Detailed chemical kinetic modeling of pyrolysis of ethylene, acetylene, and propylene at 1073-1373 K with a plug-flow reactor model

Article abstract of DOI:10.1002/kin.20302

This study examines the predictive capability of our recently proposed reaction mechanism (Norinaga and Deutschmann, Ind Eng Chem Res 2007, 46, 3547) for hydrocarbon pyrolysis at varying temperature. The conventional flow reactor experiments were conducted at 8 kPa, over the temperature range 1073-1373 K, using ethylene, acetylene, and propylene as reactants to validate the mechanism. More than 40 compounds were identified and quantitatively analyzed by on- and off-line gas chromatography. The chemical reaction schemes consisting of 227 species and 827 reactions were coupled with a plug-flow reactor model that incorporated the experimentally measured axial temperature profile of the reactor. Comparisons between the computations and the experiments are presented for more than 30 products including hydrogen and hydrocarbons ranging from methane to coronene as a function of temperature. The model can predict the compositions of major products (mole fractions larger than 10^-2) in the pyrolysis of three hydrocarbons with satisfactory accuracies over the whole temperature range considered. Mole fraction profiles of minor compounds including polycyclic aromatic hydrocarbons (PAHs) up to three ring systems, such as phenanthrene, anthracene, and phenylnaphthalene, are also fairly modeled. At temperatures lower than 1273 K, larger PAHs were underpredicted and the deviation became larger with decreasing temperature and increasing molecular mass of PAHs, while better agreements were found at temperatures higher than 1323 K.

Full text of DOI:10.1002/kin.20302

Detailed Chemical Kinetic
Modeling of Pyrolysis of
Ethylene, Acetylene, and
Propylene at 1073–1373 K
with a Plug-Flow Reactor
Model
KOYO NORINAGA,1 VINOD M. JANARDHANAN, OLAF DEUTSCHMANN1
,2
1
1
Institut f u¨ r Technische Chemie und Polymerchemie, Universit a¨ t Karlsruhe (TH), Engesserstr. 20, 76131 Karlsruhe, Germany
Center for Advanced Research of Energy Conversion Materials, Hokkaido University, N13-W8, Kitaku,
2
Sapporo 060-8628, Japan
Received 2 April 2007; revised 27 August 2007; accepted 6 November 2007
DOI 10.1002/kin.20302
Published online in Wiley InterScience (www.interscience.wiley.com).
ABSTRACT: This study examines the predictive capability of our recently proposed reaction
mechanism (Norinaga and Deutschmann, Ind Eng Chem Res 2007, 46, 3547) for hydrocarbon
pyrolysis at varying temperature. The conventional flow reactor experiments were conducted at
8
kPa, over the temperature range 1073–1373 K, using ethylene, acetylene, and propylene as re-
actants to validate the mechanism. More than 40 compounds were identified and quantitatively
analyzed by on- and off-line gas chromatography. The chemical reaction schemes consisting of
2
27 species and 827 reactions were coupled with a plug-flow reactor model that incorporated
the experimentally measured axial temperature profile of the reactor. Comparisons between
the computations and the experiments are presented for more than 30 products including
hydrogen and hydrocarbons ranging from methane to coronene as a function of temperature.
−2
The model can predict the compositions of major products (mole fractions larger than 10
)
in the pyrolysis of three hydrocarbons with satisfactory accuracies over the whole temperature
range considered. Mole fraction profiles of minor compounds including polycyclic aromatic
hydrocarbons (PAHs) up to three ring systems, such as phenanthrene, anthracene, and phenyl-
naphthalene, are also fairly modeled. At temperatures lower than 1273 K, larger PAHs were
underpredicted and the deviation became larger with decreasing temperature and increasing
molecular mass of PAHs, while better agreements were found at temperatures higher than
1
323 K. ꢀC 2008 Wiley Periodicals, Inc. Int J Chem Kinet 40: 199–208, 2008
Correspondence to: Koyo Norinaga; e-mail: norinaga@eng.
hokudai.ac.jp.
c 2008 Wiley Periodicals, Inc.
ꢀ

2
00
NORINAGA, JANARDHANAN, AND DEUTSCHMANN
INTRODUCTION
a single temperature (1173 K) and the computation
without a transport model.
When heated at around 1273 K, hydrocarbon generally
undergoes pyrolysis where hydrocarbons with increas-
ing numbers of carbon atoms and aromatic rings are
progressively formed, and theoretically they convert
almost completely into solid carbon and gaseous hy-
drogen at equilibrium [1]. The carbon formation from
gaseous light hydrocarbons has both detrimental and
positive aspects. The former is catalyst deactivations
by coke formations [2–4]. The latter is syntheses of
carbon materials such as carbon–carbon composites
In this study, we examine the predictive capability
of our recently developed mechanism at varying tem-
perature. Pyrolysis experiments were performed using
the same flow reactor at pressure of 8 kPa, residence
time of 0.5 s, and over the temperature range of 1073–
1373 K. Numerical simulation is conducted with a soft-
ware package, which can couple the detailed chemical
reaction schemes with a plug-flow reactor model, in
which the temperature profiles along the reactor are
adequately incorporated.
