Bioresource Technology 175 (2015) 569–577

Contents lists available at ScienceDirect

Bioresource Technology journal homepage: www.elsevier.com/locate/biortech

Apparent kinetics of high temperature oxidative decomposition of microalgal biomass Saad Aldin M. Ali, Shaikh A. Razzak, Mohammad M. Hossain ⇑ Department of Chemical Engineering, King Fahd University of Petroleum & Minerals, Dhahran 31261, Saudi Arabia

h i g h l i g h t s

g r a p h i c a l a b s t r a c t

 Thermal degradation characteristics

of two different algae biomass are investigated.  Each microalgae strains exhibit different thermal behavior and characteristics.  The apparent kinetic parameters are determined by fitting the experimental data.  The value of activation energy dependents on the species and culture conditions.

a r t i c l e

i n f o

Article history: Received 22 September 2014 Received in revised form 18 October 2014 Accepted 20 October 2014 Available online 28 October 2014 Keywords: Thermal characteristics Microalgae biomass Oxidation Kinetics modeling

a b s t r a c t The oxidative thermal characteristics of two microalgae species biomass Nannochloropsis oculta and Chlorella vulgaris have been investigated. The apparent kinetic parameters for the microalgal biomass oxidation process are estimated by fitting the experimental data to the nth order rate model. Also, the iso-conversional methods Kissinger–Akahira–Sunose (KAS) and Flynn–Wall–Ozawa (FWO) were used to evaluate the apparent activation energy. The results indicate that biomass of different microalgae strains exhibit different thermal behavior and characteristics. In addition, growth parameters and medium composition can affect the biomass productivity and composition. This would have significant impact on the thermal decomposition trend of the biomass. The kinetic modeling of the oxidation reaction with direct model fitting method shows good prediction to the experimental data. The apparent activation energies estimated by KAS and FWO methods for N. oculta were 149.2 and 151.8 kJ/mol, respectively, while for C. vulgaris were 214.4 and 213.4 kJ/mol, respectively. Ó 2014 Elsevier Ltd. All rights reserved.

1. Introduction Biomass is one of the attractive forms of renewable energy resources. It can effectively contribute to the sustainable energy for the future developments and industrialization (López et al., 2013). Conventionally biomass is used for direct heating and/or energy generation purposes (Amutio et al., 2012). It can also be further processed to produce more versatile fuels such as biodiesel ⇑ Corresponding author. Tel.: +966 13 860 1478; fax: +966 13 860 4234. E-mail address: [email protected] (M.M. Hossain). http://dx.doi.org/10.1016/j.biortech.2014.10.109 0960-8524/Ó 2014 Elsevier Ltd. All rights reserved.

and hydrogen via syngas (Razzak et al., 2013). Generally, agricultural wastes, municipal wastes and forests are considered as the major sources of biomass. On the other hand, microalgae biomass has many unique features represent by its rapid growth rate as compared to the other terrestrial plants. Microalgae culture can also be employed in CO2 mitigation given its high carbon dioxide uptake efficiency (López-González et al., 2014). Unlike terrestrial crops, which requires vast lands, considerable amount of water and fertilizer for growth, microalgae culture can be cultivated using wastewater as a free source of nutrients (Kebelmann et al., 2013). In this way, microalgae culture can contribute to CO2

570

S.A.M. Ali et al. / Bioresource Technology 175 (2015) 569–577

utilization, wastewater treatment and source of bioenergy (Razzak et al., 2013). Concerning the microalgae biomass processing, there are different thermal technologies one can consider. The most common techniques are pyrolysis, gasification and direct combustion. The pyrolysis is an oxygen free process while both gasification and direct combustion requires oxygen (Sanchez-Silva et al., 2013). Pyrolysis and gasification produce valuable biofuel products, while direct combustion generates heat energy. The direct biomass combustion can be a standalone process or mixed with coal for electricity generations (López et al., 2013; Tahmasebi et al., 2013; Vamvuka et al., 2003). Firing of biomass mixed with coal is considered as an effective and low cost process (Tahmasebi et al., 2013). Thus, microalgae culture integrated with power generation can significantly contribute to the global efforts on CO2 emission control (Razzak et al., 2013). In order to employ the microalgae based biomass for direct energy applications, it is very important to understand their thermal behavior, apparent kinetics and heating values. Thermogravimetric analysis (TGA) allows studying the decomposition mechanism and burning characteristics of solid substances in a controlled environment. In open literature, there are studies available on the pyrolysis and kinetic of different biomass, including woods, corn, plant leaves and stems. Surprisingly, few studies reported the thermal characteristics of microalgae biomass (Rizzo et al., 2013). Most of the articles considered processing of microalgae to produce biodiesel (Razzak et al., 2013). On the other hand, the thermal analysis of microalgal biomass allows to determine the decomposition temperature, which helps in thermal drying and subsequent processing such as high temperature lipid extraction (Razzak et al., 2013). Peng et al. (2001) investigated the pyrolytic characteristics of Spirulina platenis and Chlorella protothecoides algae species under nitrogen atmosphere using different heating rates. They found three distinguish phases (dehydration, devolatilization and decomposition) in the thermal decomposition of algal biomass. Peng et al. (2001) used a linearized fitting technique (Freeman–Carroll method) for kinetic modeling and estimation of the activation energy. Agrawal and Chakraborty (2013) conducted thermogravimetric analysis to study the pyrolytic and thermal oxidative kinetics of Chlorella vulgaris microalgae strain. The TGA analysis showed three distinct stages during the thermal decomposition of microalgae biomass. The combustion of biomass in air atmosphere showed higher conversion than the conversion in biomass pyrolysis. Sanchez-Silva et al. (2013) studied the effects of particle and sample sizes on the thermal behavior of Nannochloropsis gaditana microalgal biomass. The samples were heated up to 1000 °C in the presence of air and analyzed the product gas using mass spectroscopy. Under the studied conditions, carbon monoxide, carbon dioxide and water were the main products and mostly found at the second degradation stage. Batista et al. (2013) examined the thermal resistances of microalgae biomass during the biomass processing such as in drying and cooking. The algae species (C. vulgaris and Haematococcus pluvialis) containing higher fat/carotenoid and low protein showed higher resistances in thermal decomposition (Batista et al., 2013). A recent work by López-González et al. (2014) also reported CO, CO2 and H2O as main products besides other compounds like light hydrocarbons, nitrogen compounds an sulfur compounds. There are some studies reported the influence of oxygen concentration to minimize the carbon monoxide formation (Chen et al., 2013, 2011; Tahmasebi et al., 2013). Regarding the thermal decomposition kinetics, only few studies available in the open literature dealing with the thermal oxidation kinetics of algal biomass. The commonly used method are the isoconversional technique such as Kissinger–Akahira–Sunose (KAS) and Flynn–Wall–Ozawa (FWO) methods. Iso-conversional models

