Bioresource Technology 101 (2010) 7416–7423

Contents lists available at ScienceDirect

Bioresource Technology

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

Combination of alkaline and enzymatic treatments as a process for upgradingsisal paper-grade pulp to dissolving-grade pulp

David Ibarra a,1, Viviana Köpcke a, Per Tomas Larsson a,b, Anna-Stiina Jääskeläinen a,c, Monica Ek a,*

a Royal Institute of Technology, Dept. of Fiber and Polymer Technology, SE 10044 Stockholm, Swedenb Innventia AB, Box 5604, SE 11486 Stockholm, Swedenc Aalto University, Dept. of Forest Products Technology, FI 00076 Aalto, Finland

a r t i c l e i n f o a b s t r a c t

Article history:Received 20 January 2010Received in revised form 12 April 2010Accepted 15 April 2010Available online 20 May 2010

Keywords:Alkaline extractionDissolving-grade pulpEnzymatic treatmentPaper-grade pulpNon-wood fibers

0960-8524/$ - see front matter � 2010 Elsevier Ltd. Adoi:10.1016/j.biortech.2010.04.050

* Corresponding author. Tel.: +46 8 790 8104; fax:E-mail address: monicaek@kth.se (M. Ek).

1 Present address: CIEMAT, Renewable Energy DivMadrid, Spain.

A sequence of treatments consisting of an initial xylanase treatment followed by cold alkaline extractionand a final endoglucanase treatment was investigated as a process for upgrading non-wood paper-gradepulps to dissolving-grade pulps for viscose production. Five commercial dried bleached non-wood soda/AQ paper pulps, from flax, hemp, sisal, abaca, and jute, were studied for this purpose. Commercial driedbleached eucalyptus dissolving pulp was used as reference sample. Sisal pulp showed the highestimprovement in Fock’s reactivity, reaching levels nearly as high or even higher than that of eucalyptusdissolving pulp (65%), and a low hemicellulose content (3–4%) when was subjected to this sequence oftreatments. The viscosity, however, decreased considerably. A uniform and narrow molecular weight dis-tribution was observed by size exclusion chromatography. 13C nuclear magnetic resonance spectroscopyand Raman microspectroscopy revealed that the cellulose structure consisted of cellulose I.

� 2010 Elsevier Ltd. All rights reserved.

1. Introduction

Despite wood is still by far the main source for pulp and paperproduction, non-wood fibers occupy small niche markets providingspecial properties to a range of high added value products (Moore,1996). However, where wood-based fibers are not available, as inthe developing world, they are the major source for the pulp andpaper industry (Moore, 1996). Moreover, there is a growing needin different world organizations to consider alternative agriculturalstrategies that move an agricultural industry entirely focused onfood production to one that also supplies the needs of other indus-trial sectors, such as pulp and paper, and cellulose-based products(van Dam et al., 1994). In addition of cereal straw, the leading non-woody plant, other sources such as flax, hemp, sisal, abaca or jutecould become an important crop in this transformation.

Dissolving-grade pulps are used as raw material in the manu-facture of different cellulose-derived products, including viscoserayon, the first commercially manufactured regenerated cellulosefiber. In the viscose process, cellulose is treated with carbon disul-fide in the presence of a base to produce cellulose xanthate(Treiber, 1985). In contrast to paper-grade pulps, dissolving pulpsmust contain a high content of cellulose (90–99%), low content of

ll rights reserved.

+46 8 790 6166.

ision, Biomass Unit, E 28040

hemicelluloses (2–4%), and traces of residual lignin, extractivesand minerals. A cellulose predominantly consisted of cellulose Iwith a uniform molecular weight distribution is also desired (Sixta,2006).

Hemicelluloses are undesirable impurities in dissolving pulps,affecting the cellulose processability, e.g. the filterability and thexanthanation in the viscose process, and properties of the cellu-lose-end products such as the viscose strength (Christov and Prior,1993). Most of the hemicellulose is reduced from wood by acid sul-fite and pre-hydrolysis kraft processes, the two major methodsused to produce dissolving pulps. However, the manufacture ofthese pulps demands higher costs than the commonly used paperpulps. For this reason, other alternatives have been studied; amongthem the conversion of paper pulps to dissolving pulps by selectivereduction of the hemicellulose (Wallis and Wearne, 1990; Jacksonet al., 1998; Bajpai and Bajpai, 2001; Puls et al., 2006; Köpcke et al.,2008). Different methods have been developed for this purpose,including alkaline, nitren, and cuen extraction (Wallis and Wearne,1990; Puls et al., 2006). In the same way, the use of xylanases,alone or in combination with alkaline extraction, has been alsodemonstrated (Jackson et al., 1998; Bajpai and Bajpai, 2001;Köpcke et al., 2008).

Reactivity is often the most significant quality parameter of dis-solving pulps. High cellulose reactivity improves the homogeneityand quality of cellulose-end products and lower the demands ofreactants, e.g. use of carbon disulfide in the viscose manufacture,reducing production costs and the environmental impact.

D. Ibarra et al. / Bioresource Technology 101 (2010) 7416–7423 7417

However, increasing the accessibility and reactivity of cellulose isnot a simple task. Cellulose has a compact fibrillar structure as a re-sult of intra- and intermolecular hydrogen bonds and hydrophobicinteractions (Fengel and Wegener, 1984). Various treatments havebeen assessed to increase the reactivity of the cellulose (Krässig,1993), including the use of enzymes such as monocomponent cel-lulases (Henriksson et al., 2005; Kvarnlöf et al., 2006; Engströmet al., 2006; Köpcke et al., 2008).

After showing the feasibility of upgrading eucalyptus and birchpaper kraft pulps to dissolving pulps by combining alkaline extrac-tion with xylanase and endoglucanase treatments (Köpcke et al.,2008), these treatments are investigated and optimized on flax,hemp, sisal, abaca and jute soda/AQ paper-grade pulps for thesame purpose. The treatment effects on pulps are evaluated interms of reactivity, according to Fock’s method, viscosity, andhemicellulose content. The molecular weight distribution is re-corded by size exclusion chromatography (SEC). The supramolecu-lar structure is studied by 13C nuclear magnetic resonancespectroscopy (13C-CP/MAS NMR) and Raman microspectroscopy.

