Understanding biomass fractionation in subcritical & supercritical water

1 Understanding biomass fractionation in subcritical & supercritical 1 water 2 María José Cocero*, Álvaro Cabeza, Nerea Abad, Tijana Adamovic, Luis Vaquerizo, Celia M. 3 Martínez, María Victoria Pazo-Cepeda. 4 High Pressure Processes Group, Department of Chemical Engineering and Environmental 5 Technology, University of Valladolid (Spain). Doctor Mergelina s/n. 47011, Valladolid, Spain. 6 * Corresponding author. Tel: +34 983423174; fax: +34 983423013. 7 E-mail addresses: mjcoc[email protected] 8 Abstract 9 Biomass fractionation into its individual building blocks poses a major challenge 10 to the biorefinery concept. The recalcitrance of the lignocellulose matrix and the 11 high crystallinity of cellulose make typical feed stocks difficult to separate into 12 their components. Hydrothermal processing fractionates biomass by its 13 hydrolysis. However, a deep knowledge of hydrolysis principles is required since 14 an inappropriate selection of the operating parameters such as an excessive 15 temperature and a long residence times causes dramatic selectivity losses. This 16 review is divided in four main sections which present the fundamentals of 17 lignocellulosic biomass fractionation in hemicelluloses, cellulose and lignin. As 18 the biomass structure plays an important role, a section to study the extraction of 19 the linked phenols that joint lignin and hemicelluloses is included. 20 1. Introduction 21 Shifting the chemical industry away from petrochemical feedstocks towards 22 renewable, bio-based chemicals and materials is a long-term strategy of the 23 European Union. This “biorefinery” concept, despite being proposed as early as 24 the late 1980s, has still not come to fruition because the cost and complexity of 25 processing biomass to generate practical, usable, saleable feedstocks makes it 26 unfeasible. 27 Lignocellulose is the most abundant, cheapest and easiest grown form of 28 biomass, and it is composed of three main fractions: cellulose (40-50%), 29 2 hemicellulose (25-35%) and lignin (10-30%), in addition to minor compounds. 30 These fractions represent potential feedstocks for bio-sourced commodity 31 chemicals, but due do their differing chemical functionalities (lignin made up of 32 linked aromatic units, hemicellulose of C5 sugars and cellulose of C6 sugars) 33 separation steps are necessary to isolate the appropriate fraction and break it 34 into its individual building blocks (e.g. sugars for cellulose/hemicellulose and 35 aromatic units for lignin). 36 37 38 39 Figure 1: Lignocellulosic biomass structure 40 This fractionation of biomass into its individual building blocks poses a major 41 challenge to the biorefinery concept, because the recalcitrance of the 42 lignocellulose matrix and high crystallinity of cellulose makes typical feedstocks 43 difficult to separate into their components. For this reason, it typically requires 44 long reaction times (from 30 minutes for the hydrothermal hydrolysis to 24-70 45 hours for enzymatic hydrolysis) and the presence of strong reagents (sodium 46 hydroxide and sodium sulfide during Kraft pulping, for example). This leads to 47 degradation of the non-cellulosic fractions as well as large volumes of effluent 48 which requires expensive treatment to reduce environmental load. 49 To truly harness the potential of the biorefinery concept, this fractionation step 50 needs to be revolutionized. It needs to be considerably more process intensive 51 (ideally seconds per unit volume of biomass -as opposed to minutes or hours) to 52 enable modular units to deal with large volumes of biomass at decentralized 53 3 locations. It must not involve the use of harsh reagents in order to minimize 54 environmental impact and cost, whilst maintaining quality of the fractions. 55 Water above its critical point (Tc 374ºC, 22 MPa), is an alternative solvent for 56 dissolution/hydrolysis of biomass. Its low viscosity and high diffusivity facilitate 57 the penetration of water into the complex structure of the lignocellulosic matrix, 58 whilst its low dielectric constant, similar to non-polar organic solvents, enhances 59 solubility of organic compounds. Physical properties of water (such as density, 60 ionic product, dielectric constant) can be finely tuned by varying temperature and 61 pressure. At these conditions, the hydrolysis of biomass fractions is rapid and 62 presents a mean to achieve significantly more process intensive fractionation of 63 biomass. 64 Reaction speed – whilst being an advantage to process intensification – is also a 65 significant disadvantage to selectivity at longer reaction times, leading to 66 degradation of hydrolysis products and resulting in complex reaction mixtures. 67 This degradation and mixture complexity leads to inefficient recovery of biomass 68 derived products and intermediates. There is therefore a need for understanding 69 the hydrothermal fractionation processes to improve processes selectivity, which 70 can harness the potential of subcritical and supercritical water fractionation. 71 Even under water’s critical point, certain fractions of biomass face reactions that 72 proceed too rapidly to be controlled by conventional methods. For instance, lignin 73 undergoes rapid hydrolysis and subsequent hydrolysis product conversion in less 74 than 1 second at 350 ºC. Whilst poor selectivity is common to both suband 75 supercritical water (SCW), there are some significant differences between the 76 reaction media – most notably the difference in ionic product of water (for 77 example the H+/OHconcentration at 300ºC and 22 MPa is around 3·10-6 mol· L78 1 vs 3·10-10 at 400 and 22 MPa) which means that subcritical water has a higher 79 concentration of ions ([H+] and [OH-]) thus favoring ionic reactions vs the radical 80 reactions that are prevalent under SCW conditions. 81 82 4 83 Figure 2. Subcritical and supercritical water properties around the critical point 84 (22 MPa). 85 This manuscript studies the lignocellulosic biomass fundamentals fractionation in 86 subcritical and supercritical water, in order to improve the selectivity of the 87 hydrothermal biomass fractionation. The manuscript presents four main sections 88 to presents the fractionation of biomass in hemicellulose, sugars and lignin. As 89 the biomass structure plays an important role, a section to study the linked 90 phenols that joint lignin and hemicelluloses is included. 91 2. Hemicellulose(s) fractionation fundamentals 92 Hemicellulose is a biopolymer present in lignocellulosic materials that acts as a 93 connection between the fibbers (cellulose) and the 3-dimensional structure 94 (lignin), constituting between 25% and 35% of the whole biomass [1]. It is 95 characterized by their amorphous structure and by the fact that it is acetylated 96 [2]. Regarding its composition, it is a biopolymer mainly composed of pentoses 97 with few hexoses in between, with a maximum length around 200 or 300 98 monomeric sugars, which makes it a renewable source of chemicals based on 599 carbon molecules. The maximum molecular weight is lower than 70 kDa in most 100 cases [3]. However, there are discrepancies between species. For instance, 101 xylose is the most common monomer in hemicelluloses of hardwood trees, while 102 softwood trees are principally composed of mannans, like mannose [2]. In 103 addition, there are two different types of hemicellulose from the extraction 104 Temperature, ºC 0 100 200 300 400 500 600 700 Density, kG m-3 0 200 400 600 800 1000 1200 Ionic product of water, pKw 10 12 14 16 18 20 22 24 Dielectric constant, e 0 20 40 60 80 100 Density Kw e 5 viewpoint: one hemicellulose easy to extract and another one that is associated 105 with the fibbers of cellulose that can be recovered only when cellulose is also 106 removed (temperatures above 240 °C) [4–6]. Since hemicelluloses have some 107 potentially acidic groups (acetyl groups among others), it can be recovered by 108 Kraft pulping. However, the use of this technique leads to a degradation of 109 hemicellulose, so a different technique is required to obtain it with a high quality. 110 Thus, hydrothermal extraction would be one of the most promising options since 111 it only requires water and mild temperatures (160-210 °C) to extract it [7,8]. If the 112 operational temperature is around 180 °C, 60% of the initial hemicellulose can be 113 recovered as oligomers and sugars [4,9]. Higher yields can be obtained if 114 temperature increases but undesired degradation products appear [10,11]. 115 However, hemicellulose can be recovered also at low temperatures (90 ºC) if the 116 operating time is high enough (days) [12]. Hemicellulose extraction has been 117 performed in both systems, batch and packed bed reactors. Therefore, it should 118 be also marked that two different operating times can be defined, the solid and 119 liquid time. The former is the time used to treat the solid. The liquid time has the 120 same value as the solid time in batch systems. However, it is fixed by the 121 volumetric flow when semi-batch or continuous system are used, being the 122 residence time (see appendix 1 for more details about the different residence 123 times). Moreover, hemicellulose hydrothermal fractionation is a complex process 124 that involves several physical phenomena [6,8,13,14], which are present as in 125 batch as in continuous systems, and a good knowledge of them is mandatory for 126 designing a profitable and sustainable hemicellulose extraction plant. These 127 phenomena are: 128  Hemicellulose cleaving into decreasing molecular weight oligomers 129  Hemicellulose deacetylation (autohydrolysis) 130  Hemicelluloses dissolution and mass transfer between the solid and the 131 liquid 132  Production of sugars & sugars degradation into furfural or other 133 substances 134  Porosity changes: extraction, swelling and biomass compaction 135 6 Additionally, and once the phenomenology is explained, a short summary about 136 the effect of the main operational variables on hemicellulose selectivity is 137 included. 