Carbon Dioxide Torrefaction of Woody BiomassSiva Sankar Thanapal,† Wei Chen,† Kalyan Annamalai,† Nicholas Carlin,‡ Robert James Ansley,§

and Devesh Ranjan†,*†MS 3123, Department of Mechanical Engineering, Texas A&M University, College Station, Texas 77843, United States of America‡Agni Corporation, Concord, California 94520, United States of America§Texas Agrilife Research, Vernon, Texas 76384, United States of America

ABSTRACT: The major drawback of biomass for direct combustion applications is its lower heating values and poorgrindability when compared to conventional fossil fuels. Torrefaction is one of the thermochemical pretreatment techniques usedto improve the properties of biomass. An inert environment is maintained to prevent oxidation of biomass during torrefaction. Anovel method for utilization of carbon dioxide as the pretreatment medium for woody biomass has been investigated in thecurrent study. Previous studies on smaller samples using thermogravimetric analysis (TGA) showed an increased mass loss in theCO2 environment, which was attributed to possible structural changes in the biomass and potential effect of ash constituents inthe biomass. However, those claims were not validated. The current study on bigger batches of samples also resulted in increasedmass loss when using CO2 compared to using N2 as the torrefaction medium. Scanning electron microscopy (SEM) studies andBrunauer−Emmett−Teller (BET) surface area investigation on the torrefied samples from different environments showedincreased internal surface area indicating a mild effect of the CO2 reacting with the samples at temperatures normally employedfor torrefaction conditions (200−300 °C). Further, grindability studies were performed on the samples pretreated in CO2 andtorrefied in N2. The results on grindability showed improved grindability on using CO2 as the pretreatment medium. Proximateand heating value analysis on the pretreated samples showed an increasing trend in the heating value of the samples withincreased temperature. Comparable mass loss at lower temparatures improved grindability and improved fuel properties, makesutilization of carbon dioxide as a torrefaction medium for pretreating biomass for combustion applications an attractivetechnology.

1. INTRODUCTION

World energy consumption is projected to increase withincreasing world population and economic growth indeveloping countries. Biofuels are one of the renewable energysources that contribute about 10% of the total energy in theworld.1 Desired characteristics of energy crops that can be usedfor power generation were listed by Mckendry, 2002.2 Theyinclude high yield, low energy input for production, low cost,low contaminants, and low nutrient requirements. Large areasof the southwestern North America (60 million ha) have beenencroached by shrubs and woody plants. Climate change, highlevels of herbivory (livestock grazing), change in fire frequency,changes in grass competitive ability, spread of seed by livestock,small mammal populations, elevated levels of carbon dioxide(CO2), and a combination of these factors are considered to besome of the reasons for such a change.3 These woody plantscan be harvested and can be used as sources of bioenergy. Themain advantage of these woody plants is that they do not needto grow on more arable land better suited for growing food orfiber.4

Honey mesquite is a polymorphic woody legume under thedivision Magnoliophyta that occurs on grasslands and range-lands in southwestern U.S.A. and occupies over 21 million ha inTexas alone.4 The rate of increase in honey mesquite coverincreased significantly with increase percentage of 2.2% unitsper year.5 Redberry juniper is a basal sprouting conifer underthe division Pinophyta that has several stems arising from thebase.6 Its infestation has also increased by about 60% during the

period 1948 to 1982 in a 65 county region in northwest Texas.It was estimated that by the year 2000, redberry juniper wouldhave invaded around 4.9 million ha, or nearly a third, of the 65county region.Both mesquite and juniper, which have invaded the

grasslands, have a good heating value. A good heat contentcoupled with increased availability makes it a renewable energycrop that can be used as fuel for direct combustion orgasification.7 Harvesting the mesquite and utilizing it as abioenergy feedstock has the following advantages. There are noplanting, cultivation, irrigation, and fertilization costs for thisnaturally occurring species. Also, the dry mass of 10 year oldregrown mesquite (mesquite grown after harvest) was found tobe 29.4 kg/tree with a typical tree density of 750 trees/ha,which in turn gives an annual production of 2.2 t/ha/y.Mesquite and juniper occur in warm, dry climate, and they canbe harvested year round, thereby reducing fuel storage costs.4

Different thermochemical methods used for energy extrac-tion from biomass include gasification, combustion, pyrolysis,and torrefaction.8 The major drawback of biomass to be used indirect combustion applications is its lower heating value, highermoisture content, poor grindability, and lower bulk density.9

Torrefaction is one of the thermochemical pretreatmenttechniques that have been used to improve the biomass

Received: November 16, 2013Revised: January 3, 2014Published: January 5, 2014

Article

pubs.acs.org/EF

© 2014 American Chemical Society 1147 dx.doi.org/10.1021/ef4022625 | Energy Fuels 2014, 28, 1147−1157

properties with respect to heating value and grindability.Torrefaction is carried out in a temperature range from 200 to300 °C in an inert environment to prevent biomass oxidation.Even though volatile matter in the biomass also contains a smallpercentage of carbon, more oxygenated compounds arebelieved to be removed from the sample during torrefaction.Higher heat value also results from removal of moisture (H2O).Different gases that are used to maintain an inert environmentinclude nitrogen9−11 and argon,12 and recently, wet torrefactionusing hot compressed water was studied to improve the energydensity of biomass.13 The effect of using a small amount ofoxygen on torrefaction was studied by Rousset et al.14 andWang et al.15 The effect of using CO2 as the pretreatmentmedium was studied by Eseltine et al.16 Using CO2 resulted incomparatively higher mass loss at the temperature rangecommonly used for torrefaction.Biomass composition influences the effect torrefaction has on

the final product. Depending on whether the biomass is offibrous type and woody type the percentage of lignin, cellulose,and hemicellulose that make up the biomass will vary. Woodybiomass may be either a hardwood or softwood. Thepercentage of hemicelluloses in the softwood is lower whencompared to that of hardwood.17 During torrefaction andpyrolysis studies, it was observed that hemicelluose is thecomponent that degrades at a lower temperature range (220−315 °C) followed by cellulose (315−400 °C) and lignin (160−900 °C).18 Hence, the mass loss percentage will vary for thetorrefied samples based on the percentage of hemicelluose inthe raw sample for different temperatures used.Mesquite is a hardwood species while juniper is a softwood

as evidenced from their divisions. Plants under Magnoliophytaare angiosperms (hardwood), and those under Pinophyta aregymnosperms (softwood).17 Analysis of hardwood showsincreased presence of hemicelluloses when compared tosoftwood. Under the temperature range considered fortorrefaction studies, it was observed that the hemicellulosedegrades first followed by cellulose. Hence, a sample with ahigher amount of hemicelluloses would be expected to showincreased mass loss with increase in temperature duringtorrefaction due to breakdown of hemicelluloses.The hydrophilic nature of biomass is related to the presence

