Bioresource Technology 150 (2013) 129–138

Contents lists available at ScienceDirect

Bioresource Technology

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

Impact of the lignocellulosic material on fast pyrolysis yieldsand product quality

0960-8524/$ - see front matter � 2013 Elsevier Ltd. All rights reserved.http://dx.doi.org/10.1016/j.biortech.2013.09.134

⇑ Corresponding author. Tel.: +27 21 808 9485; fax: +27 21 808 2059.E-mail addresses: carrier@sun.ac.za, marion_carrier@msn.com (M. Carrier).

Marion Carrier ⇑, Jan-Erns Joubert, Stephen Danje, Thomas Hugo, Johann Görgens,Johannes (Hansie) KnoetzeDepartment of Process Engineering, University of Stellenbosch, Private Bag X1, Matieland 7602, South Africa

h i g h l i g h t s

� Fast pyrolysis of South African feedstocks.� Overlap of chemical and transfer phenomena in the bubbling fluidized bed reactor.� Key role of the hemicelluloses content on the quality of fast pyrolysis bio-oil.� Dependence of biochars’ properties with the carbohydrates content.

a r t i c l e i n f o

Article history:Received 13 June 2013Received in revised form 25 September 2013Accepted 28 September 2013Available online 8 October 2013

Keywords:Fast pyrolysisLignocelluloseThermodynamic regimeBio-oilBiochar

a b s t r a c t

The paper describes the fast pyrolysis conversion of lignocellulosic materials inside a bubbling fluidizedbed. The impact of biopolymers distribution in the biomass feed, namely hemicelluloses, cellulose andlignin, on the yields and properties of pyrolytic bio-oils and chars was investigated.

Although it is not possible to deconvoluate chemical phenomena from transfer phenomena using bub-bling fluidized bed reactors, the key role of hemicelluloses in biomass feedstocks was illustrated by: (i) itsinfluence on the production of pyrolytic water, (ii) its impact on the production of organics, apparentlydue to its bonding relationship with the lignin and (iii) its ability to inhibit the development of charsporosity, while the cellulose appeared to be the precursor for the microporous character of the biochars.These results are of interest for the selection of suitable feedstocks aimed at producing bio-oil and char asfuels and soil amendment, respectively.

� 2013 Elsevier Ltd. All rights reserved.

1. Introduction

Plant biomass has potential to be a sustainable energy sourceand could become a viable alternative to the use of fossil resources.It is of particular interest to researchers and policy-makers inSouth Africa where CO2 emissions are a problem (Tomascheket al., 2012). The assessment of existing and available biomassfor bioenergy applications is a critical issue for optimization bio-mass utilization. The Council for Scientific and Industrial Research(CSIR) developed an analytical frame-work and decision-supporttools to assist in assessing, managing and monitoring the sustain-ability of bioenergy (Von Maltitz and Stafford, 2012). This workmapped the woody biomass distribution in South Africa. The out-comes of the work were a national resource map for woody bio-mass projecting the growth and the estimation of potentialquantities. Resources such as the invasive alien plants and bushencroachment, which have the highest potential as biomass

sources, are only available for a defined period of time, estimatedto be 20 years. Substantial amounts of waste biomass from agricul-ture (11.8 Mton per year), wood and forestry residues (9.2 Mtonper year) and sugar cane bagasse (7.5 Mton per year) can also beused for bioenergy purposes (Vosloo, 2013). Conversion of SouthAfrican forest and agricultural residues through thermochemicalprocesses is not only considered as a potential source for the pro-duction of heat, electricity and transport fuels, but can also pro-duce added-value materials and chemicals, e.g. activated carbonand biochar as a soil amendment (Uras et al., 2012; Carrier et al.,2012).

Lignocellulosic plant biomasses such as forest and agriculturalwastes are composed mainly of carbohydrates (i.e. hemicellulosesand cellulose) and lignin. The proportion of these basic polymerblocks varies according to the origin and the type of biomass(Carrier et al., 2011). In addition to the major components, minoramounts of extractives, inorganics and water made up the finalcomposition. The impact of lignocellulosic composition on thequantity and quality of pyrolysis products has often been reportedat the milligram scale using a pyrolyzer oven or a microbalance

130 M. Carrier et al. / Bioresource Technology 150 (2013) 129–138

coupled with gas chromatography or/and mass or infrared spec-trometry. These studies described the chemistry behind fast pyroly-sis degradation of individual and pure compounds, to identify themajor chemical degradation mechanisms (Shen et al., 2010; Hosoyaet al., 2007). Recently, additional milligram scale studies showedthe limitations in using model lignocellulosic compounds to predictyields (Pasangulapati et al., 2012; Couhert et al., 2009). These stud-ies showed the significant role of the chemical composition of thelignocellulosic feedstock on the composition and yields of gasesgenerated by pyrolysis processes. They demonstrated that it is notpossible to predict pyrolysis gas yields from a recombined mixturewith model compounds representing the lignocelluloses feedstock.This failure could be explained by the lack of interactions betweenthe biopolymers and the absence of ash intimately linked to the lig-nocellulosic structure, which influences pyrolysis (Pasangulapatiet al., 2012; Couhert et al., 2009). The use of native lignocellulosicbiopolymers is then critical to fully understand the role of lignocel-lulosic composition on pyrolysis reactions and products.

The influence of the original lignocellulosic composition on fastpyrolysis (FP) products obtained from kilogram scale plants(approx. 1 kg/h) is not well reported. Previous investigations mainlyfocused on the yields and the quality of the products: liquids(moisture, pH, density, viscosity, high heating values, elementalanalysis and organics composition), solid residue (pH, surface area,moisture content) or non-condensables (gas composition) (Oasmaaet al., 2010; Mullen, 2010). Only Oasmaa et al. (2010) conducted afull analysis of the lignocellulosic composition of feedstocks topoint out the impact of this on yields and the quality of productsfrom the 20 kg/h FP conversion. They concluded that the highercontent of hemicelluloses in grass and straw feedstocks, comparedto woody biomass, resulted in higher gas yields, in particularhigher CO2 amounts. The bio-oil analyses suggested that no cleardifference between grassy and woody biomass could be made interms of cellulose degradation into a mixture of sugars and water(levoglucosan and hydroxyacetaldehyde). Mullen (2010) suggestedthat hemicelluloses were degraded mainly into acids and gasesduring fast pyrolysis, while the decomposition of the ligninresulted in the production of phenolic compounds condensed inthe pyrolytic liquid. They also described the properties of biochar,where adsorptive capacity was mainly affected by the presence ofinorganics in the raw lignocellulosic material.

In general, the conclusions around the influence of the lignocel-lulosic biomass composition on the chemical reactions and compo-sition of products in a 1 kg/h fast pyrolysis plant are limited.Indeed, the design of a fast pyrolysis reactor, in particular bubblingfluidized bed reactor type, is not well adapted to control the regimeof the pyrolytic reaction (Mettler et al., 2012) as the transport andchemical-controlled pyrolysis mechanisms are numerous andinterconnected, which does not allow deconvolution of chemicaland transport phenomena (Di Blasi, 2008).

Therefore the present work introduces the chemical and struc-tural influence of lignocellulosic composition of feedstocks on theyield and quality of fast pyrolysis products by describing the basicbio-oil and char characteristics and heat and mass transfer limita-tions within the raw biomass and a 1 kg/h fast pyrolysis unit.

