Algal Research 35 (2018) 168–177

Contents lists available at ScienceDirect

Algal Research
journal homepage: www.elsevier.com/locate/algal

Application of the distributed activation energy model to the kinetic study of T

pyrolysis of Nannochloropsis oculata
⁎
Daniel Viju, Ribhu Gautam, R. Vinu
Department of Chemical Engineering and National Center for Combustion Research and Development, Indian Institute of Technology Madras, Chennai 600036, India

A R T I C LE I N FO A B S T R A C T

Keywords: Pyrolysis of algae is a promising route to produce high quality bio-oil and renewable chemicals. Owing to its
Microalgae complex structural composition, multiple pseudo-components are required to describe its thermal decomposition
Nannochloropsis oculata in a wide temperature range and evaluate the reaction kinetics. In this study, the pyrolysis behavior of the
Distributed activation energy model microalga, Nannochloropsis oculata (N. oculata), was studied by means of a thermogravimetric analyzer at various
Kinetics
heating rates. A four-parallel-reaction scheme characterizing the pyrolysis of carbohydrate, protein, lipid and the
Pyrolysis
TGA
secondary decomposition of char was employed to model thermal degradation using distributed activation
Py-GC/MS energy model (DAEM). The average and standard deviation of activation energy, pre-exponential factor, and
composition of the model components for pyrolysis of N. oculata were estimated. The model mass loss and
differential mass loss profiles matched well with the experimental data at different heating rates. Based on the
model predictions, the decomposition of proteins, carbohydrates, lipids and char occurred in the temperature
regimes of 200–450 °C, 200–300 °C, 400–500 °C, and 750–900 °C, respectively. To gain valuable insights on the
pyrolysate composition at various temperature regimes, analytical pyrolysis-gas chromatography/mass spec-
trometry experiments were performed. Indole and phenol, aliphatic and aromatic hydrocarbons, and long chain
oxygenates were observed as the major pyrolysates in the temperature regimes of 30–350 °C, 350–600 °C and
600–1000 °C, respectively.

1. Introduction [6]. For effective utilization of microalgae via thermochemical tech-

A thorough understanding of the physico-chemical properties and
chemistry of transformation of renewable and carbon neutral species
like microalgae are inevitable for their efficient utilization for biofuels
production, and design of efficient systems for processing them [1,2]. In
recent years, Nannochloropsis sp. has been proposed as an excellent
candidate for biofuel production due to their ability to generate high
quantity of lipids [2]. The bio-chemical composition of algae is fun-
damentally different from terrestrial biomass. Cellulose, hemicellulose,
and lignin constitute lignocellulosic biomass, while microalgae are
made up of proteins, lipids, carbohydrates and nucleic acids [3]. A wide
variety of products are possible from the conversion of microalgae, as
lipids, after extraction, can be converted to biodiesel, and carbohy-
drates, via fermentation, can be converted to bio-ethanol [4].
Pyrolysis, which involves heating the feedstock at moderate tem-
peratures (400–600 °C) in the absence of oxygen, has gained significant
attention, as this thermochemical process converts the whole of mi-
croalgae into liquid bio-oil [5]. Microalgae pyrolysis results in bio-oil
with better properties than that derived from lignocellulosic biomass
nologies, an understanding of their thermal behavior and pyrolysis ki-
netics is necessary. This understanding can be applied to reactor design
and to decide the optimum operating parameters for better process
efficiency [6]. The inherent composition of the microalgae leads to
lower decomposition temperatures than lignocellulosic biomass during
thermochemical conversion [7]. Global kinetic schemes are used to
explain the mass volatilization and to capture the overall yields of bio-
oil, gas, and char from pyrolysis. Investigating the intrinsic (or me-
chanistic) kinetics of algae pyrolysis is complicated due to its complex
composition. Recently, Ranzi and co-workers developed a multi-step
semi-detailed kinetic mechanism of algae pyrolysis using different re-
ference compounds, and validated the model with the experimental
data obtained under dynamic and isothermal slow and fast pyrolysis
conditions [8].
Thermogravimetric analysis (TGA) is extensively used to study the
thermal behavior and evaluation of apparent kinetic parameters of
decomposition. TGA precisely gives the mass loss data as a function of
time and temperature at low heating rates (5–100 °C min−1). TG data
obtained at single or multiple heating rates are used to calculate the

⁎
Corresponding author.
E-mail address: vinu@iitm.ac.in (R. Vinu).

