Energy Conversion and Management 89 (2015) 251–259

Contents lists available at ScienceDirect

Energy Conversion and Management

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

Determination of pyrolysis characteristics and kinetics of palm kernel

shell using TGA–FTIR and model-free integral methods
Zhongqing Ma a,b,⇑, Dengyu Chen c, Jie Gu c, Binfu Bao a,b, Qisheng Zhang c
a
School of Engineering, Zhejiang Agriculture & Forestry University, Lin’an, Zhejiang 311300, PR China
b
National Engineering & Technology Research Center of Wood-Based Resources Comprehensive Utilization, Lin’an, Zhejiang 311300, PR China
c
School of Materials Science & Engineering, Nanjing Forestry University, Nanjing 210037, Jiangsu, PR China

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

Article history: Palm kernel shell (PKS) from palm oil production is a potential biomass source for bio-energy production.
Received 20 May 2014 A fundamental understanding of PKS pyrolysis behavior and kinetics is essential to its efficient thermo-
Accepted 28 September 2014 chemical conversion. The thermal degradation profile in derivative thermogravimetry (DTG) analysis
Available online 17 October 2014
shown two significant mass-loss peaks mainly related to the decomposition of hemicellulose and cellu-
lose respectively. This characteristic differentiated with other biomass (e.g. wheat straw and corn stover)
Keywords: presented just one peak or accompanied with an extra ‘‘shoulder’’ peak (e.g. wheat straw). According to
Biomass
the Fourier transform infrared spectrometry (FTIR) analysis, the prominent volatile components gener-
Palm kernel shell
Pyrolysis
ated by the pyrolysis of PKS were CO2 (2400–2250 cm1 and 586–726 cm1), aldehydes, ketones, organic
Kinetic acids (1900–1650 cm1), and alkanes, phenols (1475–1000 cm1). The activation energy dependent on
Model free the conversion rate was estimated by two model-free integral methods: Flynn–Wall–Ozawa (FWO)
TGA–FTIR and Kissinger–Akahira–Sunose (KAS) method at different heating rates. The fluctuation of activation
energy can be interpreted as a result of interactive reactions related to cellulose, hemicellulose and lignin
degradation, occurred in the pyrolysis process. Based on TGA–FTIR analysis and model free integral
kinetics method, the pyrolysis mechanism of PKS was elaborated in this paper.
Ó 2014 Elsevier Ltd. All rights reserved.

1. Introduction turbine [4–6]. In addition, pyrolysis is an important sub-step of

gasification technology yielded producer gas for electricity supply
Biomass is a promising, clean, renewable energy source due to [7,8]. Therefore, a fundamental understanding of PKS pyrolysis
its abundance, wide distribution, and CO2 neutrality [1]. Palm ker- behavior and kinetics is essential to its efficient thermochemical
nel shell (PKS) is the primary residue left over from palm oil pro- conversion.
duction in Malaysia, the largest palm oil producer in the world. Thermogravimetric analysis (TGA), coupled with Fourier trans-
Malaysia generated the PKS reached up to 471 thousand tons in form infrared spectrometry (FTIR), is a good means by which to
2000 [2]. Currently, the majority of the PKS is used for primary study not only the mass-loss characteristics and kinetics parame-
energy generation with low thermal conversion efficiency via an ters of the thermal decomposition process, but also identify the
oversimplified utilization pattern. This typically involves co-com- volatile components generated in real-time. With the single use
bustion with coal in boilers to supply heat [3]. However PKS can of TGA, the thermogravimetric characteristics of PKS was investi-
be converted into high-quality bio-fuels through thermochemical gated in inert atmosphere by Kim [3], Lee [9] and Asadullah [10].
conversion. Asadullah [10] observed that the thermal decomposition of PKS
Pyrolysis technology is attracting more attention, because at was slower compared to other biomasses because of higher frac-
fast heating rate, biomass can be converted into higher energy con- tion of lignin. With the combined use of TGA and FTIR, the volatile
tent transportable liquid (bio-oil). Using upgrading technology (e.g. components of biomass pyrolysis process, such as poplar wood
hydrodeoxygenation technology), crude bio-oil will own higher [11] and pine wood sawdust [12], were identified. The main com-
calorific value for extensive application in boiler, engine and ponents were some small moleculars gases (CO, CO2, H2O, CH4) and
various kinds of organic compounds. The identification is on the
⇑ Corresponding author at: School of Engineering, Zhejiang Agriculture & Forestry basis of the characteristic absorbances of the functional groups in
University, Lin’an, Zhejiang 311300, PR China. Tel./fax: +86 571 6374 1098. the evolved gases. However, less research focused on the PKS
E-mail address: mazqzafu@163.com (Z. Ma). biomass using TGA–FTIR.

http://dx.doi.org/10.1016/j.enconman.2014.09.074
0196-8904/Ó 2014 Elsevier Ltd. All rights reserved.
252 Z. Ma et al. / Energy Conversion and Management 89 (2015) 251–259

