You are on page 1of 9

Bioresource Technology 196 (2015) 136–144

Contents lists available at ScienceDirect

Bioresource Technology
journal homepage:

Isoconversional kinetic study of the thermal decomposition of sugarcane

straw for thermal conversion processes
Yesid Javier Rueda-Ordóñez, Katia Tannous ⇑
School of Chemical Engineering, University of Campinas, 500 Albert Einstein Avenue, Cidade Universitária ‘‘Zeferino Vaz’’, 13083-852 Campinas, SP, Brazil

h i g h l i g h t s

 First study about the kinetics of thermal decomposition of sugarcane straw.

 Similar kinetic, thermal and physicochemical characteristics with sugarcane bagasse.
 Two dimensional diffusion reaction model represented the thermal decomposition reaction.
 Mathematical model proposed for conversion rate was in good agreement with experimental data.
 Current industrial technology for sugarcane bagasse is suitable for straw processing.

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

Article history: The aim of this work was investigate the kinetics of the thermal decomposition reaction of sugarcane
Received 19 June 2015 straw. The thermal decomposition experiments were conducted at four heating rates (1.25, 2.5, 5 and
Received in revised form 17 July 2015 10 °C/min) in a thermogravimetric analyzer using nitrogen as inert atmosphere. The kinetic analysis
Accepted 18 July 2015
was carried out applying the isoconversional method of Friedman, and the activation energies obtained
Available online 23 July 2015
varied from 154.1 kJ/mol to 177.8 kJ/mol. The reaction model was determined through master plots,
corresponding to a two-dimensional diffusion. The pre-exponential factor of 1.82 * 109 s1 was
determined by linearization of the conversion rate equation as a function of the inverse of absolute
Activation energy
temperature, concerning to activation energy of 149.7 kJ/mol, which are in the order of magnitude for
Model free biomass thermal decomposition reported in literature. Finally, the theoretical and experimental conver-
Pyrolysis sion data showed a very good agreement, indicating that these results could be used for future process
Thermogravimetry modeling involving sugarcane straw.
Ó 2015 Elsevier Ltd. All rights reserved.

1. Introduction Annually, Brazil produces and processes more than 300 million
metric tons of sugarcane, which correspond to a quarter of the
Biomass presents interesting characteristics as raw material for 1300 million tons planted worldwide. The sugarcane straw is com-
several energy generation processes, as combustion, gasification, posed by dry leaves, green leaves and tops of the sugarcane plant.
liquefaction, and pyrolysis. The biomass or its derivative products Each ton of stalks corresponds to 140 kg of dry residues. Part of the
(e.g., ethanol, methane, bio-oil, and char) when used as fuels sugarcane is mechanically harvested and the straw is left on the
release CO2, which was absorbed from the atmosphere during ground, and other great part is burned before the harvesting of
the photosynthesis process, and consequently, does not have an sugarcane in order to reduce the cost of this operation. The São
increase in the global emissions (Abbasi and Abbasi, 2010). The Paulo and Minas Gerais states in Brazil, which are the major pro-
vegetal biomass is composed mainly by hemicellulose, cellulose ducers of sugarcane (around two thirds of the production) are com-
and lignin, as well as low moisture, extractives, and mineral matter mitted for the total removal of burning process until 2018 (Leal
contents (White et al., 2011). According to Sluiter et al. (2008), et al., 2013). This process has been controlled by the Brazilian gov-
extractives can be inorganic material, non-structural sugars, ernment by law No. 11241 (September 19th, 2002) to promote the
chlorophyll, waxes and minor components. gradual reduction of the sugarcane burning until 2031. Due to the
high generation of this agricultural residue in the ethanol produc-
tion, it turns necessary to find one or more energetic applications
⇑ Corresponding author. Tel.: +55 19 3521 3927; fax: +55 19 3521 3910. for the sugarcane straw. The combustion, gasification or pyrolysis
E-mail address: (K. Tannous). processes could be some possibilities.
0960-8524/Ó 2015 Elsevier Ltd. All rights reserved.
Y.J. Rueda-Ordóñez, K. Tannous / Bioresource Technology 196 (2015) 136–144 137

