Fuel 113 (2013) 379–388

Contents lists available at SciVerse ScienceDirect

Fuel

journal homepage: www.elsevier .com/locate / fuel

Torrefaction of woody biomass (Juniper and Mesquite) using inertand non-inert gases

0016-2361/$ - see front matter � 2013 Elsevier Ltd. All rights reserved.http://dx.doi.org/10.1016/j.fuel.2013.04.085

⇑ Corresponding author. Tel.: +1 979 458 3580; fax: +1 979 458 3081.E-mail addresses: eseltined@tamu.edu (D. Eseltine), sivsan.siva@tamu.edu

(S.S. Thanapal), kannamalai@tamu.edu (K. Annamalai), dranjan@tamu.edu(D. Ranjan).

Dustin Eseltine, Siva Sankar Thanapal, Kalyan Annamalai, Devesh Ranjan ⇑Department of Mechanical Engineering, MS 3123 Texas A&M University, College Station, TX 77843, United States

h i g h l i g h t s

� Encroachment of woody biomass in rangelands and grasslands of central Texas.� Torrefaction as viable pretreatment option for Mesquite and Juniper.� Effect of carbon dioxide as medium for torrefaction.� Increased weight loss and improved grindability on using CO2 for torrefaction.

a r t i c l e i n f o

Article history:Received 30 May 2012Received in revised form 25 April 2013Accepted 29 April 2013Available online 29 May 2013

Keywords:TorrefactionCarbon-dioxideTGA

a b s t r a c t

Torrefaction has been shown to be a viable pre-treatment process for thermally upgrading biomass andmaking it better suited for use as a fuel. Results are presented here from the torrefaction of two differentwoody biomass (Mesquite and Juniper) using thermogravimetric analysis (TGA). Samples were torrefiedin two different mediums (Nitrogen and Carbon-dioxide) to investigate the impact of inert and non-inertgases on the thermal degradation of the biomass, specifically mass loss as well as the breakdown of hemi-cellulose, cellulose, and lignin. A wide range of torrefaction temperatures ranging from 200 to 300 �C wasinvestigated at a single residence time. TGA thermograms showed increased amounts of weight losswhen the biomass was torrefied in a Carbon-dioxide environment compared to inert environments, whilederivative thermogravimetric analysis (DTA) showed a drop in the peak rate of hemicellulose and cellu-lose weight loss when using Carbon-dioxide as the torrefaction medium. Fuel properties analysis of sam-ples subjected to a mild torrefaction process (240 �C) showed increases in ash and fixed carbon content ofthe biomass, independent of torrefaction medium. Furthermore, comparison of the grindability of rawand torrefied samples showed torrefaction in a Carbon-dioxide environment further improved the grin-dability of the torrefied biomass. The effect of inert and non-inert torrefaction mediums on the volatilespecies (CO2, CO, CH4, and C2H6) released from torrefied Mesquite during pyrolysis was further investi-gated via TGA–FTIR analysis.

� 2013 Elsevier Ltd. All rights reserved.

1. Introduction reburning. Pyrolysis is a process of treating the biomass at very

The continued use of fossil fuels worldwide has lead to in-creased research in the area of non-conventional energy technolo-gies. Of the different non-conventional energy sources available(wind, solar, tidal, nuclear), biomass is a widely available energysource which is considered to be carbon neutral as well as renew-able. The energy stored within the biomass can be extractedthrough different techniques which include biochemical (such asanaerobic digestion producing biogas) and thermo-chemical meth-ods. Thermo-chemical ways of extracting the energy from the bio-mass includes pyrolysis, gasification, co-combustion and

high temperatures (temperatures greater than 300 �C) in an inertenvironment which results in the production of fuel in all threephases oil (l), char (s) and gases (g) which are combustible. Gasifi-cation and combustion of raw biomass are direct combustion tech-niques which involve the use of air. Gasification is carried out infuel rich conditions (Equivalence ratio (ER) > 1; partial oxidation)and combustion is carried out in fuel lean condition (ER < 1, excessair) [1]. Ash fouling [1–3], poor grindability [4], and lower heat va-lue of raw biomass makes it difficult to burn the raw biomass di-rectly in coal fired boilers. The overall conversion efficiency ofraw biomass is limited by its high moisture content which can alsohave significant effects on grindability [5] and energy yield [6]. Ithas been noted that biomasses with high moisture contents arebetter suited for biological conversion processes [7] (anaerobicdigestion) unless they undergo some pre-treatment process that

380 D. Eseltine et al. / Fuel 113 (2013) 379–388

reduces the moisture content, such as drying. Torrefaction is one ofthe techniques which has been studied recently to improve thegrindability and energy density or heat value (HV) of biomass. It in-volves the partial pyrolysis of biomass in order to improve its fuelproperties [8–14]. Torrefaction is similar to pyrolysis, with the heattreatment being carried out in a lower temperature range (200–300 �C) and the resulting product biomass having a higher energydensity.

Torrefaction mainly removes moisture in the biomass and caus-ing it to become hydrophobic [14–16] due to destruction of hydro-xyl (OH) groups in the biomass [17] allowing for easier storage andtransportation. Furthermore, the destruction of OH groups resultsin a biomass with a larger heating value due to a decrease in oxy-gen content. The hydrophobic nature of torrefied biomass also re-sults in less energy being required to process the biomass. Theimproved properties of torrefied biomass suggest that it is viableas a renewable energy source that is better suited for use in co-fir-ing and gasification applications [17–19] when compared to virgin(untreated) biomass. Further studies [20] have shown that gasifica-tion of torrefied woody biomass produces more usable CO and H2

than raw biomass. A simplistic rationale for explaining the im-proved HV of biomass is to use empirical Boie equation for HV ofbiomass [21].

