Thermogravimetric and mass-spectrometric analyses of combustion of
spent potlining under N
2
/O
2
and CO
2
/O
2
atmospheres
Guang Sun
a,b, Gang Zhang
b, Jingyong Liu
a,⇑, Wuming Xie
a, Jiahong Kuo
a, Xingwen Lu
a, Musa Buyukada
c,
Fatih Evrendilek
d,e, Shuiyu Sun
aa
Guangzhou Key Laboratory Environmental Catalysis and Pollution Control, Guangdong Key Laboratory of Environmental Catalysis and Health Risk Control, School of Environmental Science and Engineering, Institute of Environmental Health and Pollution Control, Guangdong University of Technology, Guangzhou 510006, China b
Department of Energy and Chemical Engineering, Dongguan University of Technology, Dongguan 523808, China c
Department of Chemical Engineering, Bolu Abant Izzet Baysal University, Bolu 14052, Turkey d
Department of Environmental Engineering, Bolu Abant Izzet Baysal University, Bolu 14052, Turkey e
Department of Environmental Engineering, Ardahan University, Ardahan 75002, Turkey
a r t i c l e i n f o
Article history:
Received 26 September 2018 Revised 24 January 2019 Accepted 31 January 2019 Available online 12 February 2019 Keywords: Spent potlining Oxy-fuel combustion Thermodynamic analysis TG-MS
a b s t r a c t
Thermal decomposition and gaseous evolution of the spent potlining (SPL) combustion were quantified using thermogravimetric and mass-spectrometric analyses in CO2/O2and N2/O2atmospheres using three
heating rates (15, 20 and 25°C/min). The thermal decomposition of SPL occurred mainly between 450 and 800°C. Based on the four kinetic methods of Friedman, Starink, Kissinger-Akahira-Sunose and Flynn-Wall-Ozawa under the various conversion degrees (a) from 0.1 to 0.7, the lowest apparent activa-tion energy was estimated at 149.81 kJ/mol in the 70% CO2/30% O2atmosphere. The pre-exponential
fac-tor, and changes in entropy, enthalpy and free Gibbs energy were also estimated. The reaction model did not suggest a single reaction of the SPL combustion. With theavalue of 0.25–0.7, the following function best described the reaction based on the Malek method: f(a) = 1/2aand G(a) = lna2. The gases released
during the combustion process included CO2, CO, NOx, HCN, and HF.
Ó 2019 Elsevier Ltd. All rights reserved.
1. Introduction
Spent potlining (SPL) is a hazardous solid waste of the alu-minum production process and consists of carbon, fluoride (NaF, CaF2, and Na3AlF6), cyanide (NaCN, Na4Fe(CN)6, or Na3Fe(CN)6),
and other inorganic compounds (Al2O3, CaO, and NaAl11O17)
(Yuan et al., 2018). Globally, 1–1.5 million tons of SPL are gener-ated annually with a SPL/primary aluminum ratio of about 1/40 (Gao et al., 2016). However, the disposal of untreated SPL harms the environment due to its high contents of cyanide, soluble fluo-ride, and inorganic compounds (Xiao et al., 2018). Leachates of sol-uble fluoride, cyanide, and other toxic substances from landfills have polluted soils, and surface and groundwater resources (Gao et al., 2016). In 1988, the U.S. Environmental Protection Agency banned the storage of SPL in landfills without its adequate pre-treatment. Biochar was reported to adsorb hazardous leachates from landfills (Maroušek et al., 2016). However, given the increas-ing amount of the waste stream of carbon and inorganic com-pounds of SPL in landfills, there is an urgent need to encourage
its environmentally and economically alternative disposals
(Courbariaux et al., 2004). Many disposal methods for SPL focused on the recovery of inorganic compounds, and the reuse of carbon for energy. For example, flotation (Li et al., 2014), caustic leaching (Birry et al., 2016), leaching with aluminum nitrate and nitric acid (Lisbona et al., 2013), co-treatment with acid and alkali (Shi et al., 2012), and ultrasound-assisted alkali leaching (Xiao et al., 2018) are some of such disposal methods. However, the gaseous releases of H2, a low leaching rate, a long leaching time, and a high
treat-ment cost are the major shortcomings of the existing methods (Xiao et al., 2018).
The co-combustion technology has been widely adopted to pro-cess SPL in many countries such as Canada, United States, France, Brazil, India, and South Africa (Holywell and Breault, 2013). There have been a few studies about thermogravimetric (TG) and differential scanning calorimetric (DSC) analyses of SPL in the air atmosphere. For example,Yuan et al. (2018)showed that the main decomposition stage of carbon took place in the range of 500– 800°C. Coal replaced by SPL for the tuyere injection was found to reduce coke rates and greenhouse gas emissions during the iron-making process (Gao et al., 2016). The use of SPL combined with pulverized coal in the bauxite sintering process led to the full
https://doi.org/10.1016/j.wasman.2019.01.047
0956-053X/Ó 2019 Elsevier Ltd. All rights reserved. ⇑ Corresponding author.
E-mail address:Liujy@gdut.edu.cn(J. Liu).
Contents lists available atScienceDirect
Waste Management
utilization of carbon, and the decompositions of cyanides into N2
and CO2 and of fluorides into calcium fluoride, and finally, into
red mud (Li and Chen, 2010). When its fixed carbon content was above 60%, SPL performed well as a secondary raw fuel in the cement plants, provided substantial heating for certain thermal
processes and helped to reduce NOx emissions (Ospina and
Hassan, 2017).Courbariaux et al. (2004)pointed out that the SPL combustion at a high temperature converted fluoride into CaF2
with the addition of limestone, while heat provided from the com-bustion of its fixed carbon fraction lowered the operational costs.
Courbariaux et al. (2004)pointed to the relatively low investment costs of the SPL combustion owing to the short residence time and low corrosive characteristics of its combustion gases. The chief advantages of the SPL combustion can be summarized thus: (1) the utilization of carbon in energy generation, (2) the decomposi-tion of cyanides into N2and CO2to reduce SPL toxicity and
dis-posal, and (3) the avoidance of costs associated with the SPL reduction and disposal.
Since the SPL combustion for energy production can release CO2
in high amounts, it is essential to develop carbon capture and stor-age technology for its flue gases. The use of the oxy-fuel combustion technology has been suggested as one of the promising technolo-gies to capture CO2 emissions from power plants (Chen et al., 2018). The traditional combustion atmosphere presents a low (around 15%) CO2concentration which renders the separation of
CO2from the exhaust gases more difficult, thus increasing the
oper-ating costs (Irfan et al., 2012; Mondal et al., 2012). The oxy-fuel combustion technology not only reduces CO2and NOxemissions
but also makes CO2capture and separation highly convenient (Bu et al., 2014; Hu and Yan, 2012). Oxy-fuel (co-)combustion studies were reported for coal (Zhuo et al., 2017), plastic, rubber and leather (Tang et al., 2015), sewage sludge with water hyacinth (Huang et al., 2016) and coffee grounds (Chen et al., 2018), and textile dyeing sludge with pomelo peel (Xie et al., 2018a). The advantages of the oxy-fuel combustion technology include the control by the heat exchanges of recycled flue gases over the combustion temperature, and the promotion of char burnout in the higher temperature.
