You are on page 1of 8

877682

research-article2019
WMR0010.1177/0734242X19877682Waste Management & ResearchYao et al.

Original Article

Waste Management & Research

Comparative study on the pyrolysis


1­–8
© The Author(s) 2019
Article reuse guidelines:
kinetics of polyurethane foam from sagepub.com/journals-permissions
DOI: 10.1177/0734242X19877682
https://doi.org/10.1177/0734242X19877682

waste refrigerators journals.sagepub.com/home/wmr

Zhitong Yao1 , Shaoqi Yu1, Weiping Su1, Weihong Wu1,


Junhong Tang1 and Wei Qi2

Abstract
Thermal treatment offers advantages of significant volume reduction and energy recovery for the polyurethane foam from waste
refrigerators. In this work, the pyrolysis kinetics of polyurethane foam was investigated using the model-fitting, model-free and
distributed activation energy model methods. The thermogravimetric analysis indicated that the polyurethane foam decomposition
could be divided into three stages with temperatures of 38°C–400°C, 400°C–550°C and 550°C–1000°C. Peak temperatures for
the major decomposition stage (<400°C) were determined as 324°C, 342°C and 344°C for heating rates of 5, 15 and 25 K min-1,
respectively. The activation energy (Eα) from the Friedman, Flynn–Wall–Ozawa and Tang methods increased with degree of
conversion (α) in the range of 0.05 to 0.5. The coefficients from the Flynn–Wall–Ozawa method were larger and the resulted Eα
values fell into the range of 163.980–328.190 kJ mol-1 with an average of 206.099 kJ mol-1. For the Coats–Redfern method, the
diffusion models offered higher coefficients, but the E values were smaller than that from the Flynn–Wall–Ozawa method. The
Eα values derived from the distributed activation energy model method were determined as 163.536–334.231 kJ mol-1, with an
average of 206.799 kJ mol-1. The peak of activation energy distribution curve was located at 205.929 kJ mol-1, consistent with
the thermogravimetric results. The Flynn–Wall–Ozawa and distributed activation energy model methods were more reliable for
describing the polyurethane foam pyrolysis process.

Keywords
Polyurethane foam, electronic waste, pyrolysis kinetics, model-fitting and -free method, distributed activation energy model
Received 21st June 2019, accepted 1st September 2019 by Editor in Chief Arne Ragossnig.

Introduction Recent treatment technologies for PUR can be grouped into


four categories: Physical (Abdel Rahman et al., 2019; Junco
The rapid rate of technologies innovations and consequent fre- et al., 2018), chemical (Sheel and Pant, 2018), biological
quent replacement of electrical and electronic equipment has (Magnin et al., 2019) and thermal (Li et al., 2016). Physical
resulted in shorter lifespans for these products (Yao et al., 2017, treatment could destroy the chemical structure of PUR to some
2018; Zeng et al., 2018). The recent report from United Nations extent, resulting into a degradation of performance for the regen-
University revealed that 44.7 million tonnes of electronic waste erated products. For the chemical treatment of alcoholysis, it is
was generated in 2016 and it will reach 52.2 million tonnes by difficult to separate various chemical components. Biological
2021. According to the report ‘White paper on WEEE recycling treatment is time-consuming and inefficient. Thermal treatment,
industry in China 2017’ (China Household Electric Appliance especially pyrolysis, is an alternative that offers advantages of
Research Institute, 2018), the theoretical amount of scrapped significant volume reduction and energy recovery (Cheng et al.,
residential equipment reached 1.25 billion units in 2017. The pro- 2019; Qi et al., 2019; Yu et al., 2019). Modelling the thermal
portion of dismantled TV sets in formal recycling plants decreased
from 70% in 2015 to 61% in 2017, whereas, the proportion of
refrigerators increased 81.8% over that in 2015. During the 1Collegeof Materials and Environmental Engineering,
refrigerators dismantling, they are first crushed under negative Hangzhou Dianzi University, Hangzhou, China
pressure, preventing the uncontrolled release of chlorofluorocar- 2Guangzhou Institute of Energy Conversion, Chinese

bons (Montzka et al., 2018). The metals, glass and plastics are Academy of Sciences, Guangzhou, China

then mechanically separated using eddy currents and electrostatic


separation, leaving low-value polyurethane foam (PUR) Corresponding author:
(Tantisattayakul et al., 2018). PUR is light weight but bulky, so Zhitong Yao, College of Materials and Environmental
Engineering, Hangzhou Dianzi University, Hangzhou 310018,
that it takes up a lot of storage space. It is imperious to develop China.
an alternative treatment technology for PUR. Email: sxyzt@126.com
2 Waste Management & Research 00(0)