of OH groups in biomass. Hemicellulose was found to have thehighest potential to adsorb water, followed by cellulose andlignin. The reason behind the hydrophobic nature of torrefiedbiomass can be attributed to reduced amount of hemicelluloses

and OH groups in the torrefied biomass during torrefaction.17

Investigation on the moisture absorption tendency of thetorrefied biomass by Acharjee et al., 2011,19 and Medic et al.,2012,20 revealed lower moisture adsorption tendency of thetorrefied biomass when compared to the raw biomass.Carbon dioxide is one of the green house gases that is

released into the environment during combustion of fossil andrenewable sources. Availability of hot gases from boiler exhaustwith higher percentage of CO2 makes it an attractive option tobe used as the biomass pretreatment medium. Studies onutilizing CO2 for the pyrolysis of lignocellulosic biomassbetween 25 and 900 °C showed enhanced cracking of releasedvolatile species resulting in increased concentration of H2, CH4,and CO upon using CO2 compared to N2 at a heating rates of10 °C per minute and 500 °C per minute.21 CO2 also showed atendency to mitigate the production of polycyclic aromatichydrocarbons (PAH) during the pyrolysis of styrene butadienerubber from 25 to 1000 °C. The presence of CO2 as themedium of pyrolysis resulted in increased cracking of benzenederivatives and reduced gas phase addition to form PAH.22

Limited studies were done on the torrefaction capability ofCO2. Studies done on smaller amount of samples in TGAshowed an increased mass loss with increase in pretreatmenttemperature on using CO2 when compared to using N2 as thetorrefaction medium.16 However, the factors that might causesuch an increased mass loss were not fully understood.Considering the temperature limits for Boudouard reaction,the effect of it under the pretreatment conditions (200−300°C) should be studied further. In the current study, the effect ofpretreating mesquite and juniper in the presence of CO2 wasstudied in a batch type facility. The influence of using CO2 andN2 for pretreatment of biomass on mass yield, energy retained,grindability, and structural variation of biomass was studied.Further, the effect of particle size on the softwood andhardwood torrefaction was studied.

2. MATERIALS AND METHODS2.1. Torrefaction. Figure 1 shows the schematic of the batch type

facility used for the current study. A well insulated batch type reactorthat can pretreat around 500 g sample per batch was used for thecurrent study. A batch of sample (4) was first loaded into the reactor.The reactor was then closed with an assembly of auger (5) andbidirectional motor (6) in place to mix the samples and maintain thedesired pretreatment temperature within the reactor. N2/CO2 wasused to purge the reactor depending on the medium used fortorrefaction. A constant flow of 30 SCFH (0.85 m3/h) of N2/CO2 was

Figure 1. Schematic of the batch torrefaction facility: (1) flow controller, (2) thermocouples, (3) band heater, (4) biomass, (5) auger, (6)bidirectional motor, (7) condenser, (8) line filters, (9) mass spectrometer, (10) exhaust fan.

Energy & Fuels Article

dx.doi.org/10.1021/ef4022625 | Energy Fuels 2014, 28, 1147−11571148

set using variable area mass flow controller (1) to maintain an inert/nonreacting environment during the pretreatment period. A 1.8 kWband type electrical heater (3) was then turned on to heat the samplesat a rate of 20 °C per minute. Two k-type thermocouples (2)connected to the electrical heater were used to monitor thetemperature and control the supply of power to the heaters. Thesamples were heated from room temperature to the desiredtemperature and kept constant at that temperature for 30 min.Residence time of 30 min was chosen based on the results from Ariaset al., 2008,10 and Sadaka and Negi, 2009,23 wherein it was observedthat higher residence times (more than 30 min) had minor effect onmass loss behavior. An auger coupled to the bidirectional motor wasused to maintain a uniform temperature throughout the batch duringthe pretreatment period. A slightly negative pressure was maintainedwithin the reactor by means of a vacuum fan (10) to remove the gasesproduced during torrefaction as well as the medium used fortorrefaction. A shell and tube heat exchanger type condenser (7)was used to condense out any condensables from the gases producedduring torrefaction. Since some of the condensables were condensedalong the pathway, an accurate quantification of the condensables wasnot made. However, a change in color of the condensables from lightyellow to dark viscous liquid was observed with increase inpretreatment temperature. A small amount of the gases were filteredusing inline filters (8), and their composition was analyzed using aThermo Scientific Prolab mass spectrometer (9). The torrefactionmedium had the highest concentration of the different speciesmeasured. The mass spectrometer was calibrated with gas mixtures ofknown composition to get accurate measurements.2.2. Grindability. In order to study the grindability of torrefied

samples, all the samples were ground for a constant time period of 20min in a Sweco DM1vibro energy grinding mill. This procedure wasfollowed to have a constant power input for the mill to grind all thesamples. The particle size distribution was obtained using Ro-taptesting sieve shaker with U.S. standard sieves of numbers 8, 10, 20, 30,100, 200, and 270 (these sieve numbers represent the following sizes:2.4 mm, 2.2 mm, 1.42 mm, 715 μm, 370 μm, 112.5 μm, and 64 μmrespectively). The variation in particle size distribution with increase intorrefaction temperature was studied for both mesquite and junipersamples.2.3. TGA-DTA. Thermogravimetric analysis (TGA) of the raw

samples was carried out using TA Instruments SDT-Q600. Thesample (10 mg) was loaded into the sample pan, and 100 mL/min ofnitrogen was used to maintain an inert environment during the TGAstudy. The samples were heated at a constant rate of 20 °C/min fromroom temperature to 900 °C to study the sample behavior duringpyrolysis. Simultaneous measurements of weight loss in the samplepan and temperature difference between sample and reference panwere made to get the TGA and differential thermal analysis (DTA)trace. Torrefaction studies on the raw samples were performed usingN2 and CO2 to study for any difference in the DTA trace. Procedure ofthe torrefaction using TGA unit is available elsewhere.16

2.4. BET Analysis. BET analysis was performed on the torrefiedsamples using a Quantachrome NOVA 4200e instrument. Thesamples tested were ground and sieved to a size of between 300 to500 μm. The samples were initially degassed for 16 h at 75 °C toremove any adsorbed moisture and impurities. The adsoptionisotherms were then obtained at a constant temperature of 77 Kusing nitrogen as the medium for adsorption to determine the surfacearea of the torrefied samples.2.5. SEM Analysis. SEM images of the samples treated using

nitrogen and CO2 were obtained using JEOL JSM-7600 F. The imageswere obtained at a magnification of 2000× to identify the poresformed on the samples when using different torrefaction mediums.

