PERGAMON Carbon 39 (2001) 15–23

Formation of pyrolytic carbon during the pyrolysis of ethane at

high conversions
Greg F. Glasier, Philip D. Pacey*
Department of Chemistry, Dalhousie University, Halifax, Nova Scotia, B3 H4 J3, Canada
Received 15 August 1999; accepted 28 January 2000

Abstract

This study reports the amounts of gaseous products and pyrolytic carbon formed during the pyrolysis of ethane under
industrially significant conditions. Analytical techniques included gas chromatography, mass spectrometry, and gravimetric
analysis. Changes in the product composition were measured in a flow system as a function of residence time (0.3 to 21 s),
pressure (6.7 to 53 kPa) and temperature (1185 to 1100 K). When the initial ethane concentration was varied, the rate of
carbon deposition was found to be proportional to the local benzene concentration. The rate of benzene formation was
proportional to the concentration of acetylene cubed. The rate of carbon deposition, at 40 kPa, reached a maximum after
ethane had reacted for |6 s, when the concentration of benzene produced was also at a maximum. The changes in product
composition with residence time were consistent with benzene, or an intermediate which has a concentration proportional to
benzene, reacting in a rate limiting step on the complex pathway to pyrolytic carbon formation.  2000 Elsevier Science
Ltd. All rights reserved.

Keywords: A. Pyrolytic carbon; Carbon precursor; B. Pyrolysis; D. Coke yield

1. Introduction Most of the early studies attempted to correlate the

In the industrial steam cracking of ethane to produce
ethylene, coke deposits on the walls of the reaction tubes,
limiting heat transfer and ultimately blocking flow [1,2]. It
is estimated that more than 25 billion kg of ethylene are
produced annually in the United States [1]. The early
stages of the pyrolysis of ethane and its principal product,
ethylene, have been investigated thoroughly [3,4]. It is
important to investigate and understand the formation of
carbon in order to optimize the industrial process.
The deposition of pyrolytic carbon ( pyc) during the
thermal decomposition of hydrocarbons at relatively low
temperatures (1000–1500 K) has been studied extensively
for several decades [5–17] and has been the subject of
renewed interest over the last few years [18–29]. The term
pyrolytic carbon ( pyc) refers to the collection of carbon-
rich species which are deposited on a substrate during the
pyrolysis of organic compounds. The mechanisms leading
to pyc are still the subject of sharp controversy.
*Corresponding author. Tel.: 11-902-494-3305; fax: 11-902-
494-1310.
E-mail address: philip.pacey@dal.ca (P.D. Pacey).
pyrolysis conditions and starting material with the micro-
structure (isotropic, rough laminar, smooth laminar) of the
solid carbon product [7] or attempted to correlate the initial
hydrocarbon partial pressure directly to the deposition rate
[8]. If the pyc was deposited at a certain rate, the resulting
microstructure was apparently independent of the hydro-
carbon starting material, indicating there may be common
intermediates in the complex processes leading to pyc
formation [6]. The structure of the reactant does affect the
extent of conversion [17,21,25]. In one early study, Pierson
and Leiberman correlated pyrolysis products with the
structure of the pyc deposited, and obtained an empirical
correlation between the structure of the solid carbon
product and the acetylene / benzene ratio (a / b) in the
hydrocarbon product [9]. Furthermore, Pierson and Leiber-
man developed a thermodynamic model empirically relat-
ing a / b to the structure [9]. Unfortunately this result was
virtually ignored until recently when Benzinger et al. [22]
suggested a rationale for the a / b correlation and Lewis et
al. [23] applied a similar model to describe a chemical
vapor infiltration system. Lewis et al. [23] and Benzinger
et al. [22] have argued the assumption of thermodynamic
equilibrium may not be warranted under these conditions.