2. Methods

2.1. Pulp samples

Commercial dried ECF-bleached soda/AQ paper-grade pulpsfrom flax (Linum usitatissimun), hemp (Cannabis sativa), sisal (Agavesisalana), abaca (Musa textilis), and jute (Corchorus capsuloris), pro-vided by Celesa (Spain), were investigated. The results were com-pared with those of a commercial dried TCF-bleached sulfite pulpfrom eucalyptus (Eucalyptus globulus), provided by Sniace (Spain).In general, these pulps presented a low kappa number (0.6–1), highbrightness (near 90%) and alpha cellulose content around 88–91%.The viscosity values were different, depending on the pulp(530 mL g�1 for E. globulus sulfite pulp, and 802, 683, 654, 1195and 692 mL g�1 for flax, hemp, sisal, abaca and jute soda/AQ pulp).Prior to the treatments, the dried sheets were maintained in deion-ized water for 24 h, disintegrated in Lorentzen & Wettre equip-ment at 1.5% consistency and 30,000 revolutions, according tothe ISO standard 5263-1:2004, and were finally filtrated.

2.2. Enzymes

Monocomponent endoglucanase preparation (Novozyme 476)and xylanase preparation (Pulpzyme HC), both supplied by Novo-zymes Denmark, were used. Novozyme 476 is produced from agenetically modified Aspergillus species. The cellulolytic activitywas determined by the manufacturer and expressed in Endo Cellu-lase Units (ECU) per unit mass of material as 5000 ECU g�1. Pulp-zyme HC is produced from a genetically modified Bacillus species.The xylanase activity was determined by the manufacturer and ex-pressed in Endo Xylanase Units (EXU) per unit mass of material as1000 EXU g�1.

2.3. Enzymatic and chemical treatments

Enzymatic treatments were carried out according to Köpckeet al. (2008) on 10 g (dry weight) of pulp at 3% pulp consistencyin phosphate buffer solution (11 mM NaH2PO4 and 9 mM Na2H-PO4), pH 7 (the optimal pH of the enzymes, as described by themanufacturer Novozymes). For a homogeneous distribution, theenzymes were added to the buffer and then to the pulp. The enzy-matic treatments were performed in plastic bags in a water bath at50 �C for Novozyme 476 and 60 �C for Pulpzyme HC (the optimaltemperatures of the enzymes, as described by the manufacturerNovozymes). The pulps were kneaded every 30 min. After treat-

ment, the enzymes were denatured by filtration on a Büchner fun-nel and mixed with deionized water at 90 �C. The treated pulpswere placed in a 90 �C water bath for 30 min and subsequently fil-tered and washed with 1000 mL of deionized water. As a control,pulps were treated under identical conditions without enzymes.

The effects of Novozyme 476 dosage and incubation time on thedifferent pulps were investigated, as described in previous study(Köpcke et al., 2008). Different enzyme dosages were tested (0,50, and 250 ECU g�1 dry weight pulp), keeping the incubation timeconstant at 1 h. In the same way, different incubation times wereconsidered (0, 15, 30, 45, 60, and 120 min), keeping the enzymedosage constant at 250 ECU g�1 dry weight pulp.

Similarly, in order to optimize the removal of xylan by enzy-matic treatment, different dosages of Pulpzyme HC were tested(0, 10, 80, 500, and 1000 EXU g�1 dry weight pulp), with incubationtime of 2 h.

Chemical treatment consisted of an alkaline extraction with 9%NaOH solution at room temperature for 1 h and 4% pulp consis-tency (Köpcke et al., 2008). Extracted pulps were filtered andwashed with deionized water until the filtrate pH was neutral.

2.4. Reactivity measurements

The reactivity of the treated pulps was analyzed according to aslightly modified version of Fock’s method (Fock, 1959; Henrikssonet al., 2005). This test is a micro-scale process similar to the viscoseprocess. The test was carried out in two steps. Prior to Fock’s anal-ysis, the treated pulps were dried at 50 �C.

2.4.1. Step 1. Preparation of viscose from treated pulps and collectionof regenerated cellulose

0.5 g of pulp samples were weight in a 100 mL Erlenmeyer witha stopper. Fifty milliliters of 9% NaOH and 1.3 mL of CS2 wereadded, and the solutions were stirred with a magnetic stirrer(300 rpm) for 4 h at room temperature. The solution was dilutedto 100 g using deionized water and carefully shaken. The solutionwas then left for 2 h in order to allow any undissolved celluloseto settle. An aliquot (10 mL) from the upper clear solution was thentransferred to another 100 mL Erlenmeyer flask and neutralizedusing 29% H2SO4. The yellow solution turned transparent andwas left overnight in a fume cupboard.

2.4.2. Step 2. Oxidation and titration of the regenerated celluloseThe regenerated cellulose samples were mixed with 20 mL of

68% H2SO4 and stirred with a magnetic stirrer for 1 h. The milkysolution was diluted to 50 mL with deionized water. Ten millilitersof 1 N K2Cr2O7 was added, and the solution was refluxed for 1 h tofully oxidize the regenerated cellulose and thereby clear the solu-tion. The solution was transferred to a 100 mL measuring and di-luted with deionized water. Forty milliliters of the solution wasthen transferred to a 250 mL beaker containing 0.5 g of KI, stirredwith a magnetic stirrer, and titrated with 0.1 N of Na2S2O3. Whenthe brown solution started to change the color, 1.5 g of starchwas added, and the solution turned blue-violet. The titration con-tinued until all of the I2 was reduced and the solution turned paleblue. The volume of Na2S2O3 required in each case was determined.

Reactivity measurements were carried out in triplicate and ex-pressed as the regenerated cellulose yield (Eq. (1))

X ¼ ð100Þ9:62a MðV1C1 � ðV2C2100=40bÞ=6Þ4Y

ð1Þ

where, X is the reacted cellulose (%), Y is the weight of sample (g), Mis the molecular mass of glucopyranosyl residue, C6H10O5

(162 g mol�1), V1 is the volume of added K2Cr2O7 (L), V2 is the vol-ume of titrated Na2S2O3 (L), C1 is the concentration of K2Cr2O7

7418 D. Ibarra et al. / Bioresource Technology 101 (2010) 7416–7423

(mol L�1), C2 is the concentration of Na2S2O3 (mol L�1), a is the firstdilution to 100 g and outtake of 10 mL (10.4 g) = 100/10.4 = 9.62,and b is the second dilution of the sample to 100 mL and outtakeof 40 mL = 100/40.

2.5. Determination of viscosity

The viscosity of the treated pulps, defined as the intrinsic vis-cosity of a sample of cellulose dissolved in a diluted solution of cu-pri-ethylenendiamene, was calculated according to SCAN-CM15:99. Viscosity measurements were carried out in duplicate.

2.6. Carbohydrate analysis

The carbohydrate composition of the treated pulps was ana-lyzed after acid hydrolysis and reduction with sodium borohy-dride, according to Theander and Westerlund (1986). Afterreduction, the corresponding alditols were acetylated and analyzedby gas chromatography (GC), using a Hewlett–Packard HP-6890chromatograph. Inositol was used as an internal standard. Separa-tion was performed in a BP X70 capillary column (12 m � 0.32 mmi.d.) (SGE Analytical Science) operated at 210 �C with He as the car-rier gas. The injector temperature was 230 �C and the detectortemperature was 250 �C.