[5] by utilizing a chemical vapor deposition (CVD)
technique. This study is performed at conditions rele-
vant to the CVD of carbon in which a great variety of
hydrocarbons and hydrocarbon radicals are formed by
gas-phase reactions, and any of these species has a po-
tential to chemisorb or physisorb at the growing solid
carbon surface [6]. The mechanism of carbon forma-
tion from gas phase is still obscure, since underlying
chemistry is quiet complex involving both gas phase
and surface.
We conducted an experimental study on a detailed
product analysis of the pyrolysis of ethylene, acetylene,
and propylene [7], because quantitative experimental
data on the hydrocarbon pyrolysis with a wide range of
product analysis including large polycyclic aromatic
hydrocarbons (PAHs) had been still limited. The ex-
periments were performed with a conventional flow
reactor at conditions relevant to the CVD of pyrolytic
carbon, i.e., at 1173 K and pressure from 2 to 15 kPa.
Gas-phase components were analyzed by both on- and
off-line gas chromatography. More than 40 compounds
including hydrogen and hydrocarbons from methane to
coronene were identified and quantitatively determined
as a function of the residence time varied up to 1.6
s. Product recoveries were generally more than 90%,
providing a useful experimental database for kinetic-
modeling studies of gas-phase reactions with detailed
chemistry.
EXPERIMENTAL
The experimental data for the model validation are ob-
tained using a vertical flow reactor with a total length of
440 mm as shown in Fig. 1 (left). The reactor employed
here is identical with that used in previous CVD exper-
iments [9]. A channel structure, made out of cordierite,
with 400 channels per square inch, is fitted in an alu-
mina ceramic tube (22-mm i.d. and 40 mm long), which
is located at the center of the reactor and called the
We developed a gas-phase reaction mechanism con-
sisting of 227 species and 827 reactions [8], which was
evaluated using a software package for time-dependent
homogeneous reaction systems. Comparisons between
the experiments and computations for more than 30
compounds showed that the mechanism has a compre-
hensive capability in predicting concentration profiles
of major components found in the pyrolysis of three
different hydrocarbons with an acceptable level of ac-
curacy, although disagreements are still observed for
minor species, especially for PAHs, for which forma-
tions are underpredicted. Obvious shortcomings of the
previous work [8] were the model validation at only
Figure 1 Schematic diagrams of the real reactor (left) and
the reactor idealized for a plug-flow simulation (right); unit:
mm.
International Journal of Chemical Kinetics DOI 10.1002/kin

CHEMICAL KINETIC MODELING OF PYROLYSIS OF ETHYLENE, ACETYLENE, AND PROPYLENE
201
deposition space. The inlet and outlet tubes (8-mm i.d.)
are connected to the deposition space through conical
nozzles. The temperature of the deposition space was
varied in the range between 1073 and 1373 K. Tem-
perature profile for the reactor was measured under
an argon flow with a type K thermocouple (R o¨ ssel
Messtechnik GmbH & Co.) that was moved axially
along the reactor length.
Ethylene, acetylene, and propylene, purchased from
Air Liquide Co. Ltd., were used as source hydrocar-
bons. The purities of the source hydrocarbons and the
initial compositions used in the computations are sum-
marized in our previous work [8]. The pressure was
maintained at 8 kPa without inert diluent gas. The flow
rate of inlet source gas was controlled so that the space-
time at the isothermal zone was fixed to 0.5 s. In the
determination of the inlet velocity, the density change
of the flowing gas by chemical reactions was not
considered.
The products were analyzed by both on- and off-line
gas chromatography. Gaseous products up to C4 com-
pounds were analyzed on-line with a Sichromat 3 gas
chromatograph (Siemens) equipped with a vacuum-
dosing system. A Porapak N column (Chrompak) and a
thermal conductivity detector were used for separation
and peak detection, respectively. We conducted GC
measurements for gaseous products with two different
modes, one for measurements of hydrogen and C1–
C2 hydrocarbons using argon as a carrier gas, and the
other for C3–C4 hydrocarbons using hydrogen as a car-
rier gas. Liquid products larger than benzene were col-
lected in two cold traps set at 195 K, dissolved in a mea-
sured amount of acetone and analyzed by a Sichromat
data for all species and the species abbreviations,
and the corresponding chemical formula are given in
the previous work [8] and can also be downloaded
(www.detchem.com). The conversion of source hydro-
carbon into the solid carbon is generally a few percent
(C1 base) for ethylene and propylene, but it accounts
for several percent for acetylene [9]. The development
of surface reaction mechanism is currently in progress
for more accurate description of the overall process in
the reactor.
