Sei sulla pagina 1di 8

Ind. Eng. Chem. Res.

2002, 41, 5619-5626 5619

Kinetic Modeling of Dehydrogenation of Isobutane on Chromia/


Alumina Catalyst
Sanna M. K. Airaksinen,* M. Elina Harlin,† and A. Outi I. Krause
Helsinki University of Technology, Department of Chemical Technology, P.O. Box 6100,
FIN-02015 HUT, Finland

The kinetics of isobutane dehydrogenation was studied on a chromia/alumina catalyst developed


for fluidized-bed operation. The dehydrogenation activity measurements were carried out in a
laboratory-scale plug-flow reactor at 520-580 °C under atmospheric pressure. Several kinetic
models were derived from different mechanisms, and the suitability of the models in describing
the rate of the dehydrogenation reaction was tested. The results of the kinetic modeling suggested
that the rate-determining step is isobutane adsorption, possibly on a pair of chromium and oxygen
ions. A satisfactory description of the reaction rate depended on inclusion of the isobutene and
hydrogen adsorption parameters in the mathematical model. The activation energy of the
dehydrogenation reaction was estimated to be 133-142 kJ/mol.

Introduction in crystalline chromia. On alumina-supported catalysts,


redox Cr3+ is the most abundant species at low chro-
Light alkenes, such as propene and butenes, are mium contents, and its amount stabilizes to about 2-3
important intermediates in the manufacture of polymers atoms of Cr/nm2 (2-3 wt % chromium, depending on
and chemicals. The most selective way to produce these the support surface area) for total chromium contents
short-chain alkenes is the direct catalytic dehydroge- above 5 atoms of Cr/nm2 (4-8 wt %).5,6,8 The amount of
nation of the corresponding alkanes. Industrial dehy- amorphous chromia increases with the total chromium
drogenation processes use chromia- or platinum-based content, and crystalline chromia is mainly observed, by
catalysts generally supported on alumina and promoted X-ray diffraction, above about 8-10 atoms of Cr/nm2 (6-
with alkali metals.1 The structure and activity of 16 wt %).5,6,8 It has been suggested that both redox and
chromia catalysts have been studied extensively in an nonredox Cr3+ are active in dehydrogenation,5,6 whereas
attempt to understand better how the properties of crystalline chromia is less active.6
supported chromia influence the dehydrogenation In addition to the Cr3+ ions, surface oxygen ions might
activity.2-10 However, an understanding of the system also participate in the reaction.3,9,10,21,22
also requires information about the reaction mechanism.
Dehydrogenation might involve a stage where the
The kinetics and mechanism of alkane dehydrogena-
alkane adsorbs and dissociates, in one or two steps, onto
tion on chromia/alumina catalysts have been explored
a pair of chromium and oxygen ions to give an alkyl
in various publications.11-19 However, the findings of
group bonded to surface chromium and a hydrogen atom
these studies have not always been in agreement. Not
bonded to surface oxygen,3,21,22 as schematized in eq 1
only different mechanisms but also different rate-
determining steps and, thus, different kinetic models
have been proposed. The rate-determining step in the Cr-O + R-H f R‚ ‚ ‚Cr-O‚ ‚ ‚H (1)
dehydrogenation of isobutane has been suggested to
involve alkane adsorption,15,16 whereas in the dehydro- In kinetic modeling studies, the active site of the
genation of propane and n-butane, it has been proposed catalyst has generally been described as a single ion.
to involve the surface reaction of the adsorbed alkyl However, if both the chromium and oxygen ions take
group.11,12,18,19 Furthermore, results of tracer studies part in the reaction, as has been suggested, the active
with 13C-labeled isobutane and deuterium on chromia site should be described as a Cr-O pair.
gel20 and with deutero-propane on chromia/alumina17 In this work, the dehydrogenation of isobutane on a
suggest a mechanism with the surface reaction as the chromia/alumina catalyst was studied to clarify the
slow step. reaction kinetics of alkane dehydrogenation. Kinetic
The dehydrogenation activity of chromia catalysts is models were derived on the basis of different mecha-
most often attributed to coordinatively unsaturated Cr3+ nisms and rate-determining steps, and the suitabilities
ions.3-10 Varying with the sample properties and treat- of the models in describing the measured dehydroge-
ment conditions, three types of Cr3+ have been observed: nation activities were tested.
2,3,5,6 (i) redox Cr3+ formed in the reduction of Cr5+ and

Cr6+ present on oxidized samples, (ii) nonredox Cr3+ in Experimental Section


an amorphous chromia phase that is present in both
reduced and oxidized samples, and (iii) the Cr3+ present Materials. The chromia/alumina catalyst used in the
study was a sample developed for fluidized-bed opera-
* To whom correspondence should be addressed. E-mail tion. The chromium content of the catalyst was 12 wt
address: airaksinen@polte.hut.fi. Fax: +358-9-451 2622. %, and the average particle size was 96 µm. The gases
† Current address: Fortum Oil and Gas Oy, P. O. Box 310, were supplied by Aga, and their purities were N2,
FIN-06101 Porvoo, Finland. 99.999%; air, 99.5%; H2, 99.999%; i-C4H10, 99.95%; and
10.1021/ie020371j CCC: $22.00 © 2002 American Chemical Society
Published on Web 10/17/2002
5620 Ind. Eng. Chem. Res., Vol. 41, No. 23, 2002