https://doi.org/10.1016/j.algal.2018.08.026
Received 21 June 2018; Received in revised form 28 July 2018; Accepted 19 August 2018
2211-9264/ © 2018 Elsevier B.V. All rights reserved.
D. Viju et al.
Nomenclature
A pre-exponential (or frequency) factor (s−1)
ci mass fraction of the i-th pseudo component
E activation energy (J mol−1)
E0 mean activation energy of the Gaussian distribution
(J mol−1)
f(E) activation energy distribution (mol J−1)
nd number of data points used
R universal gas constant (J mol−1 K−1)
apparent kinetic parameters via integral or differential isoconversional
methods and n-th order model fitting techniques [9]. TG data is useful
not only to determine the kinetic parameters, but also to understand the
overall pyrolysis mechanisms involved [10]. Owing to the complexity
of microalgae pyrolysis and the involvement of multiple parallel reac-
tions, it is difficult to describe the entire process with a single-step re-
action kinetic model. Kim et al. [11] utilized a differential TG and
evolved gas analysis–mass spectrometric technique to identify the de-
composition regimes of carbohydrates (< 300 °C), proteins and lipids
(301–380 °C) and secondary cracking of proteins and lipids (> 380 °C)
during pyrolysis of Porphyra tenera. In another study, apparent activa-
tion energies for pyrolysis of Chlorella sp. (~300 kJ mol−1) and Tetra-
selmis suecica (~100 kJ mol−1) were evaluated using TGA employing
isoconversional methods, and found to be lower than lignocellulosic
biomasses [12].
Vo et al. [13] described Aurantiochytrium sp. pyrolysis using a
lumped model involving three reactions, which included algae →
bio–oil → gases, and algae → gases. The apparent activation energy of
bio-oil formation was found to be 71.16 kJ mol−1. Bach and Chen [14]
utilized one, two, three, four and seven-reaction models, and found that
the seven-reaction model described well the pyrolytic mass loss profiles
of Chlorella vulgaris. Bui et al. [15] used five pseudo–component me-
chanism to model the pyrolysis of Chlamydomonas sp. and -Chlorella
sorokiniana, and determined the apparent kinetic parameters. A multi-
step model involving devolatilization and oxidation to describe the
pyrolysis of Nannochloropsis gaditana, Scenedesmus almeriensis and
Chlorella vulgaris is reported to fit the pyrolysis mass loss data [16]. In a
recent study by Gautam et al. [17], first order and one-dimensional
diffusion models matched the experimental data obtained under iso-
thermal fast pyrolysis conditions in the temperature range 400–700 °C
for Nannochloropsis oculata, Chlorella vulgaris, Arthrospira platensis and
Schizochytrium limacinum. The models employed in these studies are
based on a simplified assumption that each microalgae component
decomposes with a single activation energy. Nevertheless, the complex
structure of the components requires a range of activation energies to
describe the dissociation of various types of bonds present in proteins,
carbohydrates and lipids.
Distributed activation energy model (DAEM) is a multiple reaction
model, which is widely used to describe the decomposition behavior of
Table 1
Characterization of N. oculata used in this study. The data are presented as the mean ± S.D., n = 3.
Proximate analysisa Ultimate analysisa
VM (%) FC (%) A (%) C (%) H (%) N (%)
64.9 ± 2.4 30.8 ± 1.3 4.3 ± 0.2 35.6 ± 1.2 6.7 ± 0.1 5.2 ± 0.2
VM = volatile matter, FC = fixed carbon, A = ash. All percentages are wt%.
a
% dry basis
b
By difference, O(%) = 100 − C − H − N − S − ash.
c
Algal bio-chemical composition from http://www.oilgae.com/algae/comp/comp.html (accessed on 9 Jun 2018).
d
Reed Mariculture Inc. reedmariculture.com/support_algae_nutritional_proximate_profile.php (accessed on 9 Jun 2018).
e
Dry ash free basis.
Algal Research 35 (2018) 168–177
S objective function for differential thermogravimetric data
(K−2)
S1 objective function for conversion data
T temperature (K)
T0 initial temperature (K)
α degree of conversion
β linear heating rate (K min−1 or °C min−1)
σ standard deviation of the activation energy distribution
(J mol−1)
solid and liquid fuels [18]. The model assumes that decomposition
occurs via large number of independent, parallel, first-order reactions
with different activation energies reflecting the varying bond strength
of the species [19]. Models based on three parallel first-order reactions
corresponding to the decomposition of hemicellulose, cellulose and
lignin have been successfully used to describe the pyrolysis of lig-
nocellulosic biomass [20]. The DAEM has been successfully extended
for the kinetic study of pyrolysis of the freshwater alga, Chlorococcum
humicola [21]. Ceylan and Kazan [22] investigated the pyrolysis be-
havior and kinetics of Nannochloropsis oculata and Tetraselmis sp. The
devolatilization behavior was explained using a simplified DAEM ap-
proach. In a recent study, Simao et al. [23] used multiple models such
as Miura-Maki DAEM, independent parallel reaction model and iso-
conversional kinetic models of Friedman and Flynn-Wall-Ozawa to
study the pyrolysis behavior of Spirulina maxima. The use of zeolites
was found to enhance the production of aromatic hydrocarbons by five
times. Soria-Verdugo et al. [24] comprehensively applied and com-
pared isoconversional models and DAEM to study the pyrolysis beha-
vior of six different microalgae species, viz., Chlorella vulgaris, Isochrysis
galbana, Nannochloropsis gaditana, Nannochloropsis limnetica, Phaeo-
dactylum tricornutum, and Spirulina platensis.
In this study, a four-parallel-reaction DAEM is developed to describe
the pyrolysis kinetics of Nannochloropsis oculata. As the fundamental
composition of microalgae is different from lignocellulosic biomass, TG
mass loss experiments of bovine serum albumin, starch and sunflower
oil, representing protein, carbohydrate and lipid, respectively, were
first conducted to map the decomposition of the pseudo-components to
the primary constituents in the algae. This study is unique as the de-
composition regimes of the pseudo-components described by the model
were validated using the experimental composition of the pyrolysis
products obtained at different temperature regimes. More importantly,
the high temperature decomposition regime at 800–1000 °C observed
for this microalga was assigned to the fourth pseudo-component, which
is nothing but secondary decomposition of char, based on both model
and experimental product composition data. The decomposition re-
gimes were then deconvoluted and the kinetic parameters were ob-
tained for the pseudo-components.
Nutritional analysisc,d,e HHV (MJ kg−1)
S (%) Ob (%) Proteins (%) Lipids (%) Carbohydrates (%)
0.5 ± 0.06 47.7 ± 2.0 69 21 10 18.0 ± 0.5
169
D. Viju et al. Algal Research 35 (2018) 168–177

2. Material and methods
2.1. Characterization of raw materials
Nannochloropsis oculata (N. oculata) was procured from Reed
Mariculture Inc., U.S.A., in the form of fine powder. Bovine serum al-
bumin, starch, and sunflower oil were procured from SRL Chemicals
Pvt. Ltd. India, Loba Chemie Pvt. Ltd., India, and a local store, re-
spectively. Table 1 depicts the salient characteristics of the microalga.
The proximate analysis of the microalga was carried out in a multiple
sample TGA (TGA-2000A, Navas Instruments, U.S.A.) according to
ASTM E-1131-08 standard procedure [25]. CHNS content of the mi-
croalga was determined using Elementar Vario Micro CHNS analyzer
(Elementar, Germany) by taking c.a. 3 mg of sample. The HHV (higher
heating value) of the microalga was determined in an isoperibolic bomb
calorimeter (IKA C2000, IKA, Germany). Typically 0.5 g of the sample
was used for this analysis.
2.2. Thermogravimetric analysis (TGA)
Thermal stability of N. oculata was studied using a TGA (SDT Q600,
T.A. Instruments, U.S.A.). Approximately 5 mg of the samples were
heated to 1000 °C from ambient temperature at heating rates of 5, 10
and 20 °C min−1 in the presence of nitrogen supplied at a flow rate of
200 mL min−1. Small sample sizes taken for this analysis ensured re-
action controlled regime for thermochemical conversion with minimum
heat and mass transport effects [26]. The mass loss and derivative mass
loss data were acquired to understand the decomposition pattern of the
alga at high temperatures. The pyrolysis of the model compounds, viz.
bovine serum albumin, starch, and sunflower oil, was also carried out at
a heating rate of 5 °C min−1. The TGA experiments were repeated in
triplicate, and the variation in the temperature corresponding to the
maximum mass loss rate in any temperature regime was within ± 1 °C.
Non-isothermal TGA offers certain advantages over the isothermal
method as it eliminates the error induced by the thermal induction
period from the start temperature to the end temperature [27]. It also
permits a rapid scan of the entire temperature range of interest.
Moreover, the kinetic parameters can be determined by using data from
limited number of experiments in case of non-isothermal kinetics as
compared to isothermal experiments [28].
2.3. Analytical pyrolysis–Gas Chromatography/Mass Spectrometry
helium gas (99.9995%) at a flow rate of 0.8 mL min−1 to the GC column
(HP-5 MS (30 m × 0.2 mm × 0.25 μm) capillary column). The GC/MS
interface and transfer line were maintained at 300 °C and 280 °C, re-
spectively. The injector and ion source were maintained at 280 and
300 °C, respectively. The oven was programmed to initially hold at
35 °C for 4 min, followed by a temperature ramp to 280 °C at a rate of
10 °C min−1, and finally held at 280 °C for 5 min. The separated pyr-
olysates were scanned in the range of 50–400 Da at an electron ioni-
zation voltage of 70 eV. GC/MS total ion chromatograms were analyzed
and the pyrolysates were matched with the NIST MS library. The
compounds with high match factor (> 85%) were considered for the
composition analysis. The relative peak area% was used to represent
relative composition or selectivity of various compounds present in the
pyrolysates. The experiments were repeated three times non-con-
secutively, and the average data are reported along with standard de-
viation.
2.4. Distributed activation energy model
Global kinetic schemes explain the mass devolatilization and cap-
ture the overall yields of char and volatiles [29]. Generally, the global
model assumes that the pyrolysis of biomass occurs through a single
reaction. This assumption is far too ideal because biomass is made up of
different components, which exhibit distinct behavior. In contrast,
DAEM assumes that the decomposition of the organic material occurs
via large number of independent and irreversible reactions, each with
its own activation energy. Further, it is assumed that all the reactions
belonging to a specific family share the same pre-exponential factor.
Thus, the reactivity is represented by a continuous distribution of ac-
tivation energies [30]. Generally, the activation energy is assumed to
follow a Gaussian distribution because of the success with this type of
modeling [31].
The DAEM equation is given by Eq. (1) [18]. The conversion is
normalized with respect to (100–wt% of char), so that it varies from
zero to 100%. The reaction is assumed to obey first-order reaction ki-
netics.
T
⎡ ⎤
∫ Aβ exp ⎛⎝ −RTE ⎞⎠ dT⎥ f (E ) dE
∞
α (T ) = 1 − ∫0 exp ⎢−
⎢
⎣ T0 ⎥
⎦ (1)
dα
By differentiating Eq. (1) with respect to T, the expression of is
dT
obtained as follows:

In order to ascertain the nature of organic compounds evolved at d (α ) ∞ A⎡ E T A E

= ∫0 − − ∫T exp ⎛− ⎞ dT⎤ f (E ) dE
different temperature regimes, analytical pyrolysis of the alga coupled dT β⎢
⎣ RT 0 β ⎝ RT ⎠ ⎥ ⎦ (2)
with gas chromatograph/mass spectrometric (GC/MS) analysis of the
As stated earlier, the popular choice for the activation energy dis-
products was performed. The experiments were conducted in an ana-
tribution is the Gaussian distribution centered at mean activation en-
lytical pyrolyzer Pyroprobe® 5200 (C.D.S. Analytical, U.S.A.), and the
ergy (E0) with standard deviation (σ).
pyrolysates were separated, identified and quantified using a GC/MS
(Agilent 7890/5975C, Agilent Technologies, U.S.A.). The three dif- 1 −(E − E0 )2 ⎞
ferent temperature regimes were 30–350 °C, 350–600 °C and f (E ) = exp ⎛ ⎜ ⎟

σ 2π ⎝ 2σ 2 ⎠ (3)
600–1000 °C. 2.0 ± 0.05 mg of the microalga was weighed in a high
accuracy microbalance (CUBIS Series, Sartorius AG, Germany) with an It is important to note that, the inner integrals of Eqs. (1) and (2) do
accuracy of 1 μg. The sample was placed in a cylindrical quartz tube not have exact analytical solutions. The following expression is widely
and blocked from both the sides using glass wool to avoid any spillage used to approximate the inner integral [32],
due to swelling of the sample during pyrolysis. The quartz tube con- T −E RT 2 E (0.99954E + 0.58058RT )
taining the sample was placed inside a resistively heated platinum coil g (T ) = ∫0 exp ⎛
⎝ RT
⎞ dT =
⎠ E
exp ⎛− ⎞
⎝ RT ⎠ (E + 2.54RT )
probe for pyrolysis. The pyrolysis experiments were conducted at a
heating rate of 5 °C min−1 in different temperature regimes (30–350 °C, (4)
350–600 °C and 600–1000 °C). The experiments were conducted in trap In general, the conversion rate curve (i.e., dα/dT vs T curve) of the
mode, where the pyrolysates from the reactor were adsorbed in a pyrolysis of biomass shows a peak and a peak shoulder to the left side of
Tenax®-containing cartridge at 45 °C. The pyrolysates were then des- the peak [33]. However, according to the numerical results of the
orbed by ramping the temperature of the cartridge at a heating rate of DAEM process, the conversion rate curves of the DAEM with certain
100 °C min−1 to 300 °C. The pyrolysates were swept by ultra-high pure activation energy distribution and pre-exponential factor can exhibit

170
D. Viju et al. Algal Research 35 (2018) 168–177

one single peak and no peak shoulder [34]. Comprehensive DAEM optimization techniques. Therefore, a robust and derivative-free direct
consisting of two or three parallel reactions are proposed to describe the pattern search algorithm is required [36]. The pattern search function
overall decomposition kinetics of lignocellulosic biomass [35]. A si- of MATLAB® was used to estimate the parameters. The model para-
milar approach is used here to describe the pyrolysis kinetics of N. meters estimated were used to generate a derivative weight loss curve,
oculata as it is a complex material with bio-chemically different species. which was compared with the experimental data using the fitness
The composition of the alga is considered as a sum of several pseudo- parameter. The fitness parameter is given by Eq. (8).
components, and the pyrolysis of each pseudo-component, is described
S
by first order Gaussian DAEM. ⎛ nd
⎞
Fit (%)DTG = ⎜1 − ⎟. 100%
d (α )
=
n
∑ ci ∫0
∞ A⎡ E
β⎢
− − ∫T0
T A
exp ⎛−
E
⎞ dT⎤
1
exp ⎛
− (E − E 0 ) 2 ⎞
⎜ dE
⎟
⎜
⎝
( )
dα
dT m
⎟
⎠ (8)
dT i=1 ⎣ RT β ⎝ RT ⎠ ⎥ ⎦ σ 2π ⎝ 2σ 2 ⎠
(5) The estimation of the best-fit parameters was performed by mea-
suring the average deviation between the experimental and calculated
n T
⎡ ⎤ 2 derivative thermogravimetric (DTG) curves, and expressing it as a
∫ Aβ exp ⎛⎝ −RTE ⎞⎠ dT⎥ σ 12π exp ⎛ −(E2−σ 2E0 )
∞
α (T ) = ∑ 1 − ∫0 exp ⎢− ⎞ dE dα
( )
⎜ ⎟

i=1 ⎢ T0 ⎥ ⎝ ⎠ percentage of the maximum experimental value, dT , as shown in Eq.