The appropriate treatment of pressure-dependent re-
actions such as dissociation and recombination reac-
tions is important in this study because the pyrolysis
experiments run at low pressures where the pressure
has a significant effect on the rate constant. The mech-
anism involves 33 reactions with third-body species,
and 31 reactions of which are given using the so-called
Troe parameters [10–13]. The impact of the pressure in
the comparisons of experiment and prediction data was
examined in our previous paper at 1173 K and pres-
sures ranged from 2 to 15 kPa [8]. It was found that
the kinetic model adequately captures the trends of the
pressure effect for major species; an increase in pres-
sure enhances consumption of source hydrocarbons
and also enhances product formation. This suggests
that at least a portion of pressure-dependent reactions
that are sensitive for the formation/consumption of the
major species is treated adequately in the current ki-
netic model, although not all of pressure-dependent
reactions are provided with third-body species or the
Troe parameters.
Reactor Idealization
1-4 gas chromatograph (Siemens) equipped with a cap-
illary column (CP-Sil 8 CB LB/MS, Chrompak) and
a flame ionization detector. Species were identified by
the retention time matching. The estimated uncertain-
ties in determining the concentrations of gaseous, ma-
jor condensing products, and minor condensing prod-
ucts (mainly PAHs) were ± 9%, ± 28%, and ± 32%, re-
spectively. The detailed descriptions of the experimen-
tal setup and product analysis are given elsewhere [7].
Figure 1 shows schematic diagrams of the real reactor
(left) and idealized reactor (right). The real reactor was
idealized to a tube reactor with a constant inner diam-
eter for one-dimensional plug-flow simulation so that
introduced gas experiences the same temperature his-
tory as it does in the real reactor. The idealized reactor
has the same length of the real reactor, and its inner
diameter was determined to make it to have the same
total volume of the real reactor. The length of the de-
position space of the idealized reactor was determined
to make it to have the same volume of that of the real
reactor. This idealization is valid as long as laminar
flow is maintained throughout the reactor. The flow
characteristics in the same reactor were studied with a
1:1 acrylic-glass model [14], in which an ammonium
chloride smoke gas was used for flow visualization. It
was shown that a clear plug-flow profile was gener-
ated by the conical nozzles, which avoid an occurrence
of turbulent flow at the reactor parts of varying tube
diameter.
MODELING
Mechanism
The gas-phase reaction mechanism used in this study
consists of 227 species and 827 elementary reac-
tions, of which 798 reactions are reversible. No ad-
justments were made to the kinetic parameters of
any chemical reaction. The mechanism, the reference
of each of elementary reactions, the thermodynamic
International Journal of Chemical Kinetics DOI 10.1002/kin

2
02
NORINAGA, JANARDHANAN, AND DEUTSCHMANN
.
Plug-Flow Computation
of species k by surface reaction, ω is the molar rate
k
of production of species k by the gas-phase reaction,
Mk is the molecular mass of species k, Yk is the mass
fraction of species k, Cp is the specific heat capacity of
species k, hk is the specific enthalpy of species k, U is
the overall heat transfer coefficient, Tw is the wall tem-
perature, T is the gas temperature, p is the pressure,
and M is the average molecular weight. However, since
The calculations were performed using the gas-phase
reaction mechanism with the PLUG code in the
PLUG
DETCHEM program package (DETHCEM
The DETHCEM
) [15].
PLUG
is designed for nondispersive
one-dimensional flow of chemically reacting ideal gas
mixture under steady-state conditions. The system of
differential algebraic equations describing the plug-
flow reactor consists of the continuity equation
the study carried out here does not account for surface
.
reactions, all the terms containing the term sk vanishes.
Furthermore, the energy equation is not solved since
the measured temperature profile in the reactor is in-
corporated into the code using a separate subroutine.
However, the equation is given here for the sake of
completeness. The system of equations is solved using
the differential algebraic equation solver LIMEX.
The measured (symbols) and modeled (lines) tem-
perature profiles of the reactor are shown in Fig. 2.
The linear lines connecting the neighboring points cor-
respond to the temperature profile input for the com-
putations. The other input parameters were obtained
from the experimental conditions such as inlet velocity,
molar composition, pressure, and so on. The computa-
tions were performed for seven different temperature
profiles with the temperature at quasi-isothermal zone
ranging from 1073 to 1373 K.
Kg
ꢀ
d(ρu)
dz
.
Ac
= As
skMk
k=1
the species conservation equation
.
d(ρuYk)
dz
.
Ac
= Mk(As sk +Ac ωk)
the energy equation
Kg
Kg
ꢀ
ꢀ
d(CpT )
dz
.
.