3. RESULTS AND DISCUSSION

3.1. Fuel Preparation. Biomass fuels used for the currentstudy, mesquite and juniper, were harvested from therangelands in north central Texas near Vernon. Both mesquiteand juniper are scattered over a large area estimated to be 200

000 ha. It should be noted that the species is not concentratedthroughout the entire area but dispersed with a spacing ofaround 1 km. After the fuel is harvested using a chain saw, achipper is used to reduce the particle size. It was also observedthat, when a freshly harvested biomass (moisture contentaround 45%) was sent into the chipper for reducing the particlesize, the chips produced after the chipping process had a lowermoisture percentage of between 10 to 20%. This might bebecause of the drying of the woodchips within the chipperusing the heat produced as a result of the chipping process.Vermeer wood chippers were used for processing the biomass.Further details on the preparation of the samples are availableelsewhere.24 Previous small scale study on mesquite and juniperof particle size between 540 and 800 μm showed lower massloss for juniper when compared to mesquite in TGA.16 In orderto study for the effect of particle size on softwood andhardwood, smaller softwood juniper wood chips of size 2−4mm and comparatively larger mesquite wood chips of 4−6 mmwere used for the current torrefaction study.

3.2. Fuel Properties. The raw samples of mesquite andjuniper were sent to a commercial testing facility to get theultimate and proximate analyses. Table 1 gives the ultimate andproximate analyses of mesquite and juniper.

3.3. Experiments and Samples. Torrefaction studies weredone for temperatures from 200 to 300 °C in steps of 20 °C formesquite (M) and juniper (J). The temperatures selected werebelow the temperature of maximum volatile release rate. Twopretreatment mediums in form of N2(N) and CO2(C) wereused. Samples torrefied will be represented by the followingnomenclature: temperature-biomass-pretreatment mediumused. For example, 200-M-C will represent mesquite pretreatedwith CO2 at 200 °C. A few of the tests were repeated to checkfor repeatability of experiments. In total, 24 samples wereobtained from the torrefaction of two woody biomass undertwo different mediums for the six temperatures studied.

3.4. Torrefied Biomass Properties. Proximate analysiswas done on the torrefied samples. Moisture, volatile matter,and ash were determined using ASTM standards E871, E872,and E1755, respectively. The remaining mass in the sample isfixed carbon (Fixed carbon was estimated from the difference).Gross heating values of the samples were determined using abomb calorimeter according to ASTM test method E711. Table2 shows the results obtained for the proximate analysis of thesamples. The values for the volatile matter (VM), fixed carbon

Table 1. Properties of Mesquite and Juniper, Adopted fromChen et al., 201224

mesquite juniper

moisture 15.5 5.85volatile matter 66.1 78.0fixed carbon 16.7 14.3ash 1.67 1.91carbon 43.6 49.3oxygen 33.6 37.0hydrogen 4.98 5.68nitrogen 0.62 0.28sulfur 0.03 0.01HHV (kJ/kg) 16700 19000HHVDAF (kJ/kg) 20100 20600empirical formulaa CH1.37N0.01O0.58S0.0003 CH1.38N0.01O0.56S0.0001

aFormula carbon normalized.

Energy & Fuels Article

dx.doi.org/10.1021/ef4022625 | Energy Fuels 2014, 28, 1147−11571149

(FC), and higher heating value (HHV) or gross heat valueobtained on a dry basis and dry ash free basis (DAF) are alsopresented in Table 2.It can be observed from Table 2 that the ash percentage of

the samples treated using CO2 and N2 are higher than the rawbiomass samples. Though the ash percentage shows someminor deviations, it may be attributed to the nonuniformdistribution of ash within the samples. Assuming there is noloss in the ash components during torrefaction, using ashbalance

− =m m y m y( )0 loss ash 0 ash,0 (1)

− =⎛⎝⎜

⎞⎠⎟y

mm

y1ashloss

0ash,0

(2)

where mo, mloss, yash, and yash,0 stands for mass of the rawbiomass samples, mass lost during torrefaction from thebiomass samples, mass fraction of the ash in the torrefiedsample, and the initial percentage of ash in the raw biomasssample, respectively. From eq 2, the product of remaining dryash mass fraction and remaining mass fraction of fuel (= (m/m0) = ((m0 − mloss)/m0)) should remain constant. Using thedata obtained from the torrefied samples, the plot in Figure 2 isobtained for torrefied mesquite and juniper. With increase intemperature, the ash percentage fluctuated.From Figure 2, it can be seen that product approximately

remains constant (around 2% on a dry basis) and the ash tracertechnique is valid for the torrefied samples. Further work iscurrently underway to determine the mineral content in the rawbiomass ash and the torrefied biomass ash.

3.5. Mass Yield. Mass yield after torrefaction is definedaccording to eq 3 as the ratio of the amount of mass left afterpretreatment to the original mass of the raw biomass.

=mm

mass yield 100TB

RB (3)

where mTB and mRB represent the mass of the torrefied biomassand raw biomass, respectively. Variation in mass yield undertwo environments, CO2 (C) and N2 (N), is shown below inFigure 3 for mesquite (M) and juniper (J) on a dry ash free(DAF) basis. It can be seen that the mass loss wascomparatively higher when using CO2 as the pretreatmentmedium for the hardwood species mesquite, which has a highermoisture content when compared to juniper. Juniper, withlower moisture content, showed similar mass losses under twotorrefaction conditions at temperatures below 280 °C. Athigher temperatures, the mass loss was much higher for both

Table 2. Proximate Analysis of the Torrefied Samplesa

samplemoisture,ar (%)

ash, ar(%)

VM, ar(%)

FC, ar(%)

HHV, ar(kJ/kg)

ash, dry(%)

VM, dry(%)

FC, dry(%)

HHV, dry(kJ/kg)

VM, DAF(%)

FC, DAF(%)