HHVf ðkJ=kgÞ ¼ 35;160 � YC þ 116;225 � YH � 11;090

� YO þ 6;280 � YN þ 10;465 � YS ð1Þ

where YC, YH, YO, YN and YS denote the mass fractions of carbon,hydrogen, oxygen, nitrogen and sulfur respectively. Increasing thecarbon percentage and lowering the oxygen percentage results inincreases in the HV of fuel. Increases in heating value with increas-ing pyrolysis temperature are due to the almost complete removalof oxygenated compounds [9,17]. Studies conducted by Bridgemanet al. [9] on the torrefaction of canary grass, wheat straw, and wil-low observed that the moisture, oxygen, and hydrogen percentagesdecreased while carbon percentage increased with increases in thetorrefaction temperature. The decrease in oxygen percentage cou-pled with the increase in carbon percentage thereby resulted inhigher heating values. The mass and energy yields of willow (awoody biomass) were also noticeably higher when compared tothat of canary grass and wheat straw. Similar research by Sadakaet al. in 2009 [22] on the torrefaction of wheat straw, rice straw,and cotton gin waste observed that the heating value increasedwith increasing residence temperature and residence time. How-ever, increasing torrefaction temperatures resulted in higheramounts of mass loss during the torrefaction process. Furthermore,it has been shown that torrefaction temperature has a much great-er effect on mass loss and energy yield than residence time byPimchuai et al. and others [23,14]. The lower limit is set by the dev-olatilization temperature of hemicellulose which is below 250 �C[9]. Hemicellulose and cellulose have been found to devolatilizebefore lignin in biomass. The devolatilization of hemicellulosebeginning around 200 �C with complete devolatilization occurringby 350 �C. Also cellulose exhibits devolatilization beginning at250 �C [24–29]. Although the upper temperature limit of torrefac-tion does not allow for the full devolatilization of cellulose andhemicellulose, depolymerization has been shown to occur. Fur-thermore, recent studies by Chen and Kuo [26] have concluded thatthere are no apparent interactions between the three basic constit-uents of biomass (hemicellulose, cellulose, and lignin).

Woody biomass is a particular type of biomass that could ben-efit greatly from a pre-treatment process such as torrefaction giventhat it is initially very difficult to grind due to its fibrous nature.Furthermore, woody biomass has invaded grasslands and range-lands causing a reduction in herbaceous plant life [30,31]. Ansleyet al. [32] have pointed out that using woody biomass as a source

of bioenergy could not only provide an alternative energy sourcebut also reduce the impact of woody plants on grasslands andrangelands. Mesquite and Juniper wood are two specific types ofwoody biomass found abundantly in Texas that could be used forsuch a purpose.

The effect of residence time (tres) and residence temperature (T)on the properties of various biomass using nitrogen as the torrefac-tion medium has been widely documented. However, despite largeamounts of research on the torrefaction of biomass it appears thatthere is scant literature on investigating the effect of non-inertgases upon the weight loss characteristics and thermal degradationof torrefied biomass. Given that Carbon-dioxide is one of the mainemissions from fossil fuel combustion, recycling of CO2 to ther-mally upgrade biomass could improve the overall efficiency ofeither process. This coupled with the fact that biomass is a sourceof energy that is carbon neutral make Carbon-dioxide an intriguingalternative to Nitrogen as a torrefaction environment. The presentstudy focuses on investigating the effect CO2 has upon the thermaldegradation of two woody biomass during torrefaction and pyroly-sis using a thermogravimetric analyzer. By studying a wide rangeof torrefaction temperatures the effect of temperature can be seenupon the weight loss characteristics of both biomass. Furthermore,an optimal torrefaction temperature can be determined from massloss data and estimated heating values. For the present study theoptimal torrefaction temperature is utilized to further study the ef-fect of Carbon-dioxide as a torrefaction medium upon the fuelproperties, grindability, and emissions profile of torrefied biomass.

2. Experimental procedure

2.1. Thermal degradation analysis via TGA

The torrefaction characteristics of the Mesquite and Juniperwood were investigated using a TGA (TA Instruments Q600). Dur-ing experiments the flow rate of the purge gas was controlled andregulated via the internal mass flow controller within the TGA withthe purge gas having a flow rate of 200 ml/min. Samples wereloaded into TGA at ambient temperatures; roughly 10 mg of sam-ple were used for each test. Once the samples had been loadedthe furnace was purged. Samples were then heated at a constantheat rate of 20 �C/min from an ambient temperature to the torre-faction temperature, where the biomass was torrefied for 1 h atthat fixed temperature. This residence time was chosen based uponthe mass loss during torrefaction of biomass torrefied between30 min and 3 h being extremely similar [8]. After torrefaction ofthe biomass was completed the furnace was heated from the torre-faction temperature to 1000 �C at a rate of 20 �C/min. The entireprocess served to torrefy the biomass and then pyrolyze the torr-efied biomass. Samples were torrefied between 200 and 300 �C in20 �C increments to investigate a wide range of torrefactiontemperatures.

2.2. Batch torrefaction

A laboratory oven was used to torrefy larger amounts of bio-mass for fuel property analysis. For batch torrefaction in the labo-ratory oven the initial particle size of 2–4 mm was used. Sampleswere initially placed in 5 ml ceramic crucibles and the crucibleswere put into the oven at room temperature. The purge gas flowwas started and maintained at 6 LPM by an external flow meter.Once the oven had been purged the samples were heated fromambient temperatures to the torrefaction temperature at a rateof 20 �C/min. The samples were then held at the torrefaction tem-perature for 1 h, removed from the oven, and placed in a dessicatorto cool.