processes is significant for understanding their behaviour at an model-free method, mostly employed for the study of solids
industrial scale (Fernandez-Lopez et al., 2016; Qi et al., 2018) decomposition (Khawam and Flanagan, 2005). Model-fitting
and will provide references for the reactors designing and method is commonly applied, attributing to its availability of
parameters optimising. However, there are sparse researches on determining the kinetic parameters directly. Different reaction
this. Xu et al. (2012) studied the pyrolysis behaviour and kinet- models have been proposed (Aboulkas and El Bouadili, 2010;
ics of PUR using the thermal gravimetric analysis (TGA)/ Chong et al., 2017; Khawam and Flanagan, 2006; Ma et al.,
Fourier transform infrared (FT-IR) technique. The evolution 2018b) and kinetic parameters can be determined based on these
characteristics of gaseous products was mainly focused, how- models. For the model-free method, there are no previous
ever, the pyrolysis was preliminarily studied using the distribu- assumptions about the reaction model and allows the kinetic
tion activation energy model (DAEM). Rogers and Ohlemiller parameters to be calculated as a function of α (degree of conver-
(1981) investigated the pyrolysis kinetic of PUR. The raw mate- sion). Different model-free methods, such as Friedman, FWO,
rial was flexible PUR, which was distinct from the rigid PUR Kissinger–Akahira–Sunose, Starink and Tang methods (Hardi
from waste refrigerators. Therefore, in this work, a series of et al., 2018; Srivastava et al., 2017), have been proposed for the
thermogravimetric experiments of PUR were conducted. Based kinetic study.
on the thermogravimetric analysis, the activation energies (E) The thermogravimetric results can be expressed as a function
were calculated using three model-free methods, including of α, which is defined based on the mass change of solids:
Friedman, Flynn–Wall–Ozawa (FWO) and Tang. The model-
fitting method Coats–Redfern (CR) and DAEM were also m0 − mt
α= (1)
applied to comparatively study the pyrolysis kinetics. m0 − m f

where m0 and mf refer to the initial and final weight of samples,


Experimental mt represents the instantaneous mass at time of t.
The conversion rate (dα/dt) can be expressed as:
Materials
dα dα
The PUR waste was collected from a typical electrical and =β = k (T ) × f (α ) (2)
electronic equipment recycling plant located in Shandong dt dT
province, China. It was compacted during refrigerator disman- where β is the heating rate (dT/dt K min-1); k(T) refers to the reac-
tling and thus crushed prior to use. The major chemical compo- tion rate constant; f(α) represents the kinetic model function
sitions of PUR were determined and consisted of following (Vyazovkin et al., 2011).
elements (in wt.%): C 61.2, O 26.1, H 6.2, N 6.5. Fourier trans- The k(T) can be described according to the Arrhenius equation:
form infrared (FT-IR) analysis indicated that the PUR sample E

could be ascribed to polyester polyurethane (Supplemental f (T ) = Ae RT (3)
Figure S1, available online).
where A refers to the pre-exponential factor (min-1); E is the activa-
tion energy (kJ mol-1); T and R represent the absolute temperature
Experimental procedure (K) and universal gas constant (8.314 J (mol K)-1), respectively.
The FT-IR analysis of the PUR sample was recorded in the range Under constant temperature ramp conditions, combining
of 400–4000 cm-1 by a NEXUS 670 FT-IR spectrometer (Thermo equations (2) and (3) yields:
Nicolet Corporation, USA) with a resolution of 2 cm-1. The ther-
dα A − RTE
mogravimetric experiments were carried out using the simultane- = e f (α ) (4)
ous thermal analyser (STA) 6000 (PerkinElmer, USA). During dT β
the thermogravimetric experiment, approximately 2 mg of the
sample was placed on the crucibles and heated from 38°C to CR method
1000°C under a nitrogen gas (purity > 99.998%) flow of The CR method is an integral model-fitting method that uses
20 mL min-1. The tests were repeated at three heating rates of 5, asymptotic series expansion for estimation of temperature inte-
15 and 25 K min-1. The sample weight and temperature changes gral. The integral form of the reaction model can be obtained by
were recorded in the thermogravimetric (TG) and derivative ther- integrating equation (4) as:
mogravimetric (DTG) profiles.
dα A −

AEα ∞ e − x AEα
g (α ) = ∫ α0 = ∫ T0 e RT dT = ∫ x 2 x
d = p ( x) (5)
Kinetic modelling f (α ) β βR x βR

The basic theories of kinetic models for solid fuels is presented where x equals to (Eα/RT). The p(x) is temperature integral and
here and more details can be found elsewhere (Brown et al., has no analytical form. Therefore, equation (5) can only be solved
2000; Criado, 1978; Papari and Hawboldt, 2015; Vyazovkin using either numerical integration or approximations to deal with
et al., 2011). They are two major methods, model-fitting and complex integral (Vyazovkin et al., 2011).
Yao et al. 3
−2 − x
By introducing an approximation p ( x) = x e (20⩽x⩽50) and integrated it into equation (8), which yields:
proposed by Murray and White (1955) into equation (5), the rela- β Eα
ln = −1.00145033 +C (11)
tion between heating rate and inverse temperature becomes: T 1.894661
RT

