Professional Documents
Culture Documents
Bioresource Technology
journal homepage: www.elsevier.com/locate/biortech
h i g h l i g h t s
a r t i c l e i n f o a b s t r a c t
Article history: The study concerns the pyrolysis kinetics of agricultural wastes, corn straw (CS) and rice husk (RH). Ther-
Received 14 July 2012 mogravimetric experiments were carried out in a thermogravimetric analyzer under inert conditions, and
Received in revised form 20 September 2012 operated at different heating rates ranging from 5 to 40 K/min. As the increment of heating rates, the vari-
Accepted 21 September 2012
ations of characteristic parameters from the TG–DTG curves were determined. Iso-conversional Starink
Available online 29 September 2012
approach and Avrami theory were used to evaluate the kinetic parameters, including apparent activation
energy and reaction order. For the range of conversion fraction investigated (20–80%), the apparent acti-
Keywords:
vation energy of CS initially increased from 98.715 to 148.062 kJ/mol and then decreased to 144.387 kJ/
Biomass
Kinetics
mol afterwards, whilst the apparent activation energy of RH increased gradually from 50.492 to
Pyrolysis 88.994 kJ/mol. With varied temperatures (517–697 K), the corresponding value of reaction order was
Apparent activation energy increased from 0.288 and 0.359 to 0.441 and 0.692, along with a decrease to 0.306 and 0.445, respectively.
Reaction order Ó 2012 Elsevier Ltd. All rights reserved.
0960-8524/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.biortech.2012.09.089
C. Gai et al. / Bioresource Technology 127 (2013) 298–305 299
Table 1
Classification scheme of thermochemical conversion of agricultural wastes for fuels.
Conversion temperature TCC technology Major products Major pollutants Typical decontamination technology Refs.
Low temperatures Pyrolysis Bio-char Ash, toxic organic pollutants Hexane extraction solvent Bernardo et al. (2012)
Intermediate temperatures Pyrolysis Bio-oil Corrosiveness of the oil Hydro treating Ozawa (1965)
High temperatures Pyrolysis Bio-gas Tars Catalytic converting Chen et al. (2003)
High temperatures Gasification Syngas Tars, dust Catalytic converting. Dedusting Karmakar and Datta (2011)
Low temperatures Hydrothermal liquefaction Bio-oil Corrosiveness of the oil Hydro treating Toor et al. (2011)
terms of gaseous compounds or bounded to the ash. In terms of the compensation effect’. And it is also called the iso-conversional
potential technologies of syngas cleaning, it can generally be di- approach, since the apparent activation energy obtained by this
vided into two categories including treatments during gasification approach is a function of the extent of conversion. Multistep
and gas cleaning after gasification. Undoubtedly, gasifier design reaction mechanism of the pyrolysis kinetics can be obtained from
parameters and operating parameters have effect on the formation the knowledge of the relationship between apparent activation
of those pollutants during gasification. Therefore, in order to min- energy and conversion degree. Besides, it is adequate to evaluate
imize the formation of such pollutants during the gasification pro- the reaction kinetics over a wider temperature region.
cess, the knowledge of pyrolysis kinetics is also needed, which is Therefore, it is more reliable for analyzing the thermal data
usually investigated by the thermogravimetric analysis (TGA). using the multi-heating rate methods together with iso-conver-
TGA can be carried out under isothermal or non-isothermal con- sional approaches in recent years. Iso-conversional approaches
ditions. The main disadvantage of the traditional isothermal meth- have been increasingly used to investigate the thermal decomposi-
od is that there is a small mass-lose before reaching the desired tion kinetics of varied materials such as cattle manure (Otero et al.,
temperature, causing a certain error when investigate the pyrolysis 2011), sensitive plants (Wongsiriamnuay and Tippayawong, 2010)
reaction mechanism of the solid state. Therefore, non-isothermal or municipal solid waste (Liu et al., 2009).
