Bioresource Technology 149 (2013) 149–154

Contents lists available at ScienceDirect

Bioresource Technology

journal homepage: www.elsevier .com/locate /bior tech

Monomeric carbohydrates production from olive tree pruningbiomass: Modeling of dilute acid hydrolysis

0960-8524/$ - see front matter � 2013 Elsevier Ltd. All rights reserved.http://dx.doi.org/10.1016/j.biortech.2013.09.046

⇑ Corresponding author. Tel.: +34 953 212780; fax: +34 953 212140.E-mail address: ajmoya@ujaen.es (A.J. Moya).

Juan G. Puentes a, Soledad Mateo a, Bruno G. Fonseca b, Inês C. Roberto b, Sebastián Sánchez a,Alberto J. Moya a,⇑a Department of Chemical, Environmental and Material Engineering, University of Jaén, 23071 Jaén, Spainb Department of Biotechnology, College of Chemical Engineering of Lorena, P.O. Box 116, Lorena, São Paulo, Brazil

h i g h l i g h t s

� Olive tree pruning was converted to monomers by a one-step hydrolysis reaction.� Response surface methodology was applied for statistical modeling and optimization.� D-Xylose recovery of 85% was achieved at optimized conditions, confirming the model.� Low concentration of toxic substances provided a high quality D-xylose substrate.

a r t i c l e i n f o

Article history:Received 25 July 2013Received in revised form 9 September 2013Accepted 11 September 2013Available online 20 September 2013

Keywords:HemicelluloseD-XyloseD-GlucoseStatistical modelingBiorefinery

a b s t r a c t

Statistical modeling and optimization of dilute sulfuric acid hydrolysis of olive tree pruning biomass hasbeen performed using response surface methodology. Central composite rotatable design was applied toassess the effect of acid concentration, reaction time and temperature on efficiency and selectivity ofhemicellulosic monomeric carbohydrates to D-xylose. Second-order polynomial model was fitted toexperimental data to find the optimum reaction conditions by multiple regression analysis. Themonomeric D-xylose recovery 85% (as predicted by the model) was achieved under optimized hydrolysisconditions (1.27% acid concentration, 96.5 �C and 138 min), confirming the high validity of the developedmodel. The content of D-glucose (8.3%) and monosaccharide degradation products (0.1% furfural and0.04% 5-hydroxymethylfurfural) provided a high quality subtract, ready for subsequent biochemicalconversion to value-added products.

� 2013 Elsevier Ltd. All rights reserved.

1. Introduction

Currently, due to the rapid depletion of fossil resources,researches of alternative renewable energy sources, such asbiomass, is taking great interest. The conversion of different bio-mass feedstocks to fuel and other products, i.e., the biorefineryconcept of biomass processing, is being considered now as a morepotential way to guarantee sustainable bio-based economy (Kammand Kamm, 2007). The agro-based lignocellulosic materials, suchas industrial crop residues and various grasses represent an abun-dant and cheap feedstock for lignocellulosic feedstock biorefinery.

Among the largest agricultural crop waste generation in Spainmay be mentioned the wheat straw and especially the olive treepruning for its high concentration in southern. The olive tree isone of the most important crops in Spain. The pruning, operationthat is usually applied to the branches and leaves after harvest,

generates 5:0 � 106 � 5:5 � 106 t/year of lignocellulosic biomass(Moya et al., 2008). The large volumes of waste generated, togetherwith the great environmental damage caused by its uncontrolledburning has suggested the possibility of exploiting this biomassresource to produce oligosaccharides and monomeric carbohy-drates (Mateo et al., 2013a).

In agro-based biomass the proportion of xylan may amount to95% of the total non-cellulosic polysaccharides (Hurter, 1988).The monomeric D-xylose can be used as substrate for a wide vari-ety of products production, such as xylitol, a five-carbon sugaralcohol that has attracted much attention because of its potentialuse in food and pharmaceutics (as a natural food sweetener, dentalcaries reducer, sugar substitute for diabetics, thin coating oftablets) (Granstrom and Leisola, 2009). Xylan isolation and depoly-merization to D-xylose can therefore be an important first step inthe complex biorefinery scheme.

