Kinetic Modeling of Brewery’s Spent Grain Autohydrolysis
Florbela Carvalheiro,† Gil Garrote,‡ Juan Carlos Parajo,‡ Helena Pereira,§ andFrancisco M. Gırio*,†
Departamento de Biotecnologia, INETI, Estrada do Paco do Lumiar 22, 1649-038 Lisboa, Portugal, Universidadede Vigo-Ourense, As Lagoas, 32004 Ourense, Spain, and Centro de Estudos Florestais, Instituto Superior deAgronomia, Universidade Tecnica de Lisboa, 1349-017 Lisboa, Portugal
Isothermal autohydrolysis treatments of brewery’s spent grain were used as a methodfor hemicellulose solubilization and xylo-oligosaccharides production. The time courseof the concentrations of residual hemicelluloses (made up of xylan and arabinan) andreaction products were determined in experiments carried out at temperatures in therange from 150 to 190 °C using liquid-to-solid ratios of 8 and 10 g/g. To model theexperimental findings concerning to brewery’s spent grain autohydrolysis severalkinetic models based on sequential pseudo-homogeneous first-order reactions weretested. Xylan and arabinan were assumed to yield oligosaccharides, monosaccharides(xylose or arabinose), furfural, and other decomposition products in consecutive reactionsteps. The models proposed provide a satisfactory interpretation of the hydrolyticconversion of xylan and arabinan. An additional model merging the two proposedmodels for xylan and arabinan degradation assuming that furfural was formed fromboth pentoses was developed and the results obtained are discussed. The dependenceof the calculated kinetic coefficients on temperature was established using Arrhenius-type equations.
Introduction
The conversion of biomass using a nonwaste approachis one of the most important ways for upgrading ligno-cellulosic residues and the major goal of the biorefineryconcept. Among the existing biomass fractionation tech-nologies, autohydrolysis is a mild, economic and envi-ronmentally friendly process, suitable for the selectivefractionation of hemicelluloses (1, 2).
During the reaction of lignocellulosic biomass withwater (autohydrolysis process), the hydrolysis of hemi-celluloses proceeds through hydronium-catalyzed reac-tions. In the first reaction stages, the hydronium ionsgenerated from water ionization lead to both the depo-lymerization of the hemicelluloses by the selective hy-drolysis of heterocyclic ether bonds and the cleavage ofacetyl groups. In further reaction stages, the hydroniumions generated from the autoionization of acetic acid alsoact as catalysts, improving reaction kinetics (3). Depend-ing on the operational conditions, the main productsobtained by autohydrolysis are a mixture of oligosaccha-rides, monosaccharides, acetic acid, and sugar degrada-tion products (e.g., furfural, hydroxymethylfurfural),which can undergo further decomposition reactions.
The sort of oligosaccharides obtained depends largelyon the source of the lignocellulosic material, since hemi-celluloses are a heterogeneous family of different poly-meric structures. For example, xylo-type oligosaccharidesare produced from the autohydrolysis of hardwoods andagricultural residues, although different xylo-oligosac-
charides (XOS) can be obtained. In nonwoody materialssuch as Gramineae the main monosaccharides of hemi-celluloses are xylose and arabinose. Uronic acids alsoappear in small amounts as substituents of xylose units,and some of the arabinose or xylose residues are substi-tuted with acetyl groups (4-6). In contrast, hardwoodxylan is almost devoid of arabinose, are much moreacetylated and with higher uronic acid content, the latterconsisting of 4-O-methylglucuronic acid (7). Therefore,XOS substituted with arabinose are obtained from thehydrolysis of the first type of lignocellulosic residues,whereas XOS substituted with methylglucuronic acid andacetyl groups are mostly obtained from hardwoods (8).
Xylo-oligosaccharides are usually considered to benondigestible oligosaccharides and enhance growth ofbifidobacteria in the large bowel, which may affect thehuman gastrointestinal tract beneficially (9-11). Inaddition, the stability over a wide range of temperaturesand pH make them useful for a variety of applicationsin the functional food market (12).
Previous studies of the production of XOS by autohy-drolysis of brewery’s spent grain (BSG) had shown thatseveral oligosaccharide mixtures of different molecularweight distributions were obtained depending on tem-perature and reaction time (8,13). Longer reaction timesled to a decreased amount of oligosaccharides and anincrease of the concentration of monosaccharides, aceticacid and sugar decomposition products (13).
