Hydrolysis of Cellobiose and Xylan over TiO2-Based Catalysts - ACS

Research Article pubs.acs.org/journal/ascecg

Cite This: ACS Sustainable Chem. Eng. XXXX, XXX, XXX−XXX

Hydrolysis of Cellobiose and Xylan over TiO2‑Based Catalysts Léa Vilcocq,* Édouard Rebmann, You Wayne Cheah, and Pascal Fongarland

LGPC, Laboratoire de Génie des Procédés Catalytiques, CNRS/CPE Lyon/UCBL-Université de Lyon, 43 Boulevard du 11 Novembre 1918, 69616 Villeurbanne, France S Supporting Information *

ABSTRACT: Hydrolysis of hemicelluloses is a key step in biorefineries processes. In this paper, TiO2-based catalysts were prepared and applied to the hydrolysis of a model compound, cellobiose, and then to xylan from corncob. Hydrolysis of cellobiose yielded glucose (60%), hydroxy-methylfurfural (HMF) and humins in various amounts depending on the catalyst nature. Activity and selectivity to glucose varied with the nature of dopant (W or Zr) and the calcination temperature of the catalyst. A kinetic model was built to elucidate the formation of humins and showed that the direct pathway “cellobiose to humins” is predominant in our reaction conditions. TiO2−W catalyst was then applied to xylan hydrolysis. Xylan reacted much faster than cellobiose and yielded xylose and furfural with low humins production. KEYWORDS: Hemicellulose, Hydrolysis, TiO2, ZrO2, Tungsten, Biomass, Kinetics, Heterogeneous catalysts

■

INTRODUCTION Biomass is considered as the only source of renewable carbon for the production of energy and chemicals. After a first generation of biofuels produced from vegetable oils (biodiesel) or from sugar culture (bioethanol) in the early 2000s, nowadays lignocellulose, coming from nonedible agricultural resources, is the center of research attention. Biorefineries were designed to transform and valorize lignocellulose. To improve both economy and carbon balance of biorefineries, the use of all the lignocellulose, including cellulose but also hemicellulose (20−40 wt %) and lignin (15−35 wt %), is essential.1 In this context, hemicellulose valorization is a key-point of the viability of biorefineries. Hemicelluloses are polymers of pentoses (C5 sugars, e.g. xylose, arabinose) and hexoses (C6 sugars, e.g. mannose, galactose, glucose), noncrystalline, with a low polymerization degree (Dp), between 50 and 300.2 Their structure varies with the source of lignocellulosic biomass. For example, xylans and arabinoxylans are the main hemicelluloses in hardwood whereas arabinogalactans are in majority in softwood. Acetyl groups are also present in hemicelluloses. Value-added compounds for fine chemistry, parapharmaceuticals and agrifood, such as rare sugars and sugars alcohols,3 furfural,4,5 hydroxymethylfurfural (HMF), or furan-dicarbloxylic acid (FDCA)6 can be produced from hemicelluloses. Hemicelluloses must be depolymerized into sugars before further chemical valorization. Such depolymerization usually occurs through hydrolysis reaction with an acid catalyst in aqueous medium. Acid hydrolysis of hemicelluloses to obtain monomer sugars is currently performed using mineral acids or enzymes.7 However, these techniques encounter major drawbacks: corrosion, need for neutralization, and production of mineral salts in the first case, low productivity in the second © XXXX American Chemical Society

case. Therefore, there is an incentive to replace mineral acids by solid heterogeneous acid catalysts that can be easily recovered and reused. The molecular mechanism of hydrolysis is well described in the literature:8,10 hydrolysis is assumed to occur via the adsorption of oligosaccharides using oxygen electron lone pairs or hydroxyl groups, then protonation of the oxygen atom of the ether linkage, followed by the insertion of a water molecule, leading to a C−O bond cleavage. A proton is then released.8 Then sugars produced by the hydrolysis reaction are degraded into several compounds, including furan derivatives such as furfural and HMF.9 In the literature, the catalytic hydrolysis of oligo- or polysaccharides using solid acid catalysts is often studied on model compounds such as sucrose, maltose, or cellobiose,12,13 which are disaccharides of glucose (and fructose in the case of sucrose). The use of solid acid catalysts for the hydrolysis of oligosaccharides was reviewed in 2014.8 There are some recent examples of solid acid catalysts for hemicelluloses hydrolysis: organic resins were used for the hydrolysis of xylan14 and Oacetyl-galactoglucomannan,11 USY zeolites were used for the hydrolysis of xylan,15 arabinogalactan,16 and arabinoxylan,17 and functionalized carbons were used for the hydrolysis of xylan.18 However, both resins and zeolites encounter severe stability issues in hydrolysis conditions. Organic resins are easily deactivated and solubilized in hot water, leading to the presence of homogeneous sulfonic species,19 whereas zeolites structures can be damaged and leach in the presence of water.15,20,21 Therefore, there is a need for robust solid acid Received: January 30, 2018 Published: February 28, 2018 A