ART 2 Eα
2 RT − RT where C is a constant and equals to:
g (α ) = (1 − )e (6)
β Eα Eα AEα
ln + 3.63504095 − 1.894661lnEα
Rg (a )
Taking the natural logarithm of both sides for equation (6) yields:
The plot of:
g (a) AR 2 RT Eα β 1
ln = ln (1 − )− (7) ln 1.894661 against
T2 β Eα Eα RT T T
gives a straight line and Eα values can be obtained from the slope
Owing to 2RT/E<<1, the equation can be simplified as: of −1.00145033Eα/R.

g (a) AR Eα
ln = ln − (8) DAEM method
T 2
β Eα RT
The DAEM model (Anthony and Howard, 1976) is a multiple
For a fixed β and proposed reaction mechanisms g(α), plotting reaction model and assumes that the decomposition of solids is
ln (g(α)/T2) versus 1/T gives a straight line. The slope of –Eα/R carried out through many independent and parallel reactions,
can be used to determine Eα. each with its own activation energy (Cai et al., 2013, 2014). Each
reaction can be described by an activation energy distribution
Friedman method focused on a medium value with a temperature range that covers
The Friedman method (Friedman, 1964) is based on the assump- this value depending on the standard deviation.
tion that solids decomposition depends only on the rate of mass The general equation of this model can be described as (Li
loss and is independent from the temperature. Therefore, the f(α) et al., 2015):
can be considered constant and taking the natural logarithms of
dα ∞ t −
E

both sids for equation (4) yields: 1− = ∫ exp[−k0 ∫ exp RT dt ] f ( E )dE (12)
dt 0 0

dα Eα
ln( β ) = ln [ Af (α ) ] − (9) where k0, E and R are the prefrequency factor, activation energy
dT RT and universal gas constant (8.314 J (mol K)-1), respectively. f(E)
Because ln[Af(α)] is a constant, so for a given α and β value, the is distribution function that encapsulates the range and variation
Eα values can be calculated from the slope of straight line by in the activation energies of the simultaneous first-order irrevers-
plotting: ible reactions and is normalised as:
dα versus 1 .
ln( β ) ∞
dT T ∫ 0
f ( E )dE = 1 (13)

FWO method Taking into the correlation among T, t and β, equation (5) can
be expressed as (Chen et al., 2016):
The FWO method (Flynn and Wall, 1966; Ozawa, 1965) uses
Doyle’s equation for approximation of temperature integral. Tak- dα ∞ k T −
E
1− = ∫ exp[− 0 ∫ exp RT
dT ] f ( E )dE (14)
ing into account of the approximation, lnp ( x) = −5.331 − 1.052 x , dT 0 β 0

equation (5) can be converted as:


For simplification, equation (14) can be rewritten as:
AEα Eα
lnβ = ln − 5.331 − 1.052 (10) dα ∞ Es
Rg (α ) RT = 1 − ∫ f ( E )dE = ∫ f ( E )dE (15)
dT Es 0

For a fixed α, each plot of lnβ versus 1/T gives a straight line. The
Miura (1995) obtained a simpler method to calculate the activa-
Eα values can be determined through the slope of −1.052Eα/R
tion energy and pre-exponential factor as follows:
over a series of α.
β kR Eα
ln 2
= In 0 + 0.6075 − (16)
T Eα RT
Tang method
The detailed procedures used to estimate Eα, k0, and f(E) are
Tang et al. (2003) developed an estimation of Arrhenius tempera-
described as follows:
ture integral:
e− x 1 1. Measure and plot α as a function of T for at least three differ-
p( x) = ( )
x 1.001989 x + 1.873912 ent heating rates.
4 Waste Management & Research 00(0)

Table 1. Peak temperatures and sample mass loss in the


first stage.

