Kinetic study of the effect of the heating rate on the waste tyre pyrolysis to maximise limonene production

Kinetic study of the effect of the heating rate on the waste tyre pyrolysis to maximise limonene production

Journal Pre-proof Kinetic study of the effect of the heating rate on the waste tyre pyrolysis to maximise limonene production N.M. Mkhize, B. Danon, P...

3MB Sizes 0 Downloads 14 Views

Journal Pre-proof Kinetic study of the effect of the heating rate on the waste tyre pyrolysis to maximise limonene production N.M. Mkhize, B. Danon, P. van der Gryp, J.F. G¨orgens

PII:

S0263-8762(19)30453-8

DOI:

https://doi.org/10.1016/j.cherd.2019.09.036

Reference:

CHERD 3826

To appear in:

Chemical Engineering Research and Design

Received Date:

15 April 2019

Revised Date:

19 September 2019

Accepted Date:

26 September 2019

Please cite this article as: Mkhize NM, Danon B, van der Gryp P, G¨orgens JF, Kinetic study of the effect of the heating rate on the waste tyre pyrolysis to maximise limonene production, Chemical Engineering Research and Design (2019), doi: https://doi.org/10.1016/j.cherd.2019.09.036

This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. © 2019 Published by Elsevier.

Kinetic study of the effect of the heating rate on the waste tyre pyrolysis to maximise limonene production

N. M. Mkhizea,b*, B. Danona, P. van der Grypa, J. F. Görgensa

a

Department of Process Engineering, Stellenbosch University, Private Bag X1, Matieland, 7602,

of

Stellenbosch, South Africa b

Discipline of Chemical Engineering, University of KwaZulu-Natal, 238 Mazisi Kunene Road, Glenwood,

-p

ro

4041, South Africa

* Corresponding author: [email protected]

The kinetic model of production of isoprene and DL-limonene from waste tire pyrolysis is proposed

lP



re

Highlights

Increase in the heating rate increased the peak temperature of isoprene and DL-limonene



The peak temperature increase was more significant in DL-limonene compared to isoprene



Rapid heating resulted in the less energy allocated to the formation of products



Increase in the heating rate favored the formation of the lower Ea compound (DL-

ur na



Jo

limonene)



Various factors other than temperature and heating rate, affect DL-limonene formation from waste tyre pyrolysis

1

Abstract The formation of isoprene and DL-limonene during waste tyre pyrolysis was investigated in terms of the effect of the heating rate (up to 100 °C/min). Ion current signals were used to track during pyrolysis the evolution of the predominant ions of isoprene (isoprene 67) and DL-limonene (limonene 93), by using a thermogravimetric analyser coupled with mass spectrometry (TGA/MS). The combined model-free and model-based kinetics were used to estimate the activation energy (Ea) for isoprene and DL-limonene

of

formation at 131 and 115 kJ/mole, respectively, based on the Kissinger method. Reaction order (n) values were estimated at 1.2 and 1.1 for isoprene and DL-limonene, respectively. Better model fit (R2 = 0.998) of

ro

the experimental data to the Arrhenius equation for isoprene and DL-limonene, respectively, was observed when the Kissinger method was used compare to Friedman method. Although the Ea values for isoprene

-p

and DL-limonene were not significantly different, the combined three kinetic parameters (Ea, pre-

re

exponential constant (A), and n) may be significantly different. Therefore, for DL-limonene formation selectivity over isoprene, the differences in the three kinetic parameters values for each compound model

lP

and heating rate on the reaction progress was significant. The reaction progress at peak isoprene and DLlimonene formation rate increased from 0.42 to 0.45 and more significantly from 0.35 to 0.44, respectively as the heating rate was increased from 15 to 100 °C, confirming that the preferred strategy to maximise DL-

ur na

limonene production is rapid heating to the moderate final pyrolysis temperature.

Keywords: heating rate, isoprene, kinetics, limonene, tyre-derived oil

Introduction

Jo

1

Waste tyre piles are increasing at a high rate compared to the rate of their elimination as provided by various recycling techniques. Among the techniques of waste tyre processing (Barlaz et al., 1993; Giugliano et al., 1999; Sienkiewicz et al., 2012), pyrolysis is one of the most promising, since it results in products that have a variety of industrial and domestic applications (Quek and Balasubramanian, 2013; Williams, 2013). 2

Pyrolysis is defined as the thermal treatment of waste tyres under inert conditions to yield pyro-gas, tyre derived oil (TDO) and pyro-char (Antoniou and Zabaniotou, 2013; Mui et al., 2004). The pyro-gas can be used as fuel for the endothermic pyrolysis process (Williams, 2013) or various other applications, e.g. when further processed to recover valuable chemicals, such as butadiene or isoprene. TDO’s significantly high hydrogen and carbon element content and high concentration of valuable chemicals, e.g., DL-limonene, terpinolene, and p-cymene characterise it as a potential alternative for liquid fuel, or as a feedstock for the

of

production of valuable chemicals (Banar et al., 2012; Quek and Balasubramanian, 2013). The pyro-char mainly contains carbon and can be upgraded into the carbon black or activated carbon (Zabaniotou et al.,

ro

2004). However, to achieve these beneficial applications for the waste tyre pyrolysis products, intensive

-p

processing to improve products, such as product refining, upgrading, and purification, are required.

re

TDO is the main pyrolysis product fraction (in weight basis), as a liquid is relatively easy to handle, hence it is the most interesting pyrolysis product (Banar et al., 2012). TDO is a potential source for the recovery

lP

of valuable chemicals, while the resulting residues can be upgraded to liquid fuels, comparable to the conventional petroleum-derived products and fuels (Banar et al., 2012; Williams, 2013). There is substantial

ur na

literature on the upgrading of the TDO to liquid fuels, however, the employed techniques are mainly focusing on the post-treatment of the TDO after pyrolysis. These processes include moisture removal, desuphurisation, hydrotreatment, thermal treatment, catalytic treatment and biological processes (Aydin and Ilkiliç, 2012; Ilkiliç and Aydin, 2011; Murugan et al., 2008; Rofiqul Islam et al., 2008), but they are

Jo

relatively complex, expensive and can generate environmental unfriendly waste streams requiring further treatment.

