2011 - coke characterization in propane dehydrogenation on pt-sn-al2o3
TRANSCRIPT
ORIGINAL PAPER
Coke Formation on Pt–Sn/Al2O3 Catalyst in PropaneDehydrogenation: Coke Characterization and Kinetic Study
Qing Li • Zhijun Sui • Xinggui Zhou •
Yian Zhu • Jinghong Zhou • De Chen
Published online: 28 July 2011
� Springer Science+Business Media, LLC 2011
Abstract The influences of gas compositions on the rates
of coke formation over a Pt–Sn/Al2O3 catalyst are studied.
The coke formed on the catalyst is characterized by ther-
mal gravimetric analysis, IR spectroscopy, Raman spec-
troscopy and elemental analysis. Two kinds of coke are
identified from the TPO profiles and assigned to the coke
on the metal and the coke on the support, respectively. The
coke formed on the metal is softer (containing more
hydrogen) than that formed on the support. The rate of
coke formation on the metal is weakly dependent on the
propylene and hydrogen pressures but increasing with
the propane pressure, while the rate of coke formation on
the support is increasing with the propane and propylene
pressures and decreasing with the hydrogen pressure.
Based on the kinetic analysis, a mechanism for the coke
formation on the Pt–Sn/Al2O3 catalyst is proposed, and the
dimerization of adsorbed C3H6 is identified to be the
kinetic relevant step for coke formation on the metal.
Keywords Coking mechanism � Kinetics � Propane
dehydrogenation � Pt–Sn/Al2O3 catalyst
1 Introduction
Recently the rapidly rising needs for propylene [1–3] have
driven searching for new propylene production techniques.
Propane dehydrogenation, developed commercially in
1980s, is an on-purpose technique for propylene production
and is now making big contributions to the world propyl-
ene supply in recent years [4].
Pt based catalyst is the most popular catalyst for propane
dehydrogenation and is used in commercial Oleflex pro-
cess. This catalyst, as well as other catalysts for alkane
dehydrogenation, is quickly deactivated, mainly by coke
formation, and the mechanism of coke formation on the Pt
catalyst is still not clear. Understanding the mechanism of
coke formation is of great significance. On one hand, based
on the mechanism of coke formation, one can optimize the
reaction conditions to reduce the frequency of dehydroge-
nation and regeneration cycle by controlling the rate of
coke formation so as to increase the propylene productiv-
ity. On the other hand, it can help the rational design of the
catalyst to reduce the rate of coke formation while main-
taining the activity and selectivity in dehydrogenation.
Coke formation on Pt or modified Pt catalysts during
propane dehydrogenation has been studied by a number of
researchers. Most of the studies were targeted to reduce
the coking rate by modifying Pt, for example, with Sn or
alkali metals, and/or using different catalyst supports.
Praserthdam et al. [5] showed that alkali metals (such as
Li, Na, and K) would help to reduce the coking rate by
providing excess mobile electrons to the Pt catalyst.
Another straightforward method to reduce the amount of
coke formed is to use supports with weak acidity, e.g.
SBA-15 [6], which is effected by weakening propylene
adsorption and suppressing propylene dehydrogenation or
polymerization.
Q. Li � Z. Sui � X. Zhou (&) � Y. Zhu � J. Zhou
State Key Laboratory of Chemical Engineering, East China
University of Science and Technology, Shanghai 200237, China
e-mail: [email protected]
D. Chen
Department of Chemical Engineering, Norwegian University
of Science and Technology (NTNU), N-7491, Trondheim,
Norway
123
Top Catal (2011) 54:888–896
DOI 10.1007/s11244-011-9708-8
The rate of coke formation depends highly on the
operating conditions. Larsson et al. [7] studied the coke
formation on Pt/Al2O3 and Pt–Sn/Al2O3 catalysts, and
suggested that only a small part of the formed coke was
responsible for catalyst deactivation, and that a major part
of the coke was formed regardless of the gas composition
but dependent on the temperature. They also concluded
that hydrogen could reduce the rate of coke formation and
correspondingly the rate of catalyst deactivation by sup-
pressing coke precursor formation, but hydrogen could not
remove the coke that had already been formed on the
catalyst. Rebo et al. [8] studied the coke formation on Pt–
Sn/Al2O3 catalyst using an oscillating microbalance reactor
and concluded that coke formation was a structure sensitive
reaction and hydrogen could decrease the rate of coke
formation as well as the deactivating effect of the coke
formed.
Coke could be formed either on the support or on the
metal, and the coke deposited on different sites is supposed
to have different effects on catalyst deactivation. Studies
on the influences of reaction conditions on the rate of coke
formation and the nature of the coke deposited on different
sites are essential for the understanding of the mechanism
of coke formation as well as the mechanism of catalyst
deactivation. However, studies on the influences of reac-
tion conditions, especially gas compositions, during pro-
pane dehydrogenation are seldom reported.
In this work, we study the influences of gas composi-
tions on coke formation on the metal and support of a Pt–
Sn/Al2O3 catalyst during propane dehydrogenation. Based
on these results, the kinetic relevant step in coke formation
on the metal is identified and a mechanism of coke for-
mation is then proposed.
