalcitrance of these lignins restricts most of the research to homogeneous BCD [ 42 ], whereas heterogeneous catalysis has been considerably much less investigated. Ru/C was selected, among other Ru-based catalysts, for conducting the reductive depolymerization of an industrial lignin-containing stillage derived from lignocellulosic bioethanol production [ 5 ]. Ru/C yielded the highest amount of monomeric phenols, but the yields were somewhat low, probably due to repolymerization reactions. In a previous work [ 6 ], samples from lignin-rich biorefinery streams from poplar and pine woodchips were successfully depolymerized under BCD conditions, obtaining remarkable monomer yields by doing a careful selection of operating conditions. Heterogeneous BCD using double-layered hydroxides known as hydrotalcites (HTC) has emerged as a promising alternative in recent years. Ni-functionalized HTCs have proven to be effective in cleaving C-O bonds of model compounds and in lignin depolymerization [ 43 ]. Up to 7–9 wt.% monomer yield was obtained due to the effect of nitrates intercalated as counter ions between the brucite-like layers that constitute the HTCs structure [ 44 ]. The efficiency of Ni-modified HTC catalysts is attributed both to their capability of specifically hydrolyzing the carbon-hydroxyl linkage at the side Biomolecules 2020,10, 1338 5 of 30 chain into alkanes and to the presence of strong binding sites for ether linkages, thus achieving their specific cleavage [ 45 ]. HTCs containing copper (Cu-HTCs) have also proven to be effective in lignin depolymerization under supercritical ethanol (scEtOH) at 573 K [ 46 ] and 613 K [ 47 ], giving monomer yields around 20 wt.% and 30 wt.%, respectively. However, these fragments are further alkylated or hydrodeoxygenated by the solvent [ 39 ], which effectively reduces the aromatic monomers yield. Noteworthy, HTC-Cu catalysts under scEtOH at 653 K produced up to 86 wt.% alkylated monomers in lignin depolymerization, with high degree of hydro-deoxygenation and hydrogenation of the aromatic ring [ 39 ]. It has been also reported that HTCs-Cu promote lignin depolymerization in supercritical methanol (scMeOH), preventing char formation [48]. Bimetallic catalysts combining Ni with noble metals and transition metals like Fe, Mo and Ti have been proposed in different studies to explore synergistic effects, thus seeking to increase the reactivity and selectivity [ 31 ]. Zhai et al. found that bimetallic NiFe supported on active carbons could effectively cleave ether linkages whilst suppressing hydrogenation of aromatic rings [ 49 ]. It was proposed that the formation of NiFe alloys tuned the reactivity of the active sites. HTC-Ni catalysts have already been used in the cleavage of model lignin dimeric compounds as well as in lignin-rich biorefinery streams derived from corn stover [ 44 ]. More specifically, enzymatically hydrolyzed lignins, some of which had been pretreated with diluted acids, were subjected to depolymerization, finding that the catalyst was effective in cleaving ether linkages, as evidenced by the formation of substantial amounts of 4-vinylphenol (4-VP) produced by the decarboxylation of p-coumaric acid. To the best of our knowledge, bimetallic HTC-NiCu catalysts have not been used in the depolymerization of lignins. In this contribution, homogeneous and heterogeneous catalytic methods were developed to depolymerize downstream biorefinery and technical lignins. The resultant reaction mixtures were thoroughly characterized to accurately determine the aromatic monomer yield and selectivity, the poly-(hydroxy)-aromatic fractions apparent masses and the extent of side-reactions. 2. Materials and Methods All high-purity liquid and solid chemicals were purchased from Carlo Erba, Fisher Scientific and Sigma-Aldrich and used as received. High-purity gases were supplied by Nippon Gases Spain. Synthetic hydrotalcite (Mg 6 Al 2 (CO 3 )(OH) 16· 4H 2 O), used as support for preparing the catalysts, was purchased from Sigma Aldrich (P.N. 652288). A commercial 5% Ru/C (50% water wet) catalyst was purchased from Alfa Aesar (P.N. 044338.14). Poplar (Populus sp.) was an industrial residue kindly provided as sawdust by Garnica Plywood Inc. (Baños de Rio Tobia, Spain). Poplar sawdust was air dried at room temperature (~295 K) for 72 h, then it was sieved to obtain a particle size distribution comprised between 0.5 and 1.0 mm (35–18 mesh). After sieving, the air-dried material was further dried in an oven at 376 K for 24 h. Biorefinery downstream lignins from pine (P. radiata) and poplar (Populus sp.), named as BioA and BioB, respectively, were kindly provided by CENER’s Biorefinery and Bioenergy Centre (Bio2C, Aoiz, Spain). Both P. radiata and Populus sp. were used as feedstocks in the production of 2 G bioethanol in a biorefinery, obtaining BioA and B as byproducts. The whole process for obtaining BioA and B starting solids is detailed in a previous study [ 50 ], although succinctly described below (please see Section 2.1). BioA and B lignins were received in powder form and used without further treatments. P. pinaster wood (PPW) chips were kindly supplied by Europac (Deocriste, Portugal). Prior to fractionation, as received PPW chips were air-dried and ground in a Retsch Cutting Mill SM 100 with a sieve of square holes of 4 mm. Further details can be found elsewhere [51]. 2.1. Lignin Isolation and Characterization Different strategies were used in this study for obtaining lignin from the four feedstocks described above (see Scheme 1), namely autohydrolysis followed by soda ethanosolv (SeolA and B lignins for pine and poplar respectively), direct organosolv (OrgB, for poplar organosolv) using isopropanol, and the use of biorefinery downstream lignins (BioA and B lignins for pine and poplar respectively). Biomolecules 2020,10, 1338 6 of 30 Biomolecules 2020, 10, x FOR PEER REVIEW 6 of 30 Scheme 1. Strategies for the isolation of lignins from the feedstocks selected in this work. Biorefinery downstream lignins, BioA and B, were obtained after acidic pretreatment at 465 K for 5 min of the pine and poplar feedstocks, respectively, which allowed to separate their corresponding hemicellulosic fractions and to valorize them thereafter. The resulting solids presented high accessibility to the cellulosic fraction that was hydrolyzed to glucose using an enzymatic cocktail, whilst yielding BioA and B as downstream biorefinery lignins [6,50]. Technical lignins, OrgB, SeolA and B, were isolated in the laboratory. SeolA and B were obtained from P. pinaster and waste poplar sawdust, respectively. Feedstocks were subjected to autohydrolysis at 448 K during 30 min to remove hemicelluloses, followed by soda ethanosolv digestion at 443 K during 90 min, using a NaOH load of 30 wt.