2.7. Size exclusion chromatography

Prior to characterization on a SEC system, the treated pulpswere dissolved and derivatized (Berthold et al., 2004). Fifteen mil-ligrams of treated pulp samples were activated for 1 h in 15 mL ofdeionized water at 4 �C. The excess water was removed, and thepulps were solvent-exchanged once with methanol and then threetimes with DMAc (N,N-dimethylacetamide) with an intermediateequilibration period of 30 min. Then, 1.9 mL of 8% LiCl/DMAc and3 mmol of ethyl isocyanate (derivatizing reagent) were added,and the samples were left at 4 �C for 5 days with mild magneticstirring. Finally, the samples were diluted to 0.5% LiCl by the addi-tion of 27.4 mL of DMAc. The dissolved treated pulps were filteredthrough a 0.45 lm PTFE (poly(tetrafluorethylene)) filter beforechromatographic characterization.

Samples were analyzed on a SEC system consisting of a DGU-20A3 degasser (Shimadzu), a LC-20AD liquid chromatograph (Shi-madzu), a CT-20A column oven (Shimadzu) equipped with a Rhe-odyne 7725i fixed-loop (100 lL) and a RID-10A refractive indexdetector (Shimadzu). The injection volume was 100 lL, and theseparations were performed at 80 �C with 0.5% LiCl/DMAc at a flowrate of 0.5 mL min�1 on four Mixed-A 20 lm columns(7.5 � 300 mm, Polymer Laboratories) connected in series and pre-ceded by a Mixed-A 20 lm guard column (7.5 � 50 mm, PolymerLaboratories). Pullulan standards 800 K, 400 K, 200 K, 110 K, 50 K,22 K, 12 K, 6 K, 1.3 K and 320 Da (Fluka) were used to calibratethe columns. The linear coefficient of determination (r2) betweenthe molecular weight of the standards and the elution time was0.996. Data acquisition and calculation were carried out with LCSolution software (Shimadzu).

2.8. NMR spectroscopy

All samples were wetted with deionized water and packed uni-formly in a zirconium oxide rotor. 13C-CP/MAS NMR spectra wererecorded in a Bruker Avance AQS 300 WB instrument operatingat 7.04 T. All measurements were performed at 290 ± 1 K. TheMAS rate was 5 kHz. A 7-mm double air-bearing probe was uti-lized, and the acquisition was performed with a CP pulse sequenceusing a 4.3-ls proton 90� pulse, a 800-ls ramped (100–50%) falling

contact pulse and a 2.5 s delay between repetitions. A TPPM 15pulse sequence was used for 1H decoupling.

2.9. Raman microspectroscopy

The Raman spectra were measured using a Horiba Jobin YvonHR 800 Raman system equipped with a confocal microscope. A la-ser diode of 785 nm wavelength was used in excitation and theRayleigh scattering was rejected by a notch filter with a cut-offof 120 cm�1. The confocal hole was set to 300 lm. 100� objective(NA 0.9) was used in the measurements. A kinematic grating of600 L mm�1 was applied to direct the Stokes scattered photonson an air-cooled CCD camera set to �65 �C. The spectral band posi-tions were calibrated daily by using a silicon standard and the cal-ibration was further checked by Teflon standard. The pulp sampleswere installed on a microscope objective glass and the measuredfibers were directed in parallel orientation with the excitation lightpolarization. Five spectra from different samples were collected(20 s acquisition time and 8 accumulations) and the average spec-tra of these five spectra were calculated. Possible cosmic peakswere removed manually from the spectra. If the cosmic peaks lo-cated at the same frequencies with the Raman bands, the spectralcollection was repeated. The spectral data treatment was per-formed with LabSpec5 software.

3. Results and discussion

After developing a sequence of treatments for the conversion ofeucalyptus and birch paper kraft pulps to dissolving pulps for vis-cose process (Köpcke et al., 2008), consisting of an initial xylanasetreatment followed by cold alkaline extraction and a final endoglu-canase treatment, this treatments are investigated on differentnon-wood soda/AQ paper pulps for the same purpose.

3.1. Effect of endoglucanase treatment: reactivity and viscosity

Dissolving-grade pulps have typically higher cellulose reactivitylevels relative to paper-grade pulps. Similar to birch and eucalyptuskraft pulps (Köpcke et al., 2008), the non-wood soda/AQ pulps alsoshowed a Focks reactivity levels considerably lower than that ofeucalyptus dissolving pulp (65%) (Fig. 1a). Sisal pulp had the highestreactivity, in the same range than eucalyptus and birch kraft pulps(35–36%), while hemp pulp had the lowest reactivity (20%). Flax,abaca and jute pulps had reactivity values around 24–26%. Thelow reactivity of the non-wood pulps might be related to the lowamount of less ordered cellulose content, 25–30% according to Yeand Farriol (2005). Only this cellulose, located between and onthe surface of the fibril aggregates, is accessible to chemicals (Kräs-sig, 1993), although part of which becomes inaccessible in driedpulps, as employed in this study, by the hornification phenomenon(Jayme, 1944). The content of hemicelluloses was higher (over 10%)compared to eucalyptus dissolving pulp (2–4%), which also stronglyinfluences their low reactivities (Krässig, 1993). This remaininghemicellulose, deposited on the surface of the cellulose fibril aggre-gates (Fengel and Wegener, 1984), reduces the surface area and,consequently, the accessibility of the cellulose to reagents (Krässig,1993). Cellulose structure, fiber morphology (Krässig, 1993), andthe species (Ye and Farriol, 2005) are other influential factors intheir accessibilities and reactivities.

To achieve reactivity levels comparable to eucalyptus dissolvingpulp (65%), the non-wood pulps need to be activated. For this pur-pose, several physical and chemical treatments have been previ-ously employed (Ye and Farriol, 2005). Endoglucanase treatmenthas also been successfully applied in wood dissolving (Henrikssonet al., 2005; Engström et al., 2006) and paper pulps (Köpcke et al.,2008). In the present study, the non-wood pulps were subjected to

Eucalyptusdissolving

Flax soda/AQ

Hempsoda/AQ

Sisal soda/AQ

Abacasoda/AQ

Jute soda/AQ

a

010

20

304050

607080

90100

Eucalyptusdissolving

Flax soda/AQ

Hempsoda/AQ

Sisal soda/AQ

Abacasoda/AQ

Jute soda/AQ

0

200

800

600

400

1000

1200

1400

Visc

osity

(mL

g-1 )

b

Flax soda/AQ

Hempsoda/AQ

Sisal soda/AQ

Abacasoda/AQ

Jute soda/AQ

010

20

304050

607080

90100

Rea

cted

cel

lulo

se (%

)

Flax soda/AQ

Hempsoda/AQ

Sisal soda/AQ

Abacasoda/AQ

Jute soda/AQ

0

200

800

600

400

1000

1200

1400

Visc

osity

(mL

g-1 )

Fig. 1. Fock’s reactivity (a) and viscosity (b) of different endoglucanase treated non-wood pulps, compared to eucalyptus dissolving pulp. Novozyme 476 dosages: 0(black bars), 50 (gray bars), and 250 ECU g�1 (white bars) dry weight pulp.Incubation time: 1 h.