However, to develop technical and economical studiesfor XOS production from BSG, additional information isneeded. As the development of rigorous equations de-scribing mass transfer and catalytic chemical reactionsin the heterogeneous autohydrolysis media is very com-plex (14), simplified approaches for mathematical inter-pretation of hemicellulose hydrolysis have been proposed.
* To whom correspondence should be addressed. Phone: +351210924721. Fax: +351 217163636. E-mail: [email protected].
† INETI.‡ Universidade de Vigo-Ourense.§ Universidade Tecnica de Lisboa.
These include either the use of severity parameters (15)or models based on sequential reaction steps with pseudo-first-order kinetics (16, 17). Since the understanding ofthe autohydrolysis kinetics is a key factor for processimprovement, the later modeling approach should bepreferably chosen because it has the advantage tohighlight the possible mechanisms involved.
Kinetic studies for the hydrolysis of hemicellulose haveshown that the solubilization of this polymer, in aqueousor acid-containing media, exhibits different hydrolysisrates: after the hydrolysis of a large fraction, there wasanother fraction that was slowly hydrolyzed. Therefore,the proposal of Kobayashi and Sakai (18) that xylan inlignocellulosic biomass is composed of two fractions withdifferent reactivity toward hydrolysis has been followedby other authors in works dealing with kinetic modeling(19-23).
In contrast to acid hydrolysis, autohydrolysis is carriedunder milder operational conditions (no acid addition)and as a consequence, the concentrations of oligosaccha-rides obtained are much higher. Also, the molecularweight of solubilized oligosaccharides is progressivelyreduced along autohydrolysis, leading to an increase oflow molecular weight oligosaccharides and monosaccha-rides in the late stages (16). For this reason, somereaction models for autohydrolysis have simplified thisby considering two consecutive reactions leading to high-and low-molecular oligosaccharides (17, 24). On thecontrary, the kinetic models for acid prehydrolysis as-sume that the oligosaccharides breakdown occurs muchmore quickly than their formation, so that the step ofoligosaccharides formation is omitted with little loss ofaccuracy (22, 25).
Pseudo-homogeneous kinetic models have been suc-cessfully applied to the kinetic modeling of autohydrolysisof hardwoods and agricultural by-products, in whichxylose is the major hemicellulosic structural unit (16, 17,20, 24). However, the kinetic study of hemicelluloses asheteropolymers is not common, although some non-woodymaterials may contain large quantities of arabinose,which makes difficult the mathematical interpretationof the data.
Considering the interest of extending the kineticmodeling to lignocellulosic materials with a relativelyhigh content of arabinose, and to provide a reliablekinetic interpretation of BSG autohydrolysis, in thisstudy the experimental data concerning to time courseof hemicellulose components (xylan and arabinan) andtheir reaction products were fitted to three differentkinetic models based on sequential pseudo-homogeneousfirst-order reactions.
Materials and MethodsFeedstock Preparation. BSG was supplied from a
local brewery (Sociedade Central de Cervejas, SA, Via-longa, Portugal). The spent grains were pretreated withwater in autoclave at 100 °C for 1 h using a liquid-to-solid ratio (LSR) of 8 g/g to remove the residual starch(final concentration <1% w/w). The solid was recoveredby filtration (filter paper Whatman no. 1), washed anddried at 50 °C until the moisture content less than 10%(w/w) has been obtained. The pretreated material wasscreened (>0.5 mm retained) and stored in PA/PEvacuum-sealed bags.
The BSG used as feedstock in this work had thefollowing average composition (dry mass basis): 18.8%glucan, 21.3% xylan, 10.5% arabinan, 21.7% Klasonlignin, 1.0% acetyl groups, 23.6% protein, 0.85% ash, and2.25% others (by difference).
Autohydrolysis. A stainless steel reactor (Parr In-struments Company, Moline, Illinois, USA) with a totalvolume of 0.6 L was used for autohydrolysis of pretreatedBSG. The reactor was fitted with two four-blade turbineimpellers, heated by an external fabric mantle and cooledby cold water circulating through an internal stainlesssteel loop. Temperature was controlled through a PIDcontroller, model 4842 (Parr Instruments Company,Moline, Illinois, USA). The feedstock material and waterwere mixed in the reactor in order to obtain the desiredLSR, taking into account the moisture content of thesample. The reactor was heated for 18-27 min to reachthe desired temperature, and stirred at 150 rpm. Timezero was set at beginning of the isothermal reactionstage. Nine to 10 experiments were carried out atdifferent reaction times (up to 9 h) for each pair ofconditions (temperature and LSR). The temperaturesstudied were 150, 170, and 190 °C for LSR of 8 g/g and10 g/g, respectively. The maximum reaction times wereselected to allow a complete observation of the time-course of oligosaccharide concentrations.