are model-free methods, since the activation energy is determined at specific reaction conversion without previous knowledge of the reaction mechanism (Slopiecka et al., 2012). These methods are simple and easy to use (Damartzis and Vamvuka, 2011). However, these methods use the linearization approach with limited experimental data points, usually in the linear region to arbitrarily draw the straight line. These methods do not thoroughly apply the entire thermal decomposition profile. On the other hand, a rigorous appropriate nonisothermal decomposition model that applies to the entire single heating rate experiment be developed. Therefore, it is worthwhile to investigate the oxidation of microalgae biomass using a rigorous approach to understand the complete behavior of the thermal oxidation process. This will eventually add new insight to this subject and broaden the comprehension of the subject, which to the best of our knowledge, has not been done before. In view of the above discussions, this study was aimed to investigate the high temperature decomposition behavior of microalgae biomass for N. oculta and C. vulgaris, different approaches were used to evaluate the reaction kinetics including direct model fitting technique, and model-free methods. 2. Methods 2.1. Microalgae culturing and biomass production In this investigation two microalgae strain N. oculta and C. vulgaris have been considered based on their CO2 biofixation capabilities and wastewater treatment. Both species were acquired from Algae Depot (USA) and cultured in BBM media. Media consist of NaNO3, CaCl2.2H2O, MgSO47H2O, K2HPO4, KH2PO4, NaCl, EDTA, KOH, FeSO47H2O, H2SO4, H3BO3, ZnSO47H2O, MnCl24H2O, CuSO4 5H2O, CoCl26H2O, CH4N2O. The media was prepared by adding the salts in deionized water. The microalgal species samples obtained from starter culture was incubated in two different batch photo bioreactors (2 L Erlenmeyer flasks) with 1700 ml working volume of each reactor. The initial biomass concentration adjusted to 1 mg/L approximately for each flask. The culture was placed in fume hood with provision of plant culture Grolux fluorescent light. After filtration, the ambient air was mixed with CO2 using gas mixing device. The CO2 concentration of 4% was mixed with air fed to the reactor. Fluorescent light was provided at the surface of reactors. Cultures were set for 10 days, samples were analyzed during the growth to determine the optimal cell density, cell concentration, and dry weight, effect of CO2 utilization by two different species, pH changes, and growth rate measurements were made. The biomass harvested using centrifugation with 9000 RPM for 3 min and then farther dried using freeze dryer. 2.2. Thermogravimetric experiments The high temperature thermal decomposition of the microalgae biomass samples were evaluated using TGA (SDTG 600, TA Instruments, USA). For each experimental run, 6–6.2 mg biomass sample were placed on the sample holder. The samples were heated at an air flow rate of 100 ml/min. The weight loss of the sample was recorded from ambient temperature to 800 °C with heating rates of 5, 10, 15, 20 °C/min. Each experiment was repeated 2–4 times in order to confirm the minimum standard deviation of the percentage weight loss. 2.3. Proximate and ultimate analysis In order to determine the moisture content, the samples were exposed to constant temperature of 105 °C until stabilized their sample weight (Kebelmann et al., 2013). Following the moisture

S.A.M. Ali et al. / Bioresource Technology 175 (2015) 569–577

2.4. Kinetic modeling The high temperature non-isothermal degradation of biomass is a function of temperature and the fractional amount of the remnant solid. Thus, the reaction rate can be described by the following equation.

da ¼ kðTÞf ðaÞ dt

ð1Þ

where t is the reaction time (in min); T is absolute temperature (in K); a is the fractional conversion and is defined as;

a ¼ 1  Wf

da ¼ ko exp dt

 

 Ea f ðaÞ RT

ð3Þ

where R is universal gas constant (in kJ/mol K); Ea is the apparent activation energy (in kJ/mol); ko is the frequency factor (in 1/min); and n is the overall order of degradation. The apparent activation energy (Ea) can be obtained by two approaches. The first approach required to specify f(a) depending on the proposed reaction mechanism, and by direct fitting procedure Ea and other reaction kinetic parameters can be estimated. While the second approach is independent of the reaction mechanism and it is called model-free methods. 2.4.1. Direct model fitting We assume f(a) as the nth order reaction rate model, then the overall rate of degradation can be expressed as follows, (Lu et al., 2013; Peng et al., 2001; Rizzo et al., 2013)

da ¼ ko exp dt

 