An alternative to post-pyrolysis TDO upgrading is the modification of the pyrolysis process to improve the TDO quality in terms of chemical composition (Mkhize et al., 2016; Quek and Balasubramanian, 2013). For successful process optimisation, kinetic mechanisms of the waste tyre devolatilisation need to be understood (Chen et al., 2001; Friedman, 1964; Kissinger, 1957; Senneca et al., 1999). Activation energies 3

(Ea) and pre-exponential rate constants (A) in relation to the pyrolysis temperature and heating rate have been reported (Cheung et al., 2011; Danon et al., 2015a; L et al., 2010; Seidelt et al., 2006). These studies mainly concluded that devolatilisation kinetics consist of various parallel and serial devolatilisation reaction pathways, such as devolatilisation of the tyre processing oils followed by devolatilisation of the natural rubber and finally synthetic rubbers (styrene-butadiene rubber and butadiene rubber). Additionally, waste tyre pyrolysis has been kinetically presented according to products volatilisation, such as gaseous products,

of

variety of TDO sub-fractions (aromatics, aliphatics, polyaromatic hydrocarbons and heteroatom compounds) and char (Olazar et al., 2005). Some effort has been done to study waste tyre pyrolysis kinetics

ro

based on the formation of chemical compounds (Aguado et al., 2005; Arabiourrutia et al., 2019; Olazar et

re

kinetic mechanism study (Leung and Wang, 1998).

-p

al., 2008). In most of the studies, the heating rate has been an important parameter in the waste tyre pyrolysis

For example, Lam et al. (2010) proposed a multi-stage pyrolysis process with 1) pyrolysis of tyre additives,

lP

2) pyrolysis of polymerised rubbers, and 3) pyrolysis of the cross-linked and/or cyclised rubbers (Lam et al., 2010). Cheung et al. (2011) propounded the following pyrolysis stages: 1) pyrolysis of the volatiles, 2)

ur na

main chain scission (depolymerisation) to intermediates, 3) further depolymerisation of the intermediates (pyrolysis of cross-linked rubbers), and 4) cracking of the intermediates to shorter-chain compounds (degradation) (Cheung et al., 2011). Danon et al. (2015) in the study of devolatilisation kinetics of tyre rubbers defined tyre devolatilisation as a three-stage process: 1) devolatilisation of additives, 2) crosslinking

Jo

and depolymerisation of the rubbers into intermediates and volatiles, respectively, and 3) finally the degradation of the intermediates to volatiles (Danon et al., 2015a). There has been therefore substantial literature data mainly focusing on tyre components devolatilisation mechanism kinetics and there seem to be agreements on the observations, however little is known about product formation kinetics.

The present work focuses on the kinetic mechanism of waste tyre pyrolysis based on the products formed in the hot volatiles, rather than considering the devolatilisation of the tyre components. Polyisoprene 4

(natural rubber) devolatilisation forms two compounds (isoprene or DL-limonene) via two supposedly competing reaction pathways. Kinetic parameters are evaluated for the formation of isoprene and DLlimonene to elucidate these mechanisms, through the application of thermogravimetric analysis (TGA), with various constant heating rates up to 100 °C/min. Isoprene and DL-limonene formation are monitored using mass spectrometry (MS) (67 and 93 amu, respectively). DL-limonene formation is more interesting than isoprene since it is the most significant compound in the tyre derived oil (TDO) fraction while isoprene

of

is in gaseous phase at atmospheric conditions, i.e., remain uncondensed in the gas fraction. It is relatively easier to handle liquid phase (consists mainly of DL-limonene) than the gaseous phase (consists mainly of

-p

Theoretical background

re

2

ro

isoprene) products and relatively high valuable compound.

The kinetic model used in the current study entails a description of the formation of the isoprene and DL-

lP

limonene from waste tyre or polyisoprene (tyre rubbery component) pyrolysis. In the literature, it has been pointed out that devolatilisation of the polyisoprene results into two competing product formation reaction pathways, i.e. the formation of allylic radicals that will either i) depropagate (unzip) to form isoprene or, ii)

ur na

undergo intramolecular cyclisation followed by chain scission to form DL-limonene (Danon et al., 2015b), see Error! Reference source not found..

Kinetic analysis requires tracking of the formation of isoprene and DL-limonene. This is achieved by MS,

Jo

where the ion current signals are obtained through the ion extraction method using predominant mass-overcharge-ratios of 67 and 93 amu, from the fragmentation of the isoprene and DL-limonene, respectively. It is assumed that tracking of isoprene and DL-limonene formation ions for ion current signals is independent of one another, i.e., do not influence one another or influenced by other ions that are present due to fragmentation of various molecules from other tyre components. The proposed model is therefore based on the devolatilisation of the polyisoprene component of the tyre rubber as well as the formation of the isoprene 5

and DL-limonene. The conversion rate of polyisoprene or natural rubber (NR) is given by the mass fraction devolatilisation equation, Eq. (1).

𝑑𝑚 = 𝑘(𝑇)𝑓(𝑚) 𝑑𝑡 Eq. (1) where: m is a mass fraction, t is time, k(T) is reaction rate constant at temperature, T (K), and f(m) is the

ro

of

reaction model.

Subsequent to devolatilisation of the polyisoprene is the formation of isoprene and DL-limonene and can be

-p

represented by the Eq. (2).

re

𝑑𝑆𝑖 = 𝑘(𝑇)𝑓(𝑆) 𝑑𝑡

lP

Eq. (2)

where: 𝑆𝑖 is MS ion current fraction of either the isoprene or DL-limonene and f(S) is the reaction model for isoprene or DL-limonene formation reaction.

by Eq.(3).

𝛼𝑖 =

𝑆𝑖,𝑐𝑢𝑚𝑠𝑢𝑚 𝑆𝑖,𝑠𝑢𝑚

Jo

Eq. (3)

ur na

The reaction progress, 𝛼𝑖 , of the formation reaction rate of either isoprene and DL-limonene is represented

where: 𝛼𝑖 is the reaction progress for the formation of either isoprene or DL-limonene, 𝑆𝑖,𝑐𝑢𝑚𝑠𝑢𝑚 is the cumulative sum of the ion current at time t, and 𝑆𝑖,𝑠𝑢𝑚 is the sum of ion current up to final time for either isoprene or DL-limonene formation.