2. Methods

2.1. Materials and preparation

Sugar cane bagasse (BG) from the Kwazulu-Natal area of SouthAfrica was provided by the Sugar Milling Research Institute(Durban, South Africa). Corn cobs (CC) and corn stover (CS) werecollected in the Lichtenburg area in the Northwest province ofSouth Africa. Eucalyptus grandis (EC) woodchips were provided by

Mondi from the Kwazulu-Natal coastline. A corn residues mixturecombining 70 wt.% of corn stover with 30 wt.% of corn cob was alsoprepared. All feedstocks (40 kg) were air-dried and stored in anoutdoor shed and milled using a SM 100 Retsch mill, first with a4 mm sieve and then with a 2 mm sieve. An AS 200 Retsch shakerwas used to remove the fines (<250 lm) and oversize (>2000 lm)fractions. The mass median particle diameter (dp50) was calculatedfor the log-normal distribution dataset, and was found to be 408,1129, 868 and 573 lm for CS, CC, BG and EC, respectively (resultsnot shown). The shape of CS, BG and EC particles was considered ascylindrical, while CC particles were considered as spherical.

2.2. Analytical methods

Sub-samples were taken from the individual batches of ligno-cellulosic biomass according to the cone and quartering technique(DD CEN/TS 14780:2005) and subjected to a series of physico-chemical characterizations. Water content (ASTM E871/MetrohmKF Titrino), higher heating value (HHV) (IKA C200 Bomb calorime-ter) and pH (509900 Utech Instruments) were determined for theliquid phase. The analytical strategy to characterize the initialbiomass and solid residue included water content (TechnicalAssociation of the Pulp and Paper Industry (TAPPI) T264 and ASTMD-871–82), ash content (TAPPI T211), proximate analysis (ASTME11311/Mettler Toledo TGA/DSC 1 – LF 1100), HHV (IKA C5000bomb calorimeter) and elemental analysis (TruSpec� Micro SeriesAnalyser) (Table 1). The characterization of CC, CS and EC biocharsencompassed a series of analyses such as the determination ofpHKCl and pHH2O (Titrator IKA�KS 260), cation exchange capacity(CEC) (Ion chromatography-Dionex DX-120 Ion Chromatograph),citric acid digestible nutrients (Macro-element/ICP-MS Agilent7700-Micro-element/Varian Liberty II Radical ICP instrument)and inorganics content (XRF/PANalytical HighScore Plus Software,Axios 2.4 kW instrument with Rh Tube) (Table 2).

The BET surface area of chars was measured by using a Micro-metrics ASAP 2010. The porous structure was described using amethod previously developed and presented in Carrier et al.(2012), which has been found to be adequate for the microporosityand mesoporosity evaluation of pyrolysis chars containing littleorganic material. However, the large standard deviation obtained(Table 3) indicated that a more rigorous sampling procedureshould be established in order to take into account the heterogene-ity of solid residues and then identify the inconsistencies betweensub-samples.

2.3. Determination of the lignocellulosic composition

Sub-samples of lignocellulosic feedstocks were milled andsieved as mentioned in Section 2.1, isolating the biomass particlesize range +250 lm and �425 lm for the lignocellulosic character-isation. Analyses were done in accordance with the Technical Asso-ciation of the Pulp and Paper Industry (TAPPI) standard methodsfor characterizing woody biomass (T264 om-88, T211 om-85) usedin the pulp and paper industry (Sluiter et al., 2010). As the optimi-zation of the extraction methods was not within the scope of thisstudy, the same extraction methods have been applied to the dif-ferent feedstocks to compare holocellulose, a-cellulose and lignincontents between feedstocks. Experimental errors of 2 wt.%,5.7 wt.%, 5 wt.% and 2 wt.%, respectively, for the determination oflignin, hemicelluloses, cellulose and extractives contents havebeen found.

2.3.1. ExtractivesThe biomass extractives content was determined in accordance

to T264 om-88 of the TAPPI standard methods (Preparation ofwood for chemical analysis). Approximately 5 g of biomass was

Table 1Initial characterization of feedstocks: Sugar cane bagasse (BG); Corn cob (CC), Corn Stover (CS), Eucalyptus grandis (EC) and mixture of 70 wt.% of corn stover and 30 wt.% of corncob (CR).

BG CC CS EC CR

Raw Char Raw Char Raw Char Raw Char Char

Lignin (wt.%, daf)a 22.4 16.7 13.0 15.2Hemicellulosesb 23.8 38.1 42.3 24.7Cellulose 44.2 35.9 37.0 57.5Extractives 9.7 9.5 7.7 2.6

C (wt.%) 45.0 62.35 43.8 74.63 41.9 56.73 44.4 79.49 59.91N 0.5 0.60 0.3 0.47 0.6 0.72 0.4 0.17 0.65H 6.0 2.85 6.5 2.24 6.4 1.89 6.3 2.77 1.94S 0.1 0.01 0.0 0.27 0.2 0.22 0.1 0.27 0.33Ob 38.8 18.09 47.8 3.59 44.8 10.04 48.2 12.30 5.87O/C molar 0.65 0.22 0.82 0.04 0.80 0.13 0.81 0.12 0.07H/C molar 1.60 0.55 1.78 0.36 1.83 0.40 1.70 0.42 0.39

Volatiles (wt.%) 76.1 19.1 79.9 15.9 76.7 17.7 80.9 15.7 24FCb 6.9 62 13.7 61.3 8.2 47.9 19.1 75.3 43.1AC 9.6 16.1 1.6 18.8 6.1 30.4 0.5 5 31.3WC 6.7 2.8 4.6 4.0 8.5 4.0 6.2 4.0 2.1HHV (MJ/kg) 17.6 24 17.3 27.4 15.9 19.8 14.3 30.75 21.6

a daf: dry and ash free weight basis.b By difference.

Table 2Micro and macro elements composition of pyrolysis feedstocks and chars.

Sample (g kg�1) BG CC CS EC

Raw Char Raw Char Raw Char Raw Char

Al 1.1109 72.93 bd 3.3 1.6399 17.51 bd 0Ca 0.4998 10.78 0.1428 2.9 4.3554 26.06 1.428 4.7124Cr bd 0 0.0684 0.1 bd 0.14 bd 0Fe 1.7475 26.38 0.2097 3.7 2.6562 13.56 0.2097 0.7689K 0.913 38.58 3.569 24.9 6.64 30.38 0.083 6.557Mg 0.603 6.35 0.7236 2.2 3.015 14.53 0.7236 1.2663Mn 0.0775 0.51 0.0775 0.1 0.155 0.47 0.155 0.5425Na bd 19.03 bd 0 bd 2.52 bd 1.3356P 0.1748 0.93 0.2185 0.6 0.437 2.27 0.0874 0.2185Si 8.424 326.6 bd 69.9 30.2328 145.6 bd 0.468Ti 0.18 3.24 0.06 0.4 0.24 1.32 0.06 0.12

bd, below detection.

Table 3Description of the porous structure of biochars obtained from the fast pyrolysis of BG, CC, CS, EC and mixture of CC and CS.