2 Experimental
2.1 Catalyst Preparation and Characterization
The Pt–Sn/Al2O3 catalyst used for this study was prepared
by incipient impregnation of alumina (Pural 200) with
H2PtCl6 and SnCl4 solutions, followed by calcination at
530 �C and treatment in steam to remove chlorine. A Pt/
MgO catalyst was also prepared by impregnation method.
The catalysts were characterized by N2 physisorption on
ASAP 2010 (Micromeritics, USA) at -196 �C after out-
gassing the samples for at least 5 h at 190 �C and 1 mm Hg
vacuum. The specific surface areas were calculated with
BET equation, and the pore volumes and pore size distri-
butions were determined from the N2 desorption isotherms
by using the BJH method. Table 1 summarizes the char-
acteristics of the catalysts.
Besides, alumina (Pural 200), which was the support for
the Pt–Sn catalyst, was also used as a blank catalyst for
comparison.
2.2 Coking Experiments
The catalysts were evaluated in a tubular stainless steel
reactor with an inner diameter of 6 mm, the temperature of
which was maintained by an electrical furnace jacketing
outside of the reactor. Inserted inside the catalyst packing
(100 mg) was a thermocouple to indicate the real temper-
ature of reaction. Silica particles as inert packing were used
for uniform flow distribution. An online gas chromatogra-
phy (Agilent 4890D) was used to measure the outlet gas
concentrations. The catalyst (or the alumina), which had
particle sizes between 0.1 and 0.15 mm, was firstly reduced
at 500 �C in flowing hydrogen (10 ml/min) for 100 min
and the temperature was then ramped to 575 �C in argon
(40 ml/min). The feed gas was then introduced into the
reactor for coking experiment. In all the experiments, argon
was used as a balance gas and the total flow rate was
maintained at 80 ml/min. After 80 min of reaction, the
reactor was switched to pure argon flow and then cooled
down to room temperature. Table 2 shows the feed com-
positions and averaged gas compositions in the reactor
during the experiments.
2.3 Coke Characterization
The amounts of coke formed on the spent catalysts (or
coked alumina) were determined by TG (thermal gravi-
metric) analysis (SDT Q600, TA Company, USA) in air.
During the TG analysis, the temperature was increased
from ambient temperature to 600 �C (sometimes higher) at
a rate of 10 �C/min, during which the temperature pro-
grammed oxidation (TPO) profiles and heat flow curves
were recorded. Elemental analysis was carried out on a
Vario EL III Elementar analyzer. The coked samples were
firstly dissolved in a hydrofluoric acid solution (40%) at
room temperature to liberate the coke from the support.
Then the coke was dried at 50 �C and collected for testing.
Table 1 Catalyst characteristics
Pt–Sn/Al2O3 Pt/MgO
Shape Pellet Pellet
Diameter (mm) 1.0 0.13
Surface area (m2/g) 56.6 47.7
Pore volume (m3/g) 0.25 0.24
Average pore diameter (nm) 18.5 19.2
Pt (%) 0.50 0.75
Sn (%) 1.50 –
Top Catal (2011) 54:888–896 889
123
FTIR analysis was carried out on a Bruker Equinox-55
with a resolution of 4.0 cm-1 and the scanned wave
number was ranged from 4000 to 400 cm-1. Raman
analysis was performed at room temperature under ambient
conditions on a Renishaw inVia ? Reflex Raman spec-
trometer with a 514.5 nm Ar-ion laser beam.
3 Results
3.1 Thermal Gravimetric Analysis
Figures 1–3 show the TPO profiles of the spent Pt–Sn/
Al2O3 catalysts coked with different gas compositions at
575 �C for 80 min. Two peaks, locating in the interval of
150–280 �C (Peak I) and in the interval of 380–430 �C
(Peak II), respectively, are present in each profile. How-
ever, for coked alumina (Sample S9) and Pt/MgO catalyst
(Sample S10), only one peak, at around 470 and 310 �C,
respectively, is present in the TPO profiles (not shown
here). To determine the amount of coke losses, the TPO
profiles are deconvoluted by multiple Gaussian functions
using a non-linear least-squared optimization procedure
based on Levenberg–Marquardt algorithm. This method
fits the TPO profiles quite well and the results are sum-
marized in Table 3.
Figure 1 shows that the area of Peak II increases while
that of Peak I is almost unchanged when the partial pres-
sure of propylene is increased (Samples S1–S3).
Consequently the total amount of coke increases with the
partial pressure of propylene. Figure 2 indicates that the
addition of hydrogen in the feed has greatly suppressed
coke formation: if there is only 1.9% hydrogen (Sample
S4) in the gas, the total amount of coke is as high as
3.80 wt%, while if the hydrogen content in the gas is
increased to 13.6 wt% (Sample S5), the total coke amount
is decreased to 1.44 wt%. It is also noted that for sample S4
or S5, the decrease in the total amount of coke, denoted by
the areas of Peak I and Peak II, is mainly due to the
decrease of the area of Peak II, as seen from Table 3.