HHV, DAF(kJ/kg)

mesqutie,raw

15.5 1.67 66.1 16.7 16700 1.98 78.2 19.8 20169 79.8 20.2 20169

juniper, raw 5.85 1.91 78.0 14.3 19000 2.03 82.8 15.1 20598 84.6 15.4 20598200-M-C 3.64 2.56 67.5 26.3 19298 2.66 70.0 27.3 20026 72.0 28.0 20573220-M-C 3.50 2.33 66.9 27.3 19658 2.41 69.3 28.3 20371 71.0 29.0 20874240-M-C 3.10 2.74 66.7 27.5 20785 2.82 68.8 28.3 21450 70.8 29.2 22074260-M-C 3.32 1.59 62.9 32.2 20661 1.64 65.1 33.3 21371 66.2 33.8 21727280-M-C 2.29 3.22 58.2 36.3 22274 3.29 59.5 37.2 22796 61.6 38.4 23572300-M-C 2.88 4.05 52.0 41.1 23101 4.17 53.5 42.3 23786 55.9 44.1 24821200-M-N 4.94 1.92 67.5 25.7 19125 2.02 71.0 27.0 20119 72.4 27.6 20535220-M-N 4.27 1.72 67.2 26.8 19416 1.79 70.2 28.0 20283 71.5 28.5 20653240-M-N 3.58 2.55 66.4 27.5 19869 2.64 68.9 28.5 20606 70.7 29.3 21166260-M-N 2.87 2.26 65.0 29.8 20621 2.32 67.0 30.7 21230 68.6 31.4 21735280-M-N 2.22 2.43 61.3 34.1 21971 2.48 62.7 34.9 22469 64.3 35.7 23042300-M-N 2.04 2.60 57.9 37.5 22733 2.66 59.1 38.3 23206 60.7 39.3 23839200-J-C 5.03 1.41 68.5 25.1 19372 1.48 72.1 26.4 20398 73.2 26.8 20705220-J-C 3.99 2.25 69.9 23.9 19602 2.34 72.8 24.9 20417 74.5 25.5 20906240-J-C 3.42 2.31 68.5 25.8 20242 2.39 70.9 26.7 20959 72.6 27.4 21472260-J-C 2.97 1.55 67.9 27.6 20418 1.60 70.0 28.4 21043 71.1 28.9 21385280-J-C 1.78 1.83 64.5 31.9 22905 1.87 65.6 32.5 23320 66.9 33.1 23764300-J-C 2.10 1.22 57.9 38.8 24206 1.25 59.2 39.6 24725 59.9 40.1 25038200-J-N 3.48 1.96 69.2 25.4 19909 2.03 71.7 26.3 20626 73.2 26.8 21054220-J-N 4.04 2.88 70.5 22.6 19492 3.00 73.5 23.5 20313 75.8 24.2 20941240-J-N 4.53 2.13 69.1 24.3 20906 2.23 72.4 25.4 21898 74.0 26.0 22397260-J-N 3.09 1.45 68.9 26.6 21577 1.49 71.1 27.4 22265 72.2 27.8 22602280-J-N 3.04 1.81 65.3 29.8 21829 1.87 67.4 30.8 22513 68.6 31.4 22941300-J-N 2.66 1.27 63.8 32.2 23390 1.30 65.6 33.1 24029 66.4 33.6 24347aar, as received basis; dry, dry basis; DAF, dry ash free basis; M, mesquite; J, juniper; C, carbon dioxide; N, nitrogen.

Figure 2. Ash tracer technique to show the ash balance in the raw andtorrefied samples. M, mesqutie; C, carbon dioxide; J, juniper; N,nitrogen.

Energy & Fuels Article

dx.doi.org/10.1021/ef4022625 | Energy Fuels 2014, 28, 1147−11571150

the species under study on using CO2 as the torrefactionmedium.Juniper with lower particle size showed higher VM loss when

compared to mesquite. Studies conducted on similar particlesizes (589−840 μm) for both the woody biomass (mesquiteand juniper) using a TGA showed increased mass losses formesquite samples when compared to juniper during thetorrefaction process.16 Softwood species (juniper) with lowerhemicellulose content showed higher mass loss than mesquite,which is a hardwood with higher hemicellulose. Hence, theeffect of particle size on the mass loss behavior of thelignocelluosic samples was compared in the current study tounderstand the torrefaction process on different wood types.CO2 had a minor effect on the softwood species whentemperature was lower than 280 °C compared to that ofmesquite. Different phenomenon that can be accounted forsuch behavior of biomass under these pretreatment mediumsinclude (a) higher specific heat of carbon dioxide whencompared to that of nitrogen which results in some heat beingremoved by the pretreatment mediums during the heatingprocess, (b) reaction of the pretreatment medium (CO2) withthe biomass fuels, (c) effect of ash contents in biomass, whichcan catalyze the reaction between pretreatment medium CO2with biomass, and (d) effect of particle size of the biomass.From the results obtained, it can be concluded that the particlesize can be altered in addition to using different torrefactionmediums to obtain desired mass loss from the samples.The behavior of hemicellulose, cellulose, and lignin content

in both juniper and mesquite can be predicted from the TGA-DTA trace. A biomass sample that has a higher percentage ofhemicellulose will exhibit a hump at lower temperatures ofaround 200−300 °C during pyrolysis process.25−28 TGA-DTG(thermogravimetric and differential thermograms) curvesobtained for mesquite and juniper pyrolysis under nitrogenenvironment is available elsewhere.16 Lower amount ofhemicelluloses in the softwood species juniper is evidentfrom the smaller hump in the DTG curve (dotted line) whencompared to that of mesquite. Since lower particle size ofjuniper was used in the current study, more hemicellulose islost at the temperature range of torrefaction for juniper whencompared to that of larger mesquite fuel particles. Though

mesquite has larger amount of hemicelluloses, larger particlesize has restricted the passage for the hemicelluloses at thecenter of the particle to escape.16

Based on equilibrium concepts for reaction C + CO2 →2CO, it has been shown that the Boudouard reaction isthermodynamically favorable only at temperatures above 710°C21 (i.e., ΔG < 0 at T > 710 °C), called the transitiontemperature, which leads to an equilibrium constant valuegreater than 1. A higher value for the equilibrium constantindicates that the mole fraction of CO will be much higher thanthe mole fraction of CO2 at temperatures above 710 °C. Theeffect of CO2 reacting with fixed carbon in the biomass attemperatures used for the present study was considered to benegligible. In order to validate the temperature and timedependence, the Boudouard reaction kinetics was obtainedfrom the literature.29,30 Assuming CO2 reacts with the carbonin the biomass, the mass loss rate can be given by

= −⎜ ⎟⎛⎝

⎞⎠

Wt

k p S WE

RTdd

expm0 s m (4)

where ps is the partial pressure of species, Sm is the specificsurface area of the particle, and W is the weight of the particle.The values of the constants m, k0, and E in eq 4 are availableelsewhere.30 The effect of the Boudouard reaction at lowertemperatures and increased residence times (60 min) wasstudied. Curves obtained for the percent mass loss for differenttemperatures with respect to residence time shows anincreasing trend in mass loss with increased residence times.Figure 4 shows the results obtained from the Boudouard

reaction kinetics for the case of coal char. Higher mass loss wasobserved when the temperatures were increased beyond thetemperature range used for the current torrefaction studiesindicating temperature and time dependent mass loss. It shouldbe noted that the value of the constants used in eq 4 werederived for coal chars with higher surface area. Though thebiomass undergoing torrefaction will not have a high surfacearea, the above model can be used as a reference to validate thetime dependency of Boudouard reaction at the temperaturerange used for torrefaction study.