In the last few years, there has been a growing interest in the Topçu, 2014; Baroni et al., 2015; Rueda-Ordóñez et al., 2015).
sugarcane straw, and some studies have been conducted on pyrol- Yao et al. (2008) determined the activation energy for ten different
ysis application. Moraes et al. (2012) analyzed the chemical com- species of fibers, considering a general trend, remains around
position of volatile compounds produced during the sugarcane 168 kJ/mol, with the exception of hemp, jute, and rice straw, which
straw pyrolysis, identifying more than 120 compounds, mainly presented values, 180.9 kJ/mol, 183.1 kJ/mol, and 197.6 kJ/mol,
oxygenated ones (acids, aldehydes, alcohols, phenols, ethers and respectively. According to the authors, the higher is activation
ketones). Mesa-Pérez et al. (2013) investigated the pyrolysis in a energy, the higher is the biomass stability comparative to the
bubbling fluidized bed reactor, obtaining maximum of bio-oil and others fibers.
bio-char yields of 35.5 wt% and 48.2 wt%, respectively, at a temper- Ounas et al. (2011) studied the thermal decomposition of sugar-
ature of 470 °C. An important contribution to the field was done by cane bagasse and olive residue from Morocco. The authors divided
these authors, exhibiting the importance and viability of the sugar- the kinetic results into two conversion sections from 0.00 to 0.50
cane straw as an energetic source. and 0.50 to 0.80 concerning the hemicellulose and cellulose reac-
The biomass pyrolysis process is composed by three stages: tions, respectively. The activation energies for hemicellulose were
drying, devolatilization, and carbonization (Basu, 2010). The drying between 168–180 kJ/mol and 153–162 kJ/mol, and for cellulose
or dehydration stage occurs in the temperature range between were between 168–180 kJ/mol and 153–162 kJ/mol, for sugarcane
room temperature and 150 °C. The devolatilization stage is charac- bagasse and olive residue, respectively. The woody residues
terized by the hemicellulose and cellulose contents volatile release, showed in Table 1, pine (Poletto et al., 2010; Anca-Couce et al.,
in which the higher mass loss is registered in the temperature 2014), eucalyptus (Poletto et al., 2010), and poplar (Slopiecka
interval between 150 °C and 400 °C. The hemicelluloses are mainly et al., 2012), presented activation energy around of 155 kJ/mol,
composed by xylose (hardwoods), mannose (softwoods) and other lower than the reported for fibers. The woody shell residues such
polysaccharides such as glucose, galactose, and arabinose, and the as hazelnut husk (Ceylan and Topçu, 2014), Tucumã endocarp
devolatilization temperature is between 200 °C and 300 °C (Di (Baroni et al., 2015), and Brazil nut woody shell (Rueda-Ordóñez
Blasi, 2008). The cellulose, according to Vassilev et al. (2012), is et al., 2015) presented the lowest activation energy compared with
composed by a long network of hydrogen bonds, which links long the fibrous and woody biomasses, around 140 kJ/mol. Thus, sum-
chains one to another, providing thermal stability and resistance. marizing the data presented in Table 1, the activation energies of
The cellulose devolatilization temperature is between 250 °C and biomass thermal decomposition in inert atmosphere using isocon-
350 °C (Di Blasi, 2008). Above 400 °C, begins the carbonization versional methods remained between 150 kJ/mol and 200 kJ/mol.
stage and the char formation, which is the main component of The aim of this work was the determination of the kinetics of
the final solid residue in the pyrolysis process. According to the thermal decomposition reaction and the physicochemical char-
Pasangulapati et al. (2012), the lignin is the higher precursor of acteristics of sugarcane straw in order to investigate the viability of
char, about 50% yield, whereas the contribution of cellulose and this biomass as a source for thermal conversion processes.
hemicellulose are very low, 1% for cellulose and 7% for
hemicellulose. 2. Methods
Table 1 presents the activation energy obtained by different iso-
conversional methods for biomasses (Friedman, Flynn–Wall– 2.1. Material used
Ozawa, Coats–Redfern modified, and Vyazovkin methods), classi-
fied in fiber (Yao et al., 2008; Ounas et al., 2011; Mishra and In this work, sugarcane straw (Saccharum officinarum Linnaeus)
Bhaskar, 2014), wood (Poletto et al., 2010; Slopiecka et al., 2012; was used as raw material composed mainly by dry leaves. The bio-
Anca-Couce et al., 2014), and woody shell residues (Ceylan and mass was ground (batches of 10 g in 20 s) in a hammer mill (Tigre

Table 1
Activation energy of several biomasses obtained by isoconversional methods.

References Biomass Activation energy (kJ/mol)

Yao et al. (2008) Bagasse 168.5 169.5 168.7 –
Bamboo 164.1 162.8 161.9 –
Cotton stalk 165.3 169.9 169.1 –
Hemp 180.9 177.9 177.7 –
Jute 183.1 184.2 184.3 –
Kenaf 169.8 170.3 169.6 –
Rice husk 168.2 167.4 166.5 –
Rice straw 197.6 195.9 196.9 –
Wood-maple 156.0 155.8 154.3 –
Wood-pine 161.5 161.8 160.4 –
Ounas et al. (2011) Sugarcane bagasse – 199.5 210.0 –
Olive residue – 178.3 188.5 –
Mishra and Bhaskar (2014) Rice straw 195.0 179.4 178.4 179.6
Poletto et al. (2010) Pine wood – 158.0 – –
Eucalyptus wood – 155.5 – –
Slopiecka et al. (2012) Poplar wood – 158.6 157.3 –
Anca-Couce et al. (2014) Beech wood – – 183.7 –
Pine wood – – 143.6 –
Ceylan and Topçu (2014) Hazelnut husk – 131.1 127.8 –
Baroni et al. (2015) Tucumã endocarp 160.5 147.3 144.6 145.0
Rueda-Ordóñez et al. (2015) Brazil nut woody shell 144.5 145.7 142.7 –
Friedman method.
Flynn–Wall–Ozawa method.
Coats–Redfern modified method.
Vyazovkin method.
138 Y.J. Rueda-Ordóñez, K. Tannous / Bioresource Technology 196 (2015) 136–144