Table 2 Summary of the experimental conditions (temperature,
575 �C)
Samples Feed
compositionsd (%)
Averaged gas
compositionse (%)
C3H8 C3H6 H2 C3H8 C3H6 H2
S1a 34.3 0.0 8.3 30.0 2.9 11.9
S2a 35.6 3.0 6.6 31.2 5.6 10.5
S3a 37.0 9.5 9.1 34.5 11.1 10.1
S4a 35.0 7.8 0.0 32.4 9.4 1.9
S5a 36.0 7.3 12.0 34.0 8.5 13.6
S6a 20.0 0.0 0.0 15.9 4.0 5.3
S7a 34.9 0.0 0.0 30.0 3.3 4.0
S8a 49.1 0.0 0.0 42.7 4.3 4.5
S9b 0.0 10.0 0.0 0.0 9.9 0.0
S10c 35.3 0.0 0.0 35.0 0.2 0.2
a Pt–Sn/Al2O3
b Al2O3
c Pt/MgOd Gas compositions at the inlet of the reactore Average compositions of gas between the inlet and outlet of the
reactor
100 200 300 400 500 600
0.000
0.005
0.010
0.015
0.020
-(D
eriv
. Wei
ght)
(%
)
Temperature (oC)
S1 S2 S3
Fig. 1 TPO profiles of Pt–Sn/Al2O3 catalyst coked under different
partial pressures of propylene (575 �C; S1: C3H8, 30.0%; C3H6, 2.9%;
H2, 11.9%; S2: C3H8, 31.2%; C3H6, 5.6%; H2, 10.5%; S3: C3H8,
34.5%; C3H6, 11.1%; H2, 10.1%; Ar, balance)
Table 3 Characterization results of coked samples
Samples AI AII Total AI/AII QI/QII ID/IG H/C
S1 1.10 0.33 1.43 3.33 – – 1.89
S2 0.96 0.49 1.45 1.96 3.02 – 1.83
S3 1.16 1.20 2.36 0.97 2.48 – 1.78
S4 0.82 2.98 3.80 0.28 – – 1.18
S5 0.78 0.66 1.44 1.18 2.45 – 2.00
S6 0.31 0.39 0.70 0.79 – 0.66 –
S7 0.70 0.54 1.24 1.30 – 0.68 1.81
S8 1.18 1.19 2.37 0.99 – 0.60 1.79
S9 n.d. 3.21 3.21 / – 0.64 1.17
S10 1.21 n.d. 1.21 / – 0.80 2.66
AI and AII: peak areas of TPO Peak I and Peak II in Figs. 1, 2 and 3
AI/AII: ratio of area of Peak I to area of Peak II
QI/QII: ratio of released heats during the oxidation of coke repre-
sented by Peak I and Peak II
ID/IG: intensity ratio of D band to G band in the Raman spectrum
H/C: ratio of hydrogen atoms to carbon atoms of the coke
n.d. not detected, – not characterized, / not calculated
890 Top Catal (2011) 54:888–896
123
Figure 3 shows that the areas of both Peak I and Peak II
increase with the partial pressure of propane.
Moreover, Peak II shifts to higher temperature, from
385.3 to 389.6 and then to 398.0 �C, as the partial pressure
of propylene increases (Fig. 1) and shifts to lower tem-
perature as the partial pressure of hydrogen increases
(Fig. 2). Peak I also shifts to higher temperature as the
partial pressure of propane increases (Fig. 3).
The released heats by oxidation of the cokes on Samples
S2, S3, and S5 at the two peak temperatures are calculated
from the integration of the heat flow curve and have
already been listed in Table 3 in terms of the heat ratios. It
is observed that the heat ratio is always larger than the area
ratio.
3.2 Elemental Analysis
The results of elemental analysis of the coke are listed in
Table 3 in terms of H/C ratio. For most samples (except for
Samples S4, S9, and S10), the H/C ratios are in the range of
1.7–2.0, which is an indication of the relatively high
hydrogen content of the coke formed on Pt–Sn/Al2O3
catalysts. However, the H/C ratio of S4 is much lower than
those of other coked Pt–Sn samples, which is because of
the very low hydrogen concentration in the feed flow. The
coked alumina (S9) has the lowest while the coked Pt/MgO
(S10) has the highest H/C ratio, which indicates that the
coke formed on the alumina is relatively deficient in
hydrogen while the coke formed on the Pt surfaces is rich
in hydrogen.
3.3 IR Characterization
Figures 4 and 5 show the IR spectra of the Pt–Sn/Al2O3
catalysts coked with different gas compositions. Several
absorption bands appear between 600 and 3000 cm-1, from
which different characteristics of the cokes can be identified.
Generally, the bands at 1350–1470 cm-1 reflect the bend-
ing vibrations of C–H in CH2 and CH3 groups [9], while
the stronger absorption at 2850–2960 cm-1 reflects the
stretching vibrations of C–H in CH, CH2 and CH3 groups
[9–12]. The vibration bands of C–H in alkenes (between 675
and 1000 cm-1) [13] overlap the deformation vibration
bands of C–H in aromatics (between 680 and 880 cm-1)
[14], while the bands between 1640 and 1680 cm-1 (rep-
resenting the stretching vibrations of C=C in alkenes [13])
overlap the bands 1660–2000 cm-1 (representing the
external-plane-bending vibration of C–H in aromatic rings
[15]). The bands between 1450 and 1600 cm-1 are origi-
nated from the skeleton vibration of C=C in aromatic rings
[12, 16]. The absorption at 3000–3020 cm-1 is the charac-
teristic of C=C bonds in aliphatic hydrocarbon, while the
absorption at 3020–3200 cm-1 is the characteristic of C=C
bonds in aromatics [9, 17]. Based on this information, the
strongest absorption at about 2920 cm-1 is chosen to rep-
resent the aliphatic nature of the coke and the absorption at
around 3060 cm-1 is chosen as the sign of the aromatic
nature of the coke. The involved bands are shown in Figs. 4
and 5, in which, the absorption intensities at 2920 cm-1 are
scaled to the same magnitude for convenience of
comparison.