After the reaction time had elapsed, the reactor wasrapidly cooled and the liquor and solid phase were recov-ered by filtration (filter paper Whatman no. 1). The solidphase was washed with water, dried at 40 °C and theyield and composition were determined as describedbelow.
Analytical Methods. Feedstock material and pro-cessed solids were ground in a knife mill (<0.5 mm) andanalyzed for glucan, xylan, arabinan, and acetyl groupsby quantitative acid hydrolysis according to standardmethods (26). The acid insoluble residue was consideredas Klason lignin, after correction for ash. The monosac-charides and acetic acid in hydrolyzates were analyzedby HPLC as described elsewhere (13).
Protein content in the feedstock was estimated by theKjeldahl method (27) using the N × 6.25 conversionfactor. Starch was enzimatically determined using a testkit supplied by Boheringer.
Moisture content of the samples was determined byoven-drying at 105 °C to constant weight.
Liquors were centrifuged and filtered through 0.45 µmmembranes and used for direct HPLC determination ofmonosaccharides, furfural and acetic acid. A secondsample of liquors was subjected to quantitative acidhydrolysis with 4% (w/w) H2SO4 at 121 °C for 60 min,before HPLC analysis. XOS and arabino-oligosaccharides(AOS) concentrations were calculated from the increasesin the concentrations of xylose and arabinose, as analyzedby HPLC, after liquor hydrolysis.
Fitting of Data. The experimental data were fittedto the kinetic models by minimization of the sum ofsquares using commercial software with a built-in opti-mization routine dealing with the Newton’s method.
Definition of Variables. The nomenclature of thedependent variables selected to follow the autohydrolysisprocess is as follows: SY is the solid yield (g of solidrecovered after treatments per 100 g feedstock, oven drybasis), Xn and Arn are the percentages of xylan andarabinan in the processed solids (g per 100 g oven dryprocessed solids), XnFS and ArnFS are the percentages ofxylan and arabinan in feedstock material (g per 100 goven dry feedstock). XOS, AOS, Xyl, Ara, and F are theconcentrations in the liquid phase of xylo-oligosaccharides(expressed as xylose equivalent), arabino-oligosaccharides(expressed as arabinose equivalent), xylose, arabinoseand furfural, respectively (g/L).
The percentage of xylan and arabinan remaining in theprocessed solids (XnR and ArnR) and the percentages of
234 Biotechnol. Prog., 2005, Vol. 21, No. 1
feedstock xylan and arabinan converted into the corre-sponding oligosaccharides (XOSR and AOSR), xylose(XylR), arabinose (AraR) and furfural (FR) were calculatedusing the eqs 1-7, where, WL and WFS are the weightsof liquor and feedstock material (g), respectively. Theterms (132/150) and (132/96) are the stoichiometricfactors for the conversion of xylose into xylan (or arabi-nose into arabinan) and furfural into xylan, respectively.
Results and DiscussionKinetic Modeling of Hemicellulose Autohydroly-
sis. The autohydrolysis process was followed using aseries of dependent variables as previously defined inmaterials and methods section. On the basis of theexperimental data, three pseudo-homogeneous kineticmodels were tested to describe the hydrolysis of BSGhemicelluloses and Figure 1 shows the mechanismscorresponding to models 1 to 3.
The hemicellulose fraction of BSG, corresponds to aâ-D-(1,4)-linked xylopyranosyl backbone, mainly substi-tuted with arabinose at 2-O and/or 3-O positions. Incomparison with hardwoods, BSG has limited degrees ofacetylation and substitution by uronic acids and arabi-nose represents approximately half of the xylose in thehemicellulose fraction (8).
Considering the relatively high content of arabinosein BSG hemicellulose, and for the purpose of the calcula-tions, xylan and arabinan were considered as two sepa-rate polymers whose hydrolyses proceeded indepen-dently. The main assumption considered in the severalmodeling approaches was that xylan is made up of twofractions, one being unreactive under the operationalconditions assayed and the other yielding XOS. Thehydrolyzable fraction of xylan (XnS) is related to the totalxylan of feedstock material (XnFS) by the “susceptiblefraction” measured by the parameter R (0 < R < 1). Thehydrolyzable xylan fraction is first degraded to high-molecular weight XOS (XOSH), which are hydrolyzed tolow-molecular weight XOS (XOSL), and then convertedinto xylose (Xyl). Conversely to xylan, it was assumedthat the whole arabinan was reactive. Arabinan is firstdegraded to arabino-oligosaccharides (AOS) and these arefurther converted to arabinose (Ara).