The dilute sulfuric acid hydrolysis under moderate reactionconditions was proved to be a reliable and easily performed lowcost method for quantitative conversion of hemicellulosic xylan

150 J.G. Puentes et al. / Bioresource Technology 149 (2013) 149–154

to monomeric sugars. Hemicellulose hydrolysis of different ligno-cellulosic materials by dilute sulfuric acid solutions has been re-ported, rice straw (Karimi et al., 2006;Roberto et al., 2003),sugarcane bagasse (Rodrigues et al., 2010), sunflower stalks (Duet al., 2012), Eucaliptus wood (Gutsch et al., 2012), cauliflowermushroom (Lee et al., 2013), corn fiber (Noureddini and Byun,2010) or triticale, barley, oats, canola and mustard straws (Pronykand Mazza, 2012). The results showed that the amount of sugarsreleased during the hydrolysis treatment is dependent on the typeof raw material used and operational conditions (reaction time,temperature and acid concentration) applied for the hydrolysisreaction. The minimal monosaccharide decomposition to furansand cellulose degradation can be achieved under optimized condi-tions, providing high effectiveness and selectivity of the overallhydrolysis process.

D-Xylose production from olive tree pruning biomass has beenstudied using a low temperature dilute sulfuric acid hydrolysis.The response surface methodology (RSM) was employed for pro-cess modeling and optimization to maximize effectiveness andselectivity of xylan conversion to monomeric D-xylose withinone-step reaction.

2. Methods

2.1. Raw material and chemical

The olive tree pruning biomass, collected during the pruningseason, consisted of leaves, branches and pieces of trunks from ol-ive trees and was collected in an olive grove situated in Jaén, Spain.The material was air-dried, milled, screened to select the fractionof particles with a diameter 0.425–0.60 mm and homogenized ina single lot.

2.2. Dilute acid hydrolysis

Hydrolysis experiments (replicated for each condition set) werecarried out in a discontinuous reactor (2 dm3 volume) heated withsilicon V50 from a bath. For this study, the reactor was loaded with100 g (on dry basis) of olive tree pruning residue and 1 dm3 ofsulfuric acid solution. The process variables were reaction time(20–220 min), temperature (86–103 �C) and acid concentration(0.2–1.8 mass%). The heating-up period for each experiment wasaround 5 min. Solid residue after hydrolysis was separated fromsolution by vacuum-filtration. The collected hydrolyzate wasexamined on degree of monosaccharide recovery and degradationproducts formed. Xylan conversion after hydrolysis (Y1) was de-fined as a ratio of D-xylose content in hydrolyzate to hemicellulosiccontent in olive tree pruning raw biomass. Y2 was defined as a ratioof D-xylose to D-glucose in hydrolyzate.

2.3. Analytical methods

The quantification of carbohydrates (D-glucose, D-xylose and L-arabinose) as well as acetic acid concentrations (in order to esti-mate the acetyl groups content) were determined by high-perfor-mance liquid chromatography (HPLC) using a WATERSinstrument, in the conditions: a BIO-RAD Aminex HPX-87H (300� 7.8 mm) column at 45 �C, 0.005 M sulfuric acid as eluant, flowrate of 0.6 cm3 min�1, refraction index (RI) detector and 20 lL sam-ple volume.

Furfural and HMF were analyzed by HPLC using a WATERSinstrument with a UV detector (at 276 nm), in the following condi-tions: a Waters Resolve C18 5 lm (300 � 3.9 mm) column at ambi-ent temperature, acetonitrile/water (1/8 with 1% of acetic acid)degassed with addition of phosphoric acid for pH correction to

2.5 as eluant, flow rate of 0.8 cm3 min�1 and 20 lL sample volume.The samples were previously diluted with ultrapure water and fil-tered through membranes HAWP 04700 with 0.45 lm pores. Theconcentrations of these compounds were calculated from calibra-tion curves obtained from standard solutions.

To calculate the cellulose and hemicellulose percentage andKlason lignin, the methodology proposed by Irick et al. (1988)was utilized and moisture composition by the TAPPI norm T12os-75. Furthermore, the concentration of acid-soluble lignin wasdetermined by the method described in a previous work (Mateoet al., 2013b), and ash composition using the procedure establishedby Browning (1967).

