O R IG IN A L

Fractionation of organic substances from the SouthAfr ican Eucalyptus grandis biomass by a combinationof hot water and mild alkaline tr eatments

Jonas K. Johakimu1 • Andrew Jerome1 •

Bruce B. Sithole1,2 • Lekha Prabashni1

Received: 11 April 2015Ó Springer-Verlag Berlin Heidelberg 2015

Abstract An alternative way of fractionating lignocellulose biomass into itsindividual components, hemicelluloses, lignin and cellulose, was investigated.South African Eucalyptus grandis wood chips were fractionated using a combina-tion of hot water and alkaline treatments with or without AQ. Initially, the biomasssamples were treated in hot water to remove hemicelluloses. At optimum pre-fraction conditions, the data acquired revealed that almost 12 % of the E. grandiswood biomass could be recovered as hemicelluloses. When the hemicelluloses pre-extracted biomass was further treated using sodium hydroxide with or without AQ,the data indicated that the amount of lignin and cellulose that could be recoveredwas 22 and 36 %, respectively (as % of the wood mass). The substrate was char-acterised by a higher amount of a-cellulose (91–93 %), lower kappa no (12–13),viscosity (327–450 g mg/L) and DP (1078–1536). It was then presumed that suchpulp could meet end-user requirement of the dissolving pulps. Industrial acceptanceof this biomass fractionation concept, however, will further require carefulassessments of various options for treating and purifying the hemicelluloses andlignin in their respect streams.

& Jonas K. Johakimujjohakimu@csir.co.za

1 Forestry and Forest Products Research Centre, Council of Scientific and Industrial Research ofSouth Africa, P.O. Box 17001, Congella, Durban, South Africa

2 Discipline of Chemical Engineering, University of Kwazulu-Natal, Mazisi Kunene Rd,Glenwood, Durban 4041, South Africa

123

Wood Sci TechnolDOI 10.1007/s00226-015-0764-2

Introduction

Wood biomass is abundant not only in South Africa but also in many other countriesof the world. This renewable resource represents an alternative and cheap feedstockfor production of value-added products such as chemicals and biofuels (Hamelincket al. 2005; Ohman 2006). Typical wood biomass contains valuable organicsubstances, notably cellulose, hemicelluloses and lignin (Christensen 1998; Girioet al. 2008). These organic substances, with appropriate fractionation technologies,can be separated and each individual component can be converted to valuableproducts similar to those obtained from fossil oil in the petroleum industry(Hamelinck et al. 2005; Ohman 2006). This implies that if appropriate woodbiomass fractionation strategies are implemented, these organic substances maymake an important contribution in helping relieve the world’s dependency on fossiloils.

In current practices, only cellulose is exploited for value-added applications, e.g.wood-based pulp production in pulp and paper industry. Consequently, there arelimited economical down streams for processing hemicelluloses or lignin to value-added products (Ohman 2006; Girio et al. 2008). Furthermore, implementation ofthe biorefinery concept based on the existing pulp and paper industry seems to bedependent on specific mill operating conditions and also may be limited to a fewmills. This has necessitated the need for new strategies that will enable optimalutilisation of the organic substances present in the lignocellulosic biomass (Zhenget al. 2009; Europe 2020 strategy, AFORE programme 2014). It is thereforeenvisaged that the success of such initiatives will open new economic streams as thewood biomass will be accessible not only in the pulp and paper sector, but also inother economic sectors such as in the chemical industries. For example, thecellulose rich stream can be used for specialty pulp applications, nano-crystallinecellulose, lactic acid, ethanol etc., whereas the hemicelluloses can be used inproduction of specialty chemicals (e.g. oxygen barriers for food packaging, xylitoland furfural or furfural derivatives), and/or biofuel (butanol or ethanol). Lignin haspotential applications in carbon fibre, dust suppressants, phenolic resins technolo-gies, etc.

Wood biomass, however, has complicated structural and compositional featureswhich negatively affect the yields and the purities of these organic substances uponseparation (Hamelinck et al. 2005; Girio et al. 2008; Xiao et al. 2011). For instancein the cell wall, lignin forms a shield around the hemicelluloses and cellulose. Inaddition, part of the lignin is covalently linked to the carbohydrates forming lignin-carbohydrates complexes (LCC). Since the transformation of these organicsubstances into value-added products such as biofuel or chemicals are largelydependent on their purity (Mosier et al. 2005; Zheng et al. 2009; Yoon and VanHeiningen 2010), it is necessary that a high level of selectivity is achieved duringbiomass fractionation, which in turn will guarantee acceptable yields and purities ineach fractionated stream.

The main challenge has been developing technologies that are technicallyeffective and economically viable on an industrial scale (NNFCC Project No 10/003

Wood Sci Technol

123

2009). Typically, in existing pulping technologies such as acid bisulphite and kraftprocesses, sulphur compounds are incorporated in chemicals used to break up thewood materials, resulting in the production of spent liquor that is contaminated withsulphur. Spent liquor contaminated with sulphur is an environmental nuisance.Moreover, organic substances recovered from such system have limitation inconversion to value-added products. For example, lignin cannot be post-processedin systems that involve catalysts (NNFCC Project No 10/003 2009). Several studieswere conducted to find alternative sulphur-free cooking liquor. Amongst othersthese alternative technologies include organosolv and alkaline cooking liquors thatare fortified with pulping aids (Christensen 1998; Testova et al. 2014). Such pulpingaids include anthraquinone (AQ), NABH4, surfactants, urea. However, at industrialscale, only soda ? AQ pulping is widely applied. Its application, however, iscurrently limited to cellulose pulp production (Christensen 1998; Testova et al.2014).

It is therefore important that the potential of the soda ? AQ process alkaline isexplored in the context of biomass fractionation to recover all organic substance.This biomass fractionation strategy could be easily implemented using capitalequipment currently used in pulp and paper mills. This would be an advantage overthe other competing technologies, e.g. organosolv technologies. However, biomassfractionation strategies involving soda ? AQ process would need to differ from thatused in cellulose pulp production. This is because unit operations involvingrecovering and purifying the organic substances in their respective streams need tobe integrated with the traditional soda ? AQ pulping process. Furthermore, thepotential application of the products delivered from such process needs to bedefined.