Heating rate Temperature interval Tmax Mass loss


(K min-1) (°C) (°C) (%)
5 38–400 324 30.78
15 38–400 342 31.03
25 38–400 344 28.93

increases the peak temperature. The weight loss during the first
stage may be attributed to the rupture of urethane bonds into
isocyanates and polyols (Ciecierska et al., 2016; Dyer and
Newborn Jr, 1958; Jiao et al., 2013; Valencia et al., 2009). At
Figure 1. TG and DTG profiles of PUR at different heating temperatures above 400°C, the resulted two types of products
rates. decomposed further (Valencia et al., 2009). Further increasing
the temperature to 550°C, the material residues decompose into
2. Plot ln(β/T2) versus 1/T for different heating rates at each α. small molecules. The above results revealed that the first stage
3. For each α in the plot, linearise the data of different heating with α of 0.05–0.5 was the major decomposition stage for PUR,
rates and obtain the Eα and k0 values from slope and intercept, and thus it was selected for the further kinetic study.
respectively.
4. Plot α against E and differentiate the relationship by E to Model-free method
obtain f(E) (Ma et al., 2018a).
According to Supplemental Table S1 (available online), linear
regression graphs were plotted for the Friedman, FWO and
Results and discussion Tang methods, respectively (Figure 2). The parallel behaviour
between the straight lines of the major linear fit presented sat-
Thermogravimetric analysis isfactory coefficients (0.8887–1.0000 for Friedman, 0.8944–
The TG and DTG profiles of PUR pyrolysis at different heating 1.0000 for FWO and 0.8884–1.0000 for Tang). The low
rates are displayed in Figure 1. From the TG curve, it can be correlations for α beyond 0.45 could be attributed to the change
seen that the decreasing trends are similar among the profiles at of thermal decomposition mechanism with the heating rate
three heating rates. Overall, the decomposition of PUR can be increasing. Another factor was the difficult definition of the
divided into three stages with temperatures of 38°C–400°C, mass interval, namely the definition of the initial and final
400°C–550°C and 550°C–1000°C, which was consistent with mass, where the start-point and end-point are defined arbitrar-
literature (Xu et al., 2012). The weight loss were determined as ily (Alves et al., 2019; Anca-Couce et al., 2014). For all meth-
approximately 52%, 18% and 11% for these stages, respec- ods, there was a larger gap between the straight lines at
tively. From the DTG curve, it can be found that there were two conversions 0.05 and 0.1, which indicated a slow conversion of
significant weight losses in the first and second stages. This was PUR, namely the low temperature at the initial stage was not
consistent with reports (Ciecierska et al., 2016; Jiao et al., enough to break down the urethane bonds. Subsequently, the
2013), where the maximum decomposition temperatures were temperatures increased to rupture the molecules causing an
observed at 314°C–331°C and 530°C–562°C. For the first approaching of the straight line after the conversion of 0.1. A
stage, the weight loss rate increased from 0.72%•min-1 to larger gap between the straight lines at conversions 0.45 and
2.17%•min-1 and 3.49%•min-1 with heating rates increasing 0.5 was also observed, which indicated the change of thermal
from 5 K min-1 to 15 K min-1 and 25 K min-1. The corresponding decomposition mechanism.
peak temperatures shifted from 324°C, to 342°C and 344°C The obtained Eα values are displayed in Table 2 and Figure 3.
(Table 1), which was also reported in the pyrolysis of coal (Du It can be seen that the values display a similar trend for those
et al., 2016; Song et al., 2016), biomass (Wang et al., 2016a, methods. They gradually increased from 163.595 kJ mol-1 to
2016b), and oil shales (Bai et al., 2015). This is attributed to the 209.55 kJ mol-1 with α increasing from 0.05 to 0.45, and reaching
heat and mass transfer limitations, which cause temperature 335.852 kJ mol-1 with α = 0.5. The progressive increase observed
gradients within the sample and inside each particle. At a lower was representative of the great variety of bonds and the multipha-
heating rate, the heating of particles occurs gradually leading to sic character of its conversion. It might correspond to competitive
an improved heat transfer to the inner portions and among the or consecutive reactions (Carrier et al., 2016; Papageorgiou, 2018).
particles. However, raising the heating rate increases the tem- The average Eα values were determined as 207.632 kJ mol-1,
perature gradient between the surface and inner of the particles, 206.099 kJ mol-1 and 207.328 kJ mol-1 for Friedman, FWO and
which strengthens the inertia effect of devolatilisation and Tang methods, respectively. As a comparison, the FWO method
affects the decomposition of the inner part of particles and then was more reliably attributed to its larger R values.
Yao et al. 5