ρuAc
+
ωkhkMkAc +
skhkMkAs =
k=1
k=1
UAs(Tw − T )
and the equation of state
The gas temperature profile inside the reactor was
also examined by numerical simulations with a two-
pM = ρRT
CHANNEL
dimensional reactor model using DETCHEM
In the above equations, ρ is the density, u is the
velocity, Ac is the area of cross section of the channel,
[15]. A laminar flow model is coupled with the detailed
chemical reaction schemes identical to that used in
the present study. The simulation shows that gas tem-
perature is almost uniformly distributed in the radial
As is the surface area per unit length, Kg is the number
.
of gas-phase species, sk is the molar rate of production
Figure 2 Experimentally obtained temperature profiles along the reactor at seven temperatures of the quasi-isothermal zone.
PLUG
The measured points are connected by linear lines and used for the DETCHEM
in the online issue, which is available at www.interscience.wiley.com.]
simulations. [Color figure can be viewed
International Journal of Chemical Kinetics DOI 10.1002/kin

CHEMICAL KINETIC MODELING OF PYROLYSIS OF ETHYLENE, ACETYLENE, AND PROPYLENE
203
direction along the quasi-isothermal zone of the reac-
tor when the inner diameter of the tube equals to that
of the reactor used in this study. Hence, errors in nu-
merical simulations with neglecting radial temperature
distribution should be minimal.
Reaction Pathway Analysis
Reaction pathways leading to benzene were analyzed
to identify crucial reactions for benzene formation and
to assess the effect of temperature on the contributions
of individual reactions. Rates of formation for species
such as benzene and the precursors were determined
using a rate analysis code included in the HOMREA
software package [16]. The reaction pathway analysis
was performed at a condition of p = 8 kPa, residence
time = 0.5 s, and temperatures from 1073 to 1373 K.
RESULTS OF COMPUTATIONS
Velocity of Chemically Reacting Flow
along the Reactor
Figure 3 shows the wall temperature profile in which
the quasi-isothermal zone temperature is 1273 K used
for the computations (upper) and the computational
outputs for the velocity profiles for ethylene, acety-
Figure 3 Wall temperature profile in which the quasi-
isothermal zone temperature is 1273 K used for the com-
putations (upper) and the computational outputs for the
velocity profiles in the ethylene, acetylene, and propylene
pyrolysis (lower) along the reactor length. [Color figure
can be viewed in the online issue, which is available at
www.interscience.wiley.com.]
lene, and propylene pyrolysis (lower) along the reactor
length. Up to the reactor length of 0.12 m, all gases
exhibit the same velocity profiles and their trends are
exactly same as the wall temperature profiles. This in-
dicates that the velocity increase, in other words the
density decrease, is induced simply by the temperature
effect. Chemical reactions changing the fluid density
are here negligible. Beyond 0.12 m, at which the wall
temperature is 1120 K, three gases show different ve-
locity profiles. Rapid velocity increases are observed
for propylene and ethylene demonstrating that these
gases undergo pyrolysis reactions dominated by de-
compositions, which lead to density decrease. On the
other hand, the velocity of acetylene shows an incipi-
ent small increase and then gradually decreases. This
indicates that the acetylene pyrolysis is primarily domi-
nated by recombination reactions leading to density in-
crease. After the reactor length of 0.29 m, velocities of
three gases decrease with decreasing temperature. The
velocity at the reactor outlet is in the order of propylene
(b) acetylene, and (c) propylene at the isothermal zone
temperature of 1273 K. At 0.12 m, all source gases
start to destruct and various products start to form.
C4 species such as vinylacetylene and 1,3,-butadiene
as well as C3 species such as propyne and propadi-
ene exhibit intermediate behaviors of which concentra-
tions go through maxima at short reactor length. Little
changes in any species’ concentration occur beyond
0.3 m due to the temperature drop.
COMPARISONS OF EXPERIMENTS
AND COMPUTATIONS
>
ethylene > acetylene. These simulation results well
correspond to our previous experimental results on the
velocity measurements at the reactor outlet [7].
Major Compounds
Figures 5–7 compare the computational predictions
at the reactor outlet (length = 0.44 m) with the
experimental data for major compounds (mole frac-
Gas-Phase Composition along the Reactor
−3
Figure 4 shows the computed mole fraction profiles
of major compounds in the pyrolysis of (a) ethylene,
tions >10 ) as a function of the quasi-isothermal zone
temperature. In general, the major species predictions
International Journal of Chemical Kinetics DOI 10.1002/kin

2
04
NORINAGA, JANARDHANAN, AND DEUTSCHMANN
Figure 5 Comparison of the computational predictions at
the reactor outlet (length = 0.44 m) with the experimen-
−3
tal data for major compounds (mole fractions >10 ) in
ethylene pyrolysis at 8 kPa as a function of the quasi-
isothermal zone temperature. Lines and symbols are the
predictions and the experiments, respectively. [Color fig-
ure can be viewed in the online issue, which is available
at www.interscience.wiley.com.]