BG CC CS EC CR

Average pore diameter (Å) 21 ± 2 21 ± 3 25 ± 1 20 ± 1 23 ± 4BET (m2 g�1) 309 ± 45 159 ± 43 109 ± 31 372 ± 15 97 ± 40Micropore area (m2 g�1) 245 ± 49 83 ± 8 87 ± 24 314 ± 20 83 ± 20External surface area (m2 g�1) 64 ± 4 76 ± 50 22 ± 7 58 ± 8 14 ± 7Micropore (%) 68 ± 6 44 ± 14 57 ± 2 77 ± 9 69 ± 17Mesopore (%) 8 ± 2 13 ± 3 4 ± 1 6 ± 1 2 ± 3Macropore (%) 22 ± 5 41 ± 10 37 ± 2 15 ± 5 28 ± 14

M. Carrier et al. / Bioresource Technology 150 (2013) 129–138 131

extracted first, with a 1:2 volume ratio of 95% ethanol and cyclo-hexane (selected for health and safety reasons instead of benzeneprescribed in the standard method), followed by distilled waterextraction. Each extraction was allowed to run for 8 h under fullreflux with an extractor siphoning rate of at least four times perhour. The extracted biomass was then allowed to air dry beforecontinuing with further analyses. The extractives proportion wasdetermined gravimetrically.

2.3.2. Klason ligninAir-dried and extractives free biomass was used for Klason lig-

nin determination. After primary hydrolysis of the biomass with72% H2SO4, the T 222 om-88 TAPPI standard method specifies sec-ondary hydrolysis (3% H2SO4) to take place for 4 h under full refluxusing a reflux condenser. This secondary hydrolysis (3% H2SO4) was

replaced by 1 h autoclave at 1.25 bar and 122 �C (Sluiter et al.,2010). After secondary hydrolysis, the sample was vacuum filteredand dried to yield Klason lignin as residue and a filtrate rich inmonomeric and dimeric sugars. The Klason lignin proportion wasdetermined gravimetrically.

2.3.3. CarbohydratesHolocellulose was extracted using 5 g of extractives free bio-

mass in a beaker with 160 mL of demineralised water, glacial aceticacid and reagent grade NaClO2, set in a water bath at 70–80 �C. Avolume of 0.5 mL of glacial acetic acid and 1.5 ± 0.1 g of reagentgrade NaClO2 were added to the heated beaker under a fume hoodand stirred occasionally. A second and third dose of the sameamounts of these chemicals were added at 1 h intervals with occa-sional stirring. After 3 h of NaClO2 treatment, the flask was placed

132 M. Carrier et al. / Bioresource Technology 150 (2013) 129–138

in an ice bath to cool the contents to below 10 �C, before filtrationwith a coarse filter crucible. Minimal ice water was used to transferthe contents from the flask to the filtration crucible. The cruciblecontent was washed with acetone (P99.9% Chromasolv�, Sigma–Aldrich), followed by 100 mL of 95% ethanol (95% Chromasolv�,Sigma–Aldrich) and finally by ice water before drying in an ovenat 105 �C.

The holocellulose extract from the above mentioned procedurewas used as feedstock for a-cellulose extraction. A mass of 3 g ofthe holocellulose and 35 mL of a 17.5% NaOH solution were addedto the beaker and allowed to soak for 5 min before stirring andsoaking for a further 10 min. The sample was placed in a waterbath set at 20 �C while an additional 40 mL of the caustic soda solu-tion was added in equal portions (10 mL every 10 min) with occa-sional stirring. The sample was diluted with 75 mL of distilledwater under continuous stirring before being vacuum filteredusing a filtration crucible. With the suction removed, 40 mL of10% acetic acid was added to the crucible and allowed to soak for5 min before filtering the liquid off. The sample was finally rinsedwith distilled water (under suction) before it was dried overnightin an oven set to 102 �C.

2.4. Thermal properties determination

2.4.1. Heat of reaction for pyrolysisA TGA/DSC 1 – LF 1100 unit (Mettler Toledo) was used to mea-

sure the heat of reaction of the pyrolysis reaction. DifferentialScanning Calorimetry (DSC) curves were calculated from simulta-neous differential thermogravimetric analysis (SDTA) curves. TheSDTA measures the difference between the sample temperatureand the reference temperature. The SDTA signal was calibratedusing three different substances (indium, aluminum and gold).The experiments were performed from 313 K to 823 K, underatmospheric pressure and with nitrogen as reactive gas suppliedat 50 mL min�1 and as protective gas at 20 mL min�1. A 900 lL alu-mina crucible was submitted to a heating rate of 10 K min�1. Theobtained SDTA curve is displayed in J g�1 (Table 4).

2.4.2. Kinetic parameters of thermal degradationThe Friedman’s isoconversional method was applied to obtain

the variation of kinetic parameters, activation energy (Ea) andpre-exponential factor (A), over the entire range of conversion forthe samples (Table 4). This model-free method, which does notrequire assumption of reaction models, has been described indetail by Aboyade et al. (2011).

Thermogravimetric analyses (TGA) were conducted using aTGA/DSC 1 – LF 1100 unit (Mettler Toledo) thermogravimetric

Table 4Thermo-physical properties of agricultural and forest residues.

BG

Feedstocks type StrawParticle shape CylinderBulk density (kg m�3) 170k (W m�1 K�1) 0.0579Heat requirement for pyrolysis (J g�1) 696Biot number (Bi) 1.313Kinetic parametersEa (kJ mol�1)-slow 174.4ln A (s�1) 26.792kr (s�1) at 773 K 0.707Ea (kJ mol�1)-fast 263.04ln A (s�1) 42.786kr (s�1) at 773 K 6.4Cp (J kg�1 K�1) 0.9Py 1.254M 0.798Py0 1.646

analyser. For all TGA experiments a nitrogen flow rate of20 mL min�1 was used for the protective gas, while 50 mL min�1

was used in the oven to provide an inert environment for the sam-ple. Constant sample size of 15, 10, 15 and 10 mg for CC, CS, EC andBG, respectively, were placed in a 70 lL alumina crucible. Thesesamples were heated from 313 K to 873 K using slow (10, 20 and50 K min�1) and fast (100, 120, 150 K min�1) heating conditions.

2.5. Comparison of kinetics of heat transfer, mass transfer andchemical reactions

Different operating regimes can occur within fluidized bed fastpyrolysis. Considering thermal conditions and properties of bio-mass, heat transfer and reaction kinetics were determined. Sincenot only the external convection but also the internal conductionis important, the overall heat transfer regime was characterizedby the thermal Biot number, which represents the internal resis-tance to heat penetration divided by the external resistance to heattransfer introduced by J.B. Biot in 1816 (Eq. (1)). On the other hand,the pyrolysis numbers (Py, Py0) established by Pyle and Zaror in1984 are used to define the ratios of the different time scales (rep-resentative of kinetic rates) involved in the internal conduction,external heat transfer and chemical conversion processes. Theinternal pyrolysis number, Py, gives a measure of the relativeimportance of the internal conduction and the reaction timeconstant (Eq. (2)). The thermal Thiele Modulus (M) measures therelative importance of the mass diffusion and reaction phenomena(Lédé, 1994). The external pyrolysis number, Py0, gives the ratiobetween heat convection rate and the reaction rate (Eq. (3)):