Model 1. In addition to the reaction mechanismproposed above leading from xylan to xylose, in this
model, xylose was assumed to yield furfural (F), whichin turn is degraded to decomposition products (DP)(Figure 1, model 1). The reaction step leading fromfurfural to degradation products was introduced becausethe decomposition of furfural was evident from materialbalances concerning both xylan and xylan-degradationproducts.
By analytical integration of the kinetic model derivedfrom the mechanism proposed, it can be obtained:
Decomposition products are calculated by a materialbalance to the xylan-derived products
The integration constants can be calculated by the eqs14-29, where the subscript 0 corresponds to the concen-tration of species at the beginning of the isothermalreaction stage (t ) 0)
The parameter â corresponds to the soluble fraction ofunreacted xylan at the beginning of the isothermal stage(t ) 0), which is related to the susceptible fraction, R, bythe equation
Since XOSH and XOSL were experimentally measuredtogether, as xylose equivalents, their joint contributionwas measured by XOSR
Model 2. The model proposed for arabinan hydrolysis(Figure 1, model 2) assumes that arabinan is firstdegraded to arabino-oligosaccharides (AOS), which inturn are converted into arabinose (Ara), being the laterconverted into decomposition products (DP).
In a similar approach to the one considered for model1, the analytical solutions and stoichiometric conditionsderived from the proposed mechanism for arabinanhydrolysis leads to the following equations:
The integration constants can be calculated by eqs 36-41:
Model 3. The kinetic models reported for hemicellulosehydrolysis which include furfural formation usuallyassume that all furfural comes from xylose (17, 20, 25).However, considering the importance of arabinose in thehemicellulosic fraction of BSG, furfural formation fromarabinose should also be considered in this case. There-fore, a third model based on furfural production from bothpentoses was proposed (Figure 1, model 3). As referredfor model 1, model 3 also assumes that furfural canundergo decomposition reactions.
Considering the similarities between the various mod-els, it can be seen that the percentage of xylan andarabinan remaining in solid phase after treatments aswell as the percentages of feedstock xylan and arabinanconverted into the corresponding oligosaccharides andmonosaccharides for model 3 can be calculated by thesame equations already described for models 1 and 2 (eqs8-11 and 32-34, respectively). Furfural can be calcu-lated using eq 42, whereas decomposition products areobtained by eq 43, a material balance to the xylan- andarabinan- derived products.
In this model, the recoveries of furfural and decomposi-tion products were expressed as a percentage of feedstockarabinoxylan (xylan + arabinan). The superscript*, in eqs42 and 43 refers to xylan and arabinan remaining in thesolid after treatments or converted into oligosaccharides,monosaccharides, and furfural, as a percentage of feed-stock arabinoxylan.
In this case, the parameter â is related to the suscep-tible fraction, R, by the equation
where 67.0 is the relative percentage of xylan in thefeedstock arabinoxylan.
The integration constants C1-C15 and C17-C22 have thesame meaning as in models 1 and 2 (eqs 14-28 and 36-41, respectively). The other integration constants can becalculated by the eqs 45-48
Kinetic Coefficients. Figures 2-6 show the experi-mental values and the calculated data for xylan andarabinan and the corresponding hydrolysis products forthe experiments performed at 150, 170, and 190 °C, usingliquid-to-solid ratios of 8 and 10 g/g. The values obtainedfor the regression parameters and the statistical coef-ficient R2 in models 1-3 are presented in Tables 1-3.
The close agreement between experimental and cal-culated data for model 1, suggests the suitability of thismodel for a quantitative interpretation of the experimen-tal results (Figure 2, Table 1). As expected, the valuesobtained for kinetic parameters increase with tempera-ture. The exception is the high value for k3 obtained at150 °C and LSR of 10 g/g.
The results obtained for R are in the range from 0.708to 0.886, even though a slight tendency to an increasewith temperature was observed. The average, 0.813, isin agreement with other R values previously obtained for
Figure 2. Experimental (symbols) and calculated (lines) time courses for residual xylan and xylan hydrolysis products using model1. (a) T ) 150 °C, LSR ) 8 g/g; (b) T ) 170 °C, LSR ) 8 g/g; (c) T ) 190 °C, LSR ) 8 g/g; (d) T ) 150 °C, LSR ) 10 g/g; (e) T ) 170°C, LSR ) 10 g/g; (f) T ) 190 °C, LSR ) 10 g/g. XnR: (9, s); XOSR: (2, - - -); XylR: (b, - - - -); FR: ([, - ‚ - ‚ -); DP: (/, -‚‚-‚‚-).
the autohydrolysis of Eucalyptus globulus wood, 0.813-0.855 (17,28) and corncob, 0.826-0.842 (24).