% (on a wood basis) and a biomass to solvent ratio of 1:8 (w/w). SeolA isolation experiments were performed in a 1 L autoclave reactor (Parr Instruments), agitated by a double six-blade propeller while SeolB isolation experiments were performed in a 150 mL autoclave (Parr instruments) agitated by a four-blade propeller. On the other hand, OrgB was directly obtained following a direct organosolv procedure using several combinations of alcohols and acidic catalysts. Amongst them, organosolv digestion with ethanol and isopropanol at 463 K provided the best lignin yields. However, DOSY (see Figure S8 in Supporting Information) evidenced that OrgB obtained with ethanol presented more impurities, including some cellulosic residues. Therefore, OrgB obtained with isopropanol was selected for conducting the depolymerization assays. Lignin content was determined according to NREL/TP 510-42618. The ratio of syringyl (S) and guaiacyl (G) units, the S/G ratio, was determined as described in the literature in BioA and B (see Supporting Information SI.1) whereas 31 P NMR was used in the rest of lignins (see Supporting Information SI.2). Average molecular weight (M w ) and number molecular weight (M n ) were determined using Size Exclusion Chromatography (SEC, see details in Supporting Information SI.2), in lignins and depolymerization products. Additionally, the apparent masses were estimated using Diffusion Ordered Spectroscopy NMR (DOSY, see Supporting Information SI.3 for details). DOSY spectroscopy has been used in this work to monitor depolymerization reactions and to selectively determine the Scheme 1. Strategies for the isolation of lignins from the feedstocks selected in this work. Biorefinery downstream lignins, BioA and B, were obtained after acidic pretreatment at 465 K for 5 min of the pine and poplar feedstocks, respectively, which allowed to separate their corresponding hemicellulosic fractions and to valorize them thereafter. The resulting solids presented high accessibility to the cellulosic fraction that was hydrolyzed to glucose using an enzymatic cocktail, whilst yielding BioA and B as downstream biorefinery lignins [6,50]. Technical lignins, OrgB, SeolA and B, were isolated in the laboratory. SeolA and B were obtained from P. pinaster and waste poplar sawdust, respectively. Feedstocks were subjected to autohydrolysis at 448 K during 30 min to remove hemicelluloses, followed by soda ethanosolv digestion at 443 K during 90 min, using a NaOH load of 30 wt.% (on a wood basis) and a biomass to solvent ratio of 1:8 (w/w). SeolA isolation experiments were performed in a 1 L autoclave reactor (Parr Instruments), agitated by a double six-blade propeller while SeolB isolation experiments were performed in a 150 mL autoclave (Parr instruments) agitated by a four-blade propeller. On the other hand, OrgB was directly obtained following a direct organosolv procedure using several combinations of alcohols and acidic catalysts. Amongst them, organosolv digestion with ethanol and isopropanol at 463 K provided the best lignin yields. However, DOSY (see Figure S8 in Supporting Information) evidenced that OrgB obtained with ethanol presented more impurities, including some cellulosic residues. Therefore, OrgB obtained with isopropanol was selected for conducting the depolymerization assays. Lignin content was determined according to NREL/TP 510-42618. The ratio of syringyl (S) and guaiacyl (G) units, the S/G ratio, was determined as described in the literature in BioA and B (see Supporting Information SI.1) whereas 31 P NMR was used in the rest of lignins (see Supporting Information SI.2). Average molecular weight (M w ) and number molecular weight (M n ) were determined using Size Exclusion Chromatography (SEC, see details in Supporting Information SI.2), in lignins and depolymerization products. Additionally, the apparent masses were estimated using Diffusion Ordered Biomolecules 2020,10, 1338 7 of 30 Spectroscopy NMR (DOSY, see Supporting Information SI.3 for details). DOSY spectroscopy has been used in this work to monitor depolymerization reactions and to selectively determine the apparent mass of the poly-(hydroxy)-aromatic fraction in the reaction mixtures. Apparent masses were estimated using two calibration curves that correlated the average log D of the diffusion traces and their log MW (see reference [ 6 ] for details). The calibration curves were calculated using two different families of standards. The first was polystyrene (PS) standards that corresponded to dispersion forces, and the second was polyethylene glycol (PEG) and dimeric and monomeric phenolic standards—referred to as a PEG calibration curve)—that corresponded to Van der Waals and even hydrogen bond interactions with the solvent. Lignin and depolymerization products have been also analyzed using Heteronuclear Single Quantum Coherence Spectroscopy NMR, HSQC, and Total Correlation HSQC NMR spectroscopy, HSQC-TOCSY (see Supporting Information SI.3 for experimental details). 2.1.1. Organosolv Method (OrgB) The extraction of lignin from waste poplar sawdust was carried out in a 200 mL high-pressure stainless-steel reactor (BerghoffBR-300) equipped with a magnetic stirrer and a Teflon liner. In a typical isolation reaction, a 10% (w/v) suspension of poplar sawdust in isopropyl alcohol was pressurized with nitrogen (20 bar), stirred at 463 K for 3 h, and then allowed to cool down to room temperature. The reaction mixture was filtered, and the solid residue was washed thrice with isopropyl alcohol (3 ×20 mL) . The filtrate was concentrated to ca. 20 mL in a rotary evaporator, then poured onto distilled water (200 mL) at 273 K under vigorous stirring. The mixture was allowed to settle overnight, and the resulting brown precipitate was collected by filtration and vacuum dried at room temperature overnight, thus yielding OrgB. 2.1.2. Autohydrolysis and Soda Alkaline Ethanosolv Method (SeolA and B) SeolA and B lignins were obtained by fractionation of P. pinaster wood and poplar sawdust, respectively. Aqueous suspensions of the feedstocks in Milli-Q H 2 O (10% w/v) were prepared and readily put into the reactor. The suspensions were stirred at 448 K for 30 min and then allowed to cool down to room temperature. The reaction mixture was then filtered and the solid was washed with H2O. The solid was re-suspended in a 35% ethanol-water mixture (biomass-to-solvent mass ratio of 1:8) and reacted with 30 wt.% NaOH (on solid basis) at 443 K for 90 min, and then allowed to cool down again. The reaction mixture was then filtered, and the solids were washed with H 2 O. The combined filtrates were concentrated by rotary distillation. The resulting solution was brought to pH 5 upon addition of H 2 SO 4 and allowed to settle down. The resulting solids were collected by filtration, re-suspended in H 2 O and collected again by filtration. The as-obtained lignin was dried at 313 K for 7 days. 2.2. Preparation and Characterization of Hydrotalcite-Supported Metal Catalysts (HTC-M) A set of five HTC-M solids were prepared by impregnation with a total nominal metal content of 5 wt.%. Varying amounts of Ni and Cu were used in the preparation of the different HTC-M catalysts, resulting in HTC0-5, where the figure corresponds to the nominal nickel loading. The hydrotalcite-supported metal catalysts were prepared by wet impregnation of a commercial hydrotalcite. The support was calcined at 723 K for 24 h in a muffle furnace, thus obtaining the calcined support (HTC). Then, the HTC was rehydrated with H 2 O for 1.5 h and subjected to ultrasonic vibration for 5 min. Known amounts of stock solutions of Ni(NO 3 ) 2· 6H 2 O and Cu(NO 3 ) 2· 3H 2 O in ethanol were then added to the rehydrated HTC, to obtain catalysts with different nominal metal loadings: 5 wt.% Ni, (HTC-5), 4 wt.% Ni—1 wt.% Cu, (HTC-4), 2.5 wt.% Ni—2.5 wt.%-Cu (HTC-2.5), 1 wt.% Ni—4 wt.% Cu (HTC-1) and 5 wt.% Cu (HTC-0). The resulting suspensions were then stirred at room temperature for 5 min. The solvent was then evaporated, and the resulting solids were grinded into powder form for obtaining the different metal-HTC catalysts. Biomolecules 2020,10, 1338 8 of 30 The resulting solids were characterized by N 2 physisorption, Inductively Coupled Plasma-Optical Emission Spectrometry (ICP-OES) and X-ray Diffraction (XRD). N 2 adsorption-desorption isotherms were determined in a Micromeritics Gemini V 2380 static volumetric analyzer at 77 K. The samples were previously degassed at 473 K for 2 h under a N 2 gas flow. The specific surface area (m 2 /g) of the samples was determined by the Brunauer-Emmett-Teller (BET) method, whereas the specific pore volume and the average pore size were calculated by the Barrett–Joyner–Halenda (BJH) method. The elemental analysis of the catalysts was done by means of optical emission spectrometry with inductively coupled plasma (ICP-OES, Thermo Elemental IRIS INTREPID RADIAL, equipped with a Timberline IIS automatic). XRD analyses were done using a D-Max 2500 Rigaku diffractometer with CuK α radiation at 40 kV and 80 mA and scanning 2 θ from 5 to 95 ◦ . Both ICP-OES and XRD analyses were carried out at the “Servicio de Apoyo a la Investigaci ó n” of the Universidad de Zaragoza (Zaragoza, Spain). 