DOI: 10.1021/acssuschemeng.8b00486 ACS Sustainable Chem. Eng. XXXX, XXX, XXX−XXX

Research Article

ACS Sustainable Chemistry & Engineering

after mineralization with a ICP-OES Activa Jobin Yvon on 10 mg samples (triplicate). TGA analysis was performed on a Mettler Toledo TGA/DSC1 thermobalance equipped with an autosampler and a DSC heat flow measurement for simultaneous detection of enthalpy events. Ammonia temperature-programmed desorption (TPD-NH3) was performed on a Belcat-M analyzer: the catalyst sample was first pretreated under helium flow at 350 °C for 1 h, then cooled to 100 °C and ammonia-saturated in a stream of 1% NH3/He for 0.5 h. After purge under helium at 100 °C for 0.25 h, ammonia was desorbed using a linear heating rate of 10 °C min−1 to 600 °C. Desorbed ammonia was analyzed on a TCD detector. The relative quantification of acid sites was done by measuring the area under the curve of ammonia desorption. Catalytic Test. All catalytic reaction were carried out in batch mode using a 150 mL autoclave (Top Industrie) equipped with a temperature probe, sampling line, mechanical stirrer, and heating jacket. In a typical hydrolysis experiment, the initial reaction medium, comprising 50 g L−1 of cellobiose or xylan and 50 wt % of catalyst, i.e., 4 g of cellobiose or xylan and 2 g of catalyst in 80 mL of water, was heated at 140 °C for 6 h and stirred at 1600 rpm. The catalyst was dried at 110 °C before test. Before heating, the reaction medium was purged with N2. Samples from the reaction medium were collected regularly during the catalytic reaction. The total volume of sampling is less than 15% of the total volume of reaction medium. The catalyst was sampled with the liquid, so the catalyst concentration was constant during the test. Catalyst was then removed by filtration. Approximately 0.5 g of sample was collected from the filtrated sample and diluted with 1.5 g of distilled water. The diluted samples were filtrated and analyzed by high-performance liquid chromatography (HPLC). For xylan hydrolysis, the same experimental procedure was employed to study the catalytic reaction using xylan from corncob as substrate in a batch reactor. For test using recovered catalyst, the catalyst was dried in the oven overnight and calcined at 550 °C under air flow during 6 h with a control of exothermicity in order to remove carbonaceous compounds and unknown substances deposited on the catalyst surface and pores. When the reaction was completed, the autoclave was cooled down immediately. The reaction medium was filtrated and centrifuged to separate the solids (catalyst and humins). Product Analysis. The diluted samples from reaction medium were filtrated and analyzed by HPLC on a Shimadzu instrument equipped with refractive index detector (RID) and a diode array detector (DAD) for UV analysis. The temperature of the HPLC column Phenomenex Rezex ROA was maintained at 50 °C, and 0.005 N H2SO4 solution was used as the mobile phase. Reaction products identification was determined by commercially available standards of cellobiose, glucose, xylose, arabinose, acetic acid, HMF, and furfural. Quantification was made using calibration curves made with four distinct standard solutions of commercial compounds. Concentrations were expressed in gcarbon L−1. The amount of unreacted xylan from the catalytic reaction was determined by post-hydrolysis, following a NREL protocol.27 A liquid sample was placed in an Ace pressure tube at 121 °C for 1 h with 4% sulfuric acid. After 1 h, the hydrolysate was cooled to room temperature and neutralized to pH 6 using calcium carbonate. The reaction medium was then filtered and diluted. Diluted sample was analyzed by HPLC. The concentration of xylo-oligosaccharides and gluco-oligosaccharides (i.e., unreacted xylan) was calculated based on the increase in concentration in xylose and glucose, respectively, after posthydrolysis. The concentration of acetyl groups was calculated from the increase in concentration in acetic acid after posthydrolysis. Degree of acetylation was calculated by dividing the number of acetyl groups by the number of sugar units. Time zero was defined as the time when reaction medium reached 140 °C. Conversion at time t was calculated as the difference between starting material concentration (cellobiose, xylo-oligosaccharides, or gluco-oligosaccharides) at time zero and at time t divided by starting material concentration at time zero. The yield of a given product at time t was calculated as the concentration of this product at time t divided by the concentration of starting material at time zero. The

catalysts for hydrolysis and more generally for biomass valorization reactions in hydrothermal medium. In a recent review, Lange stated that one of the most straightforward options for the preparation of stable catalyst in aqueous medium was the use of TiO2.22 Indeed, titania has a potential stability in water and is already used as a catalyst support in numerous aqueous phase processes, particularly for wastewater treatment.23 However, TiO2 is not active in itself for acid catalysis. Acidity must be brought by dopants like tungsten or zirconium oxides. Tungstated titania was applied to aqueous phase reactions and did not show any deactivation sign.24 Titanium−zirconium mixed oxides were also used as solid acid catalysts in hydrothermal conditions.25,26 In this paper, TiO2-based acid catalysts were prepared and applied for hydrolysis of hemicelluloses for the first time. The activity, selectivity, and recyclability of the catalysts were studied first on cellobiose as a model compound. Then the catalytic hydrolysis of a xylan was investigated. Kinetic models were designed and used to determine reaction mechanisms for cellobiose and xylan hydrolysis reactions. Critical assessments on the use of model compounds for investigation of reactivity of complex polysaccharides were made.

■

EXPERIMENTAL SECTION

Materials. TiO2, zeolite HY, cellobiose, HMF, and arabinose were purchased from Alfa Aesar. Xylose, tungstic acid, furfural, and TiCl4 were purchased from Sigma-Aldrich. ZrOCl2 was purchased from Acros Organics. Corncob xylan was purchased from Carbosynth Ltd. Glucose was purchased from Fisher Chemicals. Catalyst Preparation. For TiO2−W preparation, TiO2 was mixed in distilled water to form a suspension. H2WO4 was dissolved in 30% H2O2 (concentration 0.25 M). The tungsten solution was added drop by drop over TiO2 suspension using a syringe pump. The suspension was stirred overnight and then centrifuged for 30 min. The recovered solid was dried at 110 °C overnight and crushed afterward. It was calcined at 600 or 700 °C for 3 h under air flow.24 For TiO2−ZrO2 synthesis, ZrOCl2 was added to water in an ice bath (concentration 0.15 M) and stirred 15 min (pH around 3). TiCl4 was slowly added to the Zr solution under air stream. The reaction is exothermic and produces a white vapor (HCl). The solution was then stirred at 0 °C for 3 h. Aqueous ammonia 2.5% was then added drop by drop and formed a white precipitate. The suspension was stirred overnight and then filtered and washed with deionized water until disappearance of chloride ions in the filtrate. The gel obtained was dried at 110 °C overnight and then calcined at 600 or 700 °C for 3 h under air flow.26 Table 1 summarizes the preparation of solid acid catalysts. All catalysts were used in the form of fine powder. Zeolite HY was calcined at 500 °C under air stream before use. Catalysts Characterization. X-ray diffraction (XRD) was performed on a diffractometer Bruker D8 Advance. BET measurements have been performed with a Micrometrics ASAP 2010 apparatus by nitrogen adsorption at 77 K. All the samples were first degassed at 120 °C for 1 h and then at 350 °C for 3 h under vacuum. Tungsten and zirconium loading were measured using ICP analysis