method has become a widespread analytical technique in recent In terms of the studies of pyrolysis kinetics of corn straw and
decades due to the high sensitivity to experimental noise compared rice husk, most investigations adopt the model-fitting method,
to isothermal methods. Characteristic parameters of the sample can and usually take the first-order reaction as the presumed reaction
be obtained from the TG and DTG curves during the TGA experi- model. Some other researches generally assume a certain reaction
ments; then in order to investigate the pyrolysis mechanism, the ki- order (e.g. normally range from zero-order to three-order) and a
netic parameters of pyrolysis reactions can be calculated using the particular reaction model. However, just as the former discussion,
characteristic parameters, including apparent activation energy E, the pyrolysis of the solid-state sample is a complicated physical
pre-exponential factor A, reaction order n and reaction model f(a). and chemical process. Not only does the reaction order may vary
Assessing these kinetic parameters of pyrolysis of the sample rely greatly during the reaction process, but such significant difference
on different mathematical approaches. may also exist in the reaction model preliminary selected. There-
Many mathematical approaches have been developed to calcu- fore, the main objective of this study is to investigate the variation
late the kinetic parameters, in terms of Coats–Redfern (Daood of pyrolysis kinetics of corn straw and rice husk at different con-
et al., 2010), Freeman and Carroll method (Font and Williams, version fractions and reaction temperature. It is hoped that the
1995), et al. These methods are usually based on the preliminary pyrolysis kinetics of agricultural wastes obtained in the current
assumption of a certain reaction order and reaction model, such study can give a better and more comprehensive understanding
as mampel, nucleation, autocatalytic, diffusional and contracting of the pyrolysis process, as well as to help establish the kinetic
geometry, et al. Then the apparent activation energy and pre- model of biomass gasification for design and scaling of industrial
exponential factor are calculated simultaneous, but not separately. gasification reactors.
Consequently, these approaches are viewed as the model-fitting
method. This kind of method often employs the single and simple
2. Methods
reaction model during the reaction process, which can result in a
substantial divergence; this discrepancy results from the difference
2.1. Raw material properties
between the ideal reaction model and actual heterogeneous
reaction process consisting of series of parallel and sequential reac-
As typical kinds of agricultural residues, corn straw and rice
tions. Besides, the values of E and A will vary greatly on condition
husk were selected to carry out thermogravimetric experiments.
that different reaction model is preliminary assumed; but the two
They were both locally available materials. Before the pyroylsis
parameters can always get good linear regression between each
tests, the samples were air-dried, crushed and then sieved. The
other. This manifestation of ‘kinetic compensation effect’ could
fraction 20–40 mesh was used as power samples for the experi-
result in some difficult in determining the accurate reaction model,
mental runs. As for the properties of the applied feedstocks, the
since more than one reaction model can be obtained using this
methods of proximate and ultimate analysis are described by Gai
method. Therefore, adopting the model-fitting method may bring
and Dong (2012). The proximate, ultimate and biochemical analy-
about the highly uncertain values of the kinetic parameters.
ses of the corn straw and rice husk are listed in Table 2.
In respect to these faults of the model-fitting method, another
approach called model-free method has been more widespread
employed to assessing the pyrolysis kinetics of the solid state in 2.2. TGA experiments
recent years, such as Friedman (Aboyade et al., 2011), Flynn–
Wall–Ozawa (Ounas et al., 2011) and Kissinger–Akahira–Sunose Thermogravimetric experiments were performed using a ther-
(Damartzis et al., 2011) approach. In this method, the apparent mogravimetric analyzer (TGA-SDTA 851e, METTLER TOLEDO),
activation energy can be calculated at the same fraction conversion which detects the mass-loss signal with a resolution of 0.1 lg. In
of the sample in multiple TG curves, without assuming the reaction order to avoid the heat-transfer limitation generated by the sample
model. Hence this approach eliminates the phenomenon of ‘kinetic itself, a small sample mass of 10 mg was placed in the pan of the
300 C. Gai et al. / Bioresource Technology 127 (2013) 298–305
Table 2
Proximate, ultimate and biochemical analyses of the CS and RH.