Therefore, Eq. (2) can be presented by Eq. (4). 6

𝑑𝛼𝑖 = 𝑘(𝑇)𝑓(𝑆𝑖 ) 𝑑𝑡 Eq. (4) The time, t, and heating rate, β, can be related by Eq. (5).

𝑑𝑇 𝑑𝑡

of

𝛽=

where: β is the heating rate, and T is the absolute temperature.

-p

The reaction rate constant, k, can be presented by Eq. (6).

ro

Eq. (5)

re

−𝐸𝑎 𝑘(𝑡) = 𝐴𝑒𝑥𝑝( ) 𝑅𝑇 Eq. (6)

ur na

the universal gas constant.

lP

where: k is the reaction rate constant, A is a pre-reaction rate constant, Ea is the activation energy, and R is

Therefore, by substituting Eq. (4) and Eq. (5) in Eq. (2): 𝑑𝛼 1 𝐸𝑎 = 𝐴𝑒𝑥𝑝(− )𝑓(𝑆) 𝑑𝑇 𝛽 𝑅𝑇

Jo

Eq. (7)

2.1

Determination of kinetic parameters

Kinetic parameters (Ea, A, and n) can be determined using various techniques classified according to i) pyrolysis conditions, i.e., isothermal or non-isothermal, or ii) mathematical analysis of the experimental results, i.e., linear or nonlinear. In the current study, non-isothermal pyrolysis conditions were used since 7

the experiments were carried out at various heating rates (15, 25, 50, 75 and 100 °C/min). The linear mathematical analysis methods for experimental results analysis were selected due to their conventional application in the analysis of thermal devolatilisation of the solids. These methods allow the determination of a linear relationship between the kinetic parameters using MS ion current data generated at various reaction rates. Linear regression methods using experimental results are applied to determine coefficients of the linear relation. Typical non-isothermal and linear methods were considered in the present study, in

of

particular, the iso-conversional methods of Friedman (differential approach) and Kissinger (integration approach) (Friedman, 1964; Kissinger, 1957). Matlab 2015© was used in the analysis of the experimental

ro

results. Moreover, MS ion current curves peak areas were used to qualitatively and quantitatively analyse isoprene and DL-limonene using predominant fragmentation spectrum ions isoprene-67 and limonene-93,

re

Friedman method

lP

2.2

-p

respectively.

Taking a natural logarithm of Eq. (7), the linear relationship is employed in the equation and resulted in Eq.

𝑙𝑛 (𝛽

ur na

(8).

𝑑𝛼 𝐸𝑎 ) = 𝑙𝑛𝐴 + 𝑙𝑛𝑓(𝛼) − 𝑑𝑇 𝑅𝑇

Jo

Eq. (8)

Plotting 𝑙𝑛 (𝛽

with slope

𝑑𝛼

) versus 𝑑𝑇

−𝐸𝑎 𝑅

1 𝑇

at given reaction progress, α, for various heating rates yields a straight line

. Activation energy can be obtained from this slope without knowing the reaction function

𝑓(𝛼). The pre-exponential reaction rate constant, A, is the y-intercept of a straight line. Friedman method is one of the model-free iso-conversional method (Friedman, 1964). 8

2.3

Kissinger method

The Kissinger method is based on the determination of the temperature at the maximum devolatilisation rate of the polyisoprene, or the temperature associated with the maximum formation rate of isoprene or DLlimonene temperatures, both at various heating rates (Kissinger, 1957). In the Kissinger method, Eq. 7 is

𝛽

𝐴𝑅

of

modified to Eq. (9).

𝐸

ro

𝑙𝑛 (𝑇 2 ) = 𝑙𝑛 (𝑓(𝛼)) − 𝑅𝑇𝑎

2 𝑇𝑚𝑎𝑥

) versus 𝑇

1

𝑚𝑎𝑥

at various heating rates yields a straight line that allows determination of

re

𝛽

A plot of 𝑙𝑛 (

-p

Eq. (9)

lP

the activation energy, Ea, from the gradient of the straight line,

−𝐸𝑎 𝑅

. The y-intercept can be used to estimate

the pre-exponential reaction rate factor. Similar to the Friedman method, Kissinger method is another

2.4

ur na

example of a model-free iso-conversional method.

Master plots

Jo

Introduction of the explicit value of the heating rate reduces the applicability of Eq. (6) to processes in which the sample temperature does not deviate significantly from the reference temperature. Therefore, the integration of Eq. (7) results in Eq. (10):

𝑔(𝛼) = ∫

(−𝑥)𝑑𝑥 𝑑𝛼 𝐴 𝐸𝑎 𝐴𝐸𝑎 𝐴𝐸𝑎 = ∫ 𝑒𝑥𝑝 (− ) 𝑑𝑇 = ( ) ∫ 𝑒𝑥𝑝 =( ) 𝑝(𝑥) 2 𝑓(𝛼) 𝛽 𝑅𝑇 𝑅𝛽 𝑥 𝑅𝛽

Eq. (10) 9

where: x =𝐸𝑎 ⁄𝑅𝑇, and p(x) is the temperature integral. By using a reference value at the midpoint of the reaction progress, α = ref., Eq. (11) is obtained.

𝑔(𝑟𝑒𝑓. ) = (

𝐴𝐸𝑎 ) 𝑝(𝑟𝑒𝑓. ) 𝑅𝛽

of

Eq. (11) Typical theoretical kinetic models summary for solids devolatilisation are represented in Table 1 (Criado,

ro

1978; Perez-Maqueda et al., 1996).

𝑇

𝑟𝑒𝑓

2

𝑑𝛼

𝑑𝛼

) [( 𝑑𝑡 ) / ( 𝑑𝑡 )

𝑟𝑒𝑓

]

lP

Eq. (12)

re

Experimental: (𝑇

-p

At given reaction progress, experimental and theoretical progress are represented by Eqs. (12) and (13).