0008-6223 / 01 / $ – see front matter  2000 Elsevier Science Ltd. All rights reserved.
PII: S0008-6223( 00 )00079-8
16 G.F. Glasier, P.D. Pacey / Carbon 39 (2001) 15 – 23
This indicates that a consideration of the kinetics is of
particular importance.
Early observations of the formation of higher molecular
weight polycyclic aromatic hydrocarbons (PAHs) before or
during the deposition of pyc [5] and electron microscopy
of the pyc surface [12] led to the suggestion that PAHs
build in size until their vapor pressure becomes high
enough that, even at pyrolysis temperatures, they con-
dense, forming liquid droplets. These may carbonize in the
gas phase or deposit on surfaces within the reactor. Lahaye
et al. performed a substantial amount of work on the
formation of particulate carbon and indicated that the
growth of PAHs are important in the formation of carbon
black [11] but surface decomposition reactions may be
more important in the formation of pyc [12]. Chen and
Back [13] studied pyc deposition during the pyrolysis of
methane and provided some empirical evidence that PAHs
may grow to the point of condensation. Lucas and Mar-
chand [18] identified and estimated the relative amounts of
PAHs up to coronene and indicated higher molecular
weight products were present but not observable by the
analytical technique used. A number of researchers have
observed PAHs in flames and suggested they are important
in particulate carbon formation [30–33]. Murphy et al.
reported that the composition of PAHs in the pyrolysis
product mixture is independent of the hydrocarbon reactant
[20,25] but suggested that PAHs are a parallel product
resulting from the adsorption and polymerization of
acetylene molecules on the carbon surface, followed by
cyclization and desorption to form the large aromatics
found in the gas phase.
Marek and Albright showed that metal surfaces enhance
coke formation [14] but deposition is also significant on
inert surfaces.
Tesner made more strenuous efforts than any other
researcher to measure the kinetics of pyc deposition and
particle growth [16]. He performed many pyrolysis experi-
ments involving the flow of gases through a heated
cylindrical reactor packed with quartz fibers. Commonly,
he worked with temperatures of 1000–1500 K, low
conversion of reactant to products, high surface to volume
ratios, and conditions of turbulent flow. He and his group
studied pyc formation from methane, ethane [8], ethylene,
acetylene, propylene, butadiene, benzene, and many other
hydrocarbons [16] but did not usually consider the com-
position of the gas phase during deposition. Generally he
stated that the rates of pyc formation were directly
proportional to the initial concentrations of the hydro-
carbons pyrolysed. From the assumption that homogeneous
gas phase reactions were not significant, a model based on
a Langmuir adsorption isotherm with competitive inhibi-
tion by hydrogen was applied. This model has been
modified over the years but the same assumption of first
order kinetics has still been made [16].
Recent work by Benzinger, Becker and Huttinger
[22,26,27], under conditions of low surface area and
laminar flow, indicated that the rate of pyc deposition is
not directly proportional to the initial concentration of the
reactants tested. Surface decomposition of products of
hydrocarbon pyrolysis was thought to be important. They
developed and tested a model including the decomposition
of pyrolysis products as well as the initial hydrocarbon
reactant. This description considered the complex gas
phase chemistry [22]. A model was used to quantitatively
describe the deposition rate, R c . In this model a seven
parameter equation was fit to the two observables, R c and
residence time [26]. Their work indicated that direct
decomposition of the original hydrocarbon is likely im-
portant at low conversion of reactant to product but, as
more products are formed, R c is a complex function of the
concentrations of many products. Benzinger and Huttinger
[27] and others [28] have applied direct decomposition
models in order to rationalize the observation of changes in
the carbon microstructure with the gas phase product
composition.
Many researchers have considered deposition from the
pyrolysis of a wide variety of compounds. In many studies
particular aspects of a reaction system have been examined
and each investigator has proposed a mechanism to
describe their individual results. The mechanisms for
deposition near industrial steam cracking conditions fall
into two categories: hydrocarbon adsorption or formation
of liquid products. The purpose of this study will be
twofold, firstly to measure the deposition rate as a function
of residence time and initial ethane concentration and
secondly to measure the concentration of gaseous species
present at the same residence time and extent of conver-
sion. In this study well-established techniques are utilized,
including the measurement of the deposition rate using a
micro-balance and the identification and quantification of
gas phase products using chromatography and mass spec-
trometry [5,13,18,25,28,29]. A consideration of the rate of
pyc deposition with respect to the gas composition serves
to extend the conclusions of research on other hydro-
carbons to the pyrolysis of ethane and to provide new
insight into the problem.
2. Experimental
2.1. Apparatus
Ethane (Matheson, 99.95% pure) was passed at a
constant flow rate through a heated, cylindrical, quartz
reactor (1.01 cm inside diameter) shown in Fig. 1. A quartz
substrate (5.0 cm long and 0.31 cm diameter), was
suspended by a quartz thread from the micro-balance
(Cahn, model RG2000, 0.1–1000 mg) encased in a vac-
uum tight casing. The reactor was heated to 1185 K by
connecting a vertical Nichrome resistive furnace to a
power supply and a temperature controller (Omega, model
CN7000). Temperature profiles were obtained by measur-
 G.F. Glasier, P.D. Pacey / Carbon 39 (2001) 15 – 23
Fig. 1. The reactor and associated apparatus (not to scale) and the temperature profiles for both furnaces used in this work.
ing (Pt / Pt13%Rh thermocouple) the temperature at 1 cm
intervals down the length of the reactor, as illustrated in
Fig. 1. A flow line connected the gas supply system to the
reactor. Reagent flow rates were measured and controlled,
to within 0.1 standard cubic centimeters per min, with
mass flow controllers (MKS, 1159b). A vacuum system
with an oil diffusion pump and a rotary pump (Welch,
model 1405) was connected to the system for evacuation
purposes. Gases from the reactor exited through an o-ring
seal, into a trap which was used for collecting liquid
products. Flow exited from the trap, through another o-ring
seal, into a manifold which contained a valve for sampling
of product gases and a pressure transducer (MKS, model
122A, 0–1000 Torr) which was connected to a pressure
controller (Omega, model CN-2000). The flow exited into
a rotary pump (Welch, model 1397).
2.2. Sampling and analysis
Pyrolytic carbon deposition rates were measured by
following the increase in real time of the mass of the
quartz substrate suspended in the middle of the hot zone of
the reactor. Typical rates measured were of the order of 0.1