Thermal analysis kinetics is an important approach to study the 2.2. TGA–FTIR analysis
mechanisms of the thermochemical conversion of biomass.
Non-isothermal kinetics can be classified into model-free and The TGA–FTIR test setup consisted of a thermogravimetric ana-
model-fitting categories. Both methods have their benefits. They lyzer (TGA Q5000, TA Instruments, USA) and a Fourier transform
are complementary rather than competition [13]. Recently, the infrared spectrometry (Nicolet 6700, Thermo Fisher Scientific,
model-free method, also called the iso-conversional method, was USA) apparatus. Approximately 15 mg of PKS sample was used
the most common used methods in the kinetics study of biomass for each test. The temperature was raised from room temperature
pyrolysis process [14–19]. The ICTAC Kinetic Committee recom- to 800 °C under heating rates of 10, 20, 30 and 40 °C min1. And the
mended that using multiple heating rate programs will obtain heating rate of the three biomass components (cellulose, xylan and
more reliable kinetic parameters instead of single heating rate lignin) was 20 °C min1. The flow rate of the carrier gas (high-pur-
program [20]. For this method, the essential assumption is that ity N2) was 70 ml min1. The temperature of the transfer line
the reaction rate for a constant extent of conversion (a) depends between the TGA and FTIR apparatuses was 210 °C. The resolution
only on the temperature (T). and spectral region of the FTIR were 4 cm1 and 4000 to 400 cm1,
The model free method can generally be split in two categories: respectively, and the spectrum scan was conducted with 8-second
differential and integral. Due to employ the instantaneous rate intervals.
value, the differential iso-conversional method is sensitive to
experimental noise, and makes the numerical value unstable. 2.3. Kinetic modeling
However, this phenomenon will be effectively avoided by
using integral method, especially in TGA experiment [20]. The 2.3.1. The basis of non-isothermal kinetic model
Flynn–Wall–Ozawa (FWO) integral method [21,22] and Kissinger– Solid reaction kinetics is suitable for biomass pyrolysis to pro-
Akahira–Sunose (KAS) integral method [23,24] using different duce char and gas with a linearly increasing heating rate. The
approximations are two typical methods. Thus they were used to non-isothermal thermal reaction rate equation (Arrhenius equa-
estimate the activation energy of PKS pyrolysis in this paper. tion) is shown below,
The key objective of this study was to investigate the pyrolysis  
da E
behavior and kinetics of PKS using TGA–FTIR analysis. This study ¼ A exp  f ðaÞ ð1Þ
dt RT
first focused on the mass-loss characteristics and volatile compo-
m  mt
nents of the PKS and the biomass three components (cellulose, a¼ o ð2Þ
xylan and lignin) at the heating rate of 20 °C min1. Then, two mo  m1
model free integral method (FWO and KAS methods) were used where a is the mass-loss fraction, m0 is the initial mass of the
to calculate the activation energy describing the thermal devolatil- sample, mt is the mass of sample at the time t, m1 is the final,
ization mechanism of the PKS pyrolysis processes with different non-decomposable mass of the sample following the completed
conversion rate (a) using multi-heating rate method (heating rates pyrolysis reaction, A is the frequency factor, E is the activation
of 10, 20, 30, and 40 °C min1). This study would be helpful in energy, R is the universal gas constant, T is the temperature in
effective design and operation of thermochemical conversion units Kelvin, and f(a) is the differential mechanism function. The mass-
fed by PKS. loss fraction a is defined as Eq. (2).
The heating rate b is defined as dT/dt. Substituting this into Eq.
2. Experimental (1) yields,
 
da A E
2.1. Materials ¼ exp  f ðaÞ ð3Þ
dT b RT

PKS, obtained from a palm oil factory in Malaysia, was ground The integral form of Eq. (3) is expressed as follows,
to a fine powder. The powder was passed through 40- to 60-mesh Z a Z T  
da A E AE
sieves, allowing particles of sizes from 250 to 380 lm to pass suit- GðaÞ ¼ ¼ exp  dT ¼ PðuÞ ð4Þ
0 f ðaÞ b 0 RT bR
able for component and TGA–FTIR analyses. The proximate analy-
sis of the PKS was performed according to ASTM D3172–07a. The where G(a) is the integral form of f(a), P(u) is an approximation, and
ultimate analysis was carried out following the CHNS/O model u is defined as the equation of u = E/R.
using an elemental analyzer (Vario EL III, Elementary, Germany),
and the oxygen content was estimated as the balance. The results 2.3.2. Model free integral method
are listed in Table 1. The PKS powder was oven-dried for 5 h at According to Eqs. (1)–(4), different integral kinetic models can
105 °C before TGA–FTIR analysis was conducted. The samples of be fitted to the PKS pyrolysis stage. To obtain the quantities E,
cellulose, xylan and alkali lignin are purchased from Sigma–Aldrich two model free integral methods, namely the Flynn–Wall–Ozawa
Co., Ltd. (USA), and they were all in a form of powder. Because of (FWO) method and Kissinger–Akahira–Sunose (KAS) method, were
commercial hemicellulose difficultly obtained. Thus xylan was used in this paper. The activation energy estimated in this method
used to be a model compound of hemicelluloses. And this method is a function of the conversion rate (a). The FWO and KAS equations
was widely used in the study of biomass pyrolysis process. are defined as Eqs. (5) and (6), respectively.

Table 1
Ultimate, proximate, and biochemical analysis of PKS.