Eq. (13)

Equipment and method

Jo

3

ur na

Theoretical: 𝑔(𝛼) ∙ 𝑓(𝛼)/[𝑔(𝛼) ∙ 𝑓(𝛼)]𝑟𝑒𝑓

Truck tyre sample with the particle size range of 0.6 to 0.8 mm was sieved from a bulk of approximately 500 kg crumbed (steel- and fabric-free) waste tyres, consisting of particle sizes up to 5 mm. The crumb was supplied by a local waste tyre recycler. A slightly adjusted version of ASTM E1131 – 08 (with X = 275 °C) was used to determine proximate analysis. Additionally, the rubber composition of the crumb was

10

determined using a procedure described by (Danon and Görgens, 2015). The results of the proximate analysis and rubber composition are shown in Table 2.

A Mettler Toledo TGA/DCS 1 thermogravimetric analyser (TGA) connected to a ThermoStar GSD 320 T3 (Pfeiffer Vacuum, Germany) mass spectrometer (MS) was used for all experiments. The ionisation energy was set to 70 eV, using a Secondary Electron Multiplier (SEM) to amplify the MS signal. Argon gas

of

(99.999% purity, Afrox, South Africa) was used as a carrier gas. The sample size of 10 mg in these experiments was pyrolysed up to 800 ° C at five different constant heating rates (15, 25, 50, 75, and 100

ro

°C/min), which correspond to experimental length (52, 31, 16, 10, 8 min), respectively. The mass-overcharge-ratios of the two most predominant fragments of isoprene and DL-limonene (67 and 93 amu,

-p

respectively) were monitored (Stein, 2015). It was assumed that secondary reactions of the hot volatiles

re

since the vacuum pressure of at most 10-6 bar was maintained throughout the experiments to sweep out the volatiles from the TGA furnace to the mass spectrometer detector, resulting into relatively short residence

4.1

Results and discussion

ur na

4

lP

time. All experiments were conducted in duplicates to ensure the repeatability of the results.

Thermogravimetric analysis

Thermogravimetry (TG) and it’s derivative (DTG) profiles on the pyrolysis temperature basis at various

Jo

heating rates up to 100 °C/min are illustrated in Figure 1 a) and b), respectively. From the TG profiles in Figure 1 a) it can be observed that, regardless of the heating rate, the profiles converge at the same value of solid residual at the end of pyrolysis. In Figure 1 b), the shape of the curves at a lower heating rate show two peaks, i.e. the main devolatilisation peak at low temperature and a small peak (shoulder) at the high temperature. As the heating rate increases the main peak becomes higher, while the shoulder disappears, as has been observed by other researchers (Danon et al., 2015a). This can be attributed to the elimination of 11

the secondary reactions (primary products cracking) at higher heating rates while promoting devolatilisation of the rubbers. The effect of the heating rate on the devolatilisation peak temperature is characterised by an increase in the temperature associated with the maximum devolatilisation rate, as the heating rate increases.

4.2

Mass spectrometric analysis

of

Illustrated in Figure 2 a) and b) are the MS ion current profiles of the isoprene and DL-limonene, respectively. These profiles resemble the DTG curves of the devolatilisation of the tyre rubber component

ro

(Figure 2 b)), in that the shape of the curves at a lower heating rate show two peaks, i.e., main devolatilisation peak at low temperature and a small peak (shoulder) at the high temperature. However,

-p

zooming in on both the peaks of the tyre devolatilisation temperature (DTG) and the peaks of the products

re

volatilisation temperature (MS) as shown in Table 3, a slight difference in profiles peak temperatures were observed. For example, the DTG curves showed that increasing the heating rate resulted in i) a decrease in

lP

the secondary degradation reactions, and ii) an increase in the temperature at which the maximum depolymerisation rate occurred. At the same heating rate, MS ion current signals showed that maximum DL-limonene

formation occurred at slightly higher temperatures compared to the maximum isoprene

ur na

formation rate temperature. An increase in the heating rate from 15 to 100 °C/min increased the isoprene formation peak temperature from 380 to 432 °C, while that for DL-limonene was increased from 380 to 438 °C. This indicated that isoprene and DL-limonene formation pathways were not identical and the effect of the heating rate on the DL-limonene formation was more significant compared to the isoprene formation.

Jo

Moreover, the increases in the peak temperatures for DL-limonene formation reaction were less severe at higher heating rates. Devolatilisation of the polyisoprene (the rubbery component in the tyre) is characterised by volatilisation of unstable allylic radicals. These allylic radicals further react by i) depropagation (unzipping) to form isoprene or, ii) intramolecular cyclisation followed by chain scission to form DL-limonene. Therefore, the depropagation reaction pathway suggests that as the heating rate increases, its effect on the isoprene formation is becoming less significant, see Table 3. 12

4.3

Reaction progress

Figure 3 presents isoprene and DL-limonene formation or reaction progress as a function of temperature. The effect of the heating rate on the progress of the DL-limonene formation reaction was more significant than on isoprene formation, as per Figure 3 a) and b). The curves are relatively more vertical and parallel

of

for isoprene (Figure 3 a) while being less vertical and cease to be parallel towards the end of the reaction for DL-limonene (Figure 3 b). Maximum DL-limonene formation rate at peak reaction progress which

ro

corresponds to the peak temperature increased from 0.35 to 0.44, while that of isoprene merely increased from 0.42 to 0.45 as the heating rate was increased from 15 to 100 °C/min, see Figure 3 c). The relation

-p

between the heating rate and reaction progress at the maximum formation rate indicated a more rapid

re

conversion of the tyre rubber to DL-limonene than to isoprene, at similar temperature increases. Therefore, lower activation energy (Ea) at higher heating rates favoured DL-limonene formation compared to isoprene

Friedman method

ur na

4.4

lP

formation.