Table 1. Catalysts Preparation catalyst Ti− W600 Ti− W700 Ti− Zr600 Ti− Zr700

preparation method

dopant

precursor

calcination temperature (°C)

wet impregnation wet impregnation coprecipitation

W

H2WO4

600

W

H2WO4

700

Zr

ZrOCl2

600

coprecipitation

Zr

ZrOCl2

700

B


Research Article

ACS Sustainable Chemistry & Engineering Table 2. Catalysts Characterizations catalyst

BET specific surface area (m2 g−1)

average pore diameter (nm)

crystalline phase (from XRD)

elemental analysis (wt %)

no. of acid sites (μmol gcat−1)

Ti−W600 Ti−W700 Ti−Zr600 Ti−Zr700 FAU zeolite

97 89 36 16 730a

13.0 12.8 7.7 16.0 1.5a

anatase anatase n.d. n.d. faujasitea

W 5.6% W 5.6% Zr 30% Zr 20% SiO2/Al2O3 = 5/1

245 210 145 44 n.d.

a

Data given by purchaser.

selectivity of a product was defined as the amount of product divided by the amount of all reaction products and was calculated by dividing yields by conversion. Carbon balance was calculated by dividing the sum of concentrations of all identified products and starting material at time t by the concentration of starting material at time zero. The loss of carbon during the reaction was attributed to the formation of humins, i.e., non-identified degradation products in liquid or solid phase. Kinetic Model. In this study, we consider a stirred tank reactor in batch mode (no output or input after t0). We consider our system as a pseudo-homogeneous reaction occurring in the liquid phase which mainly consists of water. The reaction medium volume was assumed to remain constant over the reaction. The density of the liquid density and the mass of active catalyst are reasonably assumed to be constant in our range of experimental conditions. Series of recycling test were performed to check the stability of the catalyst (cf. previous section). The absence of diffusional or thermal transfer limitation has been checked and details about the methodology are supplied in the Supporting Information. On the basis of the initial quantity of matter, on the amount of identified products, it is admitted that the amount of water produced or consumed during the hydrolysis or other side reactions is not significant regarding the initial amount of water used as a solvent. Consequently, the concentration of water is assumed to be constant. In addition, all the experiments were performed with a constant mass of catalyst (25 g L−1). Hence, this study considers apparent kinetic constants. Nevertheless, we took care to strictly keep the same reaction conditions when comparing different substrates and catalysts. Other things being equal, the kinetic constants obtained from different catalysts can be compared. Typically, we assumed first order reaction for all elementary steps. For example, for mechanism A, the following set of equation was considered: dCell(t ) = − (k1 + k5) Cell(t ) dt

(1)

dGlc(t ) = k1 Cell(t ) − (k4 + k 2) Glc(t ) dt

(2)

dHMF(t ) = k 2 Glc(t ) − k 3 HMF(t ) dt

(3)

dHum(t ) = k5 Cell(t ) + k4 Glc(t ) + k 3 HMF(t ) dt

(4)

calculated values. This optimization was simply achieved with the generalized reduced gradient algorithm proposed in the Excel Solver. n

min β ̂ SSE(β) =

̂ 2 ∑ (yi − f (xi , β )) i=1

(5)

where SSE is the error sum of squares,yi the experimental concentration data, f(xi,β̂) the analytical solution of the ODE system, i.e., the calculated concentrations as a function of xi (time, initial conditions, and temperature), and β̂ the adjusted vector which contains the final estimates of the kinetic parameters.

■

RESULTS Catalyst Preparation and Characterization. Titanium oxide based catalysts are known to be stable in hydrothermal conditions and, as such, are often used in catalytic processes for water treatment and biomass conversion. In this work, TiO2based catalysts were prepared and characterized. Physical and chemical properties of the prepared catalysts are summarized in Table 2. Tungstated titania Ti−W600 and Ti−W700 were prepared by wet impregnation of tungstic acid in hydrogen peroxide solution, a method already described in the literature. Two batches were prepared and calcined at 600 and 700 °C, respectively. The support is a low crystalline anatase TiO2 with a large BET specific surface area (100 m2 g−1) and mesoporous volume (0.34 mL g−1). The porosity does not change drastically during tungsten impregnation and calcination. The final tungsten content was 5.6 wt %. The calcination temperature does not influence the final properties of the solid. Mixed titania−zirconia was prepared by hydrolysis and coprecipitation of chloride salts TiCl4 and ZrOCl2. Two batches were prepared and calcined at 600 and 700 °C, respectively. After calcination, Zr content was 30 wt %. Ti− Zr600 was a low crystalline oxide with medium surface area (35 m2 g−1) and small mesoporous volume (0.07 mL g−1). Ti− Zr700 had a small surface area (16 m2 g−1) and mesoporous volume in accordance (0.06 mL g−1). The characterization of acidity of solid catalysts was done with TPD-NH3. This method gives an estimation of acid sites number and strength but does not differentiate Brønsted and Lewis acid sites. The desorption curves are presented on Figure 1. Tungstated titania catalysts exhibit large desorption peaks with two maxima at 205 and 438 °C, corresponding to medium to strong acid sites. The calcination temperature of Ti−W catalysts does not influence drastically the acidity. Ti−Zr600 shows a bimodal repartition of acid sites with two peaks centered on 216 and 502 °C, corresponding to medium and strong acid sites. Ti−Zr700 only gives a small and broad peak at 216 °C. Thus, calcination TiO2−ZrO2 at 700 °C instead of 600 °C led to lose some specific surface area and most of the strong acid sites. The number of acid sites (Table 2) was comparatively lower for Ti−Zr samples and particularly for Ti−Zr700.

with Cell, Glc, HMF, Hum, for cellobiose, glucose, hydroxymethylfurfural, and humins, respectively, and these initial conditions Cell(t = 0) = Cellini, Glc(t = 0) = Glcini, HMF(t = 0) = HMFini, Hum(t = 0) = 0. The concentrations are expressed in carbon amounts (cf. Kinetic Model), so there is no need for stoichiometric factors. The analytical solutions of the previous ordinary differential equation (ODE) system (see the Supporting Information part S3) were provided by the function “dsolve” of Maple software and have been implemented in Excel. Hence, the adjustments of the kinetic parameters for these nonlinear systems have been performed numerically to fit the data. We used a traditional nonlinear leastsquares technique in which we minimized the objective function, i.e., the error sum of squares between the experimental data and the C