Biot number ¼ h � Lk

ð1Þ

Py ¼ 1M¼ k

krqcpL2 ð2Þ

Py0 ¼ hkrqcpL

ð3Þ

where h is the effective heat transfer coefficient (W m�2 K�1), L isthe particle characteristic length (m), k is the thermal conductivityof the biomass (W m�1 K�1), kr is the reaction rate (s�1), q is thebulk density (kg m�3), cp is the specific heat capacity (J kg�1 K�1)

To determine these numbers, a series of thermo-physical prop-erties pertaining to transport or thermodynamic characteristicswere determined as follow:

CC CS EC

Starchy Straw WoodySphere Cylinder Sphere290 170 1600.0819 0.0579 0.0559540 683 5090.804 0.616 0.898

289.97 223.32 186.9050.752 36.757 29.147279 7.45 1.067532.81 328.66 174.4996.547 56.14 26.898,34,504 148.69 0.7750.6 0.9 1.20.002 0.245 18.268551.733 4.084 0.0551.28 ⁄ 10�5 0.151 16.410

M. Carrier et al. / Bioresource Technology 150 (2013) 129–138 133

– The characteristic length (L) is defined as the ratio of the volume

to surface area of the solid particle L ¼ VAs

� �, which differs

between a cylinder, particle diameter, dp50/4, and a sphere,dp50/6.

– The heat transfer coefficient (h) varies within the particle sizeand ranges from 190 to 474 W m�2 K�1, which is a typical rangefor fluidized bed reactors using particle size of sand between 40and 500 lm and a particle density between 1.4 and 4 g/cm3

(Geldart B group particles) (Papadikis et al., 2009). An averagevalue for h of 350 W m�2 K�1 was used for this study.

– The thermal conductivity (k) was determined using the correla-tion (4) developed by MacLean (1941) for a moisture percentageof the biomass on a dry basis lower than 40 wt.%.

k ¼ qbulk

qwater� 0:2þ 0:0055 �WCð Þ þ 0:0238 ð4Þ

where k is the thermal conductivity (W m�1 K�1), qbulk is the bulkdensity (kg m�3), qwater is the water density (kg m�3) and WC isthe water content (wt.%).

– The specific heat capacity, cp,– The apparent and measured rate constant, kr, is a strong func-

tion of the temperature and follows the Arrhenius equation(Eq. (5)).

kr ¼ Aexp �Ea=RTð Þ ð5Þ

where kr is the overall reaction rate (s�1), A is the pre-exponentialfactor (s�1) and Ea is the activation energy (kJ mol�1) obtained infast heating conditions using thermogravimetric analysis, R is thegas constant in J mol�1 K�1 and T is the temperature in K.

Although the reaction rate depends strongly on the heating rateapplied (Van de Velden et al., 2010), these reaction rates have beendetermined through the use of the thermogravimetric instrument(TGA/DSC 1 – LF 1100, Mettler Toledo), which is limited to0.002–2.5 K s�1 (0.1–150 K min�1) as external heating rates againstheating rates in excess of 100 K s�1 in a fluidized bed (Papadikiset al., 2009). Kinetic parameters, Ea and A, have been determinedin slow (10, 20 and 50 K min�1) and fast (100, 120, 150 K min�1)heating conditions in a TGA to illustrate the influence of heatingconditions and feedstock nature on the rate of reaction (Table 4).The use of small amounts of biomass, reduced particle size andlow heat-transfer coefficient (10 W m�2 K�1) in the thermogravi-metric device ensure the dominance of a kinetically-controlledregime (Bi < 0.1).

Fig. 1. The schematic diagram of the fast pyrolysis system, consisting of a bubbling flprecipitator.

2.6. Fast pyrolysis conversion

The fast pyrolysis (FP) plant (Fig. 1) can be divided into threesections: the biomass feeding unit, the furnace housing the bub-bling fluidised bed reactor (BFBR) with two cyclone separators,and the condensation chain. The biomass feeding unit consistedof a hopper with a screw feeder, which introduced biomass (parti-cle size 250–2000 lm) at a feed rate of ±0.85 kg h�1 into the reac-tor. The hopper was maintained at a slight N2 overpressure(±0.02 bar gauge) to prevent hot gas and product vapour pushingback from the reactor into the feeding system. A 6.6 kW cylindricalfurnace housed the BFBR with an internal diameter of 75 mm and aheight of 680 mm in stainless steel (SS 316), followed by two cy-clones in series for solid and vapour separation. Sand (AFS 45 fusedsilica, from Consol Minerals) was used as the heat carrier inside thereactor. The cylindrical furnace was divided into two sections, topand bottom, which met at the feeder inlet that was pre-heated to773 K. The pre-heated N2 carrier gas (Technical nitrogen, Afrox)was fed at a flow rate of 2.4 m3 h�1. Pyrolysis product vapours leftthe furnace via a heated pipe (maintained at 673 K to preventintermediate condensation) before entering the cooling tower.The condensation chain consisted of a spray cooling tower fol-lowed by two electrostatic precipitators (ESPs) in series, set to 14and 12 kV, respectively. An iso-paraffinic hydrocarbon (Isopar fromEngen Petroleum Limited) was used as direct contact quenchingmedium in the spray cooling tower. A 13 kW Daikin water chillerwas used to control the temperature of the water bath which, inturn, maintained the Isopar circulation loop at the desired set pointtemperature. All incondensable gasses were purged to atmosphere.

Char yield (Ychar), bio-oil yield (Ybio-oil), water yield (Ywater),pyrolytic water (Ypyrolytic water), organics yield (Yorganics) and gasyield (Ygas) were calculated on a weight basis (wt.%) for thebiomass used for each experimental run.

Ychar wt:%ð Þ ¼ Mchar pots þ Mreactor content �Msandð ÞMbiomass

� 100 ð6Þ

Ybio-oil wt:%ð Þ ¼ Mbulk liquid þMtarry phase

Mbiomass� 100 ð7Þ

Ywater wt:%ð Þ ¼ MKFwater

Mbiomass� 100 ð8Þ

Ypyrolyticwater wt:%ð Þ ¼ MKFwater �Minitialwater

Mbiomass� 100 ð9Þ

uidized bed reactor (BFBR), cyclones and a condensation train. ESP, Electrostatic

134 M. Carrier et al. / Bioresource Technology 150 (2013) 129–138

Yorganics wt:%ð Þ ¼ Mbulk liquid þMtarry phase �MKFwater

Mbiomass� 100 ð10Þ

Ygas wt:%ð Þ ¼ 100%� YLiquid � YSolid ð11Þ

where M is the mass of product in g and KFwater (Eq. (10)) corre-sponds to water content determined by Karl–Fisher method.

The bio-oil yield (Ybio-oil) was calculated by adding the mass ofliquid products recovered from the cooling tower reservoir andfrom the acetone wash. The viscous tar resulting from the acetonewash was not mixed back into the bulk liquid and not subjected tofurther analysis. The determination of product yields was repeatedthree times at 500 �C; uncertainties correspond to the standarddeviations of average yields.

The linear regression and the coefficient correlation have beenstatistically verified using the parametric data analysis function‘regression’ in Excel. Uncertainty from the linear regression witha sample size of 4–5 points has been evaluated by calculating a95% t-confidence interval (DCI).