Ultimate analysis, ash free (mass%) Proximate analysis, dry basis (mass%) Biochemical analysisb (mass%)
Carbon 51.56 Volatiles 75.14 Cellulose 33.03
Hydrogen 6.31 Fixed carbon 22.05 Hemicelluloses 23.82
Oxygena 41.33 Ash 2.81 Lignin 45.59
Nitrogen 0.7 Moisture content 12.69
Sulfur 0.1 LHV/MJ kg1 17.3
a
By difference.
b
Information from [3].
 Z. Ma et al. / Energy Conversion and Management 89 (2015) 251–259 253

AE E 185–325 °C. The mass percentage of pyrolysis residue of xylan was

lg b ¼ lg  2:315  0:4567 ð5Þ
RGðaÞ RT about 26.81 mass%. Compared to the xylan, cellulose pyrolyzed at
 
b AR E higher temperature range of 270–400 °C, but had a more narrow
ln 2 ¼ ln  ð6Þ range of maximum weight loss at 290–380 °C. The maximum
T EGðaÞ RT
rate of weight loss and the final residue of decomposition were
The activation energy E can be obtained from the linear plots of 2.317%/°C and 3.64 mass%, respectively. In contrast to the sharper
log(b) vs. 1/T and log(b/T2) vs. 1/T for each conversion rate (a), DTG peaks of xylan and cellulose, lignin pyrolyzed slowly in a
where the slope is 0.4567E/R and E/R, respectively. Generally, wider temperature of 100–800 °C and produced highest mass
three or more heating rates (10, 20, 30 and 40 °C min1 in this percentage of residue along with 44.74%. Therefore, the order of
study) should be used to obtain reliable values of the activation thermal stability was established: lignin > cellulose > xylan
energy. (representing hemicellulose). The result was also confirmed by
Yang et al. [25] and Pasangulapati et al. [26].

3. Results and discussion

3.2. TGA analysis of PKS
3.1. TGA analysis of cellulose, xylan and lignin
The results of TGA analysis during the PKS pyrolysis process
The pyrolysis characteristics, both the weight loss (thermo- taking heating rate of 20 °C min1 as an example are shown in
gravimetry curves, in units of mass%) and the rate of weight loss Fig. 2. The pyrolysis process consisted of the three stages: 139–
(derivative thermogravimetry curves, in units of mass%/°C), of cel- 323 °C, 323–389 °C and 390–762 °C. The two former stages were
lulose, xylan and lignin at the heating rate of 20 °C min1 are fast devolatilization stages during which a large proportion of
shown in Fig. 1. Obvious differences were found among the pyroly- weight of the sample lost with two significant mass-loss peaks
sis behaviors of the three model components. The weight loss of (shown in derivative thermogravimetry (DTG) curve), then the
xylan (representing hemicellulose) occurred early in the tempera- third stage was a slow degradation stage.
ture range of 100–365 °C, and the maximum weight loss focused at It is well known that ligno-cellulosic biomass is mainly com-
posed of hemicellulose, cellulose and lignin. In the first stage
(139–323 °C), the weight loss (30.52 mass%) was mainly attributed
to the degradation of the hemicellulose. Because hemicellulose is a
mixture of various polymerized monosaccharides (xylose, man-
nose, glucose, galactose, arabinose etc.) with lower degree of poly-
merization that the thermal stability was lower than cellulose. The
main temperature of hemicellulose degradation was from 185 to
325 °C (shown in Fig. 1). However, the weight loss of 30.52 mass%
was higher than the content of hemicellulose in PKS (23.82 mass%).
This might be caused by the higher content of lignin (45.59%) in
PKS, while only 10–25% for hard wood [27], 21.4% for switchgrass
[26]. In part 3.1, it had stated that lignin owned wider temperature
range of degradation at 100–800 °C. Thus, the extra weight loss
was caused by the degradation of part of lignin.
In the second stage (323–389 °C), a contiguous and/or simulta-
neous degradation process was presented mainly due to the com-
position of cellulose with weight loss of 29.92 mass%. Because
cellulose was a high-molecular compound with long linear chain
composed of D-glucosyl group [28], and part of cellulose has crys-
talline structure made of ordered microfibrils that resulted in ther-
mal degradation more difficulty than hemicellulose [25]. This

Fig. 1. TG (a) and DTG (b) curves of cellulose, xylan and lignin pyrolysis with a
heating rate of 20 °C min1. Fig. 2. TG/DTG curves of PKS pyrolysis with a heating rate of 20 °C min1.
254 Z. Ma et al. / Energy Conversion and Management 89 (2015) 251–259

Fig. 3. 3D FTIR analysis of PKS pyrolysis with a heating rate of 20 °C min1. (a) 3D FTIR diagram, (b) FTIR spectrum for volatiles with different temperatures at 281 and 357 °C,
and (c) evolution of volatile components with increasing temperature.

result was also confirmed by the single cellulose degradation TGA in the minute interspace between microfibrils of cellulose, and
experiment along with the temperature of 290–380 °C. covalently linked to hemicellulose and cross linked to polysaccha-
However, the two distinct weight loss peaks in DTG curve ride. Therefore higher content lignin made the thermal degradation
(shown in Fig. 2) of PKS significantly differentiated with other bio- of PKS slower than other biomass and led to mass-losses of hemi-
mass presented only one peak (e.g. pine wood [11], bamboo [14], cellulose and cellulose separate clearly. This result was also con-
corn stover [29]) or accompanied with an extra ‘‘shoulder’’ peak firmed by Asadullah et al. using PKS biomass [10]. Similar result
(e.g. wheat straw [30]). This also might be caused by the higher of two distinct mass-loss peaks were also found in hazelnut husk
content of lignin in PKS. Lignin is an amorphous substrate, exists with high content of lignin (39%) studied by Clean et al. [31]. Liu

Table 2
Typical FTIR Analysis of PKS pyrolysis.