S.A., CV2, Brazil) coupled to an induction motor of 3800 rpm, CV5 Afterward, the DTG curves were determined with the thermo-
(General Electric, 25.4062.405, Brazil) and separated by sieving gravimetric analyzer software TA-50WS (Shimadzu Corporation)
using granutest sieves with mesh of 28 and 35 from Tyler series and normalized (dW/dt) using Eq. (3).
and then obtained the mean diameter between sieves of 510 lm.
After that, a rotary cone sample divider (model laborette 27, dW dm 1
¼ ð3Þ
Fritsch, Germany) was utilized to obtain a representative sample dt dt m0
to apply in the thermogravimetric analysis.
For the kinetic analysis, it was selected the main volatile release
range from the normalized mass data. Then, using Eq. (4), were
2.1.1. Sample characterization
converted in conversion, a, in which W0, W, and Wf are the normal-
The sample characterization was divided into four analyses:
ized mass at the beginning, at a time t, and at the end of the main
proximate, ultimate, heating value, and chemical composition.
volatile release, respectively. The experimental conversion rate
These analyses were performed using the following standards:
(da/dt)exp is given by Eq. (5).
moisture content (ASTM E871–82), Ash content (ASTM E1755–
01), volatile content (ASTM E872–82) and fixed carbon by W0  W
difference on a dry basis; elemental analysis (CHN-O) using an aexp ¼ ð4Þ
W0  Wf
elemental analyzer (Perkin Elmer, Series II 2400, USA), and the
oxygen content was determined by difference on a dry and ash free  
da dW 1
basis. The higher heating value (HHV) was determined using an ¼ ð5Þ
oxygen bomb calorimeter (IKA, C200, Germany), applying the dt exp dt ðW 0  W f Þ
standard ASTM D240–09. The lower heating value (LHV) in MJ/kg The theoretical conversion rate (da/dt)theoretical is given by Eq.
was determined using Eq. (1), in which C, H, O, S, and W are the per- (7), in which E is the activation energy, A is the pre-exponential
centage of carbon, hydrogen, oxygen, sulphur and moisture in the factor, R is the universal gas constant (8.314 J/mol K), and T is the
biomass, respectively. The sulphur content in Eq. (1) was not taken absolute temperature.
into account because the percentage was lower than 0.1% and the   
equipment resolution was not sufficient for an accurate E
k ¼ A exp ð6Þ
measurement. RT
LHV ¼ 4:187½81C þ 300H  26ðO  SÞ  6ðW þ 9HÞ ð1Þ     
da E
¼ k½f ðaÞ ¼ A½f ðaÞ exp ð7Þ
The chemical compositions determination of hemicellulose and dt theoretical RT
cellulose were carried out using the ANKOM A200 filter bag tech-
nique based on Van Soest’s method. The lignin content in the sam-
2.4. Isoconversional method
ple was determined by the Klason’s method based on acid
hydrolysis (Sluiter et al., 2012).
The kinetic analysis of the thermal decomposition reaction was
done applying the isoconversional method of Friedman (1964) for
2.2. Experimental set-up
four heating rates between 1.25 °C/min and 10 °C/min.
The method developed by Friedman (1964) solves the Eq. (7)
The thermal decomposition experiments were performed with
using natural logarithm, as presented in Eq. (8). In this method,
a thermogravimetric analyzer Shimadzu Corporation (TGA-50,
the conversion was evaluated from 0.05 to 0.95 with a step of
Japan). This apparatus measures the sample mass with a resolution
0.05, resulting in 19 conversion levels. For each conversion level
of 0.001 mg. The samples were installed in an open sample
evaluated, it was determined a value of (da/dt)exp obtained by Eq.
platinum pan (6 mm internal diameter and 2.5 mm depth) with
(5) and its temperature associated. After that, it was plotted for
an initial mass of 3.0 ± 0.2 mg. The temperature was controlled
each heating rate the value of ln(da/dt) as a function of the 1/T,
from room temperature (25 ± 3 °C) up to 900 °C at seven heating
resulting in a straight line for each conversion, then the slopes
rates, 1.25 °C/min, 2.5 °C/min, 5 °C/min, 10 °C/min, 15 °C/min,
are multiplied by the universal gas constant to obtain the activa-
20 °C/min, and 40 °C/min. The International Confederation of
tion energies.
Thermal Analysis and Calorimetry (ICTAC) since 2011 (Vyazovkin
et al., 2011) recommend the use of at least three temperature da E
ln ¼ ln A þ ln f ðaÞ  ð8Þ
ramps lower than 20 °C/min to perform kinetic analysis from ther- dt RT
mogravimetry. The highest heating rate chose in this work was to
evaluate the effect of transport phenomena in the sample that
could affect the thermogravimetric curves, and consequently the 2.4.1. Reaction model determination
kinetic parameters. The reaction model was determined following the recommen-
The acquisition rate of sample mass was 12 sample/min corre- dations provided by the International Confederation of Thermal
sponding to a step of 5 s. Inert gas of high purity (N2 = 99.996%, Analysis and Calorimetry (ICTAC). According to Vyazovkin et al.
4.6 FID, White Martins, Campinas, Brazil) with flow rate of (2011) the use of master plots is the correct way to determine
50 mL/min was used. the reaction model showed in Table 2. However, since its concep-
tion raised from the same definition of the isoconversional meth-
2.3. Mathematical approach ods, the master plots emerged from the definition of the
integrated solution of the conversion rate equation (Eq. (7)). For
Firstly, the TG data was converted to normalized mass, W, using that, the Eq. (7) is modified in order to obtain the derivative of con-
the Eq. (2), where m and m0 are the mass at a time t, and the initial version with respect to the temperature, and therefore, introducing
mass, respectively. the heating rate effect, b = dT/dt, is obtained the Eq. (9). From the
integration of the reaction model, f(a), it was obtained the function
m g(a) (Table 2).
W¼ ð2Þ
Y.J. Rueda-Ordóñez, K. Tannous / Bioresource Technology 196 (2015) 136–144 139

Table 2
Integral and differential form of several reaction models for solid-state kinetics (adapted from Vyazovkin et al., 2011).

Reaction model Code f(a)a g(a)b

Power law P1 4a a1/4
Power law P2 3a2/3 a1/3
Power law P3 2a1/2 a1/2
Power law P4 2/3a1/2 a3/2
Phase-boundary controlled reaction (contracting area) R2 2(1a)1/2 [1(1a)1/2]
Phase-boundary controlled reaction (contracting volume) R3 3(1a)2/3 [1(1a)1/3]
Avrami Erofe’ev (m = 2) A2 2(1a)[ln(1a)]1/2 [ln(1a)]1/2
Avrami Erofe’ev (m = 3) A3 3(1a)[ln(1a)]2/3 [ln(1a)]1/3
Avrami Erofe’ev (m = 4) A4 4(1a)[ln(1a)]3/4 [ln(1a)]1/4
First order F1 (1a) ln(1a)
nth order Fn (1a)n [1(1a)(1n)](1n)
One-dimensional diffusion D1 1/2a a2
Two-dimensional diffusion Valensi–Carter equation D2 [ln(1a)]1 (1a)ln(1a)+a
Three-dimensional diffusion Jander equation D3 (3/2)(1a)2/3[1(1a)1/3]1 [1(1a)1/3]2
Three-dimensional diffusion Ginstling–Brounshtein D4 (3/2)[(1a)1/31]1 (12a/3)(1a)2/3
Differential form.
Integrated form.

    Z af
da A E da gðaÞ
Ea A
pðxÞ pðxÞ
¼ f ðaÞexp  ¼ bR
¼ Ea A ¼ ð11Þ
dT b RT ao f ðaÞ gð0:5Þ bR pðx0:5 Þ pðx0:5 Þ
  Z Tf  
¼ exp  dT ð9Þ In order to obtain the reaction model which better describes
b 0 RT
the decomposition reaction, the theoretical and experimental
The last term of the right side of Eq. (9) is an integral without master plots are plotted as a function of conversion. Then, the
analytical solution. In order to overcome this problem, according model which better represents the experimental data at the dif-
to Flynn (1997) the integral is replaced by (E/R)[p(x)] with ferent heating rates (1.25 °C/min, 2.5 °C/min, 5 °C/min, and
x = E/(RT), as detailed in Eq. (10). 10 °C/min) is considered the suitable for the reaction
  Z Tf   description.
A E The solution of p(x) function in Eq. (11) was performed applying
gðaÞ ¼ exp dT
b 0 RT the 8th degree rational approximation developed by
 Z 1 
AE  expðxÞ expðxÞ AE Pérez-Maqueda and Criado (2000), and presented in Eq. (12). The
¼  ð Þdx ¼ pðxÞ ð10Þ
bR x x x2 bR activation energy used to solve the x = E/RT in Eq. (12) was the
average of the set of activation energies provided by the isoconver-
In Eq. (10), the unknown pre-exponential factor is directly sional method of Friedman (1964).
affected by the reaction model, g(a), and therefore, it is necessary
to eliminate its influence. Thus, it was selected a conversion refer-  
expðxÞ x7 þ70x6 þ1886x5 þ24920x4 þ
ence point (a = 0.5), and performed the ratio g(a)/g(0.5) as pre-
pðxÞ ¼ x x8 þ72x7 þ2024x6 þ28560x5 þ216720x4 þ
sented in Eq. (11), which represents the theoretical master plots. þ170136x3 þ577584x2 þ844560xþ357120
þ880320x3 þ1794240x2 þ1572480xþ403200
Also, the experimental master plots could be obtained from the
ratio p(x)/p(0.5).