Proximate analysis (wt.% dba) Ultimate analysis (wt.% dafb) Biochemical analysis (daf b)
Corn straw Rice husk Corn straw Rice husk Corn straw Rice husk
Fixed carbon 13.21 13.85 Carbon 43.83 41.56 Cellulose 36.4e 29.2f
Ash 11.80 18.91 Hydrogen 5.75 5.23 Hemicelluloses 22.6e 20.1f
Volatile matter 75.0 64.27 Oxygend 45.01 39.75 Lignin 16.6e 18.2f
Moisture (arc) 6.57 7.26 Nitrogen 0.97 0.59 Extractives 7.82e 6.71f
a
Dry basis.
b
Dry and ash free basis.
c
As received basis.
d
By difference.
e
Ref. White et al. (2011).
f
Ref. Mansaray and Ghaly (1999).
Z T
thermobalance, which was enough to fill the pan due to the low E E
exp dT ¼ ð0:00484e1:0516u Þ ð4Þ
bulk density of the ground samples. Afterwards, experiments were T0 RT R
carried out at heating rates of 5, 10, 20 and 40 K/min, respectively.
Samples were firstly held at 298 K for 1 min and then heated to Substituting of Eq. (4) into Eq. (3), and then taking the logarithm
1073 K at the specific heating rate. After that, samples were held of both sides result in the expression for the FWO integral iso-con-
at 1073 K for 1 min. For suppressing mass transfer effect to a min- versional method (Ozawa, 1965):
imum, pure nitrogen (99.99% purity) was used as the carrier gas,
and it was kept at a steady flow rate of 80 mL/min during all exper- AE E E
ln bi ¼ ln 2:315 0:4567 ¼ C 0 0:4567 ð5Þ
iments. Under each test of heating rate, a separate blank run was RGðaÞ RT RT
carried out for baseline correction, using an empty pan. In terms of the Kissinger–Akahira–Sunose (KAS) method, the
employed empirical approximation is the following expression:
2.3. Theory
Z T
E Eb
As the typical kind of heterogeneous solid-state reaction, the exp dT ¼ ð6Þ
T0 RT RAT
primary pyrolysis process of biomass is represented by the follow-
ing reaction scheme: Biomass ? char + (volatiles + gases) (Kumar Substituting of Eq. (6) into Eq. (3), and then taking the logarithm
Sadhukhan et al., 2009). The global kinetics of the biomass pyroly- of both sides result in the expression for the KAS integral iso-
sis reaction can be described as: conversional method (Kissinger, 1957):
!
da bi Ak R Ek Ek
¼ kðTÞf ðaÞ ð1Þ ln ¼ ln ¼ CK ½i ¼ 1; 2; 3; 4; 5; 6 ð7Þ
dt T 2pi Ek RT pi RT pi
where a is the fraction of conversion, a = (W0 Wt)/(W0 W1) (W0 According to the Starink method (Starink, 1996), the approxi-
and W1 are sample masses at the beginning and at the end of mass mate expression of FWO method (Eq. (5)) and KAS method (Eq.
loss reaction, respectively. Wt is the sample mass at time t/temper- (7)) can both be transformed into the same general formula as:
ature T). f(a) is the differential function of conversion. T is the tem-
perature, k(T) is the rate constant which is described by Arrhenius b BE
ln ¼ CS ð8Þ
equation, k(T) = A exp(E/RT) (A is the pre-exponential factor, E is TS RT
the apparent activation energy, R is the gas constant (8.314 J/ Where for the FWO method (Eq. (7)) s = 0, B = 0.4567; for the
(mol K)).Eq. (1) can be converted into Eq. (2) expressed as: KAS method (Eq. (9)) s = 2, B = 1. After a further exact analysis by
Starink, the parameters of s and B has been adjusted to s = 1.8,
da A E
¼ exp f ðaÞ ð2Þ B = 1.0033, respectively. Besides, it has been verified by Starink
dT b RT
that the precision of apparent activation energy calculated by this
Where b is the linear heating rate (b = dT/dt) and it is a constant. method is higher than those of FWO method and KAS method.