Table 1. Reaction Conditions Applied in the Kinetic The hydrogen prereduction was done at 590 °C with
Activity Measurements 10% H2/N2 and a reduction time of 15 min. The reduc-
feed composition tion products were analyzed by FTIR spectroscopy. The
amount of WHSV T i-C4H10/H2/N2 influence of hydrogen prereduction on the catalyst
series catalyst (g) (h-1) (°C) (mol:mol:mol) activity and product distribution was studied in cycles
1 0.1 15, 30 520-580 1:0:1, 3:0:7 1-12: cycles 1-6 and 12 did not include a prereduction
2 0.2 7.5, 15, 30 520-580 1:0:1 stage, whereas cycles 7-11 did. The kinetic measure-
3 0.1 15, 30 520-580 1:1:1, 1:1/2:1, 1:1/4:1 ments were done with the prereduced catalyst com-
mencing at about cycle 16. During the dehydrogenation
Ar, 99.999%. Nitrogen and argon were purified with stage, the isobutane feed was continued for 11 or 15 min.
Oxisorb (Messer-Griesheim). The reaction products were monitored by FTIR spec-
Dehydrogenation Equipment and Analytical troscopy for the first 6 min on stream, a GC sample was
Methods. The dehydrogenation activities were mea- taken after 8 or 10 min, and the FTIR analysis was then
sured in a continuous-flow fixed-bed microreactor. The continued. The catalyst was regenerated after the
catalyst sample (0.1 or 0.2 g) was loaded between two dehydrogenation stage with 2% O2/N2 for 2 min, with
layers of silicon carbide into a quartz-glass reactor with 5% O2/N2 for 2 min, and with 10% O2/N2 for 30 min.
a 7-mm inside diameter. The catalyst bed length was 3 The amounts of the regeneration products, carbon
or 6 mm for 0.1- or 0.2-g loading, respectively. The oxides and water, were determined by FTIR spectros-
reaction temperature was monitored with a thermo- copy. The amount of coke deposited on the catalyst
couple in the catalyst bed, and the flow rates of gases during dehydrogenation was calculated from the mea-
were regulated with mass flow controllers. sured amounts of carbon oxides.
The reaction products were analyzed on-line with a Thermal reactions in the system were investigated
Gasmet Fourier transform infrared (FTIR) gas analyzer by replacing the catalyst with silicon carbide but
(Temet Instruments Ltd.) equipped with a Peltier-cooled otherwise implementing the dehydrogenation procedure
mercury-cadmium telluride detector and with an HP as usual. Internal mass transfer was considered with
6890 gas chromatograph (GC) equipped with an HP the Weisz-Prater criterion, and external mass transfer
PLOT/Al2O3 “M” column and a flame ionization detector. by conducting dehydrogenation experiments with two
The FTIR spectra were recorded in the wavenumber catalyst loadings at the same reaction temperature and
range 4000-850 cm-1 with a resolution of 8 cm-1 and WHSV value.
a scanning rate of 10 scans/s. The analysis cuvette (9 Calculation of Conversions and Selectivities.
cm3) was maintained at constant temperature (175 °C) The conversions of isobutane and the product selectivi-
and pressure (1.03 bar). The spectra were measured ties determined from the FTIR results were calculated
every 3 s during the first 2 min on stream, every 5 s on the basis of the measured amounts of carbon-
during the next 2 min, and every 30 or 60 s thereafter. containing compounds. The molar amount of carbon in
The compounds analyzed by FTIR spectroscopy were the feed was assumed to be equal to the molar amount
methane, ethane, ethene, propane, propene, n-butane, of carbon in the product; the amount of coke deposited
isobutane, 1-butene, cis-2-butene, trans-2-butene, 1,3- on the catalyst during dehydrogenation was ignored.
butadiene, isobutene, benzene, toluene, carbon monox- The amount of isobutane fed CA,in was thus
ide, carbon dioxide, and water. The composition of the
∑C1 × 4 + ∑C2 × 4 + ∑C3 × 4 + ∑C4 +
product stream was calculated by the classical least- 1 2 3
squares multicomponent analysis method using an
CA,in )
analysis program (Calcmet, Temet Instruments Ltd.) 6 7
with calibration spectra for the measured compounds. C6 × + C7 × (2)
4 4
The method for analyzing light hydrocarbons by FTIR
spectroscopy has been described in detail elsewhere.23 The accuracy of this calculation was verified by calcu-
Activity Measurements. Three series of activity lating the mass balance of carbon in the reactor sys-
measurements were carried out at atmospheric pressure tem: the calculated inlet isobutane concentration CA,in
(1.03 bar) under different reaction conditions. As de- was compared with the measured inlet concentration.
scribed in Table 1, the catalyst loading, reaction tem- The conversion of isobutane X was calculated as
perature, isobutane weight hourly space velocity (WHSV),
and molar ratios of isobutane to hydrogen and to CA,in - CA,out
nitrogen in the feed were varied in the experiments. X ) 100% × (3)
CA,in
Each series consisted of about 50 (pre)reduction-
dehydrogenation-regeneration cycles with the same
catalyst load. During the first 15-20 cycles, the catalyst The selectivity Si to product i was calculated as
activity stabilized to a constant level. Then, the kinetic
Ci
measurements were started. The stability of the catalyst Si ) 100% × (4)
activity was verified regularly with standard experi- CA,in - CA,out
ments. The stabilization and standard dehydrogenation
stages were carried out at 580 °C and a WHSV value of The yield of product i was calculated by multiplying the
15 h-1. The isobutane feed was diluted with nitrogen, conversion of isobutane X by the selectivity Si to product
the molar ratio of isobutane to nitrogen being 1:1. No i. The conversions and selectivities determined from the
hydrogen was added to the feed. Some experiments were GC results were calculated on a molar basis.
performed with argon instead of nitrogen as the inert Kinetic Models. Several kinetic models were tested
gas. to describe the rate of dehydrogenation; the equations
The catalyst was reduced either (i) with hydrogen are presented in Table 2. Model 1 is a power-law model
before the dehydrogenation stage or (ii) with isobutane that includes both the forward and the reverse reaction
during the first few minutes on the isobutane stream. rates. Models 2-8 were derived from possible dehydro-
Ind. Eng. Chem. Res., Vol. 41, No. 23, 2002 5621