138 139 2.1. Hemicellulose cleaving into decreasing molecular weight 140 oligomers 141 The following discussion is focused on the behavior observed in a semi142 continuous system since only globalized values can be obtained from a batch 143 reactor. Hemicellulose cleaving is one of the first phenomena that takes place 144 inside a chip or a particle of biomass. Due to the mild operating temperature (e.g. 145 120 to 185 ºC), the bonds between the monomeric sugars start breaking 146 randomly, producing progressively shorter oligomers. This process continues 147 until the moment in which the oligomer has a length low enough to be extracted 148 from the solid by solubilization or dragging [6,14,15]. In this moment, both 149 phenomena are present, oligomer dissolution and oligomer cleaving, and two 150 distinct stages can be differentiated: (1) solid oligomer cleaving, which is present 151 from the beginning, and (2) solid & liquid oligomer cleaving with hemicellulose 152 dissolution. The fact that these two phases are present at the same time explain 153 why there is a delay in the extraction profiles (Figure 3.a). Before this first soluble 154 oligomer releasing, only raw material free sugars and a little number of cleaving 155 products (small oligomers and monomers) could be removed. Nevertheless, 156 there are cases where no delay is present due to the biomass diversity [16]. This 157 is possible when the initial hemicellulose length is so low that it is initially soluble 158 or it is so acetylated that only stage 2 is present. Therefore, if both stages are 159 present, the molecular weight evolution during the extraction should have a 160 maximum (the first soluble oligomer) near the time (tm) when the concentration in 161 the liquid reaches the highest value (Figure 3.b). After this molecular weight peak, 162 it would continuously decrease due to the cleaving. This Behavior was already 163 observed in literature [15,16]. However, when only stage 2 is present the 164 molecular weight would decrease with time. 165 166 7 167 168 169 170 171 172 Figure 3: Liquid profiles at the output of a packed bed reactor during a 173 hydrothermal extraction process: (a) TOC evolution, (b) molecular weight 174 evolution (Mw) when both stages are presents and molecular weight evolution 175 when only stage 2 is present. 176 MW TOC t,min TOC MW Mw t, min Both stages Stage 2 a) Stage 1 Stage 1 & Stage 2 tm b) Delay 8 177 To sum up, temperature, the molecular weight and de acetyl contents plays an 178 essential role in hemicellulose extraction since they directly affect hemicellulose 179 solubility. 180 2.2. Sugar production from the cleaving processes 181 As it was explained in the previous section, the cleaving can also produce 182 monomeric sugars and, if temperature is high enough, all the hemicellulose could 183 be converted into monomeric sugars. However, the operational conditions 184 required to achieve a complete conversion are so high that they also imply sugar 185 degradations. Gallina et al. [4] studied the optimal conditions for the hydrothermal 186 fractionation of eucalyptus in a semi-continuous reactor. They found that the 187 optimum monomeric sugar yield was at 185 ºC (67.41%), starting to decrease at 188 higher temperatures. Yedro et al. [9] assessed the hemicellulose extraction from 189 holm oak in a batch system, obtaining that the highest monomeric yield was at 190 170 ºC (60%) and that degradation started at temperatures as low as 150 ºC. 191 Rissanen et al. [10] analyzed the hydrothermal degradation of spruce in the same 192 reactor as Yedro et al. [9], reaching a similar optimum. Sukhbaatar et al. [11] 193 worked with sugarcane bagase also in a batch system, being their monomeric 194 yield optimum at 180 ºC and observing a huge degradation above 190 ºC. The 195 same biomass was considered by Vallejos et al. [17] who reached the best 196 monomeric yield at 180 ºC too (70%). Similar result were reported by Thomsen 197 et al.[18], dos Santos Rocha et al.[19] and Makishima et al. [20] for wheat straw, 198 sugarcane straw and corncob, respectively. 199 Focusing on direct sugar production is of interest since they can be used to 200 produce fuels (bioethanol) or chemicals (like xylitol via hydrogenation). These so201 called “degradation products” can also be the target [8]. For instance, furfural and 202 its derivatives can be used as fungicides or lubricants [21] while lactic acid is a 203 precursor for biodegradable polymers production [22]. Therefore, to 204 avoid/promote sugar degradation the operating temperature and the volumetric 205 flow (the less time in the reactor, the lower degradation [4,7,18,20] are the main 206 9 involved variables. It is worth highlighting that when the reactor is a batch system, 207 the liquid/solid ratio has the same role as residence time. 208 2.3. Hemicellulose deacetylation 209 Hemicellulose deacetylation and cleaving take place in parallel, which is reflected 210 in a releasing of acetic acid during the hemicellulose extraction, decreasing the 211 pH of the water. It should be remarked that this acetic acid production only 212 happens in the solid phase [6,14,23–25]. However, acetic acid can be obtained 213 from sugar degradation in liquid phase too [8]. Similarly, uronic acid can be also 214 released during the hydrothermal treatment [26,27]. Nevertheless, it is not 215 completely clear if the pH change accelerates extraction or if this change is only 216 a consequence of the extraction [10,12]. This phenomenon is deeply related with 217 the hemicellulose extraction process selectivity since these acids are a source of 218 protons that catalyze the cleaving and degradation reactions in liquid phase if the 219 residence time is high enough [6,14]. A statement that was verified by Song et al. 220 [28], showing that degradation is much lower if the pH is maintained above 4-5. 221 Moreover, the releasing of acetyl groups also means that the solubility of the 222 remained part of the hemicellulose would be lower since the capacity of linking 223 by hydrogen bonds with water would be lower. Additionally, the steric hindrance 224 also would be higher. Following this idea, it can be seen in Figure 4. that the 225 minimum of the pH corresponds to the maximum in the TOC profile. Thus, the 226 oligomers involved in the stage (2) defined in Figure 3 will be more soluble since 227 they are smaller but, at the same time, their solubility also decreases due to the 228 lack of acetyl groups, explaining why the extraction is more difficult after the 229 maximum (decreasing slope). Moreover, extraction would also be slower 230 because the available amount of hemicellulose is much lower. Thereby, the 231 acetylation degree (and uronic content) is another variable to consider. 232 16 cellulose is a water insoluble polymer so that it is not possible to simply dissolve 375 and hydrolyze cellulose in water at ambient conditions. As a result, the hydrolysis 376 of cellulose in lignocellulosic biomass usually involves the use of strong acids as 377 catalysts [46], which cause a negative impact in the environment and yields a 378 high concentration of degradation products. However, when using supercritical 379 water cellulose is more effectively converted to oligomers and monomer sugars 380 instead of yielding mainly degradation products. 381 Therefore, the objective of this section is to clarify the mechanisms involved in 382 both the dissolution and the hydrolysis of cellulose in water as well as to discuss 383 the influence of the key parameters which affect both processes. 384 4.1. Cellulose dissolution 385 The dissolution of cellulose in water have been explained [48–50] from a 386 thermodynamic point of view. The Gibbs free energy is a thermodynamic 387 magnitude commonly considered to analyze whether a chemical process is 388 spontaneous or not. Its variation is expressed as a combination of the variation 389 of the enthalpy and the variation of the entropy of the system: 390 ∆𝐺 = ∆𝐻 − 𝑇∆𝑆 (1) 391 “G” is the Gibbs free energy, “H” the enthalpy, “T” the temperature and “S” the 392 entropy. When the variation of the Gibbs free energy is negative, the process is 393 spontaneous. In the reaction of two different compounds, the variation of enthalpy 394 represents the heat of reaction or the heat of mixing. In the combination of 395 cellulose and water this parameter is almost negligible since no additional heat is 396 generated or consumed. Therefore, the previous expression is reduced to: 397 398 ∆𝐺 = −𝑇∆𝑆 (2) 399 400 Consequently, the dissolution and hydrolysis of cellulose in water is carried out 401 (spontaneous process) when the entropy variation is positive. From a structural 402 point of view, the entropy of cellulose increases when its molecular conformation 403 changes from a rigid structure to a more flexible one which benefits dissolution. 