D. Eseltine et al. / Fuel 113 (2013) 379–388 381

Biomass samples torrefied in the laboratory oven were used todetermine the effect of varying the torrefaction medium upon thefuel properties and grindability of the biomass. Fuel property anal-ysis was done by Hazen Laboratories in Golden, CO. Grindabilityanalysis was done via sieving and obtaining a size distribution be-fore and after grinding the torrefied biomass. Grinding of the torr-efied biomass was done using a vibrating ball mill grinder (SwecoModel DM1).

2.3. Exhaust gas composition analysis via FTIR

Samples torrefied in the laboratory oven were further analyzedusing a TGA/FTIR setup to investigate the effect of torrefaction andthe different torrefaction mediums upon the emissions profile. Thetorrefied samples were ground and sieved in the same manner asthe raw biomass samples outlined previously, with the 589–840 lm fraction being removed and used. The thermogravimetricanalyzer was connected to an MKS FTIR (Multigas Model 2030)via a heated gas transfer line maintained at 200 �C. The heatedgas transfer line was used to avoid condensing of any gases duringthe transfer process which could adversely affect the emissionsmeasurements. The working temperature of the FTIR during emis-sions analysis was set at 191 �C and the FTIR was set to measureconcentrations of the following gases: CO2, CO, CH4, and Ethane.Due to flow requirements of the FTIR the flow rate of the purgegas was set at 500 ml/min. Samples were heated from ambienttemperatures to 800 �C at a heating rate of 20 �C/min in an nitro-gen environment with emissions data being collected every 20 s.

3. Results and discussion

3.1. Biomass properties and preparation

The raw Mesquite (Prosopis Glandulosa) and Juniper (JuniperusPinchotti) wood samples were supplied by the Texas Agrilife Re-search and Extension Center in Vernon, Texas. The initial woodsamples were received in chip form with sizes ranging from 2 to4 mm. The proximate, ultimate (elemental), and heat value analy-ses of the raw biomass are listed in Table 1. As shown in the table,the moisture content of Mesquite is much higher (15.5%) than thatof Juniper (5.85%). However, the ash content of both biomass arerelatively low (1.67% for Mesquite and 1.91% for Juniper).

For TGA experimentation the raw wood required further pro-cessing. The wood was further ground and sieved, with the 589–840 lm fraction being selected for TGA tests. The ground samplesthat were used for TGA experiments were stored in airtight plasticbags and kept at the laboratory temperature.

From the proximate analysis one can determine the approxi-mate value of average heat value of volatile matter (VM).

Table 1Properties of Raw Mesquite and Juniper.

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) 16,700 19,000HHVDAF (kJ/kg) 20,100 20,600Empirical formulaa CH1.37N0.01O0.58S0.0003 CH1.38N0.01O0.56S0.0001

a Formula carbon normalized.

HHVDAFkJkg

� �ffi HHVVM � VM0 þHVFC � FC0 ð2Þ

where HHVDAF is the dry ash free higher heating value, VM0 isthe initial VM in the DAF raw biomass (RB), HHVVM is the higherheating value of the VM, FC0 is the initial amount of FC in theDAF RB and HVFC is the heating value of the FC. Assuming onlythe volatile matter within the biomass is released during torrefac-tion, the heating value of the dry ash free torrefied biomass can beapproximately estimated using the following equation:

HHVDAF;TBkJkg

� �¼ ðHHVVM � VMTBÞ þ ðHVFC � FC0Þ

VMTB þ FC0ð3Þ

where HHVDAF,TB is the higher heating value of the dry ash free torr-efied biomass and VMTB is the volatile matter in the torrefied bio-mass (TB). Eq. (3) is based on the presumption that HHVVM doesnot change in TB during torrefaction process. Fig. 1. shows the var-iation of heating value and the heat content of the torrefied biomasswhen the VM in the biomass is released. It is seen from Fig. 1 thatHHVTB can range from HHVDAF with 0% torrefaction to HHVFC with100% of VM released from biomass.

3.2. Thermogravimetric analysis

3.2.1. Mass loss profileWeight trace results from thermogravimetric analysis of the

Mesquite and Juniper woods are shown in Figs. 2 and 3 respec-tively. The weight trace curve can be divided into several differentzones independent of the biomass being tested. Initial weight lossat temperatures below 200� is due to the moisture loss during dry-ing of the biomass. At temperatures between 200 and 500 �C thebiomass undergoes significant thermal decomposition resultingin high amounts of mass loss as the biomass is pyrolyzed. Thepyrolysis zone of lignocellulosic biomass can be divided into twozones as done by Mansaray and Ghaly [34]. The first portion,termed the ‘‘active’’ pyrolysis zone (roughly 200–375 �C), attrib-uted to the liberation of low weight volatiles and breakdown ofhemicellulose and cellulose. The second portion, the ‘‘passive’’pyrolysis zone (roughly 375–500 �C), attributed to further evolu-tion of volatiles and the conversion of lignin which is a slow pro-cess. The change from active to passive pyrolysis is indicated bya clear change in slope between 375� and 400�. For simplicity thethermograms for each torrefaction temperature were overlaid onthe same plot. For torrefaction, each thermogram followed theweight loss curve for pyrolysis until reaching the torrefaction tem-perature. Since the sample is held at this constant temperature, avertical line results which shows the weight loss at that torrefac-tion temperature.

As expected, increases in torrefaction temperature resulted inhigher amounts of mass loss during the torrefaction process which

0

10

20

30

40

50

60

70

80

90

100

1

1.2

1.4

1.6

1.8

2

2.2

2.4

0 0.2 0.4 0.6 0.8 1

Heat T

B/H

eat RB

(%)H

V o

f T

B/ H

V o

f R

B

VM fraction released

Mesquite

Juniper

Fig. 1. Estimated variation of heating value of the TB and the heat content in the TBwith respect to the amount of volatile matter released during torrefaction.