To the best of our knowledge, there still remains a gap in related literature about the quantitative understanding of thermodynamic and kinetic parameters for the SPL combustion and of its behavior in the oxy-fuel atmospheres. Therefore, the objectives of this study were to (1) determine ignition temperature, burnout performance and comprehensive combustion performance of SPL using TG and mass-spectrometric (TG-MS) analyses and (2) compare effects of CO2/O2versus N2/O2atmospheres on the SPL combustion
perfor-mance using four kinetic methods to estimate apparent activation energy (Ea) and four thermodynamic parameters under three
heat-ing rates.
2. Materials and methods
2.1. Sample collection and preparation
SPL was collected from an aluminum smelter plant in Guangz-hou of the Guangdong Province, China. About 5 kg of SPL were ran-domly sampled, air-dried in an oven at 105°C for 24 h, crushed, sieved with a 74
l
m sieve, homogenized and prepared for tests. 2.2. Experimental methodsThermogravimetric experiments were carried out using a NETZSCH STA 409 PC (Luxx, Germany) simultaneous analyzer. Its detection limit was 0.001 mg. Around 5.500 ± 0.500 mg of SPL for each experimental run were used in Al2O3crucible. It was heated
from room temperature to 900°C at the three heating rates of 15, 20 and 25°C/min in the following four atmospheres: 80%
CO2/20% O2, 70% CO2/30% O2, 60% CO2/40% O2and 70% N2/30% O2
at a flow rate of 50 mL/min. Prior to the experiments, several tests without samples were performed to obtain the baselines to mini-mize instrumental errors. Each experiment was conducted in three times to ensure that errors were within ±2%. Thermogravimetric and derivative TG (DTG) curves were obtained directly from the NETZSCH-T4-Kinetic 2 software. An elemental analyzer (Elemen-tary Analysen Systeme Gmbh, Germany), and a Parr 6300 Oxygen Bomb Calorimeter (Parr Instrument Company, United States) were used to obtain ultimate analysis results and higher heating values, respectively. The proximate and ultimate analyses, and higher heating values are shown inTable 1.
The SPL samples were analyzed using X-ray diffraction (XRD, MiniFlex 600, Rigaku Corporation, Japan) in a scanning range of 10–90° at a scanning velocity of 2°/min. The X-ray tube was per-formed at 40 kV and 40 mA. The powdered samples were analyzed using Cu K
a
(k = 0.15418 nm) radiation with a step size of 0.02°. The main SPL compounds of the XRD pattern included C and NaF together with small amounts of Na3AlF6, CaF2, NaAl11O17, andAl2O3 as well as NaF, CaF2, NaAl11O17, NaAlSiO4, and Al2O3 after
the combustion in the 70% CO2/30% O2atmosphere at 900°C (Fig. 1).
Fluoride content of SPL was determined using an Ion Selective Electrode (ISE, Pinnacle-315P, USA). Alumina (Al) was determined using the chemical titration method, while sodium (Na) was deter-mined using an atomic absorption spectrophotometer (AAS-240, USA). Other mineral compositions and trace metals of SPL were determined using an inductively coupled plasma optical emission spectrometer (ICP-OES, ICAP7400, Thermo, USA). All the results are shown inTable 1.
The TG-MS experiments were performed using a thermo mass photo TG-DTA-PIMS 410/S (Rigaku Corporation, Japan) with an electron bombardment ionization (EI) source. It was heated from room temperature to 900°C at a heating rate of 20 °C/min in the 80% He/20% O2atmosphere at a flow rate of 150 mL/min.
2.3. Characterization of combustion performances
Combustion characteristic parameters were obtained to evalu-ate the effects of the heating revalu-ates and the atmospheres on the combustion performances of SPL and included peak temperature (Tp), ignition temperature (Ti)—defined as the intersection between
the tangent line of the point at which decomposition started and the tangent line of the maximum weight loss rate (Rp), average
weight loss rate (Rv), temperature interval at the half value of
maximum weight loss rate (DT1/2), and burnout or final
tempera-ture (Tb) (Cheng et al., 2018; Jayaraman et al., 2017).
So as to directly understand the SPL combustion performances under the different conditions, the following combustion indices were also calculated: volatile matter release index (Dv), ignition
index (Ci) (López-González et al., 2014), burnout index (Cb)
(Wang et al., 2018), and comprehensive combustibility index (S) (Wang et al., 2012). Ciand Cbcan be described as a function of
char-acteristic temperatures and weight loss rates (Chen et al., 2015). Dv
represents the release performance of volatiles of fuels (Chen et al., 2017c). S represents the comprehensive characteristics including ignition and burnout (Hao et al., 2018). These combustion indices are expressed as follows:
Dv¼ Rp Tp Tv
D
T1=2 ð1Þ Ci¼ R p ti tp ð2Þ Cb¼ R pD
t1=2 tp tb ð3ÞS¼ Rp Rð vÞ T2 i Tb ð4Þ
where ti, tp, tb,Dt1/2and Tvrefer to ignition time, peak time, burnout
time, time interval at the half value ofRp, and initial
devolatiliza-tion temperature, respectively. 2.4. Kinetic analyses
2.4.1. Kinetic models
The kinetic analysis is essential to the evaluation of the charac-teristic behaviors of the SPL combustion in a changing condition. Apparent activation energy (Ea) and the pre-exponential factor
(A) were determined, and the best-fit reaction model was found. A kinetic equation for the thermal decomposition processes was expressed as follows:
d
a
dt ¼ fð Þ k T
a
ð Þ ð5Þwhere t is time (min); d
a
/dt is reaction rate (min1); T is absolute temperature (K); k(T) is rate constant; f(a
) is reaction model of a dif-ferential form; anda
is conversion rate which can be derived from TG curves thus:a
¼m0 mtm0 mf
ð6Þ
where m0, mtand mfare the initial, actual and final sample masses,
respectively. According to the Arrhenius law, the following rate con-stant (k(T)) was described:
k Tð Þ ¼ Aexp Ea RT
ð7Þ
where Ea is apparent activation energy (kJ/mol); A is the
pre-exponential factor (s1); and R is the universal gas constant (8.314 J/molK1). For the non-isothermal reaction of solid fuels,
the heating rate (b) is generally a constant. b was estimated as a function of temperature (T) over time (t) thus:
b ¼dTdt ð8Þ
Based on the Vallet equation, Eq.(8)can be rewritten as dt = dT/ b, then Eq.(5)can be replaced by the following:
d
a
dT¼ A bexp Ea RT fð Þa
ð9Þwhere A, Ea, and f(
a
) are usually determined using (D)TG data basedon the Arrhenius equation. The integral equation of Eq.(9)can be written as follows: Gð Þ ¼
a
Z a 0 dð Þa
fð Þa
¼ A b Z T 0 exp Ea RT d Tð Þ ¼AEa bR Z 1 x expð Þx x2 dx¼ AEa bRp xð Þ ð10Þwhere x = Ea/RT; and p(x) is temperature integral, not a convergent
one with an exact analytic solution; and G(
a
) is the integral form of reaction model (1/f(a
)). Different reaction models (f(a
) and G(a
)) are shown in supplementary Table S1.Many different kinetic analyses have been used to estimate Ea
and can be grouped into the differential versus integral methods. Both methods have their advantages and disadvantages. For exam-ple, the differential methods need actual TG values (d
a
/dt or da
/dT), whereas the others need to calculate the temperature integral. In this study, Eawas estimated using the four methods of Friedman,Starink (differential form), Flynn-Wall-Ozawa (FWO), and
Kissinger-Akahira-Sunose (KAS) (integral form). 2.4.2. Integral iso-conversional methods (FWO and KAS)
The FWO method (Chen et al., 2017b, Fernandez-Lopez et al., 2016) is the integral iso-conversional one that uses the Doyle approximation method (Flynn, 1966) as follows:
lnb ¼ ln AEa
RGð Þ
a
5:3305 1:052 EaRT ð11Þ
where
a
is constant, and G(a
) is a constant assuming that G(a
) is only related toa
. From the plot of 1/T against lnb, the slope was derived as1.052Ea/R. Thus, Eaestimate of the reaction at a certainconversion rate was calculated.