 Ea ð1  aÞn RT

ð4Þ

The kinetics parameters of Eq. (4) which are Ea, ko and n, were evaluated by fitting Eq. (4) to the experimental data. FindFit and NonlinearModelFit commands in Mathemaica 9 software were used to evaluate the unknown parameters for each heating rate. Maximum 1000 iterations was used to achieve a specified

(b)

(c)

(d)

TG (wt%)

DTG (% wt/ min)

TG (wt%)

(a)

ð2Þ

Since Wf represents the remnant weight fraction of the sample. k(T) is the reaction rate constant which is function of temperature. Using the Arrhenius expression for the reaction rate constant, the overall rate of reaction becomes,

DTG (% wt/ min)

removal, the ash and other volatile matter fractions were estimated by modifying the traditional standard methods for biomass as described by Cantrell et al. (2010). For ash measurement, the sample was heated at 600 °C under air flow with heating rate of 11 °C/min and held at isotherm temperature for 10 min. In order to determine the amount of the volatile substance, the sample equilibrated at 110 °C then heated to 950 °C with 100 °C/min heating rate in nitrogen atmosphere and held for 7 min. The gas flow rate for both test were kept at 80 ml/min. The fixed carbon content was calculated by difference. The elemental composition represented by the percentage weight of the carbon, hydrogen and oxygen were determined using correlations developed by Parikh et al. (2007), while the corresponding higher heating value calculated using correlation developed by Parikh et al. (2005). All correlations are based on proximate analysis.

571

Fig. 1. (a) TG curve for Nannochloropsis oculta for different heating rate. (b) DTG curve for Nannochloropsis oculta for different heating rate. (c) TG curve for Chlorella vulgaris for different heating rate. (d) DTG curve for Chlorella vulgaris for different heating rate.

572

S.A.M. Ali et al. / Bioresource Technology 175 (2015) 569–577

(a)

(b) Experimental

TG (wt%)

Experimental

TG (wt%)

Model

(c)

Model

(d) Experimental

TG (wt%)

TG (wt%)

Experimental Model

Model

Fig. 2. Experimental and model TG curve for Nannochloropsis oculta at (a) b = 5 °C/min, (b) b = 10 °C/min, (c) b = 15 °C/min and (d) b = 20 °C/min.

precision, since that the fitting commands generate many solutions and the selection was based on the highest coefficient of determination R2. The coefficient of determination was estimated using ‘‘RSquared’’ command. In order to compare between the thermogravimetric (TG) profile of the model and the experiments, we need a relationship between a and T to obtain the TG curve for the model after substitution of the estimated parameters. Therefore, we must firstly expressing Eq. (4) to be as a function of temperature only and then solving it. This was done by replacing the time differential according to the given relation between temperature, time and the heating rate (b = dT/dt), where b is the heating rate (°C/min). Thus, Eq. (4) becomes

da ko ¼ exp dT b

  Ea ð1  aÞn  RT

ð5Þ

Eq. (5) can be solved by two procedure, either by the approximate analytical solution or by the numerical solution. Approximate analytical solution Firstly, Eq. (5) was arranged, followed by taking the integral of the both side of the equation

Z

0

a

da ko ¼ b ð1  aÞn

Z

T

To

exp





 Ea dT RT

ð6Þ

where To refers to the initial experiment temperature (in K). And the left hand sight of was defined as g(a).

gðaÞ ¼

Z 0

a

Although, the right hand side of Eq. (6) has no analytical solution (Ebrahimi-Kahrizsangi and Abbasi, 2008), the Coats– Redfern approximation could be used (Coats and Redfern, 1964). Given that the value of Ea/RT  1, then the right hand side of Eq. (6) could be approximated as follow (López et al., 2013).

Z

T

exp

To

    Ea R Ea dT  T 2 exp   Ea RT RT

ð8Þ

After substitution of Eq. (7) and Eq. (8) in Eq. (6), and by taking the natural logarithmic for both sides we get

ln

" # 1  ð1  aÞ1n ð1  nÞT 2

¼ ln



 ko R Ea  bEa RT

ð9Þ

The same results of Eq. (9) is obtained by (Lu et al., 2013), further rearrangement of Eq. (9) yields;

   1 ko R Ea 1n W f ¼ ð1  aÞ ¼ 1  ð1  nÞ T 2 exp  bEa RT

ð10Þ

Then, TG curve from model was plotted after substituting the estimated parameters values in Eq. (10). For thermogravimetric (TG) and differential thermogravimetric (DTG) profiles modeling, data were selected from experimental results starting from temperature of 30 °C and with increment of 10 °C the total number of point selected were in the range of (65– 78 points), for plotting of the experimental data for Fig. 2 and Fig. 3 half of the points are chosen in order to get a clear and neat graphs. Numerical solution

1n

da 1  ð1  aÞ ¼ ð1  nÞ ð1  aÞn

ð7Þ

Numerical solution for a as a function of temperature was obtained by solving Eq. (5) using NDSolve command in Mathemaica

573

S.A.M. Ali et al. / Bioresource Technology 175 (2015) 569–577

(a)

(b) Experimental

TG (wt%)

TG (wt%)

Experimental Model

(c)

Model

(d) TG (wt%)

TG (wt%)

Experimental Model

Experimental Model

Fig. 3. Experimental and model TG curve for Chlorella vulgaris at (a) b = 5 °C/min, (b) b = 10 °C/min, (c) b = 15 °C/min and (d) b = 20 °C/min.