Research Article


catalyst by varying the stirring rate from 800 to 1600 rpm (see the Supporting Information part S1 for more details). The performances of TiO2-based catalysts are presented in Figure 2. A commercial HY zeolite (faujasite structure) was also tested for benchmark. Cellobiose conversion (Figure 2a) varied with the type of catalyst used. The rate of conversion was highest for Ti−W catalysts, reaching 90% conversion after 6 h. Both Ti−W catalysts exhibited similar catalysts performances, in accordance with their similar solid properties. In contrast, Ti−Zr catalysts led to lower cellobiose conversion rates, with 76% final conversion for Ti−Zr600 and 43% for Ti−Zr700. The faujasite HY zeolite led to a low conversion rate with 21% final conversion. One can note that the final conversion of cellobiose obtained with HY zeolite is similar to the one obtained without catalyst. This indicates a trivial catalytic activity for the hydrolysis reaction. Glucose selectivity (Figure 2b) was highest for both Ti−W catalysts, decreasing slowly from 73% to 65% when cellobiose conversion increased, indicating that glucose degradation reactions are favored at high conversion. Ti−Zr700 led to 72% glucose selectivity whereas Ti−Zr600 yielded to 55% glucose selectivity. It should be noted that glucose selectivity was stable over the studied conversion range for both Ti−Zr catalysts. Glucose selectivity for HY zeolite was rather low (around 40%) particularly when compared to the blank test, which shows a catalytic activity of HY zeolite for glucose degradation reactions. Besides glucose, HMF was detected in significant amounts. HMF selectivity (Figure 2c) tends to follow the reverse trend of glucose selectivity, being low when glucose selectivity is high,

Figure 1. TPD-NH3 desorption curves for TiO2-based catalysts.

Catalytic Hydrolysis of Cellobiose. The prepared catalysts were tested on the hydrolysis of a model compound, cellobiose. Hydrolysis of cellobiose produces glucose, which can undergo isomerization and dehydration to form hydroxymethylfurfural (HMF) in acidic conditions. Fructose is often described as an intermediate for HMF production from glucose.28 Sugars and HMF can also be degraded in nonidentified condensed products, i.e., humins. Cellobiose reactivity was studied in a batch reactor at 140 °C. In these conditions, the conversion in absence of catalyst was 18% after 6 h and so was considered as negligible. The absence of external mass transfer limitations was checked for Ti−W600

Figure 2. Cellobiose hydrolysis over solid acid catalysts. (a) Conversion as a function of time, (b) glucose selectivity as a function of cellobiose conversion, (c) HMF selectivity as a function of cellobiose conversion, and (d) carbon balance as a function of cellobiose conversion. Reaction conditions: 140 °C, 50 g L−1 cellobiose initial concentration, ratio of catalyst−cellobiose 1:2. Lines are drawn for a guide to the eye. D


Research Article


Kinetic Model of Cellobiose Hydrolysis. The kinetics of hydrolysis of cellobiose was already described in the literature.13 After hydrolysis of cellobiose, glucose can be dehydrated to HMF; sugars and HMF can also produce non-identified humins. The reaction system is depicted on Scheme 1. It was assumed that all reactions were catalytic since a negligible rate of conversion was observed for either cellobiose, glucose, or HMF in water at 140 °C without catalyst.

e.g., for Ti−W catalysts (6% HMF selectivity at 90% cellobiose conversion) and high when glucose selectivity is low, e.g., for Ti−Zr600 (14% HMF selectivity). However, in the case of zeolite HY, selectivity in both glucose and HMF was low (40% and 12%, respectively), indicating the massive production of non-identified degradation products from cellobiose. Other compounds were detected by HPLC-RI in traces. Particularly, fructose was identified but in concentrations too small to be quantified. Carbon balance (Figure 2d) was found to follow a general trend for all catalysts, decreasing slowly with cellobiose conversion. This loss of carbon can be attributed to the formation of non-identified degradation products, i.e., humins, either in solution or adsorbed on the catalyst surface. Indeed, spent catalysts were brown and TGA analysis of Ti− W600 catalyst after reaction and filtration (see the Supporting Information part S2) under air showed a mass loss around 400−500 °C, corresponding to the oxidation of carbonaceous matter in CO2. Catalyst Recycling. The stability of solid acid catalysts during aqueous phase reactions is a major issue for the development of biomass valorization processes. Catalyst deactivation can occur by leaching of the active phase, loss of structure (e.g., amorphization) or of specific surface area, or by coking mechanisms.21 The recyclability of our best hydrolysis catalyst, Ti−W600, was evaluated over four runs. At the end of each run, the catalyst was filtered, dried, and calcined under air before being used again. Figure 3 depicts the conversion and

Scheme 1. Mechanism of Cellobiose Hydrolysis over a Solid Acid Catalyst

Kinetic modeling was based on the following hypothesis: (i) the system is a batch reactor with a constant volume of liquid; (ii) all reactions follow pseudo-homogeneous, first order kinetics; (iii) the water content changes are negligible; (iv) kinetic constants ki are apparent pseudo-first order rate constants including a factor corresponding to the catalyst concentration. Several reaction pathways were considered. In model A, k3, k4, and k5 were set to zero. In model B, k3 was considered but k4 and k5 were set to zero. In model C, k3 and k4 were considered but k5 was set to zero. Finally in model D, all degradation pathways (k3, k4, and k5) were considered. Figure 4 represents the trends in concentrations of cellobiose, glucose, HMF, and humins over 24 h obtained experimentally and with models A, B, C, and D after adjustment with experimental data. The trend for all plots from all kinetic models generally conformed to the predicted theory showing an increase of glucose concentration due to the hydrolysis of cellobiose and the formation of HMF from dehydration of glucose. However, the first kinetic model A poorly fits the experimental values. This can be validated by observing that all the scatters for cellobiose, glucose, and HMF concentration deviate extensively from the theoretical trend obtained from this kinetic model. By addressing such issue, an alternative degradation of HMF to humins was taken into account in the reaction pathway for model B. Hence, such a model seems to improve the fitting of values by lowering the sum of errors despite the fitting for glucose and humins concentrations was still irrelevant. Consequently, an additional degradation reaction of glucose to humins was considered for model C. A consistency with the experimental values appears to be obvious in this case with even lower sum of errors as compared to model B. The reaction scheme was then further improved to model D by considering the formation of humins from all compounds during the catalytic reaction. The last kinetic model came out to be the best fit and shows the lowest sum of errors among all kinetic models. Finally, the comparison of four different models with experimental data obtained with Ti−W600 catalyst on 24 h (Figure 4) shows that the best fit between kinetic model and experimental data is obtained for model D, i.e., when all degradation pathways were considered. Additional data obtained with glucose as a starting material with the same