Estimation of the apparent activation energy, Ea, as a function of reaction conversion with the differentialbased, iso-conversional Friedman method is presented in Figure 4 a) and b), for isoprene and DL-limonene formation, respectively. In Figure 4 a) the Ea of isoprene formation remained relatively constant at

Jo

approximately 141 kJ/mole for a reaction progress up to α = 0.5. For the DL-limonene formation reaction, an Ea of approximately 145 kJ/mole was observed at reaction progress between 0.1 and 0.5. For the substantial increases in the Ea for both isoprene and DL-limonene above α = 0.5, the Friedman method ceases to be reliable as indicated by the large relative standard deviation. Capart et al. (2004), also observed a large standard error at α above 0.7 as well a sudden change in the slopes of the curves (Capart et al., 2004). The changes in the apparent Ea for DL-limonene formation, for α above 0.5, was more severe 13

than for isoprene formation. In Figure 4 b), the Ea not only ceases to be constant above α = 0.5 but became negative towards the completion of the DL-limonene formation.

The observations in Figure 4 a) and b) agree with the trends as illustrated in Figure 3 above, i.e. more uniformity in isoprene formation compared to DL-limonene formation. Based on merely Figure 3 which is within the reliable extent of reaction progress, this distinction can be attributed to the different reaction

of

pathways by which isoprene and DL-limonene are formed. Moreover, isoprene at ambient temperature is a gas, while limonene is normally a liquid. Therefore, DL-limonene must be maintained at the gaseous phase

ro

as it moves from the TGA furnace to the mass spectrometry detector, while isoprene may be less susceptible to condensation. This indicates that isoprene formation may be relatively less depended on the investigated

-p

parameters (temperature of 475 °C and heating rate up to 100 °C/min) compare to the DL-limonene

Kissinger method

lP

4.5

re

formation.

An alternative to Friedman method, the Kissinger method is an iso-conversional method based on an integration approach, which was also applied to estimate Ea for both isoprene and DL-limonene formation.

ur na

Figure 5 a) and b) show the Kissinger plots for isoprene and DL-limonene, respectively. The Kissinger method plots seem to be uniform and constitute adequately straight-line fits for both isoprene and DLlimonene. This can be attributed to the use of the peak temperatures when constructing the Kissinger plots, which are below nominal reaction progress of 0.5. Nominal reaction progress is reaction progress by which

Jo

the kinetic mechanism can be mathematically analysed using linear methods. The estimated Eas from Kissinger method were 129 and 113 kJ/mole for isoprene and DL-limonene, respectively.

Using Friedman method, the Ea for isoprene (141 kJ/mole) and DL-limonene (145 kJ/mole) formation were not significantly different, which may indicate the limitations of the Friedman method in estimating the Ea values. Friedman method limitations can be associated with the lack of taking into account the variation of 14

the Ea with reaction progress. Similar Ea values would suggest similar kinetic mechanisms for isoprene and DL-limonene

formation, which is unlikely since isoprene formation entails depropagation (unzipping) of

the polyisoprene allylic radicals, while DL-limonene is formed through intramolecular cyclisation of the allylic radicals followed by chain scission, see Error! Reference source not found. (Danon et al., 2015b).

However, using the Kissinger method the estimated Eas for isoprene (129 kJ/mole) and DL-limonene (113

Jo

ur na

lP

re

-p

ro

of

kJ/mole) formation were significantly different, see

15

Table 4. Aguado et al (2005) also observed higher Ea in the isoprene (108 kJ/mole) formation compared to DL-limonene

(97 kJ/mole), although the actual values were lower than those observed in the present study,

Jo

ur na

lP

re

-p

ro

of

see

16

Table 4. This is not surprising since the rate of the increase in the peak temperature for DL-limonene formation is higher than for the isoprene formation at the same heating rate.

4.6

Master plots

Estimation of the kinetic parameters (Ea, A, and n) for isoprene and DL-limonene formation model, reaction

of

order (n) was also performed graphically using model-based plots, called master plots. Various established models based on the solids devolatilisation were examined using master plots, i.e. half order (P3), first-

ro

order (F1), second-order (F2), phase boundary (R3) and two-dimensional diffusion (D2); see Table 1. The comparisons of the various models and experimental data are presented in Figure 6. Both isoprene (Figure

-p

6 a)) and DL-limonene (Figure 6 b)) experimental data profiles were between P3 and F1. Figure 6 c) and d)

re

illustrate a more detailed comparison of the P3 and F1 models to the experimental data for isoprene and DL-limonene formation, respectively. This observation implies that n (reaction order) for both isoprene and

Reaction model

ur na

4.7

formation is between 0.5 and 1.

lP

DL-limonene

The sum of the least square difference between the model and experimental data was applied to optimise the reaction models for isoprene and DL-limonene formation, using Matlab 2015a© programming software. As a first-guess, the Ea and A values obtained from the iso-conversional methods using the Kissinger method

Jo

were substituted in Eq. (12), together with a value of n = 0.7, between 0.5 and 1; results are shown in Figure 7. The subsequent regression further improved the model so that the model fit to the experimental data had the R2 (a measure of the amount of reduction in the variability) values 0.998 and 0.997 isoprene and DLlimonene, respectively. As shown in Figure 7, the best-fit models were comparable in terms of prediction of the trends of the main peak at all heating rates. The second peak (shoulder) on the high temperature side was poorly modelled, particularly for low heating rate experimental data. This was attributed to the fact that 17

at the lower heating rate the shoulder is relatively more significant, while at a high heating rate the shoulder is insignificant.

The values of Ea for isoprene and DL-limonene formation in the best-fit models were estimated at 132 and 115 kJ/mole, respectively, while the corresponding n values first guess as per master plots (Figure 6) was at 1, for both isoprene and DL-limonene. The Ea and A kinetic parameters did not change significantly during

of

model-optimisation, indicating that the model-free estimations of these values were effective. The n value

ro

of 2 for both isoprene and DL-limonene suggested second-order formation reactions.

The comparison of the experimental data to the model for both isoprene and DL-limonene was satisfactory

-p

and resulted in an R2 values closer to 1, i.e., 0.998 and 0.997, respectively. The difference in the Ea for

re

isoprene and DL-limonene indicate that their formation pathways differ. Therefore, it is possible to optimise the formation of DL-limonene over isoprene by setting the operating parameters. Although temperature and

lP

heating rate were the only parameters investigated in the present study, DL-limonene high yield and formation selectivity may have been depended on the other reactor operating parameters, such as residence

ur na

time of the hot volatiles on the reaction zones and the rate of cooling and condensing of the hot volatiles from the reactor (Mkhize et al., 2017).