to 3310 26 kg / m 2 s. Due to background noise, results
were found to vary by |0.05–0.1310 26 kg / m 2 s. The
total rate of carbon deposition in the reactor was obtained
17
by multiplying the rate per unit area on the substrate by the
total surface area of quartz in the hot zone of the reactor.
Liquid samples were collected at the reactor exit using
the trap illustrated in Fig. 1. The neck of the trap extended
into the reactor exit and allowed for the capture of all tars
exiting the reactor. The main body of the trap allowed for
maximum surface area with minimum weight. At short
residence times the trap was kept at room temperature so
that benzene could be analyzed in the gas phase. It was
suspected that, as the partial pressure of benzene increased
with residence time, a portion of the benzene product may
have been dissolved in the higher molecular weight liquid
products collected at the exit of the reactor. Therefore, for
residence times greater than 3 s, the concentration of
benzene was obtained from HPLC analysis of the liquid
products. To condense benzene as a liquid, the trap was
kept at approximately 2608C using a dry ice and chloro-
form bath. Samples were collected under stable conditions
for 60 min. The trap was then removed from the apparatus
and weighed on an analytical balance (Mettler, AJ100).
The increase in mass was divided by the collection time to
obtain the total rate of formation of liquid products. The
liquid samples were analyzed by reversed phase high
performance liquid chromatography (HPLC) (Hewlett
Packard, HP 1100 series) with an ultraviolet detector used
to quantify the benzene content. The average molecular
18 G.F. Glasier, P.D. Pacey / Carbon 39 (2001) 15 – 23
weight of the liquid products was assumed to be 200 g
mol 21 , in order to convert the wt.% to mol.%.
Samples of gaseous products were extracted from the
system by means of a gas sampling valve (Varian, 57-
000034-00) and injected into the gas chromatograph
(Tracor, model 550). All gaseous products, excluding
acetylene, were separated on a 1.2-m Poropac (type Q)
column with a temperature ramp (25–2008C in 10 min and
held for the duration of the run, 30 min). Acetylene was
separated on a 1-m (10% NaOH on alumina) column. A
flame ionization detector (Varian) was used for analysis of
all compounds except hydrogen, for which a thermal
conductivity detector (Gow-Mac, 10952-6) was used.
Product identification was achieved by comparison of
sample retention times to those of known standards and
confirmed by mass spectrometry (VG Quattro).
The total rate of formation of gaseous products was
obtained as the difference between the rate of flow of
ethane into the reactor, and the sum of the rates of
formation of carbon and liquids. The measurement of
deposition rate, the sampling of liquid products for off-line
analysis and the on-line analysis of gas phase products
were each performed in separate runs.
2.3. Calculation of residence times
The residence time, t r , was calculated from Eq. (1)
t r 5 PAl /(RT(Dn /Dt)). (1)
Here P, A and T are, respectively, the pressure, cross-
sectional area, and average temperature in the reactor. R is
the gas constant, and (Dn /Dt) is the average molar flow
rate in the reactor. In considering residence times for
carbon deposition, only the distance, l, from the top of the
hot zone to the mid-point of the substrate was included. To
find residence times for gaseous and liquid products, the
total length of the hot zone was included in Eq. (1). In
order to compare product yields at a particular residence
time, the flow rate for the measurement of deposition onto
the quartz substrate was set at one half the flow rate which
was used for the gas and liquid analyses. The deposition
rate and gas composition were then compared at the same
residence time. The length of the hot zone and the average
reaction temperature were calculated [34] from the mea-
sured temperature profiles. This calculation took into
account the fact that the reaction was slower on the
shoulders of the temperature profile at the ends of the
reactor than it was at the peak temperature in the center.
Temperatures were weighted by an Arrhenius exponen-
tial factor to find an average reaction temperature. In a
typical experiment 85% of the reaction was found to have
occurred within 10 K of the average reaction temperature.
Changing from an activation energy of 25 000 K to 30 000
K, only changed the calculated average reaction tempera-
ture by 1 K and the effective reactor length by 4%.
The dominant reaction in ethane pyrolysis is [4]
C2H6 → C2H4 1 H2 (2)
The resulting doubling of the number of moles leads to an
acceleration of flow, resulting in shorter residence times.
The residence times for the pyrolysis of pure ethane have
been calculated assuming (Dn /Dt) in Eq. (1) is twice the
flow rate of ethane entering the reactor. This assumption
will be tested in Section 3. In order to establish the effect
of the concentration of ethane several experiments were
performed by diluting the ethane with an inert gas (argon).
The partial pressure of ethane was varied while holding the
total reactor pressure constant. In this case the residence
time was calculated by assuming that (Dn /Dt) in Eq. (1) is
the sum of the argon flow rate entering the reactor and
twice the flow rate of ethane entering the reactor.
2.4. Reaction, transport and flow
The goal of this work was to consider the pyrolysis of
ethane under industrially relevant conditions. These con-
ditions are rather extreme and as a consequence some of
the usual assumptions are not valid. The following discus-
sion deals with some of the potential errors in the flow
method.
In the use of a flow reactor, it is assumed that the
entering gas molecules warm to reaction temperature
instantaneously at the inlet of the reactor. However, when
a gas is passed through the reactor at a fast flow rate, the
gas molecules will travel a certain distance down the
reactor before attaining the reactor temperature. This delay
in warming up the reagent results from incomplete radial
heat transfer. Furue and Pacey [35] found that the fraction-
2
al experimental error is approximately equal to r /k t r ,
where r is the radius of the reactor and k is the coefficient
of thermal diffusivity. Under the conditions of these
experiments, the error, for the shortest residence times
considered, has been calculated to be less than 10% at
reactor pressures below 40 kPa and no greater than 20% at
all pressures.
High conversion results in concentration gradients down
the axis of the reactor. Reactant molecules move faster
toward the exit of the reactor due to diffusion. Product
molecules diffuse in the opposite direction. Furue and
Pacey [35] reported that an upper limit to the fractional
error in residence time caused by mass transfer is equal to
2 2 2
r / 48D 1 Dt r /l , where D is the diffusion coefficient. The
error due to mass transfer was calculated to be less than
10% at residence times less than 10 s and no greater than
40% at longer reaction times.
3. Results and discussion
3.1. Products as a function of residence time
A typical product analysis is provided in Table 1. This
 G.F. Glasier, P.D. Pacey / Carbon 39 (2001) 15 – 23 19