404 Since cellulose structure is characterize by its complexity and rigidity, at lower 405 17 temperatures no conformational changes will be produced, the entropy will not 406 increase nor the Gibbs free energy will decrease and therefore no dissolution will 407 be produced. Only in the cases in which the temperature is considerably 408 increased and therefore the internal energy of the structure, conformational 409 changes could be produced. 410 In literature, three main characteristics of the cellulose structure are considered 411 of fundamental interest in its dissolution in water: 412 413 1) The presence of intra and intermolecular hydrogen bonds [48,49,51]. 414 Cellulose is constituted by glucose molecules joined together forming long 415 fibbers which are connected by hydrogen bonds. This fact results in a rigid 416 and cohesive structure which avoids the penetration of water molecules and 417 consequently the dissolution of the structure. 418 2) Cellulose is considered an amphiphilic molecule [48,49,52]. Its structure has 419 both hydrophobic and hydrophilic zones as a consequence of the orientation 420 of its functional groups. While the hydroxyl groups located in equatorial 421 position create the hydrophilic regions, the axial glycosidic bonds produce 422 hydrophobicity. This is considered the reason why the water molecules are 423 not able to easily create hydrogen bonds with the cellulose which will 424 produce its dissolution. 425 3) Crystallinity: crystallinity has always been considered a key parameter when 426 analyzing the dissolution of cellulose in water [48,53]. The high crystallinity 427 of the cellulose molecule is responsible of its rigid structure avoiding 428 conformational changes which could facilitate its dissolution. 429 Considering the lack of a robust model which explains the dissolution of cellulose 430 in water, several authors have performed experiments with the objective of 431 analyzing the influence of the process parameters. 432 From a structural point of view, [54,55] studied the influence of the raw cellulose 433 used in the dissolution. They demonstrated that the cellulose allomorph directly 434 affects the dissolution process. For example, although cellulose I is the most 435 abundant type in nature, cellulose II is more stable [49]. Moreover, not only the 436 cellulose type influences the dissolution, also the amount of water has to be 437 considered [48,56]. Regarding to the crystallinity of the structure, [57] analyzed 438 18 the dissolution of cellulose after milling. Milling produces an amorphous structure 439 which facilitates the action of water. They stated that the critical factor is not the 440 reduction of the particle diameter but the cleavage of the hydrogen bonds and 441 the consequent generation of amorphous zones. Amorphous and semi-crystalline 442 zones are easier to be hydrolyzed since water molecules can avoid the 443 hydrophobic zones which are present in the structure as a consequence of the 444 amphiphilic nature of the cellulose [58]. 445 From an operating point of view, the majority of experiments analyzed the 446 process focusing in the variation of the pressure, the temperature and the 447 reaction time. As it has been explained in this section, due to the physical 448 structure and the nature of cellulose, its dissolution is greatly limited by 449 temperature. Common working temperatures usually range from 200ºC to more 450 than 400ºC. Therefore, in order to maintain water in liquid or supercritical state 451 when the working conditions surpasses its critical point, (Tc=374ºC, Pc=22.1 452 MPa) the pressure shall be increased. The analysis of the influence of pressure 453 has been studied by [53]. They proved that when the pressure is increased above 454 50MPa (reaching pressures up to 700MPa), even at relative low temperatures 455 the water molecules are able to enter inside the cellulose structure and swell the 456 polymeric matrix which finally collapses. When the pressure is only considered in 457 order to maintain the water in liquid or supercritical state, its influence is negligible 458 and the fundamental parameters to be considered are the temperature and the 459 reaction time. An increase of temperature clearly benefits the dissolution of 460 cellulose since it modifies its structure favoring the combination of cellulose and 461 water molecules. However, it also accelerates its hydrolysis consuming the 462 cellulose which is being dissolved. Consequently, the only possibility of dissolving 463 cellulose and reduce its hydrolysis rate is selecting an optimum combination of 464 temperature and reaction time. In literature, the analysis of cellulose dissolution 465 at high temperatures is generally combined with hydrolysis studies. Hydrolysis is 466 considered one of the fundamental processes in green chemistry since it allows 467 obtaining high value products from renewable resources such as biomass. As 468 biomass is a complex raw material and due to the lack of enough know-how in 469 this field, the majority of authors have started working with cellulose instead of 470 with biomass. When cellulose is mixed with water at high temperatures, it is first 471 19 dissolved and subsequently it reacts with the water molecules present in the liquid 472 medium producing the cleavage (hydrolysis) of the glycosidic bonds. As the 473 cleavage of these bonds is not completely simultaneous nor instantaneous, first, 474 oligosaccharides are generated which are then hydrolyzed to monosaccharides. 475 Finally, if the hydrolysis reaction is not stopped, the monosaccharides are 476 degraded to organic compounds such as acids [59,60]. As it has been stated, it 477 is fundamental both in dissolution and in hydrolysis to find the optimum pair of 478 temperature and reaction time values in order to reduce the generation of 479 undesired products. 480 The experiments presented in literature are clearly divided in three zones: 481 subcritical region, vicinities of the critical point and supercritical region. 482 In this paper, the subcritical region is considered the one in which the temperature 483 remains below 320ºC. In this zone the dissolution and subsequent hydrolysis is 484 produced as a result of the consumption of superficial cellulose which is able to 485 interact with water molecules [61]. Furthermore, the cellulose which can be easily 486 dissolved is the one which was present in an amorphous state. Below 280ºC, it 487 is observed that the cellulose dissolution rate decreases with time since water is 488 not able to dissolve crystalline cellulose once the amorphous cellulose has been 489 already dissolved [55]. At temperatures between 280ºC and 320ºC increasing 490 either the reaction time or the temperature only increases the degradation of the 491 cellulose, mostly amorphous, which has been already dissolved [55,62]. 492 Therefore, working with low reaction times produces high DP (degree of 493 polymerization) molecules [63]. At temperatures below 250ºC [64] proved that 494 cellulose is dissolved but not hydrolyzed and therefore that it is possible to obtain 495 high DP molecules. However, the process is limited by the amount of amorphous 496 cellulose available. In these cases reaction times in the order of hours are 497 required which implies the operation in batch and semi-continuous reactors. 498 Finally, milling the raw cellulose creates amorphous zones which can be easily 499 dissolved, even at temperatures below 230ºC, generating high DP molecules 500 [57]. At this temperatures, no modifications are observed in the solid residue 501 when the cellulose structure is crystalline instead of amorphous [65]. 502 In the region near the critical point, when the reaction time is increased, the 503 dissolved cellulose is hydrolyzed to glucose and lately to degradation products. 504 20 It has been experimentally demonstrated [51] that at temperatures between 505 320ºC and 330ºC (25MPa) a transition from a crystalline to an amorphous 506 structure is produced. This transition explains the rapid dissolution of cellulose in 507 water and the absence of any swelling phenomena [51]. The fact that when 508 cellulose and water react at these or higher temperatures during a short reaction 509 time the final product obtained is cellulose II when the initial cellulose allomorph 510 is cellulose I is justified as a consequence of the higher stability of cellulose II. 511 When the temperature is increased above 330ºC cellulose I is converted into 512 amorphous cellulose. Then, when the temperature decreases, the amorphous 513 cellulose is converted into the more stable cellulose II allomorph [59,61]. This is 514 also confirmed working at temperatures below 320ºC since only cellulose I is 515 obtained [66]. 516 Finally, in the supercritical region the dissolution and hydrolysis of cellulose when 517 working at low concentrations is produced simultaneously, in homogeneous 518 phase and without mass transfer limitations [53,67]. The transition between 519 crystalline cellulose to amorphous cellulose at 330ºC, the high temperatures of 520 reaction which produce the cleavage of the hydrogen bonds [53,68] and the 521 properties of supercritical water such as high diffusivity, high density compared 522 with water in vapor state and its ability to dissolve organic compounds, observing 523 the total dissolution of cellulose [69,70], explain the homogeneity of the process. 