9 software. The boundary condition is a [298 K] = 0, which at the initial experimental temperature (normally 298 K) the sample weight loss not started.

3. Results and discussion

2.4.2. Model-free methods These methods are used to determine the unknown apparent activation energy (Ea) without specifying a model to describe the reaction mechanism, there is no need to define f(a) or g(a). The following two iso-conversional methods are commonly used (Slopiecka et al., 2012):

For this investigation N. oculta and C. vulgaris microalgae samples were collected after 10 days of cultivation in batch processes. The biomass yields of N. oculta and C. vulgaris are 0.86 g/L and 0.9 g/L. The corresponding biomass productivities are 0.0859 g/L/ day and 0.0899 g/L/day, respectively. The biomass productivities was determined from the difference between initial and final culture concentration in (g/L) divided by the culture duration. One can consider the biomass productivity for the both the algae species are in the lower side. The possible reason for this low biomass productivity may be due to the low starting concentration of the culture (1 mg/L). According to Chiu et al. (2008) the biomass yields is get affected by the initial cell concentration, however when high density inoculums is used the biomass yield will be significantly greater than when low density used for culturing.

Kissinger–Akahira–Sunose (KAS) method This methods is briefly described as (Zhao et al., 2013).

 ln

b T2





¼ ln

 ko R Ea  Ea gðaÞ RT

ð11Þ h i

According to this method, the plots of ln Tb2 versus 1/T for different a values is a straight line and the Ea can be calculated from the slope of the line. Though, there is an Ea value at each fractional conversion a. Flynn–Ozawa–Wall (FWO) method The alternative linearized form of the model equation (Zhao et al., 2013)

 ln ½b ¼ ln

 0:0048ko Ea Ea  1:0516 R gðaÞ RT

ð12Þ

Again, ln [b] versus 1/T gives straight line. The Ea can be calculated from the slope of the line. Also, there is an Ea value at each fractional conversion a.

3.1. Biomass yields

3.2. Thermal characteristics of microalgae strains Table 1 shows the approximate and ultimate analysis of N. oculta and C. vulgaris. Approximate analysis is found that the moisture content, ash, volatile matter, fixed carbon and heating values are different for N. oculta than those of C. vulgaris in the culture product. It can be even more confirmed that the elemental carbon, hydrogen and oxygen content also higher for N. oculta. TG and DTG profiles show in Fig. 1 for N. oculta and C. vulgaris at different heating rates 5, 10, 15 and 20 °C/min. Fig. 1 shows the TG and DTG profiles of both the N. oculta and C. vulgaris algae species considered for this study. One can see both the microalgae species

574

S.A.M. Ali et al. / Bioresource Technology 175 (2015) 569–577

Table 1 Approximate analysis and ultimate analysis.

⁄ ⁄

Nannochloropsis oculta

Chlorella vulgaris

Approximate analysis (%) Moisture Ash Volatile matter Fixed carbon Heating value (MJ/kg)

6.71 ± 0.29 6.40 ± 0.47 78.94 ± 2.61 7.95 15.07

6.89 ± 0.44 6.86 ± 1.41 78.40 ± 2.05 7.85 14.94

Ultimate analysis (%) C H O

40.98 5.31 39.99

40.67 5.27 39.70

The standard deviation calculated using STDEV in Microsoft Excel 2010. Fixed carbon calculated by difference.

shows similar thermal behavior in the first burning stage up to temperature of 220 °C. In case of C. vulgaris a small change in trend was observed just before the maximum peak while this change not recorded clearly for N. oculta. Also, N. oculta shows high decomposition rate at the last peaks than C. vulgaris confirmed from DTG plots for both strains. In general, the thermal decomposition processes can be classified into three stages, first stage where the devolatilization of volatile compounds and moisture occur, at the second sages most of the weight loss is observed, and the last stage wherein weight change slowly at high temperature near the end of decomposition process (Peng et al., 2001). However, at earliest stage there is slight weight change at temperature around 100 °C which is mainly due to moisture evaporation. The second decomposition stage start from 200 °C to 400 °C roughly, though this weight loss is a result of combustion of proteins and carbohydrates compounds (Batista et al., 2013), other small peaks appears till a temperature of 650 °C approximately. Finally at temperature higher than 650 °C,

the weight loss decreases slowly and this stage is associated with the decomposition of the solid residual (Wu et al., 2014). Moreover, the maximum peak temperature (Tpeak) for N. oculta is slightly higher than C. vulgaris for all heating rates. Also, the corresponding maximum decomposition rates (da/dt)max and (da/dT)max is higher in N. oculta than C. vulgaris as shown in Table 2. If these temperatures at the maximum peak temperature (Tpeak) for microalgae biomass decomposition compared to other biomass like corncob, sawdust, palm shell and coconut shell as described by Chaiwong et al. (2013), we found that N. oculta and C. vulgaris have the lowest temperatures. Ash is considered as byproduct for biomass thermal decomposition, however in real operation like in pyrolysis and gasification high values of ash content is undesirable and its would influence the process design and operation since it decreases the final products quality and consequently the need for purification process (Bi and He, 2013). Ashes in microalgae mainly formed by metal compounds, however different pre-treatment processes such as washing before the combustion processes are suggested by several researches to reduce the mineral content (López-González et al., 2014). In this study, C. vulgaris showed slightly high ash content than N. oculta and in term of heating value N. oculta is better as recorded in Table 1. 3.3. Effects of heating rate Generally, when the heating rate increased the temperature at the maximum peak is tend to be delayed, a similar results was reported by Chen et al. (2011) and Chen et al. (2013) for C. vulgaris decomposition. Gai et al. (2013) postulate that since the biomass has poor thermal conductivity, so with increasing the heating rate the temperature gradient inside the biomass is increased resulting into limited heat transfer from the furnace pan to the sample, and