Extractives (nonstructural components such as pectins, fattymatters, terpenes, phenols, tannins, uronic acids, etc.) were deter-mined gravimetrically using a two-step sequential extraction pro-cess by Soxhlet to remove water and ethanol soluble materialaccording to a procedure adapted from Sluiter et al. (2008).

2.4. Statistical modeling

Response surface methodology (RSM) was employed for statis-tical data treatment and optimization of hydrolysis conditions bymultiple regression analysis, using Statistica 6.0 (Statsoft, USA)software. The 23 central composite rotatable design (CCRD) withthree independent variables at five different levels, six star (axial)points and five central points (total 19 runs) was adopted to findlinear, quadratic and interaction effects of independent processvariables on experimental responses. A second-order polynomialmodel was fitted to each set of experimental data to predict opti-mal reaction conditions by the following equation:

Yz ¼ b0 þX3

i¼1

biXi þX3

i¼1

biiX2i þ

X3

i<j;j¼2

bijXiXj ð1Þ

where Y is a predicted response (xylan conversion or D-xylose/D-glucose ratio), b0 is an interception coefficient (regressioncoefficient at central point), bi are the linear coefficients; bii arethe quadratic coefficients, bij are the interaction coefficients, Xi

and Xj are the independent variables (temperature, time and acidconcentration).

The statistical significance of regression coefficients and effectswas checked by analysis of variance (ANOVA) using the softwareSTATISTICA 6.0 (Statsoft, USA).

3. Results and discussion

3.1. Compositional analysis of raw material

Detailed chemical analysis of olive tree pruning biomass used inthis study revealed some general features typical for other indus-trially important agro-crops and woody species. The hemicellulosecontent represent 18.63 � 0.27 % of dry matter.

The content of cellulose, as the principal chemical constituent,accounting for 33.85 � 0.76 of dry residue, does not differ greatlyfrom wheat straw (29–35%), bamboo (26–43%) and sugarcanebagasse (32–44%), but somewhat lower in comparison with woods(38–50%) (Hurter, 1988). However, it should be mentioned thatcellulose and hemicellulose contents depend on the methods usedfor the determination of these components. Olive tree pruning hasless lignin (23.13 � 0.04 % of dry matter, being 18.93 � 0.08acid-insoluble and 4.20 � 0.03 acid soluble) and more extractives(19.20 � 0.39 % of dry matter), in relation to woods (25–30% and1–5%, respectively) (Atchison, 1987). The particularly high propor-tion of water-soluble substances, (17.39 � 0.28 % of dry matter),indicates the high accessibility and therefore reactivity of this bio-mass during chemical processing. The minerals (ash) comprise

Table 1Experimental data on formation and decomposition of monomeric sugars during hydrolysis of olive tree pruning.

T (�C) t (min) c (%) Yield (% on dry material)

Glua Xyl Ara AcA F HMF

90 60 0.5 2.54 ± 0.08 0.77 ± 0.03 3.20 ± 0.07 1.74 ± 0.07 Traces Traces1.0 4.03 ± 0.09 3.66 ± 0.09 3.54 ± 0.06 2.61 ± 0.09 Traces Traces1.5 6.48 ± 0.14 8.73 ± 0.11 3.54 ± 0.07 2.94 ± 0.12 0.01 ± 0.00 Traces

180 0.5 5.58 ± 0.09 7.04 ± 0.08 3.50 ± 0.05 2.77 ± 0.05 0.01 ± 0.00 0.01 ± 0.011.0 9.01 ± 0.18 13.83 ± 0.21 3.78 ± 0.04 3.46 ± 0.07 0.06 ± 0.02 0.03 ± 0.011.5 10.12 ± 0.34 15.99 ± 0.31 3.92 ± 0.07 3.59 ± 0.04 0.13 ± 0.04 0.04 ± 0.01