Figure 3. Mass retained after pretreatment in CO2 and N2. Graphs arepresented for the dry ash free (DAF) case.

Figure 4. Effect of residence time and temperature on the Boudouardreaction. Temperatures above 300 °C show higher mass loss withrespect to residence time.

Energy & Fuels Article

dx.doi.org/10.1021/ef4022625 | Energy Fuels 2014, 28, 1147−11571151

Use of carbon dioxide as the pretreatment medium has animpact on the mass loss behavior of the biomass. A smallamount of mass loss due to the reaction of carbon with carbondioxide will cause a slight increase in the pore spaces availablefor the volatiles trapped within the biomass particle to leave theparticle. A TGA unit was used to study the time dependency ofCO2 reacting with the biomass. A juniper sample (10 mg) ofparticle size 300 μm was used for this study. The biomass washeated at a constant rate of 20 °C per minute from roomtemperature to 240 °C, and the temperature was maintained at240 °C for three different time periods (15 min, 30 min, and 60min). After the isothermal stage, the samples were heated againto 1000 °C at a heating rate of 20 °C/min. More details on theprocedure for torrefaction using TGA is available elsewhere.16

Two different mediums (N2 and CO2) were used to study themass loss behavior during the torrefaction stage (isothermalperiod). Figure 5 shows the mass loss for different mediums atthree different residence times.

It can be seen that lower residence time (15 min) did nothave any impact on the mass loss upon using CO2 as similarmass losses were observed with both mediums. However, withincrease in residence times (30 and 60 min), using CO2

resulted in higher mass loss indicating a mild effect of thereaction of CO2 with biomass carbon at higher residence times.Biomass treated with different mediums will, in turn, havedifferent kinetic parameters. Effect of the pretreatmentmediums on the kinetic parameters of biomass will bepresented in a future publication.3.6. Energy Yield. Pretreated biomass with higher fixed

carbon and lower oxygenated compounds and moisture contentwill have higher heating value than the raw virgin biomass.However, in order to account for the mass loss associated withthe pretreatment process, a term called Energy yield is used andis defined according to eq 5.31

=energy yield mass yieldHHVHHV

100TB

Raw (5)

where HHVTB and HHVRaw are the higher heating values of thetorrefied biomass and raw biomass, respectively. Heating valueof the volatile matter of the biomass can be estimated on a dryash free basis according to approximate eq 6,32,33 which ignoresheat of pyrolysis

= × + −HHV VM HHV (1 VM)HVfuel,DAF VM FC (6)

where HHVfuel,DAF is the dry ash free heating value of thebiomass, VM is the fraction of volatile matter in the biomass,HHVVM is the heating value of the volatile matter, and HVFC isthe heating value of fixed carbon. Using eq 4, the averageheating value of the VM for mesquite and juniper are estimatedto be 17 000 kJ/kg and 18 400 kJ/kg on a dry ash free basis.Equation 6 can be used to estimate the increase in heating valueof the treated biomass with respect to the raw biomass whensome of the volatiles are released during pretreatment. Figure 6

is a plot of the results from the model and experiments. In themodel, it was assumed that the heating value of the volatilematter remains constant and does not change through theliberation process.However, it can be seen that the variation of heating value

with respect to the amount of volatile matter released was muchhigher in the experiments than the model value due to thevariation in VM heating value; that is, the model presumes thatHV of volatiles remain constant. As the oxygenated compoundsare released, the heating value of the remaining volatile matterin the biomass will have a higher value than the initial heatingvalue of the biomass VM. It can also be observed from Figure 6that juniper with lower particle size shows a higher VM releaseat lower temperatures resulting in higher heating values of thetorrefied biomass. Both the pretreatment mediums showedcomparatively similar loss in volatile matter at temperaturesbelow 250 °C and higher mass loss at higher temperatures forthe mesquite samples. However, juniper shows a higher releasein VM when pretreated with CO2.

Figure 5. Mass loss behavior of juniper samples torrefied at 240 °C fordifferent residence times using N2 and CO2 as the torrefaction mediumin a TA Instruments SDT Q600 TGA unit. J, juniper; C, carbondioxide; N, nitrogen; 15, 30, and 60 denote the isothermal time periodin minutes for torrefaction.

Figure 6. Variation of biomass heating value with release of VM fromthe biomass. M model and J model represents the modeled increase inthe heating value of mesquite and juniper, respectively, with release ofVM.

Energy & Fuels Article

dx.doi.org/10.1021/ef4022625 | Energy Fuels 2014, 28, 1147−11571152

The energy yield of the torrefied samples obtained using eq 5is presented in Figure 7. A rise in the energy yield valueindicates an increase in heating value and lower mass lossbehavior at that particular temperature.

Both mesquite and juniper showed an increase in energyyield at temperatures around 250 °C, which suggested that thisis the optimum temperature for torrefaction of these woodybiomass samples. Though further increase in temperatureshows an increase in energy yield for juniper samples, it willresult in higher loss in combustible volatile matter from thebiomass. Typical temperatures at which maximum volatilerelease rate occurs is 710 K or 437 °C for coal; 604 K or 331 °Cfor dairy biomass, 651 K or 378 °C for juniper; and 628 K or355 °C for mesquite.34 All values obtained are at heating rate of20 °C/min. Torrefaction at these temperatures will lead to veryrapid volatile loss. If the heating rate is raised to 100 °C/min,these temperatures raise from 25 to 50 °C. The recommendedtemperature of 250 °C is much below these temperature values.Though mesquite has an energy yield that is greater than theheat content of the raw biomass, it can be attributed primarilyto the loss of oxygenated compounds in the volatile matter, andthere is also an uncertainty associated with the measuredheating values.3.7. Grindability. Torrefaction has been shown to improve

the grindability of biomass resulting from breakdown of fibresand increased porosity of the pretreated biomass.35,36 In thepresent study, samples treated in nitrogen and CO2 wereground for a constant time period (20 min) in a grinding mill.The total energy consumed by the grinding mill for grindingthe samples for 20 min was estimated to be 0.8 MJ. Sizedistribution of ground samples was studied. Figures 8 and 9show the size distribution of mesquite samples pretreated inCO2 and N2, respectively.Size distribution of a sub-bituminous powder river basin

(PRB) coal ground in a Vortec mill and raw mesquite groundfor 20 min in the grinding mill is also presented for comparison.It should be noted that the PRB coal was ground in a vortecmill for a sufficient time so that 70% of particles are less than 75μm. However, the torrefied biomass was ground only for 20min to study improvement in grindability. Results obtained

from the size distribution of torrefied juniper ground in thegrinding mill for 20 min showed a trend similar to mesquite onusing CO2 and N2.It can be observed from Figures 8 and 9 that the use of CO2

as the pretreatment medium improves the grindability of thebiomass. A higher percent of ground samples passes throughthe smaller sieves, indicating better size reduction. Analysis ofSauter Mean Diameter (SMD), defined according to eq 7,33

will give a better understanding on the grindability of pretreatedbiomass.