3. Results and discussion

3.1. Characterization of feedstocks

The proportions of the main lignocellulosic components (lignin,holocellulose and cellulose) of feedstocks (Table 1) were determinedusing a two-step acid hydrolysis and chlorination method or a basichydrolysis. The use of different fractionation methods can lead todifferences in the lignocellulosic composition of feedstocks, whichlimits the comparison with literature (Carrier et al., 2011). Corn res-idues contained more hemicelluloses than bagasse and wood, whilea surprisingly low content of lignin was recorded for Eucalyptusgrandis in this study and previous work (Vena et al., 2013). Such vari-ations within a specific species and genus can be attributed to,amongst others, genetic, climatic and geographical parameters thatdirectly influence the growth conditions of the biomass (e.g. rainfall,altitude above sea level, soil quality and exposure to sunlight)(Fengel and Wegener, 2003). Moreover, the introduction of reducedlignin trees to improve the yield of kraft pulp production (MacLeod,2007) could explain the low lignin content quantified for theEucalyptus grandis (Table 1). The lignocellulosic composition of BGwith the highest lignin content, 22.4 wt.%, resulted in the lowestO/C and H/C molar ratios, 0.65 and 1.60 respectively.

The XRF analysis (Table 2) showed that the inorganic content inlignocelluloses was mainly composed of SiO2: 1.8 wt.% for the su-gar cane bagasse (BG) and 6.5 wt.% for corn stover (CS) feedstocks.The presence of particularly high levels of inorganics in the raw

Fig. 2. Char, liquid (organics + water), gas yields of forest (EC: Eucalyptus grandis)and agricultural (BG, Sugar cane bagasse; CC, Corn cob; CS, Corn stover) residues in1 kg/h fast pyrolysis plant.

materials such as K (6.64 g kg�1), Fe (2.66 g kg�1), Ca (4.36 g kg�1)and Al (1.64 g kg�1) for the corn stover feedstock was confirmed bythe ash content (AC) of 30.4 wt.% obtained from the proximateanalysis (Table 1). Processing and harvesting practices or anatom-ical composition could explain the large amounts of inorganics.

3.2. Pyrolytic yields – mass balance

As per design specifications for liquids production (Bridgewater,2012), the fast pyrolysis of the five feedstocks (agricultural and for-est residues) in the present study achieved high liquid yields, rang-ing between 51.1 and 66.9 wt.% for corn stover and corn cob,respectively (Fig. 2). However, char yields were high for both feed-stocks containing large ash content, in particular for corn residues,and varied from 9.3 wt.% to 22.3 wt.% (Table 1). This last observa-tion combined with the result showing a significant linear trendbetween char yields and original hemicelluloses contents(Ychar (wt.%) = 0.7046⁄HEMI �6.9949, R2 = 0.9200, p-value = 0.05,DCI = 3.1 wt.%) could indicate that the presence of hemicelluloseswould limit the devolatilization of inorganics. According to theirnature, these inorganics could affect the char formation mecha-nisms, which include gas-phase repolymerization of volatile prod-ucts, polymerization of condensed phase species, and dehydrationof biopolymers (Mettler et al., 2012). In particular, the highest levelof silicon, 30.2 g kg�1, was detected for the raw material CS(Table 2). Silicon, a non-metal compound in the oxide form ofSiO2, is used as a secondary catalyst for the cracking of large poly-meric molecules such as oxygenates (Bulushev and Ross, 2011).This oxide could have catalyzed the repolymerization reactionsfor the benefit of the char formation and gas production. In the caseof CC, a high liquid yield (66.9 wt.%) was also recorded, indicatingthat the dehydration of biopolymers could have been promoted.

The range of fast pyrolysis product yields observed for thevarious feedstocks (Fig. 2) is the result of a series of rate-limitingphenomena that occur simultaneously. The highest liquid and charyields were recorded with CC, for which a heat transport (conduc-tion/convection)-limited pyrolysis regime (Py and Py0 < 10�1)coupled to a mass transfer (diffusion)-limited regime (M > 1) wasestablished (Table 4). In general, a longer volatiles residence timemaximizes secondary reactions such as cracking and re-polymeri-zation reactions, which increase the production of gases and char,respectively (Mettler et al., 2012). However the lowest gas yield of10.8 wt.% was recorded for CC (Fig. 2) despite the long vapor resi-dence time. This result indicated that the mass and heat transferlimitations that impact the volatiles residence time promoted theproduction of condensable compounds in this case.

3.3. Characterization of bio-oils

The liquid phase from fast pyrolysis is composed of organiccompounds and water. The water content affects the phase stabil-ity, energy density, viscosity and high heating value of pyrolysisbio-oils and it is of interest to understand its origin (Westerhofet al., 2010). The water content of bio-oil is a mixture of pyrolyticwater, produced during the pyrolysis process, and water content ofthe initial biomass. The latter water can remain in two forms, withfree moisture generally residing outside the cell walls, while inher-ent moisture is contained within the cell walls (Basu, 2010). Thefree moisture determined after biomass preparation was in therange of 4.6–8.5 wt.% (Table 1) and was considered to be con-densed with organics into the bio-oil. The pyrolytic water alsopresent in the bio-oil originates primarily from the dehydrationreactions of carbohydrates (Garcia-Perez et al., 2008).

Water content and pyrolytic water of bio-oils obtained from fastpyrolysis in the present study were respectively found linearly andquadratically correlated with the hemicelluloses (HEMI) content

M. Carrier et al. / Bioresource Technology 150 (2013) 129–138 135

and the mass ratio between the hemicelluloses (HEMI) and lignin(LIG) of feedstocks (Water content (wt.%) = 0.7313⁄HEMI(wt.%) + 4.2678, R2 = 0.8169, p-value = 0.10, DCI = 5.2 wt.%;Ypyrolytic water (wt.%) = �2.1938⁄(HEMI/LIG)2 + 9.5696⁄(HEMI/LIG) �0.9076, R2 = 0.78, DCI = 2.2 wt.%). This result suggests thatthe presence of hemicelluloses (HEMI) could affect the productionof water through dehydration reactions. Indeed, Patwardhan(2010) showed that the main fast pyrolysis products of switchgrasshemicelluloses were compounds originating from the doublehydration products of xylose.

Organics products present in bio-oils are obtained from thepyrolytic conversion of hemicelluloses, cellulose and lignin as or-ganic components of lignocelluloses. For the bubbling fluidizedbed fast pyrolysis system the organics liquid yield was significantlycorrelated to the mass ratio between the hemicelluloses (HEMI)and lignin (LIG) contents (Yorganics (wt.%) = �6.2603⁄HEMI/LIG + 59.670, R2 = 0.8501, p-value = 0.05, DCI = 4.0 wt.%). This evi-dence indicates the existence of interactions between lignin andhemicelluloses, whose weight ratio (HEMI/LIG) equal to one couldfavor the production of organics. This result showed that a chemi-cal interaction exists between the biopolymers, hemicelluloses andlignin, which affected the formation of organic compounds in thebio-oil from fast pyrolysis. Indeed the lignin is covalently boundto the hemicelluloses which in turn, are connected to cellulosethrough extensive hydrogen bonding. Therefore, both covalentand hydrogen bonding between organic components of lignocellu-loses may affect pyrolysis reactions. Covalent bonding between lig-nin and hemicelluloses may either be an aliphatic–aromatic ether

Table 5Citric acid extractable nutrients (g kg�1) present in agricultural and forest-basedbiochars.