⎣ ⎦ m
(8) [10]. Importantly, similar objective function (S1) and fitness para-
(6)
meter (Fit(%)conv.) were used for the conversion, and are given by Eqs.
In the above expression, the composition factor (ci) for each of the (9) and (10), respectively. It is to be noted that the maximum value of
pseudo-component represents the amount of volatiles produced by the conversion is unity, which eliminates the need for scaling.
ith pseudo component. It is necessary to normalize the composition nd
n
factor to 1, such that ∑i = 1 ci , as the conversion, α, is defined to account
for the volatile fraction from the model compounds.
S1 = ∑ [(α )i,exp − (α )i,cal ]2
i=1 (9)

2.5. Numerical method for parameter estimation S1 ⎞

Fit (%)Conv = ⎜⎛1 − ⎟ . 100%

⎝ nd ⎠ (10)
Using Eq. (5), an objective function was formulated, which was
optimized to estimate the parameters. The objective function is defined
as the difference between the experimental data and the data calculated 3. Results and discussion
from the model given by Eq. (7). The subscript i refers to the data points
used, nd is the number of data points used, dT
dα
and dT
dα
( )
re-
i, exp
( )i, cal
3.1. Thermogravimetric Analysis of N. oculata
present the experimental and calculated values, respectively, for a given
set of parameters. Fig. 1 presents the mass loss curve in the form of fraction of volatiles
evolved (α vs T) and the mass loss rate in the derivative form ( dα vs T)
2 dT
nd
dα dα of the microalga during pyrolysis in the temperature range of
S= ∑ ⎡⎢ ⎛ dT ⎞ −⎛ ⎞ ⎤
i=1 ⎣⎝ ⎠i, exp ⎝ dT ⎠i, cal ⎥
⎦ (7)
150–1000 °C. There are three regimes of decomposition, as evidenced
from the DTG curve of the microalga. In the first stage of decomposition
There are 3n unknown parameters to be evaluated (E0,i, Ai, σi for (150–350 °C) a peak is observed at 256 °C, and to the right of this peak,
i = 1:n). Since the objective function has no explicit solution, the op- a shoulder is noticed. The shoulder indicates the decomposition of two
timization problem is difficult to be solved using conventional components occurring in the same temperature range. Based on the

Fig. 1. (a) Thermogravimetric (TG) and (b) differential thermogravimetric (DTG) curve of N. oculata at 5 °C min−1.