524 Recently it has been proved that when the concentration of cellulose is increased 525 the dissolution and hydrolysis processes are not completely simultaneous nor 526 homogeneous [71]. 527 4.2. Cellulose hydrolysis 528 It is noted that in this reaction zone the hydrolysis of biomass has gained a lot of 529 attention [67,68]. In fact, special attention has been paid to the hydrolysis of 530 cellulose, since it is the major component of lignocellulosic biomass and therefore 531 is the key to better understand the reaction mechanisms, kinetics and 532 performance of supercritical water hydrolysis of real biomass [71,72]. 533 4.2.1. Production of sugars from cellulose hydrolysis in supercritical 534 water 535 21 The conversion of cellulose to sugars in supercritical water has been extensively 536 studied using different kinds of reactors. The hydrolysis in batch-type reactors is 537 usually carried out with long reaction times, favoring the decomposition of glucose 538 to degradation products [73,74]. However, the flow-type system makes it possible 539 to reduce the reaction time and therefore increasing the yields of sugars instead 540 of degradation products [61,70]. Recently our research group developed an 541 experimental set up to perform the hydrolysis of cellulose suspensions in 542 supercritical water by using a continuous micro-reactor, giving as a result a total 543 conversion of cellulose in milliseconds and yielding a sugars production of 96 % 544 w/w [67]. This continuous micro-reactor is shown in Figure S2, where it can be 545 seen that the reaction section consisted of a tee junction (M) where the cellulose 546 (or biomass) was instantaneously heated up by mixing it with a supercritical water 547 stream. In order to effectively stop the hydrolysis reaction, a sudden 548 depressurization through a needle valve was carried out, so that the effluent was 549 immediately cooled down from 400ºC to around 100ºC and therefore reaction 550 was over. Then, depending on the dimensions of the pipe between the junction 551 and the depressurization valve, the reaction time was calculated as a function of 552 reactor volume and flow to the reactor, so that just by changing the dimensions 553 of the reactor of the pumped flow, different reaction times would be provided. In 554 terms of sugars yield from cellulose hydrolysis in hydrothermal medium, several 555 conditions were tested by changing temperature, pressure and reaction time in 556 the micro-reactor mentioned above. As a result, it was found that the optimal 557 conditions to obtain soluble sugars (up to six units of glucose) were achieved at 558 400 ºC with extremely short reaction times (around 0.01 s). If the reaction time 559 was increased, the sugars were hydrolyzed and the yield decreased, as it can be 560 seen in Figure 8. The combination of supercritical water medium and the effective 561 method for the reaction time control allowed such a high sugars yield from 562 cellulose hydrolysis. This fact can be explained taking into account than under 563 those conditions, the cellulose hydrolysis kinetics are improved and the glucose 564 hydrolysis kinetics are slow enough so that using the sudden expansion micro565 reactor is possible to stop the reactions after complete cellulose hydrolysis but 566 before glucose degradation [67]. It was also proven that cellulose hydrolysis 567 reactions were highly influenced by temperature, meanwhile pressure did not 568 affected cellulose hydrolysis rate in the studied range [75,76]. 569 22 570 Figure 8. Sugars yield from cellulose hydrolysis in hydrothermal medium along 571 reaction time. Experiment temperature: red = 400ºC; yellow = 350 ºC; blue = 572 300ºC. Experiment pressure ( ♦ ) 27 MPa; ( ■ ) 25 / 23 MPa and ( ▲ ) 23 / 18 MPa 573 [76]. 574 575 4.2.2. Cellulose hydrolysis kinetics in supercritical water 576 Cellulose was hydrolyzed following the main hydrolysis reaction pathway in 577 supercritical water which is shown in Figure 9 [71], where it can be seen that 578 cellulose is firstly hydrolyzed into oligosaccharides and then into glucose. Once 579 the glucose has been produced, it can be isomerized to fructose and then 580 converted into dehydrated (5-HMF) or retro-aldol condensation products 581 (glycolaldehyde, pyruvaldehyde and/or glyceraldehyde). As mentioned above, 582 working at 400 ºC and very short reaction times, the reaction would be stopped 583 at glucose. However, if the reaction time is increased, retro-aldol condensation 584 products would be produced, yielding aldehydes as glycolaldehyde, 585 pyruvaldehyde and/or glyceraldehyde. Therefore, the control of reaction time was 586 the key factor to selectively hydrolyze cellulose in supercritical water. 587 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 0.01 0.1 1 10 100 Sugars Yield, % w·w-1 tr, s 400-27 400-25 400-23 350-27 350-23 350-18 300-27 300-23 300-18 23 Glucose O H H H H OH OH H OH OH OH OOH OH H H OH H OH H H OH Fructose O H H H H OH OH H OH O OH O H H H HOH H OH OH OH Cellobiose O H H H H OH OH H OH O OH OH H H HOH H OH O OH O H H H H OH H OH O OH O H H H HOH H OH OH OH Oligosaccharides O H H H H OH H OH O OH O H H H HOH H OH OH OH Cellulose O OH O 5-HMF O OH OH Glyceraldehyde OH O Glycolaldehyde CH3 O O Pyruvaldehyde 588 Figure 9. Reaction pathway for cellulose hydrolysis in supercritical water based 589 on [71]. 590 The properties of water may vary considerably when changing the conditions 591 form subcritical to supercritical, affecting to the products yielded from cellulose 592 hydrolysis [72]. Just by changing pressure and temperature, different reaction 593 mechanisms are favored. Water at 25MPa and temperatures below 300 ºC has 594 a density around 800 kg/m3 and an ionic product (as pK) between 11 and 14. 595 Under those conditions, water is highly dissociated and H+/OHions are highly 596 available in the reaction medium and therefore ionic reactions are favored [77,78]. 597 However, when temperature is increased up to 400 ºC at constant pressure, the 598 density considerably decreases (being around 150 kg/m3) and the ionic product 599 increases up to 21 [79]. This change in the ionic product affects the kinetics of 600 glucose and fructose degradation, avoiding the ionic degradation reactions 601 (which are the governing chemistry when using acid catalysts) and favoring the 602 radical reactions [72]. In fact, it was found that the concentration of H+/OHdue 603 to water dissociation was a determining factor in the selectivity of cellulose 604 hydrolysis in supercritical water [76]. So far, kinetic models for cellulose 605 hydrolysis only considered the concentration of cellulose and its derived products 606 24 into the equations, so that first order kinetics were selected to predict cellulose 607 hydrolysis in supercritical water. Following those traditional kinetic models 608 Cantero et al. [76] found an incongruity for the kinetic constants of fructose 609 dehydration to 5-HMF when carrying out the hydrolysis of cellulose in supercritical 610 water at temperatures between 300 – 400 ºC and 25 MPa. In Figure 10 it can be 611 seen the fitted kinetic constants of 5-HMF formation (khmf) versus the reciprocal 612 temperature, according to Arrhenius law. A break point can be clearly observed 613 in Figure 10a, corresponding to the surroundings of the critical point of water, 614 which represents a deviation from Arrhenius law. So that, the traditional models 615 where only cellulose concentration was taken into account in a first order kinetics 616 equation were only capable to predict the kinetic constants of fructose 617 dehydration to 5-HMF at subcritical conditions. That suggested that another 618 factor was not taken into account into the kinetic equation. To solve the problem, 619 the concentration of protons and hydroxide ions were added to the kinetic model, 620 turning it into a second order kinetic equation. As a consequence of that 621 transformation, the kinetic constants followed the Arrhenius law for the full 622 temperature spectra, meaning that the dehydration to 5-HMF under both 623 subcritical and supercritical conditions was lineally fitted as it can be observed in 624 Figure 10b. That would suggest that the selectivity of the process was strongly 625 affected by the protons and hydroxide ions concentration in the reaction medium, 626 so that improving the understanding of the reaction mechanisms of the hydrolysis 627 of cellulose in supercritical water. In that way, retroaldol condensation reactions 628 from glucose and fructose (to produce aldehydes) are not very demanding of ions 629 and therefore they are favored when water is highly associated (as it occurs at 630 supercritical state). On the other hand, isomerization glucose-fructose and 631 dehydration reactions are not favored since these reactions take place forming 632 transition states with OHand H+ and thus they are diminished when water is 633 highly dissociated [76]. 634 25 635 636 Figure 10. Kinetic constants Arrhenius fitting for fructose dehydration to 5-HMF 637 at 25 MPa and temperatures between 300 and 400ºC [76]. a) Kinetic evaluation 638 just considering cellulose and derived products concentration. b) Kinetic 639 evaluation also considering protons and hydroxide ions concentrations as 640 reagents. 