CC CS EC

Ca 0.755 1.315 0.675Mg 0.1625 0.3080 0.0685Na 0.3075 0.4350 0.8625K 33.23 21.00 1.37Cl 6.875 6.800 0.288NO�3 0 0 0NO�2 0 0 0

PO2�4

0 0 0

Table 6Physico-chemical characteristics of chars as soil amendment agent.

Biochar CC CS EC Linearly co

pHKCl = 1.0pHKCl 9.2 10.2 8.57 pHKCl = 0.1

pHH2O = 1.pHH2O = 0.

pHH2O 9.9 10.9 8.77 pHH2O = 0.EcC = �78EcC = �43EcC = �31

EcC (dS m�1) 2.1 1.6 0.20 EcC = �54CEC = 4.01CEC = �4.1CEC = 2.57

CEC (cmol kg�1) 89.0 83.6 140.8 CEC = �1.1Aciditysurf

Aciditysurf

Aciditysurf. (mmol g�1) 1.05 1.5 0.55 Aciditysurf

Basicitysur

Basicitysurf. (mmol g�1) 0.4 0 0.95 Basicitysur

CAEt nutrients CAEt nutriBET (m2 g�1) BET = �13Macropores (%) MACRO =Micropores area (m2 g�1) MICROare

(phenyl glycoside) bond or an aliphatic–aromatic ester bond(Iversen, 1985).

3.4. Characterization of char

The solid residue obtained from pyrolysis process can be usedfor many purposes, e.g., as a solid fuel (Bridgewater, 2012), in acti-vated form as a sorbent material (Carrier et al., 2012), as a methodof carbon sequestration in soil, and as soil amendment (Laird et al.,2010). Although the fuel characteristics of fast pyrolysis charproduced in this study such as HHV (19.8–30.8 MJ kg�1), S (lowcontent of sulfur, 0.66–1.3 wt.%), N (low levels of nitrogen,0.53–1.8 wt.%) and ash contents (Table 1) confirmed its potentialapplication as replacement for combustible solid fuels. Futher-more, the influence of lignocellulosic biomass composition on thechar properties, and its impact on application as soil amendmentagent, known as biochars, could also be demonstrated. This alter-native use for the fast pyrolysis solid may enhance soil qualityand then crop production (Joseph et al., 2010). If the biocharsremain stable it could also be an effective means of sequesteringlarge amounts of carbon in soils (Atkinson et al., 2010). The ele-mental (C, H, N, S–O) composition of biochars (Table 1) influencestheir stability in the soil (Preston and Schmidt, 2006). CC-basedbiochar with the lowest O/C and H/C molar ratios (0.04–0.13 and0.36–0.40, respectively) should persist in soils for the longest time,while other biochars, such as the BG-based biochars, with the high-est O/C and H/C molar ratios (0.22 and 0.55, respectively), is likelyto degrade more quickly (Preston and Schmidt, 2006). The citricacid extractable nutrients analyses (Table 5) indicated the presenceof the major plant nutrient contents such as Mg2+ and K+ in bioch-ars (especially high for CC and CS biochars), and the absence ofNO�3 , an essential nutrient for plant growth (Atkinson et al.,2010). Finally, the high levels of Al in BG and CS based biocharscould be toxic for soil biota in very acidic soils by releasing the sol-uble form of Al3+ (Gercel et al., 2007).

Biochar characteristics such as the pH, cation exchange capacity(CEC) and electrical conductivity (EcC) have been reported todepend mainly on both the content and composition of the mineralfraction (Uras et al., 2012). In this study, the biochars from fastpyrolysis were alkaline (Table 6), which could be attributed tothe presence of ash composed of salt base metals (Enders et al.,2012). However pHH2O and pHKCl were linearly correlated to the

rrelated with R2 DCI p-Value

145 ⁄ HEMI/LIG + 6.8987 0.998958 ⁄MESO + 7.6635 0.9812562⁄HEMI/LIG + 6.8435 0.938036⁄ESA + 7.902 0.9402475⁄MESO + 7.7483 0.995.827 ⁄ CELL + 4722.4 0.9440.31 ⁄ AveragePoreDia + 11043 0.974.967 ⁄ BJHAdsorPoreDia + 4153 0.976.278 ⁄ BJHDesorPoreDia + 4483 0.99759 ⁄ AC + 17.06 0.93274 ⁄ HEMI + 254.32 0.96856 ⁄ CELL � 7.4791 0.983133 ⁄ ESA + 164.46 0.981

. = 0.0207⁄Oxidest + 0.5532 0.978

. = 0.0163⁄ESA + 0.1567 0.927

. = 0.1128⁄MESO + 0.0792 0.998f.= �0.0508⁄HEMI + 2.2308 0.960f. = 0.0166 ⁄ ESA + 1.3432 0.955ents = 39.362⁄EC – 2477 0.999.37 ⁄ HEMI + 663.5 0.946 47.4 m2 g�1 0.019�1.008 ⁄ CELL + 71.192 0.764 8.4% 0.052a = 11.37 ⁄ CELL � 318.205 0.902 56.3 m2 g�1 0.013

136 M. Carrier et al. / Bioresource Technology 150 (2013) 129–138

mass ratio between hemicelluloses and lignin in the raw materials(Table 6). This observation suggested that an equivalent proportionin HEMI and LIG will decrease the pH. The measured pH values canbe considered together with the acidity and basicity surface resultsto better characterize the chemistry of biochars surface. Similar tothe measured pH values, the surface basicity values of biocharswere correlated to the lignocellulosic feedstock composition, inparticular with the presence of hemicelluloses (Table 6). This cor-relation indicated that high levels of hemicelluloses in the feed-stock will decrease the amount of basic functionalities on thechar surface, which may include ketones, carbonates and otheralkalinity causing species (Demeyer et al., 2001). On the otherhand, the acidic surface functionalities of biochars, typically com-prising carboxyl, phenolic and lactonic groups (Demeyer et al.,2001) appeared to be correlated to the total amount of oxides(Oxidest) in the biochars (Table 6), including acid (SiO2), ampho-teric (Al2O3) and basic (CaO, Na2O, MgO, K2O, Fe2O3) oxides(Table 2). The high electrical conductivity (EcC) values of fast pyro-lysis biochars, 0.20 and 2.1 dS m�1 (Table 6), which may be relatedto salination problems and nutrient imbalances (Joseph et al.,2010) when biochars are applied to soils, were mostly correlatedto the cellulose content of the lignocellulosic feedstocks (Table 1).The cation exchange capacity (CEC), which is an important indica-tor of the ability of biochars to enhance soil fertility, is also affectedby the presence of ash. Furthermore, the presence of negativelycharged functional groups on the biochar surface gives the abilityto attract, retain and exchange basic cations, which can be usedto enhance nutrient holding capacity of the soil, thus minimizingnutrient losses by leaching and enhancing soil fertility (Josephet al., 2010). The high CEC values observed for EC, CC and CS basedbiochars were strongly correlated to the carbohydrates contents ofthe lignocellulosic feedstocks (Table 6). Increasing hemicellulosescontents appeared to decrease the CEC, while increasing cellulosecontents will increase it.