95 60 0.5 3.42 ± 0.07 1.97 ± 0.04 3.23 ± 0.03 2.60 ± 0.07 0.03 ± 0.01 0.02 ± 0.001.0 6.88 ± 0.13 9.13 ± 0.09 3.13 ± 0.09 3.21 ± 0.06 0.03 ± 0.01 0.03 ± 0.011.5 10.50 ± 0.21 13.20 ± 0.17 3.30 ± 0.06 3.62 ± 0.08 0.05 ± 0.01 0.04 ± 0.01

180 0.5 7.49 ± 0.11 10.88 ± 0.24 3.65 ± 0.07 3.42 ± 0.06 0.03 ± 0.01 0.03 ± 0.011.0 10.53 ± 0.30 15.92 ± 0.26 3.28 ± 0.05 3.50 ± 0.06 0.08 ± 0.03 0.03 ± 0.01

12.75 ± 0.27 16.03 ± 0.20 3.40 ± 0.05 3.57 ± 0.05 0.16 ± 0.04 0.05 ± 0.01

100 60 0.5 4.56 ± 0.04 4.02 ± 0.05 3.27 ± 0.06 3.00 ± 0.07 0.01 ± 0.00 0.02 ± 0.001.0 8.77 ± 0.11 12.07 ± 0.16 3.21 ± 0.04 3.37 ± 0.06 0.02 ± 0.00 0.03 ± 0.011.5 10.31 ± 0.19 14.60 ± 0.37 2.97 ± 0.02 3.71 ± 0.05 0.06 ± 0.00 0.04 ± 0.01

180 0.5 11.80 ± 0.26 16.17 ± 0.29 3.28 ± 0.03 3.63 ± 0.04 0.11 ± 0.01 0.05 ± 0.011.0 9.96 ± 0.16 16.09 ± 0.27 2.40 ± 0.06 3.53 ± 0.07 0.12 ± 0.02 0.06 ± 0.021.5 10.76 ± 0.17 16.10 ± 0.32 2.36 ± 0.07 3.52 ± 0.05 0.13 ± 0.04 0.06 ± 0.01

a Glu, Xyl, Ara, AcA, F, HMF as D-glucose, D-xylose, L-arabinose, acetic acid, furfural and 5-hydroxymethylfurfural, respectively.

Table 2Range and levels of independent process variables, (X1: time (min); X2: temperature(�C); X3: acid concentration (%)), used in experimental design.

Variable Range and levels

�a �1 0 +1 +a

X1 19 60 120 180 221X2 86.6 90 95 100 103.4X3 0.2 0.5 1.0 1.5 1.8

J.G. Puentes et al. / Bioresource Technology 149 (2013) 149–154 151

4.62 � 0.28 % of dry matter, similar to wheat straw (4–9%), flax (2–5%), kenaf (2–5%) and substantially higher of wood species (Hurter,1988).

3.2. Effect of hydrolysis conditions on D-xylose formation anddegradation

Some series of preliminary hydrolysis experiments have beencarried out under variable conditions of sulfuric acid concentra-tion, c (0.5, 1.0 and 1.5%), temperature, T (90, 95 and 100 �C) andreaction time, t (60 and 180 min) to define the current levels(settings) of the independent process variables to be used in statis-tical experimental design for process modeling and optimization.

As can be seen from Table 1, although significant amount ofmonomeric D-xylose can be recovered in solution after dilute-acidhydrolysis, the xylose recovery (yield) is highly dependent of ap-plied reaction conditions. In general, increase in process severity(i.e., increase in acid concentration, temperature and duration),while accelerates xylan hydrolysis to D-xylose, intensifies substan-tially the secondary degradation reactions of monomeric sugars,thereby decreasing the final yield of this pentose in solution. Someother sugars, such as L-arabinose and D-glucose, particularly, areformed during dilute hydrolysis. Whereas L-arabinose take a partof heteroxylan structure, the presence of D-glucose is a result ofcellulose degradation, namely, of its less ordered (amorphous) por-tion having the same (or close) reactivity than hemicelluloses (Fen-gel and Wegener, 1989). Like the furans, D-glucose can haveharmful effect on subsequent D-xylose bioconversion, e.g., toxylitol. Hydrolysis selectivity should therefore assure the minimalconcentration of D-glucose units in the hydrolyzate as well as a richD-xylose substrate. Obviously, the temperature of 90 �C provides