0

20

40

60

80

100 700

650

600

550

500200 40 60 120

Rea

cted

cellu

lose

(%)

Visc

osity

(mL

g-1 )

0

20

40

60

80

100 700

650

600

550

500200 40 60 80 100 120

Time (min)

Rea

cted

cellu

lose

(%)

Visc

osity

(mL

g-1 )

Fig. 2. Fock’s reactivity (d) and viscosity (s) of endoglucanase sisal soda/AQtreated pulp as a function of the incubation time (0, 15, 30, 45, 60 and 120 min).Novozyme 476 dosage: 250 ECU g�1 dry weight pulp.

Table 1Carbohydrate composition of sisal soda/AQ pulp after xylanase treatment usingdifferent enzyme dosages. Activity of Pulpzyme HC: (EXU g�1 dry weight pulp).

PulpzymeHC

Glu(%)

Man(%)

Xyl(%)

Ara(%)

Gal(%)

Eucalyptus dissolving pulp 97.6 0.0 2.4 0.0 0.0

Sisal soda/AQ pulp 81.0 <0.5 18.1 <1 0.010 82.7 <0.5 16.5 <1 0.080 84.6 <0.5 14.6 <1 0.0500 85.6 <0.5 14.0 <1 0.01000 85.5 <0.5 13.9 <1 0.0

D. Ibarra et al. / Bioresource Technology 101 (2010) 7416–7423 7419

a Novozyme 476 endoglucanase treatment (Fig. 1a). The highestreactivity increase (17 units) was obtained when 50 ECU g�1 dryweight pulp of Novozyme 476 was applied on sisal pulp, reachingaround 50% of reactivity, in the same rate than eucalyptus andbirch kraft pulps after endoglucanase treatment (Köpcke et al.,2008). A similar increase was obtained for abaca and jute pulps(15 units); in contrast, the reactivity values did not reach morethan 40%. Flax and hemp pulps did not show any remarkableimprovement. Some hypotheses have been described to explainthe positive effect of the monocomponent endoglucanases on thereactivity of pulps (Rabinovich et al., 2002; Henriksson et al.,2005). One of them suggests the attack of the less ordered celluloseregions by endoglucanase, leading to fiber wall swelling and, there-fore, an increase in the accessibility to solvents. However, someparameters limit the endoglucanase action on the different non-wood pulps. As explained for eucalyptus and birch kraft pulps(Köpcke et al., 2008), the high hemicellulose content and the hor-nification phenomenon might hinder the penetration of the endo-glucanase into the fibers, limiting the rise in their reactivities.Moreover, the species can be other influencing factor on the endo-glucanase behavior (Ye and Farriol, 2005).

The highest reactivity increase was achieved for sisal pulp (25units) when 250 ECU g�1 dry weight of Novozyme 476 was used,reaching values close (60%) to that of eucalyptus dissolving pulp(Fig. 1a) after 1 h of treatment (Fig. 2). An equivalent tendencywas observed for birch and eucalyptus kraft pulps (Köpcke et al.,2008). Abaca and jute pulps showed a reactivity increase of 20units, although their reactivities did not exceed more than 45%.In contrast, an increased endoglucanase dosage did not entail areactivity increase of flax and hemp pulps. None of the pulpsshowed a reactivity enhancement by elevating the endoglucanasedosage over 250 ECU g�1 dry weight pulp (data not shown).

Another important aspect of dissolving pulps for viscose pro-duction is their low degree of polymerization (Sixta, 2006). In gen-eral, the non-wood pulps showed higher viscosity levels than thatof eucalyptus dissolving pulp (530 mL g�1) (Fig. 1b). Abaca pulphad the highest viscosity (1200 mL g�1), while sisal, hemp, and jutepulps had the lowest viscosity (650–690 mL g�1). Flax had an inter-mediate viscosity (800 mL g�1).

A high viscosity affects the cellulose processability in the vis-cose process. Thus, it is necessary to lower the viscosities of non-wood pulps to upgrade them to dissolving pulps. Besides ofincreasing the cellulose reactivity, physical and chemical treat-ments have shown their efficiency lowering the viscosity (Ye andFarriol, 2005). The endoglucanase treatment also offers this possi-bility. As described for eucalyptus and birch kraft pulps (Köpckeet al., 2008), Novozyme 476 endoglucanase (50 ECU g�1 dry weightpulp) produced a viscosity decrease in all enzymatic treated non-wood pulps (Fig. 1b), being considerably intense for abaca pulp(310 units). Flax and hemp pulps presented a viscosity decreaseof 220 and 170 units, respectively. The lowest decrease was ob-served for jute and sisal pulps (70–90 units). Lower values werereached using a higher endoglucanase dosage (250 ECU g�1 dryweight pulp) (Fig. 1b), although the influence of the dosage wasdependent on the pulp type. As illustrated for sisal pulp (Fig. 2),the highest viscosity decrease was achieved within the first15 min, although the lowest viscosity was reached by extendingthe enzymatic treatment to 2 h.

3.2. Sisal soda/AQ pulp

3.2.1. Effect of xylanase treatment: carbohydrate compositionIt is known that excessive amounts of hemicelluloses are con-

sidered to be undesirable impurities in the viscose process (Sixta,

Table 3Carbohydrate composition of sisal soda/AQ pulp after xylanase treatment, alkalineextraction, xylanase treatment followed by alkaline extraction, and xylanase treat-ment followed by alkaline extraction and endoglucanase treatment.