2.3. Lignin Depolymerization Different strategies were planned for each of the isolated lignins depending on their physicochemical properties (see Scheme 2). These will be described in detail in the following subsections, including homogeneous base catalyzed depolymerization using aqueous NaOH, NaOH in ethanol-water mixtures, and heterogeneously catalyzed depolymerization using Ru/C or Ni-, Cuand NiCu-HTCs. Biomolecules 2020, 10, x FOR PEER REVIEW 8 of 30 isotherms were determined in a Micromeritics Gemini V 2380 static volumetric analyzer at 77 K. The samples were previously degassed at 473 K for 2 h under a N 2 gas flow. The specific surface area (m 2 /g) of the samples was determined by the Brunauer-Emmett-Teller (BET) method, whereas the specific pore volume and the average pore size were calculated by the Barrett– Joyner–Halenda (BJH) method. The elemental analysis of the catalysts was done by means of optical emission spectrometry with inductively coupled plasma (ICP-OES, Thermo Elemental IRIS INTREPID RADIAL, equipped with a Timberline IIS automatic). XRD analyses were done using a D-Max 2500 Rigaku diffractometer with CuKα radiation at 40 kV and 80 mA and scanning 2θ from 5 to 95°. Both ICP-OES and XRD analyses were carried out at the “Servicio de Apoyo a la Investigación” of the Universidad de Zaragoza (Zaragoza, Spain). 2.3. Lignin Depolymerization Different strategies were planned for each of the isolated lignins depending on their physicochemical properties (see Scheme 2). These will be described in detail in the following subsections, including homogeneous base catalyzed depolymerization using aqueous NaOH, NaOH in ethanolwater mixtures, and heterogeneously catalyzed depolymerization using Ru/C or Ni-, Cuand NiCuHTCs. Scheme 2. Lignin depolymerization strategies. 2.3.1. Base Catalyzed Depolymerization Using Aqueous NaOH or NaOH aq. (BioA473, BioB473, OrgB463) Depolymerization reactions were carried out using aqueous solutions of NaOH in a 100 mL batch stirred pressurized stainless-steel vessel reactor (Autoclave Engineers, EZ100RXR) equipped with a PID controller. Following this, 50 mL of lignin suspension in 0.25 M NaOH aq. (1% w/v; pH Scheme 2. Lignin depolymerization strategies. 2.3.1. Base Catalyzed Depolymerization Using Aqueous NaOH or NaOH aq. (BioA473, BioB473, OrgB463) Depolymerization reactions were carried out using aqueous solutions of NaOH in a 100 mL batch stirred pressurized stainless-steel vessel reactor (Autoclave Engineers, EZ100RXR) equipped with a PID controller. Following this, 50 mL of lignin suspension in 0.25 M NaOH aq. (1% w/v; pH 13.4, NaOH/solid ratio 25:1) was pressurized with N 2 (20 bar) and stirred at 463 K (OrgB) or 473 K (BioA or B) during 240 min. The resulting solutions were filtered, and the liquid was acidified to pH 2 upon addition Biomolecules 2020,10, 1338 9 of 30 of 1 M HCl. Then, 25 mL of a solution of bromobenzene (0.0015 M) in ethyl acetate, and additional ethyl acetate when necessary, was then added and the organic phase was separated. 1.0 mL of the organic phase was taken for the quantification of monomeric phenols by GC-FID. The identity of the monomeric phenols was confirmed by GC-MS (see Supporting Information SI.2 ). The rest was evaporated to dryness and the yield of the oily fraction recovered (herein referred to as bio-oil) was gravimetrically calculated and used for further characterization (Supporting Information SI.2 and SI.3). 2.3.2. Base Catalyzed Depolymerization Using NaOH in Aqueous EtOH or EtOH aq. (OrgB523, SeolA493, SeolA523 and SeolB523) Lignin (1.900 g) was suspended in a NaOH (1.800 g) solution in EtOH/H 2 O (35 wt.%, 50 mL) . The mixture was pressurized with nitrogen (20 bar), readily stirred at 493 K or 523 K in an Autoclave Engineers reactor for 1 h and then allowed to cool down. The reaction mixture was filtered, and the solids were washed. The ethanol in the combined filtrate was removed by rotary evaporation and the resulting solution was brought to pH 2 upon addition of 1 M HCl. The subsequent procedure for quantifying monomeric phenols in the samples by GC-FID and to determine the bio-oil yield is the same as that detailed in the previous subsection (see above). 2.3.3. Depolymerization Using Ru/C or HTC-M Catalysts A solution of lignin or solids SeolA 493 or A 523 (250 mg) in ethanol (50 mL) and a load of catalyst (250 mg of HTC-M or 500 mg of Ru/C -50% water wet-) were placed in the autoclave reactor, which was then sealed. Air was purged with nitrogen thrice and the reactor was subsequently pressurized with nitrogen (20 bar) or hydrogen (30 bar). The reaction medium was then stirred at 523 K during 30 min (Ru/C catalyst) or 543 K during 60 min (HTC-M catalyst). The reactor was then allowed to cool down, the reaction mixture was filtered and the solid was washed with ethanol (20 mL). The combined filtrates were concentrated by rotary evaporation and then distilled water (50 mL) was added. The next steps for quantifying monomeric phenols in the samples are the same as those detailed in the two previous subsections (see above). 3. Results 3.1. Characterization of the Different Lignins Table 2shows the main characteristics of the lignins used in this study. As expected, the purity of the technical lignins SeolA, and OrgB was much higher (98.0 wt.% and 96.1 wt.%, respectively) than those of downstream biorefinery lignins, BioA and B (51.1 wt.% and 69.1 wt.%, respectively). It is important to note that BioA and B presented a relatively high sulfur content coming from the acidic thermal pretreatment with sulfuric acid [ 50 ], which causes some sulfonation in the aromatic rings. Most of the impurities in biorefinery lignins, 38.8 wt.% in BioB and 23.2 wt.% in BioA, corresponded to glycans that were not fully removed upon enzymatic hydrolysis. Opposite to this, neither hemicelluloses nor celluloses or glycans were detected in SeolA and B, whereas only 0.4 wt.% of glycans were found in OrgB. In the case of SeolB, lignin content in the isolated solid was surprisingly low, 68.7 wt.%, with a high ash content, 30.1 wt.%. The reason for this high content in ashes could be a non-efficient removal of salts formed during lignin isolation. According to the SEC analyses, downstream biorefinery lignins (see Figure S2 in Supporting Information) presented much broader mass distributions than technical lignins. BioB presented its maximum at 1038 Da and BioA at 228 Da. However, fractions beyond 32000 Da were found in both cases. Although SEC for biorefinery lignins was carried out using 0.1% LiBr in N,N-dimethylformamide, neither BioA nor B could be completely solubilized. This is probably the reason for the high dispersity index found for these solids. Presumably, low molecular weight fractions are much more soluble and provide the relatively low M n in both cases, 535 Da and 414 Da, respectively. Size distributions in SeolA, B and OrgB were much narrower, as it is pointed by the dispersity index, which is, in all Biomolecules 2020,10, 1338 16 of 30 and A 523-Ru , but there is a strong increase in 10 (from 20% to 3%) together with a decrease in 18 (10% to 2%, at 30 min of reaction time), and gigantol, 19, (6% to 2%, at 30 min of reaction time). Next, SeolB was reacted following the optimized procedure for SeolA. Monomer yields for SeolB 523 (9.7 wt.