=∑

∑

Nd

NdSauter Mean Diameter (SMD) i i

i i

p3

p2

(7)

where dp is the diameter of the particle collected in each sieveand N is the number of particles in each size group. Figure 10

Figure 7. Variation in the energy yield for both mesquite and junipersamples. Uncertainty for all the measurements was around 7%.

Figure 8. Comparison of grindbility of the CO2 pretreated mesquiteexpressed according to the percent biomass passing the sieves ofdifferent sizes. MES-R-G refers to raw mesquite samples ground for 20min in the grinding mill.

Figure 9. Grindability of mesquite torrefied in nitrogen.

Energy & Fuels Article

dx.doi.org/10.1021/ef4022625 | Energy Fuels 2014, 28, 1147−11571153

shows the variation in Sauter Mean Diameter (SMD) of thesamples torrefied in CO2 and N2.

3.8. Surface Area Analysis. Such an improved grindabilitycan be linked to the increase in the number of pores or in otherwords increased porosity of the samples pretreated in CO2.BET surface area analysis and SEM image analysis of thetorrefied samples were done to study the effect of CO2 mediumon the development of pores. In order to verify the effect ofusing CO2 as the torrefaction medium, BET surface area of thesamples torrefied with CO2 and N2 was analyzed usingquantochrome NOVA e4200 instrument. Samples torrefied at300 °C under N2 and CO2 were analyzed for their surface area.Adsorption isotherms obtained for the samples resembled atypical type-I isotherm37 for samples with micropores (poressmaller than 2 nm). Table 3 shows the results obtained fromthe BET analysis.

It is noted that BET surface area in CO2 is 55% more forJuniper and 29% more for mesquite. The samples pretreatedwith CO2 showed a comparatively higher surface area whencompared to the samples torrefied with N2. The average poreradius obtained was much higher than the average diameter ofCO2 (3.94 Å) and N2 (3.798 Å) molecules estimated usingcollision diameter values.38 Hence, the CO2 and N2 moleculescan easily diffuse into the voids created in the biomass particlesat higher residence times. Though the numbers for the surfacearea are much smaller, it should be noted that these are thenumbers obtained for the ground samples. This is consistentwith the study conducted by Pilon and Lavoie, 2011,39 andGray et al., 1985.40 Pilon and Lavoie39 showed the BET surfacearea of the ground switch grass sample treated at 300 °C also to

be less than 1 m2/g. Gray et al., 1985,40 listed the surface area ofthe wood char obtained from the pyrolysis of wood waste at atemperature of 330 °C in nitrogen environment to be less than0.8 m2/g. The effect of treatment temperature on the apparentchar surface area was reported by Valenzuela-Calahorro et al.,1987.41 Figure 11 shows the plot of surface area of chardetermined at various temperatures.

The surface area of the char pretreated at lower temperaturessay less than 500 °C has a low surface area as evidenced fromFigure 11. Since torrefaction is carried out at much lowertemperatures, and there is only partial pyrolysis, very highsurface areas will not be observed for the torrefied samples. Theexternal surface area per unit mass of a particle can be estimatedusing the following eq 8.33

ρ=S

d6

( )ext,mp p (8)

where ρp is the apparent particle density and dp is the diameterof the particle. Since the diameter of the particles of mesquiteand juniper used for the current study was around 2 mm, theexternal surface area per unit mass was estimated to be 0.075m2/g assuming an apparent particle density of 400 kg/m3 forraw wood.36 The results obtained from the BET internalsurface area were around 0.750 m2/g for the case of mesquitetorrefied at 300 °C with CO2 (Table 3). Further, internalsurface area increases with increase in carbon burnout.33 SEMimages of the torrefied and raw biomass was studied to analyzethe presence of pores on samples torrefied under twoenvironments. Figure 12 shows the SEM images of the rawjuniper sample and torrefied juniper sample.It is evident from the SEM images that more pores were

visible in the samples torrefied using CO2. It is clear from theBET analysis and SEM images that Boudouard reaction indeedhas a minor effect on the torrefaction temperature range owingto increased residence times. Though not much fixed carbonreacts with the CO2 during the pretreatment process, sufficient

Figure 10. Variation of SMD of the ground torrefied samples ofmesquite and Juniper.

Table 3. BET Surface Area of the Ground Torrefied Biomass

fuel mediumBET surface area

(m2/g)total pore vol.

(cc/g)avg. poreradius (Å)

300-J-C CO2 0.358 ± 0.040 0.000182 10.12300-J-N N2 0.231 ± 0.150 0.000116 9.708300-M-C CO2 0.747 ± 0.350 0.000365 9.845300-M-N N2 0.577 ± 0.280 0.000275 9.746

Figure 11. Effect of temperature on the apparent surface area of charobtained from holm-oak wood. Different symbols stand for methodsused to pyrolyze the char. Adapted from Valenzuela-Calahorro et al.,1987.41

Energy & Fuels Article

dx.doi.org/10.1021/ef4022625 | Energy Fuels 2014, 28, 1147−11571154

reaction occurs to form small voids on the surface of thebiomass (i.e., creating more pore space). These voids pave theway for the release of volatile matter from the biomass.Increased mass loss on using CO2 as the torrefaction mediumcan be attributed to this behavior of the biomass.The effect of increased porosity on using CO2 as the

torrefaction medium can also be observed from the proximateanalysis results of the samples (Table 2). The amount ofvolatile matter in the biomass treated with CO2 on a dry ashfree basis was lower than the biomass treated with N2 indicating

the removal of volatile matter from the formed pores. Acorresponding increase in FC was observed with decrease inVM within the torrefied biomass.

3.9. DTA Analysis of the Samples. Mesquite and junipersamples were torrefied in the TGA instrument using nitrogenand CO2. DTA traces were obtained along with the weight losstrace during the torrefaction period. Figure 13 shows the DTA

plot and weight % for juniper sample torrefied at 240 °C for 30min. DTA is plotted with respect to time and TGA trace(Figure 14) is plotted with respect to temperature. The use ofCO2 as the pretreatment medium causes an increased mass loss.DTA trace gives the difference in sample temperature duringtorrefaction in CO2 and N2. Torrefaction is known to beslightly exothermic, although many times moisture evaporation(endothermic) usually dwarfs the heat released from

Figure 12. SEM images. (a) Raw juniper sample, (b) juniper sampletorrefied at 300 °C with CO2, (c) juniper sample torrefied at 300 °Cwith N2.