Figure 4 Computed mole fraction profiles of major com-
pounds in the pyrolysis of (a) ethylene, (b) acetylene, (c)
and propylene at the isothermal zone temperature of 1273
K, pressure of 8 kPa, and reactant inlet velocity of 0.05 m/s.
ered when the space-time at quasi-isothermal zone is
fixed to 0.5 s. The residence time is an important pa-
rameter to consider. Although the impact of residence
time on the comparison between the model and the
experiments was thoroughly examined at 1173 K [8],
this does not guarantee that the present kinetic model
can predict successfully the pyrolysis behaviors of the
three hydrocarbons over a range of residence time at
other temperatures.
[Color figure can be viewed in the online issue, which is
available at www.interscience.wiley.com.]
agree well with the experimental data over the whole
temperature range, considering that the model param-
eters were never fitted to match the experimental data.
Exceptions are the intermediate species such as 1,3-
butadine for ethylene pyrolysis and vinylacetylene for
both ethylene and acetylene pyrolysis, which are over-
predicted by factors of 3–4, but their trends are captured
well. The hydrogen mole fraction in acetylene pyroly-
sis at high temperatures (ꢁ1223 K) is underpredicted
by about 35–100%. Carbon deposition from hydrocar-
bon always accompanies hydrogen formation. The ex-
clusion of heterogeneous reactions in the present simu-
lation results in the significant disagreements in hydro-
gen formation from acetylene, which has the highest
propensity to form carbon deposit among three hydro-
carbons examined [9].
Minor Compounds
Figure 8 presents the computational and experimen-
tal mole fractions of minor compounds in the ethy-
lene pyrolysis. The minor compounds are classified
into C3, C4 hydrocarbons (upper), benzene substitutes
(middle), and compounds with two rings and acenaph-
thylene (bottom). Vinylacetylene is ∼3-fold overpre-
dicted, but the profiles of other species are fairly pre-
dicted and their trends are well captured. For the ben-
zene substitutes, an overprediction (∼3-fold) of styrene
at low temperatures is observed. All the prediction
curves show maxima, and the same trends are ob-
served experimentally for toluene and styrene, though
the mole fraction peaks appear at higher temperatures
The model can predict the profiles of major products
in the pyrolysis of three hydrocarbons with satisfactory
accuracies over the whole temperature range consid-
International Journal of Chemical Kinetics DOI 10.1002/kin

CHEMICAL KINETIC MODELING OF PYROLYSIS OF ETHYLENE, ACETYLENE, AND PROPYLENE
205
Figure 6 Comparison of the computational predictions at
the reactor outlet (length = 0.44 m) with the experimen-
Figure 7 Comparison of the computational predictions at
the reactor outlet (length = 0.44 m) with the experimen-
−3
−3
tal data for major compounds (mole fractions >10 ) in
acetylene pyrolysis at 8 kPa as a function of the quasi-
isothermal zone temperature. Lines and symbols are the
predictions and the experiments, respectively. [Color fig-
ure can be viewed in the online issue, which is available
at www.interscience.wiley.com.]
tal data for major compounds (mole fractions >10 ) in
propylene pyrolysis at 8 kPa as a function of the quasi-
isothermal zone temperature. Lines and symbols are the
predictions and the experiments, respectively. [Color fig-
ure can be viewed in the online issue, which is available
at www.interscience.wiley.com.]
to separate the peaks from xylene, ethylbenzene, and
phenylacetylene in our current GC measurement. The
agreements for the other species would generally be
acceptable except for a significant overprediction of
biphenyl (by a factor of about 40) at lower tempera-
ture, considering that the absolute concentration values
than those by predictions. Experimental mole fraction
of phenylacetylene increases monotonously with in-
creasing temperature unlike the prediction. The reac-
tion pathway analysis shows that phenylacetylene is
consumed almost solely by phenylacetylene + H →
C6H5 (phenyl radical) + acetylene at 1073 K. With
increasing temperature, contribution of this reaction is
decreased instead the contributions of phenylacetylene
−6
−3
are as small as 10 to 10 . Similar results obtained
for the minor products in acetylene and propylene py-
1
rolysis are given in the Supporting Information.
+
C2H3 → C6H5 + vinylacetylene and phenylacety-
lene + phenyl → phenanthrene + H become signifi-
cant. The pathway analysis also indicates that styrene
dehydrogenation (styrene + H → C6H5CCH2 + H2
and C6H5CCH2 + H → phenylacetyrene + H2) is the
preferred phenylacetylene production route over the
temperature range examined. With increasing temper-
ature, the C6H5 + acetylene → phenylacetylene +
H pathway becomes important and contributes about
Polycyclic Aromatic Hydrocarbons
The computational and experimental mole fractions of
large PAHs (≥ phenanthrene) are shown in Fig. 9 for
acetylene pyrolysis. The same figures for other two hy-
drocarbon sources (ethylene and propylene) are given
in the Supporting Information. Mole fraction profiles
of PAHs such as phenanthrene, anthracene, phenyl-
naphthalene, and fluoranthene are fairly predicted. At
temperatures lower than 1273 K, larger PAHs are
20% of the phenylacetylene formation at 1373 K. The
examinations of the rate and thermodynamic param-
eters of the above-mentioned reactions may improve
the prediction of the phenylacetylene concentration.