5

Conclusions

Jo

The increase in the heating rate up to 100 °C/min increased the peak temperature of both the isoprene and DL-limonene

formation reactions, during waste tyre pyrolysis. The peak temperature increase was more

significant in the DL-limonene compared to the isoprene with increasing heating rate. In addition, maximum DL-limonene

formation rate was relatively higher compared to maximum isoprene formation rate as the

heating rate increased. This suggested that rapid heating resulted in the less energy was being allocated to the formation of products, therefore the increase in the heating rate favored the formation of the low Ea (i.e., 18

DL-limonene).

The model first-guess as provided from the master plots was able to precisely predict the

model for both for isoprene and DL-limonene. Model parameter estimations through the Kissinger method provided activation energies of approx. 133 and 115 kJ/mol for isoprene and DL-limonene formation reactions, respectively. Above α = 0.5, the Ea strongly increased for both isoprene and DL-limonene and the Friedman method ceased to be accurate as indicated by the large relative standard deviation. However, other factors, such as the hot volatiles residence time in the hot reaction zones and the rate of cooling of the hot

of

volatiles also affect DL-limonene formation from waste tyre pyrolysis. Therefore, to maximise DL-limonene

zones and the rate of cooling the hot volatiles should be considered.

re

-p

Acknowledgements

ro

yield and recovery, pyrolysis temperature, heating rate, the hot volatiles residence time in the hot reaction

This research was supported by the Recycling and Economic Development Initiative of South Africa

lP

(REDISA) and the National Research Foundation (NRF). The authors acknowledge that opinions, findings and conclusions or recommendations expressed are those of the authors only, and the sponsors accept no

ur na

liability whatsoever in this regard.

Declaration of Interest Statement The authors whose names are listed immediately below certify that they have NO affiliations with or involvement in any organization or entity with any financial interest (such as honoraria; educational grants;

Jo

participation in speakers’ bureaus; membership, employment, consultancies, stock ownership, or other equity interest; and expert testimony or patent-licensing arrangements), or non-financial interest (such as personal or professional relationships, affiliations, knowledge or beliefs) in the subject matter or materials discussed in this manuscript.

19

References

Aguado, R., Olazar, M., Vélez, D., Arabiourrutia, M., Bilbao, J., 2005. Kinetics of scrap tyre pyrolysis under

fast

heating

conditions.

J.

Anal.

Appl.

Pyrolysis

73,

290–298.

https://doi.org/10.1016/J.JAAP.2005.02.006 Antoniou, N., Zabaniotou, A., 2013. Features of an efficient and environmentally attractive used tyres

of

pyrolysis with energy and material recovery. Renew. Sustain. Energy Rev. 20, 539-558. https://doi.org/10.1016/j.rser.2012.12.005

kinetics

for

tyre

pyrolysis

modelling.

Eng.

Sci.

201,

362–372.

-p

https://doi.org/10.1016/J.CES.2019.02.025

Chem.

ro

Arabiourrutia, M., Lopez, G., Aguado, R., Bilbao, J., Olazar, M., 2019. Coupling gas flow pattern and

resembling

to

diesel

fuel

by

re

Aydin, H., Ilkiliç, C., 2012. Optimization of fuel production from waste vehicle tires by pyrolysis and various

lP

https://doi.org/10.1016/j.fuel.2012.06.067

desulfurization

methods.

Fuel

102,

605-612.

Banar, M., Akyildiz, V., Özkan, A., Çokaygil, Z., Onay, Ö., 2012. Characterization of pyrolytic oil obtained from pyrolysis

of

TDF

(Tire

Derived

Fuel).

Energy Convers.

Manag.

62,

22-30.

ur na

https://doi.org/10.1016/j.enconman.2012.03.019

Barlaz, M.A., Eleazer, W.E., Whittle, D.J., 1993. Potential To Use Waste Tires As Supplemental Fuel In Pulp And Paper Mill Boilers, Cement Kilns And In Road Pavement. Waste Manag. Res. 11, 463-480. https://doi.org/10.1006/wmre.1993.1050

Jo

Capart, R., Khezami, L., Burnham, A.K., 2004. Assessment of various kinetic models for the pyrolysis of a microgranular cellulose. Thermochim. Acta 417, 79-89. https://doi.org/10.1016/j.tca.2004.01.029

Chen, J.H., Chen, K.S., Tong, L.Y., 2001. On the pyrolysis kinetics of scrap automotive tires. J. Hazard. Mater. 84, 43–55. https://doi.org/10.1016/S0304-3894(01)00180-7 Cheung, K.Y., Lee, K.L., Lam, K.L., Chan, T.Y., Lee, C.W., Hui, C.W., 2011. Operation strategy for multistage pyrolysis. J. Anal. Appl. Pyrolysis. 91, 165-182. https://doi.org/10.1016/j.jaap.2011.02.004 20

Criado, J.M., 1978. Kinetic analysis of DTG data from master curves. Thermochim. Acta 24, 186-189. https://doi.org/10.1016/0040-6031(78)85151-X Danon, B., Görgens, J., 2015. Determining rubber composition of waste tyres using devolatilisation kinetics. Thermochim. Acta 621, 56-60. https://doi.org/10.1016/j.tca.2015.10.008 Danon, B., Mkhize, N.M., Van Der Gryp, P., Görgens, J.F., 2015a. Combined model-free and model-based devolatilisation

kinetics

of

tyre

rubbers.

Thermochim.

Acta

601,

45-53.

of

https://doi.org/10.1016/j.tca.2014.12.003 Danon, B., Van Der Gryp, P., Schwarz, C.E., Görgens, J.F., 2015b. A review of dipentene (dl-limonene) from

waste

tire

pyrolysis.

J.

Anal.

https://doi.org/10.1016/j.jaap.2014.12.025

Appl.

Pyrolysis

ro

production

112,

1-13.