Table 2. Kinetic Models Tested for the Dehydrogenation Data


model rate equation model rate equation

1 (
k pA -
pEpH2
K ) 7
(
k′ pA -
pEpH2
K )
1/2
(III-b) [1 + KApA + K′pEpH2 + (KH2pH2)1/2]2

2
(
k′ pA -
K
pEpH2
) 8
(
k′ pA -
K
pEpH2
)
(I-a) 1 + KEH2pEpH2 + KEpE + KH2pH2 (IV-a) 1 + KEH2pEpH2 + KH2pH2

3
(
k′ pA -
K
pEpH2
) 9
(
k′ pA -
K
pEpH2
)
(I-b) (1 + KApA + KEpE + KH2pH2)2 (IV-b) 1 + KApA + KH2pH2

4 K (
k′ pA -
pEpH2
) 10
(
k′ pA -
K
pEpH2
)
(II-a) [1 + K′pEpH2 + KEpE + (K′′′pH2)1/2 + KH2pH2]2
1/2 1 + KEH2pEpH2 + KEpE

(
k′ pA -
pEpH2
K ) (
k′ pA -
K
pEpH2
)
[ ]
5 11
(II-b) pA 2 1 + KEpE + KH2pH2
(K′′′pH2)1/2 1 + K′′ 1/2 + KEpE + (K′′′pH2)1/2 + KH2pH2
pH2

6
(
k′ pA -
K
pEpH2
) 12
(
k′ pA -
K
pEpH2
)
(III-a) 1 + KEH2pEpH2 + K′pEpH21/2 + (KH2pH2)1/2 (1 + KH2pH2)(1 + KApA + KEpE)

genation mechanisms with use of the general principles rate of isobutane dehydrogenation on unsupported
of the Langmuir-Hinshelwood approach.27 These mod- chromia gel.
els will be referred to in the form “model I-a”, where Mechanisms I and II imply that the hydrocarbon and
the Roman numeral indicates the mechanism and the hydrogen species occupy the same type of active surface
letter (a or b) the reaction rate-determining step (RDS), site S, for example, chromium. Mechanisms III and IV,
with the letter a referring to the adsorption of isobutane however, assume that (i) the active site requires both a
and b to the surface reaction. Models with isobutene or chromium and an oxygen ion, as has been suggested;3,21,22
hydrogen desorption as the slow step were tested in (ii) some steps occur on the same Cr-O pair, with the
some calculations, but these models described the use of both the chromium and the oxygen ion; and (iii)
activity results poorly. Moreover, it has been concluded it is not necessary in the model derivation to write
from tracer experiments that the desorptions of isobutene separate site balances for the chromium and oxygen
and hydrogen are in equilibrium.20 Desorption models ions. In mechanism III, proposed by Carrà et al.12 for
were, therefore, not considered more closely. Mecha- n-butane dehydrogenation on chromia/alumina, the
nisms I and II are presented in eqs 5a-d and 6a-e, alkane adsorbs dissociatively on a Cr-O pair (step 7a).
respectively. In the surface reaction, the adsorbed alkyl group reacts
with a neighboring Cr-O pair (step 7b), and finally, the
Mechanism I alkene and the hydrogen desorb from the surface (steps
i-C4H10(g) + S T i-C4H10‚ ‚ ‚S (5a) 7c and d).

i-C4H10‚ ‚ ‚S + S T i-C4H8‚ ‚ ‚S + H2‚ ‚ ‚S (5b) Mechanism III


i-C4H8‚ ‚ ‚S T i-C4H8(g) + S (5c) i-C4H10(g) + Cr-O T i-C4H9‚ ‚ ‚Cr-O‚ ‚ ‚H (7a)

H2‚ ‚ ‚S T H2(g) + S (5d) i-C4H9‚ ‚ ‚Cr-O‚ ‚ ‚H + Cr-O T


i-C4H8‚ ‚ ‚Cr-O‚ ‚ ‚H + H‚ ‚ ‚Cr-O (7b)
Mechanism II
i-C4H10(g) + 2S T i-C4H9‚ ‚ ‚S + H‚ ‚ ‚S (6a) i-C4H8‚ ‚ ‚Cr-O‚ ‚ ‚H T i-C4H8(g) + Cr-O‚ ‚ ‚H (7c)

i-C4H9‚ ‚ ‚S + S T i-C4H8‚ ‚ ‚S + H‚ ‚ ‚S (6b) Cr-O‚ ‚ ‚H + H‚ ‚ ‚Cr-O T H2(g) + 2 Cr-O (7d)


i-C4H8‚ ‚ ‚S T i-C4H8(g) + S (6c) In the reaction equations, the notation i-C4H9‚ ‚ ‚Cr-
2H‚ ‚ ‚S T H2‚ ‚ ‚S + S (6d) O‚ ‚ ‚H, for example, means that the alkyl group and
the hydrogen atom are adsorbed on the Cr-O pair.
H2‚ ‚ ‚S T H2(g) + S (6e) Mechanism III can be simplified to mechanism IV,
suggested by Burwell et al.,22 by assuming that the
Kao et al.20 used mechanism II with a surface reaction dehydrogenation reaction takes place on a single Cr-O
(step 6b) as the rate-determining step to describe the pair. Recently, Weckhuysen and Schoonheydt3 proposed
5622 Ind. Eng. Chem. Res., Vol. 41, No. 23, 2002