Table 3
Proximate and ultimate analyses and heating values of sugarcane straw.

This work Mesa-Pérez et al. (2013) Leal et al. (2013) Hoi and Martincigh (2013)
Proximate analysis wt.%
Moisture 8.42 ± 0.30 10.4 13.5 67.7 82.3 – – –
Asha 3.85 ± 0.21 16.4 2.7 3.7 4.3 – – –
Volatilesa 86.64 ± 0.53 74.0 84.5 80.6 79.3 – – –
Fixed carbona 9.51 ± 0.53 13.0 11.6 15.7 16.4 – – –
Ultimate analysis wt.%a,b
C 42.94 ± 0.25 43.2 46.2 45.7 43.9 45.2 39.1 41.5
H 6.26 ± 0.16 6.7 6.2 6.2 6.1 5.8 3.0 5.0
N 0.31 ± 0.05 0.3 0.5 1.0 0.8 – – –
S – 0.2 0.1 0.4 0.1 – – –
O 46.65 ± 0.18 33.2 43.0 42.8 44.0 47.4 45.1 46.7
Heating value MJ/kg
HHV 18.61 18.0 17.4 17.4 16.4 18.3 – –
LHV 15.72 17.0 – – – – – –

DL – dry leaves.
GL – green leaves.
T – tops.
wt.% on dry basis.
wt% on ash free basis.
140 Y.J. Rueda-Ordóñez, K. Tannous / Bioresource Technology 196 (2015) 136–144

2.4.2. Pre-exponential factor determination large range of moisture from dry leaves to green leaves with an
The pre-exponential factor determination was carried out as a increasing of 55 wt.%, and from dry leaves to tops around 70 wt.%.
forward step, with the knowledge of the reaction model, f(a). Regarding to the ash content, our result (3.85 wt.%) was similar
Therewith, rearranging Eq. (7) and plotting the values of to the average of 3.56 ± 0.81 wt.% obtained from Leal et al. (2013),
ln[(da/dt)/f(a)] as a function of 1/T, it was obtained a straight line, whilst Mesa-Pérez et al. (2013) showed a significantly higher value
in which the intercept represents lnA, and with the slope is deter- (16.4 wt.%). Concerning to the volatile contents, our result
mined the final activation energy. This activation energy should (86.6 wt.%) was comparative to the average of 81.47 ± 2.71 wt.%
not differ more than 10% from the values provided by the obtained from Leal et al. (2013), nevertheless Mesa-Pérez et al.
Friedman’s isoconversional method (Janković, 2008). (2013) showed a higher difference (12.6 wt.%) probably due the
" # high mineral contents in their samples. About fixed carbon, it
ðddtaÞ E was observed that the results of Mesa-Pérez et al. (2013) and
ln ¼ lnA  ð13Þ
f ðaÞ RT Leal et al. (2013) are close (13 wt.% and 14.57 wt.%, respectively),
however our results presented lower data. From the ultimate anal-
ysis, it was observed that the values of C, H, N, S, and O are quite
2.5. Evaluation of data modeling similar, with difference lower than 3 wt.%, with exception of Hoi
and Martincigh (2013) for green leaves. Comparing the higher
The quality of fit (<5%) was evaluated through the average devi- heating values the variation (max. 6.8%) could be considered neg-
ation percentage (AVP) proposed by Órfão et al. (1999), presented ligible with exception of Leal et al. (2013) for tops of plant (13.5%).
in the Eq. (14), where N is the number of TG experimental data, Finally, the properties obtained in this work does not differ sig-
and SS, the least square method (Eq. (15)). In Eq. (15) the experi- nificantly with the values obtained by other authors who studied
mental conversion (aexp) was determined by Eq. (4), and the theo- the sugarcane straw, indicating that exists a uniformity of the
retical conversion (atheoretical) was performed with the fourth order material under study, providing reliability in the raw material.
Runge–Kutta method, solving Eq. (7), implemented in an Excel In Table 4 is presented the chemical composition of sugarcane
spreadsheet (software MS Office Excel 2007 version straw compared with Saad et al. (2008), Luz et al. (2010), and
12.0.6683.5002). The number of TG experimental data considered Costa et al. (2013) works, and also are presented the results of
were 1687, 688, 392, and 168 for the heating rates of researches (Luz et al., 2010; Rocha et al., 2011; Aboyade et al.,
1.25 °C/min, 2.5 °C/min, 5 °C/min, and 10 °C/min, respectively.
AVP ¼ 100 ð14Þ

N h
X i2
SS ¼ ðaÞi;exp  ðaÞi;theoretical ð15Þ

3. Results and discussion

3.1. Biomass characterization

Table 3 presents the proximate and ultimate analyses of sugar-

cane straw obtained in this work and compared with the results
obtained by Mesa-Pérez et al. (2013), Leal et al. (2013), and Hoi
and Martincigh (2013). Considering the composition of the sugar-
cane straw as dry leaves, green leaves and tops, Mesa-Pérez et al.
(2013) did not present clearly the part of plant used, while Leal
et al. (2013) and Hoi and Martincigh (2013) analyzed separately
these three parts.
In general, the proximate and ultimate analyses, as well as heat-
ing value are in the same magnitude order excluding the moisture
content. From the proximate analysis, it was observed that the
moisture was around 8% lower than Mesa-Pérez et al. (2013) and
Leal et al. (2013) results. However, the latest authors present a

Table 4
Chemical composition of sugarcane straw and sugarcane bagasse.