more preserving conditions for the formed monosaccharides andonly traces of furfural can be detected in reaction solution, Table 1.Limited cellulose degradation and acetic acid formation (due tosplitting out of acid-liable acetic groups of heteroxylan) was alsoobserved. At the same time, the D-xylose recovery in solution (asa main objective) was low under this process temperature, point-ing to incomplete xylan conversion. By contrast, the drastic condi-tions of 100 �C caused substantial monosaccharide (basically D-xylose) degradation and cellulose depolymerization (up to 5.3%of furfural and 4% of D-glucose in solution), decreasing substan-tially D-xylose recovery after hydrolysis and lowering substratequality as a whole, particularly under elevated acid concentration.

Maximum D-xylose recovery (ca. 16% of dry material, or 98% oftotal D-xylose) was observed when olive tree pruning was hydro-lyzed at 95 �C for 180 min in 1.5% acid solution or at 100 �C for180 min. Under these conditions, the contents of furfural,5-hydroxymethylfurfural, acetic acid and D-glucose were found as0.13%� 0.02, 0.06%� 0.01, 3.56%� 0.05 and 11.3%� 1.2, respec-tively, pointing to fairly good quality of D-xylose hydrolyzate withlow concentration of inhibitors.

Since high values of D-xylose recovery were achieved (90% andmore), the tested ranges of the principal independent process vari-ables were used later in the statistical experimental design, tomaximize the reaction outputs.

3.3. Hydrolysis modeling and optimization

Statistical modeling and optimization of dilute sulfuric acidhydrolysis of olive tree pruning biomass was done using responsesurface methodology (RSM) (Myers et al., 2009). To optimize theeffect of the principal independent variables (reaction time (X1),temperature (X2) and acid concentration (X3)) on efficiency ofxylan conversion to D-xylose (Y1) and D-xylose/D-glucose ratio(Y2), the 23 central composite rotatable design (CCRD) was em-ployed. The current settings of process variables, Table 2 were de-fined based on results of the preliminary experiments, discussedabove. According to CCRD, the RSM experimental design matrixfor 3 coded independent variables at 5 levels each, with 6 star (ax-ial) points and 5 replicates at the central point (total 19 runs) wasdeveloped (Table 3) and the significant effects having the greatestimpact on reaction outputs (Y1 and Y2) were calculated usingexperimental data.

Table 3Central composite rotatable design (CCRD) applied for olive tree pruning hydrolysisand the corresponding experimental responses on xylan conversion (Y1) and D-xylose/D-glucose ratio (Y2) used for RSM modeling.

Run N� Coded variables Responses

X1 X2 X3 Y1 Y2

1 �1 �1 �1 0.05 0.302 �1 �1 +1 0.54 1.353 �1 +1 �1 0.25 0.884 �1 +1 +1 0.90 1.425 +1 �1 �1 0.57 1.666 +1 �1 +1 0.98 1.587 +1 +1 �1 0.99 1.388 +1 +1 +1 0.99 1.509 �a 0 0 0.13 0.8310 +a 0 0 0.98 1.4311 0 -a 0 0.61 1.4212 0 +a 0 0.95 1.5713 0 0 -a 0.35 1.0514 0 0 +a 0.99 1.4515(C) 0 0 0 0.85 1.5216(C) 0 0 0 0.86 1.5117(C) 0 0 0 0.85 1.5218(C) 0 0 0 0.86 1.5219(C) 0 0 0 0.87 1.52

152 J.G. Puentes et al. / Bioresource Technology 149 (2013) 149–154

The Pareto charts of standardized linear, quadratic and interac-tion effects of the independent process variables, sorted by theirabsolute magnitude in relation to the statistical significance p-levelof 0.05 shown that the effectiveness of xylan conversion to D-xy-lose was mainly affected by time and acid concentration (linear ef-fects) and in a lesser degree by reaction temperature andinteraction effect between time and acid concentration. The D-xy-lose/D-glucose ratio was also primarily controlled by time and acidconcentration (linear and quadratic effects) having no significanceeffects the temperature and its interactions. The statistical signifi-cance of estimated effects was checked by analysis of variance(ANOVA), Table 4. The low p-values of the main effects (p < 0.01)indicated high statistical significance of the estimated relations be-tween variables within a 99% confidence interval. To define theoptimum levels (conditions) of the independent process variables,the second-order polynomial model (Eq. (1)) was fitted to experi-mental data and the regression coefficients were calculated bymultiple regression analysis. Two model equations were obtainedusing more statistically significant regression coefficients(p<0.05):