a similar mechanism for dehydrogenation. Although where Amean is A0e-E/RTmean and z is 1/R(1/T - 1/Tmean).
they described the adsorption of the alkane and the The temperature 550 °C was used as Tmean. The rep-
dissociation to an alkyl group and a hydrogen atom as arametrization was done to reduce the correlation
two separate steps, their mechanism leads to a math- between the parameters. During preliminary modeling,
ematical expression similar to that obtained with mech- the adsorption/desorption equilibrium constants were
anism IV. described with an equation similar to eq 9. However,
this resulted in poor identifiability of the models.
Mechanism IV Therefore, the adsorption equilibrium constants were
i-C4H10(g) + Cr-O T i-C4H9‚ ‚ ‚Cr-O‚ ‚ ‚H (8a) assumed to be temperature-independent. Consequently,
the parameters are not real adsorption equilibrium
i-C4H9‚ ‚ ‚Cr-O‚ ‚ ‚H T constants but represent mean values in the temperature
range studied (520-580 °C). The equilibrium constant
i-C4H8(g) + H‚ ‚ ‚Cr-O‚ ‚ ‚H (8b)
of the dehydrogenation reaction was calculated from the
thermodynamic data in the Flowbat program26
H‚ ‚ ‚Cr-O‚ ‚ ‚H T H2(g) + Cr-O (8c)
122 kJ/mol
In mechanism IV, step 8a is the adsorption of isobu-
tane, and step 8b the surface reaction. Mechanisms III
K ) 2.1 × 107 exp - ( RT )
bar (10)

and IV both include the formation of Cr-H and O-H


bonds, in agreement with the results of FTIR studies Results and Discussion
by Busca,24 who detected Cr-H and O-H species on General Remarks. Mass Balance of Carbon. The
unsupported chromia during hydrogen adsorption at mass balance of carbon in the reactor system was
room temperature. The main difference between the two calculated from the molar amounts of hydrocarbons
mechanisms is the surface reaction: mechanism III measured by FTIR spectroscopy during dehydrogena-
requires two chromium ions and mechanism IV only one tion. The mass balance was generally 97-99%. Coke
ion. De Rossi and co-workers8-10 suggested the active deposition on the catalyst during dehydrogenation was
site to consist of one Cr3+ ion, with neighboring oxygen low, and the effect of coke on the total mass balance
ion(s) involved in the reaction. This might support a was less than 1%.
mechanism with one chromium ion. However, Cavani Thermal Reactions. The thermal reactions were stud-
et al.6 proposed amorphous chromia to be the most ied by replacing the catalyst with an equal amount of
active species, which might support a mechanism silicon carbide. Without the catalyst, the yield of
involving multiple chromium ions or else a mechanism isobutene was 0-3.7% in the studied temperature range
involving one chromium ion but with the activity of the of 520-580 °C, with the highest yields being obtained
single ion higher in an environment with several at high temperatures and low feed rates.
chromium ions nearby. Mass Transfer. The effect of external mass transfer
In addition, three other models presented in earlier was studied in dehydrogenation experiments done at
studies were tested: models 10 and 11 (Table 2), with 580 °C with a WHSV value of 15 h-1 with use of two
adsorption as the RDS, proposed by Zwahlen and different catalyst loadings (0.1 and 0.2 g). The isobutane
Agnew15 and Happel et al.,16 respectively, for the conversion was 31% with both loadings, and the selec-
dehydrogenation of isobutane on chromia/alumina, and tivities to isobutene were 93 and 96% with 0.1 and 0.2
model 12, with surface reaction on two different types g, respectively. The similar results indicated that ex-
of sites as the RDS, proposed by Suzuki and Kaneko19 ternal mass-transfer limitations were negligible.27 In-
for the dehydrogenation of propane on a potassium- ternal mass transfer was considered with the Weisz-
modified chromia/alumina catalyst. Prater criterion. The calculated value was about 10-7
Parameter Estimation. The parameters of the in magnitude (,1), which implied that internal mass-
kinetic equations were estimated from the results of 59 transfer effects were not significant.
dehydrogenation stages carried out with a hydrogen- Effect of Nitrogen Dilution. Noda et al.18 accounted
prereduced catalyst. The amounts of isobutane, isobutene, for nitrogen adsorption onto the catalyst surface in their
hydrogen, and nitrogen were used in the estimation. The kinetic investigation of n-butane dehydrogenation on
molar amounts of isobutane and isobutene were deter- chromia/alumina. Even though nitrogen was not ex-
mined from the results of FTIR analyses after 5 min on pected to interact with the oxide catalyst,28 the adsorp-
the isobutane stream, and the amount of hydrogen was tion effect of nitrogen was investigated in the present
assumed to equal that of isobutene. The data set did study by doing some experiments with argon instead of
not include the results of the stabilization or the nitrogen dilution. The experiments were carried out
standard dehydrogenation stages so that the effects of with a 0.2-g catalyst loading at 580 °C and a WHSV
deactivation and the overemphasis of certain reaction value of 15 h-1. The conversion of isobutane and the
selectivity to isobutene were 30 and 95%, respectively,
conditions could be ruled out. The parameters were
for both nitrogen and argon dilution. Thus, the nitrogen
estimated with the Kinfit program25 by minimizing the
adsorption parameter was not included in the kinetic
sum of squares of the residuals (SSR) between the
equations.
measured and calculated compositions of the product
Activity in the Dehydrogenation of Isobutane. To
stream. The kinetic model was combined with an determine whether to perform the kinetic measure-
integral plug-flow reactor model, and the minimization ments with a hydrogen-prereduced or a nonprereduced
was performed by the Levenberg-Marquardt method. catalyst, the effect of the reducing medium on the
The temperature dependence of the reaction rate product distribution was studied in the first 12 cycles
constant was described with an Arrhenius equation that of the series. A typical product distribution for the
was rewritten as nonprereduced catalyst is shown in Figure 1. During
the first few minutes on stream, carbon oxides and
k ) Ameane-zE (9) water were formed in the reduction of Cr6+ and Cr5+ to
Ind. Eng. Chem. Res., Vol. 41, No. 23, 2002 5623