20

30

40

50

60

70

80

90

100

0 100 200 300 400 500 600 700 800

Wei

ght

%

Temperature (°°C)

Pyrolysis200°C220°C240°C260°C280°C300°C

Fig. 2. TGA thermograms of Mesquite wood pyrolysis and torrefaction using N2 as the purge gas.

20

30

40

50

60

70

80

90

100

0 100 200 300 400 500 600 700 800

Wei

ght

%

Temperature (°°C)

Pyrolysis200°C220°C240°C260°C280°C300°C

Fig. 3. TGA thermograms of Juniper wood pyrolysis and torrefaction using N2 as the purge gas.

382 D. Eseltine et al. / Fuel 113 (2013) 379–388

is in agreement with previous studies [8–10]. Furthermore, as thetorrefaction temperature was increased, larger amount of weightloss was observed, increasing from roughly 14% at 200 �C to 60%at 300 �C. Comparison of weight traces of biomass torrefied inNitrogen and Carbon-dioxide (Figs. 4 and 5) showed increasedamounts of weight loss when using CO2 as a torrefaction medium.Initially the weight trace curves showed less than 1% difference atlower torrefaction temperatures (200–220 �C). However, as thetorrefaction temperature was increased, biomass torrefied in aCO2 environment exhibited increased amounts of weight loss asshown in Table 2. The difference between the two curves becominglarger as the torrefaction temperature increased. Although bothMesquite and Juniper are woody biomass, Mesquite exhibited aslightly larger amount of weight (Fig. 4) loss than Juniper (Fig. 5)(19% Mesquite vs. 15% for Juniper at 220 �C in Nitrogen). The differ-ence in weight loss is due to variations in lignocellulose composi-tion of the two woods. Previous studies [31] have shown thatdeciduous woods (Mesquite) lose more mass than coniferouswoods (Juniper) during torrefaction due to difference in hemicellu-lose composition.

3.2.2. Derivative thermogravimetric analysisSeveral studies [10,25,26] have used DTG distributions as a way

of identifying the three major constituents of lignocellulosic bio-mass (hemicellulose, cellulose, and lignin) due to differences inchemical structure and the temperature range over which eachof the components thermally degrades (which has been well doc-umented [25]). As previously stated, hemicellulose is the mostreactive of the three constituents, thermally degrading over thetemperature range 220–315 �C. Cellulose, which is the most ther-mally sensitive, degrades between 315 and 400 �C. Finally lignin,which is the most chemically complex, gradually breaks down overa wide temperature range between 160 and 900 �C [25].

Analyzing the DTG distributions of Mesquite and Juniper (Figs. 6and 7 respectively, where Peak M: moisture loss peak; Peak C: cel-lulose loss peak; Peak L: Lignin loss peak) subjected to a pure pyro-lysis process shows a clear difference in the hemicellulose peaksbetween the two biomass, with Mesquite having a more pro-nounced peak (seen at roughly 280 �C). The variance in thermaldegradation behavior of the two woods is due mainly to the factthat Mesquite is a hardwood versus Juniper being a softwood.

20

30

40

50

60

70

80

90

100

0 100 200 300 400 500 600 700 800

Wei

ght

%

Temperature (°C)

Pyrolysis -Nitrogen

Pyrolysis -Carbon-dioxide

200°C -Nitrogen

200°C -Carbon-dioxide

220°C -Nitrogen

220°C -Carbon-dioxide

240°C -Nitrogen

240°C -Carbon-dioxide

260°C -Nitrogen

260°C -Carbon-dioxide

280°C -Nitrogen

280°C -Carbon-dioxide

300°C -Nitrogen

300°C -Carbon-dioxide

Fig. 4. Comparison of thermograms from pyrolysis and torrefaction of Mesquite wood using N2 and CO2 as the purge gases.

20

30

40

50

60

70

80

90

100

0 100 200 300 400 500 600 700 800

Wei

ght

%

Temperature (°°C)

Pyrolysis -Nitrogen

Pyrolysis -Carbon-dioxide

200°C -Nitrogen

200°C -Carbon-dioxide

220°C -Nitrogen

220°C -Carbon-dioxide

240°C -Nitrogen

240°C -Carbon-dioxide

260°C -Nitrogen

260°C -Carbon-dioxide

280°C -Nitrogen

280°C -Carbon-dioxide

300°C -Nitrogen

300°C -Carbon-dioxide

Fig. 5. Comparison of thermograms from pyrolysis and torrefaction of Juniper wood using N2 and CO2 as the purge gases.

Table 2Percent mass loss due to torrefaction.

Temp. (�C) Mesquite Juniper

Nitrogen Carbon-dioxide Nitrogen Carbon-dioxide

200 14.7 13.6 11.6 12.4220 18.7 18.9 15.0 15.5240 25.4 27.7 20.9 24.6260 32.9 37.4 28.7 30.4280 45.7 50.5 41.3 45.2300 58.9 62.7 57.9 62.2

D. Eseltine et al. / Fuel 113 (2013) 379–388 383

The percentage of holocellulose (the combination of hemicellu-loses and cellulose) and lignin in hardwoods and softwoods is in-nately different with hardwoods containing a larger percentageof holocellulose (71.7 ± 5.7% for hardwoods and 64.5 ± 4.6% forsoftwoods) [35]. Furthermore, thermogravimetric studies con-ducted by Sebio-Punal et al. [36] on the thermal degradation of

holocellulose and lignin obtained from hard and softwoods hasshown that hardwood lignins exhibit a faster degradation overthe temperature range from 290 to 300 �C. This rapid degradationof lignin results in the appearance of a more pronounced ‘‘shoul-der’’ near 300 �C for Mesquite.