Table 1
Product analysis for an ethane flow rate of 2.01310 26 kg s 21 at a
temperature of 1185 K and a pressure of 39.9 kPa
Compound Wt.% Mol.%
Hydrogen 5.3 43.8
Methane 15.3 15.8
Acetylene 6.6 4.2
Ethylene 53 31.3
Ethane 2.8 1.5
Propylene 1.1 0.4
Butadiene 1.9 0.6
Benzene 9.1 1.9
Liquid products 4.9 0.40
Pyrolytic carbon 0.05 0.07

Fig. 3. The mole percent composition of the major gaseous

analysis gives an indication of the relative proportions of
95 wt.% of the individual products. High selectivities for
hydrogen, ethylene and methane were observed, with
relatively large amounts of acetylene and benzene also
present.
In order to establish trends in the pyc deposition rate,
experiments were run at a variety of residence times. From
Fig. 2 it is apparent that the deposition rate at 39.9 kPa
reaches a maximum at |6 s. At lower pressures it appears
the rate of deposition would also reach a maximum, but at
longer residence times. The results agree with the work of
Lahaye and Prado on benzene [10] and Feron et al. on
methane and carbon tetrachloride mixtures [29] where a
maximum deposition rate was observed. Fig. 2 also
provides the yield of solids deposited within the reactor
(calculated by integrating the deposition rate over the
residence time). The maximum rate at 6 s is matched by a
slight maximum in the slope of the yield.
The main gaseous products were hydrogen, ethylene,
products, hydrogen (s), ethylene (h), and methane (,), and of
the reactant, ethane (d) as a function of residence time, t r , of
gases in the hot zone of the reactor at a pressure of 39.9 kPa.
and methane. Analysis of these products as a function of
residence time, shown in Fig. 3, revealed high conversions
of ethane to products (.97 mol.%) with high selectivities
for hydrogen and ethylene at shorter residence times.
The minor products, shown in Fig. 4, included
acetylene, propylene, butadiene and benzene. Trace
amounts of other species such as vinylacetylene, toluene
and styrene were also found in the gas phase. At relatively
short residence times and low pyc yields acetylene and
benzene were present in appreciable quantities, whereas
propylene and butadiene did not appear to be important
products at any residence time. Acetylene reached a
maximum at about 1 s and declined. Benzene had a
maximum at about the same time as the deposition rate.