171
D. Viju et al.
study conducted by Maddi et al. [37], this can be attributed to the
possible overlap between thermal degradation of proteins and carbo-
hydrates.
The second degradation stage occurs between 350 and 600 °C with a
peak at 440 °C. Considering the previous studies reporting the TGA of
microalgae, a peak was observed in the same region for lipids extracted
from C. vulgaris. So, the peak observed with N. oculata in this region can
be attributed to the decomposition of the lipids present in the microalga
[38]. Nearly 69% of the volatiles are evolved in the temperature range
of 150–700 °C, which are predominantly due to the nutritional com-
ponents in the algae undergoing depolymerization, decarboxylation
and cracking reactions [39]. There is, however, a final high tempera-
ture decomposition stage present, which cannot be attributed to any of
the organic components in the microalga. Algae typically accumulate
high quantities of metals such as Na, Mg, Ca, and often Si, during their
culture. Yainik et al. [40] attributed this to the decomposition of in-
organic compounds like carbonates and silicates present in the ash. As
the amount of ash in N. oculata is only 4.3 wt%, this last decomposition
regime can be attributed to secondary decomposition of char as re-
ported in literature [41]. The formation of a peak in this high tem-
perature region is, thus, due to the secondary decomposition of char
catalyzed by ash. In order to confirm that this peak is not observed due
to water washable impurities, the alga was washed in distilled water,
and dried at 100 °C for 2 h to remove the physisorbed moisture.
Thereafter, TGA of this washed and dried microalga was conducted
under identical conditions. The peak in this high temperature zone was
still observed. This proves that this last decomposition step can be at-
tributed, at least partially, to the secondary decomposition of char.
3.2. TGA of the model compounds
In order to propose the reactions of pseudo-components for the
DAEM, and to experimentally verify the regimes of decomposition of
the pseudo-components, TGA of the model compounds was performed.
The model compounds were bovine serum albumin for protein, sun-
flower oil for lipid, and starch for carbohydrate. Fig. 2 depicts the DTG
curves of bovine serum albumin, sunflower oil, and starch. The de-
composition of bovine serum albumin starts at about 200 °C, and de-
volatilization occurred till 500 °C. The mass loss rate was observed to be
highest at 291 °C. Generally, proteins and polypeptides are reported to
degrade in the similar temperature range [37]. Lysozyme, also a pro-
tein, is reported to degrade in 200–500 °C range [37]. Although the
172
Algal Research 35 (2018) 168–177
model protein used here is not derived from N. oculata, this analysis is
helpful to understand the decomposition regime characteristic of pro-
teins.
Sunflower oil is a mixture of monounsaturated, polyunsaturated,
and saturated fatty acids. The thermal decomposition profile for sun-
flower oil shows three distinct steps in the DTG. The first step
(230–380 °C) is attributed to the decomposition of polyunsaturated
fatty acids, the second step (380–480 °C) corresponds to the decom-
position of monounsaturated fatty acids, and the third step
(480–550 °C) corresponds to the thermal decomposition of saturated
fatty acids [42]. Starch decomposes in a relatively broad temperature
range (250–500 °C). Studies focusing on the pyrolysis of other carbo-
hydrates also reported a similar decomposition behavior. The thermal
behavior of starch is similar to the decomposition of alginic acid [39].
The TGA of the model components, thus, provides the typical decom-
position regimes of the bio-chemical components of the microalga.
These temperature regimes can be used in the DAEM to predict the
pyrolysis behavior of the pseudo-components.
3.3. Kinetic study of N. oculata pyrolysis
This section focuses on the determination of kinetic parameters of
pyrolysis of N. oculata. It is evident from the previous discussion that
the organic components in the algae decompose in a wide temperature
range from 150 to 700 °C, and the total volatiles contributed by these
components is around 69%. The rest of the volatiles are evolved during
the secondary decomposition of char in the temperature range of
750–1000 °C. Using the nutritional analysis of the alga in Table 1, the
composition of pseudo-components are calculated. The percentages of
protein, carbohydrate, and lipid-derived fractions in the volatiles are
47%, 7%, and 15%, respectively. As the framework for the model is
firmly established, 12 unknown parameters (E0,i, Ai, σi for i = 1:4)
corresponding to the four pseudo-components have to be estimated.
Using the DTG data obtained at 5 and 10 °C min−1, the parameters were
determined, and with the help of parameters obtained, the pyrolysis
behavior at 20 °C min−1 was predicted, and compared with experi-
mental data. Table 2 provides the optimal set of parameters that best
describe the experimental data.
As seen in Figs. 3, 4 and 5, the quality of the fit obtained in all the
cases is above 90%, which indicates the suitability of the four pseudo-
component assumption. It is important to note that increasing the
number of pseudo-components will increase the uncertainty of the
Fig. 2. Differential thermogravimetric (DTG) curves ( dα vs
dT
T) of bovine serum albumin, sunflower oil, and starch at
5 °C min−1.
D. Viju et al.
Table 2
The kinetic parameters of decomposition of the pseudo-components obtained
from optimization.
Parameter Protein Carbohydrate Lipid Char
log10(A) 7.17 8.9 8.16 8.13
E0 (kJ mol−1) 106.0 115.0 146.0 247.0
σ (kJ mol−1) 16.10 2.12 0.32 5.15
cj 0.47 0.07 0.15 0.31
model owing to the many fitted parameters. In this regard, the choice of
pseudo-components used in this study is optimal. Moreover, they are
based on the biochemical composition of the alga and the experimen-
tally validated high temperature decomposition step. As the parameters
are highly sensitive to the initial guess values, there is a need to com-
pare the consistency of the parameters obtained with the ones pre-
sented in the literature. The pre-exponential factors for the pyrolytic
decomposition of the model components are observed in the range of
107–109 s−1. The activation energy for the nutritional fraction of the
microalga lies between 106 and 146 kJ mol−1. These values are in
agreement with the reported mean apparent activation energy and pre-
exponential factor for N. oculata estimated by a simplified DAEM [22].
According to the report of a kinetic study on pyrolysis of residues ob-
tained after the oil extraction from Chlamydomonas and Chlorella sor-
okiniana, the calculated activation energies using a five pseudo-com-
ponent model for protein and lipid were found to be
90.74–99.31 kJ mol−1 and 104.93–131.97 kJ mol−1, respectively [15].
This is comparable with the parameters obtained in the present study.
Based on a recent study conducted by Bach and Chen [14], the com-
bined activation energy of decomposition of carbohydrates and proteins
estimated by the two reaction DAEM for Chlorella vulgaris is reported to
be 135.83 kJ mol−1, which is in close agreement with the values esti-
mated in this study. The variations in the parameters with respect to the
literature can be attributed to the multiplicity of kinetic parameters,
which is explained by the kinetic compensation effect (KCE) [9,17,18].
KCE states that a linear relationship exists between activation energy
and the natural logarithm of pre-exponential factor of the form ln
(A) = m × E + n [9]. By knowing the values of the parameters, m and
Fig. 3. Experimental and predicted differential thermogravimetric (DTG) and conversion curves at a heating rate of 5 °C min−1. (The residuals between the ex-
periments and model predictions of the DTG are plotted).
173
Algal Research 35 (2018) 168–177
n, the pre-exponential factor can be determined when the apparent
activation energy is known, and vice versa.
Based on the model, the decomposition regimes of pseudo-compo-
nents, viz., proteins, carbohydrates, and lipids are 200–450 °C,
200–300 °C and 400–500 °C, respectively. As reported by Kebelmann
et al. [38], proteins extracted from green microalga such as Chlorella
vulgaris exhibited weight loss in the range of 200–500 °C. Muniz et al.
[43] pointed out that lipids extracted from Bertholletia excelsa H.B.K.
almond degraded in the temperature range of 270–580 °C. Secondary
decomposition of char is found to occur from 750 to 950 °C. The ash
present in N. oculata essentially contains inorganic metals, such as Na,
Mg, Ca, Al, Si and Cl [2]. This is expected to catalyze the secondary
decomposition of char, which results in different products being
formed. The pyrolysis products formed in the different temperature
regimes are determined using Py-GC/MS analysis, which is discussed in
the following section.
3.4. Pyrolysis-Gas Chromatography/Mass Spectrometry of N. oculata
Fig. 6 depicts the product distribution from the pyrolysis of N.
oculata carried out at three regimes, viz., 30–350 °C, 350–600 °C and
600–1000 °C. The compounds were classified based on the functional
groups, viz., low molecular weight oxygenates, phenolics, N-containing
compounds, aliphatic hydrocarbons and benzene derivatives. The ra-
tionale behind choosing 350 and 600 °C as the intermediate points are
due to the fact that most of the carbohydrates along with the protein
fraction decompose by 350 °C, and lipids by 600 °C. The selectivities of
the major compounds observed in the above three temperature zones
are shown in Tables 3, 4 and 5. The total ion chromatograms obtained
from the Py-GC/MS in these three temperature regimes also reflect the
distinct nature of the products in these three zones (Fig. 7).
As microalgae are fundamentally made up of lipids, carbohydrates
and proteins, each of the components in the pyrolysate can be related to
the bio-chemical family. The composition of pyrolysates varied greatly
with the temperature regime. Nearly 35% of the products collected in
the first temperature zone comprised of N-containing compounds, such
as nitriles, amines, indole, and pyrroles. The decomposition of proteins
in this regime is explained by the evolution of nitrogenous compounds
D. Viju et al. Algal Research 35 (2018) 168–177

Fig. 4. Experimental and predicted differential thermogravimetric (DTG) and conversion curves at a heating rate of 10 °C min−1. (The residuals between the
experiments and model predictions of the DTG are plotted).

[44]. The evolution of aromatic hydrocarbons, such as toluene and
styrene, can possibly be due to the decarboxylation, deamination and
concerted rupturing of CeC bonds during the thermal decomposition of
amino acids present in proteins. There is a possibility of amino acid first
undergoing individual decomposition by scission of the polymer chain
and cleavage of the side chains. Kebelmann et al. [38], stated that (i)
pyrrole is derived from the amino acids, such as hydroxyproline and
glutamine, (ii) phenol and cresol are formed from tyrosine, and (iii)
toluene and styrene are derived from phenylalanine. Protein decom-
position is ascertained to be the source of compounds like indole and
aniline in the pyrolysates [17,44]. The products observed in this tem-
perature regime suggest significant decomposition of protein fraction
present in the microalga. It is clear from the nutritional analysis of N.
oculata that carbohydrates constitute a mere 10 wt%. Therefore, the
compounds evolved from the decomposition of carbohydrate fraction
were not observed with significant selectivity. Nevertheless, the pre-
sence of cyclic oxygenates such as 2-cyclopenten-1-one and isophorone
suggests the decomposition of carbohydrates in this regime. It is to be
noted that cyclic oxygenates are observed only in this temperature re-
gime making this argument stronger. Cyclic oxygenates are formed via

Fig. 5. Experimental and predicted differential thermogravimetric (DTG) and conversion curves at a heating rate of 20 °C min−1. (The residuals between the
experiments and model predictions of the DTG are plotted).