By overlaying the DTG distributions of both biomass subjectedto a torrefaction process (N2 environment) the effect of increasingtorrefaction temperature is readily seen (Figs. 6 and 7). Increases intorrefaction temperature resulted in significant impacts on thehemicellulose and cellulose within the biomass. Due to hemicellu-lose being the most reactive of the lignocellulosic components,even the lowest torrefaction temperature (200 �C) resulted in deg-radation of hemicellulose. However, no degradation of cellulosewas seen at torrefaction temperatures below 240 �C. Further in-creases in the torrefaction temperature resulted in the degradationof hemicellulose and cellulose. Furthermore, the most severe torre-faction temperature (300 �C) appears to result in the almost com-plete degradation of hemicellulose and cellulose. The lignin peak

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0 100 200 300 400 500 600 700 800

DT

G (

wt%

/°C

)

Temperature (°C)

Pyrolysis

200°C

220°C

240°C

260°C

280°C

300°C

Peak C

Peak M

Peak L

Fig. 6. DTG distribution of Mesquite wood pyrolysis and torrefaction using N2 as the purge gas.

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0 100 200 300 400 500 600 700 800

DT

G (

wt%

/°C

)

Temperature (°C)

Pyrolysis

200°C

220°C

240°C

260°C

280°C

300°C

Peak L

Peak C

Peak M

Fig. 7. DTG distribution of Juniper wood pyrolysis and torrefaction using N2 as the purge gas.

384 D. Eseltine et al. / Fuel 113 (2013) 379–388

seen on the DTG distributions near 500 �C were not affected byincreasing torrefaction temperatures.

The DTG curves for Mesquite pyrolysis and torrefaction usingCO2 as the purge gas is shown below in Fig. 8. A comparison ofthe DTG plots when using Nitrogen and Carbon-Dioxide shows dif-ferences in thermal breakdown of Mesquite depending on thepurge gas used with the most noticeable difference being a de-crease in the peak rate (% per �C) of cellulose breakdown (350 �C)when CO2 was used as the purge gas. This difference is due to lossof some hydrocarbons during torrefaction since torrefied biomasscontained less VM when pyrolyzed in CO2. There is also a notice-able shift in the peak lignin decomposition to a slightly higher tem-perature when using Carbon-dioxide as the purge gas which can bedue to higher specific heat of carbon dioxide when compared tonitrogen [37]. DTG distributions of Juniper wood using Nitrogenand Carbon-dioxide as the purge gas showed the same trends asobserved with Mesquite wood.

3.2.3. Energy retention estimationsBased upon the mass loss data obtained from the TGA an opti-

mal torrefaction temperature was determined. Although the heat-ing value of the biomass increases after torrefaction there is aninherent mass loss, therefore, the amount of energy retained with-in the torrefied biomass will be less than the raw biomass. Theamount of energy retained within the torrefied biomass can be cal-culated using the following equation:

Energy retainedð%Þ ¼ MTB �HHVTB

MRB �HHVRB� 100 ð4Þ

where MTB and MRB are the mass of torrefied biomass and mass ofraw biomass respectively. Assuming the mass lost during torrefac-tion is solely from the degradation of volatile matter the dry ashfree HHV of the torrefied biomass can be estimated from Eqs. (2)and (3). From the mass loss data obtained the dry ash free HHVand energy retention of the torrefied biomass was estimated (Ta-

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0 100 200 300 400 500 600 700 800

DT

G (

%w

t/°C

)

Temperature (°C)

Pyrolysis

200°C

220°C

240°C

260°C

280°C

300°C

Peak C

Peak L

Peak M

Fig. 8. DTG distribution of Mesquite wood pyrolysis and torrefaction using CO2 as the purge gas.

Table 3Estimated energy retention values for the entire range of torrefaction temperaturesinvestigated.

Biomass Temperature (�C)

200 220 240 260 280 300

Mesquite – N2 93.0 90.7 86.8 80.9 65.9 56.2Mesquite – CO2 93.6 90.7 84.7 77.6 65.7 51.6

Juniper – N2 94.2 92.5 88.8 82.8 73.3 56.4Juniper – CO2 93.7 91.4 86.1 81.7 68.7 57.5

Table 5Mesquite and Juniper ash composition.

Mesquite Juniper

SiO2 8.49 22.4AL2O3 2.14 4.93TiO2 0.11 0.29Fe2O3 2.81 4.14CaO 41.9 38.4MgO 2.38 2.21Na2O 0.98 0.63K2O 14.3 8.30P2O5 3.10 2.17SO3 0.88 1.01Cl 0.24 0.06CO2 15.6 11.8S 0.03 0.01

D. Eseltine et al. / Fuel 113 (2013) 379–388 385

ble 3). Torrefaction at temperatures exceeding 240 �C resulted insignificant increases in mass loss (nearly 15%), which is mainlydue to degradation of cellulose beyond 240 �C. Therefore, the en-ergy retention drops below 85% for both Mesquite and Juniperregardless of the torrefaction environment. Based upon the esti-mated energy retention values 240 �C was chosen as the best torre-faction temperature and was used for as the torrefactiontemperature for laboratory oven studies.

Table 4Properties of Mesquite and Juniper torrefied in N2 and CO2 at 240 �C.