Glu (%) Man (%) Xyl (%) Ara (%) Gal (%)

Eucalyptus dissolving pulp 97.6 0.0 2.4 0.0 0.0Sisal soda/AQ pulp 81.0 <0.5 18.1 <1 0.0Xyla 85.3 <0.5 14.0 <1 0.0Xyl + Endoc 84.7 <0.5 14.6 <1 0.0Xyl + Endod 85.0 <0.5 14.2 <1 0.0Alkb 94.1 <0.5 5.1 <1 0.0Xyl + Alk 95.6 <0.5 3.6 <1 0.0Xyl + Alk + Endoc 95.2 <0.5 4.2 <1 0.0Xyl + Alk + Endod 95.5 <0.5 3.9 <1 0.0Eucalyptus kraft pulpe 97.6 0.0 2.4 0.0 0.0Birch kraft pulpe 95.2 <1 3.8 0.0 0.0

a Xylanase treatment (500 EXU g�1 dry weight pulp).b Alkaline extraction.

c,d Endoglucanase treatment (50 and 250 ECU g�1 dry weight pulp, respectively).e After xylanase + Alk + endoglucanase treatment (Köpcke et al., 2008).

7420 D. Ibarra et al. / Bioresource Technology 101 (2010) 7416–7423

2006). Compared to eucalyptus dissolving pulp, sisal soda/AQ pulpshowed a significantly higher content of xylan (Table 1). For its useas dissolving pulp, sisal pulp was also subjected to a xylanase(Pulpzyme HC) treatment for a selective reduction of hemicellulosecontent.

Ca. 22% of the initial xylan present in the sisal pulp dissolvedafter the treatment with 500 EXU g�1 dry weight pulp of PulpzymeHC (Table 1), lower compared to 47% and 37% of the initial xylansolubilized in birch and eucalyptus kraft pulps, respectively(Köpcke et al., 2008). This restriction in the hemicellulosessolubilization has been also reported for dissolving pulps (Gübitzet al., 1997). The resistance of the hemicelluloses towards enzy-matic hydrolysis may be related to a partial inaccessibility of theresidual hemicelluloses (Christov and Prior, 1993). This could bedue to many factors such as xylanase size (Stone and Scallan,1968), fiber porosity (Wong et al., 1988), medium pore size(Suurnäkki et al., 1997), and accessible surface area (Stone et al.,1969). Moreover, the more accessible fraction of hemicellulose isdissolved from the cell walls during the bleaching stage, leavinga fraction in locations less accessible to the enzymatic attack(Christov and Prior, 1993; Suurnäkki et al., 1996). Higher enzymedosages did not increase the extent of hemicelluloses hydrolysis(Table 1).

3.2.2. Influence of xylanase prior to endoglucanase treatment:carbohydrate composition, reactivity and viscosity

The reactivity of sisal soda/AQ pulp showed an improvement of6–8 units when xylanase treatment (500 EXU g�1 dry weight pulp)was performed prior to endoglucanase treatment (Table 2), reach-ing 65% with 250 ECU g�1 dry weight pulp of endoglucanase. Thesame phenomenon was observed for eucalyptus kraft pulp (Köpckeet al., 2008). This reactivity enhancement can be as a result ofchanges in the pulp fibers and their surface. The enzymatic re-moval of xylan prior to endoglucanase treatment suggests an in-crease of both surface area and pulp porosity which enables asmoother and easier penetration of the endoglucanase to cellulose.Wong et al. (1988) reported an improvement in the enzymaticdigestibility of the lignocellulose with the increase in fiber poros-ity. This effect was supported by the viscosity results (Table 2),which decreased 90–115 units when the xylanase was appliedprior to endoglucanase treatment, compared to values obtainedwithout the supplement xylanase stage. None variation was ob-served in the hemicelluloses content (Table 3).

Table 2Fock’s reactivity and viscosity of sisal soda/AQ pulp after endoglucanase treatment,xylanase followed endoglucanase treatment, xylanase treatment followed by alkalineextraction, and xylanase treatment followed by alkaline extraction and endoglucan-ase treatment.

Reacted cellulose (%) Viscosity (mL g�1)

Eucalyptus dissolving pulp 64.8 530Sisal soda/AQ pulp 34.8 655Endoa 51.4 585Endob 59.4 555Xylc + Endoa 59.4 495Xyl + Endob 65.4 440Xyl + Alkd 31.1 725Xyl + Alk + Endoa 57.5 375Xyl + Alk + Endob 66.2 290Eucalyptus kraft pulpe 70.3 220Birch kraft pulpe 66.0 190

a,b Endoglucanase treatment (50 and 250 ECU g�1 dry weight pulp, respectively).c Xylanase treatment (500 EXU g�1 dry weight pulp).d Alkaline extraction.e After xylanase + Alk + endoglucanase treatment (Köpcke et al., 2008).

3.2.3. Alkaline extraction between xylanase and endoglucanasetreatment: carbohydrate composition, reactivity and viscosity

Despite enhancing the reactivity and reducing the hemicellu-loses content of sisal soda/AQ pulp after xylanase and endoglucan-ase treatments, high xylan levels (14%) remained in the pulp toproduce high quality viscose. Alkaline extraction could be a feasi-ble alternative to achieve lower xylan content. The potential ofalkaline extraction to upgrade paper pulps to dissolving pulps iswell known (Wallis and Wearne, 1990; Puls et al., 2006). Similarto birch and eucalyptus kraft pulps (Köpcke et al., 2008), 70% ofthe initial xylan was solubilized for sisal pulp using 9% NaOH(Table 3). This solubilization was extended to 80% (Table 3) whenalkaline extraction was performed after the xylanase treatment(500 EXU g�1 dry weight pulp), achieving satisfactory hemicellu-lose levels (3–4%) for viscose production. The synergistic effect ofxylanase treatment and alkaline extraction in the production ofdissolving pulps has already been described with successful results(Jackson et al., 1998). The cellulose reactivity was negatively af-fected after the alkaline extraction, whereas the viscosity valuewas slightly higher (Table 2).

The integration of alkaline extraction between the xylanase andendoglucanase treatments did not vary the reactivity values (Ta-ble 2). By contrast, a considerable reduction in viscosity was ob-tained (Table 2), being stronger when 250 ECU g�1 dry weightpulp of endoglucanase was used. This effect confirms the idea thatgreat xylan removals allow a higher accessibility of endoglucanaseto the cellulose, behavior that was also described for birch andeucalyptus kraft pulps (Köpcke et al., 2008) and hardwood dissolv-ing pulp (Rahkamo et al., 1998).