Model-fitting method
From equation (8), the activation energy for all g(α) functions in
Table 3 can be obtained, and the resulted kinetic parameters are
displayed in Table 4 and Figure 3. For each model, the Eα values
slightly increased with an increasement in heating rate. However,
the three groups (nucleation models, diffusion models and reac-
tion order models) of reaction models manifested different pro-
files. The diffusion models endowed larger Eα values, however,
the nucleation models gave lower values.
It was noted that the E values from the model-fitting method
was significantly smaller than that from model-free method,
which was also reported in literatures (Dhaundiyal et al., 2018;
Wang et al., 2018). Unlike the model-free method, which is
based on α, the CR method depends on β. There are two main
reasons for the disagreement between the two methods in case of
non-isothermal runs. The first is a result of the force-fitting of
non-isothermal data to hypothetical reaction model (Dhaundiyal
et al., 2018). Arrhenius parameters are determined by a form of
g(α), which is already assumed. Moreover, both parameters T
and α are simultaneously varying with time. The model-fitting
model is not able to distinguish separately temperature depend-
ence of k(T) and α. Consequently, any assumed model can be
easily synchronised with thermoanalytical data at the expense of
drastic variation between the assumed model and the true
unknown model. Owing to this reason, the model-fitting method
provide ambiguous values of Arrhenius parameters. Another
reason is the existence of multi-reaction mechanisms with dif-
ferent activation energies, as contribution of these steps to the
overall decomposition rate is influenced by T and α, which
implies activation energy calculated from the model-fitting
method is a function of T and α. However, the calculated value
of activation energy represents the average value for the overall
process. This value is derived in such a way that it is invariant to
the reaction mechanism and kinetics with change in T and α,
therefore, the isoconversional method is prefer to the model-
based scheme to overcome aforementioned drawbacks, as these
models check dependency of date with respect to α or T without
Figure 2. Linear fitting curves under different conversions prior judgement about the assumed model.
for Friedman, FWO and Tang methods. Recently, there have been ongoing debates about the use of
isoconversional methods for the solids decomposition of the E
Table 2. The kinetic parameters determined using the values. Some researchers claimed that the model-fitting method
model-free methods. might not describe the complex processes, such as biomass
α Friedman FWO Tang decomposition, sufficiently, since numbers of reactions are
involved simultaneously in the process. At this point, the model-
E R E R E R free method played a critical role for the investigation of activa-
(kJ mol-1) (kJ mol-1) (kJ mol-1)
tion energy according to varying α. It has been reported that the
0.05 163.595 0.9906 163.980 0.9900 164.010 0.9889 activation energy produced by the Friedman method was consist-
0.1 177.180 0.9981 176.404 0.9982 176.656 0.9980 ent with that of the FWO and Kissinger-Akahira-Sunose (KAS)
0.15 185.078 1.0000 183.738 1.0000 184.155 1.0000
methods. In this work, it has also confirmed that the Ea values
0.2 194.265 0.9996 193.735 0.9996 194.506 0.9996
from three model-free methods was comparable.
0.25 197.715 0.9948 196.580 0.9956 197.366 0.9952
0.3 200.126 0.9912 200.160 0.9920 201.008 0.9913
0.35 206.179 0.9913 205.471 0.9921 206.470 0.9913
0.4 206.778 0.9968 205.408 0.9968 206.254 0.9966 DAEM method
0.45 209.550 0.9939 207.328 0.9941 208.050 0.9935 The fitting curves of ln(β/T2) versus 1/T were plotted according
0.5 335.852 0.8887 328.190 0.8944 334.804 0.8884
to equation (16) and the results are displayed in Figure 4. The Eα
6 Waste Management & Research 00(0)

Table 4. The kinetic parameters calculated using the CR


method.

Reaction β (K min-1))
models
15 25 35

E R E R E R
(kJ mol-1) (kJ mol-1) (kJ mol-1)
P2 22.812 0.9764 23.217 0.9858 24.551 0.9674
P3 12.021 0.9608 12.210 0.9760 13.048 0.9474
P4 6.626 0.9262 6.707 0.9536 7.297 0.9055
A2 26.632 0.9870 27.084 0.9927 28.646 0.9804
A3 14.568 0.9800 14.788 0.9889 15.777 0.9705
A4 8.536 0.9662 8.640 0.9812 9.343 0.9518
D1 119.929 0.9872 122.282 0.9925 128.077 0.9818
D2 124.760 0.9893 127.171 0.9939 133.257 0.9842
D3 129.940 0.9910 132.409 0.9950 138.802 0.9864
D4 126.481 0.9899 128.917 0.9943 135.102 0.9850
F0 55.184 0.9846 56.238 0.9909 59.061 0.9782
F1 62.824 0.9909 63.970 0.9949 67.249 0.9862
F2 71.269 0.9932 72.513 0.9950 76.307 0.9900
F3 80.509 0.9922 81.854 0.9921 86.216 0.9904

Figure 3. Comparison of kinetic parameters from model-free


and model-fitting methods.

Table 3. Expressions of g(α) for reaction mechanisms used in


this work.