Figure 3. Activity of the catalyst as a function of time on stream


Figure 1. Product distribution with a nonprereduced catalyst. after hydrogen prereduction. (Reaction conditions: 580 °C, 0.2 g,
15 h-1, and 50% i-C4H10/N2.)

Figure 2. Yield of isobutene at 10 min on the isobutane stream Figure 4. Results of selected kinetic activity measurements.
in the stabilization and standard dehydrogenation stages for the (Reaction conditions: 520-580 °C, 0.1 g, 15 or 30 h-1, and 50%
first series, as measured by GC. (Reaction conditions: 580 °C, 0.1 i-C4H10/N2.)
g, 15 h-1, and 50% i-C4H10/N2.)
Cr3+ with isobutane, and water was formed in the and 93 ( 1%, respectively, in the standard dehydroge-
prereduction with hydrogen. With the isobutane reduc- nation stages of the first series. The results were similar
tion, the amount of dehydrogenation products stabilized for the second and third series. In the kinetic activity
only after the formation of carbon oxides terminated, measurements, the conversion was 3.6-34%, and the
whereas with hydrogen prereduction, no carbon oxides selectivity to isobutene was 90-99%, depending on the
or water were formed during the dehydrogenation stage. reaction conditions. Figure 4 shows the conversion of
The kinetic experiments were, therefore, carried out isobutane at different temperatures with a catalyst
with a prereduced catalyst. The most significant side loading of 0.1 g and WHSV values of 15 and 30 h-1
reaction in all dehydrogenations was the cracking of C4 without hydrogen in the feed. At 550 °C, the thermo-
hydrocarbons to C1-C3 hydrocarbons, mostly methane dynamic equilibrium conversion is about 61% with the
and propene. In addition, a low concentration (<1%) of isobutane feed diluted with nitrogen at a molar ratio of
C4 hydrocarbons other than isobutane and isobutene 1:1.29
was detected. Addition of hydrogen to the isobutane feed decreased
The experimental data consisted of dehydrogenation the conversion and the selectivity. The main reason for
activities measured in three series. Figure 2 shows, as the lower selectivity was that the cracking reaction of
an example for the first series, the yield of isobutene at C4 hydrocarbons was mainly a thermal reaction. Thus,
10 min on the isobutane stream for the first 16 dehy- hydrogen affected only the yield of isobutene and not
drogenation stages and for the standard stages. Figure the yield of cracking products. The additional hydrogen
3 presents a typical example of the catalyst activity with in the gas phase might have decreased the conversion
time on the isobutane stream after hydrogen prereduc- by slowing the desorption of hydrogen from the surface,
tion. During the initial 16 cycles, the catalyst activity thereby decreasing the number of vacant surface sites.
decreased, most likely as a result of structural changes The effect might also have been purely kinetic: the
on the catalyst5 and possibly also accumulation of reaction rate decreased because of the increased hydro-
unburned coke on the surface. After the deactivation gen partial pressure. In any case, the results suggest
period, the catalyst activity stabilized as a function both that the hydrogen partial pressure is required in the
of the number of cycles and of the time on the isobutane kinetic model.
stream. With the stabilized catalyst, the conversion of The effect of the reducing medium on the yield of
isobutane and the selectivity to isobutene were 31 ( 1 isobutene can be seen from Figure 2 (cycles 1-12). The
5624 Ind. Eng. Chem. Res., Vol. 41, No. 23, 2002