The KAS method (Doyle, 1961) uses the following Coats-Redfern approximation:
Table 1
Proximate and ultimate analyses, higher heating value, and mineral and heavy metal contents of SPL.
Ultimate analysis (wt%) Proximate analysis (wt%) Qnet(MJ/kg)
Cad Had Nad S Mad Vad Aad FCad
69.11 0.40 0.16 0.26 0.73 1.56 29.04 68.67 22.21
Mineral matter (air dry) (wt%) Element content (wt%)
Na2O Al2O3 SiO2 Fe2O3 CaO Li2O K2O MgO F
11.44 7.58 2.21 1.42 1.17 0.79 0.27 0.13 10.40
Trace metal content (mg/kg)
Cr Ba V Sr Zr Ni Zn Pb Cu Y Nb
218.70 84.24 70.77 47.98 30.52 21.13 17.04 7.91 6.42 5.58 4.46
Mad: moisture; Vad: volatile matter; Aad: ash; FCad: fixed carbon; Qnet: higher heating value on a dry basis; andad: air-dried basis.
Fig. 1. XRD patterns of raw SPL and ash after combustion in the 70% CO2/30% O2 atmosphere at 900°C.
lnb T2¼ ln AR EaGð Þ
a
Ea RT ð12ÞOn the premise that G(
a
) is known whena
andb are constant, the plot of ln(b/T2) against 1/T can be fitted a linear regression line. Apparent activation energy of the sample was obtained from the slope of the fitted line(Ea/R).
2.4.3. Differential iso-conversional methods (Friedman and Starink) The Friedman method (Friedman, 1964) is the common differ-ential iso-conversional one to calculate Ea according to Eqs.(5) and (9)as follows: ln d
a
dt ¼ lnA þ lnfð Þa
Ea RT ð13ÞAccording to Eq.(11), when A and
a
are constant, and lnf(a
) is a function of onlya
, then lnA and lnf(a
) become constant. From the plot of ln(da
/dt) versus (1/T), the slope was directly derived asEa/R at a given
a
.The Starink method (Cai et al., 2016) is based on the iso-conversional FWO and KAS methods thus:
lnb Ts¼ Cs
BEa
RT ð14Þ
where S (1.8), B (1.0037), and Csare constant (Cai et al., 2016).
Hence, Eq.(14)can be converted into the following equation:
ln b
T1:8¼ Cs 1:0037 Ea
RT ð15Þ
where1.0037 Ea/R is the slope of the linear regression line of the
plot of ln(b/T1.8) versus 1/T.
2.4.4. Selection of reaction model
The reaction model was determined using the Malek method (Diaz and Phan, 2016) which can yield holonomic kinetic results via the iso-conversional method. The Malek method also serves to avoid the issue of constantly determining f(
a
) values and a kinetic compensation effect to calculate A, Eaand f(a
)simultane-ously (Huang et al., 2017). The two master plots of y(
a
) and Z(a
) can be used to select the reaction model based on the Malek method. In this study, the master plot of y(a
)-a
was adapted to determine the reaction model.According to Eqs.(5) and (16)(the Coats-Redfern equation), G (
a
) can be turned into Eq.(17):Gð Þ ¼
a
Z a 0 da
fð Þa
¼ ART2 Eab exp Ea RT ð16Þ Gð Þ ¼a
RT 2 Eab da
dt 1 fð Þa
ð17ÞWhen
a
= 0.5, G (0.5) can be described as follows:G 0:5ð Þ ¼RT 2 0:5 Eab d
a
dt 0:5 1 f 0:5ð Þ ð18Þwhere T0.5and (d
a
/dt)0.5 were the temperature and reaction rate,respectively, with
a
= 0.5. The Malek method used in the calculation was shown as follows (Huang et al., 2017):yð Þ ¼
a
TT 0:5 2 da dt da dt 0:5 ! ¼f 0:5ðfð Þ Ga
Þ G 0:5ð Þða
Þ ð19ÞWhen the theoretical data were placed in Eq.(20), then the the-oretical curves were obtained at different
a
values.yð Þ
a
theo¼fð Þ G
a
ð Þa
f 0:5ð Þ G 0:5ð Þ ð20Þ
When the experimental data were put into Eq.(21), then the experimental curves were obtained at different
a
values.yð Þ
a
exp¼ T T0:5 2 da dt da dt 0:5 ! ð21ÞThe Malek method consists of the two plots to determine the reaction model. When the experimental curves are similar to or match the theoretical curves for a given
a
, then the reaction model can be determined (Zou et al., 2017).Mean squared error (MSE) as calculated by Eq.(22)was adopted to select the reaction model and assess the experimental validity.
MSE¼1 n Xn 1 ytheo yexp 2 ð22Þ
where n is sample size; and ytheoand yexpwere calculated using Eqs. (20) and (21), respectively. Based on the model validation, the low MSE values indicate that the predictive model describes the exper-imental data with a better accuracy (Diaz and Phan, 2016). The MSE values > 5 between the theoretical and experimental plots were assumed to indicate a poor model performance.
2.4.5. Thermodynamic parameters
Apart from the kinetic parameters, the calculation of the ther-modynamic ones is essential to the evaluation of the feasibility of the combustion process. The four thermodynamics parameters of A, and changes in entropy (DS), enthalpy (DH) and free Gibbs energy (DG) were also estimated using the following equations (Huang et al., 2018): A¼bEae Ea= RTð Þp ð Þ RT2p ð23Þ
D
H¼ Ea RTa ð24ÞD
G¼ Eaþ RTpln kB Tp h A ð25ÞD
S¼D
HTD
G p ð26Þwhere kBis Boltzmann constant (1.381 1023JK1); h is Planck’s
constant (6.626 1034Js); T
pis temperature of maximum weight
loss rate; Tais temperature at a given
a
; and Eais apparent activa-tion energy at a givena
.3. Results and discussion
3.1. SPL combustion in changing CO2/O2atmospheres
The (D)TG curves of the SPL combustion in the three CO2/O2
(80%/20%, 70%/30% and 60%/40%) atmospheres at 20°C/min are presented inFig. 2. A single peak observed in the range of 450– 800°C corresponded to the fixed carbon combustion. The low vola-tiles and high carbon concentrations of SPL (Table 1) contributed to the peak. The decomposition of inorganic compounds including fluoride and char gasification was slow at above 800°C, and thus, was neglected. With the increased O2concentration, the SPL
com-bustion curves were similar under the oxy-fuel atmospheres at below 800°C. The final mass percentages were 30.59, 31.79 and 32.36 wt% with 20, 30 and 40% O2, respectively. This finding could
be explained by the presence of the higher CO2in the oxy-fuel
atmosphere that most probably favored the burnout of char at above 800°C (Li et al., 2009; Hecht et al., 2012). Therefore, the char burnout contributed to the lower char content.