The activation phenomenon seems to be related not only to vis-cosity decrement. A cellulose with lower DP (lower viscosity), i.e.when alkaline extraction was introduced between the xylanaseand endoglucanase treatments, should be more accessible tochemicals and therefore more reactive (Henriksson et al., 2005;Kvarnlöf et al., 2006). However, no reactivity increase was ob-served for sisal pulp after the endoglucanase treatment (Table 2).The same results were observed in our previous study (Köpckeet al., 2008). This effect was described by Kvarnlöf et al. (2006),who observed a viscosity decrease without reactivity increasewhen applied a multicomponent cellulase on softwood dissolvingpulp. In the same way, Engström et al. (2006) compared the reac-tivity of softwood dissolving pulp after acid hydrolysis andendoglucanase treatment. It was demonstrated that, at a given vis-cosity, the dissolving pulp had lower reactivity after acid hydroly-sis compared to endoglucanase treated dissolving pulp.

log Mw (x106)2.53

dw/d

logM

456 5.35.47 5.56.57.5log Mw (x106)

2.53

dw/d

logM

456 5.35.47 5.56.57.5

2.53

dw/d

logM

456 5.35.47 5.56.57.5 2.53

dw/d

logM

456 5.35.47 5.56.57.5log Mw (x106 Mgol) w (x106)

2.53

dw/d

logM

456 5.35.47 5.56.57.5log Mw (x106)

2.53

dw/d

logM

456 5.35.47 5.56.57.5log Mw (x106)

log Mw (x106)2.53

dw/d

logM

456 5.35.47 5.56.57.5

a

log Mw (x106)2.53

dw/d

logM

456 5.35.47 5.56.57.5

b

2.53

dw/d

logM

456 5.35.47 5.56.57.5

c

2.53

dw/d

logM

456 5.35.47 5.56.57.5

d

log Mw (x106 Mgol) w (x106)

2.53

dw/d

logM

456 5.35.47 5.56.57.5

e

log Mw (x106)2.53

dw/d

logM

456 5.35.47 5.56.57.5

f

log Mw (x106)

Fig. 3. Molecular weight distribution relative to pullulan of sisal soda/AQ pulp after different enzymatic and chemical treatments: (a) untreated pulp (N) compared toeucalyptus dissolving pulp ( ); (b) compared to pulp after endoglucanase treatment ( ); (c) compared to pulp after xylanase treatment followed by endoglucanasetreatment ( ); (d) compared to pulp after xylanase treatment followed by alkaline extraction ( ); (e and f) compared to pulp after xylanase followed by alkaline extraction andendoglucanase treatment ( ) and eucalyptus dissolving pulp ( ). Novozyme 476 dosage: 50 (e) and 250 ECU g�1 (b, c, and f) dry weight pulp. Incubation time: 1 h.

D. Ibarra et al. / Bioresource Technology 101 (2010) 7416–7423 7421

3.2.4. Molecular weight distributionPulp source with a uniform molecular weight distribution

(MWD) is of primary importance for viscose production. Incontrast to eucalyptus dissolving pulp, the MWD of sisal soda/AQpulp was rather broad (Fig. 3a) revealing a bimodal distributionwith two well-defined maxima; a low molecular mass peak ofhemicelluloses and lignin fraction and a high molecular mass peakof cellulose fraction (Janzon et al., 2008). On the other hand, a sin-gle peak of cellulose fraction, with the lowest molecular mass, andthe absence of hemicelluloses fraction was observed in the MWDof eucalyptus dissolving pulp, in agreement with other dissolvingpulps (Sixta, 2006).

A numerical evaluation from their MWDs was determined (Ta-ble 4). The average molecular weight distribution (Mw), expressedas the average degree of polymerization (DPw), was based on thetotal distribution and not only cellulose fraction. The width of

the MWD was expressed by the polydispersity index (PDI), whichconfirmed the broader distribution depicted for sisal pulp (9.6)compared to that of eucalyptus dissolving pulp (7.4), mainly dueto the larger amounts of xylans in the sisal fibers. Nevertheless,as described for other hardwood sulfite dissolving pulps, a broadcellulose distribution was showed for eucalyptus dissolving pulp(Sixta, 2006), as indicated by the high PDI (Table 4). This is also re-flected in the high amount of short-chain molecules (DP > 50)(Table 4).

Changes in the MWD of sisal pulp were observed after the endo-glucanase treatment. The cellulose peak shifted slightly towardsthe low molecular mass region (Fig. 3b), accompanied by a decre-ment of Mw and PDI values (higher with 250 ECU g�1 dry weightpulp of endoglucanase) (Table 4). A similar behavior has been re-ported on wood dissolving (Engström et al., 2006) and paper pulps(Ramos et al., 1999).

Table 4Numerical evaluation of molecular weight distribution, relative to pullulan, of sisalsoda/AQ pulp after endoglucanase treatment, xylanase treatment followed byendoglucanase treatment, xylanase treatment followed alkaline extraction, andxylanase treatment followed by alkaline extraction and endoglucanase treatment.

DPw DPn PDI DP < 50(%)

DP < 200(%)

DP > 2000(%)

Eucalyptus dissolvingpulp

2390 320 7.4 7.9 15.9 26.8

Sisal soda/AQ pulp 3575 370 9.6 na na naEndoa 3320 365 9.1 na na naEndob 3150 370 8.6 na na naXylc + Endoa 2540 300 8.5 na na naXyl + Endob 2260 280 8.1 na na naXyl + Alkd 4115 615 6.7 na na naXyl + Alk + Endoa 2070 300 6.9 2.9 14.3 28.1Xyl + Alk + Endob 1200 225 5.3 3.9 17.9 14.5Eucalyptus kraft pulpe 1095 230 4.8 5 19 12.5Birch kraft pulpe 615 135 4.5 7.4 27.3 5.5

a,b Endoglucanase treatment (50 and 250 ECU g�1 dry weight pulp, respectively).c Xylanase treatment (500 EXU g�1 dry weight pulp).d Alkaline extraction.e After xylanase + Alk + endoglucanase treatment (Köpcke et al., 2008); DPw,

weight average degree of polymerization; DPn, number average degree of poly-merization; PDI, polydispersity index (DPw/DPn); na, not applicable.

7422 D. Ibarra et al. / Bioresource Technology 101 (2010) 7416–7423

The reduction of Mw and PDI were more pronounced when axylanase treatment was performed prior to endoglucanase treat-ment (Table 4), proving a higher accessibility of the sisal fibersafter xylanase treatment. In this sense, a greater shift of the cellu-lose peak towards the low molecular mass region was observed(Fig. 3c), in comparison to the small change after the endoglucan-ase treatment alone. A slight shift of the xylan peak towards lowermolecular mass by the action of xylanase was also visible.

The alkaline extraction after the xylanase treatment let to ahomogenization of the sisal MWD (Fig. 3d). A remarkable decreasein the height of the low molecular mass peak of hemicellulose wasnoticed, in accordance with the high xylan extraction describedabove (80%). This decrease was also apparent from the PDI of sisalpulp (Table 4), which was reduced in 30%. Moreover, no cellulosedegradation could be detected, observing even a slight increaseof the Mw value (Table 4) accompanied by increased viscosity(Table 2). This effect have also been described for alkaline extrac-tion on hardwood dissolving pulp (Rahkamo et al., 1998), and nit-ren extraction on different hardwood and softwood paper pulps(Janzon et al., 2008).