The reaction rates obtained for the conversion of xylaninto high-DP XOS (k1), at the highest temperatureassayed (190 °C) were of similar magnitude to thosereported for the autohydrolysis of corncob but substantiallower to the described for the autohydrolysis of E.globulus wood at the same temperature (17, 24). Also,the rates obtained in this work for the hydrolysis of low-DP XOS into xylose were lower than the reported for E.globulus, and may explain the low xylose concentrationsactually obtained, which could be important to attain ahigh recovery of XOS. The differences observed for thosekinetic parameters, emphasize the role of hemicellulose/oligosaccharides structure on the hydrolysis. Compara-tively, more complex structures have been described forxylan and XOS obtained from E. globulus than from BSG,the former with higher uronic acids and acetyl groupscontent (8, 29). These compounds could play a double rolein the hydrolysis once the detached acetic acid can act
as catalyst although both have been associated to theslow of the hydrolysis rates because of the stabilizationthey bring to the glycosidic bonds (30).
In a similar way to model 1, the model 2 predictionsare in good agreement with the experimental data,confirming that the proposed kinetic model and associ-ated kinetic parameters are valid for data interpretation(Figure 3, Table 2). The hydrolysis of the “polymer”occurred faster than the corresponding oligomers attemperatures of 150 and 170 °C. Conversely, for thehighest temperature, higher rates of oligosaccharideshydrolysis were obtained. The chemical nature of arabi-nan, which consists of branches in xylopyranosyl back-bone, would make it more susceptible to the hydrolysisthan xylan polymer. This is evident by the comparativelyshorter reaction times needed to reach the highestconcentrations of arabinose oligomers in comparison withxylo-oligomers. In addition, a considerable decrease inoligomers to arabinose was observed, which in turn alsodecrease for the longer reaction times, confirming the
Figure 3. Experimental (symbols) and calculated (lines) time courses for residual arabinan and arabinan hydrolysis products usingmodel 2. Experiments (a) to (f) as in the caption of Figure 2. ArnR: (0, s); AOSR: (4, - - -); AraR: (O, - - - -); DP: (×, -‚‚-‚‚-).
238 Biotechnol. Prog., 2005, Vol. 21, No. 1
participation of parasitic reactions leading to sugarconsumption in the overall reaction scheme.
Model 3 was used to highlight the role of arabinosedegradation products on the hydrolysis of BSG hemicel-lulose. Since it assumes the same reaction steps asmodels 1 and 2 for the hydrolysis of xylan and arabinanto xylose and arabinose, respectively, the values obtainedfor the several parameters involved were similar to thoseobtained in the previous cases. However, the kineticparameter k5 controlling the dehydration of pentoses(xylose and arabinose) to furfural, was dramaticallychanged (Table 3). In this model, an increase in devia-tions of experimental from calculated results is particu-larly evident for furfural and in some cases for xylose.Even though this mechanism seems to be the mostrealistic one, the interpretation obtained for furfuralconcentrations is poor. Improved fitting for furfuralconcentration was obtained when the experimental datacorresponding to longer reaction times were omitted (datanot shown), suggesting that additional reactions (not
included in the mechanism) can take place under harshoperational conditions. In fact, even under the mildconditions of the autohydrolysis, many different byprod-ucts are obtained, originated from sugar degradationreactions or due to some hydrolysis of lignin. In addition,these compounds can undergo further decomposition and/or condensation reactions, the most common reported ofthe latter involving sugars and sugar-derived byproductsand lignin to give insoluble products (2, 31-33) whichbecome important at high severities. Such additionalreactions were not integrated in the kinetic model sincethose insoluble products are ill defined and difficultlyquantifiable. Nevertheless, models 1 and 2 could providea satisfactory interpretation of the experimental resultseven without considering such additional reaction steps.
Contrarily to the situation reported for acid prehy-drolysis (34), the rate of xylan hydrolysis for models 1and 3 was higher than the one calculated for theconversion of high-DP into low-DP XOS. This typicaldifference between these processes, due to the availability
Figure 4. Experimental (symbols) and calculated (lines) time courses for residual xylan, xylo-oligosaccharides and xylose usingmodel 3. Experiments (a) to (f) as in the caption of Figure 2. XnR: (9, s); XOSR: (2, - - -); XylR: (b, - - - -).