Figure 13. DTA trace for torrefaction of juniper at 240 °C usingnitrogen and carbon dioxide. DTA trace shows the differential thermalanalysis with respect to time.

Figure 14. TGA trace for torrefaction of juniper at 240 °C using twodifferent mediums. TGA trace shows the weight loss with respect totemperature.

Energy & Fuels Article

dx.doi.org/10.1021/ef4022625 | Energy Fuels 2014, 28, 1147−11571155

torrefaction. A slightly more exothermic behavior was observedfor the CO2 torrefaction when compared to N2 torrefactionduring the first 15 min followed by a slight endothermicreaction indicating the time dependence of endothermicBoudouard reaction.

4. SUMMARY AND CONCLUSION

Mesquite and juniper, which have invaded the southwest part ofthe U.S., can be used as a fuel for combustion applications.Torrefaction of these woody biomass serves to improve theproperties of biomass in terms of increased heating value,improved grindability, and hydrophobicity. Effect of using N2and CO2 as the medium for torrefaction was studied.

(1) Comparable mass losses were observed on using the twomediums (CO2 and N2) for temperatures lesser than 250°C. Lower particle size results in higher mass losses whencompared to that of larger particles.

(2) Comparing the mass and energy yield, torrefaction at 240°C seems to be the optimum temperature fortorrefaction of mesquite and juniper biomass.

(3) Effect of using CO2 as the torrefaction medium serves toimprove the grindability of the biomass because of theincreased surface area caused due to the formation ofpores on the biomass samples.

Hence, the exhaust gases from the boilers with higherpercentage of CO2 can be used as a medium for torrefaction.Though gases from boiler exhaust also contains a small amountof oxygen due to slightly lean combustion, influence of 3−6%oxygen on the torrefaction process is found to be minimal.15

Hence, the exhaust gases with CO2, O2 and N2 can be utilizedfor torrefaction applications. Though a small amount of CO2may react with the biomass, the effect of such reaction can becontrolled by varying the particle size of biomass torrefied.Considering the results obtained, it should be noted thatfurther studies should be done to characterize the compositionof the condensed tar, the gases released, and the changes to thepercentages of the three components (hemicellulose, cellulose,and lignin) along with three component modeling studies inorder to gain a better understanding about increased masslosses upon using CO2. Increased surface area of the particlestorrefied with CO2 improves the grindability. The effect ofincreased surface area on the adsorption of harmful emissionssuch as mercury should be investigated along with the foulingpotential of torrefied biomass. If the boiler exhaust gases can beutilized for the pretreatment process, it will serve to improvethe thermal efficiency of the plant in addition to producing afuel with a good heating value and improved grindability.Hence, CO2 pretreatment seems to be an attractive option topretreat the biomass for combustion applications.

■ AUTHOR INFORMATION

Corresponding Author*E-mail: dranjan@tamu.edu.

Author Contributions∥The manuscript was written through contributions of allauthors. All authors have given approval to the final version ofthe manuscript. All authors contributed equally.

NotesThe authors declare no competing financial interest.

■ ACKNOWLEDGMENTS

The authors acknowledge the help and support from Dr. SergioCapareda and Dr. Bjorn Santos for the BET surface areaanalysis runs on the torrefied samples in Bioenergy Testing andAnalysis Lab (BETA lab), Texas A&M University. The authorsthank Dr. Yardanos Bisrat of Materials CharacterizationFacility, Texas A&M University, for her help with SEMimage analysis of the torrefied samples. The authors acknowl-edge the financial support from 2013 Texas A&M University-CONACYT: Collaborative Research Grant Program.

■ REFERENCES(1) IEA. www.iea.org/publications/freepublications/publication/kwes.pdf (accessed May 2013).(2) McKendry, P. Energy production from biomass (part 1):overview of biomass. Bioresour. Technol. 2002, 83, 37−46.(3) Van Auken, O. W. Shrub invasions of North American semiaridgrasslands. Annu. Rev. Ecol. Syst. 2000, 197−215.(4) Ansley, R. J.; Mirik, M.; Castellano, M. J. Structural biomasspartitioning in regrowth and undisturbed mesquite (Prosopisglandulosa): Implications for bioenergy uses. GCB Bioenergy 2010, 2,26−36.(5) Ansley, R. J.; Ben Wu, X.; Kramp, B. A. Observation: Long-termincreases in mesquite canopy cover in a north Texas savanna. J. RangeManage. 2001, 171−176.(6) Ueckert, D. N.; Phillips, R. A.; Petersen, J. L.; Ben Wu, X.;Waldron, D. F. Redberry juniper canopy cover dynamics on westernTexas rangelands. J. Range Manage. 2001, 603−610.(7) Chen, W.; Thanapal, S. S.; Annamalai, K.; Ansley, R. J.; Mirik, M.Updraft gasification of mesquite fuel using air-steam and CO2/O2

mixture. Energy Fuels 2013, 27, 7460−7469.(8) Thanapal, S. S.; Annamalai, K.; Sweeten, J. M.; Gordillo, G. Fixedbed gasification of dairy biomass with enriched air mixture. Appl.Energy 2012, 97, 525−531.(9) Bridgeman, T. G.; Jones, J. M.; Shield, I.; Williams, P. T.Torrefaction of reed canary grass, wheat straw, and willow to enhancesolid fuel qualities and combustion properties. Fuel 2008, 87, 844−856.(10) Arias, B.; Pevida, C.; Fermoso, J.; Plaza, M. G.; Rubiera, F.; Pis,J. J. Influence of torrefaction on the grindability and reactivity ofwoody biomass. Fuel Process. Technol. 2008, 89, 169−175.(11) Almeida, G.; Brito, J. O.; Perre,́ P. Alterations in energyproperties of eucalyptus wood and bark subjected to torrefaction: Thepotential of mass loss as a synthetic indicator. Bioresour. Technol. 2010,101, 9778−9784.(12) Prins, M. J.; Ptasinski, K. J.; Janssen, F. J. Torrefaction of wood:Part 2. Analysis of products. J. Anal. Appl. Pyrolysis 2006, 77, 35−40.(13) Yan, W.; Hastings, J. T.; Acharjee, T. C.; Coronella, C. J.;Vasquez, V. R. Mass and energy balances of wet torrefaction oflignocellulosic biomass. Energy Fuels 2010, 24, 4738−4742.(14) Rousset, P.; Macedo, L.; Commandre,́ J. M.; Moreira, A.Biomass torrefaction under different oxygen concentrations and itseffect on the composition of the solid by-product. J. Anal. Appl.Pyrolysis 2012, 96, 86−91.(15) Wang, C.; Peng, J.; Li, H.; Bi, X. T.; Legros, R.; Lim, C. J.;Sokhansanj, S. Oxidative torrefaction of biomass residues anddensification of torrefied sawdust to pellets. Bioresour. Technol. 2012,127, 318−325.(16) Esletine, D.; Thanapal, S. S.; Annamalai, K.; Ranjan, D.Torrefaction of woody biomass (juniper and mesquite) using inert andnon-inert gases. Fuel 2013, 113, 379−388.(17) Rowell, R. M. Handbook of Wood Chemistry and WoodComposites; CRC press: Boca Raton, FL, 2012.(18) Yang, H.; Yan, R.; Chen, H.; Lee, D. H.; Zheng, C.Characteristics of hemicellulose, cellulose, and lignin pyrolysis. Fuel2007, 86, 1781−1788.