641 4.2.3. Cellulose concentrations as a mass transfer limitation 642 Another factor recently revised concerning cellulose hydrolysis and dissolution in 643 supercritical water was the effect of cellulose concentration itself [71]. So far, 644 existing models describing the conversion rate of cellulose assumed that the 645 hydrolysis of cellulose particles takes place at their surface and therefore the 646 particle size was considered the key parameter for the conversion rate. That 647 shrinking-core model implied the use of a nonconventional kinetic equation 648 [66,72]. On the other hand, to take into account the reagent concentration, a first 649 order kinetic was assumed to describe the conversion rate of cellulose in 650 supercritical water. As it can be seen in Figure 11, experimental results of 651 cellulose hydrolysis in supercritical water at 400 ºC and 25 MPa and different 652 concentrations were fitted to the first order kinetic equation by plotting the 653 logarithm against the reaction time. In all cases, a linear dependence was found, 654 where the slope represented the kinetic constant, k. In Figure 11 it can be 655 observed that when increasing the cellulose inlet concentration the reaction rate 656 is slower, suggesting that mass transfer resistances must have an important 657 effect over cellulose hydrolysis kinetics. Also, combining those data with the ones 658 from a previous work [67] it was possible to calculate the so called mass transfer 659 limit for cellulose hydrolysis in hydrothermal media. Those calculations are 660 a b subcritical supercritical 32 hour the lowest yield was given when just water was used as solvent. The best 836 performing solvent was a water-phenol mixture, which achieved nearly total 837 suppression of char formation with 99% TS molecules [96]. 838 Fang at al. followed decomposition of organosolv lignin in water/phenol solution 839 in micro-reactor coupled with optical microscopies at temperatures up to 600˚C 840 and water densities up to 1165 kg/m3. The microreactor, diamond anvil cell (DAC) 841 allows for in-situ observations of samples in the fully-visible chamber via optical 842 microscopy. The DAC consisted of a hole and sealed by compression of two 843 opposing anvils made of diamond. The chamber was rapidly heated by two 844 electric microheaters by cutting power, which is convenient for the study of phase 845 behavior and chemical reactions. Experiments have been done at different water 846 densities, heating rate, maximum temperatures and lignin concentration. Three 847 different types of products were obtained: a non-dissolved black residue, a 848 precipitated residue and reddish oil. A homogenous phase was formed for the 849 phenol + lignin system where phenolic char precipitated as the main product. 850 Adding water to this system de-polymerization of lignin was promoted by 851 hydrolysis in a homogeneous phase and its re-polymerization was inhibited by 852 phenol. The homogenous phase was not found in the case of lignin + water 853 system. After initial dissolution at above 377 °C lignin underwent hydrolysis and 854 pyrolysis to phenolic, which are further changed to oil in the aqueous phase. At 855 higher temperatures, solid particles precipitated from the aqueous via 856 homogeneous re-polymerization of the phenolics and water soluble compounds 857 to form a phenolic char. At these same conditions, non-dissolved lignin underwent 858 heterogeneous pyrolysis and formed polyaromatic char. Higher water density 859 decrease lignin dissolution. Therefore, polyaromatic char, with a lighter molecular 860 weight was the main product along with a smaller fraction of phenolic char. It can 861 be conclude that for water and phenol mixtures, lignin can be completely 862 solubilized and undergoes homogeneous hydrolysis and pyrolysis that prevents 863 further re-polymerization [94]. 864 5.2.2. Water without co-solvent 865 Sasaki and Goto presented a work in which the chemical conversion of alkali 866 lignin in near and supercritical water at 350 ºC and 400 ºC and a pressure of 25867 33 40 MPa using a batch reactor without catalyst, having 5-240 minutes residence 868 time was studied. The products were separated into two fractions, methanol 869 soluble (MS) and methanol insoluble (MI). The main products observed in the MS 870 fraction were catechol, phenol, and o, m, pcresols, while MI product was defined 871 as a residual solid. It was proposed that catechol is formed via hydrolysis of 872 guaiacol which is the main compound in structure of lignin. In further hydrolysis, 873 phenol, m, p and o-cresol were obtained. Dependence of reaction time showed 874 that the yield of catechol rapidly increased with reaction time (till 30 min) and then 875 decreased, especially at 400 ˚C, while the yields of phenol, m, p and o-cresol 876 increased with reaction time. After 90 min the yields of m, p and o-cresols were 877 almost constant while the yield of phenol slightly increased. At 400 ˚C after 878 catechol was consumed, the majority of the reaction most likely terminated. The 879 decreasing of catechol was not followed by the increasing of phenol, m, p and o880 cresol significantly. Water density influence yields of products where the yield of 881 catechol was gradually decreased with increasing the water density at 350 ˚C 882 and dramatically decreased at 400 ˚C. The yields of phenol, m, p and o-cresol 883 increased gradually with increasing the water density at 350 ˚C and 400 ˚C. It 884 was suggested that an increase in water density enhanced the hydrolysis rate of 885 ether and carbon-carbon bonds of alkylphenol in lignin. According to this results 886 it was proposed reaction mechanism showed in scheme below (Figure 13) where 887 lignin was degraded into its derivate compounds by dealkylation and hydrolysis 888 reaction. Under SCW conditions hydrolysis takes place at ether and ester bonds 889 in lignin. Hydrolysis is accelerated by a high ion product of water. Dealkylation of 890 lignin gives catechol, which is then hydrolyzed into phenol. This reaction pathway 891 suggests that some useful chemical intermediates (MS fraction) might be 892 recovered in a rapid and selective manner by changing the temperature, reaction 893 time under near and supercritical water condition. At the same time, re894 polymerization of low molecular weight compounds occurs as seen by the 895 formation of char through condensation reaction [97]. 896 897 34 898 Figure 13. Proposed scheme for degradation of lignin under near and 899 supercritical condition 900 Lignin conversion was also investigated in the continuous system for short 901 residence time 0.5-10 s, pressure of 25 MPa and different temperatures under 902 supercritical conditions at 390˚C and 450˚C and subcritical condition at 300 ˚C 903 35 and 370 ˚C [98] [99]. Temperature plays an important factor in deciding the 904 dominant pathway because of the existence of the parallel ionic and radical based 905 pathways. Lignin products were divided in char, gas, TOC, phenolic and aromatic 906 hydrocarbons. Under hydrothermal condition lignin was rapidly converted into 907 lower molecular weight products for all temperatures which was followed with 908 high yields of TOC, phenolic compounds and aromatic hydrocarbons, while 909 decomposition was accelerated under supercritical condition [99]. Increasing 910 decomposition rate with temperature follows Arrhenius behavior of lignin 911 degradation what was already obtained by Zhang and Ramaswamy [100]. The 912 rapid depolymerization is cause by cleavage of ether bonds from abundant β-aryl 913 ether (β-O-4) linkages in softwood lignin [101] [102]. The low dissociation 914 enthalpies of the ether bond in the β-O-4 linkage initiated the reaction to form a 915 phenoxy radical and a secondary alkyl aromatic radical [103]. The Arrhenius 916 behavior shown by lignin decomposition under hydrothermal conditions even in 917 subcritical region further supported the conclusion that the initial decomposition 918 was a radical reaction. TOC yield decreased with temperature and the yield was 919 much higher under subcritical condition. The TOC yield in subcritical water 920 increased within short residence time and remained stable or decrease slowly 921 despite longer residence time. The increasing in the polymerization during the 922 increase of temperature should be reflected in the TOC yield. Low TOC yield 923 under supercritical condition suggested the occurrence of secondary reaction. 924 This could be due to the cross-linking between reactive degradation fragments 925 obtained from the lignin depolymerization to produce fragments with higher 926 molecular weights. Increase in the lower molecular weight compounds during the 927 time resulted with the simultaneous formation of the higher molecular weight 928 compounds because of the repolymerization. The minimal decrease in TOC yield 929 for subcritical temperatures implied that the crosslinking reactions between these 930 lower-molecular weight compounds did not take place actively under these 931 conditions. This suggested the significance of radical’s involvement in enhancing 932 the reaction [98][99]. 