References Hemicellulose Cellulose Lignin Extractives

Sugarcane straw wt.%
This work 33.28 39.81 21.63 5.28
Saad et al. (2008) 26.2 39.4 21.5 –
Luz et al. (2010) 27.4 33.3 26.1 10.6
Costa et al. (2013) 27.1 33.5 25.8 –
Sugarcane bagasse wt.%
Luz et al. (2010) 28.6 43.8 23.5 2.8
Rocha et al. (2011) 27.0 45.5 21.1 4.6
Aboyade et al. (2013) 23.8 44.2 22.4 9.7
Fig. 1. Normalized mass (a) and normalized DTG (b) as a function of temperature.
Y.J. Rueda-Ordóñez, K. Tannous / Bioresource Technology 196 (2015) 136–144 141

2013) about sugarcane bagasse in order to correlate both one peak. This behavior indicates that the reaction time is too short
biomasses. to obtain a correct representation of the complexity of thermal
Our results of hemicellulose content was slightly higher decomposition of sugarcane straw. Also, because in TG analysis is
(6 wt.%) than the average reported in literature (26.9 ± 0.6 wt.%). assumed that the inter-particle and intra-particle transport phe-
Nevertheless, concerning the literature results of sugarcane straw nomena effects are avoided.
and bagasse no quite difference (26.46 wt.%) was observed. The According to the literature review presented in the Section 1,
cellulose content in our sample was 39.81 wt% similar obtained the hemicellulose component reacts between 200 °C and 300 °C
by Saad et al. (2008), but 6 wt.% higher than (33 wt.%) other sug- corresponding to the first peak in the DTG curve (Fig. 1b). The cel-
arcane straw from literature. lulose component reacts between 250 °C and 350 °C corresponding
Comparing this component between sugarcane bagasse and to the main peak, and the lignin is the responsible for the final
straw, the average value from literature of bagasse was remaining solid. However, the latest component reacts in a broad
44.5 ± 0.9 wt.%, which was 5 wt.% and 11 wt.% higher than our temperature range, from 200 °C to 600 °C, which means that when
sample and other sugarcane straw, respectively. Concerning to hemicellulose and cellulose react, the lignin also is decomposing,
the lignin content in our sample presented 5 wt.% of variation in but not governing the overall reaction.
comparison with literature and between biomasses, i.e., for sugar-
cane straw was from 21.5 wt.% to 26.1 wt.%, and for bagasse from
3.2.3. Carbonization
21.1 wt.% to 23.5 wt.%.
The final stage in a pyrolysis process is the carbonization occur-
Finally, both biomasses on a dry basis presented close charac-
ring at temperatures above 450 °C. As observed in Fig. 1a, the mass
teristics. This is very important to the reactor designs, where could
released in this stage was 12.46%, 10.72%, 8.29%, 5.67%, 15.73%,
be used the same technology for sugarcane straw or a mixture of
16.66% and 16.70% for the heating rates from 1.25 °C/min to
both biomasses.
40 °C/min, respectively. The final residues, formed by the mineral
content and the char fraction, were 8.61%, 11.52%, 16.30%,
3.2. Thermal decomposition analysis
17.30%, 7.65%, 8.09%, and 4.31% for the same heating rates, respec-
tively. In Fig. 1b the DTG curves above 450 °C presented a small
Fig. 1(a) and (b) present the normalized mass (W) and the nor-
shoulder decreasing its amplitude with the temperature increas-
malized DTG (dW/dt) as a function of temperature, according to Eq.
ing, resembling a tail shape. This behavior is due to the lignin ther-
(2) and (3), respectively. The thermal decomposition of the sugar-
mal decomposition, which is characterized by low reaction rate
cane straw was divided into three stages such as dehydration,
and broad temperature range. However, the curves at
devolatilization and carbonization.
1.25 °C/min, 20 °C/min, and 40 °C/min presented different curve
patterns with higher conversion rates and higher volatile release.
3.2.1. Dehydration
The behavior at 20 °C/min and 40 °C/min confirms that were
The first stage in a thermal conversion process corresponds to
affected by transport phenomena.
the sample dehydration, and takes place in the temperature range
In conclusion, it was observed that the higher the heating rate,
between the room temperature and 150 °C. The moisture content
the higher the final remaining mass. In other words, decreasing the
in the sugarcane straw shown in Fig. 1a were 9.32%, 8.13%,
heating rate, the reaction time is increased, allowing the occur-
7.17%, 7.31%, 8.35%, 7.37% and 9.35% for the heating rates of
rence of more reactions, breaking the polymeric linkages of bio-
1.25 °C/min, 2.5 °C/min, 5 °C/min, 10 °C/min, 15 °C/min,
mass compounds and increasing the volatile production, and
20 °C/min, and 40 °C/min, respectively. In the DTG curve (Fig. 1b)
therefore, greater biomass decomposition.
this stage presented a peak in the temperature range of 25 °C–
100 °C for all the heating rates. Above 150 °C up to 200 °C, was
not registered significant mass release (<0.5%). 3.3. Isoconversional method results

3.2.2. Volatile release The isoconversional method of Friedman was used to estimate
The second stage or intermediate step in a pyrolysis process is the activation energy. Due to the heat and mass transfer effects,
the volatile release, which takes place in the temperature range
between 200 °C–450 °C. At this stage, the biomass is completely
dried, and its main components (hemicellulose, cellulose and lig-
nin) react with the temperature increasing. Different gaseous com-
pounds are released, and a fraction of them are condensable, which
are the precursors of the liquid fraction (Di Blasi, 2008). The vola-
tilized mass of sugarcane straw as shown in Fig. 1a was 69.71%,
69.63%, 68.29%, 69.73%, 69.05%, 67.87%, and 69.70% for the heating
rates from 1.25 °C/min to 40 °C/min, respectively. Also, in the same
figure is observed the overlapping of the curves at 15 °C/min,
20 °C/min and 40 °C/min, indicating that the use of heating rates
higher than 10 °C/min involves mass and heat transfer effects
which cannot be measurable. In our case, the last three heating
rates are not recommended to determine the kinetic parameters.
The normalized DTG curve, shown in Fig. 1b, was characterized
by the formation of two well defined peaks at 298.7 °C and
347.3 °C, 311.6 °C and 360.3 °C, 324.7 °C and 370.8 °C, and
340.0 °C and 385.0 °C, for 1.25 °C/min, 2.5 °C/min, 5 °C/min, and
10 °C/min, respectively. However, for the heating rates higher than
10 °C/min the curve pattern (two peaks) suffered modifications, at
15 °C/min and 20 °C/min the curve pattern turns into a shoulder Fig. 2. Derivative of conversion with respect to temperature as a function of
followed by a peak, and at 40 °C/min the curve is almost only conversion.
142 Y.J. Rueda-Ordóñez, K. Tannous / Bioresource Technology 196 (2015) 136–144