174
D. Viju et al. Algal Research 35 (2018) 168–177

Table 5
Major products obtained from the pyrolysis of N. oculata in the temperature
range of 600–1000 °C. The data are presented as the mean ± S.D., n = 3.
Functional groups Compounds Selectivity %

Oxygenates 2-Heptadecanone 23.7 ± 6.4

Phytol 13.5 ± 4.5
2-Nonadecanone 2.8 ± 1.1
Phytol acetate 0.8 ± 0.37
Hydrocarbons Heptadecane 2.85 ± 1.1
1-Nonadecene 1.63 ± 0.4
Heneicosane 1.47 ± 0.35
Polyaromatics Naphthalene 0.23 ± 0.05
Biphenyl 0.40 ± 0.04
Acids Tetradecanoic acid 1.3 ± 0.02
Benzene derivatives Toluene 0.5 ± 0.18
Nitrogen compounds Phenylpropanamide 2.86 ± 0.33
Heptadecanenitrile 1.0 ± 0.15

In the temperature range of 350–600 °C, the selectivity of hydro-

carbons was high (60%). This is attributed to the decomposition of lipid
in the microalgae. Interestingly, fast pyrolysis of N. oculata at 500 °C
Fig. 6. Composition of pyrolysates obtained from the pyrolysis of N. oculata
also resulted in aliphatic and aromatic hydrocarbons to the tune of 51%
carried out at heating rate of 5 °C min−1 at three temperature regimes. The data
are presented as the mean ± S.D., n = 3.
selectivity owing to the catalytic effects of inorganic minerals present in
the ash [46]. The fatty acids that constitute lipids undergo decarbox-
ylation and CeC cracking to form products such as alkenes. The carbon
Table 3 number of major hydrocarbons observed was in the range of C10–C17.
Major products obtained from the pyrolysis of N. oculata in the temperature
The presence of long chain fatty amides and nitriles can be attributed to
range of 30–350 °C. The data are presented as the mean ± S.D., n = 3.
the interaction of proteins and lipids in the vapour phase. Ammonia
Functional groups Compounds Selectivity % evolved from protein pyrolysis reacts with fatty acids to form long chain
fatty amides under pyrolysis conditions [17]. These fatty amides then
Nitrogen compounds Indole 11.7 ± 1.1
Benzenepropanenitrile 5.1 ± 0.4
undergo dehydration to form long chain nitriles such as pentadecani-
1-H-pyrrole,3-methyl 2.0 ± 0.6 trile (C15). Gautam et al. [17] reported the formation of amides and
Ethane-1,2-diamino 0.23 ± 0.05 nitriles as a result of deamination of proteins, ammoniation of car-
Pentadecanenitrile 0.9 ± 0.3 boxylic acids and dehydration of amides. As the pyrolysis of proteins is
Butanamide 0.3 ± 0.1
expected to occur until 450 °C, minor amounts of benzene derivatives
Benzene derivatives Toluene 4.7 ± 0.24
Styrene 1.96 ± 1.2 and N-containing compounds are also observed in this regime. The
Ethylbenzene 1.3 ± 0.4 presence of phytol and its derivatives in this temperature regime can be
Benzene 0.2 ± 0.07 attributed to the thermal decomposition of terpenoid chain of chlor-
Benzaldehyde 0.3 ± 0.17
ophyll [38].
Phenolics Phenol 6.25 ± 2.2
p-Cresol 1.74 ± 0.57
The last regime from 600 to 1000 °C is mainly composed of long-
Phenol,4-ethyl 3.26 ± 0.52 chain oxygenates, such as 2-heptadecanone (C17) and 2-nonadecanone
Cyclic oxygenates Isophorone 0.63 ± 0.3 (C19), which constitute nearly 26% of the total pyrolysates obtained in
1,2-Cyclopentanediol,trans- 0.33 ± 0.1 this regime. The selectivity of compounds in the carbon number range
2-Cyclopenten-1-one, 3-methyl- 0.3 ± 0.16
of C14-C22 was observed to be around 47% with C17–C21 hydro-
carbons, fatty acids and carbonyl compounds in majority. The dec-
arboxylation of fatty acids resulted in the formation of long chain hy-
Table 4
Major products obtained from the pyrolysis of N. oculata in the temperature drocarbons. Lipids under fast pyrolysis conditions exhibited a delay in
range of 350–600 °C. The data are presented as the mean ± S.D., n = 3. product evolution when compared to proteins and carbohydrates [17].
The formation of these long chain compounds in the pyrolysates can be
Functional groups Compounds Selectivity %
due possibly to the thermal degradation of the lipid fraction left in the
Hydrocarbons Pentadecane 6.0 ± 0.6 solid residue. Anand et al. [47] reported the formation of 2-heptade-
Hexadecane 4.6 ± 0.12 canone from fast pyrolysis of Schizochytrium limacinum, a lipid-rich alga,
Tetradecane 4.1 ± 0.2 via ketonization reaction in the presence of MgO catalyst. The presence
Heptadecane 3.4 ± 0.16
of tetradecanoic acid in this temperature regime ascertained this ar-
1-Tridecene 2.2 ± 0.18
Dodecane 2.4 ± 0.24 gument. Metal content in the ash could have catalyzed the secondary
1-Decene 1.3 ± 0.3 decomposition of lipid residues in char, and led to the formation of long
Cetene 2.45 ± 0.08 chain oxygenates. A similar phenomenon of the catalytic activity of
Benzene derivatives Toluene 4.5 ± 0.4 char owing to the metal contents in it was reported by Ronsse et al.
Styrene 4.56 ± 0.3
Ethylbenzene 1.2 ± 0.20
[41]. Moreover, slower heating rates facilitate the secondary cracking
Nitrogen compounds Pentadecanenitrile 3.45 ± 0.94 of pyrolysates, repolymerization reactions and condensation of smaller
Oxygenates Phytol 2.9 ± 1.0 molecules. Aromatic compounds such as biphenyl and naphthalene are
also observed in this region as high pyrolysis temperatures favour the
formation of these hydrocarbons [8].
successive dehydration steps during the pyrolysis of carbohydrates. To summarize, this study has demonstrated the applicability of four
These observations are in line with Zhou et al. [45], who proposed the pseudo-component DAEM to understand different decomposition zones
mechanisms for the formation of similar oxygenates from cellulose. during pyrolysis of N. oculata. More importantly, the fourth pseudo-