Biotechnol. Prog., 2005, Vol. 21, No. 1 239
of catalyst, leads to a reduced rate of oligosaccharideshydrolysis, enabling the recovery of XOS (which have acomparatively high added-value) at relatively high yields.The models predicted up to 53-61% of xylan recovery asXOS, with a deviation from experimental values below5%. Higher temperatures resulted in slightly increasedXOS recovery. However, the LSR seems not to affect therecovery obtained, excepting the experiments performedat 170 °C, where the conversion of xylan into XOS wasslightly increased for the higher value of this parameter.The maximum xylan conversion into XOS is within therange reported in the literature for typical xylan-contain-ing raw materials (13, 16, 17, 35).
Activation Energies. The dependence of the kineticcoefficients on temperature has been established bymeans of the Arrhenius equation. Tables 4-6 show thepreexponential factors, activation energies and the valuesof R2 determined using models 1, 2, and 3, respectively.Excepting the hydrolysis of XOS, no evident effects of theliquid-to-solid ratio on preexponential factors and activa-
tion energies were observed for all the models presented,even though these parameters could not be calculated forthe hydrolysis of low-DP XOS at the LSR of 10 g/g.
The values of activation energy obtained for the hy-drolysis of xylan, ranged from 96 to 108 kJ/mol, whichare slightly lower than the typical values reported in theliterature for the degradation of hardwood xylan andsome agricultural byproducts using hydrothermal treat-ments. Activation energies for the degradation of hard-wood xylan in the range of 103-146 kJ/mol have beenreported (16, 17, 28, 36) depending on the assumption ofthe models (xylan made up of one or two fractions). Thevalues reported for corncob autohydrolysis lie in the samerange (24).
It is worth noting that the values of activation energiesobtained for the degradation of arabinan were onlyslightly lower than those obtained for xylan. The valuesobtained for the hydrolysis of the corresponding oligosac-charides were lower than those reported for the noniso-thermal autohydrolysis of rice husks (37).
Figure 5. Experimental (symbols) and calculated (lines) time courses for residual arabinan, arabino-oligosaccharides and arabinose,using model 3. Experiments (a) to (f) as in the caption of Figure 2. ArnR: (0, s); AOSR: (4, - - -); AraR: (O, - - - -).
240 Biotechnol. Prog., 2005, Vol. 21, No. 1
The activation energies for the dehydration of xyloseto furfural, which can be considered independent offeedstock material, ranged from 122 to 128 kJ/mol, lyingin the range previously reported for autohydrolysisexperiments (16, 17, 24, 28).
ConclusionHydrothermal processing of BSG causes the solubili-
zation of hemicellulose, leading to oligosaccharides as themajor products obtained. The hydrolysis of the mainhemicellulosic components (xylan and arabinan) wasdescribed using kinetic models based on sequentialpseudo-homogeneous first-order reactions. For this pur-pose, three different kinetic models were compared. The
Figure 6. Experimental (symbols) and calculated (lines) time courses for furfural and decomposition products using model 3.Experiments (a) to (f) as in the caption of Figure 2. FR*: ([, - ‚ - ‚ -); DP*: (/, -‚‚-‚‚-).
Table 1. Values of Kinetic Coefficients, Parameter r andR2 Obtained for the Hydrolysis of BSG Xylan UsingModel 1
first two models were developed for the individualhydrolysis of xylan and arabinan, respectively, while forthe third one, the formation of furfural was assumed tobe consequence of the dehydration of both xylose andarabinose generated during autohydrolysis.
The close agreement between experimental and cal-culated data for models describing the hydrolysis of xylanand arabinan, suggests the suitability of these modelsfor a quantitative interpretation of the experimentalresults. The arabinoxylan model (model 3), even thoughpresenting a more rational mechanism for the autohy-drolysis of BSG hemicellulose lacks in accuracy forpredicting the furfural concentrations.
The temperature dependence of the kinetic coefficientscalculated by the models was established using Arrhe-
nius-type equations, and both preexponential factors andactivation energies were calculated for the several kineticcoefficients involved in the models, providing a theoreticalbackground for further process analysis namely for fine-tuning of the operational conditions and scale-up.
AcknowledgmentThe authors are grateful to the European Commission
for the financial support (Project FP4-FAIR-CT98-3811).L.C. Duarte is gratefully acknowledged for the criticalreading of the manuscript.