%) were, however, much higher than for SeolA 523-Ru , because of lower cross-linking in the starting lignin. The most abundant monomers were 4-hydroxy-3-methoxyphenylacetone (36%), 15, followed by syringol, 20 (17%). Further hydrogenolysis in the presence of Ru/C produced a slight increase in the aromatics yield up to 11.2 wt.% with 15 being the most abundant bioaromatic (45%) followed by 12 (16%) and 20 (15%). Finally, treatment of OrgB with NaOH at 523 K in EtOH/H 2 O, OrgB 523 , provided 7.5 wt.% yield to phenolic monomers, being 20 (34%) the most abundant followed by phenols, while 15 was not detected and 26 accounted for just 11% of the bioaromatics. Hydrogenation of OrgB 523 also produced a slight increase in aromatic monomers, 8.7 wt.%, with 15 (36%) and 20 (19%) being the most abundant monomers. This suggested that most of the depolymerization in SeolB and OrgB occurred under BCD in EtOH/H 2 O at 523 K, while Ru/C catalyzed hydrogenolysis had little effect in monomer yields. 3.4. Poly-(Hydroxy)-Aromatic Fractions BioB was subjected to BCD in the presence of NaOH at 473 K, BioB 473 . DOSY spectra (see Figure 1a) clearly showed the difference in apparent mass between the aromatic hydrogen atoms in the starting lignin and those corresponding to the BCD reaction mixture, with fractions in the range of 360 Da–1060 Da. Moreover, some diffusion traces were detected in the aliphatic region (see Figure S5 in Supporting Information), which correspond to degradation products from the peeling reactions of saccharides in the basic medium [ 58 ]. A similar observation can be made when OrgB is reacted under similar reaction conditions at 463 K (see Figure 1c), OrgB 463 . The apparent mass decreased from 1454 Da to 709 Da (see Table 6) according to DOSY measurements in the aromatic region, although the SEC analyses did not show any noticeable change but a slight shift to lower masses (see Figure 1d). However, when the same reaction was carried out in ethanol under the same conditions, the mass distribution changed drastically and the size distribution narrowed with a prominent peak at ca. 530 Da, although the monomer yield decreased to 2.1 wt.% (see Table 4, Table 6and Figure S9). The maximum monomer yield in the depolymerization of OrgB was obtained using HTC-1 in scEtOH (543 K), with an increase in the monomers yield up to 6.8 wt.%. M n determined by SEC was 694 Da while the apparent mass determined by DOSY was 428 Da, which represents a noticeable decrease in apparent mass from the starting OrgB and B 463 in H 2 O. Size distribution was clearly shifted to lower masses when BCD was catalyzed by HTC-M, OrgB HTCx, whatever the Ni and/or Cu loading. In addition, DOSY measurements (see Figure 1) showed traces with noticeable lower apparent masses than OrgB 463 , which evidenced that depolymerization is much more extensive when the reaction is carried in ethanol than when using H2O as solvent, which can be due to the highest solubility of OrgB in ethanol than in H2O. SeolA and B depolymerization reactions were conducted using a different strategy that consisted of a two-step approach. Firstly, NaOH treatment in EtOH/H 2 O was carried out at 493 K or 523 K for 60 min runs, SeolA 493 and A 523 . The monomer yields in both cases were low, 0.6 wt.% and 1.0 wt.%, respectively (see Table 5). However, DOSY for SeolA 493 and A 523 showed in both cases that depolymerization of SeolA did occur, being more extensive in SeolA 523 , 660 Da, than in SeolA 493 1039 Da (see Table 6). The most representative diffusion trace for SeolA 493 was centred at 800 Da but representative traces were present in the 800–1600 Da range, whereas in the case of SeolA 523 , the most representative diffusion traces appeared centred at 630 Da with some small diffusion traces around 1024 Da (see Figure 2). Hydrogenolysis of SeolA 493 and A 523 was done in scEtOH at 523 K in the presence of Ru/C (20 bar) . As expected, the monomer yields were higher than in starting SeolA 493 and A 523 , reaching 4.2 wt.% in SeolA 493-Ru and 7.3 wt.% in SeolA 523-Ru . DOSY spectra showed in this case a more pronounced effect of the treatment in SeolA 493-Ru , since its apparent mass decreased from 1039 Da to 691 Da (see Table 6) . In the case of SeolA 523-Ru the apparent mass was similar to that of SeolA 523 (ca. 660 Da) Biomolecules 2020,10, 1338 17 of 30 and, although DOSY spectra showed some slight differences between SeolA 523 and A 523-Ru , it can be appreciated that an important fraction of the diffusion in SeolA523 overlaps with those of SeolA523-Ru (see Figure 2). Noteworthy, the estimated apparent masses, 746 Da, were only slightly higher when SeolA was directly treated with Ru/C at 523 K in ethanol and diffusion traces were essentially similar to those of SeolA 523-Ru (see Table 6and Figure 3) although its chemical monomer yield was much lower. Biomolecules 2020, 10, x FOR PEER REVIEW 18 of 30 SeolA was directly treated with Ru/C at 523 K in ethanol and diffusion traces were essentially similar to those of SeolA 523-Ru (see Table 6 and Figure 3) although its chemical monomer yield was much lower. Figure 1. (a) Aromatic region for the DOSY spectra of BioB (black) and BioB 473 (blue); (b) SEC for BioB (red) and BioB 473 (blue); (c) DOSY spectra and (d) SEC for OrgB (black), OrgB 463 (red) and OrgB HTC2.5 (blue). Figures in black, red and blue correspond to polystyrene, PS, calibration. Data provided by DOSY spectroscopy shed light about the somewhat strange SEC chromatograms obtained after depolymerization of SeolA (see Figures 2 and 3). Apparent masses determined by DOSY do not seem to match with the most prominent peaks determined by SEC. Thus, in all analysed samples, a peak at ca. 130 Da appeared. This peak is the most intense in the samples treated with Ru/C in EtOH, but it is also important in OrgB samples depolymerized using HTC in EtOH. This is consistent with the detection of diffusion traces in the aliphatic region of the DOSY spectra that might correspond to compounds arising from ethanol via hydrogen transfer mechanism (see Figure 4 and Figure S7). Besides this peak, the rest of SEC chromatograms presented a broad distribution which is centred at similar mass values than those estimated by DOSY for the aromatic region (i.e., poly- (hydroxy)-aromatic fraction). In the case of SeolA 493 and A 493-Ru , broad mass fractions were present, which were more important in SeolA 493, centred at 1360 Da, and at 570 Da in SeolA 493-Ru. In SeolA 523 , A 523-Ru and A Ru mass distributions were much narrower and presented a prominent peak at ca. 360 Da (see Figure 2). OrgB HTC presented also a broad distribution with two major peaks centred at 526 Da and 758 Da that may correspond to three and four-unit aromatic compounds. Figure 1. ( a ) Aromatic region for the DOSY spectra of BioB (black) and BioB 473 (blue); ( b ) SEC for BioB (red) and BioB 473 (blue); ( c ) DOSY spectra and ( d ) SEC for OrgB (black), OrgB 463 (red) and OrgB HTC2.5 (blue). Figures in black, red and blue correspond to polystyrene, PS, calibration. Data provided by DOSY spectroscopy shed light about the somewhat strange SEC chromatograms obtained after depolymerization of SeolA (see Figures 2and 3). Apparent masses determined by DOSY do not seem to match with the most prominent peaks determined by SEC. Thus, in all analysed samples, a peak at ca. 130 Da appeared. This peak is the most intense in the samples treated with Ru/C in EtOH, but it is also important in OrgB samples depolymerized using HTC in EtOH. This is consistent with the detection of diffusion traces in the aliphatic region of the DOSY spectra that might correspond to compounds arising from ethanol via hydrogen transfer mechanism (see Figure 4and Figure S7). Besides this peak, the rest of SEC chromatograms presented a broad distribution which is centred at similar mass values than those estimated by DOSY for the aromatic region (i.e., poly-(hydroxy)-aromatic fraction). In the case of SeolA 493 and A 493-Ru , broad mass fractions were present, which were more important in SeolA 493, centred at 1360 Da, and at 570 Da in SeolA 493-Ru. In SeolA 523 , A 523-Ru and A Ru mass distributions were much narrower and presented a prominent peak at ca. 360 Da (see Figure 2). OrgB HTC presented also a broad distribution with two major peaks centred at 526 Da and 758 Da that may correspond to three and four-unit aromatic compounds. Biomolecules 2020,10, 1338 18 of 30 Table 5. Distribution and yields (Y%) of monomers afrom different depolymerization experiments carried at 523 K. Lignin Cat bSolvent t (Min) Y (wt.