933 Char has significantly higher yields in the supercritical region and formation was 934 enhanced at elevated temperatures. Formation of char from lignin follows 935 Arrhenius behavior and it is not affected with change in water properties under 936 36 subcritical condition what point radical reaction. In order to examine the 937 suggested hypothesis of formation of low molecular-weight fragments and 938 formation of higher molecular weight fragments by cross linking of the smaller 939 fragment it was determine the yields of the phenolic compounds and aromatic 940 hydrocarbons. The main phenolic compound from lignin decomposition is 941 guaiacol, which is followed by minor composition of other phenolic compounds 942 such as o, m, p-cresol, catechol and phenol. Formation of guaiacol was higher in 943 supercritical temperature, but rapidly decreased at longer residence time [98] 944 [99]. Guaiacol is an intermediate degradation product and highly reactive, since 945 the methyl C−O bond is the weakest in the guaiacol unit and is susceptible to 946 undergo cleaving. The aliphatic C−O bond of the methoxyl group is more likely to 947 react because the bond energy of the aliphatic C−O bond (245 kJ/mol) is smaller 948 than that of the aromatic C−O bond (256 kJ/mol). This was concluded in the study 949 of Wahyudiono et al. where also was found that guaiacol showed a fast 950 decomposition rate and the formation of high-molecular-weight substances 951 reformed to char was important for the guaiacol decomposition to reach 952 equilibrium [104]. However, high yield of guaiacol was also obtained under 953 subcritical conditions. The high yield of guaiacol under two separate regions of 954 temperature (subcritical and supercritical) with different water properties 955 indicated guaiacol formation via two different pathways. The formation of guaiacol 956 from lignin probably proceeded through hydrolysis under subcritical conditions 957 because of the high ionic product and dielectric constant of water. On the 958 contrary, under supercritical condition and high temperature free radical reaction 959 should be enhanced that lead to the formation of guaiacol from lignin [98]. In both 960 studies it was showed that the decomposition of lignin occurred rapidly with 961 residence time below one second, which indicate that kinetic study should be 962 done for residence time below 1s. 963 5.2.3. Water and base catalyst 964 In order to enhance the obtaining of monomeric phenols, basic compounds such 965 as hydroxides are used as catalyst [105][106] [107]. Studies on lignin model 966 compound dihydro-diisoeugenol, showed that the basic agent caused ether and 967 C–C bond cleavage which yielded volatile phenols [108]. Furthermore, the 968 37 analysis of products from model compound reactions revealed that phenyl ether 969 linkages were effectively broken in the base catalyzed hydrolysis reaction while 970 C-C linkages were less affected [109]. In another study, it was concluded that in 971 alkaline depolymerisation of lignin, ether bonds are hydrolyzed at random, most 972 likely from the outside of the oligomer and not in the sequence of their bond 973 strengths, forming first large units and then smaller subunits [106]. In addition, it 974 was stated that the formation of monomers is directly proportional to the 975 concentration of sodium hydroxide in the aqueous medium. Furthermore, a 976 mechanism for the NaOH catalyzed breakdown of the ether bonds of lignin is 977 proposed explaining the preferential formation of syringol derivatives, based on 978 the stabilizing effect that the methoxyl groups provides to the transition states of 979 the carbenium ions. It was also concluded that the production of monomers is 980 limited by the oligomerization and polymerization reactions of the products 981 formed. 982 Miller et al. showed that in the alkali depolymerization of lignin using water as 983 solvent the most important factor in lignin depolymerization was base 984 concentration. Moreover, it was observed that concentration excess of a strong 985 base gave better results on lignin depolymerization. In addition, a little amount of 986 a strong base (NaOH) together with a larger amount of less expensive base 987 (Ca(OH)2) produced positive results [105]. 988 Silva et al. studied the catalytic depolymerization of organosolv lignin with both 989 NaOH catalyst and with boric acid as a capping agent, aiming to produce oils of 990 monomeric and dimeric products. In the case of reactions with NaOH and no 991 capping agent, the highest oil yield was obtained at 300 ºC with a residence time 992 of 4 minutes. This gave a yield of 23% oil and no char formation. Lignin 993 conversion increased steadily with increasing temperature but char was formed 994 as well as oil. In order to increase oil yields, boric acid was used as a capping 995 agent. Without base, the boric acid increased the yield of oils to a maximum of 996 36% after 40 minutes at 300 ºC, but at longer residence times or higher 997 temperatures the yield decreased again. The results showed that the molecular 998 weights of the oils from the boric acid catalyzed reactions were around 500 Da, 999 compared to 300 Da for the base catalyzed depolymerization [88]. 1000 38 In contrast to a basic environment, leading to deprotonation of phenolic hydroxyl 1001 groups and decreased hydrogen bonding, the acidic environment enhances the 1002 degree of internal hydrogen bonding. As result, the probability of acid-catalyzed 1003 cleavage of ether bonds is reduced compared to base-catalyzed cleavage. Thus, 1004 in acid-catalyzed hydrolysis the primary products produced are larger (dimers to 1005 tetramers) than in the base-catalyzed route. For both cases, the primary products 1006 undergo easy addition and condensation reactions leading to higher molecular 1007 weight products [110]. 1008 Under supercritical and subcritical condition lignin is hydrolyzed and different 1009 phenolic and other aromatic compounds could be obtained. These hydrolysis 1010 reactions occur in the shortest time than the residence time that has already been 1011 used in literature (more than one second). It is very important to have better 1012 understanding of reaction pathways, intermediate reaction products and reaction 1013 products for first milliseconds of reaction time, thus the specific weaknesses and 1014 strengths of the polymer and its intermediates – i.e. the substructures which are 1015 the most susceptible to chemical attack. Kinetic models that have been obtained 1016 until today are justified with the final reaction products, without information about 1017 intermediate produced. 1018 Considerable effort is still required to address the separation challenges 1019 associated with lignin depolymerization. The supercritical water ultrafast 1020 hydrolysis could open a new way to improve the understanding of lignin 1021 depolymerization, as has been done in the cellulose hydrolysis. 1022 1023 Acknowledgements 1024 The authors thank MINECO and FEDER program for the financial support 1025 Projects CTQ2013-44143-R and CTQ2016-79777-R. 1026 1027 1028 1029 1030 1031 1032 39 1033 References: 1034 [1] P. Bajpai, Structure of Lignocellulosic Biomass, in: Pretreat. Lignocellul. 1035 Biomass Biofuel Prod., Springer Singapore, Singapore, 2016: pp. 7–12. 1036 doi:10.1007/978-981-10-0687-6_2. 1037 [2] O. Bobleter, Hydrothermal degradation of polymers derived from plants, 1038 Prog. Polym. Sci. 19 (1994) 797–841. 1039 http://cat.inist.fr/?aModele=afficheN%7B&%7Dcpsidt=4247470. 1040 [3] J. V Rissanen, H. Grénman, C. Xu, S. Willför, D.Y. Murzin, T. Salmi, 1041 Obtaining Spruce Hemicelluloses of Desired Molar Mass by using 1042 Pressurized Hot Water Extraction, ChemSusChem. 7 (2014) 2947–2953. 1043 doi:10.1002/cssc.201402282. 1044 [4] G. Gallina, Á. Cabeza, P. Biasi, J. García-Serna, Optimal conditions for 1045 hemicelluloses extraction from Eucalyptus globulus wood: hydrothermal 1046 treatment in a semi-continuous reactor, Fuel Process. Technol. 148 1047 (2016) 350–360. doi:http://doi.org/10.1016/j.fuproc.2016.03.018. 1048 [5] C. Wyman, S. Decker, M. Himmel, J. Brady, C. Skopec, L. Viikari, 1049 Hydrolysis of Cellulose and Hemicellulose, in: Polysaccharides, CRC 1050 Press, 2004. doi:doi:10.1201/9781420030822.ch43. 1051 [6] A. Cabeza, C.M. Piqueras, F. Sobrón, J. García-Serna, Modeling of 1052 biomass fractionation in a lab-scale biorefinery: Solubilization of 1053 hemicellulose and cellulose from holm oak wood using subcritical water, 1054 Bioresour. Technol. 200 (2016) 90–102. 1055 doi:http://doi.org/10.1016/j.biortech.2015.09.063. 1056 [7] F. Carvalheiro, L.C. Duarte, F. Gírio, P. Moniz, Chapter 14 - 1057 Hydrothermal/Liquid Hot Water Pretreatment (Autohydrolysis): 1058 A Multipurpose Process for Biomass Upgrading A2 - Mussatto, Solange I, 1059 in: Biomass Fractionation Technol. a Lignocellul. Feed. Based Biorefinery, 1060 Elsevier, Amsterdam, 2016: pp. 315–347. doi:http://doi.org/10.1016/B9781061 0-12-802323-5.00014-1. 1062 40 [8] C.M. Piqueras, Á. Cabeza, G. Gallina, D.A. Cantero, J. García-Serna, 1063 M.J. Cocero, Online integrated fractionation-hydrolysis of lignocellulosic 1064 biomass using suband supercritical water, Chem. Eng. J. 308 (2017) 1065 110–125. doi:http://doi.org/10.1016/j.cej.2016.09.007. 1066 [9] F.M. Yedro, H. Grénman, J. V Rissanen, T. Salmi, J. García-Serna, M.J. 1067 Cocero, Chemical composition and extraction kinetics of Holm oak 1068 (Quercus ilex) hemicelluloses using subcritical water, J. Supercrit. Fluids. 1069 (n.d.). doi:http://doi.org/10.1016/j.supflu.2017.01.016. 1070 [10] J. V Rissanen, H. Grénman, S. Willför, D.Y. Murzin, T. Salmi, Spruce 1071 Hemicellulose for Chemicals Using Aqueous Extraction: Kinetics, Mass 1072 Transfer, and Modeling, Ind. Eng. Chem. Res. 53 (2014) 6341–6350. 