References and Notes(1) Mok, W. S. L.; Antal, M. J. Uncatalyzed solvolysis of whole
biomass hemicellulose by hot compressed liquid water. Ind.Eng. Chem. Res. 1992, 31 (4), 1157-1161.
(2) Garrote, G.; Domınguez, H.; Parajo, J. C. Hydrothermalprocessing of lignocellulosic materials. Holz Roh- Werkst.1999, 57 (3), 191-202.
(3) Heitz, M.; Carrasco, F.; Rubio, M.; Chauvette, G.; Chornet,E.; Jaulin, L.; Overend, R. P. Generalized correlations for theaqueous liquefaction of lignocellulosics. Can. J. Chem. Eng.1986, 64 (4), 647-650.
(4) Wilkie, K. C. B. The hemicelluloses of grasses and cereals.Adv. Carbohydr. Chem. Biochem. 1979, 36, 215-264.
(5) Ishii, T. Acetylation at O-2 of arabinofuranose residues inferuloylated arabinoxylan from bamboo shoot cell-walls.Phytochemistry 1991, 30 (7), 2317-2320.
(6) Wende, G.; Fry, S. C. O-feruloylated, O-acetylated oligosac-charides as side-chains of grass xylans. Phytochemistry 1997,44 (6), 1011-1018.
(7) Puls, J.; Poutanen, K. Mechanisms of enzymic hydrolysisof hemicelluloses (xylans) and procedures for determinationof the enzyme activities involved. In Enzyme Systems forLignocellulosic Degradation; Coughlan, M. P., Ed.; ElsevierApplied Science: London, 1989; pp 151-165.
(8) Kabel, M. A.; Carvalheiro, F.; Garrote, G.; Avgerinos, E.;Koukios, E.; Parajo, J. C.; Gırio, F. M.; Schols, H. A.; Voragen,A. G. J. Hydrothermally treated xylan rich by-products yielddifferent classes of xylo-oligosaccharides. Carbohydr. Polym.2002, 50 (1), 47-56.
(9) Campbell, J. M.; Fahey, G. C.; Wolf, B. W. Selected indigest-ible oligosaccharides affect large bowel mass, cecal, and fecalshort-chain fatty acids, pH and microflora in rats. J. Nutr.1997, 127 (1), 130-136.
(10) Fooks, L. J.; Fuller, R.; Gibson, G. R. Prebiotics, probioticsand human gut microbiology. Int. Dairy J. 1999, 9 (1), 53-61.
Table 3. Values of Kinetic Coefficients, Parameter r andR2 Obtained for the Hydrolysis of BSG Xylan andArabinan Using Model 3
(11) Van Laere, K. M. J.; Hartemink, R.; Bosveld, M.; Schols,H. A.; Voragen, A. G. J. Fermentation of plant cell wallderived polysaccharides and their corresponding oligosaccha-rides by intestinal bacteria. J. Agri. Food Chem. 2000, 48 (5),1644-1652.
(12) Vazquez, M. J.; Alonso, J. L.; Domınguez, H.; Parajo, J. C.Xylooligosaccharides: manufacture and applications. TrendsFood Sci. Technol. 2000, 11, 387-393.
(13) Carvalheiro, F.; Esteves, M. P.; Parajo, J. C.; Pereira, H.;Gırio, F. M. Production of oligosaccharides by autohydrolysisof brewery’s spent grain. Bioresour. Technol. 2004, 91 (1), 93-100.
(14) Fengel, D.; Wegener, G. Wood. Chemistry, Ultraestructure,Reactions; Walter de Gruyter: Berlin, New York, 1984.
(15) Overend, R. P.; Chornet, E. Fractionation of lignocellulosicsby steam-aqueous pretreatments. Philos. Trans. R. Soc.London 1987, A321, 523-536.
(16) Conner, A. H.; Lorenz, L. F. Kinetic modeling of hardwoodprehydrolysis. Part III. Water and dilute acetic acid prehy-drolysis of southern red oak. Wood Fiber Sci. 1986, 18 (2),248-263.
(17) Garrote, G.; Domınguez, H.; Parajo, J. C. Mild autohy-drolysis: an environmentally friendly technology for xyloo-ligosaccharide production from wood. J. Chem. Technol.Biotechnol. 1999, 74 (11), 1101-1109.
(18) Kobayashi, T.; Sakai, Y. Hydrolysis rate of pentosan ofhardwood in dilute sulfuric acid. Bull. Agric. Chem. Soc. Jpn.1956, 20, 1-7.