On the experimental side, the quantification of phenyl-
acetylene should be further examined since it is hard
1Supplementary information is available at http://www.
interscience.wiley.com/jpages/0538-8066/suppmat/.
International Journal of Chemical Kinetics DOI 10.1002/kin

2
06
NORINAGA, JANARDHANAN, AND DEUTSCHMANN
Figure 8 Comparison of the computational predictions at
the reactor outlet (length = 0.44 m) with the experimental
data for minor compounds in ethylene pyrolysis at 8 kPa as
a function of the quasi-isothermal zone temperature. Lines
and symbols are the predictions and the experiments, respec-
tively. [Color figure can be viewed in the online issue, which
is available at www.interscience.wiley.com.]
Figure 9 Comparison of the computational predictions at
the reactor outlet (length = 0.44 m) with the experimental
data for polycyclic aromatic hydrocarbons in acetylene pyrol-
ysis at 8 kPa as a function of the quasi-isothermal zone tem-
perature. Lines and symbols are the predictions and the exper-
iments, respectively. [Color figure can be viewed in the online
issue, which is available at www.interscience.wiley.com.]
underpredicted. The deviations appear to be larger with
decreasing temperature and increasing molecular mass
of PAHs. The underpredictions of 2–5 orders of mag-
nitude are observed at low temperatures. These gaps
become smaller with increasing temperature, and bet-
ter agreements are obtained at temperature larger than
would help to close the gaps at the lower tempera-
ture. Nevertheless, the increasing gap with increas-
ing PAHs sizes and decreasing temperature strongly
suggests the further necessity of incorporation of the
transport model such as molecular diffusion of the
species in both axial and radial directions. The present
pyrolysis experiments were carried out at reduced pres-
sures and relatively low flow rates with no carrier gas.
1323 K. The PAHs formation mechanism is acquired
from the high-temperature flame modeling study by
Richter and Howard [17] and would be applicable only
at high temperature. Further explorations of new routes
to PAHs as well as improvements of kinetic parame-
ters of the reactions sensitive to PAHs concentrations
International Journal of Chemical Kinetics DOI 10.1002/kin

CHEMICAL KINETIC MODELING OF PYROLYSIS OF ETHYLENE, ACETYLENE, AND PROPYLENE
207
Under such conditions, the fluid has very low Reynolds
numbers. An axial diffusion [18] is then likely to have
a significant effect on the overall flow characteristics
in the reactor.
bination between vinylacetylene and acetylene (vinyl-
acetylene + acetylene → benzene) is the major route to
benzene. Vinylacetylene is formed almost solely by the
dimerization of acetylene. Temperature has little effect
on the reaction pathways of the benzene formation.
In the propylene pyrolysis, combinations between
C3 species such as C3H5 (allyl radical) + C3H3 → ben-
zene + H + H and propadiene + C3H3 → benzene + H
are the major routes to benzene. Eighty-nine percent
of benzene is produced by the former reaction at 1073
K, but its contribution decreases with increasing tem-
perature and exhibits 35% at 1373 K. The latter re-
action contributes a few percent of benzene forma-
tion at 1073 K, but the contribution increases with in-
creasing temperature and is 13% at 1373 K. The other
important route to benzene is 1,3-cyclohexadienyl →
benzene + H as also observed in the ethylene pyroly-
sis. The contribution becomes important with increas-
ing temperature (6% at 1073 K and 17% at 1373 K).
Unlike the ethylene pyrolysis, 1,3-cyclohexadiennyl is
formed almost solely by the isomerization of 1-methyl-
cyclopentadienyl over the temperature range exam-
ined. Most of 1-methyl-cyclopentadienyl is formed
by two reactions such as 1-methyl-cyclopentadiene
→ 1-methyl-cyclopentadienyl + H and cyclopentadi-
enyl + CH3 → 1-methyl-cyclopentadienyl + H, and
their contributions are 53% and 43% at 1373 K, respec-
tively. Temperature has little effect on the contributions
of the two reactions to 1-methyl-cyclopentadienyl for-
mation. Much abundance of cyclopentadiene, which
is mainly formed by C3H5 (allyl radical) + C2H2 →
cyclpentadiene + H, cyclopentadienyl produced from
cyclopentadiene, and methyl radical supports benzene
production route through C5 species in the pyrolysis
of propylene. Methyl and cyclopentadienyl are impor-
tant radicals as well as H radical in the propylene
pyrolysis. For instance, more than 60% of allyl rad-
ical is formed by H abstraction of propylene attacked
by methyl and cyclopentadienyl radicals, whereas the
reaction C3H6 + H → allyl + H2 contributes 25% of
allyl formation at 1223 K. Considerations of H ab-
stractions from PAHs by radicals except for H rad-
ical would improve the prediction of PAHs forma-
tion of which growth pathways are presently involved
only in the hydrogen-abstraction/acetylene-addition
(HACA) mechanism that was introduced in the high-
temperature combustion study [19–21].