Figure 3. Ti−W600 catalyst recycling for cellobiose hydrolysis. Reaction conditions: 140 °C, 50 g L−1 cellobiose initial concentration, ratio of catalyst−cellobiose 1:2.

glucose yield after 6 h during recyclability tests. Between runs 1 and 2, a decrease in catalytic performances is observed. However, between runs 2 and 4, a decrease in the conversion and glucose yield after 6 h is limited to 6% and 10%, respectively. These results point out a negligible leaching of the tungsten active phase. It was concluded that TiW catalysts are mostly stable in our reaction conditions and thus represent a promising class of robust acid catalysts for biomass valorization processes. Finally, TiO2-based catalysts were active for cellobiose hydrolysis. The type of dopant as well as the calcination temperature had an influence on the final activity and selectivity of solid acid catalysts. When compared to a commercial zeolite, our catalysts exhibited superior performances. Particularly, Ti− W catalysts were active, selective, and recyclable for the hydrolysis of cellobiose, leading to 59% yield of glucose at 90% conversion. E


Research Article


Figure 4. Concentration profiles for cellobiose hydrolysis over Ti−W600. Dots correspond to experimental data and lines correspond to kinetic modeling: (a) model A, (b) model B, (c) model C, and (d) model D. Cellobiose (●), glucose (⧫), HMF (▲). Dotted lines: humins.

major source of humins formation. Such direct degradation of a disaccharide was already described by Kumar and Wyman during hydrolysis of xylobiose with H2SO4.29 The formation of humins from glucose, fructose, and HMF was described in detail by Lund et al. using IR spectroscopy. They concluded that in aqueous phase, at 125 °C, at low pH, humins are not formed directly from glucose but by a pathway including HMF and DHH (dioxo-6-hydroxyhexanal) as intermediates and aldol condensation reactions,30 in accordance with our results. Finally, a kinetic model valid for several acid catalysts was designed for cellobiose hydrolysis. It integrates degradation pathways from cellobiose rarely found in the literature but, in our point of view, important for a global understanding of the reaction. In our reaction conditions, humins were formed in the majority from cellobiose and not from reaction products. Scheme 2 summarizes the mechanistic insights from kinetic modeling. A simplified version of solutions of ordinary differential equations with k4 equal to zero is given in the Supporting Information (model E). Catalytic Hydrolysis of Xylan. A commercial xylan from corn cob (Carbosynth Ltd.) with a low polymerization degree was used as a starting material. Xylan is a polymer of xylose units, linked by β-1,4 ether bonds (osidic bonds). Some arabinose side units can also be present as well as acetyl groups. Normally, glucose units are not present in xylan but ex-cellulose oligosaccharides (gluco-oligosaccharides) can remain after xylan purification. The average Mn of xylan was estimated around 400 g mol−1 by SEC (data not shown). The sugar composition was analyzed by post-hydrolysis method and is presented in Table 4. Xylose units are in the majority (87.5%); arabinose and glucose were also detected in minor concen-

catalyst corroborated the accuracy of model D for cellobiose hydrolysis (see the Supporting Information part S4). Therefore, model D was used for determination of apparent kinetic constants for all catalysts at 140 °C. Fits between kinetic model data and experimental data for all catalysts are presented in Figure 5. Calculated values of kinetic constants are listed in Table 3. The nature of catalyst has a strong impact on the rate of hydrolysis, the value of k1 being three times higher for Ti−W catalysts than for Ti−Zr700. The value of k1 is proportional to the number of acid sites measured by TPD-NH3, i.e., the rate of cellobiose hydrolysis is strongly dependent on the number of acid sites, which is consistent with the hydrolysis mechanism involving a proton described in the Introduction. However, the rate of glucose dehydration into HMF was similar for all catalysts. A relative ranking of catalytic activity for hydrolysis can be built based on the k1 value: Ti−W600 ≈ Ti−W700 > Ti−Zr600 > Ti−Zr700 ≫ FAU zeolite. Moreover, the catalyst selectivity for hydrolysis versus glucose dehydration into HMF, expressed as α = k1/(k1 + k2),8 follows the trend Ti−W600 ≈ Ti−W700 ≈ Ti−Zr700 > Ti−Zr600 ≫ FAU zeolite. Interestingly, the kinetic constant k4 corresponding to the degradation of glucose into humins was null for all catalysts, and k3 was also equal to zero for Ti−Zr catalysts. Thus, the degradation of glucose into humins does not seem to be a pathway in our reaction conditions. Indeed, according to model D, humins were mostly formed from cellobiose, at least at low conversion level, and partly from HMF at high conversion level. Figure 6 represents the formation of humins calculated using the constants k3, k4, and k5 (see the Supporting Information part S3 for calculations) for Ti−W600. Cellobiose appears as a F


Research Article


Figure 5. Concentration profiles for cellobiose hydrolysis over solid acid catalysts: (A) Ti−W700, (b) Ti−Zr600, (c) Ti−Zr700, (d) HY zeolite. Dots correspond to experimental data and lines correspond to kinetic modeling using model D. Cellobiose (●), glucose (⧫), HMF (▲). Dotted lines: humins.