%). 1–9 10 11 12 13 15 16 17 18 19 20 21 24 25 26 SeolA cNaOH EtOH/H2O 60 0.6 - 18 3 18 0 1 13 13 14 0 - - 5 5 - SeolA493 Ru/C EtOH 0 2.4 - 12 10 54 5 - - - 7 5 - - 4 0 - 30 d2.2 4 9 10 42 4 1 - - 11 6 - - 7 2 - 60 d3.7 - 17 11 46 5 - - - 7 6 - - 4 - - 120 d4.2 2 19 11 36 3 3 - - 9 10 - - 3 - - 240 d4.2 - 20 10 43 3 1 - - 6 8 - - 1 2 - SeolA NaOH EtOH/H2O60 d1.1 1 45 8 29 3 0 - - 5 2 - - - 5 - SeolA523 Ru/C EtOH 0 5 - 30 12 46 4 0 - - 2 4 - - 0 0 - 30 d7.3 3 31 8 43 4 0 - - 2 2 - - - 4 - 60 5.1 - 32 11 44 4 2 - - - 0 - - - - - 120 6.7 - 33 11 42 4 0 - - 3 4 - - 0 0 - 240 d6.6 - 32 12 44 4 - - - 2 4 - - - 0 - SeolB NaOH EtOH/H2O 60 9.7 2 11 3 14 9 36 - - - - 17 3 - 1 3 SeolB523 Ru/C EtOH 30 11.2 1 7 3 16 1 45 - - - - 15 8 - 2 1 OrgB NaOH EtOH/H2O 60 7.5 27 7 0 12 0 - - - - - 34 9 - - 11 OrgB523 Ru/C EtOH 30 8.7 8 5 3 9 1 36 - - - - 19 7 - 7 2 SeolA Ru/C EtOH 30 d3.5 2 7 8 51 15 1 - - 5 6 - - - - - SeolB Ru/C EtOH 30 3.9 - - - 11 4 32 - - - - 14 11 - 5 20 OrgB Ru/C EtOH 30 6.3 5 4 1 - 8 31 - - - - - - 5 - 37 a (1) Phenol; (2) 2-ethylphenol; (3) o-cresol; (4-5) m-cresol/p-cresol; (6) Catechol; (7) 4-methylcatechol; (8) 4-ethylcatechol; (10) Guaiacol; (11) 4-methylguaiacol; (12) 4-ethylguaiacol; (13) 4-Propylguaiacol; (14) Vanillin; (15) 4-hydroxy-3-methoxyphenylacetone; (16) Homovanillyl alcohol; (17) Homovanillic acid; (18) Acetovanillone; (19) Gigantol; (20) 2,6 dimethoxyphenol (Syringol); (21) 4-methylsyringol; (24) Syringaldehyde; (25) Acetosyringone; (26) Homosyringaldehyde. bCat.: catalyst; creaction carried at 493 K; dreactions were made by triplicate. Biomolecules 2020,10, 1338 19 of 30 Table 6. Apparent masses estimated by SEC and NMR DOSY for the starting lignins and samples. SEC DOSY NMR Sample Mw(Da) Mn(Da) Aromatic Region (Da) Aliphatic Regions (Da) Aromatic-OH (mmol/g) Aromatic-C (%) BioB473 1051 719 658 482 418 3.30 58 BioA473 1173 720 764 494 520 n.m. 54 SeolA 3005 2296 2200 279 553 4.57 97 SeolA493 1798 891 1039 285 463 3.58 83 SeolA493-Ru 811 523 691 441 502 2.97 61 SeolA523 1250 587 660 322 401 4.51 66 SeolA523-Ru 818 468 665 433 540 3.74 56 SeolARu 956 752 746 562 509 2.77 65 SeolB 1791 1206 1477 869 n.m. 1.69 98 SeolB523 1794 1194 742 492 447 2.54 68 SeolB523-Ru 1464 962 671 406 527 2.11 51 SeolBRu 1434 980 720 539 425 n.m. 71 OrgB 1636 1230 1454 548 950 4.72 90 OrgB (EtOH) 1650 1195 1529 286 764 n.m. 85 OrgB463 1578 1186 709 245 515 5.10 73 OrgB463-EtOH 948 702 770 414 480 2.20 62 OrgB543-EtOH 1515 924 267 191 273 n.m. 16 OrgBEtOH-543 920 668 723 365 605 n.m. 65 OrgBHTC0 1328 724 420 348 421 n.m. 38 OrgBHTC1 908 694 428 396 459 n.m. 38 OrgBHTC2.5 821 6478 583 381 462 2.61 44 OrgBHTC4 1003 725 437 406 396 nm 49 OrgBHTC5 968 681 478 276 381 nm 57 OrgB523 1163 764 601 320 450 n.m. 61 OrgB523-Ru 836 593 638 319 457 n.m. 44 OrgBRu 888 682 559 316 559 n.m. 68 Biomolecules 2020,10, 1338 20 of 30 Biomolecules 2020, 10, x FOR PEER REVIEW 19 of 30 Figure 2. (a) DOSY in the aromatic region and (b) SEC for SeolA (black), A 493 (blue) and A 493-Ru (red). (c) DOSY in the aromatic region and (d) SEC for SeolA (black), A 523 (blue) and A 523-Ru (red). Figures in red and blue correspond to PS calibration. It is important to note that in all depolymerization reactions of OrgB, carried out in EtOH in the presence of either NaOH or HTCs with different Ni-Cu loadings, no significant differences in M n (ca. 700 Da) could be found among the different Ni-Cu loadings or in the blank reaction. However, when BCD was carried out in H 2 O in the presence of NaOH, M n was much higher, ca. 1200 Da, which suggests that fractionation of lignin is caused by the ethanolic solvent at high temperatures (viz. thermos-solvolysis). Similarly, direct hydrogenolysis in the presence of Ru/C in ethanol and BCD in ethanol/H2O at 523 K of SeolA (see Figure 3), B and OrgB, M n was in the range of 600–700 Da, that also suggest the essential role of thermos-solvolysis in lignin depolymerization. Diffusion traces in the range of 300–330 Da using the PS calibration curve or in the range of 185– 300 Da using PEG curves, could be detected in all the DOSY spectra. This agrees with the peak that can be observed in most SEC chromatograms in the range of 275–350 Da. These diffusion traces in the aromatic region can be attributed to the presence of dimers (e.g., guaiacylglycerol-β-guaiacyl ether, gigantol, etc.). It must be highlighted that in all the studied samples derived from technical lignins, SeolA, B and OrgB, diffusion traces in this apparent mass range could be detected in the aromatic region, which suggests that most of this peak may come from the presence of two-unit aromatic compounds. However (see Figure 4 and below), diffusion traces corresponding to these relatively low apparent masses were much more intense in the aliphatic regions. Figure 2. ( a ) DOSY in the aromatic region and ( b ) SEC for SeolA (black), A 493 (blue) and A 493-Ru (red). ( c ) DOSY in the aromatic region and ( d ) SEC for SeolA (black), A 523 (blue) and A 523-Ru (red). Figures in red and blue correspond to PS calibration. Biomolecules 2020, 10, x FOR PEER REVIEW 20 of 30 Figure 3. (a) DOSY spectra and (b) SEC for SeolA (black), A 523-Ru (red) and A Ru (blue). Figures on the left correspond to PS calibration. Figure 4. DOSY spectrum for SeolA 523-Ru . Figures on the left correspond to PS calibration and figures on the right to PEG calibration. Red dotted squares correspond to the integration regions in the aromatic and in the aliphatic regions. DOSY spectra presented in most cases a remarkable difference between the diffusion coefficients associated to the aromatic and the aliphatic regions, being the latter significantly higher. Diffusion traces in the DOSY spectra (see Figure 4) associated to the aromatic region are related to apparent masses above 400 Da, and centered in the range of 486–630 Da, whereas the aliphatic regions are below 500 Da, according to PS calibration, and below 300 Da according to PEG calibration (for the comparison of average apparent masses in both regions see Table 6). In a previous work [6], these traces in the aliphatic region were associated to the peeling reaction of saccharide impurities in the starting lignins, as is the case for BioA and B. However, diffusion traces at low apparent masses were also observed in the depolymerization of technical lignins, as it can be clearly seen for SeolA 523-Ru (see Figure 4). GC-MS analysis from the depolymerization reactions carried in SeolA, B and OrgB presented products whose origin can be associated to side reactions from ethanol (viz. hexanol, 2ethyl-butan-1-ol, 4-methyltetrahydro-2H-pyran-2-one, etc.). Both Ru/C and Cu-containing HTC Figure 3. ( a ) DOSY spectra and ( b ) SEC for SeolA (black), A 523-Ru (red) and A Ru (blue). Figures on the left correspond to PS calibration. Biomolecules 2020,10, 1338 21 of 30 Biomolecules 2020, 10, x FOR PEER REVIEW 20 of 30 Figure 3. (a) DOSY spectra and (b) SEC for SeolA (black), A 523-Ru (red) and A Ru (blue). Figures on the left correspond to PS calibration. Figure 4. DOSY spectrum for SeolA 523-Ru . Figures on the