The estimates of the SPL combustion parameters under the three CO2/O2 atmospheres are presented in Table 2. With the
increased O2concentration, Tbshifted slightly to a lower
tempera-ture. This may be attributed to high reactivity and bimolecular dif-fusivity of O2(Shen et al., 2011). The ignition changes were slight,
and thus, were considered to remain constant for the three atmo-spheres. However, the O2concentrations did not significantly
influ-ence S, Dv, Ciand Cb.
3.2. SPL pyrolysis in CO2and N2atmospheres
The (D)TG curves of the SPL pyrolysis in the CO2and N2
atmo-spheres at 20°C/min are shown inFig. 3. The TG results showed no significant weight loss of SPL at below 800°C and a weight loss at 800–900°C in the CO2atmosphere. Meanwhile, the DTG curve
started to decrease at 800°C. This case confirmed that CO2was a
weaker oxidizing agent than O2. While the carbon combustion
can occur at below 630°C, the heterogeneous reaction of SPL with CO2 delayed the oxidation combustion. CO2gasification through
the Boudouard reaction was slow and highly endothermic. Thus, the Boudouard reaction was required to shift the reaction equilib-rium towards the CO production at the high temperature typically >700°C (Lahijani et al., 2015). At the normal atmospheric pressure, the Boudouard reaction is thermodynamically favorable at above 900°C and initiated at 800 °C (Parvez et al., 2016). Therefore, the carbon content of SPL was probably converted into CO in the high temperature during the Boudouard reaction as follows (Kumari and Vairakannu, 2017; Oh et al., 2018):
C (s) + CO2(g)$ 2CO (g)
D
H = 173 kJ/mol ð27Þ As can be seen inFig. 3b, the reaction rate of the SPL pyrolysis in the N2atmosphere was constant with the temperature whichindi-cated that the decomposition of carbon did not occur. Similar results were also reported by Yuan et al. (2018). However, the Fig. 2. (a) TG and (b) DTG curves of the SPL combustion in the three CO2/O2
atmospheres at 20°C/min.
Table 2
Thermodynamic parameters and combustion characteristics of SPL in the three CO2/O2atmospheres at 20°C/min.
CO2/O2 Ti Rp Tp Tb Ci Dv Cb S
80%/20% 552 6.77 603.0 866.3 0.83 1.23 0.60 4.03
70%/30% 551 6.54 609.5 864.2 0.79 1.11 0.54 3.86
60%/40% 551 6.86 604.8 860.2 0.84 1.15 0.57 4.04
Ti: ignition temperature (°C); Rp: maximum weight loss rate (wt%/min); Tp: peak temperature (°C); Tb: burnout temperature (°C); Ci: ignition index (102(wt%min3)); Dv: volatile matter release index (107(wt%min1°C3)); C
b: burnout index (103(wt%min4)); and S: comprehensive combustibility index (108(wt%2°C3min2)). Fig. 3. (D)TG curves of the SPL combustion in the (a) CO2and (b) N2atmospheres at 20°C/min.
slight mass loss by 4.23 wt% during the whole process in the N2
atmosphere may result from the slight decomposition of Na3AlF6
with the release of fluoride (Zhang et al., 2015). 3.3. SPL combustion in CO2/O2and N2/O2atmospheres
The SPL combustion behavior in the 70% CO2/30% O2and 70%
N2/30% O2 atmospheres were compared using the (D)TG curves
at 20°C/min (Fig. 4). The (D)TG curves in the atmospheres were similar at below 450°C since there was no significant decomposi-tion of fixed carbon, fluoride, and other inorganic compounds of SPL in the lower temperature. The carbon oxidation did not occur in the reaction system during the heating process at below 450°C. As discussed above, N2and CO2did not significantly affect
the SPL decomposition in the lower temperature. There was a negligible volatile inorganic matter during the overall thermal Fig. 4. (a) TG and (b) DTG curves of the SPL combustion in the 70% N2/30% O2and 70% CO2/30% O2atmospheres at 20°C/min.
Table 3
Thermodynamic parameters and combustion characteristics of SPL at 20°C/min.
Atmosphere Ti Rp Tp Tb Ci Dv Cb S
70% CO2/30% O2 551 6.54 609.5 864.2 0.79 1.11 0.54 3.86
70% N2/30% O2 551 7.12 605.9 854.1 0.87 1.36 0.68 4.26
Ti: ignition temperature (°C); Rp: maximum weight loss rate (wt%/min); Tp: peak temperature (°C); Tb: burnout temperature (°C); Ci: ignition index (102(wt%min3)); Dv: volatile matter release index (107(wt%min1°C3)); C
b: burnout index (103(wt%min4)); and S: comprehensive combustibility index (108(wt%2°C3min2)). Fig. 5. (a) TG, (b) DTG and (c) conversion rate of the SPL combustion in the 70% CO2/ 30% O2atmosphere at the three heating rates.
decomposition process of SPL. The TG curve was delayed in the 70% CO2/30% O2 atmosphere when compared to the 70% N2/30% O2
atmosphere at the main decomposition stage of SPL, with the final masses (Mf) of 31.49 and 31.40 wt%, respectively. This indicated
that the char burnout occurred more easily in the CO2/O2 than
N2/O2atmosphere. This case may be related to the occurrence of
the char burnout favored in the presence of CO2 in the CO2/O2
atmosphere (Irfan et al., 2012). At the same O2level, replacing N2
only by CO2was reported to significantly increase the char burnout
(Meng et al., 2013). This result was consistent with our study. The key parameters of the SPL combustion in the 70% CO2/30%
O2and 70% N2/30% O2atmospheres at 20°C/min are presented in Table 3. Compared to the 70% CO2/30% O2atmosphere, the
maxi-mum mass loss rate was observed in the 70% N2/30% O2
atmo-sphere at a slightly lower temperature. As presented inTable 3, S, Cb, Dvand Ciwere higher in the N2/O2than CO2/O2atmosphere.
The fixed carbon of the SPL combustion was delayed in the CO2/O2
atmosphere from 450 to 800°C. This case may be related to the fact
that CO2has a higher specific heat and density than does N2as well
as to the different transport properties due to the different masses of CO2and N2(Irfan et al., 2012).
3.4. Effect of heating rate on SPL combustion
The (D)TG curves of the SPL combustion in the 70% CO2/30% O2
atmosphere at the three heating rates are shown inFig. 5. With the increased heating rate, the TG curves shifted to a higher tempera-ture gradually, with the final masses of 30.88, 31.49 and 33.34 wt% at 15, 20 and 25°C/min, respectively. The reaction and heat trans-fer times were shortened by the increased heating rate which made it less likely for the sample to be completely volatilized in the second reaction (Chen et al., 2017a). The conversion rate grad-ually decreased with the increased heating rate which was in close agreement with the delayed SPL decomposition at the increased heating rate (Fig. 5c).
Table 4
Thermodynamic parameters and combustion characteristics of SPL in the 70% CO2/30% O2atmosphere.
b Ti Rp Tp Tb Ci Dv Cb S
15 547 5.35 598.5 857.2 0.38 1.05 0.22 2.45
20 551 6.54 609.5 864.2 0.79 1.11 0.54 3.86
25 552 7.43 611.0 869.5 1.37 1.12 1.05 5.27
b: heating rate (°C/min); Ti: ignition temperature (°C); Rp: maximum weight loss rate (wt%/min); Tp: peak temperature (°C); Tb: burnout temperature (°C); Ci: ignition index (102(wt%min3)); D
v: volatile matter release index (107(wt%min1°C3)); Cb: burnout index (103(wt%min4)); and S: comprehensive combustibility index (108(wt %2°C3min2)).