Table 3. Results of the Parameter Estimation for the Kinetic Models


model parameter valuea model parameter valuea model parameter valuea
1 Ameanb 0.0313 ( 0.0003 5 (II-b) Ameanb 0.0160 ( 0.0139 9 (IV-b) Ameanb 0.0371 ( 0.0008
E (kJ/mol) 99 ( 5 E (kJ/mol) 147 ( 6 E (kJ/mol) 115 ( 4
SSR 218 K′′ (bar-1/2) 0.33 ( 0.09 KA (bar-1) 1 × 10-8 c
2 (I-a) Ameanb 0.0507 ( 0.0028 KE (bar-1) 4.02 ( 1.10 KH2 (bar-1) 2.75 ( 0.36
E (kJ/mol) 141 ( 3 K′′′ (bar-1) 0.33 ( 0.37 SSR 89
KEH2 (bar-2) 9.72 ( 12.89 KH2 (bar-1) 6 × 10-6 c 10 Ameanb 0.0383 ( 0.0010
KE (bar-1) 7.33 ( 2.03 SSR 85 E (kJ/mol) 133 ( 5
KH2 (bar-1) 3.87 ( 0.62 6 (III-a) Ameanb 0.0596 ( 0.0037 KEH2 (bar-2) 58.00 ( 8.09
SSR 27 E (kJ/mol) 141 ( 3 KE (bar-1) 6 × 10-8 c
3 (I-b) Ameanb 0.0540 ( 0.0046 KEH2 (bar-2) 8 × 10-6 c SSR 66
E (kJ/mol) 139 ( 3 K′ (bar-3/2) 22.90 ( 7.25 11 Ameanb 0.0524 ( 0.0021
KA (bar-1) 0.14 ( 0.10 KH2 (bar-1) 7.56 ( 2.19 E (kJ/mol) 142 ( 3
KE (bar-1) 3.14 ( 0.32 SSR 24 KE (bar-1) 8.76 ( 1.09
KH2 (bar-1) 1.56 ( 0.13 7 (III-b) Ameanb 0.0599 ( 0.0052 KH2 (bar-1) 4.28 ( 0.38
SSR 28 E (kJ/mol) 140 ( 3 SSR 27
4 (II-a) Ameanb 0.0581 ( 0.0035 KA (bar-1) 0.15 ( 0.10 12 Ameanb 0.0557 ( 0.0049
E (kJ/mol) 141 ( 3 K′ (bar-3/2) 7.56 ( 0.82 E (kJ/mol) 140 ( 3
K′ (bar-3/2) 2.69 ( 2.39 KH2 (bar-1) 0.93 ( 0.17 KH2 (bar-1) 3.55 ( 0.27
KE (bar-1) 1.70 ( 0.92 SSR 24 KA (bar-1) 0.35 ( 0.23
K′′′ (bar-1) 1.11 ( 0.25 8 (IV-a) Ameanb 0.0414 ( 0.0008 KE (bar-1) 6.37 ( 0.81
KH2 (bar-1) 1 × 10-6 c E (kJ/mol) 137 ( 4 SSR 27
SSR 24 KEH2 (bar-2) 51.17 ( 6.26
KH2 (bar-1) 1.95 ( 0.25
SSR 33
a ( 95% confidence limit. b Units of mol/(kgcat s bar). c 95% confidence limit very large.

conversion of isobutane was lower after prereduction of they might have lowered the activity by changing the
the catalyst with hydrogen than after its reduction with dispersion of the chromia phase.32 Hydrogen addition
the isobutane feed; the selectivity to isobutene was the to the feed, in turn, might have decreased the reaction
same. The observation that hydrogen prereduction rate purely through the influence of the increased
lowers the alkane conversion on chromia/alumina is in partial pressure on the catalyst surface.
accordance with the results of Hakuli30 and Gorriz et Kinetic Modeling. Comparison of the Models. The
al.31 but contrary to those of De Rossi et al.9 In general, 12 models presented in Table 2 were tested for the
the reduction of the chromia phase liberates water in measured dehydrogenation data, and the results of the
hydrogen prereduction and carbon oxides and water in parameter estimation are presented at Table 3. The sum
isobutane reduction according to eqs 11 and 12, respec- of squares of the residuals (SSR) was significantly
tively higher with the power-law model (SSR 218) than with
the models assuming some adsorption effect (SSR 24-
H2 + O2- f H2O + 2e- (11) 89). Thus, the least accurate model was the power-law
equation. Models II-b, IV-b, and 10 were also rejected,
as their SSRs were in the range 66-89.
i-C4H10 + (4x + 5)O2- f The lowest SSR value of 24 was obtained with models
4COx + 5H2O + (8x + 10)e- (12) II-a, III-a, and III-b. For model III-a, which assumed
adsorption as the RDS, the value of the isobutane
In the present study, the amount of water detected adsorption term KEH2pEpH2 was near 0. Thus, the model
by FTIR spectroscopy during hydrogen prereduction and could be simplified to eq 13 by omitting this term from
isobutane reduction was lower than anticipated from the model. The parameter values were not significantly
the amount of carbon oxides released in the hydrocarbon changed from those presented in Table 3.

( )
reduction. Some water must have remained on the
catalyst, as proposed by Hakuli et al.4 and De Rossi et pEpH2
al.9,10 The water remaining on the catalyst, in the form k′ pA -
K
of OH/H groups for example, has been suggested to -r′A ) 1/2
(13)
modify the active sites,30 possibly by influencing their 1 + K′pEpH2 + (KH2pH2)1/2
coordinative unsaturation. Moreover, De Rossi et al.10
observed that addition of water to the alkane feed The additional parameters and the squared denomi-
decreased the dehydrogenation activity of chromia/ nators in models II-a and III-b compared to eq 13 did
zirconia catalysts. They suggested that the water ad- not decrease the SSR. The similar descriptions of the
sorbed on the surface and saturated a fraction of the three models can be explained by the estimation results,
active sites. Here, saturation of some of the active sites as the values for the reaction rate constant factors Amean
might explain the lower activity measured after hydro- and E are similar for the three models (Table 3).
gen prereduction compared to that measured after Therefore, the values of the denominators should also
hydrocarbon reduction. It is questionable, however, be similar, indicating that the parameter values were
whether the lower activity observed after hydrogen determined so that they compensated for the modifica-
prereduction and the lower activity we observed upon tions in the form of the equation. Thus, the three models
hydrogen addition to the feed were due to the same can be expressed with the simplest of them, eq 13.
cause, for example, the decrease in the number of vacant Although the best modeling results were obtained
surface sites. The oxygen-containing species formed in with eq 13, models I-a, I-b, IV-a, 11, and 12 (Table 2)
the prereduction might have had an additional effect: were also satisfactory (SSR 27-33). These five models
in addition to influencing the coordinative unsaturation, derive from quite similar mechanisms with adsorption
Ind. Eng. Chem. Res., Vol. 41, No. 23, 2002 5625