1073 doi:10.1021/ie500234t. 1074 [11] B. Sukhbaatar, E.B. Hassan, M. Kim, P. Steele, L. Ingram, Optimization of 1075 hot-compressed water pretreatment of bagasse and characterization of 1076 extracted hemicelluloses, Carbohydr. Polym. 101 (2014) 196–202. 1077 doi:http://doi.org/10.1016/j.carbpol.2013.09.027. 1078 [12] J. V Rissanen, D.Y. Murzin, T. Salmi, H. Grénman, Aqueous extraction of 1079 hemicelluloses from spruce – From hot to warm, Bioresour. Technol. 199 1080 (2016) 279–282. doi:http://doi.org/10.1016/j.biortech.2015.08.116. 1081 [13] W. Reynolds, H. Singer, S. Schug, I. Smirnova, Hydrothermal flow1082 through treatment of wheat-straw: Detailed characterization of fixed-bed 1083 properties and axial dispersion, Chem. Eng. J. 281 (2015) 696–703. 1084 doi:http://doi.org/10.1016/j.cej.2015.06.117. 1085 [14] A. Cabeza, F. Sobrón, F.M. Yedro, J. García-Serna, Two-phase 1086 modelling and simulation of the hydrothermal fractionation of holm oak in 1087 a packed bed reactor with hot pressurized water, Chem. Eng. Sci. 138 1088 (2015) 59–70. doi:http://doi.org/10.1016/j.ces.2015.07.024. 1089 [15] X. Chen, M. Lawoko, A. van Heiningen, Kinetics and mechanism of 1090 autohydrolysis of hardwoods, Bioresour. Technol. 101 (2010) 7812–7819. 1091 doi:http://doi.org/10.1016/j.biortech.2010.05.006. 1092 [16] X.J. Ma, X.F. Yang, X. Zheng, L. Lin, L.H. Chen, L.L. Huang, S.L. Cao, 1093 Degradation and dissolution of hemicelluloses during bamboo 1094 41 hydrothermal pretreatment, Bioresour. Technol. 161 (2014) 215–220. 1095 doi:http://doi.org/10.1016/j.biortech.2014.03.044. 1096 [17] M.E. Vallejos, F.E. Felissia, J. Kruyeniski, M.C. Area, Kinetic study of the 1097 extraction of hemicellulosic carbohydrates from sugarcane bagasse by 1098 hot water treatment, Ind. Crops Prod. 67 (2015) 1–6. 1099 doi:http://doi.org/10.1016/j.indcrop.2014.12.058. 1100 [18] M.H. Thomsen, A. Thygesen, A.B. Thomsen, Hydrothermal treatment of 1101 wheat straw at pilot plant scale using a three-step reactor system aiming 1102 at high hemicellulose recovery, high cellulose digestibility and low lignin 1103 hydrolysis, Bioresour. Technol. 99 (2008) 4221–4228. 1104 http://linkinghub.elsevier.com/retrieve/pii/S0960852407007158. 1105 [19] M.S.R. dos Santos Rocha, B. Pratto, R. de Sousa Júnior, R.M.R.G. 1106 Almeida, A.J.G. da Cruz, A kinetic model for hydrothermal pretreatment of 1107 sugarcane straw, Bioresour. Technol. 228 (2017) 176–185. 1108 doi:http://doi.org/10.1016/j.biortech.2016.12.087. 1109 [20] S. Makishima, M. Mizuno, N. Sato, K. Shinji, M. Suzuki, K. Nozaki, F. 1110 Takahashi, T. Kanda, Y. Amano, Development of continuous flow type 1111 hydrothermal reactor for hemicellulose fraction recovery from corncob, 1112 Bioresour. Technol. 100 (2009) 2842–2848. 1113 http://linkinghub.elsevier.com/retrieve/pii/S0960852408010791. 1114 [21] A. Eseyin E., P. Steele H., An overview of the applications of furfural and 1115 its derivatives, 2015. 3 (2015) 6. doi:10.14419/ijac.v3i2.5048. 1116 [22] F.A. Castillo Martinez, E.M. Balciunas, J.M. Salgado, J.M. Domínguez 1117 González, A. Converti, R.P. de S. Oliveira, Lactic acid properties, 1118 applications and production: A review, Trends Food Sci. Technol. 30 1119 (2013) 70–83. doi:https://doi.org/10.1016/j.tifs.2012.11.007. 1120 [23] P. Gao, G. Li, F. Yang, X.-N. Lv, H. Fan, L. Meng, X.-Q. Yu, Preparation 1121 of lactic acid, formic acid and acetic acid from cotton cellulose by the 1122 alkaline pre-treatment and hydrothermal degradation, Ind. Crops Prod. 48 1123 (2013) 61–67. doi:http://doi.org/10.1016/j.indcrop.2013.04.002. 1124 [24] J.C. Parajó, G. Garrote, J.M. Cruz, H. Dominguez, Production of 1125 xylooligosaccharides by autohydrolysis of lignocellulosic materials, 1126 48 Cellulose Hydrolysis in Supercritical Water, ChemSusChem. 8 (2015) 1316 1026–1033. doi:10.1002/cssc.201403385. 1317 [77] N. Akiya, P.E. Savage, Roles of Water for Chemical Reactions in High1318 Temperature Water, Chem. Rev. 102 (2002) 2725–2750. 1319 [78] A. Kruse, A. Gawlik, Biomass Conversion in Water at 330−410 °C and 1320 30−50 MPa. Identification of Key Compounds for Indicating Different 1321 Chemical Reaction Pathways, Ind. {&} Eng. Chem. Res. 42 (2003) 267– 1322 279. doi:10.1021/ie0202773. 1323 [79] C. Promdej, Y. Matsumura, Temperature Effect on Hydrothermal 1324 Decomposition of Glucose in SubAnd Supercritical Water, Ind. {&} Eng. 1325 Chem. Res. 50 (2011) 8492–8497. 1326 http://pubs.acs.org/doi/abs/10.1021/ie200298c. 1327 [80] D.A. Cantero, C. Martínez, M.D. Bermejo, M.J. Cocero, Simultaneous and 1328 selective recovery of cellulose and hemicellulose fractions from wheat 1329 bran by supercritical water hydrolysis, Green Chem. 17 (2015) 610–618. 1330 doi:10.1039/c4gc01359j. 1331 [81] A. Romero, D.A. Cantero, A. Nieto-Márquez, C. Martínez, E. Alonso, M.J. 1332 Cocero, Supercritical water hydrolysis of cellulosic biomass as effective 1333 pretreatment to catalytic production of hexitols and ethylene glycol over 1334 Ru/MCM-48, Green Chem. 18 (2016) 4051–4062. 1335 doi:10.1039/C6GC00374E. 1336 [82] R.J.A. Gosselink, E. De Jong, B. Guran, A. Abächerli, Co-ordination 1337 network for lignin - Standardisation, production and applications adapted 1338 to market requirements (EUROLIGNIN), Ind. Crops Prod. 20 (2004) 121– 1339 129. doi:10.1016/j.indcrop.2004.04.015. 1340 [83] M.P. Pandey, C.S. Kim, Lignin Depolymerization and Conversion: A 1341 Review of Thermochemical Methods, Chem. Eng. Technol. 34 (2011) 29– 1342 41. doi:10.1002/ceat.201000270. 1343 [84] J.H. Lora, W.G. Glasser, Recent industrial applications of lignin: A 1344 sustainable alternative to nonrenewable materials, J. Polym. Environ. 10 1345 (2002) 39–48. doi:10.1023/A:1021070006895. 1346 49 [85] J. Ralph, J. Peng, F. Lu, Isochroman structures in lignin: A new β-1 1347 pathway, Tetrahedron Lett. 39 (1998) 4963–4964. doi:10.1016/S00401348 4039(98)00968-X. 1349 [86] M.M. Campbell, R.R. Sederoff, Variation in Lignin Content and 1350 Composition (Mechanisms of Control and Implications for the Genetic 1351 Improvement of Plants)., Plant Physiol. 110 (1996) 3–13. 1352 doi:10.1104/pp.110.1.3. 1353 [87] W.-J. Liu, H. Jiang, H.-Q. Yu, Thermochemical conversion of lignin to 1354 functional materials: a review and future directions, Green Chem. 17 1355 (2015) 4888–4907. doi:10.1039/C5GC01054C. 1356 [88] E.A.B. da Silva, M. Zabkova, J.D. Araújo, C.A. Cateto, M.F. Barreiro, M.N. 1357 Belgacem, A.E. Rodrigues, An integrated process to produce vanillin and 1358 lignin-based polyurethanes from Kraft lignin, Chem. Eng. Res. Des. 87 1359 (2009) 1276–1292. doi:10.1016/j.cherd.2009.05.008. 1360 [89] G. Gonzalez, J. Salvado, D. Montane, Reactions of vanillic acid in sub1361 and supercritical water, J. Supercrit. Fluids. 31 (2004) 57–66. 1362 doi:10.1016/j.supflu.2003.09.015. 1363 [90] G.L. Huppert, B.C. Wu, S.H. Townsend, M.T. Klein, S.C. Paspek, 1364 Hydrolysis in supercritical water: identification and implications of a polar 1365 transition state, Ind. Eng. Chem. Res. 28 (1989) 161–165. 1366 doi:10.1021/ie00086a006. 1367 [91] Wahyudiono, T. Kanetake, M. Sasaki, M. Goto, Decomposition of a Lignin 1368 Model Compound under Hydrothermal Conditions, Chem. Eng. Technol. 1369 30 (2007) 1113–1122. doi:10.1002/ceat.200700066. 1370 [92] L. Pan, Z. Shen, L. Wu, Y. Zhang, X. Zhou, F. Jin, Hydrothermal 1371 production of formic and acetic acids from syringol, J. Zhejiang Univ. Sci. 1372 A. 11 (2010) 613–618. doi:10.1631/jzus.A1000043. 1373 [93] K. Yoshida, J. Kusaki, K. Ehara, S. Saka, Characterization of low 1374 molecular weight organic acids from beech wood treated in supercritical 1375 water, Appl. Biochem. Biotechnol. 121–124 (2005) 795–806. 1376 [94] Z. Fang, T. Sato, R.L. Smith, H. Inomata, K. Arai, J.A. Kozinski, Reaction 1377 50 chemistry and phase behavior of lignin in high-temperature and 1378 supercritical water, Bioresour. Technol. 99 (2008) 3424–3430. 1379 doi:10.1016/j.biortech.2007.08.008. 1380 [95] M. Saisu, T. Sato, M. Watanabe, T. Adschiri, K. Arai, Conversion of Lignin 1381 with Supercritical Water - Phenol Mixtures, Energy & Fuels. (2003) 922– 1382 928. 1383 [96] K. Okuda, M. Umetsu, S. Takami, T. Adschiri, Disassembly of lignin and 1384 chemical recovery - Rapid depolymerization of lignin without char 1385 formation in water-phenol mixtures, Fuel Process. Technol. 85 (2004) 1386 803–813. doi:10.1016/j.fuproc.2003.11.027. 1387 [97] Wahyudiono, M. Sasaki, M. Goto, Recovery of phenolic compounds 1388 through the decomposition of lignin in near and supercritical water, Chem. 1389 Eng. Process. Process Intensif. 47 (2008) 1609–1619. 1390 doi:10.1016/j.cep.2007.09.001. 1391 [98] T.L.K. Yong, M. Yukihiko, Kinetic Analysis of Lignin Hydrotermal 1392 Conversion in Suband Supercritical Water, Ind. Eng. Chem. Res. 52 1393 (2013) 9048–9059. 1394 [99] T.L.K. Yong, Y. Matsumura, Reaction Kinetics of the Lignin Conversion in 1395 Supercritical Water, (2012) 0–8. 1396 [100] B. Zhang, H.J. Huang, S. Ramaswamy, Reaction kinetics of the 1397 hydrothermal treatment of lignin, Appl. Biochem. Biotechnol. 147 (2008) 1398 119–131. doi:10.1007/s12010-007-8070-6. 1399 [101] J. Li, G. Henriksson, G. Gellerstedt, Lignin 1400 depolymerization/repolymerization and its critical role for delignification of 1401 aspen wood by steam explosion, Bioresour. Technol. 98 (2007) 3061– 1402 3068. doi:10.1016/j.biortech.2006.10.018. 1403 [102] S. Kang, X. Li, J. Fan, J. Chang, Classified separation of lignin 1404 hydrothermal liquefied products, Ind. Eng. Chem. Res. 50 (2011) 11288– 1405 11296. doi:10.1021/ie2011356. 