in the kinetic analysis were used only the four lower heating rates Above 0.2 up to 0.5 an increasing from 153.9 kJ/mol to
(1.25 °C/min, 2.5 °C/min, 5 °C/min, and 10 °C/min) in order to 174.0 kJ/mol was observed. This increasing of the activation energy
obtain reliable kinetic parameters. The biomass main decomposi- was due to the polymeric differences between hemicellulose and
tion occurred from the normalized mass, W, of 90% to 25% for all cellulose. According to Di Blasi (2008) the hemicellulose is the least
the heating rates, related to the temperature range between thermally stable of the biomass components, and therefore, its
200 °C and 450 °C (Fig. 1a), and thus, it was the range chosen for activation energy in an inert thermal decomposition process is
the kinetic analysis. lower than the cellulose one. The cellulose is formed by links of b
Fig. 2 presents the derivative of conversion with respect to tem- (1–4)-glucose (Liu et al., 2011), very stable links, thus requiring
perature (da/dT) as a function of conversion (a). Two peaks of the more energy for the breaking and decomposition. The maximum
DTG curves were observed to the volatile release corresponding to activation energy was achieved when the cellulose decomposition
hemicellulose, cellulose, and lignin decomposition reaction, as was reached, corresponding to 0.5 of conversion and no higher
described in the volatile release section. Besides that, the conversion energy was necessary to continue the biomass decomposition.
range from 0.00 to 0.45 was related to the temperature range from Therefore, after this conversion (0.5 to 0.8), the activation energy
200 °C to 320 °C for all heating rates, concerning mainly to hemi- decreased from 177. 5 kJ/mol to 160.0 kJ/mol.
cellulose and lignin decomposition reactions. At the last conversion Above the conversion of 0.8, the biomass was almost decom-
level (a = 0.45), the DTG curve was marked by a minimum, becom- posed and a carbonaceous solid was formed constituting the
ing clear the division between hemicellulose and cellulose main majority of the residue, and the activation energy became stable
reactions. in 155.8 ± 1.86 kJ/mol, concerning mainly to the lignin thermal
It is important to remark in this conversion range, that the small decomposition reaction, as commented in the Section 1.
effect of heating rate on the variation of the conversion with tem-
perature (da/dT) leads similar conversion curve patterns for the
3.4. Reaction model and pre-exponential factor determination
four heating rates. Also, verified by the similar shape in the nor-
malized mass curve (Fig, 1a) between 85% and 55%.
3.4.1. Reaction model determination
For the conversion range between 0.45 and 1.00, the main
The reaction model determination is a fundamental step in
decomposition reaction corresponds to cellulose and lignin. In
kinetic studies, since it is a theoretical function that helps to
Fig. 2, the curves of da/dT as a function of conversion were similar
describe and improve the understanding of the reaction. In bio-
for the heating rates of 1.25 °C/min and 2.5 °C/min, but for
mass studies applying isoconversional methods, it was observed
5 °C/min and 10 °C/min the derivative decreased with the increas-
that it is commonly assumed a first or nth order reaction model
ing of the heating rate. In this case, the normalized mass curves
(Damartzis et al., 2011; Slopiecka et al., 2012; Ceylan and Topçu,
(Fig. 1a) changed their shapes from 50% to 30%.
2014; Baroni et al., 2015). Nevertheless, according to Vyazovkin
The activation energies obtained varied from 154.1 kJ/mol to
et al. (2011) is not recommended the assumption of a reaction
177.8 kJ/mol, representing a variation of 15%, and average value
model without a previous verification through master plots, since
of 163.4 kJ/mol. This variation is explained by the fact that the
other reaction model could be more suitable to describe the
Arrhenius rate law was developed for the study of the kinetics of
gases which are well described by a one-step global reaction,
In Fig. 3 is presented the theoretical and experimental master
whereas in the solid-state kinetics the activation energy presents
plots as a function of conversion. The theoretical curves (lines)
a strong dependency with the conversion evolution (Janković,
are for a few integrated reaction models, g(a), (F1, D1, D2, D3
2008). In the conversion range from 0.1 to 0.2, the average of acti-
and D4) which could represent the experimental behavior at the
vation energy remained in 154.66 ± 0.44 kJ/mol, corresponding to
different heating rates evaluated. It is observed in Fig. 3 that the
the temperature interval between 200 °C and 300 °C, related to
experimental master plots (symbols) at 1.25 °C/min (x),
the biomass components depolimerization, which is the first step
2.5 °C/min (e), 5 °C/min (D) and 10 °C/min (o) are consistent with
in the thermal decomposition with low energy requirements.
the theoretical master plot of the reaction model D2
(Valensi-Carter equation) corresponding to two-dimensional

Fig. 3. Theoretical and experimental master plots for the thermal decomposition of Fig. 4. Linearization of the conversion rate equation as a function of the inverse of
sugarcane straw at the heating rates of (x)1.25 °C/min, (e)2.5 °C/min, (D)5 °C/min, absolute temperature for our experimental data of the thermal decomposition of
(o)10 °C/min as a function of conversion. sugarcane straw.
Y.J. Rueda-Ordóñez, K. Tannous / Bioresource Technology 196 (2015) 136–144 143

diffusion. According to Khawam and Flanagan (2006), the

two-dimensional diffusion is derived of the assumption of a cylin-
drical shape solid with a reaction occurring radially from the exter-
nal boundary to the center, increasing the reaction zone with time.
In this present study, the two-dimensional diffusion (D2) provides
an approximation to the real behavior of the biomass decomposi-
tion reaction, and taken into account that the sample of sugarcane
straw used in this work is a long fiber with cylindrical shape, the
mathematical form is correct. From a kinetic point of view, it is
widely discussed the approximation of the thermal decomposition
reaction as a one-step reaction, since it is governed by multiple
complex reactions affected by the interactions between biomass