Table 2 Maximum peak characteristics. b

5 10 15 20 ⁄ ⁄

Nannochloropsis oculta

Chlorella vulgaris

T peak

ðda=dtÞmax

ðda=dTÞmax

T peak

ðda=dtÞmax

ðda=dTÞmax

278.20 ± 0.35 283.47 ± 0.16 285.50 ± 0.02 287.24 ± 0.23

3.42 ± 0.01 6.78 ± 0.04 10.82 ± 0.21 15.26 ± 0.01

0.683 ± 0.003 0.678 ± 0.004 0.722 ± 0.014 0.763 ± 0.001

277.38 ± 0.82 282.61 ± 0.42 286.19 ± 0.18 287.16 ± 0.60

3.31 ± 0.01 6.49 ± 0.12 10.42 ± 0.19 14.02 ± 0.65

0.662 ± 0.001 0.649 ± 0.012 0.695 ± 0.013 0.701 ± 0.032

The standard deviation calculated using STDEV in Microsoft Excel 2010. T in (°C), ðda=dtÞmax in (wt%/min), ðda=dTÞmax in (wt%/K).

Table 3 Estimated reaction kinetic parameters using the nth order model fitting. b, °C/min

5 10 15 20

Nannochloropsis oculta

Chlorella vulgaris

ko, min1

n

Ea, kJ/mol

R2

ko, min1

n

Ea, kJ/mol

R2

3.37748  109 4.2756  109 9.1346  109 1.04168  1010

6.06432 6.48867 6.79387 6.97785

106.428 105.660 107.722 107.653

0.877888 0.847409 0.826895 0.797467

3.74618  108 3.31982  108 3.62118  108 4.59909  108

5.50554 5.90607 5.69797 5.95000

95.7672 92.884 92.1558 92.0378

0.884248 0.852943 0.854144 0.837901

Table 4 Comparison between average reaction kinetics parameters evaluated using nth order model from previous studies and this work. Microalgae

b, °C/min

Atmosphere

Ea, kJ/mol

ko, min1

n

References

Chlorococcum humicola Nannochloropsis oculta Chlorella vulgaris

5, 10, 20 and 80 5, 10, 15 and 20 5, 10, 15 and 20

Nitrogen Air Air

189.99 106.87 92.36

2.4359  1016 6.8011  109 3.8216  108

7.25 6.58 5.76

Kirtania and Bhattacharya (2013) Present work Present work

575

S.A.M. Ali et al. / Bioresource Technology 175 (2015) 569–577

(a)

mass when high heating rates is used. In addition, Idris et al. (2010) reported that when heating rate is decreased, residual of the thermal decomposition reaction decreased too.

-9.6

Ln(B/T2)

-10

α=0.2

3.4. Evaluation of the model-fitting method

α=0.4

-10.4

α=0.6 -10.8 α=0.8 -11.2

-11.6 0.0012

0.0014

0.0016

0.0018

0.002

0.0022

1/T , (K-1)

(a)

-9.6

Ln(B/T2)

-10

α=0.2 α=0.4

-10.4

α=0.6

-10.8

α=0.8 -11.2

-11.6 0.0012

0.0014

0.0016

0.0018

0.002

0.0022

1/T , (K-1) h i Fig. 4. The plot of ln Tb2 versus 1/T for different a values for KAS method for (a) Nannochloropsis oculta and (b) Chlorella vulgaris.

therefore shifting the decomposition temperatures, on other words delay will occur in the sample temperature due to the rapid change in the furnace temperature. Also they argued that when heating rate is increased the residence time of the sample inside the combustion furnace is shorten and as consequences the corresponding initial and final decomposition temperatures are get affected (delayed). This evidences supported again by Liang et al. (2014) for the thermal decomposition process of smooth cordgrass, hence they refer the shifting of the initial temperature for main decomposition stage, to the temperature gradient across the bio-

Experimental and modeling TG curves at (a) b = 5 °C/min, (b) b = 10 °C/min, (c) b = 15 °C/min and (d) b = 20 °C/min for N. oculta shown in Fig. 2 and for C. vulgaris in Fig. 3. The fitting of the thermal weight loss with the nth order rate model described by Eq. (4) shows excellent prediction to the experimental values of the decomposition processes for both species. However, model-fitting enables using of just one single heating rated each time for determination of reaction kinetics parameters including the activation energy, and actually this is considered as drawback, since the activation energy is get affected by the heating rate due to mass and heat transfer consequences (López et al., 2013). Table 3 shows the estimated reaction kinetic parameters using the nth order rate model described by Eq. (4). Although, the proposed model shows best fitting the experimental data but the coefficient of determination (R2) values from Table 3 are to some extent much lesser than unity. This because for such nonlinear model is misleading to use the coefficient of determination R2. Therefore, the coefficient of determination R2 is an inappropriate measure for the goodness of fit in case of nonlinear models (Spiess and Neumeyer, 2010). More discussions about fitting data to nonlinear models, R2 determination and its suitability, and about nonlinear methods for regression are provided by (Filzmoser, 2008; Spiess and Neumeyer, 2010). Other statistical expression to examine the fitting accuracy for nonlinear modeling of the thermal decomposition of biomass is used by Vamvuka et al. (2003) which is based on the deviation between the observed and calculated (da/dt) with respect to the maximum observed (da/dt). The estimated average reaction kinetic parameters described in Table 4, shows high value of reaction order (n), since the reaction orders were 6.58 and 5.76 for N. oculta and C. vulgaris, respectively. A similar result of such high value of (n) for nth order reaction model was reported by Kirtania and Bhattacharya (2013) in pyrolysis kinetics of fresh water alga strain Chlorococcum humicola. In addition, the approximate analytical solution developed by Coats–Redfern method gave identical TG curve to those evaluated by the numerical solution generated using NDSolve command in Mathematica 9 software, which is mean that the Coats–Redfern method is an accurate approximation, and it can be used safely.