of the alkane has been suggested to determine the rate


of isobutane dehydrogenation on supported Pt/Sn cata-
lysts33 and the molecular adsorption of the alkane the
rate of propane dehydrogenation on silicalite-supported
zinc oxide.34
The activation energy of the reaction rate-determining
step was estimated to be 133-142 and 115-147 kJ/mol
with models assuming, respectively, adsorption and
surface reaction as the RDS. Lower values than these
have been estimated before for isobutane dehydrogena-
tion on chromia/alumina with models assuming alkane
adsorption as the RDS: Zwahlen and Agnew15 esti-
mated a value of 112 kJ/mol with model 10 and Happel
et al.16 a value of 73 kJ/mol with model 11. Both models
10 and 11 are simplified forms of model I-a (Table 2),
with model 10 leaving out the KH2pH2 term of model I-a
and model 11 the KEH2pEpH2 term. Zwahlen and Agnew
Figure 5. Correlation between weight fractions in the product modeled dehydrogenation and deactivation of the cata-
stream estimated with model 11 and eq 13 and the measured lyst simultaneously, which increased the number of
weight fractions. parameters to be fitted and might have reduced the
accuracy of the estimates. In addition, the values for
on one site, S or Cr-O, and desorption of alkene and the reaction equilibrium constants were different in the
hydrogen also from single sites. When comparing the earlier studies from the values used in our study. These
form and the parameter estimation results of the differences might explain the higher values obtained
equations that adequately described the reaction, cer- here for the activation energy.
tain generalizations can be made. First, the adsorption Comparison of the Mechanisms. The modeling results
term of isobutane was either poorly determined, or its support mechanisms I and IIIsboth assuming adsorp-
value was near zero. Thus, it could be omitted from the tion on single site and surface reaction on dual sitess
equations without increasing the SSR or changing the approximately equally. Mechanism II, with adsorption
values of the other parameters significantly. Second, on dual sites, seems less probable as the results suggest
both the isobutene and hydrogen adsorption terms were that the adsorption might require only one site. The
required in the equations for a satisfactory description weakness of mechanism I, however, is the unlikely
of the reaction rate. The effect of hydrogen was also clear abstraction of the hydrogen molecule from the alkane
from the activity results: addition of hydrogen to the in a single step; the weakness of mechanisms II and III
feed lowered the conversion. In addition, the results of is that they require three and two similar active sites.
model II-a (Table 3) indicated that the hydrogen atom Mechanism IV, suggested by Weckhuysen and
combination and desorption could be treated as a single Schoonheydt3 and by Burwell et al.,22 would seem to be
step. Third, squaring the denominator, corresponding the most realistic on the basis of the conclusions of De
to an RDS with two similar active sites, only adjusted Rossi et al.8-10 that the reaction requires only one
the parameter values but did not decrease the SSR. The
chromium ion. However, our modeling results do not
simplest model studied to incorporate only the isobutene
support mechanism IV as strongly as the other mech-
and hydrogen adsorption terms was model 11, with
anisms. It might be beneficial to treat the chromium
adsorption as the RDS, suggested earlier by Happel et
and oxygen ions as separate SCr and SO sites in the
al.16 Figure 5 shows the weight fractions of isobutane,
model derivation rather than as a Cr-O pair. However,
isobutene, and hydrogen in the outlet stream as esti-
if the adsorption/dissociation of alkane is the slow step
mated with eq 13 and model 11 versus the measured
in the reaction, the other steps are more difficult to
weight fractions. Both equations can be used to describe
elucidate with kinetic modeling of reaction rate data.
the isobutane dehydrogenation reaction with good ac-
Isotopic tracer studies might also prove helpful in
curacy.
deciding among the different mechanisms, and such an
The finding that the isobutane adsorption term was
approach is under study.
not required in the equations for a satisfactory descrip-
tion of the reaction rate supports a model with adsorp-
tion of isobutane as the slow step; the two proposed Conclusions
models, eq 13 and model 11, both assume this RDS. In
addition, the results suggest that the adsorption re- The kinetics of isobutane dehydrogenation was stud-
quires only one site because squaring the denominator ied on a chromia/alumina catalyst developed for fluid-
did not improve the description. This agrees with the ized-bed operation. The catalyst used was active and
idea that the adsorption can occur dissociatively on the selective in isobutane dehydrogenation. The activity was
suggested3,21,22 active site, a Cr-O pair. If the surface lower after prereduction of the catalyst with hydrogen
reaction of the adsorbed species to the adsorbed alkene than after reduction with the isobutane feed. The lower
were the slow step, the amount of adsorbed isobutane activity might have been caused by the modification of
or adsorbed i-C4H9- would be significant and, thus, the active sitesspresumably coordinatively unsaturated
necessary in the equation. Our speculation for the rate- Cr-O pairssby the water formed during prereduction
determining step is in accordance with models suggested and subsequently retained on the catalyst. The results
by Zwahlen and Agnew15 and Happel et al.16 for isobu- of the kinetic modeling suggest that the rate-determin-
tane dehydrogenation on chromia/alumina but contrary ing step is isobutane adsorption, possibly with dissocia-
to the proposal of Kao et al.20 that the surface reaction tion on a Cr-O pair. The isobutene and hydrogen
of the adsorbed alkyl group is the RDS in isobutane adsorption parameters were required in the kinetic
dehydrogenation on chromia gel. Dissociative adsorption equation for a satisfactory description of the reaction
5626 Ind. Eng. Chem. Res., Vol. 41, No. 23, 2002