Fig. 6. Linear regression analyses to determine activation energy at different conversion rates according to (a) Friedman, (b) Starink, (c) KAS and (d) FWO methods in the 70% CO2/30% O2atmosphere.
Ti, Tpand Tbincreased with the increased heating rate (Table 4).
The reason for these findings may be related to the difference between external and internal temperatures of the sample in response to the rapidly increased temperature of the reaction sys-tem with the increased heating rate, non-uniform heat transfer, and the delayed volatility of the substances (Herce et al., 2014). However, at the higher heating rate, a certain temperature was achieved in a shorter time with the higher thermal energy to favor the heat transfer from the surrounding to the inside of the sample. Therefore, the SPL combustion resulted in a faster mass loss rate with the increased heating rate (Xie and Ma, 2013). Meanwhile, the increased Dv, Ci, S and Cbwith the increasedb indicated a more
concentrated combustion as well as improved ignition, combustion and burnout performances.
3.5. Kinetic analysis of activation energy
Apparent activation energy is the energy barrier for a chemical reaction to occur (Yuan et al., 2017) and an important indicator to evaluate the combustion performance. Combustion involves a set of complicated chemical reactions where the oxidation of the fuel not only releases heat but also produces new chemical species, thus rendering Eavalues difficult to calculate (Deng et al., 2016).
Given the unstable start and end of the reaction due to the complex chemical and physical structures of SPL, Eawas estimated for the
decomposition at the conversion degrees of 0.10–0.70 according to the Friedman, Starink (differential form), FWO and KAS (integral form) methods in the CO2/O2 (80%/20%, 70%/30% and 60%/40%)
atmospheres at the three heating rates (Figs. 6 and 7).
The fluctuated Eavalues estimated according to the Friedman
method were lower by 1.21–1.49 times than those according to the Starink, KAS and FWO methods. This case confirmed that Friedman method emphasized the relationship between the reac-tion rate (d
a
/dt) and 1/T as a function of the instantaneous conver-sion rate which was prone to the experimental noise (Dhyani et al., 2017). Furthermore, a minor change in the power of the tempera-ture term in the Starink’s equation led to an increment in Eavaluesby less than 0.3% (Dhyani et al., 2017). Therefore, the Eaestimates
based on the Starink method were used to calculate A,DH,DS, and
DG.
The average activation energy increased when the O2
concen-tration grew beyond 30% in the oxy-fuel atmosphere. Similar results were obtained in the recent studies (Huang et al., 2016; Xie et al., 2018b). This may be because Ea was affected by a
decrease in activated molecule proportion, diffusion limitation, and inorganic impurities during the combustion process (Fang et al., 2006). The increased O2 concentration was reported to
increase the heat release from the semi-coke oxidization, and thus, the surface temperature of semi-coke (Chen et al., 2011). Semi-coke structure also expanded the size of grain and increased the ash content with the increased final temperature (Chen et al., 2011). Therefore, Eaincreased with the increased O2concentration
(Chen et al., 2011). The lower Eavalue in the initial stage may be
the result of the decomposition of some extractives with small molecules. The activation energy peaked (181.25 kJ/mol) at the conversion rate of 0.2 that corresponded to the maximum mass loss stage during which more energy may be needed to provide the thermal decomposition of fixed carbon. Overall, Eaincreased
in the start of the thermal decomposition process and decreased after the conversion rate of 0.25.
3.6. Kinetic analysis of reaction model
The uses of the conversion rate of 0.5 as the reference point and of the theoretical data of
a
i, and y(a
i) in Eq.(20)yielded 41stan-dard curves according to the reaction function in Supplementary Table S1. The use of the experimental data of
a
i, and y(a
i) in Eq. (21) helped to obtain the experimental curves. A comparison of the theoretical versus experimental plots at the different conver-sion rates served to determine the most likely mechanism function of the reaction.Fig. 8a shows the theoretical curves, while the experimental curves in the 80% CO2/20% O2, 70% CO2/30% O2and60% CO2/40% O2atmospheres at the three heating rates are shown
inFig. 8b–d, respectively.
As shown in Fig. 8b–d, the experimental and theoretical data better matched in the range of 0.25–0.70 which may be attributed to the occurrence of a single-step reaction. The inadequate match between the experimental and theoretical data in the range of
0.10–0.25 may result from the involvement of other complex chemical reactions. All the MSE values were lower than 0.015 in response to the different atmospheres and heating rates (Table 5). All the R2values between the theoretical and experimental plots
were above 0.894 (Table 5). All these findings indicated a better validation and accuracy of the reaction model (Diaz and Phan, 2016). In conclusion, the reaction model was best described by the following function of f(
a
) = 1/2a
, G(a
) = lna
2 in the CO2/O2
(80%/20%, 70%/30% and 60%/40%) atmospheres with the
a
value of 0.25–0.7 at the three heating rates.3.7. Thermodynamic analyses
The apparent activation energy estimated from the Starink method was used to calculate A values of the SPL combustion under the three CO2/O2 (80%/20%, 70%/30% and 60%/40%)
atmo-spheres at 20°C/min. The low A values (h1 09) may imply surface
reactions in most cases but a closed complex if the reactions did not depend on surface area. However, the high A values (109)
indicated a simple complex and a higher reactive system (Yuan
Fig. 8. Theoretical (a) and experimental curves in the (b) 80% CO2/20% O2, (c) 70% CO2/30% O2and (d) 60% CO2/40% O2atmospheres at different conversion rates according to the function in Supplementary Table S1.
Table 5
Mean squared error (MSE) and coefficients of determination (R2
) for reaction model.
15°C/min 20°C/min 25°C/min
CO2/O2 MSE R2 MSE R2 MSE R2
80%/20% 0.015 0.898 0.011 0.910 0.006 0.907
70%/30% 0.011 0.925 0.008 0.912 0.006 0.904
et al., 2017). In particular, when the A values were between 1010
and 1012, the activated complex was probably more restricted in
terms of rotation than the initial reagent (Yuan et al., 2017). The reason behind this finding is that the A values in the 70%
CO2/30% O2 atmosphere ranged widely from 1.64 106 to
3.10 1010, with its mean lowest value of 8.24 109, and were
above 1010 in the range of 0.10–0.25. This suggested that more
reaction energy was required with the start of the fixed carbon decomposition (0.10 <
a
< 0.25).TheDH,DS,DG and Eavalues of the SPL decomposition are
shown inFig. 9. TheDH values (Fig. 9b) indicated the energy differ-ence between the reagent and the activated complex. The lowDH value indicated the favored formation of the activated complex (Chen et al., 2017c). As shown inFig. 9b,DH was in close agree-ment with Eaand peaked (DH = 174.45 kJ/mol) around the start
of the SPL decomposition in the solid phase. The case is most likely related to the relatively high degrees of reaction and conversion at the start of the SPL combustion where more heat was required (Yuan et al., 2017). With the increased conversion rate, DH decreased illustrating the energy between reagent and the acti-vated simple complex (Chen et al., 2017c). The fixed carbon
com-bustion in this stage provided heat energy. Decreased DH
indicated less heat energy required for a reaction system which was in close agreement with the activation energy changes. The
average DH values in the CO2/O2 (80%/20%, 70%/30% and
60%/40%) atmospheres at 20°C/min showed that the least heat requirement was by the SPL decomposition in the 70% CO2/30%
O2atmosphere.