left correspond to PS calibration and figures on the right to PEG calibration. Red dotted squares correspond to the integration regions in the aromatic and in the aliphatic regions. DOSY spectra presented in most cases a remarkable difference between the diffusion coefficients associated to the aromatic and the aliphatic regions, being the latter significantly higher. Diffusion traces in the DOSY spectra (see Figure 4) associated to the aromatic region are related to apparent masses above 400 Da, and centered in the range of 486–630 Da, whereas the aliphatic regions are below 500 Da, according to PS calibration, and below 300 Da according to PEG calibration (for the comparison of average apparent masses in both regions see Table 6). In a previous work [6], these traces in the aliphatic region were associated to the peeling reaction of saccharide impurities in the starting lignins, as is the case for BioA and B. However, diffusion traces at low apparent masses were also observed in the depolymerization of technical lignins, as it can be clearly seen for SeolA 523-Ru (see Figure 4). GC-MS analysis from the depolymerization reactions carried in SeolA, B and OrgB presented products whose origin can be associated to side reactions from ethanol (viz. hexanol, 2ethyl-butan-1-ol, 4-methyltetrahydro-2H-pyran-2-one, etc.). Both Ru/C and Cu-containing HTC Figure 4. DOSY spectrum for SeolA523-Ru. Figures on the left correspond to PS calibration and figures on the right to PEG calibration. Red dotted squares correspond to the integration regions in the aromatic and in the aliphatic regions. It is important to note that in all depolymerization reactions of OrgB, carried out in EtOH in the presence of either NaOH or HTCs with different Ni-Cu loadings, no significant differences in M n (ca. 700 Da) could be found among the different Ni-Cu loadings or in the blank reaction. However, when BCD was carried out in H 2 O in the presence of NaOH, M n was much higher, ca. 1200 Da, which suggests that fractionation of lignin is caused by the ethanolic solvent at high temperatures (viz. thermos-solvolysis). Similarly, direct hydrogenolysis in the presence of Ru/C in ethanol and BCD in ethanol/H2O at 523 K of SeolA (see Figure 3), B and OrgB, M n was in the range of 600–700 Da, that also suggest the essential role of thermos-solvolysis in lignin depolymerization. Diffusion traces in the range of 300–330 Da using the PS calibration curve or in the range of 185–300 Da using PEG curves, could be detected in all the DOSY spectra. This agrees with the peak that can be observed in most SEC chromatograms in the range of 275–350 Da. These diffusion traces in the aromatic region can be attributed to the presence of dimers (e.g., guaiacylglycerolβ -guaiacyl ether, gigantol, etc.). It must be highlighted that in all the studied samples derived from technical lignins, SeolA, B and OrgB, diffusion traces in this apparent mass range could be detected in the aromatic region, which suggests that most of this peak may come from the presence of two-unit aromatic compounds. However (see Figure 4and below), diffusion traces corresponding to these relatively low apparent masses were much more intense in the aliphatic regions. DOSY spectra presented in most cases a remarkable difference between the diffusion coefficients associated to the aromatic and the aliphatic regions, being the latter significantly higher. Diffusion traces in the DOSY spectra (see Figure 4) associated to the aromatic region are related to apparent masses above 400 Da, and centered in the range of 486–630 Da, whereas the aliphatic regions are below 500 Da, according to PS calibration, and below 300 Da according to PEG calibration (for the comparison of average apparent masses in both regions see Table 6). In a previous work [ 6 ], these traces in the aliphatic region were associated to the peeling reaction of saccharide impurities in the starting lignins, as is the case for BioA and B. However, diffusion traces at low apparent masses were also observed in the depolymerization of technical lignins, as it can be clearly seen for SeolA 523-Ru (see Figure 4) . GC-MS analysis from the depolymerization reactions carried in SeolA, B and OrgB presented products whose origin can be associated to side reactions from ethanol (viz. hexanol, 2-ethyl-butan-1-ol, Biomolecules 2020,10, 1338 22 of 30 4-methyltetrahydro-2H-pyran-2-one, etc.). Both Ru/C and Cu-containing HTC catalysts are able to promote hydrogen transfer reactions from ethanol that prompt the Guerbet reaction [ 47 , 59 ] that produces long chain alcohols. Given the reaction conditions, it can be expected that higher mass ethanol derivatives, that were not detected by GC-MS were produced. Semi-q-DOSY was used in Section 3.1 to estimate the purity of isolated lignin. A similar approach was used to estimate the percent of C atoms that correspond to aromatic rings in the reaction mixtures (see Table 6). The percentage of aromatic C is relatively low in the depolymerization of BioA and B (54% and 58%) because of the peeling reaction of partially degraded saccharides in the starting lignins. OrgB 463 (73%) presents higher aromatic C content than BioA 473 and B 473 because no degradation of saccharides occurred upon BCD. SeolA Ru , B Ru and OrgB Ru also presented relatively low contents in aromatic C (65%, 71% and 68% respectively) according to DOSY integration. However, in this case the decrease in the aromatic percentage of C atoms accounts for the formation of aliphatic compounds upon reaction of ethanol at high temperatures, following the hydrogen transfer mechanism in a similar fashion than in the Guerbet reaction. The estimated aromatic C is higher in SeolA 493 and A 523 (83% and 66%) than in SeolA 493-Ru and A 523-Ru (61% and 56%) respectively that is due to more extensive ethanol side reaction upon reaction at 523 K. A similar trend was observed in SeolA 523 and A 523-Ru (66% and 56%) and OrgB523 and B523-Ru (61% and 44%). This effect was more pronounced when OrgB reacted with NaOH. 73% and 62% of aromatic C were estimated when the reaction was run at 463 K in H 2 O and ethanol respectively, whereas only 16% was estimated when the reaction was run in ethanol at 543 K. Even more interesting, when the BCD was carried out using HTCs at 543 K, the estimated aromatic carbon increased with increasing nickel contents. Thus, 38% of aromatic C was estimated for treatments with HTC-0 and HTC-1, 44% for HTC-2.5, 49% for HTC-4 and 57% for HTC-5. This percentage of aromatic carbons was low, suggesting that the Guerbet reaction took place to a great extent, which is consistent with previous reports that described that the Guerbet reaction is catalyzed by HTC-Cu catalysts [47]. 