As shown in Figs. 4 and 5, the intensity at 3060 cm-1
increases with the partial pressure of propylene but
decreases with the partial pressure of hydrogen. Propylene
promotes while hydrogen inhibits the formation of aro-
matic coke. These results are in good agreement with the
shift of Peak II in Figs. 1 and 2.
100 200 300 400 500 600
0.000
0.005
0.010
0.015
0.020
0.025
0.030
0.035
0.040-(
Der
iv. W
eigh
t) (
%)
Temperature (oC)
S4 S3 S5
Fig. 2 TPO profiles of Pt–Sn/Al2O3 catalyst coked under different
partial pressures of hydrogen (575 �C; S4: C3H8, 32.4%; C3H6, 9.4%;
H2, 1.9%; S3: C3H8, 34.5%; C3H6, 11.1%; H2, 10.1%; S5: C3H8,
34.0%; C3H6, 8.5%; H2, 13.6%; Ar, balance)
100 200 300 400 500 600
0.000
0.005
0.010
0.015
-(D
eriv
. Wei
ght)
(%
)
Temperature (o C)
S6 S7 S8
Fig. 3 TPO profiles of Pt–Sn/Al2O3 catalyst coked under different
partial pressures of propane (575 �C; S6: C3H8, 15.9%; C3H6, 4.0%;
H2, 5.3%; S7: C3H8, 30.0%; C3H6, 3.3%; H2, 4.0%; S8: C3H8, 42.7%;
C3H6, 4.3%; H2, 4.5%; Ar, balance)
Top Catal (2011) 54:888–896 891
123
3.4 Raman Spectral Analysis
Figure 6 shows the Raman spectra of the coked catalysts
(Samples S6, S7, S8, S9, and S10). Five peaks are identi-
fied at 1336, 1602, 2700, 2920, and 3200 cm-1, respec-
tively. The bands at 1336 and 1602 cm-1 are generally
considered as the D band and G band [18, 19] and attrib-
uted to the ring stretching in polyaromatic compounds,
which are considered to be the graphite-like carbon species
[20, 21]. The region of 2700–3000 cm-1 is supposed to be
the C–H stretching in alkyl hydrocarbon [19, 21, 22] while
the band at 3200 cm-1 is regarded as the C–H stretching in
aromatics [21, 22]. The intensity ratio of D band to G band
is always used as an important measure of the degree of
graphitization of carbon materials [23, 24]. A high ratio
indicates a low degree of graphitization. Table 3 lists the
ID/IG ratios of the coked catalysts.
As shown in Table 3, the ID/IG ratios of Samples S6–S9
are very close to each other. However, the ID/IG ratio of the
coked Pt/MgO catalyst (Sample S10) is large indicating the
coke on the Pt/MgO catalyst has a low degree of graphi-
tization. This is consistent with the results of elemental
analysis and indicates that the coke on the alumina and Pt–
Sn catalysts has a H/C ratio lower than that on the Pt/MgO
catalyst. Furthermore, as shown in Fig. 6, the lower
intensity of C–H vibration in the aromatics (3200 cm-1) on
coked Pt/MgO catalyst were observed comparing to other
samples. It indicates that the degree of graphitization of the
coke on the Pt surfaces is lower, and coke molecules are
more like olefin/paraffin.
3.5 Reaction Orders to Propane, Propylene
and Hydrogen
Kinetic experiments were carried out under different gas
compositions as summarized in Table 2, where the aver-
aged compositions of gas between inlet and outlet of the
reactor are also listed. During the course of reaction for
80 min, we observed that the amounts of the two types of
coke were both increasing approximately linearly with time
on stream (not shown here). The linear build-up of coke
with time was also reported by Kumar et al. [25]. There-
fore, the average coking rates were estimated in the present
2750 2800 2850 2900 2950 3020 3040 3060 3080 3100
2962-CH3 (a)
2920-CH2 (a)
S1
S2
S3
Inte
nsity
(a.
u.)
Wavenumbers (cm-1)
Fig. 4 FT-IR spectra of catalyst coked under different partial
pressures of propylene (575 �C; S1: C3H8, 30.0%; C3H6, 2.9%; H2,
11.9%; S2: C3H8, 31.2%; C3H6, 5.6%; H2, 10.5%; S3: C3H8, 34.5%;
C3H6, 11.1%; H2, 10.1%; Ar, balance)
2750 2800 2850 2900 2950 3020 3040 3060 3080 3100
S4
S3
S5
Inte
nsity
(a.
u.)