Carry out the integration of the both sides of Eq. (2), and it can Hence the Starink method can be described as:
be further integrated into: b E
ln ¼ C S 1:0037 ð9Þ
Z a Z T T 1:8 RT
da A E A
¼ GðaÞ ¼ exp dT ¼ pðxÞ ð3Þ
0 f ðaÞ b T0 RT b For a given conversion fraction, a, the points of ln(b/T1.8) versus
1/T at different temperature heating rates can be fitted to a straight
Where p(x) in Eq. (3) is known as the temperature integral. line, and the slope of the line corresponds to 1.0037E/R. Hence
Although it does not have an exact analytical solution in closed- the apparent activation energy E can be calculated from the slope
form, it can be approximately represented via different empirical of the straight line.
interpolation formulas in terms of Doyle, Agrawal, Gorbatchev Reaction order is also a significant parameter to investigate the
and Frank–Kameneskii approximation. pyrolysis characteristic of biomass, besides apparent activation
The Flynn–Wall–Ozawa (FWO) method and Kissinger–Akahira– energy. In order to calculate the reaction order, Avrami theory
Sunose (KAS) method are the two most common model-free (Ruitenberg et al., 2001) was used in this study, and it can be
methods of determination of kinetic parameters. The empirical described as:
approximation used by the Flynn–Wall–Ozawa (FWO) method is
Doyle’s approximation (Doyle, 1961): kðTÞ
a ¼ 1 exp ð10Þ
bn
C. Gai et al. / Bioresource Technology 127 (2013) 298–305 301
Where a, b and k(T) are the same parameters described in Eq. (1),
and as the exponent of b, n represents the reaction order.
Taking the double logarithm and transposing, Eq. (10) is trans-
formed into the following equation:
E
lnð lnð1 aÞÞ ¼ ln A n ln b ð11Þ
RT
For a given temperature, T, the points of ln(ln(1a)) versus lnb
at different temperature heating rates can be fitted to a straight
line, and the slope of the line corresponds to n. Thus, the reaction
order can be deduced from the slope of the straight line.
Table 3
Characteristic parameters of CS and RH during the pyrolysis process.
some other parameters also got higher values through the whole lated energy of activation and the square of correlation coefficient
thermal decomposition process of the two samples, including Ti (r2) are listed in Table 4. As the result of ANOVA analysis, the values
(i.e. refer in particular to the stage II) and Tf. This may have been of standard deviation (SD) are also listed in Table 4, in addition to
due to the fact that with the increment of heating rate, the time the regression analysis adopted.
consumed in reaching the settled final temperature is lesser. Con- It can be seen from Table 4 that the apparent activation energy
sequently, the heat transferred from the heating furnace within the of the two samples both varied greatly with different conversion
thermo-gravimetric instrument to the tested sample is limited, fractions, instead of keeping constant. And they showed varied var-
which may bring about a discrepancy between the recorded tem- iation trends. To be specific, as the increase of conversion fraction
perature and the actual sample temperature. In other words, the from 20% to 80%, the apparent activation energy of corn straw
reaction extent represented by horizontal axis and vertical axis of underwent the process of a slight fall to 144.387 kJ/mol after a rise
TG–DTG curves synchronously recorded by the computer are actu- from 98.715 to 148.062 kJ/mol, whereas the apparent activation
ally not coincided. In terms of the parameter of HTZ for the two energy of rice husk increased gradually from 50.492 to 88.994 kJ/
samples, it always increased with the increase of heating rates. mol. The different energies of activation at different conversion
This may also have been due to the heat-transfer limitation men- fractions suggest that the pyrolysis of corn straw and rice husk
tioned above. should proceed with varied reaction mechanisms.