Y1 ¼ 0:8579þ 0:2357X1 � 0:1067X21 þ 0:1144X2 þ 0:1923X3

� 0:0661X23 � 0:0913X1X3 ð2Þ

Table 4ANOVA of estimated linear (L), quadratic (Q) and interaction effects for xylan conversion

Factor Y1

E. e.a t-Test F-test

(1) Temperature (L) 0.4715 12.2627 150.3751Temperature (Q) � 0.2134 � 5.5498 30.7999(2) Time (L) 0.2287 5.9487 35.3874Time (Q) � 0.0543 � 1.4128 1.9961(3) Concentration (L) 0.3846 10.0034 100.0684Concentration (Q) � 0.1321 � 3.4353 11.80151L by 2L � 0.0325 � 0.6469 0.41851L by 3L � 0.1825 � 3.6329 13.19772L by 3L � 0.0625 � 1.2441 1.5479

a E. e. Estimated effect.

Y2 ¼ 1:5187þ 0:4656X1 � 0:2825X21 þ 0:0794X2 þ 0:3372X3

� 0:1977X23 � 0:2525X1X2 � 0:3875X1X3 ð3Þ

The regression coefficients obtained (R2 = 0.97 and R2adj = 0.95 for

Y1, and R2 = 0.94 and R2adj = 0.89 for Y2) and the sum of the squares

of the differences between the values of Yi and the average Yi

(Total SS) 1.756 for Y1 and 2.069 for Y2 confirm the goodness ofthe model.

Fig. 1 shows the 3D response surfaces and the correspondingcontour plots constructed on the basis of Eqs. (2) and (3) and illus-trate the modeled effects of independent variables on reaction out-puts. The response surfaces for xylan conversion (Fig. 1 top),having some maximum values (stationary points) near the centerpoint of the experimental design, allow locating and characterizingthe optimum responses. The effect of reaction temperature andtime on xylan conversion is illustrated in Fig. 1(a) (top). At fixedacid concentration set at 1% as a center point of statistical experi-mental design, the maximum D-xylose yield can be obtainedaround the maximum values of this variables. The desirable rangesof acid concentration and reaction time at constant temperature of95 �C set as a center point are particularly notable in Fig. 1(b) (top).It can be seen that a concentration range of 1.0–1.5% and time120–180 min maximize the D-xylose recovery during hydrolysis.Finally, Fig. 1(c) (top) shows that maximum reaction temperatureand acid concentration (100 �C and 1.5%, respectively), when thereaction time was kept constant at 120 min, were the best condi-tions to D-xylose recovery.

The response surfaces for D-xylose/D-glucose ratio (Fig. 1 bot-tom) are similar to those for the xylan conversion, with maximumvalues encountered at various combinations of independent vari-ables. As can be seen from Fig. 1(a) (bottom), keeping fixed acidconcentration at 1% as a center point, the maximum D-xylose/D-glucose ratio is obtained during prolonged reaction time(180 min) decreasing lighting when more time is employed. Simi-larly, at fixed time of 120 min, the higher values of Y2 are obtainedat 1.5% acid concentration, Fig. 1(c) (bottom). However, based onFig. 1(b) (bottom) it can be seen a maximum Y2 zone at acid con-centration range around 1.0% and time 120 min. This suggeststhe possibility of maximizing this response on the basis of ridgemaximum and canonical analysis.

Partial differentiation of the multivariate function described byEq. 3 was done to find a critical value of the independent processvariables for D-xylose/D-glucose ratio:

@Y2

@X1¼ 0:4656� 0:5651X1 � 0:2525X2 � 0:3875X3 ¼ 0 ð4Þ

@Y2

@X2¼ 0:0794� 0:2525X1 ¼ 0 ð5Þ

(Y1) and D-xylose/D-glucose ratio (Y2).