175
D. Viju et al.
Fig. 7. Total ion chromatograms of pyrolysates obtained in various temperature regimes from the pyrolysis of N. oculata at 5 °C min−1.
component, i.e., char, and its secondary decomposition at high tem-
peratures is validated by Py-GC/MS product distribution. The modeling
framework developed in this study can be applied to other complex
microalgae that exhibit different decomposition regimes. More im-
portantly, the high temperature decomposition behavior of the micro-
alga unravelled by this study will be valuable in understanding gasifi-
cation of algal residues, where secondary decomposition of char is a key
reaction for the formation of syngas.
4. Conclusions
This study has showcased the need of a fourth pseudo-component
like char, in addition to proteins, carbohydrates and lipids, to describe
the mass loss observed at high temperatures during microalgae pyr-
olysis. The pyrolysis of model compounds, viz. bovine serum albumin,
starch and sunflower oil, were performed to validate the decomposition
regimes of protein, carbohydrate and lipid, respectively, in the micro-
alga. The mass loss and differential mass loss data predicted by the
DAEM matched well with experimental data. The apparent activation
energies (kJ mol−1) for the decomposition of proteins, carbohydrates,
lipids and char were 106 ± 16.1, 115 ± 2.12, 146 ± 0.32 and
247 ± 5.15, respectively, and the pre-exponential factors varied in the
range of 107–109 s−1. Py-GC/MS experiments revealed the identity of
the pyrolysates in different temperature regimes. Nitrogen-containing
compounds like indoles, nitriles and pyrroles, and benzene derivatives
dominated the product spectrum in the low temperature regime
(30–350 °C), owing, primarily, to the decomposition of proteins. The
medium temperature regime (350–600 °C) was characterized by the
176
Algal Research 35 (2018) 168–177
production of long chain aliphatic and aromatic hydrocarbons from
cracking of lipids, while the high temperature regime (600–1000 °C)
was characterized by the formation of long chain oxygenates as a result
of possible condensation reactions catalyzed by the char.
Contribution of authors
All three authors took part in conceptualizing the study, performing
experiments, analysis and interpretation of data, manuscript prepara-
tion, and revision after review.
Authors' agreement
The authors agree to the presented authorship, and to the submis-
sion of the manuscript for peer review in Algal Research.
Statement of informed consent, human/animal rights
No conflicts, informed consent, human or animal rights applicable.
Conflict of interest
The authors declare no competing financial interest.
Acknowledgements
The National Center for Combustion Research and Development
(NCCRD) is funded by Department of Science and Technology, India.
D. Viju et al.
References
[1] A.B. Ross, J.M. Jones, M.L. Kubacki, T. Bridgeman, Classification of macroalgae as
fuel and its thermochemical behaviour, Bioresour. Technol. 99 (2008) 6494–6504.
[2] Sukarni, Sudjito, N. Hamidi, U. Yanuhar, I.N.G. Wardana, Potential and properties
of marine microalgae Nannochloropsis oculata as biomass fuel feedstock, Int. J.
Energy Environ. Eng. 5 (2014) 279–290.
[3] A. Pandey, D.J. Lee, Y. Chisti, C. Soccol, Biofuels from Algae, Elsevier B.V., India,
2014.
[4] Q. Al-Abdallah, B.T. Nixon, J.R. Fortwendel, The enzymatic conversion of major
algal and cyanobacterial carbohydrates to bioethanol, Front. Energy Res. 4 (2016)
1–15.
[5] T. Suganya, M. Varman, H.H. Masjuki, S. Renganathan, Macroalgae and microalgae
as a potential source for commercial applications along with biofuels production: a
biorefinery approach, Renew. Sust. Energ. Rev. 55 (2016) 909–941.
[6] S.W. Kim, B.S. Koo, D.H. Lee, A comparative study of bio-oils from pyrolysis of
microalgae and oil seed waste in a fluidized bed, Bioresour. Technol. 162 (2014)
96–102.
[7] D. López-González, M. Fernandez-Lopez, J.L. Valverde, L. Sanchez-Silva, Kinetic
analysis and thermal characterization of the microalgae combustion process by
thermal analysis coupled to mass spectrometry, Appl. Energy 114 (2014) 227–237.
[8] P.E.A. Debiagi, M. Trinchera, A. Frassoldati, T. Faravelli, R. Vinu, E. Ranzi, Algae
characterization and multistep pyrolysis mechanism, J. Anal. Appl. Pyrol. 128
(2017) 423–436.
[9] S. Vyazovkin, A.K. Burnham, J. Criado, L. Pérez-Maqueda, C. Popescu,
N. Sbirrazzuoli, ICTAC kinetics committee recommendations for performing kinetic
computations on thermal analysis data, Thermochim. Acta 520 (2011) 1–19.
[10] C. Branca, C. Di Blasi, Global kinetics of wood char devolatilization and combus-
tion, Energy Fuel 17 (2003) 1609–1615.
[11] Y. Kim, T.U. Han, B. Lee, A. Watanabe, N. Teramae, J. Kim, Y. Park, H. Park, S. Kim,
Analytical pyrolysis reaction characteristics of Porphyra tenera, Algal Res. 32 (2018)
60–69.
[12] M.A. Kassim, K. Kirtania, D. De La Cruz, N. Cura, S.C. Srivatsa, S. Bhattacharya,
Thermogravimetric analysis and kinetic characterization of lipid-extracted
Tetraselmis suecica and Chlorella sp, Algal Res. 6 (2014) 39–45.
[13] T.K. Vo, H.V. Ly, O.K. Lee, E.Y. Lee, C.H. Kim, J. Seo, J. Kim, S. Kim, Pyrolysis
characteristics and kinetics of microalgal Aurantiochytrium sp. KRS101, Energy 118
(2017) 369–376.
[14] Q.V. Bach, W.H. Chen, A comprehensive study on pyrolysis kinetics of microalgal
biomass, Energy Conv. Manag. 131 (2017) 109–116.
[15] H.H. Bui, K.Q. Tran, W.H. Chen, Pyrolysis of microalgae residues–a kinetic study,
Bioresour. Technol. 199 (2016) 362–366.
[16] D. López-González, M. Fernandez-Lopez, J.L. Valverde, L. Sanchez-Silva, Pyrolysis
of three different types of microalgae: kinetic and evolved gas analysis, Energy 73