ANALYSIS OF REACTION PATHWAYS
LEADING TO BENZENE
The results of the reaction pathway analysis for the
major species at 1173 K were reported in the previous
paper [8], and the major reaction routes to benzene
were identified. Here the effect of temperature on the
contributions of the crucial reactions leading to ben-
zene, which is the most abundant aromatics in the gas
phase, is assessed.
In the ethylene pyrolysis, benzene is formed primar-
ily by dehydrogenation of 1,3-cyclohexadiene (1,3-
cyclohexadiene → benzene + H2). The contribution
of this reaction is decreased from 82% to 34%
with increasing temperature from 1073 to 1373 K.
Benzene is produced secondarily (∼13%) by 1,3-
cyclohexadienyl → benzene + H at 1073 K, but the
contribution is decreased down to 5% at 1373 K. The
1,3-cyclohexadine is formed almost solely by isomer-
ization of 1,3-hexatriene, which is formed also almost
solely by the combination of 1,3-butadine and vinyl
radical (1,3-butadiene + C2H3 → 1,3-hexatriene + H).
1,3-Cyclohexadienyl is produced from by hydro-
gen abstraction of 1,3-cyclohexadiene such as
1
,3-cyclohexadiene → 1,3-cyclohexadienyl + H and
1,3-cyclohexadiene + H → 1,3-cyclohexadienyl + H2.
These reactions dominate the 1,3-cyclohexadienyl for-
mation at low and intermediate temperatures. Iso-
merization of 1-methyl-cyclopentadienyl becomes sig-
nificant with increasing temperature and contributes
15% of the 1,3-cyclohexadienyl formation at 1373 K.
The contribution of the cyclization of 1,3-hexadiene-
5
-yne (1,3-hexadiene-5-yne + H → benzene + H) for
the benzene formation is less than 10% over the
temperatures examined. Instead of the formation of
benzene from C6 species, contributions of combina-
tions of C3 species (propadiene + C3H3 (propargyl)
→
benzene + H, C3H5 (allyl radical) + C3H3 → ben-
zene + H + H) and combinations of C4 and C2 species
(
vinylacetylene + acetylene → benzene) become im-
portant with increasing temperature. The former and
the latter reactions account for 10% and 11% of
the benzene formation at 1373 K. Decomposition of
toluene and styrene also contributes to the benzene for-
mation at high temperatures, and the sum of their con-
tributions to the benzene formation is 17% at 1373 K.
The chemistry of acetylene pyrolysis is simplest
among the three hydrocarbons examined. The com-
CONCLUSION
Our recently developed gas-phase reaction mechanism
for the pyrolysis of ethylene, acetylene, and propy-
lene, which had been validated only at 1173 K, was
International Journal of Chemical Kinetics DOI 10.1002/kin

2
08
NORINAGA, JANARDHANAN, AND DEUTSCHMANN
found to exhibit an acceptable capability in predicting
the product distributions even at varying tempera-
ture ranging from 1073 to 1373 K. Advanced from
the previous completely mixed reactor model (zero-
dimensional simulation), the detailed chemical kinetic
schemes were coupled with a plug-flow reactor model
that incorporates the axial temperature profiles of the
reactor. The computational predictions were compared
with the experimental data obtained with a conven-
tional flow reactor for 35 species including hydrogen
and hydrocarbons ranging from methane to coronene.
Such an extensive model validation was not found in
other detailed chemical kinetic studies for light hy-
drocarbon pyrolysis [22–31]. The model can predict
the profiles of major products (mole fractions larger
than 10 ) in the pyrolysis of three hydrocarbons with
satisfactory accuracies over whole temperature range.
Mole fraction profiles of minor compounds including
PAHs up to three ring systems are also fairly mod-
eled. Although the predictions are still not perfect, it
is encouraging that the detailed mechanism developed
by compiling elementary reactions could be verified
at comprehensive operational conditions without any
modification and to have a merit in its generality which
is never expected in kinetic models with rate constants
determined by numerical fittings. Underpredictions are
observed for the larger PAHs at temperature lower than
Dr. H. H. Carstensen (Collorado School of Mines) for his
useful comments on the reaction mechanism.
BIBLIOGRAPHY
1
. Benzinger, W.; Becker, A.; H u¨ ttinger, K. J. Carbon 1996,
34, 957.