3.4.2. Pre-exponential factor determination

The pre-exponential factor was determined as a forward step,
with knowledge of a suitable reaction model, f(a), which provided a
good representation of the experimental data. The pre-exponential
factor was determined by linearization of ln[(da/dt)/f(a)] as a func-
Fig. 5. Conversion curves as a function of temperature. Modeled data as solid lines,
tion of the 1/T, as mentioned in the subsection Pre-exponential and experimental data at the heating rates of (j)1.25 °C/min, (h)2.5 °C/min,
factor determination of the Experimental section, and presented in (d)5 °C/min, (s)10 °C/min.
Fig. 4. The linear equation obtained is given by Eq. (16), in which
the slope and intercept were 18.007 and 21.322.
Table 5
Kinetic parameters of sugarcane straw.
ln½ðda=dtÞ=f ðaÞ ¼ 18:007ð1=TÞ þ 21:322 R2 ¼ 0:9658 ð16Þ
One-step global reaction
The scattering of the experimental data in Fig. 4 was reflected in
f(a) E (kJ/mol) lnA (ln s1)
the determination coefficient, R2, of 0.9658. This slight scatter
effect is due to the formation of two peaks (Fig. 2a), associated to D2 = [ln(1a)] 149.710 21.322
b (°C/min) 1.25 2.5 5 10
multiple and complex reactions in the biomass thermal decompo-
AVP (%) 2.3 2.2 2.3 3.1
sition, as stated by Vyazovkin et al. (2011).
The intercept represents lnA, corresponding to a
pre-exponential factor A = 1.82 * 109 s1. model to describe the biomass thermal decomposition, concluding
The activation energy (E = 149.710 kJ/mol) was obtained that using this reaction model the results are overestimated.
through the multiplication of slope by universal gas constant
(8.314 J/mol K). This value is slightly below (2%) of the lowest acti- 4. Conclusions
vation energy obtained by Friedman’s method (154.1 kJ/mol), but
it is in agreement with the range 150–200 kJ/mol previously ana- The sugarcane straw in Brazil would be a potential biomass for
lyzed for biomass in the Section 1 and also presented in Table 1. thermo-chemical conversion processes due to their thermal char-
acteristics and availability. Also, based on the similarities with
3.5. Evaluation of the kinetic parameters bagasse it would be possible to mix both biomasses to use indus-
trially with the current technologies. The reaction model, which
Fig. 5 presents theoretical (atheoretical) and experimental (aexp) better described the reaction, was the two-dimensional diffusion,
conversion curves obtained at different heating rates. These curves combined with activation energy of 149.71 kJ/mol and
were obtained assuming the thermal decomposition reaction of pre-exponential factor of 1.82 * 109 s1. Finally, the reaction kinet-
sugarcane straw as a one-step global reaction (Eq. (7)). It is ics was well described by a one-step global reaction (AVP < 3%),
observed that from the conversion of 0.00 to 0.20, the theoretical indicating that these results could be used for future process mod-
data is slightly underestimated, and above 0.80 is overestimated. eling involving sugarcane straw.
However, for the conversion between 0.20 and 0.80, the theoretical
curve is a good reproduction of the experimental data. The Acknowledgements
variation in the extremes suggests a reaction model modification.
In Table 5 are presented the kinetic parameters determined for The authors gratefully acknowledge the financial support from
the sugarcane straw. The kinetic parameters obtained, i.e., activa- the Coordination for the Improvement of Higher Level Personnel
tion energy, pre-exponential factor, and reaction model, remained (CAPES).
constant for all the heating rates analyzed. This fact confirms the
independency of the process with the heating rate, i.e., was References
successfully avoided the inter-particle and intra-particle effect of
Abbasi, T., Abbasi, S.A., 2010. Biomass energy and the environmental impacts
the transport phenomena. associated with its production and utilization. Renew. Sustain. Energy. Rev. 14,
The activation energy of 149.710 kJ/mol obtained in this 919–937.
research was in the literature range for biomass (Table 1), normally Aboyade, A.O., Görgens, J.F., Carrier, M., Meyer, E.L., Knoetze, J.H., 2013.
Thermogravimetric study of the pyrolysis characteristics and kinetics of coal
between 140 kJ/mol and 200 kJ/mol. Also, the reaction model
blends with corn and sugarcane residues. Fuel. Process. Technol. 106, 310–320.
determined for the one-step global reaction provides a better Anca-Couce, A., Berger, A., Zobel, N., 2014. How to determine consistent biomass
description of the experimental behavior than the reaction model pyrolysis kinetics in a parallel reaction scheme. Fuel 123, 230–240.
normally used in literature, which is a first or nth reaction order. Baroni, E.G., Tannous, K., Rueda-Ordóñez, Y.J., Tinoco-Navarro, L.K., 2015. The
applicability of isoconversional models in estimating the kinetic parameters of
This fact is in agreement with the results and conclusion of biomass pyrolysis. J. Therm. Anal. Calorim.
Baroni et al. (2015), who studied the use of first order reaction 015-4707-9.
144 Y.J. Rueda-Ordóñez, K. Tannous / Bioresource Technology 196 (2015) 136–144