Previous studies have reported on different biomass pre-treatment strategies priorto processing wood biomass for either ethanol or pulp production. When hot water isused for pre-treatment, the cell walls in the wood biomass are disrupted andconsequently the pre-treated biomass responds favourably in the subsequentdownstream processes (Cheng et al. 2010; Johakimu and Andrew 2013). Moreimportantly, during hot water pre-treatments a high proportion of the hemicellulosescan also be solubilised, this is because hemicelluloses are easily degraded intooligomeric and monomeric sugars, whereas lignin and cellulose remain insoluble(Christensen 1998; Sixta and Schild 2009). Thus, hot water pre-treatment can alsobe used to remove and recover hemicelluloses which can be used as a separateeconomic stream. Nabarlatz et al. (2007) and Boussarsar et al. (2009) studied theinfluence of hot water treatment processes on treating biomass destined for ethanolproduction and found that substantial amounts of hemicelluloses could be removedfrom the biomass. Similarly, Tunc and Van Heiningen (2008) studied the extractionof hemicelluloses from wood biomass prior to pulp production; however, thisstrategy was not beneficial due to excessive pulp yield loss in the hemicellulosesextracted pulps. This is due to the fact that conditions known to promote extractionof hemicelluloses from wood also facilitate the degradation of cellulose during post-treatments in alkaline liquor (Tunc and Van Heiningen 2008; Garcıa et al. 2011). Toavoid this drawback, it is imperative that the alkaline cooking liquors used in thepost-hemicellulose extraction are fortified with cellulose protector.

Wood Sci Technol

123

Thus, instead of processing the remaining solid residue into either ethanol or pulpas done in previous studies, the lignin and cellulose rich stream can further befractionated using soda ? AQ process. Soda ? AQ process is one of the alkalinesulphur-free technologies in which the sodium hydroxide is used as the source of thecooking liquor. The role of AQ is to protect cellulose against the peeling reactions(Christensen 1998; Gullichsen and Fogelholm 1999). During soda ? AQ treat-ments, lignins are solubilised and most of cellulose can be retained (Christensen1998). The dissolution of lignin in alkaline solvent involves cleavage of the esterbonds (Christensen 1998; Gullichsen and Fogelholm 1999). At the same time,cellulose swells and the degree of polymerisation and crystallinity are reduced. It isalso worthwhile to note that soda ? AQ process requires high alkaline concentra-tions and as a result cause cellulose cleavage reactions that are detrimental to thecellulose yields. It is therefore envisaged that a process that avoids high alkalineconcentrations whilst maintaining effective removal of lignin would yield substratethat is richer in cellulose. Previous studies on soda ? AQ pulping have shown thatlimiting the alkali concentration to about 20 g/L as Na2O during the cook offersgood retention of cellulose (Christensen 1998; Gullichsen and Fogelholm 1999;Sturgeoff and Pitl 1993). Alternatively, mild treatments, e.g. using lower cookingtemperatures can also be used to minimise the negative effect on the cellulose yield(Christensen 1998; Sturgeoff and Pitl 1993).

In the present work, fractionation of the lignocellulose biomass into its individualcomponents hemicelluloses, lignin and cellulose using a combination of hot waterand mild alkaline treatment was investigated. South African Eucalyptus grandiswood chips were used as source of raw materials. Initially, the biomass sampleswere treated in hot water to recover the hemicelluloses, and this process was called apre-fractionation process. In the subsequent process ‘‘post pre-fractionation’’ stage,separation of lignin and cellulose from the hemicelluloses pre-extracted biomasswas accomplished by performing mild alkaline treatments using sodium hydroxidewith or without AQ. Mild alkaline treatment was preferred as a means ofminimising the negative effects of the soda ? AQ process. As it will be seen later,the alkaline solution and the cooking temperature were kept in the lower range. Theproposed concept is based on the existing pulping technologies. However, it is acompletely new biomass processing strategy with the objective of recovering allorganic substances and beneficiate into various valuable products.

Mater ials and methods

Mater ials

Eucalyptus grandis woodchips obtained from a kraft pulp mill in South Africa wereused in this study. The woodchip samples were screened using a vibrating screen toremove under- and over-sized chips, knots, and bark. Air-dried chips with anaverage thickness of 3–8 mm were collected and stored in plastic bags for thesubsequent experiments. Chip moisture contents were determined according toTAPPI method T258 om-94.

Wood Sci Technol

123

In order to determine the chemical composition of the wood chips, they wereground into sawdust in the range of 40–60 meshes using a Wiley mill. Klason ligninwas determined according to TAPPI method T222 om-88. The sugars were analysedusing HPLC as reported in previous studies (Cheng et al. 2010; Johakimu andAndrew 2013). The data were used for evaluating the mass balance and/or therecovery rate of the organic substances in the various fractionated streams.

Pre-fractionation exper iments

Hot water treatments

In these experiments, water was used as the solvent. All the experiments wereperformed in a 7-L rotating digester. Wood chips were charged in digester at a rateof 600 g (oven dried equivalent mass) per each run. The water to wood ratio wasmaintained at 4.5:1. Thereafter, the digester was electrically heated up to themaximum treatment temperature of 170 °C at a constant ramping rate of 1.6 °C/min. The reaction times at maximum temperature were varied between 15 and90 min.

At the end of each pre-fractionation experiment, free spent liquor was drained outand collected from each experiment. A portion of the spent liquor was collected andstored at 4 °C until required. The determination of sugars in the extract wasperformed using HPLC. Prior to analysis, the pH of the extract was first adjusted tobetween 5 and 6 using 6 mol/L HCl, and then the sugars were hydrolysed viaheating with 4 % H2SO4 at 121 °C for 1 h. For sugar yield calculations, the HPLCresults for arabinose, galactose, glucose, xylose, and mannose were corrected forarabinan, galactan, glucan, xylan, and mannan (Yoon and Van Heiningen 2010;Janson 1974). The measured extract volume was used to estimate the amount ofhemicelluloses extracted. This was based on the monosugars contents (mg/L) whichwere determined by HPLC analysis:

C ¼ Glu �162180

�Man

b�162180

ð1Þ

where b = 1.6 (as an average value for number of mannose units per glucose inhemicelluloses of hardwood), Glu = concentration of glucose (mg/L) andMan = concentration of mannose (mg/L) as determined by HPLC analysis.