N2 medium

Mesquite Juniper

Moisture 4.84 5.69Volatile matter 69.5 74.6Fixed carbon 23.3 18.6Ash 2.39 1.08Carbon 53.4 53.6Oxygen 33.2 34.1Hydrogen 5.33 5.42Nitrogen 0.81 0.19Sulfur 0.05 0.01HHVTB (kJ/kg) 19,800 20,100HHVRB (kJ/kg) 16,700 19,000HHVTB,DAF (kJ/kg) 21,400 21,600HHVDAF,RB (kJ/kg) 20,100 20,600Empirical formula (TB)a CH1.20N0.013O0.47S0.0004 CH1.22N0.003OEmpirical formula (RB)a CH1.37N0.01O0.58S0.0003 CH1.38N0.01O0

a Formula carbon normalized, TB: torrefied biomass, RB: raw biomass.

3.3. Torrefied biomass fuel properties

3.3.1. Proximate, ultimate (elemental) and heat value analysesHaving determined an optimal torrefaction temperature

(240 �C) from TGA mass loss data and energy retention estima-tions, large batch torrefaction was done using a laboratory oven.

CO2 medium

Mesquite Juniper

4.30 6.2571.1 73.222.2 19.42.34 1.1353.3 53.233.9 33.85.37 5.410.77 0.170.04 0.0219,500 19,80016,700 19,00020,900 21,40020,100 20,600

0.48S0.0001 CH1.21N0.014O0.48S0.0002 CH1.22N0.003O0.48S0.0001

.56S0.0001 CH1.37N0.01O0.58S0.0003 CH1.38N0.01O0.56S0.0001

Table 6Size distribution analysis for unground and ground raw and torrefied biomass.

Mesquite – un-groundSample (lm) >2380 2380–2000 2000–840 840–590 590–149 149–74 74–53 <53 %< 840Raw 78.1 8.7 10.8 0.4 0.3 0.1 0.2 0.3 1.2Torrefied – N2 71.1 9.6 16.1 0.9 0.6 0.3 0.4 0.4 2.6Torrefied – CO2 77.4 7.9 12.3 0.6 0.7 0.3 0.2 0.2 2.1Juniper – un-groundSample (lm) >2380 2380–2000 2000–840 840–590 590–149 149–74 74–53 <53 %< 840Raw 44.9 17.1 34.3 2.1 0.8 0.1 0.1 0.3 3.4Torrefied – N2 36.2 14.5 44.5 2.8 1.4 0.1 0.1 0.1 4.5Torrefied – CO2 36.2 17.0 39.5 3.8 1.8 0.3 0.2 0.3 6.3Mesquite – groundSample (lm) >2380 2380–2000 2000–840 840–590 590–149 149–74 74–53 <53 %< 840Raw 60.0 10.1 20.2 1.8 2.4 1.3 1.5 2.4 9.4Torrefied – N2 42.1 8.4 25.4 3.7 9.0 3.1 1.8 4.6 22.1Torrefied – CO2 31.2 7.3 18.7 22.2 6.0 5.5 2.7 3.9 40.2Juniper – groundSample (lm) >2380 2380–2000 2000–840 840–590 590–149 149–74 74–53 <53 % < 840Raw 27.9 11.0 47.0 5.8 4.7 0.9 1.4 0.8 13.6Torrefied – N2 16.0 7.7 47.6 8.6 9.0 2.6 1.7 6.6 28.5Torrefied – CO2 14.9 7.2 44.0 8.0 9.2 4.8 4.4 6.4 32.8

(a)

(b)Fig. 9. Exhaust gas composition profiles. (a) Mesquite torrefied in Nitrogen and (b) Mesquite torrefied in carbon-dioxide.

386 D. Eseltine et al. / Fuel 113 (2013) 379–388

D. Eseltine et al. / Fuel 113 (2013) 379–388 387

Torrefied biomass samples were sent for proximate, ultimate (ele-mental), and heat value analyses. The results of fuel property anal-ysis are presented in Table 4. The mild torrefaction processresulted in reduced moisture content of the torrefied biomass, withthe exception of Juniper torrefied in a CO2 environment. Increasesin ash content was observed for torrefied Mesquite independent oftorrefaction environment while the opposite was observed for thetorrefied Juniper samples.

In terms of elemental analysis, minimal variations were ob-served between the two torrefaction environments. However, itshould be noted that torrefaction in a Carbon-dioxide environmentproduced the smallest amount of elemental Carbon and lowestHHV for both Mesquite and Juniper. Increased mass loss on usingcarbon dioxide as the torrefaction medium can be attributed to anumber of factors. One of the factors is the behavior of ash constit-uents (Table 5) when using different torrefaction mediums. Thecomposition of the ash suggests the possibility of inorganic ele-ments within the biomass acting as a catalyst when using CO2 asthe torrefaction medium, resulting in the higher weight loss ob-served on the TGA weight traces (Figs. 2 and 3). It is clear fromthe DTG trace (Fig. 8) that there is a reaction with CO2 occurring,evident in the DTG curves trending upwards after 700 �C. This ismost likely caused by the Boudouard reaction (CO2 + C ? 2CO)where the CO2 gas reacts with the fixed carbon present in the char.However there is no clear evidence showing this is the reason forincreased mass loss during torrefaction (200–300 �C), and furtherstudies should be done to clearly understand the mechanism be-hind the increased mass loss.

3.3.2. GrindabilityIn order to further investigate the effect of the torrefaction envi-

ronment upon the biomass, grindability studies were done andverified using size distribution analysis. The results of the grinda-bility studies are presented in Table 6. Prior to grinding, compari-son of the raw and torrefied biomass samples indicates that theprocess of torrefaction slightly reduced the particle size of the bio-mass. However, after grinding, the torrefied biomass had signifi-cantly smaller particle sizes compared to the raw biomass.Initially less than 7% of the biomass (raw or torrefied) passed an840 lm mesh. Although the raw biomass saw slight reduction inparticle sizes after grinding (9.4% through 840 lm for Juniper and13.6% for Mesquite), grinding of the torrefied biomass resulted insignificantly smaller particle sizes (>20% through 840 lm).