3.5. Evaluation of the iso-conversional methods Using iso-conversional models KAS and FWO the activation energy was determined using different heating rate experimental data, hence the evaluation of apparent activation energy was done

Table 5 Apparent activation estimated using KAS and FWO methods.

a Nannochloropsis oculta

Chlorella vulgaris

0.2 0.4 0.6 0.8 Average 0.2 0.4 0.6 0.8 Average

KAS

FWO

Slope

Ea, kJ/mol

R2

Slope

Ea, kJ/mol

R2

21939 26602.6 12919.6 10334.7

182.4 221.1 107.4 85.9 149.2 240.0 321.3 163.0 132.9 214.3

0.99 0.99 0.99 0.98

23014.1 27742.7 14189.9 11845

0.99561 0.999482 0.99371 0.993251

0.99 0.98 0.91 0.91

29914.6 39777.6 20837.7 17457.2

181.9 219.3 112.1 93.6 151.7 236.5 314.4 164.7 138.0 213.4

28873.2 38656.4 19607.9 15990.9

0.99 0.984617 0.927383 0.92693

576

S.A.M. Ali et al. / Bioresource Technology 175 (2015) 569–577

(a) 3.1 2.9 α=0.2

Ln(B)

2.7 2.5

α=0.4

2.3

α=0.6

2.1

α=0.8

1.9 1.7 1.5 0.0012

0.0014

0.0016

0.0018

0.002

0.0022

1/T , (K-1)

(b)

3.1 2.9 2.7

α=0.2

Ln(B)

2.5

α=0.4

2.3 α=0.6 2.1 α=0.8

1.9 1.7 1.5 0.0012

0.0014

0.0016 1/T ,

0.0018

0.002

0.0022

(K-1)

Fig. 5. The plot of ln [b] versus 1/T for different a values for FWO method for (a) Nannochloropsis oculta and (b) Chlorella vulgaris.

without any need to define the reaction mechanism and how the reaction rate look h i like (Damartzis and Vamvuka, 2011). The plot ln Tb2 versus 1/T for a values of 0.2, 0.4, 0.6 and 0.8 for KAS method shows for N. oculta in Fig. 4a), while Fig. 4b) shows similar plots for the species of C. vulgaris. Ea and ko calculated from the slope and intercept of all lines produced for four different a values and shows in Table 5 for comparison. The plot of ln [b]

versus 1/T for a values of 0.2, 0.4, 0.6 and 0.8 for FWO method shows for N. oculta in Fig. 5a) and for C. vulgaris in Fig. 5b). The Ea and ko by the same methods described before, and the values listed in Table 5. From Table 5, it can be noted that the average apparent activation energy, Ea for N. oculta is found 149.22 kJ/mol by KAS method and 151.78 kJ/mol by FWO method. For the case of C. vulgaris it is found to be 214.35 kJ/mol by KAS method and 213.44 kJ/mol by FWO method. Obviously, the values of activation energy for C. vulgaris is much greater than N. oculta, however this results is limited to biomass under study, since changing culturing condition and growth parameter may reflect on the produced biomass composition and therefore the thermal characteristics of biomass decomposition. From author’s experience, that activation energy determined by KAS and FWO methods is very sensitive towards small changes in the experimental data than the model fitting techniques described before, however a considerable change are noticed in the average activating energy if experiments replicated. For all reviewed studies the difference between the average values of activation energy estimated by KAS and FWO are in significance for various microalgae biomass under different atmosphere with different heating rate used (Table 6). Although, the apparent activation energy is not constant and it varies during the conversion as clear in Fig. 6, which means there is more than single reaction mechanism for the biomass thermal decomposition process (Slopiecka et al., 2012). Therefore, single reaction mechanism at each thermal decomposition stage can be used to evaluate the activation energy and other reaction parameters instead of single mechanism for the whole process (López-González et al., 2014). To analyze the trend of activation energy versus reaction conversion we found that for all decomposition stage a = 0.2, 0.4, 0.6, and 0.8 the average estimated activation energy by KAS and FWO methods for N. oculta is lower than that for C. vulgaris. A general trend is that at small a, activation energy increase with increasing the reaction conversion till reaching the maximum at almost a middle conversion, then the value of activation energy decreases sharply until the end of combustion process as shown in Fig. 6. Therefore, the facts that apparent activation energy change during the decomposition indicates the present of different reaction mechanisms throughout the thermal decomposition process, since these iso-conversional methods do not depend on a reaction mechanism for activation energy determination (Slopiecka et al., 2012). Multi stages kinetics modeling are normally being used for this purpose, since the thermal decomposition process is divided into devolatilization stage and oxidation stage which also includes the

Table 6 Comparison between reaction kinetics parameters evaluated KAS and FWO from previous studies and including this work.