Species Wavenumbers range (cm1) Functional groups Vibrations

H2O 4000–3400 OAH Stretching
CH4 3000–2700 CAH Stretching
CO2 2400–2250 C@O Stretching
CO 2250–2000 CAO Stretching
Aldehydes, ketones, acids 1900–1650 C@O Stretching
Aromatics 1690–1450 C@C, benzene skeleton Stretching
Alkanes, alcohols, phenols, ethers, lipids 1475–1000 CAO, CAC, carbon chain skeleton Stretching
CO2 586–726 C@O Bending
Details in the fingerprint region from 1475 to 1000 cm1
Alkanes 1460–1365 CAC, CAH Stretching
Alcohols 1200–1000 CAO Stretching
Phenols 1300–1200 CAO Stretching
Ethers 1275–1060 CAO Stretching
Lipids 1300–1050 CAO Stretching
 Z. Ma et al. / Energy Conversion and Management 89 (2015) 251–259 255

et al. [32] studied the interactions of cellulose, hemicellulose and Taking a temperature of 281 °C as an example, a detailed anal-
lignin. It was found that as the proportion of added lignin increased ysis is shown in Fig. 3(b). The typical volatile components present
from 0.33 to 3, the maximum weight loss rate of cellulose and were identified (see Table 2). A few small molecular gaseous com-
hemicellulose decreased. And their temperature range of degrada- ponents (e.g., H2O, CH4, CO2, and CO) were easily identified by their
tion became wider. This result also verified the conclusion that prominent characteristic bands [34]. CH4 came primarily from the
higher content lignin would make cellulose and hemicellulose decomposition of methoxy (AOCH3), methyl (ACH3), and methy-
pyrolysis more slowly at the same heating rate. lene (ACH2A) groups under high temperatures [35]. CO2 was
The third stage, namely slow degradation stage (390–762 °C), formed via the decarboxylation reaction and the breakage of car-
accounted for only a small fraction of the total weight loss bonyl groups [36]. The breakage of ether bonds and C@O bonds
(11.88 mass%). This stage might be ascribed to the lignin degrada-
tion based on the analysis of Fig. 1. And char was the main product
from lignin degradation during this stage [33].

3.3. FTIR analysis of PKS

Fig. 3(a) shows the 3D FTIR diagram of the PKS pyrolysis process
with a heating rate of 20 °C min1. The characteristic infrared
absorption peaks of the volatile components are clearly shown in
Fig. 3(b). The devolatilization in the first stage occurred mainly
between 10 and 20 min elapsed. The appearance of absorbance
peaks agreed well with the mass-loss in the DTG curve shown in
Fig. 2. Fig. 3(c) shows the evolution of absorbance intensity of
the volatile components, from which the temperatures associated
with the two peaks (281 and 357 °C) were evaluated. The three
substances corresponding to absorbance peaks at 2364, 1751,
and 1404 cm1 were the dominant volatile components, and the
identification of the components would be discussed in the next
paragraphs.

Fig. 4. TG (a) and DTG (b) curves of PKS pyrolysis with heating rates of 10, 20, 30 Fig. 5. Arrhenius plots of log(b) vs. 1/(T + 273.15) for PKS in different conversion
and 40 °C min1. rate ranges using FWO method: (a) 0.03–0.13, (b) 0.15–0.8, and (c) 0.81–0.92.
256 Z. Ma et al. / Energy Conversion and Management 89 (2015) 251–259

temperature was obtained and is shown in Fig. 3(c). According to

the Lambert–Beer law, absorbance intensity at a specific wave-
number is linearly dependent on relative concentration of volatile
components [35]. Therefore, the evolution of absorbance intensity
in the whole pyrolysis process represented the tendency of relative
concentrations of volatile components. In agreement with DTG
curve, each lumped component had two peaks. In the first stage
(139–322 °C), all volatile components appeared, but CO2, H2O,
aldehydes, ketones, and organic acids were the predominant
substances, mainly originating from hemicellulose. Although these
three substances were still dominant in the second stage, the
concentrations of gaseous CO2 (2400–2250 cm1), H2O, CH4, CO,
aldehydes, ketones, and organic acids increased remarkably. This
is because cellulose and little part of lignin began to devolatilize.
In the third stage (slow degradation stage), the relative concentra-
tion of all volatile components gradually decreased.