H ¼ ðArab þ xylÞ�132150

þ ðGal þ Glu þ ManÞ�162180

� C ð2Þ

Arab ? xyl = sum of the concentration of arabinose and xylose (mg/L), andGal ? Glu ? Man = sum of the concentration of galactose, glucose and mannose(mg/L) as determined by HPLC analysis.

All the remaining solid residues (pre-treated woodchips) were defiberised using adisc refiner equipped with defibration plates. The defiberised pulp samples werespin-dried to remove excess water, weighed, and stored in plastic bags at 4 °C untilfurther use. The wood loss was evaluated gravimetrically. The solid residue

Wood Sci Technol

123

obtained after performing the pre-fractionation was expressed as a percentage of thewood mass charged into the digester (on an oven dry basis). Sugars and Klasonlignin in the remaining solid residue were analysed using similar procedures asdiscussed previously.

Post pre-fractionation experiments

Mild alkaline treatments

Selected pre-hemicelluloses extracted samples that gave a relatively higher yield ofhemicelluloses were chosen for the post pre-fractionation studies. All post pre-fractionation experiments were performed in the same rotating digester that wasused for the pre-fractionation stage. The only exception was that the free spentliquor from each pre-fractionation experiment was drained out prior to performingthe post pre-fractionation experiment. A sodium hydroxide solution of 110 g/L (asNa2O) was prepared and used as the source of alkaline liquor. These samples werethen treated at 150 °C using various sodium hydroxide dosages (12–20 %) with orwithout AQ. The AQ was added to the alkaline liquor and the dosage wasmaintained at 0.08 % (on oven dried woodchips weight). The AQ in a dispersedform was supplied by Buckman laboratory SA. The liquor to wood ratio wasmaintained at 4.5:1, whereas a constant reaction time of 2 h was used for allexperiments.

At the end of each post pre-fractionation experiment, a portion of the spent liquorwas collected and stored at 4 °C until further use. Thereafter, the pulp slurry waswashed using deionized water, spin-dried and also stored at 4 °C until required. Theamount of sugars and lignin in the spent liquor and in the remaining solid residuewas quantified using the same protocols as discussed earlier during the pre-fractionation experiments. However, the carbohydrate compositions were notcorrected with the Janson formula. In addition, substrate properties; kappa, viscosityand solubility were characterised using TAPPI standard methods. In particular, thesolubility data in terms of degraded cellulose were used to calculate the content ofthe cellulose retained (a-cellulose). The degree of polymerisation (DP) wascalculated as follows (Sun et al. 2005): DP0.9 = 1.65X [g] mg/L, where g refers toCED viscosity CED values.

Lignin was recovered from the extract by acidic precipitation. The extract wasfirst filtered to remove suspended solids. Thereafter, a 50 mL portion of the filtratefrom each sample was transferred into a 150-mL beaker. The sample was stirred at aconstant rate using a magnetic stirrer plate. A few drops of 6 M sulphuric acid wereadded until the desired pH was reached. The reaction time was set for 2 h. Thesolution was then filtered and washed with deionized water. The weight ofprecipitated lignin was determined gravimetrically, by calculating the differencebetween the mass of lignin obtained after precipitation and mass of dried lignin. Theresulting mass was thereafter used to determine the concentration of lignin in theextract (i.e. lignin concentration (g/L) = (X0 - Xd)/VL), where X0 and Xd are themass of lignin (in g) before and after being dried, respectively, and VL is the volumeof filtrate used, in L.

Wood Sci Technol

123

Structural changes in substr ate biomass and lignin

Structural changes in the substrate biomass and lignin were characterised using fieldemission gun scanning electron microscopy (FEG SEM) and FT-IR Spectrometer.FEG SEM tests were limited to untreated, pre-treated and post-treated samples. Theoven dried samples were mounted onto aluminium stubs using carbon tape. Thesamples were sputter coated with gold using a Polaron SC500 sputter coater.Thereafter, the gold-coated samples were viewed at 5 kV using a Carl Zeiss FEGSEM. FT-IR tests were also conducted on the same samples as in FEG SEM, theonly exception was that recovered lignin was also analysed. The FT-IR spectra wererecorded from a Nicolet NETXUS 670 FTIR spectrometer. Samples were pre-treated by tabletting the mixture of sample and KBr into a very thin film with adiameter of 5 mm. All the spectra were recorded in the absorbance mode from 4000to 400 cm- 1 at room temperature.

Results and discussion

Pre-fractionation of hemicelluloses

Effects of pre-fractionation conditions on wood loss and pH of the resulting extracts

The wood loss and the pH of the resulting extracts are given in Table 1. The resultsshowed that an increase in the severity of the pre-fractionation conditions byincreasing the treatment time resulted in progressively increasing wood losses.Wood loss was in the range between 16 and 27 % for the pre-fractionationconditions examined in this study. Longer pre-fractionation time resulted in higherwood losses, indicating that more carbohydrates and lignins were solubilised.

As reported in previous studies (Nabarlatz et al. 2007; Tunc et al. 2010), theresulting extracts were acidic (i.e. pH 3.0–3.2). This is because during hot watertreatment, hydrolysis of the acetyl groups in the hemicelluloses occurs (Tunc andVan Heiningen 2008; Tunc et al. 2010). Hydrolysis of acetyl groups leads toformation of acetic acid which causes the pH of the resulting extract to drop toacidic levels. It has been shown in previous studies that acidic conditions lead to thedegradation of carbohydrates and can result in excessive wood losses (Xiao et al.2011). However, when pH buffering reagents are added to the water during hotwater treatment, acid neutralisation occurs. As a result, using the appropriatebuffering dosage, the pH can be maintained at a near neutral pH level (Mosier et al.

Table 1 Wood loss and pH ofthe resulting extract Treatment time (min) Wood loss (%) pH

15 15.7 ± 0.0 3.2 ± 0.245 24.0 ± 1.0 3.1 ± 0.060 25.2 ± 0.2 3.0 ± 0.490 27.1 ± 0.9 3.0 ± 0.0

Wood Sci Technol

123

2005). However, the hemicelluloses yield is significantly lower when compared tohot water treatment (Mosier et al. 2005; Sixta and Schild 2009).