-p

Friedman, H.L., 1964. Kinetics of thermal degradation of char-forming plastics from thermogrametry.

re

Application to a phenolic plastic. J. Polym. Sci. Part C Symp. 4, 183–195. Giugliano, M.M., Cernuschi, S., Ghezzi, U., Grosso, M., Giugliano, M.M., 1999. Evaluation of Waste Tires

lP

Utilization in Cement Kilns Experimental Evaluation of Waste Tires Utilization in Cement Kilns. J. Air Waste Manag. Assoc. J. Air Waste Manag. Assoc. J. Air Waste Manag. Assoc 49, 1405–1414.

ur na

https://doi.org/10.1080/10473289.1999.10463976doi.org/10.1080/10473289.1999.10463976 Ilkiliç, C., Aydin, H., 2011. Fuel production from waste vehicle tires by catalytic pyrolysis and its application

in

a

diesel

engine.

Fuel

Process.

Technol.

92,

1129-1135.

https://doi.org/10.1016/j.fuproc.2011.01.009

Jo

Kissinger, H.E., 1957. Reaction Kinetics in Differential Thermal Analysis. Anal. Chem. 27, 1702-1706. https://doi.org/10.1021/ac60131a045

Lam, K.-L., Lee, C.-W., Hui, C.-W., 2010. Multi-stage Waste Tyre Pyrolysis: An Optimisation Approach. Chem. Eng. Trans. 21, 853–858. https://doi.org/10.3303/CET1021143

Leung, D.Y.C., Wang, C.L., 1998. Kinetic study of scrap tyre pyrolysis and combustion. J. Anal. Appl. Pyrolysis 45, 153-169. https://doi.org/10.1016/S0165-2370(98)00065-5 Mkhize, N.M., Danon, B., van der Gryp, P., Görgens, J.F., 2017. Condensation of the hot volatiles from 21

waste

tyre

pyrolysis

by

quenching.

J.

Anal.

Appl.

Pyrolysis

124,

180-185.

https://doi.org/10.1016/j.jaap.2017.02.007 Mkhize, N.M., van der Gryp, P., Danon, B., Görgens, J.F., 2016. Effect of temperature and heating rate on limonene production from waste tyre pyrolysis. J. Anal. Appl. Pyrolysis 124, 180-185. https://doi.org/10.1016/j.jaap.2016.04.019 Mui, E.L.K., Ko, D.C.K., McKay, G., 2004. Production of active carbons from waste tyres––a review.

of

Carbon N. Y. 42, 2789–2805. https://doi.org/10.1016/J.CARBON.2004.06.023 Murugan, S., Ramaswamy, M.C., Nagarajan, G., 2008. Performance, emission and combustion studies of

ro

a DI diesel engine using Distilled Tyre pyrolysis oil-diesel blends. Fuel Process. Technol. 89, 152– 159. https://doi.org/10.1016/J.FUPROC.2007.08.005

-p

Olazar, M., Aguado, R., Vélez, D., Arabiourrutia, M., Bilbao, J., 2005. Kinetics of scrap tire pyrolysis in a

re

conical spouted bed reactor. Ind. Eng. Chem. Res. 44, 3918-2924. https://doi.org/10.1021/ie040259g Olazar, M., Lopez, G., Arabiourrutia, M., Elordi, G., Aguado, R., Bilbao, J., 2008. Kinetic modelling of

lP

tyre pyrolysis in a conical spouted bed reactor. J. Anal. Appl. Pyrolysis. 81, 127-132. https://doi.org/10.1016/j.jaap.2007.09.011

ur na

Perez-Maqueda, L.A., Ortega, A., Criado, J.M., 1996. The use of master plots for discriminating the kinetic model of solid state reactions from a single constant- rate thermal analysis (CRTA) experiment. Thermochim. Acta 277, 165–173.

Quek, A., Balasubramanian, R., 2013. Liquefaction of waste tires by pyrolysis for oil and chemicals - A

Jo

review. J. Anal. Appl. Pyrolysis. 101, 1-16. https://doi.org/10.1016/j.jaap.2013.02.016 Rofiqul Islam, M., Haniu, H., Rafiqul Alam Beg, M., 2008. Liquid fuels and chemicals from pyrolysis of motorcycle tire waste: Product yields, compositions and related properties. Fuel 8, 3112-3122. https://doi.org/10.1016/j.fuel.2008.04.036

Seidelt, S., Müller-Hagedorn, M., Bockhorn, H., 2006. Description of tire pyrolysis by thermal degradation behaviour

of

main

components.

J.

Anal.

Appl.

Pyrolysis

75,

11-18.

https://doi.org/10.1016/j.jaap.2005.03.002 22

Senneca, O., Salatino, P., Chirone, R., 1999. A fast heating-rate thermogravimetric study of the pyrolysis of scrap tyres. Fuel 78, 1575–1581. https://doi.org/10.1016/S0016-2361(99)00087-3 Sienkiewicz, M., Kucinska-Lipka, J., Janik, H., Balas, A., 2012. Progress in used tyres management in the European

Union:

A

review.

Waste

Manag.

32,

1742-1751.

https://doi.org/10.1016/j.wasman.2012.05.010 Stein, S., 2015. NIST Chemistry WebBook [WWW Document]. Nist Stand. Ref. Database.

of

Williams, P.T., 2013. Pyrolysis of waste tyres: A review. Waste Manag. 33, 1714-1728. https://doi.org/10.1016/j.wasman.2013.05.003

ro

Zabaniotou, A., Madau, P., Oudenne, P.D., Jung, C.G., Delplancke, M.P., Fontana, A., 2004. Active carbon production from used tire in two-stage procedure: Industrial pyrolysis and bench scale activation with

Jo

ur na

lP

re

-p

H2O-CO2mixture. J. Anal. Appl. Pyrolysis. 72, 289-297 https://doi.org/10.1016/j.jaap.2004.08.002

23

TG Weight loss (mg)

12

o

15 C/min o

10

25 C/min 50 oC/min

8

o

75 C/min 100 oC/min

6 4 2 0

200

400

600

800

o

of

Temperature ( C) a)

b)

ro

Figure 1: Profiles of the a) thermogravimetric (TG) and b) derivative thermogravimetric

Isoprene-67

-9

x 10

o

Ion current (A)

25 oC/min o

1

50 C/min

lP

1.5

75 oC/min o

100 C/min

0.5 0 0

200

o

15 C/min

ur na

Ion current (A)

15 C/min 2

Limonene-93

-10

x 10

re

2.5

-p

(DTG) of waste tyre devolatilisation.