Reaction models Symbols g(α)


Nucleation models
Power law P2 α1/3
Power law P3 α1/2
Power law P4 α3/2
Avrami–Erofeev A2 [-ln(1-α)]1/2
Avrami–Erofeev A3 [-ln(1-α)]1/3
Avrami–Erofeev A4 [-ln(1-α)]1/4
Diffusion models
One-dimensional diffusion D1 α2
Two-dimensional diffusion D2 (1-α)In(1-α)+ α
Three-dimensional D3 [1-(1-α)1/3]2
diffusion (Jander equation)
Four-dimensional D4 1-(2/3)α-(1-α)2/3
diffusion (Ginstling– Figure 4. Linear fitting curves under different conversions for
Brounshetein) the DAEM method and the activation energy distribution.
Reaction order models
Zero-order F0 α values at selected levels of α were determined from the slope of
First-order F1 -ln(1-α) plots (Gokul et al., 2019). The relationship between α versus Eα
Second-order F2 (1-α)-1-1
could be obtained by plotting α against corresponding Eα values
Third-order F3 (1/2)[(1-α)-2-1]
as shown in Figure 4. The Eα values increased from 163.536 to
Yao et al. 7

334.231 kJ mol-1 with α increasing. The average Eα value was Anca-Couce A, Berger A and Zobel N (2014) How to determine consistent bio-
mass pyrolysis kinetics in a parallel reaction scheme. Fuel 123: 230–240.
determined as 206.799 kJ mol-1, comparable with 206.099 kJ mol-1
Anthony DB and Howard JB (1976) Coal devolatilization and hydrogastifica-
from the FWO method. When the curves for relationship between tion. AIChE Journal 22: 625–656.
α and Eα were differentiated graphically by Eα , f(E) curves for Bai F, Sun Y, Liu Y, et al. (2015) Thermal and kinetic characteristics of
PUR pyrolysis were obtained. The peak of f(E) curve appeared at pyrolysis and combustion of three oil shales. Energy Conversion and
Management 97: 374–381.
205.929 kJ mol-1 and the maximum thermal decomposition takes Brown ME, Maciejewski M, Vyazovkin S, et al. (2000) Computational
place at this point. Calculated results of the fitting procedure aspects of kinetic analysis: Part A: The ICTAC kinetics project-data,
using the DAEM method showed good agreement with experi- methods and results. Thermochimica Acta 355: 125–143.
Cai J, Wu W and Liu R (2014) An overview of distributed activation energy
mental DTG data in Figure 1.
model and its application in the pyrolysis of lignocellulosic biomass.
Renewable and Sustainable Energy Reviews 36: 236–246.
Cai J, Wu W, Liu R, et al. (2013) A distributed activation energy model for the
Conclusions pyrolysis of lignocellulosic biomass. Green Chemistry 15: 1331–1340.
Carrier M, Auret L, Bridgwater A, et al. (2016) Using apparent activation
The thermogravimetric analysis indicated that the PUR decom- energy as a reactivity criterion for biomass pyrolysis. Energy & Fuels
position process could be divided into three stages with tempera- 30: 7834–7841.
Chen T, Zhang J and Wu J (2016) Kinetic and energy production analysis of
tures of 38°C–400°C, 400°C–550°C and 550°C–1000°C. Eα
pyrolysis of lignocellulosic biomass using a three-parallel Gaussian reac-
values from three model-free methods displayed a similar trend. tion model. Bioresource Technology 211: 502–508.
The R values from the FWO method were larger and the resulted Cheng S, Qiao Y, Huang J, et al. (2019) Effects of Ca and Na acetates on
E values fell into the range of 163.980–328.190 kJ mol-1. For the nitrogen transformation during sewage sludge pyrolysis. Proceedings of
the Combustion Institute 37: 2715–2722.
CR method, the diffusion models had higher R values, although China Household Electric Appliance Research Institute (2018) White Paper
the E values were smaller than that from the FWO method. The on Current Situation and Trend of WEEE Recycling Industry in China. 22
activation energy distribution from the DAEM method showed May 2018, Beijing, China. Available at: http://www.cheari.org/ (accessed
7 October 2019).
that the Eα values increased with α increasing. The average Eα
Chong YY, Thangalazhy-Gopakumar S, Gan S, et al. (2017) Kinetics