Effects of pre-fractionation conditions on dissolution of the wood organicsubstances

To ascertain the effects of the pre-fractionation conditions on these organicsubstances, chemical composition of raw wood and pre-fractionated wood sampleswere performed and the results are shown in Table 2, expressed as a percentage ofthe original wood weight.

As anticipated, most of the short-chain carbohydrates (hemicelluloses) weresolubilised during pre-fractionation; an indication that hemicelluloses depolymerizemuch faster than lignin and cellulose in hot water. As discussed earlier, similarfindings have been reported in previous studies (Tunc and Van Heiningen 2008).However, in those studies, hot water treatment processes were used to treat biomassdestined for either ethanol production or extraction of hemicelluloses from woodbiomass prior to pulp production.

Therefore, despite the fact that different approaches were used, hot watertreatment shows uniqueness in removing hemicelluloses. This finding appears tosuggest that biomass fractionation strategies could be designed to use this advantageby removing the hemicelluloses in earlier stages of the biomass fractionationprocess. Indeed, this practice has been used in production of ‘‘hemicelluloses-free’’dissolving pulps where a hot water process step ‘‘steam pre-hydrolysis’’ is employedspecifically for removing hemicelluloses (Christensen 1998). It is also worthwhile tonote that the removal of hemicelluloses from wood requires that covalent (ester andether) bonds which link the hemicelluloses mostly to lignin are broken out. Incontrast, it is only the ester bonds that are easily cleaved, whilst the ether linkagesremain stable (Christensen 1998). This implies that it is almost impossible toremove all the hemicelluloses in pure form.

Table 2 Chemical composition of the wood residue, % of original wood mass

Treatmenttime (min)

Sugars Totalsugars

Lignin

Arabinan(%)

Galactan(%)

Glucan(%)

Xylan (%) Mann(%)

Klasonlignin (%)

Wood 0.17 ± 0.1 1.05 ± 0.0 51.4 ± 0.3 11.9 ± 0.1 2.2 ± 0.0 66.7 ± 0.1 26.0 ± 0.415 \ DL \ DL 47.9 ± 0.4 7.1 ± 0.0 1.1 ± 0.2 56.1 ± 0.7 22.5 ± 0.045 \ DL \ DL 49.5 ± 0.7 3.3 ± 0.3 0.7 ± 0.1 53.5 ± 0.3 20.8 ± 0,260a \ DL \ DL 50.1 ± 0.2 2.6 ± 0.1 0.2 ± 0.0 53.0 ± 0.1 20.6 ± 0.390 \ DL \ DL 49.8 ± 0.3 2.3 ± 0.5 0.4 ± 0.0 52.5 ± 0.2 20.2 ± 0.5

\ DL denotes less than detection limitsa This sample was selected for the subsequent ‘‘post pre-fractionation studies’’

Wood Sci Technol

123

In particular, the arabinan and the galactan in pre-fractionated wood sampleswere not detected by HPLC, indicating that these hemicelluloses where completelyremoved from the wood and dissolved in extract. It can also be seen that the removalof wood sugars was more pronounced for the hemicelluloses (xylan and mannan)than for cellulose (glucan). The hemicelluloses solubilisation reached a maximumwhen a pre-fractionation time of 60 min was applied, i.e. approximately 82 % of theoriginal hemicelluloses content was solubilised.

At this pre-fractionation condition, the cellulose retained in the pre-fractionatedwood biomass was almost 98 % of the original cellulose content, whereas theamounts of lignin removed were approximately 21 % of the original content or 6 %of the original wood mass. Presumably, some of the lignin was depolymerized andgave rise to phenolic compounds that are soluble in water (Cheng et al. 2010).Therefore, although the pre-fractionation process stage may only be intended toremove the hemicelluloses up front from the wood biomass, in practice, it may benecessary to recover the lignin. The amount of lignin removed, however, isrelatively small when compared to the amounts that remained in the pre-fractionatedwood. This could be an advantage, as it could simplify the downstream operationrequired for purification/separation of the lignin from the hemicelluloses stream.

It can also be observed that solubilisation of the hemicelluloses was progressivelyas the severity of the pre-treatment conditions increased and reached the maximumlevel at 60 min. Thereafter, at longer treatment time (90 min) the concentration ofhemicelluloses started to drop. It is presumed that longer treatment time leads todegradation of some of the hemicelluloses solubilised in the extract. This isespecially true if one takes into consideration the acidic nature of the extract.Extract under acidic condition can lead to degradation of hemicelluloses andcellulose (Hamelinck et al. 2005; Girio et al. 2008). According to previous studies(Sixta and Schild 2009), hot water treatments produce extracts that are characterisedby hemicelluloses which are in oligomer and monomeric form. However, the type ofsugar present in extract was not investigated in this study.

During pre-fractionation, most of the hemicelluloses are solubilised and may befurther degraded and, as a result cannot be recovered. Therefore, it was necessary todetermine the actual hemicelluloses that could be recovered from the extract. Themethod proposed by Janson (1974) was used to establish this data. The results for

Table 3 Chemical composition in the extract, as % of original wood mass

Treatment time (min) Carbohydrates

Cellulose (%) Hemicelluloses (%) Total carbohydrates (%)

15 0.64 6.30 6.9045 0.60 9.10 9.7060a 0.50 11.70 12.2090 0.40 10.20 10.60a This sample was selected for the subsequent ‘‘post pre-fractionation studies’’

Wood Sci Technol

123

the chemical composition in the extracts after the various pre-fractionationtreatments are summarised in Table 3. The sugars dissolved in the extracts weremostly dominated by the hemicelluloses. This trend supported the sugar yield dataobserved in the hemicelluloses pre-extracted wood samples, in which most of thecellulose were retained (Table 2).

The highest hemicelluloses yield obtained was 11.7 % of the wood mass. This isequivalent to a recovery rate of 76 % (based on original hemicelluloses content inthe wood). Furthermore, the purity, defined as the ratio between hemicellulosesrecovered, was determined (Table 3) and the total amount of the hemicellulosessolubilised in the extract (Table 2). The purity was approximately 93 %. Impuritiescontent referred to the solubilised chemical compounds other than hemicelluloseswhich may include degraded hemicelluloses products (Hamelinck et al. 2005; Girioet al. 2008), such as furfural, hydroxymethyl furfural (5-HMF).