However, not all characteristic parameters are increasing. The The data in Table 4 also demonstrate that the average value of
maximum weight loss rate DTGmax for both corn straw and rice the apparent activation energy of corn straw is 129 kJ/mol, which
husk were found to decrease a certain value. Specifically, with
the heating rate increased from 5 to 40 K/min, as for the corn
straw, the value of DTGmax was decreased from 0.098%/K for
the first segment and 0.481%/K for the second segment to
0.078%/K and 0.412%/K, respectively. As for the rice husk, the cor-
responding value of DTGmax was decreased from 0.184%/K and
0.433%/K to 0.160%/K and 0.391%/K, respectively. Biagini et al. Bia-
gini et al. (2008) carried out thermogravimetric experiment of
three biomass residues (e.g. rice husks, olive cake, cacao shells).
It was found that the maximum weight loss rate of three kinds
of sample all decreased with the heating rate increasing from
5 K/min to 100 K/min. Ptáček et al. (2011) investigated the kinetic
parameters of the thermal decomposition of kaolinite and reported
the same variation trend that the maximum weight loss rate grad-
ually decreased as the heating rate was increased from 1 to 40 K/
min. Therefore, taking into account the differences in the proper-
ties of the tested sample and operating conditions, the result in
the current study showed similar trends to the results published
in literature.
The data in Table 3 also showed that the parameter of ML for
two samples both appeared a narrower fluctuation with different
heating rates. According to Jeguirim and Trouvé (2009), this is be-
cause the total mass loss at any stage is mainly determined by the
essential component of the tested sample, which is not affected by
the heating rate.
Table 4
Slope, apparent activation energy (E) and correlation coefficient (r2) deduced from Starink method for CS and RH.
Sample Conversion (%) Slope E (kJ/mol) r2 SD Sample Conversion (%) Slope E (kJ/mol) r2 SD
CS 20 11.918 98.715 0.994 0.012 RH 20 6.096 50.492 0.991 0.017
30 14.239 117.942 0.989 0.022 30 7.412 61.398 0.999 0.002
40 15.853 131.312 0.987 0.022 40 8.295 68.713 0.998 0.002
50 17.345 143.667 0.984 0.027 50 9.234 76.489 0.998 0.001
60 17.522 145.136 0.978 0.033 60 9.958 82.488 0.989 0.001
70 17.875 148.062 0.984 0.045 70 10.598 87.784 0.995 0.003
80 17.432 144.387 0.993 00.032.015 80 10.744 88.994 0.997 0.006
Average 129 79
is higher than that of rice husk, 79 kJ/mol. Since the value of appar-
ent activation energy represents the minimum energy required to
break the chemical bonds between atoms, which means the higher
value of the apparent activation energy, slower is the reaction. It is
a parameter indicating the reactivity and sensitivity of overall reac-
tion rate. Consequently, the reactivity of pyrolysis of rice husk with
relatively lower apparent activation energy is higher than that of
corn straw. It explains the experimental result that the entire ther-
mal decomposition process of the corn straw sample is later than
that of rice husk, which is indicated in Section 3.1.
Aboyade (2011) carried out non-isothermal kinetic analysis of
the pyrolysis of corn cobs in the heating rate range of 10–50 °C/
min. It was reported that the apparent activation energy in the
0.1–0.8 conversion interval ranged from 75–130 kJ/mol. Thus, the
result in this study showed the similar range to those obtained
in the present study in spite of the small property differences be-
tween corn cob and corn straw. Sharma and Rajeswara Rao
(1999) investigated the kinetics of pyrolysis of rice husk under
non-isothermal conditions in nitrogen. The experiments were con-
ducted at different temperature heating rates from 5 to 100 K/min,
and the energies of activation calculated was found to range from
72.7 to 97.1 kJ/mol. Taking into account the differences in the
experimental conditions, the result of 79 kJ/mol in this study
showed good agreement with the conclusion published in the
literature.
Researchers have also paid special attention to investigate the
three biochemical composition of biomass, including cellulose,
hemicellulose and lignin. As suggested by Huang et al. (2011),
the apparent activation energy of cellulose, hemicellulose and lig-
nin is generally about two hundreds, one hundreds and tens kJ/
mol, respectively. Thus, it indicates that property of the biomass
Fig. 4. Regression lines to reaction order proposed by Avrami theory for CS (a) and
is also a key factor determining the variation of apparent activation RH (b) at heating rates of 5, 10, 20, 40 K/min.
energy in the pyrolysis process.