31 P NMR analyses after derivatization of some reaction mixtures provided the number of aromatic hydroxy groups per gram of sample (see Table 6). This can be also a good parameter to evaluate the purity of the mixture in terms of poly-(hydroxy)-aromatic content. Indeed, this parameter followed the same trend that the aromatic C content in comparable reaction mixtures. In the case of the depolymerization of OrgB it can be observed that OrgB 463 carried out in H 2 O presented 5.09 mmol aromatic OH/g while the same reaction in ethanolic medium provided a reaction mixture with 2.20 mmol aromatic OH/g. Similarly, OrgB HTC2.5 contained 2.61 mmol aromatic OH/g. In the reaction mixtures of the SeolA series it can be also observed that the content of aromatic OH/g decreased from SeolA 493 to A 493-Ru and from SeolA 523 to A 523-Ru , which can be attributed to the increase in aliphatic carbon derived from side reactions produced by the presence of ethanol. Some of the diffusion traces detected in the aliphatic region may also come from C-and O-alkylation under reaction conditions with scEtOH, as it has been previously described [ 39 , 47 , 60 ]. HSQC of SeolA 523 and A 523-Ru in the aliphatic region (see Figure S14 in Supporting Information) presented cross-peaks that are compatible with alkylated aromatic compounds. A similar observation can be made for OrgB HTC2.5 (see Figure S15 in Supporting Information), but not in those reactions carried in H 2 O. However, the intensity of these signals was relatively low and was only detected after selective peak-picking treatment of the raw spectrum, and their diffusion coefficient was slightly higher than that of the corresponding aromatic region. Finally, O-alkylated and C-alkylated (others than gathered in Tables 4and 5) products were not detected by GC-MS. 4. Discussion The results shown above reveal that BCD in the aqueous medium is still a valuable option for the depolymerization of highly recalcitrant lignins, because the basic media promotes the formation of phenolate anions that increase the solubility of the lignin, which evolve to quinone methides Biomolecules 2020,10, 1338 23 of 30 initiating the process. It must be noted that BioA and B were obtained after acidic thermochemical treatment of poplar and pine chips. This acidic pretreatment promotes the formation of C–C bonds between the aromatic units that increased the recalcitrance of the lignin, which explains their low solubility in organic solvents [ 42 ]. Organosolv-derived lignin from poplar sawdust, OrgB, was poorly solubilized in aqueous medium, hence yielding low amounts of depolymerization products in NaOH BCD, whereas the opposite situation took place in ethanolic medium, which not only solubilized OrgB lignin effectively but also acted as reactant causing thermo-solvolysis. BCD reaction conditions for OrgB in ethanol were harsh that provoked repolymerization and, therefore, low monomer yields. As concerns heterogeneously catalyzed depolymerization of lignins using Ni-, Cuand NiCu-HTC catalysts, the results show that the yield to aromatics only increased slightly in comparison to homogeneous BCD. The catalysts primarily exerted a role in modifying the selectivity to the different products. Moreover, impregnating the HTC supports with Ni and/or Cu also caused activity and selectivity to vary, and synergistic effects were observed in the simultaneous use of Ni and Cu as active metals. Nonetheless, the formulation of bimetallic NiCu-HTC-s must be carefully done. In fact, the increase in monomers containing carbonyl groups (15 and 26) taking place upon depolymerization using the different HTC-M catalysts prepared can be associated with the decrease in copper content. These compounds are more prone to repolymerization reaction under basic conditions that can be in the origin of the decrease in the overall monomer yield at the lowest copper loadings. One important feature in HTC/Ni-Cu catalysts is that, in the opposite way to BCD using NaOH, monomers with side alkyl chains are produced with noticeable amounts of alkyl-guaiacols, 12–13 and 15 and alkyl-syringols, 26, which are hardly detected in NaOH depolymerization of BioB and OrgB in aqueous medium. This difference can be due either to the depolymerization mechanism itself or to the C-alkylation that may occur after depolymerization, as it has been already described in the presence of Cu-Mg-Al mixed oxides as reported by Huang et al. [ 47 ]. Concerning the depolymerization mechanism, in the case of Ni-HTC, Kruger et al. [ 44 ] suggested that the nitrate ion has an important role. Under the reaction temperature used in this study, nitrates can be decomposed to NO and NO 2 that would promote lignin nitration that may accelerate the β -O-4 cleavage. In the same study, it was found vinylphenol, from decarboxylation of p-coumaric acid, as the major monomer in the Ni-HTC depolymerization of lignin. In addition, as mentioned before, Ni-modified catalysts present strong binding sites for ether linkages and specifically cleave the C–O linkages, yielding side alkyl chains in the presence of hydrogen [ 45 ], which in this case would be provided by the dehydrogenation of ethanol that is in the origin of the Guerbet reaction (see below) [ 59 ]. Concerning Cand Oalkylation, the Guerbet reaction and alkylation using HTC-Cu catalysts have also been described by Huang et al. [47] . In some cases, methylation of the aromatic ring was also observed when using methanol as solvent medium and HTC-Cu catalysts [ 61 ]. In any case, neither O-alkylated phenols nor other aromatics than those gathered in Tables 4and 5were detected by GC-MS in the present study. Although the as-used reaction conditions are more similar to those described by Kruger et al. than those described by Huang et al. (543 K vs. 613 K and 653 K), the relatively high contents of homosyringaldehyde and ethylguaiacol in the HTC-catalyzed reaction do not allow us to totally discard C-alkylation. The Guerbet reaction consists of a sequence of reactions that starts with metal-catalyzed dehydrogenation followed by aldol-condensation, dehydration and metal-catalyzed hydrogenation, which can lead to long chain alcohols under our reaction conditions. Indeed, as can be inferred from the DOSY and HSQC analyses presented above, the extent of the Guerbet reaction when using the Ni-, Cuand NiCu-HTC catalysts is very large, which is in agreement with previous works [59]. As stated in the introduction section, other authors studied depolymerization of lignin-rich biorefinery streams derived from enzymatic hydrolysis of corn stover using Ni-HTC catalysts [ 44 ]. Despite selecting milder operating conditions in the present study, similar M n values were attained, and monomer yields were similar than those presented by Kruger et al. [ 44 ]. However, it seems likely that in that study neither saccharide degradation reactions nor self-condensation reactions of the alcohol used as reaction medium were considered, which meant that monomer yields were lower Biomolecules 2020,10, 1338 24 of 30 than expected from SEC. Most of the fractionation attained is likely caused by the synergistic effect of the solvent and the HTC solids. It is interesting that the role of Ni in the depolymerization is somewhat less effective than the catalytic role of nitrates, which differs with the observations found in Sturgeon et al. [43]. However, it must be noted that the yields to monomers were not reported in that work. It could be possible that the decrease in M n noticed in their SEC analyses would be mainly caused by thermal cracking reactions. In fact, the prevailing monomer was guaiacylketone, which seems logical if thermal cracking is the prevailing depolymerization route. The results obtained using the Ru/C commercial catalyst showed that depolymerization of technical lignins obtained from pine or poplar wood was moderate in terms of activity, and that the noble metal catalyst could not outperform the HTC-M catalysts in OrgB. It is worth noting, however, that bio-aromatic based compound yield for OrgB was slightly higher than in SeolA and B, which can be attributed to a milder process upon its isolation (i.e., autohydrolysis step was not carried for OrgB). Notwithstanding that cleavage of alkyl side-chain bonds readily took place, C–C double bonds and carbonyl groups are more extensively reduced in the presence of this catalyst. Moreover, a number of aliphatic compounds as octanol, hexanol or 4-methyltetrahydro-2H-pyran-2-one were also detected by GC-MS, which suggests that hydrogenolysis proceeded via hydrogen transfer from ethanol, as it has been previously described for lignin depolymerization with Ru/C [ 62 , 63 ] or molybdenum carbide [ 64 ]. The formation of these byproducts may follow a similar mechanism to that described for HTC-catalyzed reactions (see above). Indeed, the proposed