The results for the hemicelluloses recovered were plotted against the pre-fractionation time (Fig. 1). It can be seen that the hemicelluloses recovery rateincreased when pre-fractionation time was increased from 15 to 60 min andthereafter started to decrease. This may be attributed to the acidic nature of theextract that favours the degradation of the solubilised hemicelluloses (Hamelincket al. 2005; Girio et al. 2008). As a result fewer amounts of hemicelluloses wereavailable for recovering. For this reason, the pre-fractionation time of 60 min wasselected as the optimal fractionation time at 170 °C and was used to preparesamples for the post pre-fraction studies.

It is worthwhile to mention that practically, the amount of hemicellulosesextracted can be separated from the spent liquor (extract) by using a process such asmembrane filtration. Sixta and Schild (2009) demonstrated that the ultrafiltrationmembrane with a pore size of 10 kDa, operating at 40 °C and a pressure of200–800 kPa can be used to isolate and recover the hemicelluloses. In the currentscale-up trials, this method was adopted.

1

3

5

7

9

11

13

15

10 25 40 55 70 85 100 115Hem

icel

lulo

se r

ecov

ery

rate

(%)

Pre-fractionation time (Min)

Fig. 1 Relationship between hemicelluloses recovery rate and pre-fractionation time

Wood Sci Technol

123

Post pre-fractionation

Because substantial amounts of hemicelluloses were removed from the woodbiomass during the pre-fractionation treatments, in the subsequent experiments thestrategy was to achieve effective separation of lignin and cellulose.

Effects on substrates properties after post pre-fractionation

The substrate yield was defined as the ratio between wood solid residues recoveredafter post pre-fractionation and the total amount of the wood originally charged intothe digester. The biomass yield preserved after post pre-fractionation (hot watertreatment followed by mild alkaline treatment [with or without AQ)] was in therange between 35 and 39 %, whereas the kappa no was in the range of 36–9,respectively.

As shown in Table 4, an increase in the sodium hydroxide dosage resulted in adecrease in substrate yield as well as the kappa no, an indication that more lignin,residual hemicelluloses and some cellulose were degraded from the hemicellulosespre-extracted wood biomass. Surprisingly, the kappa no. attained appears to beslightly higher than expected at a yield range of 35–40 %. Presumably, hot watertreatments resulted in formation of condensed residual lignins. Condensed lignincould be formed as a result of the induced effects of the auto-hydrolysis, as the pHdropped to extremely lower values, i.e. all the pH were \ 10 (Table 1) requiredpreventing the lignin condensation reactions (Christensen 1998; Gullichsen andFogelholm 1999). Condensed lignins are known to be more resistant to deligni-fication during alkaline treatments (Christensen 1998).

Table 4 Substrate yield obtained after post-treatments of the pre-hemicelluloses extracted wood sam-ples, expressed as % of original wood mass

Sample Yield (%) Kappa no Viscosity g (mL/g) DP a-Cellulose (%)

Wood 100 NA NA NA NAHT60 75 ± 1.8 NA NA NA NAHT-MA12 37.7 ± 0.7 36 ± 0.9 NA NA NAHT-MA12 ? AQ 38.3 ± 0.2 23 ± 0.4 NA NA NAHT-MA16 35.9 ± 0.1 15 ± 0.2 327 ± 2.2 1078 91 ± 0.5HT-MA16 ? AQ 36.3 ± 0.2 12 ± 0.5 450 ± 1.5 1536 93 ± 0.2HT-MA20 35.1 ± 0.4 13 ± 0.1 NA NA NAHT-MA20 ? AQ 35.1 ± 0.2 9 ± 0.5 NA NA NA

HT60 denotes hot water pre-fractionated sample at 60 min which was subsequently post-fractionated.HT-MA denotes hot water treatment followed by mild alkaline treatment, whereas the numbers 12, 16and 20 refer to sodium hydroxide dosage used with or without AQ. Samples produced using 8 % NaOHwith or without AQ were excluded because a substantial amount of the material remained uncooked.Presumably, this dosage was insufficient and as a result the delignification reaction was limited. NAdenotes that no tests were done. DP denotes degree of depolymerisation

Wood Sci Technol

123

The effect of AQ on protecting the cellulose against degradation during alkalinetreatments was also observed when the sodium hydroxide dosage applied waslimited to the range between 12 and 16 %. At these levels all samples producedusing sodium hydroxide liquor ? AQ, exhibited an increase in substrate yield toalmost 1 %. Considering that the residual lignins in the AQ samples were not at thesame level as in the pulp samples produced without AQ (Table 5), the amount ofhemicelluloses in substrate was almost negligible (0.3 %). Furthermore, the amountof hemicelluloses in substrate was almost negligible (0.3 %). This result tended tosuggest that although AQ samples retained more cellulose, but at the same timeretained relatively less amount of lignin. This is justifiable by the fact that althoughalkaline treatment with AQ showed the amount of cellulose retained was in therange of 1–2 %, the substrate yields were only increased by 1 %. It is therefore clearthat the discrepancy in mass balance yield data was attributed to the amount oflignin that also was retained in the substrates.

To reveal the impact on carbohydrates, the viscosity and a-cellulose propertieswere further investigated; however, these tests were limited only to HT-MA16 andHT-MA16 ? AQ samples. As it will be seen later, these samples were identified tobe produced at the optimum post-fractionation conditions (Table 5). It can also beseen that when AQ was applied, pulp samples produced were characterised by areasonable value of viscosity and retained a higher amount of a-cellulose withrelatively high purity, i.e. contained less than 0.27 % of xylan (Table 5). The degreeof polymerisation was also relatively reasonable (1536). These data also support thesubstrate yield and cellulose data in Tables 4 and 5, respectively. It was thenpresumed that such pulp could meet the end-user requirements of the dissolvingpulps (Sixta and Schild 2009).