*

Microalgae

b, °C/min

Atmosphere

Ea, kJ/mol KAS

Ea, kJ/mol FWO

References

Chlorella vulgaris Chlorella vulgaris Chlorella vulgaris Chlorella vulgaris Macrocystis pyrifera* Potamogeton crispus* Sargassum thunbergii* Tetraselmis suecica Chlorella vulgaris Chlorella vulgaris Chlorella vulgaris Chlorella vulgaris Nannochloropsis oculta Chlorella vulgaris

10, 20 and 40 10, 20 and 40 10, 20 and 40 10, 20 and 40 5, 10, 20, and 30 10, 30 and 50 10, 30 and 50 5, 10, and 15 10, 20 and 40 10, 20 and 40 10, 20 and 40 10, 20 and 40 5, 10, 15 and 20 5, 10, 15 and 20

O2: O2: O2: O2: Ar N2 N2 O2: O2: O2: O2: O2: Air Air

134.53 163.49 211.24 242.33 221.4 143.2 185.6 70.09 37.58 46.13 53.75 56.27 149.23 214.35

134.03 165.74 209.92 241.04 219.7 145.3 185.4 75.81 42.11 50.22 57.31 59.50 151.78 213.44

Chen et al. (2011) Chen et al. (2011) Chen et al. (2011) Chen et al. (2011) Zhao et al. (2011) Li et al. (2012) Li et al. (2012) Tahmasebi et al. (2013) Chen et al. (2013) Chen et al. (2013) Chen et al. (2013) Chen et al. (2013) Present work Present work

Macroalgae.

N2 N2 N2 N2

(20:80) (50:50) (60:40) (80:20)

N2 (21:79) N2 (20:80) CO2 (20:80) CO2 (50:50) CO2 (80:20)

S.A.M. Ali et al. / Bioresource Technology 175 (2015) 569–577

350 300 C. vulgaris Ea , (KJ/mol)

250

N. oculta

200 150 100 50 0

0.0

0.2

0.4

0.6

0.8

1.0

α Fig. 6. Average activation energy estimated by KAS and FWO methods at different conversion stages for Nannochloropsis oculta and Chlorella vulgaris.

last step of carbonate decomposition and metal volatilization (López-González et al., 2014). Multi-zone pyrolysis analysis was also involved in study done by Wu et al. (2014) for the thermal analysis and modeling of aquatic biomass decomposition including Dunaliella tertiolecta microalgae. 4. Conclusions The TGA analysis shows that decomposition profiles are different for the two studied microalgae species (N. oculta and C. vulgaris). The TGA study also reveals that the biomass culture conditions have significant influences on the produced biomass compositions and their thermal decomposition characteristics. The thermal decomposition kinetics analysis shows that the thermal degradation of the studied microalgae biomass follows different mechanisms. The estimated values of activation energy for C. vulgaris (215 kJ/mol) is significantly higher than that of N. oculta (150 kJ/mol). Acknowledgements King Abdulaziz City for Science and Technology (KACST) – Saudi Arabia through the Science & Technology Unit at King Fahd University of Petroleum & Minerals (KFUPM) is acknowledged for funding this work through project No. KACST ARP# A-T-32-62. References Amutio, M., Lopez, G., Aguado, R., Artetxe, M., Bilbao, J., Olazar, M., 2012. Kinetic study of lignocellulosic biomass oxidative pyrolysis. Fuel 95, 305–311. Agrawal, A., Chakraborty, S., 2013. A kinetic study of pyrolysis and combustion of microalgae Chlorella vulgaris using thermo-gravimetric analysis. Bioresour. Technol. 128, 72–80. Batista, A.P., Gouveia, L., Bandarra, N.M., Franco, J.M., Raymundo, A., 2013. Comparison of microalgal biomass profiles as novel functional ingredient for food products. Algal Res. 2, 164–173. Bi, Z., He, B., 2013. Characterization of microalgae for the purpose of biofuels production. Trans. ASABE 56, 1529–1539. Cantrell, K., Martin, J., Ro, K., 2010. Application of thermogravimetric analysis for the proximate analysis of livestock wastes. J. ASTM Int. 7, 1–13. Chaiwong, K., Kiatsiriroat, T., Vorayos, N., Thararax, C., 2013. Study of bio-oil and bio-char production from algae by slow pyrolysis. Biomass Bioenergy 56, 600– 606. Chen, C., Lu, Z., Ma, X., Long, J., Peng, Y., Hu, L., Lu, Q., 2013. Oxy-fuel combustion characteristics and kinetics of microalgae Chlorella vulgaris by thermogravimetric analysis. Bioresour. Technol. 144, 563–571.