Fig. 4. The concentration of the minor products, benzene (1),

Fig. 2. The rate, R c , of deposition of solids onto a quartz
substrate, as a function of the residence time, t r , of gases from the
top of the hot zone to the substrate at 39.9 kPa (d), 26.6 kPa (j),
and 13.3 kPa (h), and at a constant temperature of 1185 K. The
cumulative yield, Yc , of carbon deposited at a pressure of 39.9 kPa
(s) is shown scaled against the right axis. The feedstock was pure
ethane.
acetylene (s), butadiene (\), and propylene (n), found in the
product mixture as a function of residence time in the hot zone of
the reactor at 39.9 kPa. The rate, R c , of deposition of solids onto
the quartz surface per unit volume of the reactor, (h), is shown
scaled against the right axis. The deposition rate and gas com-
position were measured at the same residence times by using
faster flow rates for the gas analysis, as explained in Section 2.3.
20 G.F. Glasier, P.D. Pacey / Carbon 39 (2001) 15 – 23
Acetylene and benzene have been suggested as im-
portant precursors of pyc formation in other systems [26].
It is expected that the concentration of a rate limiting
species would have reached a maximum at about the same
time as the deposition rate; this was observed for benzene.
From these results it is possible that acetylene may have
formed pyc directly but it appears more likely acetylene
reacted to form benzene and reaction of benzene or some
intermediate with a concentration proportional to the
benzene concentration was the slow step in forming pyc.
This is not to say that parallel mechanisms involving direct
decomposition of hydrocarbons are not prevalent under
different conditions.
Fig. 5 provides a plot of the experimentally determined
ratio of the moles of product exiting the reactor over the
moles of ethane entering the reactor as a function of the
mass flow rate. From the data it is apparent that the ratio
does not deviate significantly from 2, indicating the flow
rate is approximately doubled. Thus the assumption made
in Section 2.3, that (Dn /Dt) is approximately twice the
entering flow rate of ethane, is justified.
3.2. Rates as a function of concentration
If a reactant, intermediate or product is involved in a
rate limiting step, then it is expected that the rate of pyc
deposition (R c ) should be proportional to the precursor
concentration. For example, Tesner [16] proposed R c to be
directly proportional to the concentration of hydrocarbon
pyrolysed, as indicated in Eq. (3)
R c 5 2 k e [C 2 H 6 ]. (3)
In order to establish the pressure and concentration depen-
dence of the deposition rate, experiments were performed
at several residence times. Results for the reactant, ethane,
are provided in Fig. 6. The data indicate that the deposition
Fig. 5. The ratio of the total number of moles of product
calculated to be exiting the reactor over the total moles of ethane
entering the reactor as a function of mass flow rate at a constant
temperature of 1185 K and a constant pressure of 39.9 kPa.
Fig. 6. The deposition rate as a function of the initial ethane
concentration, at residence times of 3.7 s (.), 1.3 s (s) and 0.7 s
(d).
rate increased nonlinearly with increasing initial concen-
tration of ethane in the reactor. Comparing Figs. 2 and 3,
we see that R c increased between 0 and 6 s, while the
concentration of ethane decreased. These observations
support work on other hydrocarbons by Huttinger’s group
[26]. It can be concluded that the simplified assumption of
a first order reaction of ethane is not warranted for this
system.
The dependence of the deposition rate on the con-
centration of benzene was also examined. Experiments
were performed by varying the flow rate of ethane and
argon in the inlet of the reactor while holding the total
reactor pressure constant at 53.3 kPa. Based on results of
the pyrolysis of benzene it has been suggested that rate of
pyc deposition is directly proportional to the initial partial
pressure of benzene [26]. A similar relationship held for
the present system. A plot of R c as a function of the
concentration of benzene was nearly linear and passed
through the origin, as indicated in Fig. 7. R c was de-
termined when the gases had spent 1.1 s proceeding from
the start of the hot zone to the substrate. In a separate run
the benzene concentration was determined when the gases
exited after spending 1.1 s in the hot zone. A plot of the
logarithm of R c against the logarithm of [C 6 H 6 ] had a
slope of 1.160.1.
In Fig. 8 a logarithmic plot of the average rate of
formation of benzene, ([C 6 H 6 ] /t r ), as a function of
acetylene partial pressure reveals a linear relationship with