Finally, when the endoglucanase treatment was applied after thexylanase treatment and alkaline extraction, a pronounced change ofthe sisal MWD was evident (Fig. 3e and f). The low molecular masspeak of hemicellulose almost disappeared. The cellulose peak dis-placed significantly towards the low molecular mass, as showedthe lower Mw values (rather small at 250 ECU g�1 dry weight ofendoglucanase) (Table 4). The lower PDI values (Table 4) displayedthe narrowing of its MWD with regard to eucalyptus dissolving pulp.It is supported by the lower amount of short-chain mole-cules (DP > 50), which is already very close to values ofpre-hydrolysis kraft viscose pulps (Sixta, 2006). The chain-lengthdistribution in dissolving pulp is a crucial property in the physico-mechanical properties of viscose. A maximum strength and highermechanical properties are exhibited with decreasing amount ofthe short-chain molecules (Sixta, 2006). In contrast, birch and euca-lyptus kraft pulps showed a great reduction in the amount of long-chain molecules (DP > 2000), especially for birch (Table 4).

3.2.5. Supramolecular structure: 13C nuclear magnetic resonancespectroscopy and Raman microspectroscopy

The effect of the alkaline extraction on cellulose sisal structurewas investigated. It is well documented that cellulose subjected to

high alkaline concentrations is converted to cellulose II (Krässig,1993; Teleman et al., 2001; Janzon et al., 2008). Cellulose II exhibitsa more tied and complex hydrogen-bonding network in compari-son to cellulose I (Krässig, 1993), being less reactive and more dif-ficult to dissolve. As a consequence, a more inhomogeneousdissolution of cellulose will be produced in the viscose process,originating a product of lower quality.

In 13C-CP/MAS NMR spectra, cellulose I is characterized by mul-tiple signals, especially a C-1 signal at 106 ppm and a C-6 signal at66 ppm with a slight shoulder at 64 ppm. Any presence of celluloseII can be detected by a shoulder in the signal intensity of C-1 at108 ppm and an increase in relative C-6 signal intensityat 64 ppm, whereas a decrease in relative C-6 signal intensity at66 ppm is produced, which disappears after a complete transfor-mation to cellulose II (Atalla et al., 1980; Janzon et al., 2008).

The spectrum of sisal pulp after xylanase treatment(500 EXU g�1 dry weight pulp) followed by alkaline extraction(with 9% NaOH) revealed mainly a great solubilization of xylan(Supplementary Fig. S1b), approximately 80% as reported above,by the disappearance of the signal at 81.7 ppm, which is assignedto xylan located at accessible fiber surface (Teleman et al., 2001).A small presence of cellulose II was apparent by a barely discern-ible signal at 108 ppm, being the spectrum dominated by celluloseI signals at 106 ppm and 66 ppm. When the endoglucanase treat-ment (250 ECU g�1 dry weight pulp) was applied as final step,the signal at 108 ppm disappeared (Supplementary Fig. S1c),observing a great similarity with the commercial eucalyptus dis-solving pulp spectrum (Supplementary Fig. S1f), which is domi-nated by cellulose I signals (Janzon et al., 2008). In wood pulps,the beginning transformation of cellulose I to cellulose II has beendescribed at 6–7% (Teleman et al., 2001; Janzon et al., 2008), pro-cess that is completed at 14–15%. This cellulose is more susceptibleto enzymatic hydrolysis by endoglucanases (Atalla, 1977), explain-ing the disappearance of scarcely discernible cellulose II signals inthe upgrade sisal pulp spectrum, and the high viscosity decreasepreviously mentioned. In contrast, in spite of the viscosity decreaseobserved also for upgrade birch and eucalyptus pulps (Köpckeet al., 2008) their spectra still exhibited the presence of celluloseII, observing a marked shoulder at 108 ppm and an increased C-6signal intensity at 64 ppm (Supplementary Fig. S1d and e).

Raman microspectroscopy confirmed the scarcely alkaline-in-duced changes in sisal supramolecular structure. The literaturehas described the transformation of cellulose I to cellulose II byspectral changes in the relative band intensities and band positions(Schenzel and Fischer, 2001; Eronen et al., 2009), such as a strongdecrease of the band intensity at 1120 cm�1 and an increase of theband intensity at 896 cm�1. The former band assigned to the glyco-sidic and ring-related COC linkage in the cellulose chain, and thelatter band assigned to HCC and HCO localized in C-6 of anhydrog-lucopyranose unit of cellulose. None of these changes could be ob-served in the Raman spectrum of upgrade sisal pulp, showingmainly features for cellulose I, which corroborates the great simi-larity with eucalyptus dissolving pulp previously described by13C-CP/MAS NMR.

4. Conclusion

The feasibility of upgrading different non-wood paper-gradepulps to dissolving-grade pulps was investigated. Initial xylanasetreatment followed by alkaline extraction and a final endoglucan-ase treatment was used for this purpose. Among them, sisal soda/AQ pulp showed a hemicellulose content and a cellulose reactivityvalue comparable to that of commercial eucalyptus dissolvingpulps. A uniform and narrow molecular weight distribution wasobtained, displaying a pattern very similar to that of pre-hydrolysiskraft dissolving pulp, with a structure predominately consisted of

D. Ibarra et al. / Bioresource Technology 101 (2010) 7416–7423 7423

cellulose I. However, despite upgrading sisal pulp a final evaluationit is necessary by a direct conversion into viscose fibers.

Acknowledgements

Dr. D.I. acknowledges a grant from the Jacob Wallenberg Re-search Foundation (Sweden). V.K. acknowledges a financial sup-port from Stiftelsen Nils och Dorthi Troëdssons forksningsfond(Sweden). Sniace (Spain) and Celesa (Spain) are acknowledged forproviding the pulp samples. Novozymes (Denmark) is acknowl-edged for supplying the enzymes.

Appendix A. Supplementary data

Supplementary data associated with this article can be found, inthe online version, at doi:10.1016/j.biortech.2010.04.050.

References

Atalla, R.H., 1977. Conformational effects in the hydrolysis of cellulose. Adv. Chem.Ser. 181, 55–69.

Atalla, R.H., Gast, J.C., Sindorf, D.W., Bartuska, V.J., Maciel, G.E., 1980. 13C NMRspectra of cellulose polymorphs. J. Am. Chem. Soc. 102, 3249–3251.

Bajpai, P., Bajpai, P.K., 2001. Development of a process for the production ofdissolving kraft pulp using xylanase enzyme. Appita J. 54, 381–384.