Table 5 Chemical composition after post-treatment of the pre-hemicelluloses extracted wood, as %wood

Sample Sugars Lignin

Arabinan Galactan Glucan Xylan Mannan

Wood 0.17 1.05 51.4 11.9 2.2 26HT60 \ DL \ DL 50.1 2.6 0.2 20.6HT-MA12 \ DL \ DL 33.0 0.6 \ DL 3.0HT-MA 12 ? AQ \ DL \ DL 35.0 0.3 \ DL 1.8HT-MA16 \ DL \ DL 32.4 0.57 \ DL 1.7HT-MA16 1 AQ \ DL \ DL 34.4 0.27 \ DL 0.6HT-MA20 \ DL \ DL 32.0 0.5 \ DL 1.3HT-MA20 ? AQ \ DL \ DL 33.2 0.5 \ DL 0.3

HT-MA denotes hot water treatment followed by mild alkaline treatment, whereas the numbers; 12, 16and 20 refer to sodium hydroxide dosage used with or without AQ. Bold samples were used for ligninprecipitation studies\ DL indicates less than detection limits

Wood Sci Technol

123

Effects on chemical composition after post pre-fractionation

Post pre-fractionated solid residue samples contained predominantly cellulose witha small percentage of lignin remaining. This can clearly be seen from analysis of thechemical composition data in Table 5. The cellulose retained in the biomass was inthe range of 62–70 % of the original cellulose content. As anticipated, alkalinetreatment conditions with AQ exhibited a slight increase in the amount of celluloseretained (1–2 % higher than in alkaline treatment performed without AQ). At thesame time more lignin was solubilised when AQ was applied (40–65 % higher thanin the alkaline treatment performed without AQ).

The lignin solubilised in the extract was in the range between 90 and 99 % of theoriginal content, indicating that most of the a-benzyl ether linkages between ligninand hemicellulose were cleaved during the post-alkaline treatments (Pandey 1999;Yang et al. 2014). It appears that a substantial amount of cellulose can be retained,whilst at the same time more lignin can be solubilised when the sodium hydroxidedosage is maintained at 16 %. It is also very clear that lignin removal was alsoassociated with cellulose degradation. The cellulose yield loss during alkalinetreatment might also contribute to the induced effect of the auto-hydrolysistreatments (Tunc et al. 2010; Mosier et al. 2005). Auto-hydrolysis has been reportedto create a whole flock of new end reducing groups (Tunc et al. 2010).Subsequently, during the alkaline treatments severe peeling reaction occurs. Itwas therefore presumed that these reactions attributed to the cellulose loss observed.

Generally, the result suggests that alkaline treatment fortified with AQ may be aninteresting technique for improving the selectivity during the separation of ligninand cellulose. However, the application of AQ as pulping aids may be restricted inthe near future (Testova et al. 2014). Therefore, it may be of interest to usealternative pulping aids such as urea and surfactants as a substitute of AQ in theproposed biomass fractionation concept.

Changes in the structural substrate biomass

To reveal the structural changes, unfractionated, pre-fractionated (HT60) andalkaline post-fractionated samples (HT-MA16 and HT-MA16 ? AQ) were sub-jected to FEG SEM and FT-IR tests. FEG SEM and FT-IR have been shown to bereliable and versatile tools to investigate the structural changes in biomass, i.e.during biomass fractionation (Corrales et al. 2012; Sun et al. 2005; Yang et al.2014). The FEG SEM micrographs in Fig. 2a–h show the structural changes in E.grandis wood fibres for unfractionated, pre-fractionated and post-fractionatedsubstrate samples. Unfractionated fibres showed a rigid and compact fibre wallstructure (Fig. 2a, b). Upon pre-fractionation with hot water pre-treatment, debrisand residues were observed on the surface of the fibre (Fig. 2c, d). This debris couldbe described as molten lignin and some remaining hemicelluloses as discussed byXiao et al. (2011).

After the post-alkaline fractionation and irrespective of whether AQ was used ornot, no residues or debris were observed on the fibre surfaces (Fig. 2e–h). The fibresurfaces of the post-treated samples had uneven folds and appeared more porous

Wood Sci Technol

123

Fig. 2 Micrographs of E. grandis fibres a, b of unfractionated wood samples; c, d after pre-fractionationwith hot water; f, e after post pre-fractionation without AQ and g, h after post pre-fractionation with AQ

Wood Sci Technol

123

when compared with pre-fractionated and unfractionated wood fibres (Fig. 2e–h cf.Fig. 2a–d). Similar findings have been reported by Corrales et al. (2012) whostudied biomass fractionation based on steam treatment in the presence of SO2 andCO2. They found that pre-treated bagasse biomass had a more porous structure thanuntreated ones. Pre-treatment disrupts the cell wall and thus opens up the biomassmatrix. As a result, porosity or permeability increases. This phenomenon is not onlycritical for dissolution of hemicelluloses, but is also important in enhancing theremoval of lignin in the subsequent processing steps (Hamelinck et al. 2005;Nabarlatz et al. 2007).

The porous appearance of the fibre surfaces could be attributed to the removal ofthe hemicelluloses and lignin from the cell wall matrix (Fig. 2f, h). It can also beseen that the use of AQ resulted in more defibration of the fibre surfaces, exposingthe micro fibrils on the fibre surfaces (Fig. 2h). Fibrillation of the fibre wall oropening of the cell wall matrix by the use of AQ resulted in a more efficient removalof lignin. This is also supported by the chemical composition data (Table 5).

The FT-IR spectra of the biomass samples are shown in Fig. 3. The spectrum ofthe samples was quite similar, which indicates that the basic biomass structure wasnot changed during the biomass fractionation process. However, by comparing theintensity of the signals between raw wood and pre-fractionated (HT60), the O–Hstretch band (at 3500 cm- 1) appears to disappear in the pre-fractionated samples.The O–H stretch band corresponds to the aliphatic moieties in lignin andpolysaccharides (Yang et al. 2014). On the other hand, the shoulder at1742 cm- 1 is attributed to the acetyl and uronic ester groups of the hemicellulosesand appeared to disappear after the hot water pre-treatment, indicating that most ofthese bonds were cleaved (Pandey 1999). The C–O band stretch band (1000 cm- 1)

4000 3500 3000 2500 2000 1500 1000 50060

65

70

75

80

85

90

95

100

HT-MA16 + AQHT-MA16

HT60

Control (Sawdust)

Wavenumber / cm-1

Tran

smitt

ance

/ a.

u.