mechanism for Ru catalyzed hydrogenolysis in MeOH at 493 K indicates that the plausible pathway for Ru/C-catalyzed hydrogenolysis of lignin leading to C3 and C2-fragmented monophenols starts from a nucleophilic attack of MeOH on the benzylic position of β -O-4 units of lignin to give α -OMe β -O-4 structures and formaldehyde. Ru/C catalysts play an essential role on the regeneration of MeOH from formaldehyde and on the hydrogenation of the alkenyl side chains [ 17 ]. In fact, when the reaction was run under nitrogen atmosphere, no significant differences were found in comparison to the homologous run carried out under hydrogen atmosphere, which confirmed that the main reaction mechanism is hydrogen transfer from ethanol catalyzed by ruthenium. Furthermore, it has been described that using Ru/C in the presence of a base causes a decrease in the content of monomer after lignin depolymerization [ 17 ]. This justified the decision adopted in this study of splitting hydrogenolysis in basic media into two separate steps for technical lignins: BCD in EtOH/H 2 O and Ru/C catalyzed hydrogenolysis. The different nature of the lignins is in the origin of the higher aromatic monomer yields for OrgB and SeolB (7–10%) compared to SeolA (1.1%) upon treatment with NaOH in EtOH/H 2 O. Further treatment with Ru/C, however, caused a noticeable increase in phenolic monomers yields in SeolA 493 andA 523 than in SeolB 523 and OrgB 523 . Thus, as a thumbnail rule, monomer yields were lower in pine-derived lignins than in poplar derived lignins. Softwood lignin has a more condensed structure than hardwood lignin, because G subunits have one less methoxy group (in comparison with S subunits) which facilitates the formation of C–C bonds (such as 5-5 and β-5) and hinders depolymerization. In addition, as already evidenced in our previous work [ 6 ], DOSY NMR, has proved to be a very valuable tool to estimate the actual apparent mass of the poly-(hydroxy)-aromatic fraction of lignins and their structural composition. Reaction mixtures after lignin depolymerization consist of a mixture of monomeric phenols and poly-(hydroxy)-aromatic ethers. The application of this latter fraction of poly-(hydroxy)-aromatic ethers is strongly dependent on their average or number molecular weight, M w or M n respectively. However, the estimation of the apparent masses using the diffusion coefficient, D, is not trivial because D is strongly dependent, among other factors, on the viscosity of the medium and the solvent-diffusate interactions. Interestingly, it could be observed that, in the estimation of poly-(hydroxy)-aromatic ethers with relatively high apparent masses, correlation with the values estimated by SEC were better using PS curves whereas in the case of low apparent masses this correlation was better using PEG curve. The reason lies in the different diffusate-diffusate and diffusate-solvent interactions that arise from the increasing hydrophobicity of the highest mass fractions that arise Biomolecules 2020,10, 1338 25 of 30 from and increasing importance of dispersion forces as the mass increases. Although some small discrepancies can be found between the apparent masses for the aromatic region, which corresponds to the apparent mass of poly-(hydroxy)-aromatic fraction, and Mn, values are in good agreement. Finally, semi-q-DOSY combined with 31 P has been used for estimating the percent of aromatic carbon atoms in the reaction mixtures. Although this is not a quantitative measure, this has allowed for approximately determining the content of poly-(hydroxy)-aromatics in the reaction mixtures and evaluating the extent of ethanol side reactions (viz. transfer hydrogenation or the Guerbet reaction). In this case, this content decreases in the reactions that were carried out in ethanol and, amongst them, in those conducted with the harshest conditions because of the side reactions of the solvent. 5. Conclusions BioA and B downstream biorefinery lignins were obtained from P. radiata and Populus Sp., respectively. SeolA and B were obtained after autohydrolysis and soda ethanosolv of P. pinaster and waste poplar sawdust, while OrgB was obtained from the latter via organosolv in isopropanol. It has been shown that there is not a single solution to lignin depolymerization. Therefore, catalyst and reaction conditions must be carefully selected depending on the feedstock and the lignin isolation method. BioA and B could only be depolymerized by NaOH BCD in aqueous medium. However, the best aromatic monomer yields in the depolymerization of technical lignins were achieved by BCD in ethanol/H 2 O followed by Ru/C catalyzed hydrogenolysis in ethanol. The catalyst plays an essential role in the selectivity. Ni-, Cuand bimetallic NiCu-catalysts supported on calcined hydrotalcite produced slightly lower aromatic monomer yields in the heterogeneous BCD of OrgB in scEtOH, although, the synergistic effect of Ni and Cu caused a change in the selectivity yielding monomers with alkyl side chains bearing carbonyl groups. The solvent and the reaction temperature are also key in the outcome of the depolymerization reaction. On one hand, the resulting poly-(hydroxy)-aromatic ether fractions presented similar apparent masses and SEC profiles when the depolymerization reactions were carried in a given solvent at a given temperature regardless of the catalyst used, suggesting that thermo-solvolysis is an essential part in lignin depolymerization. On the other hand, when the reaction was carried out in ethanol at high temperatures, a series of compounds derived from the hydrogen transfer mechanism in Ru/C catalyzed reactions and the Guerbet reaction in HTC-M catalyzed reaction were detected. DOSY-NMR has proven to be a useful and reliable tool in selectively determining the apparent mass of the different fractions in the reaction mixtures. Semi-q-DOSY combined with 31 P has been used for the first time in the estimation of the purity, not only of lignins, but also of the poly-(hydroxy)-aromatic fractions arising from lignin depolymerization and it has been essential to estimate the extent of the solvent side reactions. Supplementary Materials: The following are available online at http://www.mdpi.com/2218-273X/10/9/1338/s1, SI. 1 XRD analyses; Figure S1. Representative XRD diffraction patterns of the fresh and calcined NiCu-HTC samples; SI. 2 Chemical characterization lignin and reaction mixtures analysis; Table S1. Aromatic monomers and retention times; SI. 3. Experimental details for NMR characterization; Table S2 Integration regions used in the determination of averaged diffusion coefficients. Table S3. Integration regions used in the determination of aromatic –OH contents and G/S ratio; SI. 4. SEC and DOSY spectra; Figure S2. Normalized SEC chromatograms for isolated lignins. BioB, SeolB, OrgB; BioA and SeolA; Figure S3. DOSY spectra for BioB and BioA; Figure S4 . Raw DOSY spectra for SeolA, SeolB, and OrgB; Figure S5 DOSY spectra for BioB 473 in the aliphatic region; Figure S6 . Aliphatic region for the DOSY spectra of OrgB, OrgB 463 and OrgB HTC2.5 ; Figure S7. DOSY in the aliphatic region for SeolA, SeolA 493 and SeolA 493-Ru . DOSY in the aromatic region for SeolA, SeolA 523 and SeolA 523-Ru ; Figure S8. DOSY spectra and SEC for OrgB obtained using ethanol and isopropyl alcohol; Figure S9. SEC chromatograms for the depolymerization of OrgB with NaOH in ethanol at 463 K and HTC-2.5 in ethanol at 543 K; SI. 5. HSQC and HSQC-TOCSY spectra; HSQC spectra for SeolA; Figure S11. HSQC spectra for OrgB; Figure S12. HSQC spectra for SeolB; Figure S13. HSQC spectra for acetylated BioB, BioA and HSQC-TOCSY for BioA; Figure S14. Aliphatic region in the HSQC spectra for SeolA523 and SeolA523-Ru; Figure S15. HSQC spectra for in the aliphatic region for OrgBHTC2.5.