Table 3. Pseudo-Apparent Kinetic Constants Values from Kinetic Model D for Cellobiose Hydrolysis kinetic constant value at 140 °C (h−1)

Scheme 2. Mechanism of Cellobiose Hydrolysis over a Solid Acid Catalyst: Simplified Version

α

catalyst

k1

k2

k3

k4

k5

k1/k1 + k2

Ti−W600 Ti−W700 Ti−Zr600 Ti−Zr700 FAU zeolite

0.259 0.278 0.156 0.071 0.020

0.043 0.053 0.073 0.045 0.049

0.291 0.544 0 0 0.260

0 0 0 0 0.003

0.065 0.067 0.070 0.016 0.019

0.858 0.806 0.690 0.816 0.289

Table 4. Sugar Composition of Xylan sugar

formula

concentration (wt %)

xylose glucose arabinose

C5H10O5 C6H12O6 C5H10O5

87.5 11.2 1.3

Xylan hydrolysis was performed in reaction conditions similar to cellobiose hydrolysis. The reaction was performed over one of the most active catalyst for cellobiose hydrolysis, Ti−W600. Monomers (glucose and xylose mainly) and oligomers were analyzed. Besides sugars, furans (HMF and furfural) were also detected by HPLC. Figure 7a depicts the evolution of reactants and products concentrations over time. Xylo-oligosaccharides hydrolysis rate was much faster than expected from cellobiose data: the conversion overcame 90% within 2.5 h while it took 6 h for cellobiose. Xylose was produced in large amount in the first hours of the reaction, yielding 68%carbon in 2.5 h but was degraded into furfural and other non-identified products, i.e., humins, in the second part of the reaction. Furfural final yield was 40%carbon. This level of

Figure 6. Humins formation calculated from model D, catalyst Ti− W600.

trations. Acetyl groups were also detected with a concentration below 2 wt %, i.e., a degree of acetylation close to 6%. G


Research Article


Kinetic Study of Xylan Hydrolysis. For the sake of simplicity, oligosaccharides of variable chain lengths were not differentiated and considered as single compounds named XOS (for xylo-oligosaccharides) and GOS (for gluco-oligosaccharides), respectively. The kinetic models A, B, C, D developed for cellobiose reaction, considering various degradation pathways where humins are produced from either oligosaccharides, sugars, and/or furans, were adjusted with xylan hydrolysis results to calculate new sets of apparent kinetic constants. First, we considered XOS hydrolysis following reaction pathways presented on Scheme 3. Figure 8 shows the Scheme 3. Mechanism of Xylan (Xylo-Oligosaccharides, XOS) Hydrolysis over Ti−W600 Catalyst

correlation between experimental concentrations of xylooligosaccharides, xylose, and furfural and the trends predicted by models A, B, C, and D. Again, the model D, considering all degradation pathways, fitted well with the experimental data. The value of apparent kinetic constant (Table 5) of the hydrolysis step k1 is three times higher for XOS than for cellobiose. However, in the case of XOS, degradation of furfural into humins was slow, i.e., k3 is low, whereas it was a major degradation pathway in the case of cellobiose. Then the hydrolysis of the minor fraction of GOS was considered. Glucose and HMF were slowly released during GOS hydrolysis. The carbon balance of the glucose-derived compounds is 100% over the 6 h of reaction, i.e., the formation of humins from GOS, glucose, or HMF is negligible. The application of kinetic model D to the glucose fraction of the reaction medium led to a good fit with experimental data after adjustment of the kinetic constant values (Figure 9). Interestingly, the apparent kinetic constants k1 and k2 were very similar for GOS and for cellobiose, which suggests a similar mechanism. Discussion on Cellobiose as a Model for Oligosaccharides Conversion. In this work, cellobiose was first used as a model compound for the investigation of hydrolysis reactions. The purpose was the evaluation of the activity of solid catalysts on a simple model before diving in the reactivity of more complex polysaccharides. It is interesting to note that xylo-oligosaccharides and glucooligosaccharides (including cellobiose) react differently over the same solid acid catalyst, Ti−W600, and under similar reaction conditions. When compared with the kinetic study of cellobiose hydrolysis, XOS gave unexpected high hydrolysis reaction rates. Indeed, the cleavage of β-1,4-osidic bond between two xylose units seems to be favored when compared with the same cleavage between two glucose units. This phenomenon was already described for homogeneous acid hydrolysis of hemicelluloses in ionic liquid32 and attributed to a protective effect of the osidic bond by the CH2OH group of glucose not included in the cycle. Moreover, the degradation of reactants and products in humins follows different pathways for xylan and for cellobiose. HMF is a source of humins during cellobiose hydrolysis whereas degradation of furfural is null during xylan

Figure 7. Xylan hydrolysis over Ti−W600 catalyst: (a) concentrations of reactants and products as a function of time and (b) carbon balance. Reaction conditions: 140 °C, 50 g L−1 xylan initial concentration, ratio of catalyst−xylan 1:2. XOS, xylo-oligosaccharides; GOS, glucooligosaccharides.

xylose yield was in line with the one obtained by Zhou et al. with USY zeolite catalyst in a longer time and with a higher amount of catalyst31 and slightly lower than the one obtained by Chung et al. using functionalized carbon catalysts, again with a higher amount of catalyst, higher dilution, and higher temperature.18 On the other hand, gluco-oligosaccharides hydrolysis was slow and started to produce glucose in a significant amount only after 2 h. Glucose yield reached 65%carbon, based on glucooligosaccharides content. Glucose was also dehydrated into HMF (17%carbon final yield). Carbon balance was remarkably higher than in the case of cellobiose, decreasing slowly over the reaction advancement to reach 84% as a final value, i.e., the formation of non-identified humins is less important in the case of xylan than in the case of cellobiose (Figure 7b). A blank test was performed in the absence of catalyst to evaluate the non catalytic hydrolysis of oligosaccharides. After 6 h, XOS conversion reached 45% with 44%carbon xylose yield whereas GOS conversion was null (Supporting Information part S5). Obviously, xylo-oligosaccharides and gluco-oligosaccharides exhibited different behaviors during catalytic and non catalytic hydrolysis. It was assumed that xylose units and glucose units belong to two different types of oligosaccharides, i.e., xylooligosaccharides are mainly xylan chains leading to xylose and gluco-oligosaccharides are mainly glucose chains coming from cellulose hydrolysis during corn cob fractionation process. From HPLC data, it is easy to conclude that XOS do not act as a cellobiose-like polysaccharide and is much more reactive in the presence as well as in the absence of a solid acid catalyst. H


Research Article


Figure 8. Concentration profiles for xylo-oligosaccharides and related products over Ti−W600. Dots correspond to experimental data and lines correspond to kinetic modeling: (a) model A, (b) model B, (c) model C, and (d) model D. XOS (●), xylose (⧫), furfural (▲). Dotted lines: humins.

the study of model disaccharides for application on complex hemicelluloses appears to be inadequate, since disaccharides of xylose are not commercially available in large amounts. A lesson from our work is also that hemicelluloses cannot be considered as “cellulose-like” polysaccharides and must be studied for themselves.