Comparison of the biomass torrefied in a Nitrogen environmentto the biomass torrefied in a Carbon-dioxide environment showsthat torrefaction in using CO2 results in further improvements ingrindability. Significantly smaller particle sizes were observed forbiomass torrefied in CO2 with 40.2% of Mesquite torrefied in CO2

passing an 840 lm mesh compared to 22.1% for torrefaction inNitrogen.

Increases in grindability are directly related to the devolatiliza-tion of the fibrious microstructure of the woody biomass. Results ofseveral other studies [10,38,39] has shown that torrefaction causesthe surface of the biomass to become smoother due to the loss ofhemicellulose. Furthermore, microporus structures develop onthe surface of the torrefied biomass [38,39].

3.4. Exhaust gas composition

The small sample size used in the TGA (9–10 mg) combinedwith the high purge gas flow resulted in a highly diluted exhaustprofile. Of all the gases measured using the FTIR, the emissions pro-file was dominated by high amounts of CO2 and CO with smalleramounts of Methane (CH4) being measured. Trace amounts of Eth-ane (C2H6) were also seen, however these were less than 1 ppm. Acomparison of the emissions profile from the pryolysis of two bio-

mass samples torrefied in N2 and CO2 respectively are shown inFig. 9.

The release profile of CO and CO2 follow the same behavior ob-served in the DTG charts, with the peak release occurring at theprimary peak of weight loss. Similar behavior was observed by Bia-gini et al. [40] using virgin biomass (wood pellets and pine wood)with IR profile temperatures in agreement with the results of thisstudy. Furthermore, CH4 exhibits a trend similar to those authorswork, producing two peaks between 400 �C and 600 �C. Slight dif-ferences between the two torrefaction mediums can be seen in theemissions profile with a slight decrease in the peak of CO2 release(�200 ppm) occurring around 350 �C. This decrease coincides witha roughly 5% increase in mass loss shown on the DTG curve. Thepeak CO release seen at the same 350 �C temperature showed onlyslight differences, 30 ppm, between the two samples. Beyond thispeak the CO and CH4 curves exhibited no significant changes be-tween the torrefaction mediums.

4. Conclusion

The results obtained from the torrefaction of two woody bio-mass in an inert (Nitrogen) and non-inert (Carbon-dioxide) envi-ronment exhibited noticeable differences between the twotorrefaction mediums. Torrefaction in a Carbon-dioxide environ-ment resulted in larger amounts of mass loss (�2–4% at 240 �C)when compared to torrefaction in a Nitrogen environment, the dif-ference between the two becoming larger as the torrefaction tem-perature increases. The increased mass loss in a Carbon-dioxideenvironment may be due to a number of factors which includesthe effect of ash constituents and effect of different torrefactionmediums on structural constituents of biomass. Further studiesshould be done to understand the higher mass loss when using car-bon dioxide as a medium for torrefaction. From thermogravimetricanalysis, estimated energy retention values for torrefied biomasswere calculated and it was determined that a mild torrefactionprocess (240 �C) is best. Ultimate (elemental) analysis of the torr-efied wood samples showed little difference between torrefactionmediums in terms of the resulting torrefied biomass. However,the HHV of biomass torrefied in a Carbon-dioxide environmentwas slightly smaller than the biomass torrefied in a Nitrogen envi-ronment, independent of the biomass. Grindability tests revealedthat biomass torrefied in a Carbon-dioxide environment producedsufficiently smaller particle sizes after grinding compared to bio-mass torrefied in a Nitrogen environment. Given that only slightreductions in HHV and energy retention were observed it con-cluded that Carbon-dioxide is a viable torrefaction medium.

References

[1] Lawrence B, Annamalai K, Sweeten JM, Heflin K. Cofiring coal and dairybiomass in a 29 KW furnace. Appl Energy 2009;86:2359–72.

[2] Baxter L. Biomass-coal co-combustion: opportunity for affordable renewableenergy. Fuel 2005;84:1295–302.

[3] Dermirbas A. Potential applications of renewable energy sources, biomasscombustion problems in boiler power systems and combustion relatedenvironmental issues. Prog Energy Combust Sci 2005;31(2):171–92.

[4] Bergman PCA, Boersma AR, Zwart RWR, Kiel JHA. Torrefaction for biomass co-firing in existing coal-fired power stations. BIOCOAL concept version, ECNreport; 2005.

[5] Mani S, Tabil LG, Sokhansanj S. Grinding performance and physical propertiesof wheat and barley straws, corn stover and switchgrass. Biomass Bioenergy2004;27:339–52.

[6] Medic D, Darr M, Shah A, Potter B, Zimmerman J. Effects of torrefaction processparameters on biomass feedstock upgrading. Fuel 2012;91:147–54.

[7] McKendry P. Energy production from biomass (part 1): overview of biomass.Bioresour Technol 2002;83:37–46.

[8] Arias B, Pevida C, Fermoso J, Plaza MG, Rubiera F, Pis JJ. Influence of torrefactionon the grindability and reactivity of woody biomass. Fuel Process Technol2008;89:169–75.

388 D. Eseltine et al. / Fuel 113 (2013) 379–388

[9] Bridgeman TG, Jones JM, Shield I, Williams PT. Torrefaction of reed canarygrass, wheat straw, and willow to enhance solid fuel qualities and combustionproperties. Fuel 2008;87:844–56.