The lowDS values (Fig. 9c) for a reaction system mean that the substance undergoes physical or chemical transformations reach-ing a new state near its decomposition equilibrium. The highDS
values are indicative of a state far from its own thermodynamic equilibrium. In this case, the reactivity is high, and the system can react faster to produce the activated complex which results in the short reaction times (Dhyani et al., 2017). Therefore, the higher the DS values are, the higher the system reactivity is (Yuan et al., 2017). The averageDS values demonstrated that the higher reactivity of the overall thermal decomposition process was in the 70% CO2/30% O2atmosphere. As shown inFig. 9c, with
the increased conversion degree, theDS values rose from61.11 to 144.15 J/mol which indicated the increased reactivity. In particu-lar, the lowestDS value (61.11 J/mol) was observed at the con-version rate of 0.25, thus resulting in a lower reactivity of the reaction system which agreed with the highest Eaat the conversion
rate of 0.25.DS increased at above 0.25 and peaked (144.15 J/-mol) at 0.65. This peak appeared to correspond to the main decom-position region of the solid phase during which enough heat was provided. Our results indicated the higher reactivity of the reaction system. Less heat energy required agreed with the activation ener-gies.Song et al. (2017)in a similar study reported that activation energy increased with the decreased reactivity.
TheDG values revealed the increased total energy in the reac-tion system with the approach of the reagents and the formareac-tion of the activated complex (Yuan et al., 2017). Fig. 9d shows increasedDG at above 0.2 which confirmed that heat energy pro-vided to the reaction system at the high temperatures was surplus (Dhyani et al., 2017). Eadecreased in the stage as a result of the
fixed carbon combustion and the increased heat energy of the sys-tem. TheDG values were higher in the 70% CO2/30% O2than CO2/O2
(80%/20% and 60%/40%) atmospheres. Overall, the SPL combustion provided more energy in the 70% CO2/30% O2 than CO2/O2
(80%/20% and 60%/40%) atmospheres. The changed patterns of
these thermodynamic parameters indicated the complex reactions during the SPL decomposition which agreed with the activation energy changes.
3.8. TG-MS analyses
The main gases released from the SPL combustion were CO2(m/
z = 44), CO (m/z = 28), NOx(m/z = 46 and 30), HCN (m/z = 27), and
HF (m/z = 20) as determined using the real-time TG-MS analyses. The advantage of the EI source was that the analyte structures and their names can be obtained from the NIST library according
to their respective ratios of mass to charge (m/z) (Striugas et al.,
2017). It should be noted that some ions could belong to various compounds when reported (López-González et al., 2017), such as ions with m/z = 28 that are related to the CO or N2 evolution. It
should be noted that the higher CO2concentration presented for
the O2diffusion to the surface of the fuel, thus incompletely
oxidiz-ing the fuel at the higher temperature despite the increased oxygen content. CO concentration rose in the near burner region due to the char gasification and lowered NO emission (Giménez-López et al., 2010). Meanwhile, higher CO2 concentration further restrained
the HCN oxidation at the higher temperature during the combus-tion (Barbas et al., 2015). He instead of CO2may have slightly
influ-enced the gas release profile during the SPL combustion. In order to monitor the CO2emissions and avoid CO2interference with MS
sig-nal, He instead of CO2(an inert atmosphere) was used.
Fig. 10shows the release of the major gases from the SPL com-bustion in the 80% He/20% O2atmosphere in the range of 450–
900°C. As shown in Fig. 10a, the highest absorbance intensities of CO2(m/z = 44) and CO (m/z = 28) occurred at about 610°C
corre-sponding to the maximum mass loss rate of the fixed carbon com-bustion stage. This may be due to the oxidation of fixed carbon and the releases of CO2and CO in this stage (Ye et al., 2018).Fig. 10b
shows the curves of absorbance intensity of NOx (m/z = 46 and
30). Although SPL was low in N (Table 1), an absorbance intensity <2 1010of NO
xwas observed during the main decomposition
zone. It appeared to come partly from the thermal decomposition of inorganic compounds. The absorbance intensity of HCN (m/ z = 27) peaked at around 620°C and decreased with the increased conversion rate. It may be related to HCN produced with the decomposition of cyanides and further converted into N2and CO2
(Li and Chen, 2010).
No peak was observed for absorbance intensity of HF (m/z = 20) at below 800°C (Fig. 10c). However, HF was observed slightly in the range of 800–900°C. This case may result from the slow decomposition of the fluoride content of SPL in the lower temper-ature and the release of fluoride gas in the higher tempertemper-ature from the decomposition of Na3AlF6(Zhang et al., 2015). In
conclu-sion, CO2was the main gas generated from the SPL combustion.
Furthermore, cyanides in SPL were probably converted into N2
and CO2and reduced the toxicity of the SPL residues when
com-pared to the initial SPL.
4. Conclusion
Our experimental results pointed to a peak observed in the range of 450–800°C during the SPL thermal decomposition in the CO2/O2and N2/O2atmospheres. The higher CO2proportion of
the oxy-fuel atmosphere favored the char burnout at above 800°C, thus improving the process of the residue burnout. The char burnout easily occurred in the 70% CO2/30% O2atmosphere when
compared to the 70% N2/30% O2atmosphere. The kinetic and
ther-modynamic analyses indicated the promising performance of the SPL combustion in the 70% CO2/30% O2atmosphere. The pyrolysis
DTG curves suggested that the initial reaction of the char gasifica-tion by CO2was observed at 800°C in the pure CO2atmosphere,
but not in the N2 atmosphere. Overall, the TG-MS experiments
showed CO, HCN, NOx and HF as the major gaseous pollutants
evolved from the combustion process. The function of f(
a
) = 1/2a
, G(a
) = lna
2in the three CO2/O2(80%/20%, 70%/30% and 60%/40%)
atmospheres with the
a
value of 0.25–0.70 at the three heating rates was determined as the best-fit reaction model with the R2values above 0.894 and the MSE values below 0.015. With proper operational conditions, the SPL combustion in the 70% CO2/30%
O2 (oxy-fuel) atmosphere appeared to be a desirable disposal
method for SPL in controlling CO2emissions.
Fig. 10. Gas release curves of the SPL combustion in the 80% He/20% O2atmosphere based on mass-spectrometric analyses: (a) m/z = 44 and 28, (b) m/z = 46 and 30, and (c) m/z = 27 and 20.
Acknowledgements
This work was financially supported by the National Natural Science Foundation of China (No. 51608129), the Science and Tech-nology Planning Project of Guangdong Province, China (No.
2019B020208017, 2018A050506046; 2017A040403044;
2017A040403047; 2016A050502059) and Natural Science Founda-tion of Guangdong Province of China (2017A030313261). Appendix A. Supplementary material
Supplementary data to this article can be found online at
https://doi.org/10.1016/j.wasman.2019.01.047. References
Barbas, M., Costa, M., Vranckx, S., Fernandes, R., 2015. Experimental and chemical kinetic study of CO and NO formation in oxy-methane premixed laminar flames doped with NH3. Combust. Flame 162, 1294–1303.