a slope of 3.160.2. This observation may indicate that
three acetylene molecules, or species with concentrations
proportional to acetylene, reacted to form benzene. Some
of the relevant literature is consistent with this idea.
Westmoreland et al. [37] provided a brief review of some
of the routes that have been suggested by several re-
searchers for the formation of benzene during pyrolysis
and combustion. Based on experimental results from
flames and computational results, Westmoreland et al. [37]
showed the rate of benzene formation can be reasonably
 G.F. Glasier, P.D. Pacey / Carbon 39 (2001) 15 – 23
Fig. 7. The rate of pyc formation (R c ) as a function of benzene
concentration at a constant total pressure of 53 kPa. Circles were
experimentally determined by pyrolysing mixtures of ethane and
argon. The gases had spent 1.1 s between the top of the hot zone
and the substrate, for measurements of R c , and between the top
and bottom of the hot zone, for measurements of [C 6 H 6 ]. The line
represents a first order relationship.
accounted for by acetylene addition to form a C4 radical
which may add to another molecule of acetylene to form
benzene. Frenklach et al. [38] support the addition of three
acetylene molecules over several steps to form a phenyl
radical. Becker and Huttinger [26] have said benzene is a
direct product of acetylene pyrolysis and that it is formed
in a third order reaction. Dimitrijevic et al. [36], based on
experimental results, have shown that the reaction is
approximately third order with respect to acetylene con-
centration. This process likely involves a complex series of
reactions of radicals. If a step in which all three acetylene
molecules have combined is rate limiting, then an empiri-
cal third order relationship is expected.
Fig. 8. Double logarithmic plot of the average rate, R [C 6 H 6 ] , of
benzene formation as a function of acetylene concentration for
experiments at a constant total pressure of 53 kPa and a residence
time of 1.1 s.
21
Under conditions of high surface area, fast residence
time, and low conversion to products it has been suggested
that the formation of surface carbon is due to the direct
decomposition of hydrocarbons on the carbon surface [16].
A large amount of quantitative data based on a first order
relationship has been tabulated by Tesner [16]. Although
the mechanistic interpretation of Tesner’s work has been
questioned [22], a consideration of the rate constants in an
empirical sense may be of some use here. From Tesner’s
work it is found that rate constants for reactions of many
aliphatic hydrocarbons are several orders of magnitude
lower than those observed for acetylene. If one supposes
that many product species are responsible for pyc forma-
tion, as has been suggested by Tesner [16], Lahaye et al.
[12], and Huttinger and co-workers [22,26], the rate of
deposition may depend on many parallel reactions as
indicated in Eq. (4)
R c 5 2 k et [C 2 H 6 ] 1 2 k ey [C 2 H 4 ] 1 2 k a [C 2 H 2 ]
1 3 k pr [C 3 H 6 ] 1 4 k bu [C 4 H 6 ] 1 6 k ab [C 6 H 6 ]. (4)
From Tesner’s results [16], k et , k ey , k pr , k bu < k a and it is
expected that R c is essentially described by Eq. (5)
R c (2 k a [C 2 H 2 ] 1 6 k ab [C 6 H 6 ]. (5)
Here the rate constants, k a and k ab , refer to the empirical
reactions (6) and (7), respectively.
C 2 H 2 → 2C 1 H 2 (6)
C 6 H 6 → 6C 1 3H 2 (7)
The values of k a and k ab , calculated from a nonlinear least
squares fit of Eq. (5) to experimental data at a 1.1 s
residence time and several starting ethane concentrations,
are (2670)310 24 and (3.560.4)310 23 s 21 , respectively.
The standard deviation of k a is more than an order of
magnitude greater than its value, indicating that the direct
contribution of acetylene to carbon formation cannot be
reliably determined from this data. Dropping this term, we
obtain
R c (6 k ab [C 6 H 6 ]. (8)
Fitting Eq. (8) to the data we also obtain a value of
(3.560.3)310 23 s 21 for k ab . The consistency of the
values of k ab calculated from Eqs. (5) and (8) agrees with
the earlier suggestion that benzene is the major precursor
of pyc in this system. Moreover, the fact, established in
Fig. 4, that the deposition rate parallels the concentration
of benzene indicates this relationship can be extended to
longer residence times. This work does not provide
evidence for the hypothesized parallel route suggested by
Benzinger, Becker and Huttinger [22,26] but does not
negate the possibility of a slower route involving the
reaction of acetylene.
These results do not prove that benzene lies on the
22 G.F. Glasier, P.D. Pacey / Carbon 39 (2001) 15 – 23