value was 206.799 kJ mol-1, consistent with that from the FWO and mechanisms for copyrolysis of palm empty fruit bunch fiber
method. The the FWO and DAEM were better for the description (EFBF) with palm oil mill effluent (POME) sludge. Energy & Fuels
31: 8217–8227.
of the PUR pyrolysis process.
Ciecierska E, Jurczyk-Kowalska M, Bazarnik P, et al. (2016) The influence
of carbon fillers on the thermal properties of polyurethane foam. Journal
Declaration of conflicting interests of Thermal Analysis and Calorimetry 123: 283–291.
The authors declared no potential conflicts of interest with respect to Criado JM (1978) Kinetic analysis of DTG data from master curves.
Thermochimica Acta 24: 186–189.
the research, authorship, and/or publication of this article.
Dhaundiyal A, Singh SB, Hanon MM, et al. (2018) Determination of kinetic
parameters for the thermal decomposition of Parthenium hysterophorus.
Funding Environmental and Climate Technologies 22: 5–21.
Du R, Wu K, Xu D, et al. (2016) A modified Arrhenius equation to predict
The authors disclosed receipt of the following financial support for
the reaction rate constant of Anyuan pulverized-coal pyrolysis at different
the research, authorship, and/or publication of this article: This work heating rates. Fuel Processing Technology 148: 295–301.
was financially supported by the National Natural Science Dyer E and Newborn GE Jr (1958) Thermal degradation of carbamates of
Foundation of China [Grant no. 51606055 and 51911530460] and methylenebis-(4-phenyl Isocyanate). Journal of the American Chemical
the Zhejiang Provincial Natural Science Foundation of China [Grant Society 80: 5495–5498.
no. LY19B070008]. Fernandez-Lopez M, Pedrosa-Castro GJ, Valverde JL, et al. (2016)
Kinetic analysis of manure pyrolysis and combustion processes. Waste
Management 58: 230–240.
ORCID iD Flynn JH and Wall LA (1966) A quick, direct method for the determination
Zhitong Yao https://orcid.org/0000-0002-9180-2329 of activation energy from thermogravimetric data. Journal of Polymer
Science Part B: Polymer Letters 4: 323–328.
Friedman HL (1964) Kinetics of thermal degradation of char-forming plas-
Supplemental material tics from thermogravimetry. Application to a phenolic plastic. Journal of
Supplemental material for this article is available online. Polymer Science Part C: Polymer Symposia 6: 183–195.
Gokul PV, Singh P, Singh VP, et al. (2019) Thermal behavior and kinet-
ics of pyrolysis of areca nut husk. Energy Sources, Part A: Recovery,
References Utilization, and Environmental Effects. Epub ahead of print 19 February
Abdel Rahman HA, Younes MM and Khattab MM (2019) Recycling of 2019. DOI: 10.1080/15567036.2019.1582733.
polyurethane foam waste in the production of lightweight cement pastes Hardi F, Imai A, Theppitak S, et al. (2018) Gasification of char derived from
and its irradiated polymer impregnated composites. Journal of Vinyl and catalytic hydrothermal liquefaction of pine sawdust under a CO2 atmos-
Additive Technology 40: 1–11. phere. Energy Fuels 32: 5999–6007.
Aboulkas A and El Bouadili A (2010) Thermal degradation behaviors of pol- Jiao L, Xiao H, Wang Q, et al. (2013) Thermal degradation characteris-
yethylene and polypropylene. Part I: Pyrolysis kinetics and mechanisms. tics of rigid polyurethane foam and the volatile products analysis with
Energy Conversion and Management 51: 1363–1369. TG-FTIR-MS. Polymer Degradation and Stability 98: 2687–2696.
Alves JLF, Da Silva JCG, Da Silva Filho VF, et al. (2019) Kinetics and ther- Junco C, Gutiérrez S, Gadea J, et al. (2018) Cement mortars lightened with
modynamics parameters evaluation of pyrolysis of invasive aquatic mac- rigid polyurethane foam waste applied on-site: Suitability and durability.
rophytes to determine their bioenergy potentials. Biomass and Bioenergy In: Taha M (ed.) International Congress on Polymers in Concrete (ICPIC
121: 28–40. 2018), Cham: Springer, pp.457–463.
8 Waste Management & Research 00(0)