Table 5. Pseudo-Apparent Kinetic Constants Values for Xylan kinetic constant value at 140 °C (h−1)

α

starting material

k1

k2

k3

k4

k5

k1/k1 + k2

XOS GOS

0.791 0.217

0.150 0.041

0.088

0

0.148

0.832 0.841

■

CONCLUSION In this work, we showed that TiO2-based materials are promising solid acid catalysts for biomass valorization reactions in aqueous phase. They represent a stable, active, and green alternative to mineral acids for hydrolysis of polysaccharides. Cellobiose hydrolysis was used as a model reaction to evaluate their catalytic performances. The activity and selectivity of such catalysts can be tuned by controlling several physical and chemical properties such as the nature of the dopant. The most active catalyst for cellobiose hydrolysis, Ti−W600, was also applied for the hydrolysis of xylan as a representative hemicellulose from corn cob. Kinetic modeling of cellobiose and xylan hydrolysis, including degradation pathways, allowed us to determine the origin of humins. The direct degradation of cellobiose and oligosaccharides to humins is a major pathway in degradation mechanisms. Finally, the reactivity of xylo-oligosaccharides cannot be predicted from model compounds. Particularly, the hydrolysis of xylan is faster than the hydrolysis of cellobiose in similar reaction conditions and led to less degradation products. At the most, the use of cellobiose as model compound indicates qualitatively the reactivity scale of the catalysts toward complex

Figure 9. Concentration profiles for gluco-oligosaccharides and related products over Ti−W600. Dots correspond to experimental data and lines correspond to kinetic modeling (model D). GOS (●), glucose (⧫), HMF (▲).

hydrolysis, where humins are formed mainly from xylan and xylose. Gluco-oligosaccharides seem to be closer to cellobiose reactivity since their rate of hydrolysis is comparable. Therefore, I


Research Article


the Aqueous-Phase Aerobic Oxidation of HMF to FDCA. Appl. Catal., A 2015, 506, 206. (7) Sella Kapu, N.; Trajano, H. L. Review of Hemicellulose Hydrolysis in Softwoods and Bamboo. Biofuels, Bioprod. Biorefin. 2014, 8 (6), 857. (8) Vilcocq, L.; Castilho, P. C.; Carvalheiro, F.; Duarte, L. C. Hydrolysis of Oligosaccharides Over Solid Acid Catalysts: A Review. ChemSusChem 2014, 7 (4), 1010. (9) Negahdar, L.; Delidovich, I.; Palkovits, R. Aqueous-Phase Hydrolysis of Cellulose and Hemicelluloses over Molecular Acidic Catalysts: Insights into the Kinetics and Reaction Mechanism. Appl. Catal., B 2016, 184, 285. (10) Jacobsen, S. E.; Wyman, C. E. Cellulose and Hemicellulose Hydrolysis Models for Application to Current and Novel Pretreatment Processes. Appl. Biochem. Biotechnol. 2000, 84−86 (1−9), 81. (11) Salmi, T.; Murzin, D. Y.; Mäki-Arvela, P.; Kusema, B.; Holmbom, B.; Willför, S.; Wärnå, J. Kinetic Modeling of Hemicellulose Hydrolysis in the Presence of Homogeneous and Heterogeneous Catalysts. AIChE J. 2014, 60 (3), 1066. (12) Kamiyama, Y.; Sakai, Y. Rate of Hydrolysis of XyloOligosaccharides in Dilute Sulfuric Acid. Carbohydr. Res. 1979, 73 (1), 151. (13) Marzo, M.; Gervasini, A.; Carniti, P. Hydrolysis of Disaccharides over Solid Acid Catalysts under Green Conditions. Carbohydr. Res. 2012, 347 (1), 23. (14) Hilpmann, G.; Becher, N.; Pahner, F.-A.; Kusema, B.; MäkiArvela, P.; Lange, R.; Murzin, D. Y.; Salmi, T. Acid Hydrolysis of Xylan. Catal. Today 2016, 259 (Part), 376. (15) Aellig, C.; Scholz, D.; Dapsens, P. Y.; Mondelli, C.; PérezRamírez, J. When Catalyst Meets Reactor: Continuous Biphasic Processing of Xylan to Furfural over GaUSY/Amberlyst-36. Catal. Sci. Technol. 2015, 5 (1), 142. (16) Murzin, D. Y.; Murzina, E. V.; Tokarev, A.; Shcherban, N. D.; Warna, J.; Salmi, T. Arabinogalactan Hydrolysis and Hydrolytic Hydrogenation Using Functionalized Carbon Materials. Catal. Today 2015, 257 (Part), 169. (17) Ennaert, T.; Feys, S.; Hendrikx, D.; Jacobs, P. A.; Sels, B. F. Reductive Splitting of Hemicellulose with Stable Ruthenium-Loaded USY Zeolites. Green Chem. 2016, 18 (19), 5295. (18) Chung, P.-W.; Charmot, A.; Olatunji-Ojo, O. A.; Durkin, K. A.; Katz, A. Hydrolysis Catalysis of Miscanthus Xylan to Xylose Using Weak-Acid Surface Sites. ACS Catal. 2014, 4 (1), 302. (19) Pérez Nebreda, A.; Salmi, T.; Murzin, D. Y.; Grénman, H. High Purity Fructose from Inulin with Heterogeneous Catalysis - Kinetics and Modelling. J. Chem. Technol. Biotechnol. 2018, 93 (1), 224−232. (20) Ravenelle, R. M.; Schüβler, F.; D’Amico, A.; Danilina, N.; van Bokhoven, J. A.; Lercher, J. A.; Jones, C. W.; Sievers, C. Stability of Zeolites in Hot Liquid Water. J. Phys. Chem. C 2010, 114 (46), 19582. (21) Lari, G. M.; Dapsens, P. Y.; Scholz, D.; Mitchell, S.; Mondelli, C.; Pérez-Ramírez, J. Deactivation Mechanisms of Tin-Zeolites in Biomass Conversions. Green Chem. 2016, 18 (5), 1249. (22) Lange, J.-P. Renewable Feedstocks: The Problem of Catalyst Deactivation and Its Mitigation. Angew. Chem., Int. Ed. 2015, 54 (45), 13186. (23) Descorme, C. Catalytic Wastewater Treatment: Oxidation and Reduction Processes. Recent Studies on Chlorophenols. Catal. Today 2017, 297, 324−334. (24) Vilcocq, L.; Koerin, R.; Cabiac, A.; Especel, C.; Lacombe, S.; Duprez, D. New Bifunctional Catalytic Systems for Sorbitol Transformation into Biofuels. Appl. Catal., B 2014, 148−149, 499. (25) Duan, J.; Kim, Y. T.; Lou, H.; Huber, G. W. Hydrothermally Stable Regenerable Catalytic Supports for Aqueous-Phase Conversion of Biomass. Catal. Today 2014, 234, 66. (26) Chareonlimkun, A.; Champreda, V.; Shotipruk, A.; Laosiripojana, N. Catalytic Conversion of Sugarcane Bagasse, Rice Husk and Corncob in the Presence of TiO2, ZrO2 and Mixed-Oxide TiO2-ZrO2 under Hot Compressed Water (HCW) Condition. Bioresour. Technol. 2010, 101 (11), 4179.