2. Guisnet, M.; Magnoux, P. Appl Catal 1989, 54, 1.
3. Guisnet, M.; Magnoux, P. Stud Surf Sci Catal 1994, 88,
53.
4
5
. Guisnet, M.; Magnoux, P. Catal Today 1997, 36, 477.
. Savage, G. Carbon-Carbon Composite; Chapmann &
Hall: London, 1993.
6
7
. H u¨ ttinger, K. J. Chem Vapor Deposition 1998, 4, 151.
. Norinaga, K.; Deutschmann, O.; H u¨ ttinger, K. J. Carbon
2
006, 44, 1790.
. Norinaga, K.; Deutschmann, O. Ind Eng Chem Res
007, 46, 3547.
. Norinaga, K.; H u¨ ttinger, K. J. Carbon 2003, 41, 1509.
−2
8
2
9
10. Troe, J. J Chem Phys 1977, 66, 4745.
11. Troe, J. J Chem Phys 1977, 66, 4758.
12. Troe, J. J Phys Chem 1979, 83, 114.
13. Troe, J. Ber Bunsen-ges Phys Chem 1983, 87, 161.
1
4. Becker, A.; H u¨ ttinger, K. J. Carbon 1998, 36, 177.
15. Deutschmann, O.; Tischer, S.; Kleditzsch, S.;
Janardhanan, V. M.; Correa, C.; Chatterjee, D.;
Mladenov, N.; Minh, H. D. DETCHEM Software
package, 2.1 ed., Karlsruhe 2007, www.detchem.com,
accessed December 1, 2007.
1273 K at which the deviation becomes larger with de-
1
6. Warnatz, J.; Maas, U.; Dibble, R. W. Combustion, 3rd
ed.; Springer-Verlag: Heidelberg, 2000.
creasing temperature and increasing molecular mass of
PAHs, while better agreements are found at tempera-
ture higher than 1323 K. Several factors such as lack of
chemistry for PAHs formation, inadequate kinetic pa-
rameters for the reactions sensitive to PAHs concentra-
tions at lower temperature, and ignorance of diffusion
of species as well as surface reactions attribute to the
gaps found in the PAHs concentrations. Future works
will include studies with the transport model such as
molecular diffusion of the species in both axial and
radial directions and surface (deposition) chemistry.
17. Richter, H.; Howard, J. B. Phys Chem Chem Phys 2002,
4, 2038.
18. Andra, H.; Langhoff, T. A.; Schnack, E.; H u¨ ttinger, K.
J. Fuel 2001, 80, 1273.
1
9. Frenklach, M.; Wang, H. Proc Combust Inst 1991, 23,
559.
0. Wang, H.; Frenklach, M. Combust Flame 1997, 110,
73.
1
2
1
2
2
1. Frenklach, M. Phys Chem Chem Phys 2002, 4, 2028.
2. Dagaut, P.; Boettner, J. C.; Cathonnet, M. Int J Chem
Kinet 1990, 22, 641.
2
2
3. Dean, A. M. J Phys Chem 1990, 94, 1432.
4. Dagaut, P.; Cathonnet, M.; Boettner, J. C. Int J Chem
Kinet 1992, 24, 813.
This research was performed in the Sonderforschungs-
bereich (SFB) 551 “Carbon from gas-phase: elementary
reactions, structures, materials,” which is funded by the
Deutsche Forschungsgemeinschaft (DFG). The Alexander
von Humboldt Foundation is acknowledged for providing
a research fellowship to K. Norinaga (KN). The authors
acknowledge the use of the computer program HOMREA
by Prof. J. Warnatz (University of Heidelberg). KN ac-
knowledges support from Kyoto University, ACCMS and
IIMC, Computing Service Group, Grant-in-Aid for Young
Scientists. We thank Mr. J. Aguilera and Prof. W. Klopper
25. Davis, S. G.; Law, C. K.; Wang, H. Combust Flame
1999, 119, 375.
26. Descamps, C.; Vignoles, G. L.; Feron, O.; Langlais, F.;
Lavenac, J. J Electrochem Soc 2001, 148, 695.
27. Matheu, D. M.; Dean, A. M.; Grenda, J. M.; Green, W.
H. J Phys Chem A 2003, 107, 8552.
28. Ziegler, I.; Fournet, R.; Marquaire, P. M. J Anal Appl
Pyro 2005, 73, 212.
29. Ziegler, I.; Fournet, R.; Marquaire, P. M. J Anal Appl
Pyro 2005, 73, 231.
30. Matheu, D. M.; Grenda, J. M. J Phys Chem A 2005, 109,
5332.
(University of Karslruhe) for providing kinetic data by DFT
calculations. We also thank Prof. K. J. H u¨ ttinger (Uni-
versity of Karslruhe) for many fruitful discussions and
31. Matheu, D. M.; Grenda, J. M. J Phys Chem A 2005, 109,
5343.
International Journal of Chemical Kinetics DOI 10.1002/kin