Energy & Fuels Article

dx.doi.org/10.1021/ef4022625 | Energy Fuels 2014, 28, 1147−11571156

(19) Acharjee, T. C.; Coronella, C. J.; Vasquez, V. R. Effect ofthermal pretreatment on equilibrium moisture content of lignocellu-losic biomass. Bioresour. Technol. 2011, 102, 4849−4854.(20) Medic, D.; Darr, M.; Shah, A.; Rahn, S. Effect of torrefaction onwater vapor adsorption properties and resistance to microbialdegradation of corn stover. Energy Fuels 2012, 26, 2386−2393.(21) Kwon, E. E.; Jeon, E. C.; Castaldi, M. J.; Jeon, Y. J. Effect ofcarbon dioxide on the thermal degradation of lignocellulosic biomass.Environ. Sci. Technol. 2013, 47, 10541−10547.(22) Kwon, E. E.; Yi, H.; Castaldi, M. J. Utilizing carbon dioxide as areaction medium to mitigate production of polycyclic aromatichydrocarbons from the thermal decomposition of styrene butadienerubber. Environ. Sci. Technol. 2012, 46, 10752−10757.(23) Sadaka, S.; Negi, S. Improvements of biomass physical andthermochemical characteristics via torrefaction process. Environ. Prog.Sustainable Energy 2009, 28, 427−434.(24) Chen, W.; Annamalai, K.; Ansley, R. J.; Mirik, M. Updraft fixedbed gasification of mesquite and juniper wood samples. Energy 2012,41, 454−461.(25) Chen, W. H.; Hsu, H. C.; Lu, K. M.; Lee, W. J.; Lin, T. C.Thermal pretreatment of wood (Lauan) block by torrefaction and itsinfluence on the properties of the biomass. Energy 2011, 36, 3012−3021.(26) Cozzani, V.; Lucchesi, A.; Stoppato, G.; Maschio, G. A newmethod to determine the composition of biomass by thermogravi-metric analysis. Can. J. Chem. Eng. 1997, 75, 127−133.(27) Biagini, E.; Barontini, F.; Tognotti, L. Devolatilization ofbiomass fuels and biomass components studied by TG/FTIRtechnique. Ind. Eng. Chem. Res. 2006, 45, 4486−4493.(28) Chen, W. H.; Tu, Y. J.; Sheen, H. K. Disruption of sugarcanebagasse lignocellulosic structure by means of dilute sulfuric acidpretreatment with microwave-assisted heating. Appl. Energy 2011, 88,2726−2734.(29) Howard, J. B., Sarofim, A. F. Gasification of coal char withcarbon dioxide and steam at 1200−1800 C. Massachusetts Institute ofTechnology, Cambridge, MA, 1978.(30) Annamalai, K.; Ryan, W. Interactive processes in gasification andcombustionII. Isolated carbon, coal and porous char particles. Prog.Energy Combust. Sci. 1993, 19, 383−446.(31) Bergman, P. C., Boersma, A. R., Kiel, J. H., Prins, M. J., Ptasinski,K. J., & Janssen, F. J. J. G.: Torrefaction for entrained-flow gasificationof biomass. In The second world conference and technology exhibition onbiomass for energy, industry and climate protection: Rome, Italy, 2004.(32) Sweeten, J. M.; Annamalai, K.; Thien, B.; McDonald, L. A. Co-firing of coal and cattle feedlot biomass (FB) fuels. Part I. Feedlotbiomass (cattle manure) fuel quality and characteristics. Fuel 2003, 82,1167−1182.(33) Annamalai, K., Puri, I. K. Combustion Science and Engineering;CRC Press: Boca Raton, FL, 2007; Vol. 7.(34) Hoffmann, L. Pyrolysis and Ignition studies on Coal andBiomass blends using TGA and FTIR. Texas A&M University, CollegeStation, TX, 2011.(35) Bridgeman, T. G.; Jones, J. M.; Williams, A.; Waldron, D. J. Aninvestigation of the grindability of two torrefied energy crops. Fuel2010, 89, 3911−3918.(36) Phanphanich, M.; Mani, S. Impact of torrefaction on thegrindability and fuel characteristics of forest biomass. Bioresour.Technol. 2011, 102, 1246−1253.(37) Sing, K. S. W.; Everett, D. H.; Haul, R. A. W.; Moscou, L.;Pierotti, R. A.; Rouquerol, J.; Siemieniewska, T. Reportingphysisorption data for gas/solid systems with special reference tothe determination of surface area and porosity. Pure Appl. Chem. 1985,57, 603−619.(38) Annamalai, K., Puri, I. K., Jog, M. A.: Advanced ThermodynamicsEngineering, 2nd ed.; CRC Press: Boca Raton, FL, 2011.(39) Pilon, G.; Lavoie, J. M. Characterization of switchgrass charproduced in torrefaction and pyrolysis conditions. Bioresources 2011, 6,4824−4839.

(40) Gray, M. R.; Corcoran, W. H.; Gavalas, G. R. Pyrolysis of awood-derived material. Effects of moisture and ash content. Ind. Eng.Chem. Process Des. Dev. 1985, 24, 646−651.(41) Valenzuela-Calahorro, C.; Bernalte-Garcia, A.; Gomez-Serrano,V.; Bernalte-García, M. J. Influence of particle size and pyrolysisconditions on yield, density and some textural parameters of charsprepared from holm-oak wood. J. Anal. Appl. Pyrolysis 1987, 12, 61−70.

Energy & Fuels Article

dx.doi.org/10.1021/ef4022625 | Energy Fuels 2014, 28, 1147−11571157