1742

1106

1054

Fig. 3 Impact on structural changes; raw wood, HT60-pre-fractionated sample, HT-MA16 and HT-MA16 ? AQ post-fractionated without and with AQ, respectively

Wood Sci Technol

123

shows deformation in cellulose and lignin present in the raw wood (saw dust) and inthe hot water pre-treated samples (Pandey 1999; Yang et al. 2014). The noted partialstructural changes upon hot water treatment were attributed to the induced effects ofthe hot water pre-treatments which resulted in opening up the cell structure (Chenget al. 2010), as a result partial dissolution or oxidation of lignin and polysaccharidesoccurred.

When the substrate samples collected after hot water treatments were furthertreated (post-fractionation), the O–H and C–O stretch bands were more evident,indicating deformation occurred in cellulose and lignin that remained in thesubstrate. A stretch band from the C–H (2890 cm- 1) indicating stretching vibrationof methyl and methylene units can also be seen (Pandey 1999). It may be this signalthat indicates breakdown of the lignin and it was more pronounced in the HT-MA16samples than in HT-MA16 ? AQ samples, indicating that lignin in HT-MA16 ? AQ sample was more fragmented and subsequently some lignin fractionwas dissolved. The bands in the region between 1500 and 1400 cm- 1 show thearomatic C=C stretch from the aromatic ring of lignin (Pandey 1999), whichindicated that the basic lignin structure was not much changed. It can also be seenthat the intensity at C–O stretch band (1000 cm- 1) decreased for the HT-MA16 ? AQ and HT-MA16 samples, an indication that most of the H-type ligninwas dissolved during the post-treatments. The absorbance in the region of1500–900 cm- 1 is associated with the typical cellulose values (Pandey 1999; Yanget al. 2014). Thus, it was concluded that partial structural changes occurred in thebiomass and the effect was more pronounced during the post-alkaline treatments.Furthermore, by comparing sample produced with and without AQ, again the resultconfirmed that AQ enhanced the dissolution of lignin whilst protecting cellulose.

Recovery of lignin

During post pre-fractionation most of the lignin was solubilised, but some may havebeen degraded to the extent that they cannot be recovered. Therefore, an attemptwas also made to quantify the amount of lignin that could possibly be recovered. Inprevious studies (Ohman 2006), it was shown that effective lignin precipitation maybe achieved when the pH is maintained at 10. However, these studies were based onthe precipitation of lignin from kraft pulp mill black liquor. Therefore, it was criticalto determine the optimum pH for lignin precipitation based on the approach used inthis study.

Preliminary trials were performed to determine the effect of pH on ligninprecipitation. It can be seen in Fig. 4 that pH in the range of 2–4 favoured a higherlignin recovery. Therefore, a higher acid consumption in the precipitation processmay be necessary as compared to acid precipitation carried out at pH 9, e.g.precipitation of lignin from kraft black liquor (Ohman 2006). The ligninprecipitation conditions were further optimised and a pH of 3.5 was selected forthe subsequent studies, in which the selected samples HT-MA16 and HT-MA16 ? AQ were further tested. The data presented in Table 6 indicate thatapproximately 76 % of the original lignin content could be recovered when a mildalkaline treatment with AQ is applied. The higher lignin recovery from the spent

Wood Sci Technol

123

liquor when AQ (HTMA16-AQ) was applied is due to the fact that liquor had arelatively higher amount of lignin than that which was found in the spent liquor ofthe treatment performed without AQ (HTMA16). Contrarily, pulp produced withoutAQ (HTMA16) has a relatively higher amount of lignin retained in pulp samples.This is clearly supported by the data in Table 5; for example, less lignin wasretained in HTMA16-AQ pulp samples than in HTMA16 pulp samples (0.6 vs1.7 %).

In practice, the amount of lignin in the spent liquor can be isolated using filtrationtechniques such as applied in LignoBoost process. This method has been reportedelsewhere (Ohman 2006). In the current scale-up trials, a lignin recovery processusing acidification followed by filtration process using a centrifuge instead of a filterpress has been adopted.

To reveal the structural changes, lignin recovered corresponding to theproduction of HT-MA16 and HT-MA16 ? AQ substrate samples were studied.The FT-IR spectra of the lignin samples are shown in Fig. 5. The bands in the regionof 1700, 1595, 1511 and 1455 cm- 1 correspond to aromatic ring vibrations (Yanget al. 2014).

The signal at 1700 cm- 1 indicates non-conjugated carbonyl group, either fromcarboxyl or the ester linkage of the lignin side chain (Yang et al. 2014). However,

Table 6 Recovery of lignin from the post pre-fractionation extracts

Sample Lignin concentration(mg/L) in the extract

Lignin as % of wood Lignin as % of the originalcontent in the wood

Wood n/a 29 100HT-MA16 211 16 55HT-MA16 ? AQ 290 22 76

0

5

10

15

20

25

30

35

40

45

50

1 2 3 4 5 6 7 8 9 10

Con

cent

ratio

n of

the

ligni

n (m

g/L)

pH of the extract

Fig. 4 Effects of pH during lignin precipitation. The initial pH of the extract was 13.5

Wood Sci Technol

123

the lignin recovered from HT-MA16 exhibited lower intensities than HT-MA16 ? AQ samples. It may indicate that most of these bonds in the ligninrecovered from HT-MA16 ? AQ samples were cleaved and subsequently werebroken down into lignin monomer units (Pandey 1999; Yang et al. 2014).

An examination of converting hemicelluloses, lignin and cellulose into value-added products for the wood biomass fractionated by using a combination of hotwater and mild alkaline treatment was beyond the scope of this study and will bereported elsewhere [the paper entitled production of carbon fibres, polylactic acidand cellulose nano-crystals (CNC) is still in preparation].

Conclusion

The overall objective of this study was to develop a better understanding of thefeasibility of fractionating the lignocellulose biomass into its individual compo-nents, namely hemicelluloses, lignin and cellulose using a combination of hot waterand mild alkaline treatments process steps. The approach adopted was to quantifythe recovery rate of the organic substances in their respective streams. Furthermore,the impact on structural changes was studied.

The results acquired have revealed that the amount of hemicelluloses, lignin andcellulose that could be recovered are 12, 22 and 36 %, respectively (as % of the

3500 3000 2500 2000 1500 1000 50050

60

70

80

90

100

Tran

smitt

ance

/ a.

u.