In addition to the ions exchange ability of biochars, the porousstructure of thereof may trigger the soil aeration and the interac-tions between micro-organisms, dissolved organic matter and

Fig. 3. Temperature lag between the oven temperature (Tref) and the sample temperatureand Eucalyptus grandis (EC, ) feedstocks under a heating rate of 150 K min�1 in th

minerals, thus enhancing fertility (Joseph et al., 2010). The porousstructure of biochars were determined by surface areas measure-ments, which was found to be relatively high, between 100 and400 m2 g�1 (Table 3), which will contribute to the enhancementof soil quality (Laird et al., 2010). These large surface areas for fastpyrolysis biochars in the present study differ from a recent reportthat found negligible surface area for CS and CC biochars (Mullen,2010). The increase in surface area for biochars from the fluidizedbed pyrolysis reactor in the present study can be attributed to thelonger char residence times in this study, caused by the position ofcyclones separators and char pots mounted in series inside the fastpyrolysis oven. The longer the char residence times of pyrolysis,the greater would be the BET surface areas (Dutta et al., 2012).

The porous structure of biochars (Table 2) was correlated to thecarbohydrates (i.e. hemicelluloses and cellulose) content in thelignocelluloses feedstocks, as illustrated by the existence of signif-icant linear trends between hemicelluloses and BET surface areavalues, or between cellulose and micropores area and/or the distri-bution of macropores (Table 6). In particular, the micropores areafor both biochars, 83–314 m2 g�1, appeared to be promoted bythe cellulose content, which were predominant in the woody feed-stock (Table 1). The EC based biochars displayed the highest BETsurface area, micropore volume, micropore area and micropore dis-tribution (Table 3). On the other hand, the CC based biocharsshowed the largest external surface areas (ESA) identified as thenon-microporous area, which is in accordance with its large distri-bution of macropores (Table 3).

In summary, corn cob (CC) based biochar presented the mostappropriate characteristics for application as fuel product, suchas low sulfur and silicon contents and a high calorific value,whereas woody based biochar was most appropriated for applica-tion as soil amendment, based on aeration and hydrology benefitsto soils. However, a lower plant-available nutrient (Ca2+, Mg2+ andK+) content makes the biochar from woody biomass less attractivefor the growing stage of plants.

The physico-chemical properties required in soil amendmentapplication such as the surface alkalinity and acidity, electrical

(Ts) for sugar cane bagasse (BG, ), corn cob (CC, ), corn stover (CS, )e thermogravimetric setup.

Fig. 4. Derivative of thermogravimetric curves for the pyrolysis of feedstocks (CC,CS, BG and EC) obtained at 150 �C min�1.

M. Carrier et al. / Bioresource Technology 150 (2013) 129–138 137

conductivity and cation exchange capacity were mostly affected bythe presence of carbohydrates (i.e. hemicelluloses and cellulose).

3.5. Thermogravimetric results

Although the above-mentioned results demonstrated the keyrole of hemicelluloses on both quality and yields of fast pyrolysisproducts, no relevant conclusions on the impact of the chemicaland transport mechanisms driven by the chemical and structuralproperties of hemicelluloses could be drawn. Indeed the calculatedvalues of Bi number (>0.1; Table 4) within the BFB reactor indicatedthat the temperature profile within the particle was not uniformfor both feedstocks and thus that the pyrolysis reactions occurredin overlapping heat and mass transfer regimes. The thermogravi-metric set-up combined with the use of reduced particle size,250–450 lm, and a high heating rate, 150 K min�1, was used toreach the chemical regime without external resistance in heattransfer (Bi < 0.1) and thus to illustrate the differences in chemicalreactions between feedstocks during truely ‘fast’ thermal degrada-tion (Fig. 3). In general, reactions occurring under higher heatingrate have a short reaction time in the same temperature regionand therefore a faster apparent rate (kr) as illustrated by kr values,compared to reactions in the same temperature region with a slowheating rate (Table 4).

The influence of the feedstock nature on the rate of reaction andpyrolysis temperature range was respectively illustrated by thelags in heat up of the biomass at 150 K min�1 and the derivativesof thermogravimetric curves for the different plant biomasses(Figs. 3 and 4). BG and CS feedstocks, which required a higher heatdemand to be converted (Table 4), displayed a similar ability toconduct heat faster than the EC and CC feedstocks, resulting inshorter thermal lags (Fig. 3). This observation suggests that thephysical (fibrous) structure of feedstocks plays a more significantrole on the heat transfer phenomena than the lignocellulosicchemical composition, while the latter, specifically the high carbo-hydrates content, impacted the activation energy (Ea) under fastheating conditions (Table 4). The apparent Ea values in the presentstudy were similar to reported values for the cellulose pseudocom-ponent, and higher than reported Ea values for lignin and hemicel-luloses (Aboyade et al., 2011).

These results showed that the main constituents of biomassthrough their different structural and chemical nature played a sig-nificant role in both chemistry and transfer phenomena occurringduring biomass fast pyrolysis. In other words, these biopolymersshould be considered not only as chemicals, but also as physicalbiomaterials whose thermal, mechanical and structural properties

affect the pyrolysis reaction through their anisotropic transportproperties, thermal conductivity, density, shape, size of particle,etc.

4. Conclusion

The influence of the lignocellulosic feedstock composition onthe yields and quality of both liquid and char products from fastpyrolysis was investigated, together with the simultaneous occur-rence of transport and chemical phenomena in the bubbling fluid-ized bed reactor. Hemicelluloses combined with lignin of thefeedstock had a significant impact on the yield of pyrolytic waterand organics contained in bio-oil, obtained from fast pyrolysis.On the other hand, an increased high carbohydrates content pro-moted the development of microporous structure in biochars,which confirmed the potential use of pyrolytic char obtained fromfast pyrolysis as soil amendment.

Acknowledgements

This work was financially supported by Sasol and the PaperManufacturers Association of South Africa (PAMSA) as well asfunding from the Technology and Human Resources for IndustryProgram (THRIP) of the Department of Trade and Industry (dti) ofSouth Africa. The authors wish to thank these organizations fortheir support. The opinions expressed in this paper are that ofthe authors and should not be attributed to the sponsors.

References

Aboyade, A.O., Hugo, T.J., Carrier, M., Meyer, E.L., Stahl, R., Knoetze, H., Görgens, J.F.,2011. Non-isothermal kinetic analysis of the devolatilization of corn cobs andsugarcane bagasse in an inert atmosphere. Thermochim. Acta 517, 81–89.

Atkinson, C.J., Fitzgerald, J.D., Hipps, N.A., 2010. Potential mechanisms for achievingagricultural benefits from biochars application to temperate soils: a review.Plant Soil 337, 1–18.

Basu, P., 2010. Biomass gasification and pyrolysis practical design and theory. In:Biomass Characteristics. Elsevier Inc., UK, pp. 53 (Chapter 3).

Bridgewater, A.V., 2012. Review of fast pyrolysis of biomass and product upgrading.Biomass Bioenerg. 38, 68–94.

Bulushev, D.A., Ross, J.R.H., 2011. Catalysis for conversion of biomass to fuels viapyrolysis and gasification: A review. Catal. Today 171, 1–13.

Carrier, M., Neomagus, H.W.J.P., Görgens, J., Knoetze, J.H., 2012. Influence ofchemical pre-treatment on the internal structure and reactivity of pyrolysischars produced from sugar cane bagasse. Energy Fuel 26, 4497–4506.