Berthold, F., Gustafsson, K., Berggren, R., Sjöholm, E., Lindström, M., 2004.Dissolution of softwood kraft pulps by direct derivatization in lithiumchloride/N,N-dimethylacetamide. J. Appl. Polym. Sci. 94, 424–431.

Christov, L.P., Prior, B.A., 1993. Xylan removal from dissolving pulp using enzymesof Aureobasidium pullulans. Biotechnol. Lett. 15, 1269–1274.

Engström, A.-C., Ek, M., Henriksson, G., 2006. Improved accessibility and reactivityof dissolving pulp for the viscose process: pretreatment with monocomponentendoglucanase. Biomacromolecules 7, 2027–2031.

Eronen, P., Österberg, M., Jääskeläinen, A.S., 2009. Effect of alkaline treatment oncellulose supramolecular structure studied with combined confocal Ramanspectroscopy and atomic force microscopy. Cellulose 16, 167–178.

Fengel, D., Wegener, G., 1984. Wood: Chemistry, Ultrastructure, Reactions. DeGruyter, Berlin.

Fock, W., 1959. A modified method for determining the reactivity of viscose-gradedissolving pulp. Papier 13, 92–95.

Gübitz, G.M., Lisching, T., Stebbing, D., Sadler, J.N., 1997. Enzymatic removal ofhemicellulose from dissolving pulps. Biotechnol. Lett. 19, 491–495.

Henriksson, G., Christiernin, M., Agnemo, R., 2005. Monocomponent endoglucanasetreatment increases the reactivity of softwood sulphite dissolving pulp. J. Ind.Microbiol. Biotechnol. 32, 211–214.

Jackson, L.S., Heitmann Jr., J.A., Joyce, T.W., 1998. Production of dissolving pulp fromrecovered paper using enzymes. Tappi J. 81, 171–178.

Janzon, R., Puls, J., Bohn, A., Potthast, A., Saake, B., 2008. Upgrading of paper-gradepulps to dissolving pulps by nitren extraction: yields, molecular andsupramolecular structures of nitren extracted pulps. Cellulose 15, 739–750.

Jayme, G., 1944. Mikro-Quellungsmessungen an Zellstoffen. Wochenbl. Papierfabr.6, 187–194.

Köpcke, V., Ibarra, D., Ek, M., 2008. Increasing accessibility and reactivity of paper-grade pulp by enzymatic treatment for use as dissolving pulp. Nord. Pulp Pap.Res. J. 4, 363–368.

Krässig, H.A., 1993. Cellulose – Structure, Accessibility and Reactivity. Yverdon,Gordon and Breach Science Publishers, Yverdon.

Kvarnlöf, N., Germgård, U., Jönsson, L., Söderlund, C.A., 2006. Enzymatic treatmentto increase the reactivity of a dissolving pulp for viscose production. Appita J.59, 242–246.

Moore, G., 1996. Nonwood Fiber Applications in Papermaking. Pira International,Leatherhead, Surrey, UK.

Puls, J., Janzon, R., Saake, B., 2006. Comparative removal of hemicelluloses frompaper pulps using nitren, cuen, NaOH, and KOH. Lezinger Berichte. 86, 63–70.

Rabinovich, M.S., Melnik, M.S., Bolobova, A.V., 2002. The structure and mechanismof action of cellulose enzymes. Biochemistry 67, 850–871.

Rahkamo, L., Viikari, L., Buchert, J., 1998. Enzymatic and alkaline treatments ofhardwood dissolving pulp. Cellulose 5, 79–88.

Ramos, P.L., Zandoná Filho, A., Deschamps, F.C., Sadler, J.N., 1999. The effect ofTrichoderma cellulases on the fine structure of a bleached softwood kraft pulp.Enzyme Microb. Technol. 24, 371–380.

Schenzel, K., Fischer, S., 2001. NIR FT Raman spectroscopy – a rapid analyticaltool for detecting the transformation of cellulose polymorphs. Cellulose 8,49–57.

Sixta, H., 2006. Handbook of Pulp. Wiley-VCH Verlag GmbH & Co, KGaA, Weinheim.Stone, J.E., Scallan, A.M., 1968. The effect of component removal upon the porous

structure of a cell wall of wood. Part III. A comparison between the sulphite andkraft processes. Pulp Paper Mag. Can. 6, 69–74.

Stone, J.E., Scallan, A.M., Donefer, E., Ahlgren, E., 1969. Digestibility as a simplefunction of a molecule of similar size to a cellulose enzyme. In: Hajny, G.J.,Reese, R.T. (Eds.), Cellulases and their Applications. American Chemical Society,Washington, DC, pp. 219–241.

Suurnäkki, A., Heijnesson, A., Buchert, J., Tenkanen, M., Viikari, L., Westermark, U.,1996. Location of xylanase and mannanase action in kraft fibers. J. Pulp Pap. Sci.29, 319–324.

Suurnäkki, A., Li, T.-Q., Buchert, J., Tenkanen, M., Viikari, L., Vuorinen, T., Ödberg, L.,1997. Effects of enzymatic removal of xylan and glucomannan on the pore sizedistribution of kraft fibers. Holzforschung 51, 27–33.

Teleman, A., Larsson, P.T., Iversen, T., 2001. On the accessibility and structure ofxylan in birch kraft pulp. Cellulose 8, 209–215.

Theander, O., Westerlund, E.A., 1986. Studies on dietary fiber. 3. Improvedprocedures for analysis of dietary fiber. J. Agric. Food Chem. 34, 330–336.

Treiber, E., 1985. Formation of fibers from cellulose solutions. In: Nevell, T.P.,Zeronian, S.H. (Eds.), Cellulose Chemistry and its Applications. Ellis HorwoodLtd., Chichester, pp. 456–457.

van Dam, J.E.G., van Vilsteren, G.E.T., Zomers, F.H.A., Shannon, W.B., Hamilton, I.T.,1994. Increased Application of Domestically Produced Plant Fibers in Textile,Pulp and Paper Production and Composite Materials. European Commission, DGXII, Science Research and Development, Brussels.

Wallis, A.F.A., Wearne, R.H., 1990. Chemical cellulose from radiate pine kraft pulp.Appita J. 43, 355–357. 366.

Wong, K.Y., Deverell, K.F., Mackie, K.L., Clark, T.A., Donaldson, L.A., 1988. Therelationship between fiber porosity and cellulose digestibility in stream-exploded Pinus radiata. Biotechnol. Bioeng. 31, 447–456.

Ye, D., Farriol, D., 2005. Improving accessibility and reactivity of celluloses of annualplants for the synthesis of methylcellulose. Cellulose 12, 507–515.