Wavenumber / cm-1

15951511

1455

14221322

HT-MA16

HT-MA16 + AQ

1212

1153

1109

1028

1700

Fig. 5 Impact on recovered lignin structural changes: HT-MA16 and HT-MA16 ? AQ post-fractionatedwithout and with AQ, respectively

Wood Sci Technol

123

wood mass), indicating that almost 70 % of the wood biomass could be recoveredand used economically. The substrate was characterised by a higher amount of a-cellulose, lower kappa no and the viscosity and DP values obtained were also in avery reasonable range. It was presumed that such pulp could meet end-userrequirement of the dissolving pulps.

Industrial acceptance of this biomass fractionation concept, however, will furtherrequire careful assessments of various options for treating and purifying hemicel-luloses and lignin in their respect streams. Such method may include application ofthe membrane filtration for recovering hemicelluloses and lignin recovery processthat is similar to the LignoBoost process, which was beyond the scope of this study.It is also worthwhile to highlight that application of AQ as pulping aids may berestricted in the near future (Testova et al. 2014). Therefore it would be interestingto explore other sulphur-free reagents such as urea and surfactants as a substitute ofAQ in the proposed biomass fractionation concept.

Acknowledgments The authors wish to acknowledge CSIR for providing financial support whichenabled the accomplishment of this research work (Project EIEB002). Special thanks are due to thetechnical staff at FFP laboratory for their assistance in performing the experiments.

References

Boussarsar H, Rage B, Mathlouthi M (2009) Optimisation of sugarcane bagasse conversion byhydrothermal treatment for recovery of xylose. J Bioresour Technol 100:6537–6542

Cheng H, Zhan H, Fu S, Lucia AL (2010) Alkali extraction of hemi cellulose from depithed corn stoverand effects on soda–AQ pulping. J Bioresour 11(1):196–206

Christensen PK (1998) Wood and pulping chemistry, vol 1. Department of Chemical Engineering, Pulpand Paper Group, The Norwegian University of Science and Technology (NTNU), Trondheim

Corrales RC et al (2012) Structural evaluation of sugar cane bagasse steam pre-treated in the presence ofCO2 and SO2. J Biotechnol Biofuel 5(36):1–8

Garcıa JC, Zamudio MAM, Perez A, Feria MJ, Lıvio Gomide J, Colodette JL, Lopez F (2011) Soda–AQpulping of Paulownia wood after hydrolysis treatment. J Bioresour 6(2):971–986

Girio FM, Carvalheiro F, Duarte LC (2008) Hemicelluloses biorefineries: a review on biomass pre-treatments. J Sci Ind Res 67:849–864

Gullichsen J, Fogelholm C-J (1999) Chemical pulping pulp and paper making technology. Published inco-operation with the Finnish Paper Engineer’s Association and Tappi, McGraw Hill BookCompany, Book 6A, Volume 6A

Hamelinck CN, Van Hooijdonk G, Faaij APC (2005) Ethanol from lignocelluloses biomass: techno-economic performance in short, middle and long term. Biomass Bioenerg 28:384–410

Janson J (1974) Analytics of polysaccharides in wood and pulp (Analytic der Polysaccharide in Holz andZellstoff). Faserforsch Textiltech 25(9):375–382 (in German)

Johakimu JK, Andrew JA (2013) Hemicellulose extraction from South African Eucalyptus grandis usinggreen liquor and its impact on kraft pulping efficiency and paper making properties. J Bioresour8:3490–3504

Mosier NS, Hendrickson R, Brewer M, Ho N, Sedlak M, Dreshel R, Welch G, Dien BS, Aden A, LadischMR (2005) Industrial scale-up of pH-controlled liquid hot water pre-treatment of corn fibre for fuelethanol production. J Appl Biochem Biotechnol 125:77–97

Nabarlatz D, Ebringerova A, Montane D (2007) Autohydrolysis of agricultural by-products for theproduction of xylose-oligosaccharides. J Carbohydr Polym 69:20–28

NNFCC Project no 10/003 (2009) Marketing study for biomass treatment technologies. www.nnfcc.co.uk.Accessed on March 2014

Ohman F (2006) Precipitation and separation of lignin from kraft black liquor. Ph.D. thesis, ChalmersTechnical University, Gothenburg, Sweden

Wood Sci Technol

123

Pandey KK (1999) Study of chemical structure of soft and hardwood and wood polymers by FTIRspectroscopy. J Appl Polym Sci 71:1969–1975

Sixta H, Schild G (2009) A new generation Kraft process. J Lenz Berich 87:26–37Sturgeoff LG, Pitl Y (1993) Low-kappa pulping without capital investment—use of anthraquinone for

low kappa pulping. In: TAPPI proceedings, Atlanta, November 14–18, pp 423–429Sun XF, Xu F, Sun RC, Fowler P, Bairdd MS (2005) Characteristics of degraded cellulose obtained from

steam-exploded wheat straw. J Carbohydr Polym 340:97–106Testova L, Borrega M, Tolonen LK, Penttila PA, Serimaa R, Larsson PT, Sixta H (2014) Dissolving-

grade birch pulps produced under various prehydrolysis intensities: quality, structure andapplications. Cellulose 21:2007–2021

Tunc MS, Van Heiningen A (2008) Hemicelluloses extraction of mixed southern hardwoods with water at150°C: effect of time. J Ind Eng Chem Res 47(18):7031–7037

Tunc MS, Lawoko M, Van Heiningen A (2010) Understanding the limitations of removal ofhemicelluloses during autohydrolysis of a mixture of southern hardwoods. J Bioresour 5(1):356–371

Xiao L-P, Sun Z-J, Shi Z-J, Xu F, Sun R-C (2011) Impact of hot compressed water pre-treatment on thestructural changes of woody biomass for bioethanol production. J Bioresour 6(2):1576–1598

Yang H, Zheng X, Yao L, Xie Y (2014) Structural changes of the lignin in the soda–AQ pulping processstudied using the carbon-13 tracer method. J Bioresour 9(1):176–190

Yoon S-H, Van Heiningen A (2010) Green liquor extraction of hemicelluloses from southern pine in anintegrated forest biorefinery. J Ind Eng Chem Res 16(1):74–80

Zheng Y, Pan Z, Zhang R (2009) Overview of biomass pre-treatment for cellulosic ethanol production.J Agric Biosci Eng Int 2(3):51–68

Wood Sci Technol

123