1406 [103] T. Faravelli, A. Frassoldati, G. Migliavacca, E. Ranzi, Detailed kinetic 1407 modeling of the thermal degradation of lignins, Biomass and Bioenergy. 1408 51 34 (2010) 290–301. doi:10.1016/j.biombioe.2009.10.018. 1409 [104] Wahyudiono, M. Sasaki, M. Goto, Thermal decomposition of guaiacol in 1410 suband supercritical water and its kinetic analysis, J. Mater. Cycles 1411 Waste Manag. 13 (2011) 68–79. doi:10.1007/s10163-010-0309-6. 1412 [105] J. Miller, L. Evans, J.E. Mudd, K.A. Brown, Batch microreactor studies of 1413 lignin depolymerization by bases. 2. Aqueous Solvents, Sandia Natl. Rep. 1414 (2002). doi:10.2172/800964. 1415 [106] V.M. Roberts, V. Stein, T. Reiner, A. Lemonidou, X. Li, J.A. Lercher, 1416 Towards quantitative catalytic lignin depolymerization, Chem. - A Eur. J. 1417 17 (2011) 5939–5948. doi:10.1002/chem.201002438. 1418 [107] A. Toledano, L. Serrano, J. Labidi, Organosolv lignin depolymerization 1419 with different base catalysts, J. Chem. Technol. Biotechnol. 87 (2012) 1420 1593–1599. doi:10.1002/jctb.3799. 1421 [108] B.F. Ward, Tall OilChemicals from a natural, renewable source, Proc. 1422 Eight Cellul. Conf. Wood Chem. Chall. 378 (1975) 332–334. 1423 [109] J.E. Miller, L. Evans, A. Littlewolf, D.E. Trudell, Batch microreactor studies 1424 of lignin and lignin model compound depolymerization by bases in alcohol 1425 solvents, Fuel. 78 (1999) 1363–1366. doi:10.1016/S0016-2361(99)000721426 1. 1427 [110] X.E. Iriarte, Study of lignin as high added value chemical compounds 1428 source, Thesis 2014 UPV San Sebastian Spain. 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 52 1440 Figures captions 1441 Figure 1: Lignocellulosic biomass structure 1442 Figure 2. Subcritical and supercritical water properties around the critical point. 1443 Figure 3: Liquid profiles at the output of a packed bed reactor during a 1444 hydrothermal extraction process: (a) TOC evolution, (b) molecular weight 1445 evolution (Mw) when both stages are presents and molecular weight evolution 1446 when only stage 2 is present. 1447 Figure 4: Relation between the pH and the extracted biomass. 1448 Figure 5: Hydrothermal extraction of eucalyptus in a semi-continuous reactor 1449 (solid time of 90 min): evolution of the hemicellulose extraction yield (Yield tot), 1450 the yield of hexoses (C6), pentoses (C5) and degradation products for eucalyptus 1451 with temperature (a) and residence time at 185 °C (b) [4] 1452 Figure 6. Effect of subcritical water temperature on the extraction of different 1453 phenolic compounds from defatted rice bran (residence time = 10 min). Obtained 1454 from Pourali et al [40] 1455 Figure 7. Cellulose formula 1456 Figure 8. Sugars yield from cellulose hydrolysis in hydrothermal medium along 1457 reaction time. Experiment temperature: red = 400ºC; yellow = 350 ºC; blue = 1458 300ºC. Experiment pressure ( ♦ ) 27 MPa; ( ■ ) 25 / 23 MPa and ( ▲ ) 23 / 18 MPa 1459 [76]. 1460 Figure 9. Reaction pathway for cellulose hydrolysis in supercritical water based 1461 on [71]. 1462 Figure 10. Kinetic constants Arrhenius fitting for fructose dehydration to 5-HMF 1463 at 25 MPa and temperatures between 300 and 400ºC [76]. a) Kinetic evaluation 1464 just considering cellulose and derived products concentration. b) Kinetic 1465 evaluation also considering protons and hydroxide ions concentrations as 1466 reagents. 1467 53 Figure 11. Kinetic analysis for cellulose concentrations of 5, 15 and 20 % w/w 1468 (corresponding to 1.5, 4.5 and 6 % w/w at the reactor inlet). The regression 1469 coefficients were: 0.90, 0.81 and 0.96, respectively [71]. 1470 Figure 12. Typical structure of lignin derived from hardwood (left) and softwood 1471 (right) [87] 1472 Figure 13. Proposed scheme for degradation of lignin under near and 1473 supercritical condition 1474 1475 54 Appendix 1. Solid and liquid residence time 1476 1477 During an extraction o reaction process where a packed bed reactor is involved (Figure S1) two 1478 different residence times can be defined, one for the solid and another one for the liquid. The 1479 solid residence time corresponds to the amount of time spent during the operation since it is 1480 fixed inside the reactor. For instance, the solid residence time in the work of Cabeza et al. [1] 1481 was 94 min because they treated 5 g of holm oak with hot pressurized water during 94 min. In 1482 contrast, the liquid is continuously flowing through the reactor. Therefore, the liquid residence 1483 depends on the reactor volume (V), the reactor porosity () and the volumetric flow (Q) fed, 1484 being this time defined as V·/Q. For this reason, it was between 2 and 15 min in the work of 1485 Cabeza et al. [1] since each experiment was done with a different volumetric flow. 1486 1487 To sum up, the solid residence time refers to the time that the solid is being treated with the 1488 liquid. And the liquid residence time is the time that the liquid is inside the reactor. 1489 1490 1491 1492 1493 1494 Figure S1: packed bed reactor scheme 1495 1496 [1] A. Cabeza, F. Sobrón, F.M. Yedro, J. García-Serna, Two-phase modelling and simulation 1497 of the hydrothermal fractionation of holm oak in a packed bed reactor with hot 1498 pressurized water, Chem. Eng. Sci. 138 (2015) 59–70. 1499 doi:http://doi.org/10.1016/j.ces.2015.07.024. 1500 1501 1502 55 Appendix 2. Sudden expansion micro-reactor 1503 1504 P-1 P-2 Biomass Tank Water Tank CV-1 CV-2 PI-1 PI-2 V-1 M PT-1 HE-1 Flash HE-2 HE-3 V-2 V-3 Heater SV-1 SV-2 Vapor Outlet Liquid Outlet TT TTTT TT TT TT TT TT PI-3/PT-2 1505 Figure S2. Experimental set-up where a micro-reactor was used to hydrolyze cellulose and biomass at 1506 supercritical water conditions [1]. 1507 1508 The results of cellulose and biomass hydrolysis in supercritical water discussed in the main 1509 manuscript [1–7] were performed in the continuous plant of the FASTSUGARS process, able to 1510 hydrolyze biomass in SCW at temperatures up to 400 ºC and pressures up to 30 MPa. A scheme 1511 of the experimental set-up designed by the High Pressure Processes Group is shown in Figure 1512 S2. 1513 Briefly, water and a biomass suspension were continuously pumped to the reactor at the operating 1514 pressure (25 MPa). At the inlet of the reactor (as a tee junction-M-) the biomass was 1515 instantaneously heated up by mixing it with a SCW stream, reaching in that way the operating 1516 temperature (400 ºC). After the desired reaction time was achieved, the reactor effluent was 1517 suddenly depressurized through a high temperature valve (V-1) obtaining an instantaneous 1518 cooling and therefore, stopping the reactions. The cooling method was an important part of the 1519 FASTSUGARS process, because it was the mechanism used to effectively stop the reactions, 1520 avoiding uncontrolled reactions and the dilution of the products, which would occur if they were 1521 cooled down by quenching. 1522 An electric heater was used to control the temperature of the water stream with an adjustable 1523 power up to 10 kW. Also, a heat exchanger (HE-1) was used to both preheat the water stream 1524 and cool down the product, introducing in that way a heat integration system. SCW was supplied 1525 up to a maximum flow rate of 5 kg/h by pump P-2 and biomass suspension was fed to a maximum 1526 flow rate of 3 kg/h by pump P-1. 1527 Finally, a flash chamber separator was installed after the reactor, allowing the separation of the 1528 products into two phases: a vapor phase mainly composed of water and a liquid phase with the 1529 concentrated product. After this stage, two heat exchangers were used to cool down the sample 1530 to room temperature (HE-2 and HE-3). 1531 1532 [1] C.M. Martínez, D.A. Cantero, M.D. Bermejo, M.J. Cocero, Hydrolysis of cellulose in 1533 supercritical water: reagent concentration as a selectivity factor, Cellulose. 22 (2015) 1534 2231–2243. doi:10.1007/s10570-015-0674-3. 1535 [2] D.A. Cantero, M.D. Bermejo, M.J. Cocero, Governing Chemistry of Cellulose Hydrolysis 1536 in Supercritical Water, ChemSusChem. 8 (2015) 1026–1033. 1537 doi:10.1002/cssc.201403385. 1538 Reaction Section 56 [3] D.A. Cantero, M.D. Bermejo, M.J. Cocero, Kinetic analysis of cellulose depolymerization 1539 reactions in near critical water, J. Supercrit. Fluids. 75 (2013) 48–57. 1540 http://www.sciencedirect.com/science/article/pii/S0896844612003841. 1541 [4] D.A. Cantero, Á. Sánchez Tapia, M.D. Bermejo, M.J. Cocero, Pressure and temperature 1542 effect on cellulose hydrolysis in pressurized water, Chem. Eng. J. 276 (2015) 145–154. 1543 doi:10.1016/j.cej.2015.04.076. 1544 [5] D.A. Cantero, M. Dolores Bermejo, M. José Cocero, High glucose selectivity in 1545 pressurized water hydrolysis of cellulose using ultra-fast reactors, Bioresour. Technol. 1546 135 (2013) 697–703. doi:10.1016/j.biortech.2012.09.035. 1547 [6] D.A. Cantero, C. Martínez, M.D. Bermejo, M.J. Cocero, Simultaneous and selective 1548 recovery of cellulose and hemicellulose fractions from wheat bran by supercritical water 1549 hydrolysis, Green Chem. 17 (2015) 610–618. doi:10.1039/c4gc01359j. 1550 [7] A. Romero, D.A. Cantero, A. Nieto-Márquez, C. Martínez, E. Alonso, M.J. Cocero, 1551 Supercritical water hydrolysis of cellulosic biomass as effective pretreatment to catalytic 1552 production of hexitols and ethylene glycol over Ru/MCM-48, Green Chem. 18 (2016) 1553 4051–4062. doi:10.1039/C6GC00374E. 1554 1555