Y2

p E. e.a t-Test F-test p

0.0000 0.4143 11.3719 56.9628 0.00000.0004 � 0.2639 � 7.2413 20.9691 0.00000.0002 0.0282 0.7729 1.6574 0.45940.1913 � 0.0058 � 0.1588 0.1570 0.87730.0000 0.2860 7.8493 29.8862 0.00000.0074 � 0.1790 � 4.9128 10.2654 0.00080.5338 � 0.1650 � 3.4663 9.8151 0.00710.0055 � 0.3000 � 6.3023 23.1161 0.00010.2449 0.0100 0.2101 0.9246 0.8383

Fig. 1. Response surfaces and contour plots of modeled xylan conversion to D-xylose (top) and D-xylose/D-glucose ratio (bottom) as a function of: (a) reaction time (t, min) andreaction temperature (T, �C) at fixed acid concentration of 1% set as a central point (b) reaction time and acid concentration (c, %) at fixed temperature of 95 �C set as a centralpoint (c) reaction temperature and acid concentration at fixed reaction time of 120 min set as a central point.

Fig. 2. Mass balance flow diagram for acid hydrolysis of olive tree pruning biomass performed under optimized conditions designed for hydrolysis process.

J.G. Puentes et al. / Bioresource Technology 149 (2013) 149–154 153

@Y2

@X3¼ 0:3372� 0:3875X1 � 0:3954X3 ¼ 0 ð6Þ

The following critical condition set was obtained after resolu-tion of Eqs. (4)–(6): reaction time of 138.2 min, reaction tempera-ture of 96.6 �C, sulfuric acid concentration of 1.27% and amaximum expected D-xylose/D-glucose ratio of 1.696 which corre-spond to a xylan conversion of 0.849 (85% of monomeric D-xyloserecovery in solution).

To validate the developed statistical model, the duplicated con-trol experiments were performed under established optimal condi-tions. The obtained experimental data on xylan conversion (0.851� 0.007, or 13.76 g D-xylose/100 g dry material) and D-xylose/D-glucose ratio (1.7� 0.4, with 8.32 g D-glucose/100 g dry material)were the same as predicted by the model. The resulting hydroly-zate revealed low concentration of toxic substances (furfural0.11 g, 5-hydroxymethylfurfural 0.04 g, and acetic acid 3.78 g per100 g of dry biomass) providing the highest quality of substratefor subsequent (bio)chemical processing.

The mass balance flow diagram, summarizing the yields of theprincipal structural components (cellulose, xylan and lignin) ofolive tree pruning material during acid hydrolysis (performedunder optimized conditions), is shown in Fig. 2.

4. Conclusions

Low temperature dilute sulfuric acid hydrolysis was very effec-tive to convert olive tree pruning biomass to monomeric sugars,providing a quality substrate for subsequent (bio)chemical pro-cessing. The statistical modeling, using RSM, made possible toidentify the main factors of the hydrolysis process affecting effi-ciency and selectivity of xylan conversion to D-xylose and to definethe critical set of reaction conditions for D-xylose/D-glucose ratio.Under these reaction conditions (time 138.2 min, temperature96.6 �C and sulfuric acid concentration 1.27%), a xylan conversionof 0.85 and a D-xylose/D-glucose ratio of 1.7 was achieved, withlimited inhibitors formation.

154 J.G. Puentes et al. / Bioresource Technology 149 (2013) 149–154

Acknowledgements

The authors are grateful to Andalusia Regional Government forthe financial support (Project Ref. AGR-6509). On the other hand,the authors also acknowledge the financial support from ‘‘CiênciaSem Fronteiras’’ of the ‘‘Conselho Nacional de DesenvolvimentoCientífico e Tecnológico’’ (CNPq), ‘‘Coordenação de Aperfeiçoamen-to de Pessoal de Nível Superior’’ (CAPES) and ‘‘Fundação de Amparoà Pesquisa do Estado de São Paulo’’ (FAPESP).