Carrier, M., Loppinet-Serani, A., Denux, D., Lasnier, J.-M., Ham-Pichavant, F.,Aymonier, C., Cansell, F., 2011. Thermogravimetric analysis as a new methodto determine the lignocellulosic composition of biomass. Biomass Bioenerg. 35,298–307.

Couhert, C., Commandre, J.-M., Salvador, S., 2009. Is it possible to predict gas yieldsof any biomass after rapid pyrolysis at high temperature from its composition incellulose, hemicellulose and lignin? Fuel 88, 408–417.

Demeyer, A., Nkana, J.C.V., Verloo, M.G., 2001. Characteristics of wood ash andinfluence on soil properties and nutrient uptake: an overview. Bioresour.Technol. 77, 287–295.

Di Blasi, C., 2008. Modeling chemical and physical processes of wood and biomasspyrolysis. Prog. Energ. Combust. 34, 47–90.

Dutta, B., Raghavan, G.S.V., Ngadi, M., 2012. Surface characterization andclassification of slow and fast pyrolyzed biochar using novel methods ofpycnometry and hyperspectral imaging. J. Wood Chem. Technol. 32 (2), 105–120.

Enders, A., Kelly Hanley, K., Whitman, T., Joseph, S., Lehmann, J., 2012.Characterization of biochars to evaluate recalcitrance and agronomicperformance. Bioresour. Technol. 114, 644–653.

Fengel, D., Wegener, G., 2003. Wood Chemistry, Ultrastructure, Reactions. VerlagKessel, Germany.

Garcia-Perez, M., Wang, S.X., Shen, J., Rhodes, M.J., Tian, F., Lee, W., Wu, H., Li, C.,2008. Fast pyrolysis of mallee woody biomass: effect of temperature on theyield and quality of pyrolysis products. Ind. Eng. Chem. Res. 47, 1846–1854.

Gercel, Ö., Özcan, A., Özcan, A.S., Gercel, H.F., 2007. Preparation of activated carbonfrom a renewable bio-plant of Euphorbia rigida by H2SO4 activation and itsadsorption behavior in aqueous solutions. Appl. Surf. Sci. 253, 4843–4852.

Hosoya, T., Kawamoto, H., Saka, S., 2007. Cellulose–hemicellulose and cellulose–lignin interactions in wood pyrolysis at gasification temperature. J. Anal. Appl.Pyrol. 80, 118–125.

Iversen, T., 1985. Lignin-carbohydrate bonds in a lignin-carbohydrate complex fromspruce. Wood Sci. Technol. 19 (3), 243–251.

138 M. Carrier et al. / Bioresource Technology 150 (2013) 129–138

Joseph, S.D., Camps-Arbestain, M., Lin, Y., Munroe, P., Chia, C.H., Hook, J., vanZwieten, L., Kimber, S., Cowie, A., Singh, B.P., Lehmann, J., Foidl, N., Smernik, R.J.,Amonette, J.E., 2010. An investigation into the reactions of biochar in soil. Aust.J. Soil Res. 48, 501–515.

Laird, D.A., Fleming, P., Davis, D.D., Horton, R., Wang, B., Karlen, D.L., 2010. Impact ofbiochar amendments on the quality of a typical Midwestern agricultural soil.Geoderma 158 (3–4), 443–449.

Lédé, J., 1994. Reaction temperature of solid particles undergoing and endothermalvolatilization. Application to the fast pyrolysis of biomass. Biomass Bioenerg. 7,49–60.

MacLean, J.D., 1941. Thermal conductivity of wood. Pip. Air Cond. 13, 380–391.MacLeod, M., 2007. The top ten factors in kraft pulp yield. Paper and Timber 89 (4),

1–7.Mettler, M.S., Vlachos, D.G., Dauenhauer, P.J., 2012. Top ten fundamental

challenges of biomass pyrolysis for biofuels. Energy Environ. Sci. 5, 7797–7809.

Mullen, C.A., 2010. Bio-oil and biochar production from corn cobs and stover by fastpyrolysis. Biomass Bioenerg. 34, 67–74.

Oasmaa, A., Solantausta, Y., Arpiainen, V., Kuoppala, E., Sipilea, K., 2010. Fastpyrolysis bio-oils from wood and agricultural residues. Energy Fuel 24, 1380–1388.

Papadikis, K., Gu, S., Bridgwater, A.V., 2009. CFD modelling of the fast pyrolysis ofbiomass in fluidised bed reactors. Part B. Heat, momentum and mass transportin bubbling fluidised beds. Chem. Eng. Sci. 64, 1036–1045.

Pasangulapati, V., Ramachandriya, K.D., Kumar, A., Wilkins, M.R., Jones, C.L.,Huhnke, R.L., 2012. Effects of cellulose, hemicellulose and lignin onthermochemical conversion characteristics of the selected biomass. Bioresour.Technol. 114, 663–669.

Patwardhan, P.R., 2010. Understanding the product distribution from biomass fastpyrolysis. Iowa State University, Dissertation.

Preston, C.M., Schmidt, M.W.I., 2006. Black (pyronegic) carbon: a synthesis ofcurrent knowledge and uncertainties with special consideration of borealregions. Biogeosciences 3, 397–420.

Shen, D.K., Gu, S., Bridgwater, A.V., 2010. The thermal performance of thepolysaccharides extracted from hardwood: cellulose and hemicelluloses.Carbohydr. Polym. 82, 39–45.

Sluiter, J.B., Ruiz, R.O., Scarlata, C.J., Sluiter, A.D., Templeton, D.W., 2010.Compositional analysis of lignocellulosic feedstocks. 1. Review anddescription of methods. J. Agric. Food Chem. 58, 9043–9053.

Tomaschek, J., Ozdemir, E.D., Fahl, U., Eltrop, L., 2012. Greenhouse gas emissions andabatement costs of biofuel production in South Africa. Glob. Change Biol.Bioenergy 4, 799–810.

Uras, U., Carrier, M., Hardie, A., Knoetze, J.H., 2012. Physico-chemicalcharacterization of biochars from vacuum pyrolysis of South Africanagricultural wastes for application as soil amendments. J. Anal. Appl. Pyrol.98, 207–213.

Van de Velden, M., Baeyens, J., Brems, A., Janssens, B., Dewil, R., 2010. Fundamentals,kinetics and endothermicity of the biomass pyrolysis reaction. Renew. Energ.35, 232–242.

Vena, P.F., García-Aparicio, M.P., Brienzo, M., Görgens, J.F., Rypstra, T., 2013. Effect ofalkaline hemicellulose extraction on kraft pulp fibers from Eucalyptus Grandis. J.Wood Chem. Technol. 33, 157–173.

Von Maltitz, G., Stafford, W., 2012. Integrating bioenergy into a green economy:identifying opportunities and constraints, 4th CSIR biennial conference: realproblems relevant solutions. CSIR, Pretoria.

Vosloo, A.C., 2013. Bio-Energy in Southern Africa? A Commodity BusinessPerspective. Sasol’s Public Presentation.

Westerhof, R.J.M., Brilman, D.W.F., van Swaaij, W.P.M., Kersten, S.R.A., 2010. Effectof temperature in fluidized bed fast pyrolysis of biomass: oil quality assessmentin test units. Ind. Eng. Chem. Res. 49, 1160–1168.