3.4. Kinetics analysis

3.4.1. The effect of heating rate

In order to use multi-heating rate method, a series of experi-
ments had to be performed at different heating rates (10, 20, 30
and 40 °C min1), and the TG/DTG curves were shown in Fig. 4.
Firstly, both points of maximum of mass loss rate in the TG and
DTG curves shifted toward higher temperatures. This could be
attributed to difference between the reference (i.e., furnace) tem-
perature and sample temperature due to the heat and mass trans-
fer limitations in the sample [20]. Further, the poor thermal
conductivity of biomass materials also caused temperature gradi-
ents in the sample particles. In other words, the temperature in
the core of particles can be lower than the temperature on their
surfaces [29,38]. Secondly, the maximum mass-loss rate in DTG
curves decreased as the heating rate increased. This was mainly
affected by mass transfer. Because it would take a longer time of
the heat conduction from the particle external to the interior as
the heating rate increased. So that, at the same temperature, the
pyrolysis in higher heating rate was less sufficient, especially for
the biomass containing high volatile component. This conclusion
was agreed well with Lah [13].

3.4.2. Analysis of the activation energy

Figs. 5 and 6 show the Arrhenius plots based on two model free
methods, namely the FWO method and KAS method, respectively.
The conversion rate is divided into three stages for further analysis,
0.03–0.13 (Figs. 5 and 6(a)), 0.15–0.8 (Figs. 5 and 6(b)), and 0.81–
0.92 (Figs. 5 and 6(c)). As shown in Fig. 5(a) and (c) using FWO
method, the very low values of correlation coefficients (R2) were
observed at a of 0.03, 0.05, 0.07, 0.89, 0.9, 0.92 that could not be
accepted to calculate activation energy, and also the same points
of a with bad correlation were found in Fig. 6(a) and (c) using
KAS method. Other points showed good correlation were used to
present the conversion rate dependence of the activation energy.
Fig. 6. Arrhenius plots of ln(b/(T + 273.15)2) vs. 1/(T + 273.15) for PKS in in different The distribution of the activation energy was presented in Fig. 7.
conversion rate ranges using KAS method: (a) 0.03–0.13, (b) 0.15–0.8, and (c) 0.81– The activation energy estimated from the FWO and KAS method
0.92.
showed excellent agreement with each other, and with only less
than 5% deviation. And the small deviation was resulted from dif-
ferent approximations used in the algorithms [39,40]. The consis-
likely formed CO [37]. Also a number of organic compounds were tency of results from both methods, and the measured TG curves
also detected at 1900–1000 cm1 including the fingerprint region. from multi-heating heating rate, validated the accuracy and reli-
They were aldehydes, ketones, organic acids, monocyclic aromat- ability of the estimated activation energy [41]. The value of activa-
ics, alkanes, alcohols, phenols, ethers and lipids ordered by the tion energy was affected by several factors, such as different
wavenumber [11,12]. Their characteristic peaks and function kinetics model, heating rate, species of biomass, particle size and
groups are clearly shown in Table 2. different types of TGA. Thus, the activation energy of PKS was only
After identifying the volatile components, the evolution of valid for this kind of experimental parameters mentioned in Sec-
absorbance intensity of volatile components with increasing tions 2.1 and 2.2.
 Z. Ma et al. / Energy Conversion and Management 89 (2015) 251–259 257

Fig. 7. Activation energy (E) distribution at different conversion rates (a) determinated from the FWO and KAS methods.

Before the discussion of the activation energy distribution 69.39 kJ mol1,179.84 kJ mol1, respectively. This comparison also
dependent on the conversion rate, it is worthy to note that activa- confirmed that part of lignin degradation in this area increased
tion energy represent the minimum energy requirement for a reac- the activation energy in this stage.
tion started, in other words, higher value of activation energy For the second area, namely 0.3 < a < 0.8, the activation energy
means slower reaction rate and more difficulty of a reaction start- initially increased slightly at 0.3 < a < 0.45, then decreased gradu-
ing. From an overall perspective of Fig. 7, the fluctuation of the acti- ally at 0.45 < a < 0.8.Compared to the first area, higher activation
vation energy on a dependence was observed both in FWO and KAS energy was observed at 0.3 < a < 0.45, namely 282 to 285 kJ mol1
method, and this revealed that the PKS pyrolysis process was a for the KAS method and 277 to 280 kJ mol1 for the FWO method.
complex reaction included parallel, competitive and consecutive According to the temperature corresponding to a of 0.3 to 0.45 in
reactions [8,20,41–43]. different heating rates, this area was just located between two
For the first area, 0.09 < a < 0.3, the activation energy increased mass-loss peak of DTG curves. Based on the analysis of the effect
from 231 to 270 kJ mol1 for the KAS method and 227 to of high content of lignin in the part of TGA analysis, higher content
261 kJ mol1 for the FWO method as the conversion rate increased, lignin made the cross-link of three main components in PKS tigh-
and this could be mainly attributed to the hemicellulose thermal ter. This indicated the competing degradation reactions occurred
degradation. Initially, the degradation started rather easily on the in the cross-linked polymer matrix, and also diffusion regime
weakly linked sites inherent to the polymeric lineal chain of the occurred [28]. Also in particular for cellulose in this area, based
hemicellulose which led to lower activation energy. Then, after on Broido–Shafizadeh kinetics model, cellulose initially pyrolyzed
the weaker bonds broke, random scission on the lineal chain to active cellulose, led to reduce the degree of polymerization
occurred that made the activation energy increased [28]. Mean- and the length of molecule chain. The active cellulose was a inter-
while, little part of lignin pyrolyzed at the sites where the functions mediate product before further pyrolysis [45,46]. Thus the above
such as methoxyl, hydroxy, carbonyl and carboxy weakly linked on analysis validated the increasing of activation energy from
three kinds of benzene–propane, i.e. guaiacyl, syringyl, p-hydroxyl- 0.09 < a < 0.3–0.3 < a < 0.45. For the area of 0.45 < a < 0.8, the acti-
phenyl, that were the basic structural unit of lignin [35]. Because of vation energy (234–266 kJ mol1 for the KAS method and 232–
the interaction of hemicellulose and lignin, it was easy to under- 262 kJ mol1 for the FWO method) decreased. This was attributed
stand that the activation energy in this stage was much higher than to the degradation of active cellulose with low molecule weight
the single component of xylan (representing the hemicellulose) in formed in the second area which required lower energy. Then
pyrolysis process listed in Table 3, i.e. 87.65 kJ mol1, active cellulose continued to degradation. Two paralleled and