Birry, L., Leclerc, S., Poirier, S., 2016. The LCL&L Process: a sustainable solution for the treatment and recycling of spent potlining. In: Light Metals 2016, John Wiley & Sons, Inc., pp. 467–471.
Bu, C., Liu, D., Chen, X., Pallarès, D., Gómez-Barea, A., 2014. Ignition behavior of single coal particle in a fluidized bed under O2/CO2and O2/N2atmospheres: a combination of visual image and particle temperature. Appl. Energ. 115, 301– 308.
Cai, Z., Ma, X., Fang, S., Yu, Z., Lin, Y., 2016. Thermogravimetric analysis of the co-combustion of eucalyptus residues and paper mill sludge. Appl. Therm. Eng. 106, 938–943.
Chen, C., Ma, X., Liu, K., 2011. Thermogravimetric analysis of microalgae combustion under different oxygen supply concentrations. Appl. Energ. 88, 3189–3196.
Chen, G., He, S., Cheng, Z., Guan, Y., Yan, B., Ma, W., Leung, D.Y.C., 2017a. Comparison of kinetic analysis methods in thermal decomposition of cattle manure by themogravimetric analysis. Bioresour. Technol. 243, 69–77.
Chen, J., Liu, J., He, Y., Huang, L., Sun, S., Sun, J., Chang, K., Kuo, J., Huang, S., Ning, X., 2017b. Investigation of co-combustion characteristics of sewage sludge and coffee grounds mixtures using thermogravimetric analysis coupled to artificial neural networks modeling. Bioresour. Technol. 225, 234–245.
Chen, J., Mu, L., Cai, J., Yao, P., Song, X., Yin, H., Li, A., 2015. Pyrolysis and oxy-fuel combustion characteristics and kinetics of petrochemical wastewater sludge using thermogravimetric analysis. Bioresour. Technol. 198, 115–123.
Chen, J., Wang, Y., Lang, X., Ren, X., Fan, S., 2017c. Comparative evaluation of thermal oxidative decomposition for oil-plant residues via thermogravimetric analysis: thermal conversion characteristics, kinetics, and thermodynamics. Bioresour. Technol. 243, 37–46.
Chen, J., Xie, C., Liu, J., He, Y., Xie, W., Zhang, X., Chang, K., Kuo, J., Sun, J., Zheng, L., Sun, S., Buyukada, M., Evrendilek, F., 2018. Co-combustion of sewage sludge and coffee grounds under increased O2/CO2 atmospheres: thermodynamic characteristics, kinetics and artificial neural network modeling. Bioresour. Technol. 250, 230–238.
Cheng, X., Zhang, M., Wang, Z., Xu, G., Ma, C., 2018. IR and kinetic study of sewage sludge combustion at different oxygen concentrations. Waste Manage. 74, 279– 287.
Courbariaux, Y., Chaouki, J., Guy, C., 2004. Update on spent potliners treatments kinetics of cyanides destruction at high temperature. Ind. Eng. Chem. Res. 43, 5828–5837.
Doyle, C.D., 1961. Kinetic analysis of thermogravimetric data. J. Appl. Polym. Sci. 5, 285–292.
Deng, S., Wang, X., Tan, H., Mikulcˇic´, H., Yang, F., Li, Z., Duic´, N., 2016. Thermogravimetric study on the Co-combustion characteristics of oily sludge with plant biomass. Thermochim. Acta 633, 69–76.
Dhyani, V., Kumar, J., Bhaskar, T., 2017. Thermal decomposition kinetics of sorghum straw via thermogravimetric analysis. Bioresour. Technol. 245, 1122–1129.
Diaz, L.S., Phan, A.N., 2016. Kinetic study of municipal plastic waste. Int. J. Hydrogen. Energ. 41, 16352–16364.
Fang, M.X., Shen, D.K., Li, Y.X., Yu, C.J., Luo, Z.Y., Cen, K.F., 2006. Kinetic study on pyrolysis and combustion of wood under different oxygen concentrations by using TG-FTIR analysis. J. Anal. Appl. Pyrol. 77, 22–27.
Fernandez-Lopez, M., Pedrosa-Castro, G.J., Valverde, J.L., Sanchez-Silva, L., 2016. Kinetic analysis of manure pyrolysis and combustion processes. Waste Manage. 58, 230–240.
Flynn, 1966. A quick, direct method for the determination of activation energy from thermogravimetric data. J. Polym. Sci. B. Polym. Lett. 4, 323–328.
Friedman, H.L., 1964. Kinetics of thermal degradation of char-forming plastics from thermogravimetry. Application to a phenolic plastic. J. Polym. Sci. C. Polym. Symp. 6, 183–195.
Gao, L., Mostaghel, S., Ray, S., Chattopadyay, K., 2016. Using SPL (Spent Pot-Lining) as an alternative fuel in metallurgical furnaces. Metall. Mater. Trans. E 3, 179– 188.
Giménez-López, J., Millera, A., Bilbao, R., Alzueta, M., 2010. HCN oxidation in an O2/ CO2atmosphere: an experimental and kinetic modeling study. Combust. Flame 157, 267–276.
Hao, Z., Yang, B., Jahng, D., 2018. Combustion characteristics of biodried sewage sludge. Waste Manage. 72, 296–305.
Hecht, E.S., Shaddix, C.R., Geier, M., Molina, A., Haynes, B.S., 2012. Effect of CO2and steam gasification reactions on the oxy-combustion of pulverized coal char. Combust. Flame 159, 3437–3447.
Herce, C., de Caprariis, B., Stendardo, S., Verdone, N., De Filippis, P., 2014. Comparison of global models of sub-bituminous coal devolatilization by means of thermogravimetric analysis. J. Therml. Anal. Calorim. 117, 507–516.
Holywell, G., Breault, R., 2013. An overview of useful methods to treat, recover, or recycle spent potlining. JOM 65, 1441–1451.
Hu, Y., Yan, J., 2012. Characterization of flue gas in oxy-coal combustion processes for CO2capture. Appl. Energ. 90, 113–121.
Huang, L., Liu, J., He, Y., Sun, S., Chen, J., Sun, J., Chang, K., Kuo, J., Ning, X., 2016. Thermodynamics and kinetics parameters of co-combustion between sewage sludge and water hyacinth in CO2/O2atmosphere as biomass to solid biofuel. Bioresour. Technol. 218, 631–642.
Huang, L., Xie, C., Liu, J., Zhang, X., Chang, K., Kuo, J., Sun, J., Xie, W., Zheng, L., Sun, S., Buyukada, M., Evrendilek, F., 2018. Influence of catalysts on co-combustion of sewage sludge and water hyacinth blends as determined by TG-MS analysis. Bioresour. Technol. 247, 217–225.
Huang, Z., Deng, Z., Chen, D., He, F., Liu, S., Zhao, K., Wei, G., Zheng, A., Zhao, Z., Li, H., 2017. Thermodynamic analysis and kinetic investigations on biomass char chemical looping gasification using Fe-Ni bimetallic oxygen carrier. Energy 141, 1836–1844.
Irfan, M.F., Arami-Niya, A., Chakrabarti, M.H., Wan Daud, W.M.A., Usman, M.R., 2012. Kinetics of gasification of coal, biomass and their blends in air (N2/O2) and different oxy-fuel (O2/CO2) atmospheres. Energy 37, 665–672.
Jayaraman, K., Kok, M.V., Gokalp, I., 2017. Thermogravimetric and mass spectrometric (TG-MS) analysis and kinetics of coal-biomass blends. Renew. Energ. 101, 293–300.