400

600

o

Temperature ( C)

a)

800

1.5

25 oC/min o

50 C/min

1

75 oC/min o

100 C/min 0.5 0 0

200

400

600

800

o

Temperature ( C)

b)

Figure 2: MS Ion current signals of a) isoprene (67 amu) and b) DL-limonene (93 amu) at 15,

Jo

25, 50, 75 and 100 °C/min.

24

MS Reaction Progress, a (-)

0.8 o

15 C/min

0.6

25 oC/min

0.4

o

50 C/min 75 oC/min

0.2 0 200

100 oC/min 400

600

800

Limonene-93 1 0.8 15 oC/min

0.6

25 oC/min 0.4

50 oC/min

0.2

75 oC/min 100 oC/min

0 200

400

800

Temperature ( C)

b)

50

25

100 °C/min

75

0.46

ro

0.44 0.42

-p

0.4 0.38 0.36

Isoprene-67 Limonene-93

0.34 380

re

Reaction Progress, a (-)

Temperature ( C) a) Heating rate = 15

600 o

o

of

MS Reaction Progress, a (-)

Isoprene-67 1

400

420

440

o

Temperature ( C)

lP

c)

Figure 3: Reaction progress of a) isoprene (67 amu) and b) DL-limonene (93 amu) at 15, 25,

Jo

ur na

50, 75 and 100 °C/min.

25

of

a)

b)

ro

Figure 4: Iso-conversional Friedman method plot for a) isoprene and b)

as

Jo

ur na

lP

re

-p

various reaction progress.

DL-limonene

26

Limonene-93 18

17.5

17.5

17 16.5 16 15.5 1.4

1.45

1.5 -1

-1

T (K ) a)

1.55

17 16.5 16 15.5 1.4

1.45

1.5 -1

-3

-1

T (K ) b)

x 10

ro

Figure 5: Iso-conversional Kissinger method plot for a) isoprene and b)

-3

x 10

DL-limonene

at

Jo

ur na

lP

re

-p

various reaction progress.

1.55

of

ln(B/T2) (1/(oCs))

ln(B/T2) (1/(oCs))

Isoprene-67 18

27

3 2

] (-)

1

]exp (-) )2[(da/dt)/(da/dt)

0.27

P3 F1 exp data

0.3

a (-)

c)

0.4

0.2

0.5

0.3

b)

0.4

0.5

a (-)

DL-limonene-93 2 1.5

P3 F1 exp data

1

0.5

re

0.2

(T/T

0.27

0.5

-2 0.1

of

0.5

1

0 0.1

0

ro

0.4

Isoprene-67

1.5

2

-p

0.3

(T/T

0.2

a (-) a) 2

4

0.27

0 -1 0.1

DL-limonene-93 P3 F2 R3 D2 F1 exp data

6

0.27

)2[(da/dt)/(da/dt)

4

lP

] (-)

0.27

)2[(da/dt)/(da/dt)

0.27

(T/T

]exp (-)

0.27

)2[(da/dt)/(da/dt)

0.27

(T/T

Isoprene-67 P3 F2 R3 D2 F1 exp data

5

0 0.1

0.2

Figure 6: Master plots of various models for isoprene a) and

0.3

d)

0.4

0.5

a (-)

DL-limonene

e b) and zoomed

Jo

ur na

in of the half and first order for isoprene c) and DL-limonene d).

28

a)

of

b)

Jo

ur na

lP

re

-p

ro

Figure 7: Model fitting on the experimental data for isoprene a) and DL-limonene b).

29

Table 1: Typical solids decomposition models. 𝒇(𝜶)

𝒈(𝜶)

Symbol

Half order

𝛼 0.5

2𝛼 0.5

P3

First order

𝛼1

𝑙𝑛𝛼

F1

Second order

𝛼2

− 1⁄𝛼

F2

1⁄ 3

R3

Phase boundary

2⁄ 3

1⁄(𝑙𝑛𝛼 + 2)

3𝛼

𝛼𝑙𝑛𝛼 + 𝛼

D2

Jo

ur na

lP

re

-p

ro

Two-dimensional diffusion

3𝛼

of

Model

30

Table 2: Proximate analysis and rubber composition of the crumb. Proximate analysis (wt.%) Moisture

Oils

Volatile matter

Fixed carbon

Ash

0.6

5.6

56.0

30.0

7.8

Rubber composition/volatile matter (wt.%)

64

36

ro

of

Synthetic rubber (SBR and BR)α

By difference

Jo

ur na

lP

re

-p

α

Polyisoprene (natural rubber)

31

Table 3: Tyre devolatilisation DTG and products formation MS ion current peak temperatures at 15, 25, 50, 75 and 100 °C/min. Heating

rate DTG peak temperature

MS ion current peak Ion

(°C)

temperature (°C) α

25

396.67

50

410.38

β

380.48

Limonene-93

380.23

Isoprene-67

392.07

Limonene-93

393.74

Isoprene-67

Isoprene-67

α

415.45 422.54

Limonene-93

430.18

Isoprene-67

432.70

Limonene-93

437.80

lP

425.88

100

re

418.71

75

409.53

-p

Limonene-93

of

385.50

15

Isoprene-67

ro

(°C/min)

predominant fragmentation spectrum ion of limonene

Jo

β

ur na

predominant fragmentation spectrum ion of isoprene

32

Table 4: Kinetic parameter estimated using iso-conversional methods for isoprene and l DLlimonene. Ea (kJ/mol) Ion

Current work: Kissinger method

Literature [Aguado et al. (2005)]

129

108

Limonene-93

113

97

Jo

ur na

lP

re

-p

ro

of

Isoprene-67

33