Basu, P., 2010. Biomass Gasification and Pyrolysis – Practical Design and Theory, 1st Pasangulapati, V., Ramachandriya, K.D., Kumar, A., Wilkins, M.R., Jones, C.L.,
ed. Academic Press, Burlington. Huhnke, R.L., 2012. Effects of cellulose, hemicellulose and lignin on
Ceylan, S., Topçu, Y., 2014. Pyrolysis kinetics of hazelnut husk using thermochemical conversion characteristics of the selected biomass. Bioresour.
thermogravimetric analysis. Bioresour. Tech. 156, 182–188. Tech. 114, 663–669.
Costa, S.M., Mazzola, P.G., Silva, J.C.A.R., Pahl, R., Pessoa Jr., A., Costa, S.A., 2013. Use Pérez-Maqueda, L.A., Criado, J.M., 2000. The accuracy of Senum and Yang’s
of sugarcane straw as a source of cellulose for textile fibre production. Ind. approximations to the Arrhenius integral. J. Therm. Anal. Calorim. 60, 909–915.
Crops Prod. 42, 189–194. Poletto, M., Dettenborn, J., Pistor, V., Zeni, M., Zattera, A.J., 2010. Materials produced
Damartzis, Th., Vamvuka, D., Sfakiotakis, S., Zabaniotou, A., 2011. Thermal from plant biomass. Part I: Evaluation of thermal stability and pyrolysis of
degradation studies and kinetic modeling of cardoon (Cynara cardunculus) wood. Mater. Res. 13, 375–379.
pyrolysis using thermogravimetric analysis. Bioresour. Tech. 102, 6230–6238. Rocha, G.J.M., Martin, C., Soares, I.B., Maior, A.M.S., Baudel, H.M., Abreu, C.A.M.,
Di Blasi, C., 2008. Modeling chemical and physical processes of wood and biomass 2011. Dilute mixed-acid pretreatment of sugarcane bagasse for ethanol
pyrolysis. Prog. Energy. Combust. Sci. 34, 47–90. production. Biomass Bioenergy 35, 663–670.
Flynn, J.H., 1997. The ‘‘Temperature Integral’’-it’s use and abuse. Thermochim. Acta. Rueda-Ordóñez, Y.J., Baroni, E.G., Tinoco-Navarro, L.K., Tannous, K., 2015. Modeling
300, 83–92. the kinetics of lignocellulosic biomass pyrolysis. In: Tannous, K. (Ed.),
Friedman, H.L., 1964. Kinetics of thermal degradation of char-forming plastics from Innovative Solutions in Fluid-Particle Systems and Renewable Energy
thermogravimetry. Application to a phenolic plastic. J. Polym. Sci. 6, 183–195. Management. IGI Global, Hershey, pp. 92–130.
Hoi, L.W.S., Martincigh, B.S., 2013. Sugar cane plant fibres: separation and 1-4666-8711-0.ch004.
characterization. Ind. Crops Prod. 47, 1–12. Saad, M.B.W., Oliveira, L.R.M., Cândido, R.G., Quintana, G., Rocha, G.J.M., Gonçalves,
Janković, B., 2008. Kinetic analysis of the nonisothermal decomposition of A.R., 2008. Preliminary studies on fungal treatment of sugarcane straw for
potassium metabisulfite using the model-fitting and isoconversional (model- organosolv pulping. Enzym. Microb. Tech. 43, 220–225.
free) methods. Chem. Eng. J. 139, 128–135. Slopiecka, K., Bartocci, P., Fantozzi, F., 2012. Thermogravimetric analysis and kinetic
Khawam, A., Flanagan, D.R., 2006. Solid-state kinetic models: basics and study of poplar wood pyrolysis. Appl. Energy. 97, 491–497.
mathematical fundamentals. J. Phys. Chem. B. 110, 17315–17328. Sluiter, A., Ruiz, R., Scarlata, C., Sluiter, J., Templeton, D., 2008. Determination of
Leal, M.R., Galdos, M.V., Scarpare, F.V., Seabra, J.E.A., Walter, A., Oliveira, C.O.F., 2013. extractives in biomass. Technical report NREL/TP-510-42619. National
Sugarcane straw availability, quality, recovery and energy use: a literature Renewable Energy Laboratory.
review. Biomass. Bioenerg. 53, 11–19. procedures.html.
Liu, Q., Zhong, Z., Wang, S., Luo, Z., 2011. Interaction of biomass components during Sluiter, A., Hames, B., Ruiz, R., Scarlata, C., Sluiter, J., Templeton, D., Crocker, D., 2012.
pyrolysis: a TG-FTIR study. J. Anal. Appl. Pyrolysis 90, 213–218. Determination of structural carbohydrates and lignin in biomass. Technical
Luz, S.M., Gonçalves, A.R., Del’Arco Jr., A.P., Leão, A.L., Ferrão, P.M.C., Rocha, G.J.M., report NREL/TP-510-42618. National Renewable Energy Laboratory. http://
2010. Thermal properties of polypropylene composites reinforced with
different vegetable fibres. Adv. Mater. Res. 123–125, 1199–1202. Vassilev, S.V., Baxter, D., Andersen, L.K., Vassileva, C.G., Morgan, T.J., 2012. An
Mesa-Pérez, J., Rocha, J., Barbosa-Cortez, L., Penedo-Medina, M., Luengo, C., overview of the organic and inorganic phase composition of biomass. Fuel 94,
Cascarosa, E., 2013. Fast oxidative pyrolysis of sugar cane straw in a fluidized 1–33.
bed reactor. Appl. Therm. Eng. 56, 167–175. Vyazovkin, S., Burnham, A.K., Criado, J.M., Pérez-Maqueda, L.A., Popescu, C.,
Mishra, G., Bhaskar, T., 2014. Non isothermal model free kinetics for pyrolysis of rice Sbirrazzuoli, N., 2011. ICTAC kinetics committee recommendations for
straw. Bioresour. Tech. 169, 614–621. performing kinetic computations on thermal analysis data. Thermochim. Acta
Moraes, M., Georges, F., Almeida, S., Damasceno, F., Maciel, G., Zini, C., Jacques, R., 520, 1–19.
Caramão, E., 2012. Analysis of products from pyrolysis of Brazilian sugar cane White, J.E., Catallo, W.J., Legendre, B.L., 2011. Biomass pyrolysis kinetics: a
straw. Fuel. Process. Technol. 101, 35–43. comparative critical review with relevant agricultural residue case studies. J.
Órfão, J.J.M., Antunes, F.J.A., Figueiredo, J.L., 1999. Pyrolysis kinetics of Anal. Appl. Pyrolysis. 91, 1–33.
lignocellulosic materials – three independent reactions model. Fuel 78, 349– Yao, F., Wu, Q., Lei, Y., Guo, W., Xu, Y., 2008. Thermal decomposition kinetics of
358. natural fibers: activation energy with dynamic thermogravimetric analysis.
Ounas, A., Aboulkas, A., El harfi, K., Bacaoui, A., Yaacoubi, A., 2011. Pyrolysis of olive Polym. Degrad. Stab. 93, 90–98.
residue and sugarcane bagasse: non-isothermal thermogravimetric analysis.
Bioresour. Tech. 102, 11234–11238.