Kumari, G., Vairakannu, P., 2017. Laboratory scale studies on CO2 oxy-fuel combustion in the context of underground coal gasification. J. CO2Util. 21, 177–190.
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. Energ. 114, 227–237.
López-González, D., Parascanu, M.M., Fernandez-Lopez, M., Puig-Gamero, M., Soreanu, G., Avalos-Ramírez, A., Valverde, J.L., Sanchez-Silva, L., 2017. Effect of different concentrations of O2under inert and CO2atmospheres on the swine manure combustion process. Fuel 195, 23–32.
Li, N., Xie, G., Xu Wang, Z., Hou, Y., Xing Li, R., 2014. Recycle of spent potlining with low carbon grade by floatation. Adv. Mater. Res. 881–883, 1660–1664.
Li, Q., Zhao, C., Chen, X., Wu, W., Li, Y., 2009. Comparison of pulverized coal combustion in air and in O2/CO2mixtures by thermo-gravimetric analysis. J. Anal. Appl. Pyrol. 85, 521–528.
Li, W., Chen, X., 2010. Development status of processing technology for spent potlining in China. Light Metals 4, 1064–1066.
Lisbona, D.F., Somerfield, C., Steel, K.M., 2013. Leaching of spent pot-lining with aluminium nitrate and nitric acid: effect of reaction conditions and thermodynamic modelling of solution speciation. Hydrometallurgy 134–135, 132–143.
Lahijani, P., Zainal, Z.A., Mohammadi, M., Mohamed, A.R., 2015. Conversion of the greenhouse gas CO2to the fuel gas CO via the Boudouard reaction: a review. Renew. Sust. Energ. Rev. 41, 615–632.
Meng, F., Yu, J., Tahmasebi, A., Han, Y., 2013. Pyrolysis and combustion behavior of coal gangue in O2/CO2and O2/N2mixtures using thermogravimetric analysis and a drop tube furnace. Energ. Fuel 27, 2923–2932.
Mondal, M.K., Balsora, H.K., Varshney, P., 2012. Progress and trends in CO2capture/ separation technologies: a review. Energy 46, 431–441.
Maroušek, J., Vochozka, M., Plachy´, J., Zˇák, J., 2016. Glory and misery of biochar. Clean. Technol. Environ. 19, 311–317.
Oh, G., Ra, H.W., Yoon, S.M., Mun, T.Y., Seo, M.W., Lee, J.G., Yoon, S.J., 2018. Gasification of coal water mixture in an entrained-flow gasifier: effect of air and oxygen mixing ratio. Appl. Therm. Eng. 129, 657–664.
Ospina, G., Hassan, M.I., 2017. Spent pot lining characterization framework. JOM 69, 1639–1645.
Parvez, A.M., Mujtaba, I.M., Pang, C., Lester, E., Wu, T., 2016. Effect of the addition of different waste carbonaceous materials on coal gasification in CO2atmosphere. Fuel. Process. Technol. 149, 231–238.
Shen, D., Gu, S., Jin, B., Fang, M., 2011. Thermal degradation mechanisms of wood under inert and oxidative environments using DAEM methods. Bioresour. Technol. 102, 2047–2052.
Shi, Z., Li, W., Hu, X., Ren, B., Gao, B., Wang, Z., 2012. Recovery of carbon and cryolite from spent pot lining of aluminium reduction cells by chemical leaching. T. Nonferr. Metal. Soc. 22, 222–227.
Song, Z., Huang, X., Luo, M., Gong, J., Pan, X., 2017. Experimental study on the diffusion–kinetics interaction in heterogeneous reaction of coal. J. Therm. Analy. Calori. 129, 1625–1637.
Striugas, N., Skvorcˇinskiene˙, R., Paulauskas, R., Zakarauskas, K., Vorotinskiene˙, L., 2017. Evaluation of straw with absorbed glycerol thermal degradation during pyrolysis and combustion by TG-FTIR and TG-GC/MS. Fuel 204, 227–235.
Tang, Y., Ma, X., Lai, Z., Fan, Y., 2015. Thermogravimetric analyses of co-combustion of plastic, rubber, leather in N2/O2and CO2/O2atmospheres. Energy 90, 1066–1074. Wang, C., Zhang, X., Liu, Y., Che, D., 2012. Pyrolysis and combustion characteristics
of coals in oxyfuel combustion. Appl. Energ. 97, 264–273.
Wang, Z., Hong, C., Xing, Y., Li, Y., Feng, L., Jia, M., 2018. Combustion behaviors and kinetics of sewage sludge blended with pulverized coal: with and without catalysts. Waste Manage. 74, 288–296.
Xiao, J., Yuan, J., Tian, Z., Yang, K., Yao, Z., Yu, B., Zhang, L., 2018. Comparison of ultrasound-assisted and traditional caustic leaching of spent cathode carbon (SCC) from aluminum electrolysis. Ultrason. Sonochem. 40, 21–29.
Xie, C., Liu, J., Zhang, X., Xie, W., Sun, J., Chang, K., Kuo, J., Xie, W., Liu, C., Sun, S., Buyukada, M., Evrendilek, F., 2018a. Co-combustion thermal conversion characteristics of textile dyeing sludge and pomelo peel using TGA and artificial neural networks. Appl. Energ. 212, 786–795.
Xie, W., Huang, J., Liu, J., Zhao, Y., Chang, K., Kuo, J., He, Y., Sun, J., Zheng, L., Xie, W., Sun, S., Buyukada, M., Evrendilek, F., 2018b. Assessing thermal behaviors and kinetics of (co-)combustion of textile dyeing sludge and sugarcane bagasse. Appl. Therm. Eng. 131, 874–883.
Xie, Z., Ma, X., 2013. The thermal behaviour of the co-combustion between paper sludge and rice straw. Bioresour. Technol. 146, 611–618.
Ye, C., Wang, Q., Yu, L., Luo, Z., Cen, K., 2018. Characteristics of coal partial gasification experiments on a circulating fluidized bed reactor under CO2/O2 atmosphere. Appl. Therm. Eng. 130, 814–821.
Yuan, J., Xiao, J., Li, F., Wang, B., Yao, Z., Yu, B., Zhang, L., 2018. Co-treatment of spent cathode carbon in caustic and acid leaching process under ultrasonic assisted for preparation of SiC. Ultrason. Sonochem. 41, 608–618.
Yuan, X., He, T., Cao, H., Yuan, Q., 2017. Cattle manure pyrolysis process: kinetic and thermodynamic analysis with isoconversional methods. Renew. Energ. 107, 489–496.
Zhang, B., Zhao, J., Liang, L., Shi, R., Tang, W.T., Li, X., 2015. Research on reaction characteristics of spent pot lining in aerobic and anaerobic environment. Light Metals 5, 27–30.
Zhuo, Z., Liu, J., Sun, S., Sun, J., Kuo, J., Chang, K., Fu, J., Wang, Y., 2017. Thermogravimetric characteristics of textile dyeing sludge, coal and their blend in N2/O2and CO2/O2atmospheres. Appl. Therm. Eng. 111, 87–94. Zou, C., Ma, C., Zhao, J., Shi, R., Li, X., 2017. Characterization and non-isothermal
kinetics of Shenmu bituminous coal devolatilization by TG-MS. J. Analy. Appl. Pyroly. 127, 309–320.