(19) Mehlberg, R.; Tsao, G. T. Low liquid hemicellulose hy-drolysis of hydrochloric acid. Presented at the 178th ACSNational Meeting, Washington, DC, 1979.
(20) Conner, A. H. Kinetic modeling of hardwood prehydrolysis.Part I. Xylan removal by water prehydrolysis. Wood FiberSci. 1984, 16 (2), 268-277.
(21) Kim, S. B.; Lee, Y. Y. Kinetics in acid-catalyzed hydrolysisof hardwood hemicellulose. Biotechnol. Bioeng. Symp. 1987,17, 71-84.
(22) Maloney, M. T.; Chapman, T. W.; Baker, A. J. An engineer-ing analysis of the production of xylose by dilute acid-hydrolysis of hardwood hemicellulose. Biotechnol. Prog. 1986,2 (4), 192-202.
(23) Parajo, J. C.; Vazquez, D.; Alonso, J. L.; Santos, V.;Domınguez, H. Prehydrolysis of Eucalyptus wood with dilutesulphuric acid: operation at atmospheric pressure. Holz Roh-Werkst. 1993, 51, 357-363.
(24) Garrote, G.; Domınguez, H.; Parajo, J. C. Kinetic modellingof corncob autohydrolysis. Process Biochem. 2001, 36 (6),571-578.
(25) Abatzoglou, N.; Koeberle, P. G.; Chornet, E.; Overend, R.P.; Koukios, E. G. Dilute acid-hydrolysis of lignocellulosics:an application to medium consistency suspensions of hard-
woods using a plug flow reactor. Can. J. Chem. Eng. 1990,68 (4), 627-638.
(26) Browning, B. L. In Methods of Wood Chemistry; Sarkanen,K. V., Ludwig, C. H., Eds.; John Wiley & Sons: New York,1967; Vol. 2, pp 795-798.
(27) AOAC Official Methods of Analysis, 11th ed.; AOACInternational: Washington, DC, 1975.
(28) Garrote, G.; Parajo, J. C. Non-isothermal autohydrolysisof Eucalyptus wood. Wood Sci. Technol. 2002, 36 (2), 111-123.
(29) Kabel, M. A.; Schols, H. A.; Voragen, A. G. J. Complex xylo-oligosaccharides identified from hydrothermally treated Eu-calyptus wood and brewery’s spent grain. Carbohydr. Polym.2002, 50 (2), 191-200.
(30) Belkacemi, K.; Abatzoglou, N.; Overend, R. P.; Chornet,E. Phenomenological kinetics of complex systemssmechanisticconsiderations in the solubilization of hemicelluloses followingaqueous steam treatments. Ind. Eng. Chem. Res. 1991, 30(11), 2416-2425.
(31) Heitz, M.; Capek-Menard, E.; Koeberle, P. G.; Gagne, J.;Chornet, E.; Overend, R. P.; Taylor, J. D.; Yu, E. Fractionationof Populus tremuloides at the pilot-plant scale: optimizationof steam pretreatment conditions using the STAKE II tech-nology. Bioresour. Technol. 1991, 35 (1), 23-32.
(32) Montane, D.; Farriol, X.; Salvado, J.; Jollez, P.; Chornet,E. Application of steam explosion to the fractionation andrapid vapor-phase alkaline pulping of wheat straw. BiomassBioenergy 1998, 14 (3), 261-276.
(33) Ramos, L. P. The chemistry involved in the steam treat-ment of lignocellulosic materials. Quim. Nova 2003, 26 (6),863-871.
(34) Carrasco; F.; Roy, C. Kinetic study of dilute-acid prehy-drolysis of xylan-containing biomass. Wood Sci. Technol.1992, 26, 189-208.
(35) Aoyama, M.; Seki, K.; Saito, N. Solubilization of bamboograss xylan by steaming treatment. Holzforschung 1995, 49(3), 193-196.
(36) Carrasco, F. C. E.; Overend, R. P. H. M. Fractionnementde deux bois tropicaux (Eucalyptus et Wapa) par traitementthermomecanique en phase aqueuse. Partie II: Caracteris-tiques chimiques des residus et considerations cinetiques surla solubilisation des hemicelluloses. Can. J. Chem. Eng. 1987,65, 71-77.
(37) Vila, C.; Garrote, G.; Domınguez, H.; Parajo, J. C. Hydro-lytic processing of rice husks in aqueous media: a kineticassessment. Collect. Czech. Chem. Commun. 2002, 67 (4),509-530.