Khawam A and Flanagan DR (2005) Complementary use of model-free and Rogers FE and Ohlemiller TJ (1981) Pyrolysis kinetics of a polyurethane
modelistic methods in the analysis of solid-state kinetics. The Journal of foam by thermogravimetry; a general kinetic method. Journal of
Physical Chemistry B 109: 10073–10080. Macromolecular Science—Chemistry 15: 169–185.
Khawam A and Flanagan DR (2006) Solid-state kinetic models: Basics and Sheel A and Pant D (2018) Chemical Depolymerization of Polyurethane
mathematical fundamentals. The Journal of Physical Chemistry B 110: Foams via Glycolysis and Hydrolysis, Recycling of Polyurethane Foams.
17315–17328. Netherlands: Elsevier, pp.67–75.
Li K, Pau DS and Zhang H (2016) Pyrolysis of polyurethane foam: Optimized Song H, Liu G and Wu J (2016) Pyrolysis characteristics and kinetics of low
search for kinetic properties via simultaneous K–K method, genetic algo- rank coals by distributed activation energy model. Energy Conversion
rithm and elemental analysis. Fire and Materials 40: 800–817. and Management 126: 1037–1046.
Li S, Chen X, Liu A, et al. (2015) Co-pyrolysis characteristic of biomass and Srivastava A, Chandel N and Mehta N (2017) Calorimetric studies of crys-
bituminous coal. Bioresource Technology 179: 414–420. tallization for multi-component glasses of Se-Te-Sn-Ag (STSA) system
Ma Y, Wang J and Zhang Y (2018a) Analysis of pyrolysis characteristics using model-free and model-fitting non-isothermal methods. Journal of
and kinetics of Euphausia superba shell waste using TG-FTIR and dis- Thermal Analysis and Calorimetry 128: 907–914.
tributed activation energy model. Biomass Conversion and Biorefinery Tang W, Liu Y, Zhang H, et al. (2003) New approximate formula for
8: 1–9. Arrhenius temperature integral. Thermochimica Acta 408: 39–43.
Ma Z, Wang J, Yang Y, et al. (2018b) Comparison of the thermal degra- Tantisattayakul T, Kanchanapiya P and Methacanon P (2018) Comparative
dation behaviors and kinetics of palm oil waste under nitrogen and air waste management options for rigid polyurethane foam waste in Thailand.
atmosphere in TGA-FTIR with a complementary use of model-free and Journal of Cleaner Production 196: 1576–1586.
model-fitting approaches. Journal of Analytical and Applied Pyrolysis Valencia LB, Rogaume T, Guillaume E, et al. (2009) Analysis of principal
134: 12–24. gas products during combustion of polyether polyurethane foam at differ-
Magnin A, Hoornaert L, Pollet E, et al. (2019) Isolation and characterization ent irradiance levels. Fire Safety Journal 44: 933–940.
of different promising fungi for biological waste management of polyure- Vyazovkin S, Burnham AK, Criado JM, et al. (2011) ICTAC Kinetics
thanes. Microbial Biotechnology 12: 544–555. Committee recommendations for performing kinetic computations on
Miura K (1995) A new and simple method to estimate f(E) and k0(E) in the thermal analysis data. Thermochimica Acta 520: 1–19.
distributed activation energy model from three sets of experimental data. Wang L, Lei H, Liu J, et al. (2018) Thermal decomposition behavior and
Energy & Fuels 9: 302–307. kinetics for pyrolysis and catalytic pyrolysis of Douglas fir. RSC
Montzka SA, Dutton GS, Yu P, et al. (2018) An unexpected and persistent Advances 8: 2196–2202.
increase in global emissions of ozone-depleting CFC-11. Nature 557: Wang S, Dai G, Ru B, et al. (2016a) Effects of torrefaction on hemicel-
413. lulose structural characteristics and pyrolysis behaviors. Bioresource
Murray P and White J (1955) Kinetics of the thermal dehydration of clays. Technology 218: 1106–1114.
Part IV. Interpretation of the differential thermal analysis of the clay Wang S, Lin H, Ru B, et al. (2016b) Kinetic modeling of biomass components
minerals. Transactions and Journal of the British Ceramic Society 54: pyrolysis using a sequential and coupling method. Fuel 185: 763–771.
204–238. Xu M, Guan J and Wang JW (2012) Pyrolysis behavior and kinetics of polyu-
Ozawa T (1965) A new method of analyzing thermogravimetric data. Bulletin rethane insulation materials from waste refrigerators. Advanced Materials
of the Chemical Society of Japan 38: 1881–1886. Research 356–360: 1752–1758.
Papageorgiou GZ (2018) Thinking green: Sustainable polymers from renew- Yao Z, Ling T, Sarker PK, et al. (2018) Recycling difficult-to-treat e-waste
able resources. Polymers 10: 952. cathode-ray-tube glass as construction and building materials: A critical
Papari S and Hawboldt K (2015) A review on the pyrolysis of woody biomass review. Renewable and Sustainable Energy Reviews 81: 595–604.
to bio-oil: Focus on kinetic models. Renewable and Sustainable Energy Yao Z, Wu D, Liu J, et al. (2017) Recycling of typical difficult-to-treat
Reviews 52: 1580–1595. e-waste: Synthesize zeolites from waste cathode-ray-tube funnel glass.
Qi W, He C, Wang Q, et al. (2018) Carbon-based solid acid pretreatment Journal of Hazardous Materials 324: 673–680.
in corncob saccharification: specific xylose production and efficient Yu S, Su W, Wu D, et al. (2019) Thermal treatment of flame retardant plas-
enzymatic hydrolysis. ACS Sustainable Chemistry & Engineering 6: tics: A case study on a waste TV plastic shell sample. Science of The Total
3640–3648. Environment 675: 651–657.
Qi W, Liu G, He C, et al. (2019) An efficient magnetic carbon-based solid acid Zeng X and Mathews JA amd Li J (2018) Urban mining of e-waste is becom-
treatment for corncob saccharification with high selectivity for xylose and ing more cost-effective than virgin mining. Environmental Science &
enhanced enzymatic digestibility. Green Chemistry 21: 1292–1304. Technology 52: 4835–4841.

You might also like