Table 3
Activation energy of cellulose, xylan and lignin from literatures.

Cellulose Xylan Lignin Ref.

Tem. (°C) E (kJ mol1) A R2 Tem. (°C) E (kJ mol1) A R2 Tem. (°C) E (kJ mol1) A R2
270–390 119.21 6.9  109 0.95 200–260 116.84 5.4  1011 0.95 200–400 43.29 1.4  103 0.92 [26]
260–315 58.45 1.66  105 0.94 680–740 98.06 5.5  104 0.92
Mean 87.65 Mean 70.68
300–340 227.02 5.6  1016 0.99 220–300 69.39 2.1  103 0.99 220–380 7.8 4.9  105 0.99 [25]
380–530 8.2 1  104 0.99
750–900 54.77 3.4  102 0.99
Mean 31.39
/ 240.23 2.6  1018 / / 179.84 6.1  1014 / 165.61 8.5  1011 / [44]
275–390 150–175a / / / / / / / / / / [47]
a
The value of activation energy of cellulose was decreasing as the conversion rate increasing from 0.15 to 0.8.
258 Z. Ma et al. / Energy Conversion and Management 89 (2015) 251–259

competitive reactions occurred. One reaction was to produce char [2] Esfahani RM, Ghani WAWA, Salleh MAM, Ali S. Hydrogen-Rich gas production
from palm kernel shell by applying air gasification in fluidized bed reactor.
favored at lower temperature, the other was produce tar and gas
Energy Fuels 2012;26:1185–91.
which prior to occur at the higher temperature [42,43,45]. The acti- [3] Kim SJ, Jung SH, Kim JS. Fast pyrolysis of palm kernel shells: influence of
vation energy in this stage was much higher than the single/pure operation parameters on the bio-oil yield and the yield of phenol and phenolic
component of cellulose listed in Table 3 (i.e. 119.21 kJ mol1 [26], compounds. Bioresour Technol 2010;101:9294–300.
[4] Grilc M, Likozar B, Levec J. Hydrodeoxygenation and hydrocracking of
227.02 kJ mol1 [25], 240.23 kJ mol1 [44], 150–175 kJ mol1 solvolysed lignocellulosic biomass by oxide, reduced and sulphide form of
[47], respectively). The interaction of cellulose and lignin in PKS, NiMo, Ni, Mo and Pd catalysts. Appl Catal B-Environ 2014;150:275–87.
namely the inherent link of their molecular structures, led to this [5] Grilc M, Likozar B, Levec J. Hydrotreatment of solvolytically liquefied
lignocellulosic biomass over NiMo/Al2O3 catalyst: reaction mechanism,
result. hydrodeoxygenation kinetics and mass transfer model based on FTIR.
For the final area, 0.8 < a < 0.87, highest activation energy (275– Biomass Bioenergy 2014;63:300–12.
545 kJ mol1 for the KAS method and 272–529 kJ mol1 for the [6] Chen DY, Zheng Y, Zhu XF. Determination of effective moisture diffusivity and
drying kinetics for poplar sawdust by thermogravimetric analysis under
FWO method) observed within the whole process, and the activa- isothermal condition. Bioresour Technol 2012;107:451–5.
tion energy fast increased as the temperature increased. This was [7] Ma ZQ, Zhang YM, Zhang QS, Qu YB, Zhou JB, Qin HF. Design and experimental
related to the degradation of lignin that had been stated in the part investigation of a 190 kW(e) biomass fixed bed gasification and polygeneration
pilot plant using a double air stage downdraft approach. Energy
of TGA analysis based on the TG/DTG curves. Because lignin was 2012;46:140–7.
mainly composed of three kinds of benzene–propane, which was [8] Narobe M, Golob J, Klinar D, Francetic V, Likozar B. Co-gasification of biomass
heavily cross-linked, and made it own highest thermal stability and plastics: pyrolysis kinetics studies, experiments on 100 kW dual fluidized
bed pilot plant and development of thermodynamic equilibrium model and
[25,41]. And the previous analysis had stated that weak-linked
balances. Bioresour Technol 2014;162:21–9.
sites on the three-dimensional network structure of lignin had bro- [9] Lee Y, Park J, Ryu C, Gang KS, Yang W, Park YK, et al. Comparison of biochar
ken in the lower temperature range. However, in the range of high properties from biomass residues produced by slow pyrolysis at 500 degrees C.
temperature, part of the three-dimensional network structure Bioresour Technol 2013;148:196–201.
[10] Asadullah M, Ab Rasid NS, Kadir SASA, Azdarpour A. Production and detailed
broke which required highest energy. And a large number of coke characterization of bio-oil from fast pyrolysis of palm kernel shell. Biomass
with low reaction activity was formed in this area which also led to Bioenergy 2013;59:316–24.
the activation energy increased fast [30]. This could also be used to [11] Gu XL, Ma X, Li LX, Liu C, Cheng KH, Li ZZ. Pyrolysis of poplar wood sawdust by
TGA–FTIR and Py–GC/MS. J Anal Appl Pyrol 2013;102:16–23.
explain the result that the activation energy in this stage was much [12] Gao NB, Li AM, Quan C, Du L, Duan Y. TGA–FTIR and Py–GC/MS analysis on
higher than the pure alkali lignin (70.68 kJ mol1, 31.39 kJ mol1, pyrolysis and combustion of pine sawdust. J Anal Appl Pyrol 2013;100:26–32.
165.61 kJ mol1 respectively, shown in Table 3). [13] Lah B, Klinar D, Likozar B. Pyrolysis of natural, butadiene, styrene–butadiene
rubber and tyre components: modelling kinetics and transport phenomena at
In general, the fluctuation of activation energy in the range of different heating rates and formulations. Chem Eng Sci 2013;87:1–13.