[10] Chen WH, Hsu HC, Lu KM, Lee WJ, Lin TC. Thermal pretreatment of wood(Lauan) block by torrefaction and its influence on the properties of thebiomass. Energy 2011;36:3012–21.

[11] Phanphanich M, Sudhagar M. Impact of Torrefaction on the grindability andfuel characteristics of forest biomass. Bioresour Technol 2011;102:1246–53.

[12] Repellin V, Govin A, Rolland M, Guyonnet R. Energy requirement for finegrinding of torrefied wood. Biomass Bioenergy 2010;34(7):923–30.

[13] Deng J, Wang GJ, Kuang JH, Zhang YL, Luo YH. Pretreatment of agriculturalresidues for co-gasification via torrefaction. J Anal Appl Pyrolysis2009;86:331–7.

[14] Bridgeman TG, Jones JM, Williams A, Waldron DJ. An investigation of thegrindability of two torrefied energy crops. Fuel 2010;89:3911–8.

[15] Hakkou M, Pétrissans M, Gérardin P, Zoulalian A. Investigations of the reasonsfor fungal durability of heat-treated beech wood. Polym Degrad Stab2006;91:393–7.

[16] Tumuluru JS, Sokhansanj S, Wright CT, Boardman R. Biomass torrefactionprocess review and moving bed torrefaction system model development. INL/EXT-10-19569. Idaho National Laboratory and Oak Ridge National Laboratory;2010.

[17] Xiao R, Chen X, Wang F, Yu G. Pyrolysis pretreatment of biomass for entrained-flow gasification. Appl Energy 2010;87:149–55.

[18] Prins MJ, Ptasinski KP, Janssen FJJG. More efficient biomass gasification viatorrefaction. Energy 2006;31:3458–70.

[19] Prins MJ, Kryzysztof JP, Janssen FJJG. Torrefaction of wood Part 2. Analysis ofproducts. J Anal Appl Pyrolysis 2006;77:35–40.

[20] Cohourt C, Salvador S, Commandre J-M. Impact of torrefaction on syngasproduction from wood. Fuel 2009;88:2286–90.

[21] Annamalai K, Puri I. Combustion science and engineering. Boca Raton,Florida: CRC Press: Taylor and Francis; 2006.

[22] Sadaka S, Negi S. Improvements of biomass physical and thermochemicalcharacteristics via torrefaction process. Environ Prog Sustain Energy2009;28:427–34.

[23] Pimchuai A, Dutta A, Basu P. Torrefaction of agriculture residue to enhancecombustible properties. Energy Fuels 2010;24:4638–45.

[24] Raveendran K, Ganesh A, Khilar KC. Pyrolysis characteristics of biomass andbiomass components. Fuel 1996;75:987–98.

[25] Yang H, Yan R, Chen H, Lee DH, Zheng C. Characteristics of hemicellulose,cellulose and lignin pyrolysis. Fuel 2007;86:1781–8.

[26] Chen WH, Kuo PC. Torrefaction and co-torrefaction characterization ofhemicellulose, cellulose and lignin as well as torrefaction of some basicconstituents in biomass. Energy 2011;36:803–11.

[27] Chen WH, Kuo PC. A study on torrefaction of various biomass materials and itsimpact on the lignocellulosic structure simulated by a thermogravimetry.Energy 2010;35(6):2580–6.

[28] Prins MJ, Ptasinski KJ, Janssen FJJG. Torrefaction of wood: Part 1. Weight losskinetics. J Anal Appl Pyrolysis 2006;77:28–34.

[29] Wu Y, Zhao Z, Li H, He F. Low temperature pyrolysis characteristics of majorcomponents of biomass. J Fuel Chem Technol 2009;37:427–32.

[30] Ansley RJ, Wu BX, Kramp BA. Observation: long-term increases in Mesquitecanopy cover in a north Texas savanna. J Range Manage 2001;54:171–6.

[31] Asner GP, Archer S, Huges RF, Ansley RJ, Wessman CA. Net changes in regionalwoody vegetation cover and carbon storage in Texas drylands, 1937–1999.Global Change Biol 2003;9:316–35.

[32] Ansley RJ, Mirik M, Castellano MJ. Structural biomass partitioning in regrowthand undisturbed Mesquite (Prosopis glandulosa): implications for bioenergyuse. Global Change Biol Bioenergy 2010;2:26–36.

[34] Mansaray KG, Ghaly AE. Thermal degradation of rice husks in nitrogenatmosphere. Bioresour Technol 1998;65:13–20.

[35] Rowell RM. Handbook of wood chemistry and wood composites. Boca Raton,Florida: CRC Press: Taylor and Francis; 2005.

[36] Sebio-Punal T, Naya S, Lopez-Beceiro J, Tarrio-Saavedra J, Artiaga R.Thermogravimetric analysis of wood, holocellulose, and lignin from fivewood species. J Therm Anal Calorim 2012;109:1163–7.

[37] Thanapal SS, Annamalai K, Sweeten J, Gordillo G. Fixed bed gasification of dairybiomass with enriched air mixture. Appl Energy 2012;97:525–31.

[38] Chen WH, Cheng WY, Lu KM, Huang YP. An evaluation on improvement ofpulverized biomass property for solid fuel through torrefaction. Appl Energy2011;88:3636–44.

[39] Biagini E, Fantozzi C, Tognotti L. Characterization of devolatilization ofsecondary fuels in different conditions. Combust Sci Technol 2004;176:685–703.

[40] Biagini E, Barontini F, Tognotti L. Devolatilization of biomass fuels and biomasscomponents studied by TG/FTIR technique. Ind Eng Chem Res 2006;45:4486–93.