Wavenumbers (cm-1)
2962-CH3 (a)
2920-CH2 (a)
Fig. 5 FT-IR spectra of catalyst coked under different partial
pressures of hydrogen (575 �C; S4: C3H8, 32.4%; C3H6, 9.4%; H2,
1.9%; S3: C3H8, 34.5%; C3H6, 11.1%; H2, 10.1%; S5: C3H8, 34.0%;
C3H6, 8.5%; H2, 13.6%; Ar, balance)
1000 1500 2000 2500 3000 3500 4000
1000 1500 2000 2500 3000 3500 4000
1000 1500 2000 2500 3000 3500 4000
0
1000 1500 2000 2500 3000 3500 40001000 1500 2000 2500 3000 3500 4000
S10
S8
S7
S6
Inte
nsity
(a.
u.)
Raman shift (cm-1)
S9
1336 1602
Fig. 6 Raman spectra of coked samples (S6–S8 are coked Pt–Sn/
Al2O3 catalysts, S9 is coked alumina, S10 is coked Pt/MgO
catalyst;575 �C; S6: C3H8, 15.9%; C3H6, 4.0%; H2, 5.3%; S7:
C3H8, 30.0%; C3H6, 3.3%; H2, 4.0%; S8: C3H8, 42.7%; C3H6, 4.3%;
H2, 4.5%; S9: C3H8, 0.0%; C3H6, 9.9%; H2, 0.0%; S10: C3H8, 35.0%;
C3H6, 0.2%; H2, 0.2%; Ar, balance)
892 Top Catal (2011) 54:888–896
123
work based on the measured coke contents in 80 min of
time on stream by assuming a constant coking rate.
The coking rates r1 and r2 were determined based on
Peak I and II from the TPO spectra. The logarithms of the
rates were then plotted against the logarithms of the partial
pressures of propane, propylene and hydrogen, respec-
tively, as shown in Figs. 7, 8 and 9.
Figure 7 shows that r1 is independent of the propylene
pressure. As a result, Samples S1–S5 and Sample S7 are all
used to find the influences of the hydrogen pressure on r1,
because these samples were obtained under similar partial
pressures of propane. Figure 8 shows that r1 is also inde-
pendent of the partial pressure of hydrogen. Samples S1–
S8 are then used to find the influences of the propane
pressure on r1. Figure 9 shows that r1 is dependent on the
partial pressure of propane with an apparent reaction order
of 1.7. Summarizing Figs. 7, 8 and 9 we see that r2 is
dependent on the partial pressures of propane, propylene
and hydrogen and the reaction orders are 1.4, 1.0 and -0.7,
respectively.
4 Discussions
4.1 Assignment of Peaks in the TPO Profiles
TG is a very useful method to identify and quantify the
coke on the catalysts [26–28]. The peaks in the TPO pro-
files can be attributed to the loss of carbonaceous deposits
with different compositions and structures and/or the car-
bonaceous deposits located at different sites on the catalyst.
The two peaks identified in the TPO profiles of coked
catalysts shown above indicate that two types of coke are
formed.
The two types of coke are usually considered to be
formed on the metal and support, respectively [29, 30]. To
confirm the locations of the coke, comparative experiments
were carried out on pure alumina, which was used as the
support for the Pt–Sn catalyst, and the Pt/MgO catalyst,
which had no acid sites on the support (Table 1). For coked
alumina, a single peak emerges at around 470 �C in the
TPO profile, while for coked Pt/MgO catalyst the peak
appears at around 310 �C. This information confirms that
Peak I and Peak II in the TPO profiles of Samples S1–S8
are corresponding to the coke formed on the metal and
support, respectively. As a result, in this study, we assign
Peak I to the coke deposited on the metal and Peak II to the
coke deposited on the support.
5.6 5.8 6.0 6.2 6.4 6.6 6.8 7.0 7.2-10.5
-10.0
-9.5
-9.0
-8.5
-8.0
-7.5
-7.0
-6.5
-6.0
S3
S2
S2S1
ln(PC3H6
)(PC3H6
, 101Pa)
r1, coking rate on the metal
r2, coking rate on the support
ln(r
) (r
, mg_
coke
/(10
2m
g_ca
t.s))
k=0.0
k=1.0
S1
Fig. 7 The relationship between coke formation rates and partial
pressures of propylene (575 �C, open square Samples S1–S3; filledcircle Samples S1–S3)
5.2 5.6 6.0 6.4 6.8 7.2-10.0
-9.5
-9.0
-8.5
-8.0
-7.5
-7.0
-6.5
-6.0
S5
S3
S4
S1
S7
S5S7S4
S1
S3
k=-0.7
k=-0.7
r1, coking rate on the metal
r2, coking rate on the support
ln(r
) (r
, mg_
coke
/(10
2m
g_ca
t.s))
ln(PH2
) (PH2
, 101Pa)
k=0.0
S2
Fig. 8 The relationship between coke formation rates and partial
pressures of hydrogen (575 �C, open square Samples S1–S5 and S7;
filled circle Samples S1 and S7, Samples S3–S5)
7.4 7.6 7.8 8.0 8.2 8.4-10.5
-10.0
-9.5
-9.0
-8.5
-8.0
-7.5
-7.0
-6.5
-6.0
S7S4
S2
r1, coking rate on the metal
r2, coking rate on the support
ln(PC3H8
) (PC3H8
, 101Pa)
ln(r
) (r
, mg_
coke
/(10
2m
g_ca
t.s))
k=1.4k=1.7
S7S6
S8
S5
S3S1
S6
Fig. 9 The relationship between coke formation rates and partial
pressures of propane (575 �C, open square Samples S1–S8; filledcircle Samples S6–S8)
Top Catal (2011) 54:888–896 893
123
4.2 The Nature of Coke
It is interesting to note from Table 3 that the ratios of the
combustion heats of the two types of coke are much higher
than the ratios of the masses determined by TPO. This fact
clearly indicates that the coke on the metal has a higher
H/C ratio than that on the support, which is consistent with
the results of elemental analysis (Samples S9 and S10).