3.3. Variation of reaction order of CS and RH and rice husk are different. From the view of the varied mathemat-
ical mutual relation of the two kinetic parameter for the two sam-
During the major pyrolysis process, in order to evaluate the ples, it can be further affirmed that the pyrolysis reaction
dependence of reaction order on temperature, seven levels of tem- mechanisms of corn straw and rice husk are obviously different.
perature were also employed at four temperature heating rates of To determine this relationship of the two kinetic parameter and
5, 10, 20 and 40 K/min. Adopting the Avrami theory, the regression the reaction mechanism, a kinetic analysis of elementary reaction
lines of corn straw and rice husk are shown in Fig. 4 and the calcu- is required, which is recommended for further study.
lated reaction order is listed in Table 5. The corresponding values of The data in Table 5 also show that the mean value of the reac-
r2 and SD are also presented in Table 5. tion order of corn straw and rice husk is 0.365 and 0.539, respec-
It can be seen from Table 5 that the Avrami theory is suitable for tively. Vuthaluru (2004) investigated the pyrolytic behavior of
determining the kinetic parameter of reaction order, based on the coal/biomass blends using thermogravimetric analysis. It was
data of r2 calculated. The reaction order of corn straw and rice husk found that the orders of reaction were in the range of 0.21–1.6,
both differed greatly rather than remained constant. However, with different blend ratios of coal/biomass. In terms of the 0:100
they showed the same variation trend. With varied temperatures blends for coal/wood waste and coal/wheat straw, the correspond-
(517–697 K), the reaction order of corn straw was first increased ing reaction orders were 0.42 and 0.52, respectively. Hence the re-
from 0.288 to 0.441 and then decreased to 0.306. As for the rice sults of reaction order obtained in the current paper were in a
husk, it increased from 0.359 to 0.692 and decreased to 0.445 similar range with the results published in literature. It also indi-
afterwards. cates that the pyrolysis kinetics is greatly affected by the type
Consequently, it can be observed that the varying currents of and composition of the biomass feedstock (Sanchez-Silva et al.,
apparent activation energy and the reaction order for corn straw 2012).
304 C. Gai et al. / Bioresource Technology 127 (2013) 298–305
Table 5
Slope, reaction order (n) and correlation coefficient (r2) deduced from Avrami theory for CS and RH.
Kumar Sadhukhan, A., Gupta, P., Kumar Saha, R., 2009. Modelling of pyrolysis of Toor, S.S., Rosendahl, L., Rudolf, A., 2011. Hydrothermal liquefaction of biomass: a
large wood particles. Bioresour. Technol. 100, 3134–3139. review of subcritical water technologies. Energy 36, 2328–2342.
Sanchez-Silva, L., López-González, D., Villaseñor, L., Sánchez, P., Valverde, J.L., 2012. Vuthaluru, H.B., 2004. Investigations into the pyrolytic behavior of coal/biomass
Thermogravimetric-mass spectrometric analysis of lignocellulosic and marine blends using thermogravimetric analysis. Bioresour. Technol. 92, 187–195.
biomass pyrolysis. Bioresour. Technol. 109, 163–172. White, J.E., Catallo, W.J., Legendre, B.L., 2011. Biomass pyrolysis kinetics: a
Sharma, A., Rajeswara Rao, T., 1999. Kinetics of pyrolysis of rice husk. Bioresour. comparative critical review with relevant agricultural residue case studies. J.
Technol. 67, 53–59. Anal. Appl. Pyrol. 91, 1–33.
Starink, M.J., 1996. A new method for the derivation of activation energies from Wongsiriamnuay, T., Tippayawong, N., 2010. Thermogravimetric analysis of giant
experiments performed at constant heating rate. Thermochim. Acta 288, 97–104. sensitive plants under air atmosphere. Bioresour. Technol. 101, 9314–9320.