sugars but no direct quantitative information is afforded except on the final hydrolysis of the GOS part of the hemicellulose. In the near future, further studies of complex hemicelluloses depolymerization, including in-depth analysis of reaction products and elaborated kinetic modeling, will be necessary to improve the understanding of biomass deconstruction reactions.

■

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acssuschemeng.8b00486. Study of transfer limitations (external and internal diffusions), TGA analysis of Ti−W600 after reaction, ODE system for kinetic modeling and analytical solutions, kinetic model D applied to glucose, and blank test (no catalyst) of xylan conversion (PDF)

■

AUTHOR INFORMATION

Corresponding Author

* Phone: +33(0)4 72 43 17 61. E-mail: [email protected]. ORCID

Léa Vilcocq: 0000-0002-3381-3927 Author Contributions

The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript. Funding

This work was supported by CNRS-INSIS and INC fundings. You Wayne Cheah thanks the Erasmus + Mobility Program for the master internship grant. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS The authors thank Stéphanie Pallier (LGPC) for the N2 adsorption−desorption measurements, Frédéric Bornette (LGPC) for his technical help, IRCELYON for ICP and TPD-NH3 analyses, Olivier Boyron and C2P2 laboratory for TGA analysis, and Ruben Vera and CDHL for XRD analysis.

■

REFERENCES

(1) Vassilev, S. V.; Baxter, D.; Andersen, L. K.; Vassileva, C. G.; Morgan, T. J. An Overview of the Organic and Inorganic Phase Composition of Biomass. Fuel 2012, 94, 1. (2) Ebringerová, A.; Hromádková, Z.; Heinze, T. Hemicellulose. In Polysaccharides I; Heinze, T., Ed.; Advances in Polymer Science, Vol. 186; Springer-Verlag: Berlin, Heidelberg, Germany, 2005; pp 1−67. DOI: 10.1007/b136816. (3) Mäki-Arvela, P.; Salmi, T.; Holmbom, B.; Willför, S.; Murzin, D. Y. Synthesis of Sugars by Hydrolysis of Hemicelluloses- A Review. Chem. Rev. 2011, 111 (9), 5638. (4) Doiseau, A.-C.; Rataboul, F.; Burel, L.; Essayem, N. Synergy Effect between Solid Acid Catalysts and Concentrated Carboxylic Acids Solutions for Efficient Furfural Production from Xylose. Catal. Today 2014, 226, 176. (5) Molina, M. J. C.; Granados, M. L.; Gervasini, A.; Carniti, P. Exploitment of Niobium Oxide Effective Acidity for Xylose Dehydration to Furfural. Catal. Today 2015, 254, 90. (6) Kerdi, F.; Ait Rass, H.; Pinel, C.; Besson, M.; Peru, G.; Leger, B.; Rio, S.; Monflier, E.; Ponchel, A. Evaluation of Surface Properties and Pore Structure of Carbon on the Activity of Supported Ru Catalysts in J


Research Article

ACS Sustainable Chemistry & Engineering (27) Sluiter, A.; Hames, B.; Ruiz, R.; Scarlata, C. Determination of Sugars, Byproducts, and Degradation Products in Liquid Fraction Process Samples; NREL Protocol TP-510-424623, 2006. (28) Xue, Z.; Ma, M.-G.; Li, Z.; Mu, T. Advances in the Conversion of Glucose and Cellulose to 5-Hydroxymethylfurfural over Heterogeneous Catalysts. RSC Adv. 2016, 6 (101), 98874. (29) Kumar, R.; Wyman, C. E. The Impact of Dilute Sulfuric Acid on the Selectivity of Xylooligomer Depolymerization to Monomers. Carbohydr. Res. 2008, 343 (2), 290. (30) Patil, S. K. R.; Heltzel, J.; Lund, C. R. F. Comparison of Structural Features of Humins Formed Catalytically from Glucose, Fructose, and 5-Hydroxymethylfurfuraldehyde. Energy Fuels 2012, 26 (8), 5281. (31) Zhou, L.; Shi, M.; Cai, Q.; Wu, L.; Hu, X.; Yang, X.; Chen, C.; Xu, J. Hydrolysis of Hemicellulose Catalyzed by Hierarchical H-USY Zeolites − The Role of Acidity and Pore Structure. Microporous Mesoporous Mater. 2013, 169, 54. (32) Enslow, K. R.; Bell, A. T. The Kinetics of Brønsted AcidCatalyzed Hydrolysis of Hemicellulose Dissolved in 1-Ethyl-3Methylimidazolium Chloride. RSC Adv. 2012, 2 (26), 10028.

K


Hydrolysis of Cellobiose and Xylan over TiO2-Based Catalysts - ACS

Recommend Documents