Based on the results of TPO, TG, and DTA of the Pt/Al2O3
and Pt–Sn/Al2O3 catalysts coked in n-butane dehydroge-
nation, Zhang et al. [31] also suggested that the coke
formed on the metal had a higher hydrogen content. The
same conclusion was obtained by Srihiranpullop et al. [32]
when studying the coke formation on Pt, Pt–Sn, and Pt–Sn–K
catalysts during n-hexane dehydrogenation. In addition, we
note that Peak I shifts to higher temperature when the
propane pressure is increased. This is because the coke
covers the metal surface and reduces the rate of coke
combustion by increasing the resistance of oxygen diffu-
sion [29].
The Raman spectral analysis of Samples S9 and S10
shows that the ID/IG ratio of Sample S9 is much lower than
that of Sample S10, which indicates that the coke formed
on the support is more graphitized. The very weak intensity
of the band at 3200 cm-1 in the Raman spectrum of coked
Pt/MgO catalyst also confirms that the coke on the metal is
less dehydrogenated.
Based on this analysis, the catalyst samples containing
more coke on the support should have lower ID/IG ratios.
Hence the ID/IG ratios of Samples S6, S7, S8, S9, and S10
should have been in the decreasing order of
S10 [ S7 [ S8 [ S6 [ S9. However, the ID/IG ratio of
Sample S8 is the lowest, and the coke is hard to burn, as
indicated by the rightward shift of the position of peak II of
Sample S8. This is because the high propane concentration
greatly increases the concentration of coke precursor on the
support, which leads to an increased degree of polymeri-
zation and aromatization of the coke.
Increasing the propylene concentration will increase the
degree of polymerization of the polymers [33], which will
be further dehydrogenated to aromatics. The degree of the
aromatization of the coke is enhanced as a result of the
increased propylene concentration and weakened as a
result of the increased hydrogen concentration, as hydrogen
will reduce the molecular weight of the polymer during
propylene polymerization [34]. This fact also explains the
shift of Peak II in Figs. 1 and 2.
In general, the H/C ratios of the cokes are very high
(Table 1), especially on the metal surfaces, indicating that
the coke contains mainly aliphatic hydrocarbons. The
high H/C ratios on metal surfaces are consistent with
the relatively large peak of 2920 cm-1 in IR spectra and
the relatively large peak of 2700–3000 cm-1 in Raman
spectra.
4.3 Dependence of Coking Rate on Gas Concentration
The rate of coke formation on the metal is found dependent
on the propane pressure (Fig. 9). The apparent reaction
order is about 1.7 with respect to propane and zero reaction
order with respect to hydrogen and propylene. Figures 7
and 9 also show that the rate of coke formation on the
support increases with the propylene or propane pressures,
while Fig. 8 shows that it decreases with the hydrogen
pressure. The apparent reaction orders of coking on the
support are about first order with respect to propylene
(Fig. 7), -0.7 order with respect to hydrogen (Fig. 8) and
1.4 order with respect to propane (Fig. 9). It suggests that
coke can be formed both from propane and propylene, but
with different reaction mechanisms. An apparent reaction
order of 1.4 with respect to propane can not explain that
propylene is the main precursor for coke formation, since
the reaction order to propane is about one for propylene
formation, and the reaction order to propylene is about one
for coke formation.
4.4 Coke Formation Mechanism
4.4.1 Coke Formation on the Metal
Many reaction mechanisms have been tested, and only the
following mechanism of coke formation on the metal can
describe the experimental observation:
C3H8ðg)þ 2� ) C3H7 � þH� ð1ÞC3H7 � þ� ) C3H6 � þH� ð2ÞC3H6� ) C3H6ðg)þ � ð3Þ2C3H6� ) C6H12 � þ� ð4ÞH � þH� , H2 þ 2� ð5Þ
The apparent reaction order of propane of about 1.7
indicates that two C3 intermediates are involved in the
carbon formation. Furthermore, the H/C ratios of the coked
Pt–Sn catalysts of 1.7–2.0 indicate a high hydrogen content
of the coke. Therefore, the formation of C6H12* from two
C3H6* intermediates is proposed as the kinetic relevant
step for coke formation (Step 4).