References

Atchison, J.E., 1987. Data on non-wood plant fibers. In: TAPPI, CPPA, Atlanta,Montreal, pp. 4–16.

Browning, B.L., 1967. Methods of Wood Chemistry. John Wiley & Sons, New York.Du, W., Ren, X., Xu, M., Zhou, A., 2012. Influencing factors in hydrolysis of sunflower

stalks by using dilute acid. Energy Procedia 17, 1468–1475.Fengel, D., Wegener, G., 1989. Wood: Chemistry, ultrastructure, reactions. Walter de

Gruyter, New York, Berlin.Granström, T.B., Leisola, M., 2009. Production and applications of xylitol. Agro Food

Ind. Hi Tech 20, 27–31.Gütsch, J.S., Nousiainen, T., Sixta, H., 2012. Comparative evaluation of

autohydrolysis and acid-catalyzed hydrolysis of eucalyptus globulus wood.Bioresour. Technol. 109, 77–85.

Hurter, A.M. 1988. Utilization of annual plants and agricultural residues for theproduction of pulp and paper, New Orlearns LA, USA, pp. 139–160.

Irick, T.J., West, K., Prownell, H.H., Schwald, W., Saddler, J.N., 1988. Comparision ofcolorimetric and HPLC techniques for quantitating the carbohydratecomponents of steam-treated wood. Appl. Biochem. Biotechnol. 17, 137–149.

Kamm, B., Kamm, M., 2007. Biorefineries – Multi product processes. Adv. Biochem.Eng. Biotechnol., 175–204.

Karimi, K., Kheradmandinia, S., Taherzadeh, M.J., 2006. Conversion of rice straw tosugars by dilute-acid hydrolysis. Biomass Bioenergy 30, 247–253.

Lee, H.J., Ahn, S.J., Seo, Y.J., Lee, J.W., 2013. A feasibility study on the multistageprocess for the oxalic acid pretreatment of a lignocellulosic biomass usingelectrodialysis. Bioresour. Technol. 130, 211–217.

Mateo, S., Puentes, J.G., Sánchez, S., Moya, A.J., 2013a. Oligosaccharides andmonomeric carbohydrates production from olive tree pruning biomass.Carbohydr. Polym. 93, 416–423.

Mateo, S., Roberto, I.C., Sánchez, S., Moya, A.J., 2013b. Detoxification ofhemicellulosic hydrolyzate from olive tree pruning residue. Ind. Crops Prod.49, 196–203.

Moya, A.J., Bravo, V., Mateo, S., Sánchez, S., 2008. Fermentation of acid hydrolysatesfrom olive-tree pruning debris by Pachysolen tannophilus. Bioprocess Biosyst.Eng. 31, 611–617.

Myers, R.H., Montgomery, D.C., Anderson-Cook, C.M., 2009, . Response surfacemethodology: process and product optimization using designed experiment,third ed. John Wiley & Sons, New York.

Noureddini, H., Byun, J., 2010. Dilute-acid pretreatment of distillers’ grains and cornfiber. Bioresour. Technol. 101, 1060–1067.

Pronyk, C., Mazza, G., 2012. Fractionation of triticale, wheat, barley, oats, canola andmustard straws for the production of carbohydrates and lignins. Bioresour.Technol. 106, 117–124.

Roberto, I.C., Mussatto, S.I., Rodrigues, R.C.L.B., 2003. Dilute-acid hydrolysis foroptimization of xylose recovery from rice straw in a semi-pilot reactor. Ind.Crops Prod. 101, 171–176.

Rodrigues, R.C.L.B., Rocha, G.J.M., Rodrigues, D., Filho, H.J.I., Felipe, M.G.A., Pessoa, A.,2010. Scale-up of diluted sulfuric acid hydrolysis for producing sugarcanebagasse hemicellulosic hydrolysate (SBHH). Bioresour. Technol. 101, 1247–1253.

Sluiter, A., Ruiz, R., Scarlata, C., Sluiter, J., Templeton, D., 2008. Determination ofExtractives in Biomass. Laboratory Analytical Procedure NREL/TP-510-42619.National Renewable Energy Laboratory, Colorado (USA). 1617 Cole Boulevard,Golden, Colorado 80401-3393.