(2014) 33–43.
[17] R. Gautam, A.K. Varma, R. Vinu, Apparent kinetics of fast pyrolysis of four different
microalgae and product analyses using pyrolysis-FTIR and pyrolysis-GC/MS, Energy
Fuel 31 (2017) 12339–12349.
[18] J.M. Cai, W. Wu, R.H. Liu, An overview of distributed activation energy model and
its application in the pyrolysis of lignocellulosic biomass, Renew. Sust. Energ. Rev.
36 (2014) 236–246.
[19] J.M. Cai, T. Li, R.H. Liu, A critical study of the Miura–Maki integral method for the
estimation of the kinetic parameters of the distributed activation energy model,
Bioresour. Technol. 102 (2011) 3894–3899.
[20] J.M. Cai, W. Wu, R.H. Liu, G.W. Huber, A distributed activation energy model for
the pyrolysis of lignocellulosic biomass, Green Chem. 15 (2013) 1331–1340.
[21] K. Kirtania, S. Bhattacharya, Application of the distributed activation energy model
to the kinetic study of pyrolysis of the freshwater algae Chlorococcum humicola,
Bioresour. Technol. 107 (2012) 476–481.
[22] S. Ceylan, D. Kazan, Pyrolysis kinetics and thermal characteristics of microalgae
Nannochloropsis oculata and Tetraselmis sp, Bioresour. Technol. 187 (2015) 1–5.
[23] B.L. Simao, J.A. Santana Junior, B.M.E. Chagas, C.R. Cardoso, C.H. Ataide, Pyrolysis
of Spirulina maxima: kinetic modelling and selectivity for aromatic hydrocarbons,
177
Algal Research 35 (2018) 168–177
Algal Res. 32 (2018) 221–232.
[24] A. Soria-Verdugo, E. Goos, N. Garcia-Hernando, U. Riedel, Analyzing the pyrolysis
kinetics of several microalgae species by various differential and integral iso-
conversional kinetic methods and the distributed activation energy model, Algal
Res. 32 (2018) 11–29.
[25] ASTM E1131-08, Standard test method for compositional analysis by thermo-
gravimetry, https://www.astm.org/standards/E1131.htm , Accessed date: July
2018.
[26] Y.C. Lin, J. Cho, G.A. Thompsett, P.R. Westmoreland, G.W. Huber, Kinetics and
mechanism of cellulose pyrolysis, J. Phys. Chem. C 113 (2009) 20097–20107.
[27] Y.J. Xia, H.Q. Xue, H.Y. Wang, Z.P. Li, C.H. Fang, Kinetics of isothermal and non-
isothermal pyrolysis of oil shale, Oil Shale 28 (2011) 415–424.
[28] F. Ma, Y. Zeng, J. Wang, Y. Yang, X. Yang, X. Zhang, Thermogravimetric study and
kinetic analysis of fungal pretreated corn stover using the distributed activation
energy model, Bioresour. Technol. 128 (2013) 417–422.
[29] K.Q. Tran, Q.V. Bach, T.T. Trinh, G. Seisenbaeva, Non-isothermal pyrolysis of tor-
refied stump – a comparative kinetic evaluation, Appl. Energy 136 (2014) 759–766.
[30] A.K. Burnham, R.L. Braun, Global kinetic analysis of complex materials, Energy
Fuel 13 (1999) 1–22.
[31] G. Várhegyi, B. Bobaly, E. Jakab, H. Chen, Thermogravimetric study of biomass
pyrolysis kinetics: a distributed activation energy model with prediction tests,
Energy Fuel 25 (2011) 24–32.
[32] J.M. Cai, R.H. Liu, New approximation for the general temperature integral, J.
Therm. Anal. Calorim. 90 (2007) 469–474.
[33] J.E. White, W.J. Catallo, B.L. Legendre, Biomass pyrolysis kinetics: a comparative
critical review with relevant agricultural residue case studies, J. Anal. Appl. Pyrol.
91 (2011) 1–33.
[34] M. Günes, S.K. Günes, The influences of various parameters on the numerical so-
lution of non-isothermal DAEM equation, Thermochim. Acta 336 (1999) 93–96.
[35] G. Várhegyi, H.G. Chen, S. Godoy, Thermal decomposition of wheat, oat, barley and
Brassica carinata straws. A kinetic study, Energy Fuels 23 (2009) 646–652.
[36] E.D. Dolan, R.M. Lewis, V. Torczon, On the local convergence of pattern search,
SIAM J. Optim. 14 (2003) 567–583.
[37] B. Maddi, S. Viamajala, S. Varanasi, Comparative study of pyrolysis of algal biomass
from natural lake blooms with lignocellulosic biomass, Bioresour. Technol. 102
(2011) 11018–11026.
[38] K. Kebelmann, A. Hornung, U. Karsten, G. Griffiths, Intermediate pyrolysis and
product identification by TGA and Py-GC/MS of green microalgae and their ex-
tracted protein and lipid components, Biomass Bioenergy 49 (2013) 38–48.
[39] W. Peng, Q. Wu, P. Tu, Pyrolytic characteristics of heterotrophic Chlorella proto-
thecoides for renewable bio-fuel production, J. Appl. Phycol. 13 (2015) 5–12.
[40] J. Yainik, R. Stahl, N. Troeger, A. Sinag, Pyrolysis of algal biomass, J. Anal. Appl.
Pyrol. 103 (2013) 134–141.
[41] F. Ronsse, X. Bai, W. Prins, R.C. Brown, Secondary reactions of levoglucosan and
char in the fast pyrolysis of cellulose, Environ. Prog. Sustain. Energy 31 (2012)
256–260.
[42] A.G. Souza, J.C.O. Santos, M.M. Conceicao, M.C.D. Silva, S. Prasad, A thermo-
analytic and kinetic study of sunflower oil, Braz. J. Chem. Eng. 21 (2004) 265–273.
[43] M.A.P. Muniz, M.N.F. dos Santos, C.E. da Costa, L. Morais, M.L.N. Lamarão,
R.M. Ribeiro-Costa, J.O. Silva-Júnior, Physicochemical characterization, fatty acid
composition, and thermal analysis of Bertholletia excelsa HBK oil, Pharmacogn. Mag.
11 (2015) 147–158.
[44] K. Wang, R.C. Brown, Catalytic pyrolysis of microalgae for production of aromatics
and ammonia, Green Chem. 15 (2013) 675–681.
[45] X. Zhou, M.W. Nolte, H.B. Mayes, B.H. Shanks, L.J. Broadbelt, Experimental and
mechanistic modelling of fast pyrolysis of neat glucose-based carbohydrates. 1.
Experiments and development of a detailed mechanistic model, Ind. Eng. Chem.
Res. 53 (2014) 13274–13289.
[46] R. Gautam, R. Vinu, Non-catalytic fast pyrolysis and catalytic fast pyrolysis of
Nannochloropsis oculata using Co-Mo/γ-Al2O3 catalyst for valuable chemicals, Algal
Res. 34 (2018) 12–24.
[47] V. Anand, R. Gautam, R. Vinu, Non-catalytic and catalytic fast pyrolysis of
Schizochytrium limacinum microalga, Fuel 205 (2017) 1–10.