It is assumed that the dehydrogenation steps (Steps 1
and 2) are both irreversible and far from equilibrium
[35–37], and the desorption of propylene (Step 3) is also
considered irreversible. Step 4 is assumed to be the kinetic
relevant step and the rate of Step 4 is considered to be
much slower than those of Steps 2 and 3.
894 Top Catal (2011) 54:888–896
123
By assuming the steady-state of the intermediates,
C3H7* and C3H6*, we obtain,
dhC3H7
dt¼ k1PC3H8h
2� � k2hC3H7h� ¼ 0 ð6Þ
dhC3H6
dt¼ k2hC3H7
h� � k3hC3H6¼ 0 ð7Þ
Then the site coverages of C3H7* and C3H6* are
determined by,
hC3H7¼ k1PC3H8
h�k2
ð8Þ
hC3H6¼ k1PC3H8
h2�
k3
ð9Þ
The coke formation rate on the metal is given by Eq.
(10) by assuming C3H7* as the most abundant surface
specie:
rc ¼ k4h2C3H6¼
kcP2C3H8
ð1þ KIPC3H8Þ4
ð10Þ
where,
kc¼k4
k1
k3
� �2
ð11Þ
KI ¼k1
k2
ð12Þ
Increasing propane concentration in the gas phase
concurrently increases the concentration of adsorbed
propylene (C3H6*) and hence the amount of coke on the
metal. A relatively low value of KI in Eq. (10) can lead to
the apparent order of 1.7 with respect to propane. The
irreversible desorption of propylene subsequently leads to
the zero order to propylene. The irreversible reaction of
steps in propane dehydrogenation and relatively weak
adsorption of hydrogen comparing to the C3 surface
intermediate result in the weak effect of hydrogen.
By fitting the rates of coking on the metal at different
gas compositions, kc and KI are estimated to be
4.15 9 10-5 mg coke/(mg cat.s) and 7.18 9 10-1,
respectively. Moreover, one can see that the model fits
the experimental data quite well (Fig. 10), which implies
the proposed kinetic model for coking on the metal is
reasonable.
It is noted here that although the hydrogen pressure has
little effect on the rate of coking on the metal, it changes
significantly the hydrogen content in the coke. At higher
hydrogen partial pressure, the coke is less dense and
compact. In addition, the coke formed on the metal can be
further softened through hydrogenation [38].
4.4.2 Coke Formation on the Support
Coke formation on the support mainly involves poly-
merization/oligomerization, condensation, cyclization,
hydride transfer, etc. Referring to the mechanism proposed
by Caeiro et al. [39], coke formation on the support can be
divided into two stages, i.e., the conversion of propane to
olefins (propylene and ethylene) by protolysis and trans-
formation of these olefins to aromatic hydrocarbons. The
second stage involved a number of reactions, such as
oligomerization–cracking reactions (producing C4–C10
olefins), hydrogen transfer (producing dienes), cyclization
(generating cycloalkenes), and further hydrogen transfer
(producing cyclic diolefins and finally aromatics). Based on
this mechanism, increasing the propylene partial pressures
will certainly increase the amount of coke on the support.
The coke precursor formed on the metal may migrate to
the support [32] and then undergoes subsequent poly-
merization/oligomerization, condensation and so on. Thus,
increasing the partial pressure of propane would increase
the rate of coke formation on the support. Sn in the Pt
catalyst will weaken the binding of hydrocarbon to the
metal [30, 40], and promote the migration of the coke
precursor from the metal to the support. The presence of
hydrogen will weaken the acidity of the support by con-
verting Brønsted acid sites to Lewis acid sites and thus
reduce the coke formation rate. Based on the above dis-
cussions, a coke formation mechanism is proposed as in
Fig. 11. Propane is firstly dissociated on the metal and the
coke precursor is formed through dehydrogenation; then
the ‘‘soft coke’’ is generated on the metal from the coke
precursor. The coke precursor generated on the metal will
also migrate to the acid sites. On these acid sites, the coke
precursor and the adsorbed propylene undergo poly-
merization/oligomerization, condensation, cyclization and
S1 S2 S3 S4 S5 S6 S7 S8-10.5
-10.0
-9.5
-9.0
-8.5
-8.0
-7.5
-7.0
-6.5
-6.0
ln(r
) (r
, mg_
coke
/(10
2m
g_ca
t.s))
Coking rate on the metal (Model)
Coking rate on the metal (Experiment)
Samples
Fig. 10 Comparison between model prediction and experiments of
coking rates on metal
Top Catal (2011) 54:888–896 895
123
hydride transfer, etc., resulting in the formation of ‘‘hard
coke’’.
5 Conclusion
Two types of coke are identified on coked Pt–Sn/Al2O3
catalyst and assigned to coke on the metal and the support,
respectively. Coke formed on the metal has aliphatic
hydrocarbon characteristic, containing more hydrogen than
that formed on the support. Coke formed on the support has
an aromatic characteristic. The rate of coke formation on
the metal is weakly dependent on the propylene and
hydrogen pressures but increases concurrently with the
propane pressure while the rate of coke formation on the
support increases concurrently with the propane and pro-
pylene pressures and decreases with the hydrogen pressure.