pathway from ethane to carbon, nor that the reaction of Since ethane contains a weaker carbon–carbon bond than
benzene is the rate-limiting step on this pathway, even benzene the radical concentration is expected to be larger
though the results are consistent with both of these in the pyrolysis of ethane. It is also expected that if large
hypotheses. It is possible that there are other species, PAHs are formed from acetylene then the concentration of
whose dependencies on time and pressure are similar to higher molecular weight aromatics may be higher in the
those of benzene, which would also be reasonable pre- ethane system. Further, Tesner [16] often worked with
cursors for carbon. One possibility is the phenyl radical reactors of higher surface to volume ratio than those used
[38]. Its time dependence is expected to be similar to that in this work and these results may indicate that the rate is
of benzene, although its concentration might also depend geometry dependent. From these results the origin of the
on the concentrations of other radicals present. discrepancy between k ab for the pyrolysis of pure benzene
Heavier aromatic compounds and radicals might also and that calculated from the pyrolysis of ethane cannot be
depend on time and pressure in a way similar to benzene. established.
It is not clear whether heavier aromatics can be formed
directly from acetylene and small radicals or whether they
are formed from benzene.
4. Conclusions
3.3. Rates as a function of temperature
This work has provided an analysis of the main products
of ethane pyrolysis coinciding with the surface deposition
The deposition rate and benzene concentration were
of pyrolytic carbon. A consideration of the product com-
examined as a function of temperature from 1100 K to
position revealed that the acetylene concentration reached
1185 K. An empirical Arrhenius analysis was used to
a maximum early. The benzene concentration reached a
investigate the rate of deposition (mol l 21 s 21 ) and rate
maximum at about the same time as the rate of formation
constants (s 21 ) calculated from Eq. (8). A plot of the
of solids on the reactor surface. The rate of benzene
natural logarithm of the deposition rate as a function of
formation was found to be third order with respect to
inverse temperature yielded an activation energy of
acetylene concentration. This implies that, as has been
412615 kJ mol 21 . The natural logarithm of the pre-
observed for the pyrolysis of pure acetylene [36], three
exponential factor in mol l 21 s 21 was 2962. A plot of the
acetylene molecules may react to form benzene. The rate
natural logarithm of rate constants, k ab (s 21 ), as a function
of carbon deposition was found to be directly proportional
of inverse temperature yielded an activation energy of
to the concentration of benzene in the system. This is
268624 kJ mol 21 . The natural logarithm of the pre-
consistent with a reaction of benzene or some other
exponential factor was 2262.
intermediate which has a concentration proportional to that
From Ref. [16], for the pyrolysis of pure benzene at
of benzene representing the bottleneck on the route to pyc
temperatures of 1023 to 1143 K, the Arrhenius parameters
deposition. A comparison of this system to the pyrolysis of
were 230 kJ mol 21 for the activation energy and 25.8 for
pure benzene indicates in both systems a first order
the natural logarithm of the pre-exponential factor given in
dependence is observed but in this system the rate con-
units of g cm 22 s 21 Pa 21 . Converting our results to the
stant, k ab , is larger. This is consistent with some other
same units, values of 259623 kJ mol 21 for the activation
intermediate which has a concentration proportional to
energy and 362 for the natural logarithm of the pre-
benzene representing the bottleneck.
exponential factor are obtained, giving a value of 2.723
Whether the benzene decomposes directly on the pyc
10 211 g cm 22 s 21 Pa 21 for k ab at 1140 K. From Tesner’s
surface as suggested by Tesner [16], Murphy and Carroll
results the value of k ab , for the pyrolysis of pure benzene at
[25] and Becker and Huttinger [26], or goes on to form
1140 K, can be calculated to give a value of 8.75310 214 g
large PAHs, which dehydrogenate and deposit on surfaces
cm 22 s 21 Pa 21 . In Ref. [26] the experimental temperature
as suggested by Lahaye and Prado [11], Chen and Back
for the pyrolysis of pure benzene was 1273 K, well above
[13], Lucas and Marchand [18], and Pierson and Lieber-
the present range. The value of k ab calculated from our
man [9] or whether the benzene concentration is simply
results is 310 times faster than that calculated from
proportional to that of some other intermediate which has
Tesner’s results [16].
not yet been observed, cannot be differentiated from these
The fact that k ab is larger for the pyrolysis of ethane
results. Interpretation of the mechanism will be examined
than for the pyrolysis of pure benzene is consistent with
in further detail in future publications.
the assumption that it is not the direct decomposition of
benzene but some intermediate with a concentration pro-
portional to benzene, such as a larger aromatic species or
radical, which represents the bottleneck to carbon forma- Acknowledgements
tion. There may be proportionally more of this inter-
mediate, relative to the benzene concentration, present in The authors wish to thank Imperial Oil Limited for the
the ethane system than for the pyrolysis of pure benzene. award of a University Research Grant, D.R. Slim for
 G.F. Glasier, P.D. Pacey / Carbon 39 (2001) 15 – 23 23