0.1 < a < 0.8 was similar to the hazelnut husk with high content [14] Jiang ZH, Liu ZJ, Fei BH, Cai ZY, Yu Y, Liu XE. The pyrolysis characteristics of
of lignin using KAS and FWO methods [10]. And the activation moso bamboo. J Anal Appl Pyrol 2012;94:48–52.
[15] Zhao H, Yan HX, Dong SS, Zhang Y, Sun BB, Zhang CW, et al. Thermogravimetry
energy rapidly increased at the range a > 0.8 was similar with study of the pyrolytic characteristics and kinetics of macro-algae Macrocystis
wheat straw [30]. In a word, the interaction of biomass compo- pyrifera residue. J Therm Anal Calorim 2013;111:1685–90.
nents with different content and structure would finally led to var- [16] Agrawal A, Chakraborty S. A kinetic study of pyrolysis and combustion of
microalgae Chlorella vulgaris using thermo-gravimetric analysis. Bioresour
iation of distribution of activation energy.
Technol 2013;128:72–80.
[17] Wongsiriamnuay T, Tippayawong N. Thermogravimetric analysis of giant
sensitive plants under air atmosphere. Bioresour Technol 2010;101:9314–20.
4. Conclusions
[18] Aboulkas A, El Harfi K, El Bouadili A. Non-isothermal kinetic studies on co-
processing of olive residue and polypropylene. Energy Convers Manager
The difference in the molecular weight and structure of three 2008;49:3666–71.
fundamental chemical compositions, namely hemicellulose, cellu- [19] Çepelioğullar Ö, Pütün AE. Thermal and kinetic behaviors of biomass and
plastic wastes in co-pyrolysis. Energy Convers Manager 2013;75:263–70.
lose and lignin, resulted in different pyrolysis characteristics. Their [20] Vyazovkin S, Burnham AK, Criado JM, Perez-Maqueda LA, Popescu C,
interaction led to the fluctuation of activation energy in PKS pyro- Sbirrazzuoli N. ICTAC Kinetics Committee recommendations for performing
lysis. (1) A higher content of lignin in PKS led to a slower degrada- kinetic computations on thermal analysis data. Thermochim Acta
2011;520:1–19.
tion thermal decomposition observed as two significant mass-loss [21] Flynn JH, Wall AL. A quick direct method for determination of activation
peaks in DTG curve that differentiated with other biomasses (e.g. energy from thermogravimetric data. J Polym Sci B 1966;4:323–8.
pine wood, bamboo, corn stover and wheat straw). (2) The fluctu- [22] Ozawa T. A new method of analyzing thermogravimetric data. Bull Chem Soc
Jpn 1965;38:1881–6.
ation of activation energy estimated by FWO and KAS methods can [23] Kissinger HE. Reaction kinetics in differential thermal analysis. Anal Chem
be interpreted as follows. First, the degradation of hemicellulose 1957;29:1702–6.
occurred at 0.09 < a < 0.3. Then cellulose initially transformed to [24] Akahira T, Sunose T. Method of determining activation deterioration constant
of electrical insulating materials. Rep. Res. Inst. Chiba Inst. Technol. (Sci.
active cellulose, soon active cellulose begun to degrade within
Technol.) 1971;16:22–31.
0.3 < a < 0.8. At last, network structure of lignin broke with the [25] Yang HP, Yan R, Chin T, Liang DT, Chen HP, Zheng CG. Thermogravimetric
highest activation energy at 0.8 < a < 0.87. (3) The prominent analysis–fourier transform infrared analysis of palm oil waste pyrolysis.
Energy Fuels 2004;18:1814–21.
volatile components generated by the pyrolysis of PKS were CO2,
[26] Pasangulapati V, Kumar A, Jones CL, Huhnke RL. Characterization of
aldehydes, ketones, organic acids, alkanes, and phenols. switchgrass, cellulose, hemicellulose and lignin for thermochemical
conversions. J Biobased Mater Bioenergy 2012;6(3):249–58.
[27] Vassilev SV, Baxter D, Andersen LK, Vassileva CG, Morgan TJ. An overview of
Acknowledgements
the organic and inorganic phase composition of biomass. Fuel
2012;94(1):1–33.
This research was supported by the Research Foundation of [28] Lopez-Velazquez MA, Santes V, Balmaseda J, Torres-Garcia E. Pyrolysis of
orange waste: a thermo-kinetic study. J Anal Appl Pyrol 2013;99:170–7.
Talented Scholars of Zhejiang A & F University and the ‘‘Twelfth
[29] Kumar A, Wang LJ, Dzenis YA, Jones DD, Hanna MA. Thermogravimetric
Five Years’’ National Science and Technology Program (NO. characterization of corn stover as gasification and pyrolysis feedstock. Biomass
2012BAD30BOO). Bioenergy 2008;32:460–7.
[30] Chen DY, Zheng Y, Zhu XF. In-depth investigation on the pyrolysis kinetics of
raw biomass. Part I: kinetic analysis for the drying and devolatilization stages.
References Bioresour Technol 2013;131:40–6.
[31] Celan S, Topcu Y. Pyrolysis kinetics of hazelnut husk using thermogravimetric
[1] Ma LL, Wang TJ, Liu QY, Zhang XH, Ma WC, Zhang Q. A review of thermal– analysis. Bioresour Technol 2014;156:182–8.
chemical conversion of lignocellulosic biomass in China. Biotechnol Adv [32] Liu QA, Zhong ZP, Wang SR, Luo ZY. Interactions of biomass components
2012;30:859–73. during pyrolysis: A TG–FTIR study. J Anal Appl Pyrol 2011;90(2):213–8.
 Z. Ma et al. / Energy Conversion and Management 89 (2015) 251–259 259