A mechanism for coke formation has been proposed based
on the kinetic analysis. The reaction between the two
strong adsorbed C3H6*, which was formed by dehydroge-
nation of propane, was identified as the kinetic relevant
step for the coke formation on the Pt surfaces. A portion of
the precursor migrates to the acid site and is involved in the
coke formation on the support. In addition, the propylene in
the gas phase can also adsorb on the support and form coke
through reactions such as polymerization/oligomerization,
condensation and so on.
Acknowledgment This work is supported by Natural Science
Foundation of China (No. 20736011)
References
1. Galvita V, Siddiqi G, Sun P, Bell AT (2010) J Catal 271:209
2. Liu H, Zhang L, Li X, Huang S, Liu S, Xin W, Xie S, Xu L
(2009) J Nat Gas Chem 18:331
3. Sun K (2004) Petrochem Des 21:25
4. Cosyns J, Chodorge J, Commereuc D, Torck B (1998) Hydrocarb
Process 77:61
5. Praserthdam P, Mongkhonsi T, Kunatippapong S, Jaikaew B,
Lim N (1997) Stud Surf Sci Catal 111:153
6. Santhosh Kumar M, Holmen A, Chen D (2009) Microporous
Mesoporous Mater 126:152
7. Larsson M, Hulten M, Blekkan EA, Andersson B (1996) J Catal
164:44
8. Rebo HP, Chen D, Blekkan EA, Holmen A (1998) Stud Surf Sci
Catal 119:617
9. Yang JL, Stansberry PG, Zondlo JW, Stiller AH (2002) Fuel
Process Technol 79:207
10. Eisenbach D, Gallei E (1979) J Catal 56:377
11. Van Doorn J, Moulijn JA (1990) Fuel Process Technol 26:39
12. Cerqueira HS, Sievers C, Joly G, Magnoux P, Lercher JA (2005)
Ind Eng Chem Res 44:2069
13. Geach A (1996) Wear Check Afr Tech Bull 2:1
14. Wolthuis E, Bossenbroek B, DeWall G, Geels E, Leegwater A
(1963) J Org Chem 28:148
15. Sato K, Ikeda S, Iida M, Oshima A, Tabata Y, Washio M (2003)
Nucl Instrum Methods Phys Res Sect B 208:424
16. Matsushita K, Hauser A, Marafi A, Koide R, Stanislaus A (2004)
Fuel 83:1031
17. Jong SJ, Pradhan AR, Wu JF, Tsai TC, Liu SB (1998) J Catal
174:210
18. Guichard B, Roy-Auberger M, Devers E, Rebours B, Quoineaud
AA, Digne M (2009) Appl Catal A 367:1
19. Korhonen ST, Airaksinen SMK, Banares MA, Krause AOI
(2007) Appl Catal A 333:30
20. Zeng Z, Natesan K (2003) Chem Mater 15:872
21. Airaksinen SMK, Banares MA, Krause AOI (2005) J Catal
230:507
22. Chua YT, Stair PC (2003) J Catal 213:39
23. Dumont M, Chollon G, Dourges M, Pailler R, Bourrat X, Naslain
R, Bruneel JL, Couzi M (2002) Carbon 40:1475
24. Li J, Naga K, Ohzawa Y, Nakajima T, Shames AP, Panich AI
(2005) J Fluor Chem 126:265
25. Santhosh Kumar M, Chen D, Holmen A, Walmsley JC (2009)
Catal Today 142:17
26. Yang Z, Zhang Y, Wang X, Zhang Y, Lu X, Ding W (2010)
Energy Fuels 24:785
27. Sun L, Guo X, Liu M, Wang X (2010) Ind Eng Chem Res 49:506
28. Li X, Zhang W, Li X, Liu S, Huang H, Han X, Xu L, Bao X
(2009) J Phys Chem C 113:8228
29. Duprez D, Hadj-Aissa M, Barbier J (1989) Appl Catal 49:67
30. Liwu L, Tao Z, Jingling Z, Zhusheng X (1990) Appl Catal 67:11
31. Tao Z, Jingling Z, Liwu L (1991) Stud Surf Sci Catal 68:143
32. Srihiranpullop S, Praserthdam P, Mongkhonsi T (2000) Korean J
Chem Eng 17:548
33. Novokshonova LA, Tsvetkova VI, Chirkov NM (1963) Russ
Chem Bull 12:1077
34. Soga K, Siono T (1982) Polym Bull 8:261
35. Li Q, Sui Z, Zhou X, Chen D (2011) Appl Catal A 398:18
36. Azzam KG, Jacobs G, Shafer WD, Davis BH (2010) Appl Catal
A 390:264
37. Biloen P, Dautzenberg FM, Sachtler WMH (1977) J Catal 50:77
38. Liu K, Fung SC, Ho TC, Rumschitzki TC (2003) Ind Eng Chem
Res 42:1543
39. Caeiro G, Carvalho RH, Wang X, Lemos MAND, Guisnet M,
Ramoa Ribeiro F (2006) J Mol Catal A 255:131
40. Lieske H, Sarkany A, Volter J (1987) Appl Catal 30:69
PropylenePropane
Coke precursorCoke on metal Coke on support
Al2O3
Migration
Pt Acid site
Fig. 11 Coke formation mechanism
896 Top Catal (2011) 54:888–896
123