valuable discussion, S. Dimitrijevic for experimental assis- vol. 202, Washington, DC: American Chemical Society,
tance and O. Knop for the loan of equipment. 1982, p. 123.
[15] Tesner PA, Borodina LM. Khimya Tverdogo Topliva
1983;17:157.
[16] Tesner PA. In: Walker Jr. PL, Thrower PA, editors, Chemis-
References
try and physics of carbon, vol. 19, New York: Marcel
Dekker, 1984, p. 65.
[1] Bodke AS, Olschki DA, Schmidt LD, Ranzi R. Science [17] Kaae JL. Carbon 1985;23(6):665.
1999;285:712. [18] Lucas P, Marchand A. Carbon 1990;28:207.
[2] Brown DE, Clark JTK, Foster AI, McCarrol JJ, Sims ML. In: [19] Blekkan EA, Myrstad R, Olsvik O, Rokstad OA. Carbon
Albright LF, Baker RTK, editors, Coke formation on metal 1992;30:665.
surfaces, ACS symposium series, vol. 202, Washington, DC: [20] Murphy DB, Carroll RW. Carbon 1992;30(1):47.
American Chemical Society, 1982, p. 23. [21] Dupel P, Bourrat X, Pailler R. Carbon 1995;33(9):1193.
[3] Williamson KD, Davis HG. In: Albright LF, Crynes BL, [22] Benzinger W, Becker A, Huttinger KJ. Carbon 1996;34:957.
editors, Industrial and laboratory pyrolysis, ACS symposium [23] Lewis JS, Lackey WJ, Vaidyaraman S. Carbon 1997;35:103.
series, vol. 32, Washington, DC: American Chemical Socie- [24] Shi R, Li HJ, Yang Z, Kang MK. Carbon 1997;35:1789.
ty, 1976, p. 123. [25] Murphy DB, Carroll RW, Klonowski JE. Carbon
[4] Pilling MJ, Smith IWM. In: Modern gas kinetics, theory, 1997;35:1819.
experiment and application, Palo Alto: Blackwell, 1987, p. [26] Becker A, Huttinger KJ. Carbon 1998;36:177–225.
316. [27] Benzinger W, Huttinger KJ. Carbon 1999;37:1311.
[5] Palmer HB, Cullis CF. In: Walker Jr. PL, editor, Chemistry [28] Feron O, Langlais F, Naslain R, Thebault J. Carbon
and physics of carbon, vol. 1, New York: Marcel Dekker, 1999;37:1343.
1965, p. 265. [29] Feron O, Langlais F, Naslain R. Carbon 1999;37:1355.
[6] Bard RJ, Baxman HR, Bertino JP, O’Rourke JA. Carbon [30] Bohm H, Jander H, Tanke D. In: 27th Symposium (Interna-
1968;6:603. tional) on Combustion, 1998, p. 1605.
[7] Bokros JC. In: Walker Jr. PL, editor, Chemistry and physics [31] Melton TR, Vincitone AM, Senkan SM. In: 27th Symposium
of carbon, vol. 5, New York: Marcel Dekker, 1969, p. 1. (International) on Combustion, 1998, p. 1631.
[8] Tesner PA, Denisevich EV. Doklady Akad Nauk SSSR [32] Yano H, Sawa Y. Combust Flame 1999;117:851.
1973;212(3):660. [33] LaFleur AL, Taghizadeh K, Howard JB, Anacleto JF,
[9] Pierson HO, Lieberman ML. Carbon 1975;13:159. Quilliam MA. J Am Soc Mass Spectrum 1996;7:276.
[10] Lahaye J, Prado G. In: Deviney ML, O’Grady TM, editors, [34] Pacey PD, Purnell JH. J Chem Soc Faraday Trans I
Petroleum derived carbons, ACS symposium series, vol. 21, 1972;68:1462.
Washington, DC: American Chemical Society, 1976, p. 335. [35] Pacey PD, Furue H. J Phys Chem 1980;84:3139.
[11] Lahaye J, Prado G. In: Walker Jr. PL, Thrower PA, editors, [36] Dimitrijevic ST, Paterson S, Pacey PD. J Anal Appl
Chemistry and physics of carbon, vol. 14, New York: Marcel Pyrolysis 1999;53:107.
Dekker, 1978, p. 167. [37] Westmoreland PR, Dean MA, Howard JB, Longwell P. J
[12] Lahaye J, Badie P, Ducret J. Carbon 1977;15:87–93. Phys Chem 1989;93:8171.
[13] Chen CJ, Back MH. Carbon 1979;17:175. [38] Frenklach M, Wang H. J Phys Chem 1994;98:11465.
[14] Marek JC, Albright LF. In: Albright LF, Baker RTK, editors,
Coke formation on metal surfaces, ACS symposium series,