[33] Amutio M, Lopez G, Aguado R, Artetxe M, Bilbao J, Olazar M. Kinetic study of [40] Ounas A, Aboulkas A, El Harfi K, Bacaoui A, Yaacoubi A. Pyrolysis of olive
lignocellulosic biomass oxidative pyrolysis. Fuel 2012;95:305–11. residue and sugar cane bagasse: Non-isothermal thermogravimetric kinetic
[34] Ren XY, Wang WL, Bai TT, Si H, Chang JM, Tian HX. TGA–FTIR study of the analysis. Bioresour Technol 2011;102:11234–8.
thermal-conversion properties of holo-cellulose derived from woody biomass. [41] Vamvuka D, Kakaras E, Kastanaki E, Grammelis P. Pyrolysis characteristics and
Spectrosc Spect Anal 2013;33:2392–7. kinetics of biomass residuals mixtures with lignite. Fuel 2003;82:1949–60.
[35] Liu Q, Wang SR, Zheng Y, Luo ZY, Cen KF. Mechanism study of wood lignin [42] Koufopanos CA, Maschio G, Lucchesi A. Kinetic modelling of the pyrolysis of
pyrolysis by using TGA–FTIR analysis. J Anal Appl Pyrol 2008;82:170–7. biomass and biomass components. Can J Chem Eng 1989;67:75–84.
[36] Fu P, Hu S, Xiang J, Sun LS, Su S, An SM. Study on the gas evolution and char [43] Anca-Couce A, Berger A, Zobel N. How to determine consistent biomass
structural change during pyrolysis of cotton stalk. J Anal Appl Pyrol pyrolysis kinetics in a parallel reaction scheme. Fuel 2014;123:230–40.
2012;97:130–6. [44] Zhang J, Chen T, Wu J, Wu J. A novel Gaussian-DAEM-reaction model for the
[37] Granada E, Eguia P, Vilan JA, Comesana JA, Comesana R. FTIR quantitative pyrolysis of cellulose, hemicellulose and lignin. RSC Adv. 2014;4:17513–20.
analysis technique for gases. Application in a biomass thermochemical [45] Broido A, Nelson MA. Char yield on pyrolysis of cellulose. Combust Flame
process. Renew Energy 2012;41:416–21. 1975;24:263–8.
[38] Lopez-Gonzalez D, Fernandez-Lopez M, Valverde JL, Sanchez-Silva L. [46] Bradbury AGW, Sakai Y, Shafizadeh F. A kinetic model for pyrolysis of
Thermogravimetric-mass spectrometric analysis on combustion of cellulose. J Appl Polym Sci 1979;23:3271–80.
lignocellulosic biomass. Bioresour Technol 2013;143:562–74. [47] Suriapparao DV, Ojha DK, Ray T, Vinu R. Kinetic analysis of co-pyrolysis of
[39] Vyazovkin S. Modification of the integral isoconversional method to account cellulose and polypropylene. J Therm Anal Calorim 2014;117:1441–51.
for variation in the activation energy. J Comput Chem 2001;22:178–83.