577

Chen, C., Ma, X., Liu, K., 2011. Thermogravimetric analysis of microalgae combustion under different oxygen supply concentrations. Appl. Energy 88, 3189–3196. Chiu, S.-Y., Kao, C.-Y., Chen, C.-H., Kuan, T.-C., Ong, S.-C., Lin, C.-S., 2008. Reduction of CO2 by a high-density culture of Chlorella sp. in a semicontinuous photobioreactor. Bioresour. Technol. 99, 3389–3396. Coats, A.W., Redfern, J.P., 1964. Kinetic parameters from thermogravimetric data. Nature 201, 68–69. Damartzis, T., Vamvuka, D., 2011. Thermal degradation studies and kinetic modeling of cardoon (Cynara cardunculus) pyrolysis using thermogravimetric analysis (TGA). Bioresour. Technol. 102, 6230–6238. Ebrahimi-Kahrizsangi, R., Abbasi, M., 2008. Evaluation of reliability of Coats-Redfern method for kinetic analysis of non-isothermal TGA. Trans. Nonferrous Met. Soc. China 18, 217–221. Filzmoser, P., 2008. Linear and Nonlinear Methods for Regression and Classification and applications in R. Gai, C., Zhang, Y., Chen, W.-T., Zhang, P., Dong, Y., 2013. Thermogravimetric and kinetic analysis of thermal decomposition characteristics of low-lipid microalgae. Bioresour. Technol. 150, 139–148. Idris, S.S., Abd Rahman, N., Ismail, K., Alias, A.B., Abd Rashid, Z., Aris, M.J., 2010. Investigation on thermochemical behaviour of low rank Malaysian coal, oil palm biomass and their blends during pyrolysis via thermogravimetric analysis (TGA). Bioresour. Technol. 101, 4584–4592. Kebelmann, K., Hornung, A., Karsten, U., Griffiths, G., 2013. Intermediate pyrolysis and product identification by TGA and Py-GC/MS of green microalgae and their extracted protein and lipid components. Biomass Bioenergy 49, 38–48. Kirtania, K., Bhattacharya, S., 2013. Pyrolysis kinetics and reactivity of algae e coal blends. Biomass Bioenergy 55, 291–298. Li, D., Chen, L., Chen, S., Zhang, X., Chen, F., Ye, N., 2012. Comparative evaluation of the pyrolytic and kinetic characteristics of a macroalga (Sargassum thunbergii) and a freshwater plant (Potamogeton crispus). Fuel 96, 185–191. Liang, Y., Cheng, B., Si, Y., Cao, D., Jiang, H., 2014. Thermal decomposition kinetics and characteristics of Spartina alterniflora via thermogravimetric analysis. Renew. Energy 68, 111–117. López, R., Fernández, C., Gómez, X., 2013. Thermogravimetric analysis of lignocellulosic and microalgae biomasses and their blends during combustion. J. Therm. Anal. Calorim. 114, 295–305. López-González, D., Fernandez-Lopez, M., Valverde, J.L., Sanchez-Silva, L., 2014. Kinetic analysis and thermal characterization of the microalgae combustion process by thermal analysis coupled to mass spectrometry. Appl. Energy 114, 227–237. Lu, K., Lee, W., Chen, W., Lin, T., 2013. Thermogravimetric analysis and kinetics of co-pyrolysis of raw/torrefied wood and coal blends. Appl. Energy 105, 57–65. Parikh, J., Channiwala, S.a., Ghosal, G.K., 2005. A correlation for calculating HHV from proximate analysis of solid fuels. Fuel 84, 487–494. Parikh, J., Channiwala, S.a., Ghosal, G.K., 2007. A correlation for calculating elemental composition from proximate analysis of biomass materials. Fuel 86, 1710–1719. Peng, W., Wu, Q., Tu, P., Zhao, N., 2001. Pyrolytic characteristics of microalgae as renewable energy source determined by thermogravimetric analysis. Bioresour. Technol. 80, 1–7. Razzak, S., Hossain, M.M., Lucky, R., Bassi, A.S., de Lasa, H., 2013. Integrated CO2 capture, wastewater treatment and biofuel production by microalgae culturing—A review. Renew. Sustain. Energy Rev. 27, 622–653. Rizzo, A.M., Prussi, M., Bettucci, L., Libelli, I.M., Chiaramonti, D., 2013. Characterization of microalga Chlorella as a fuel and its thermogravimetric behavior. Appl. Energy 102, 24–31. Sanchez-Silva, L., López-González, D., Garcia-Minguillan, a.M., Valverde, J.L., 2013. Pyrolysis, combustion and gasification characteristics of Nannochloropsis gaditana microalgae. Bioresour. Technol. 130, 321–331. Slopiecka, K., Bartocci, P., Fantozzi, F., 2012. Thermogravimetric analysis and kinetic study of poplar wood pyrolysis. Appl. Energy 97, 491–497. Spiess, A., Neumeyer, N., 2010. An evaluation of R2 as an inadequate measure for nonlinear models in pharmacological and biochemical research: a Monte Carlo approach. Spiess Neumeyer BMC Pharmacol. 10 (6), 1–11. Tahmasebi, A., Kassim, M.A., Yu, J., Bhattacharya, S., 2013. Thermogravimetric study of the combustion of Tetraselmis suecica microalgae and its blend with a Victorian brown coal in O2/N2 and O2/CO2 atmospheres. Bioresour. Technol. 150, 15–27. Vamvuka, D., Pasadakis, N., Kastanaki, E., Grammelis, P., Kakaras, E., 2003. Kinetic modeling of coal/agricultural by-product blends. Energy Fuels 17, 549–558. Wu, K., Liu, J., Wu, Y., Chen, Y., Li, Q., Xiao, X., Yang, M., 2014. Pyrolysis characteristics and kinetics of aquatic biomass using thermogravimetric analyzer. Bioresour. Technol. 163, 18–25. Zhao, H., Yan, H., Dong, S., Zhang, Y., Sun, B., Zhang, C., Ai, Y., Chen, B., Liu, Q., Sui, T., Qin, S., 2013. Thermogravimetry study of the pyrolytic characteristics and kinetics of macro-algae Macrocystis pyrifera residue. J. Therm. Anal. Calorim. 111, 1685–1690.