rate by the model. A value of 133-142 kJ/mol was (10) De Rossi, S.; Ferraris, G.; Fremiotti, S.; Cimino, A.;
calculated for the reaction activation energy. Indovina, V. Propane Dehydrogenation on Chromia/Zirconia Cata-
lysts. Appl. Catal. A 1992, 81, 113.
(11) Dodd, R. H.; Watson, K. M. Process Design of Catalytic
Acknowledgment ReactorssDehydrogenation of Butane. Trans. Inst. Chem. Eng.
1946, 42, 263.
Financial support from the Academy of Finland is (12) Carrà, S.; Forni, L.; Vintani, C. Kinetics and Mechanism
gratefully acknowledged. We thank Ms. Jaana Kanervo in Catalytic Dehydrogenation of n-Butane over Chromia-Alumina.
and Dr. Juha Linnekoski for valuable discussions and J. Catal. 1967, 9, 154.
help with the modeling. (13) Happel, J.; Atkins, R. S. Rate-Controlling Path for Cata-
lytic Butane Dehydrogenation. Ind. Eng. Chem. Fundam. 1970,
9, 11.
Nomenclature (14) Happel, J.; Hnatow, M. A.; Mezaki, R. Rate Modeling for
the Butane-Butenes System. Adv. Chem. Ser. 1970, 97, 92.
A0 ) preexponential factor, mol/(kgcat s bar) (15) Zwahlen, A. G.; Agnew, J. B. Isobutane Dehydrogenation
Cn ) molar amount of compound containing n carbon Kinetics Determination in a Modified Berty Gradientless Reactor.
atoms, including hydrocarbons and carbon oxides Ind. Eng. Chem. Res. 1992, 31, 2088.
E ) activation energy, J/mol (16) Happel, J.; Kamholz, K.; Walsh, D.; Strangio, V. Kinetics
k′ ) (apparent) rate constant, mol/(kgcat s bar) of the Isobutane-Isobutene-Hydrogen System Using Tracers. Ind.
K ) reaction equilibrium constant, bar Eng. Chem. Fundam. 1973, 12, 263.
(17) Jackson, S. D.; Grenfell, J.; Matheson, I. M.; Webb, G.
Ki ) adsorption equilibrium constant of component i, bar-1 Modelling of Alkane Dehydrogenation under Transient and Steady-
KEH2, K′, K′′, K′′′ ) combined equilibrium constants State Conditions over a Chromia Catalyst Using Isotopic Labeling.
pi ) partial pressure of component i, bar Stud. Surf. Sci. Catal. 1999, 122, 149.
ri′ ) reaction rate of component i, mol/(kgcat s) (18) Noda, S.; Hudgins, R. R.; Silveston, P. L. A Kinetic Study
R ) 8.314 J/(mol K) of Catalytic Butane Dehydrogenation. Can. J. Chem. Eng. 1967,
RDS ) rate-determining step 45, 294.
(19) Suzuki, I.; Kaneko, Y. Dehydrogenation of Propane over
SSR ) sum of squares of the residuals Chromia-Alumina-Potassium Oxide Catalyst. J. Catal. 1977, 47,
S ) adsorption site on the catalyst 239.
Si ) selectivity to component i (20) Kao, J.-Y.; Piet-Lahanier, H.; Walter, E.; Happel, J.
T ) temperature, K or °C Reversible Reactions over Non-Metallic Catalysts: A Transient
X ) conversion of isobutane Isotopic Tracing on the Isobutane-Isobutene-Hydrogen System
over Chromia. J. Catal. 1992, 133, 383.
Subscripts (21) Resasco, D. E.; Haller, G. L. Catalytic Dehydrogenation
of Lower Alkanes. Catalysis 1994, 11, 379.
A ) isobutane (22) Burwell, R. L., Jr.; Littlewood, A. B.; Cardew, M.; Pass,
E ) isobutene G.; Stoddart, C. T. H. Reaction between Hydrocarbons and
A,in ) isobutane in the feed Deuterium on Chromium Oxide Gel. I. General. J. Am. Chem. Soc.
A,out ) isobutane in the product stream 1960, 82, 6272.
(23) Hakuli, A.; Kytökivi, A.; Lakomaa, E.-L.; Krause, O. FT-
IR in the Quantitative Analysis of Hydrocarbon Mixtures. Anal.
Literature Cited Chem. 1995, 67, 1881.
(24) Busca, G. Fourier Transform-Infrared Spectroscopic Study
(1) Bhasin, M. M.; McCain, J. H.; Vora, B. V.; Imai, T.; Pujadó,
of the Adsorption of Hydrogen on Chromia and on Some Metal
P. R. Dehydrogenation and Oxydehydrogenation of Paraffins to
Chromites. J. Catal. 1989, 120, 303.
Olefins. Appl. Catal. A 2001, 221, 397.
(25) Aittamaa, J.; Keskinen, K. I. Kinfit; Laboratory of Chemi-
(2) Weckhuysen B. M.; Wachs, I. E.; Schoonheydt, R. A. Surface cal Engineering, Helsinki University of Technology, Espoo,
Chemistry and Spectroscopy of Chromium in Inorganic Oxides. Finland, 2001.
Chem. Rev. 1996, 96, 3327. (26) Aittamaa, J.; Keskinen, K. I. FLOWBAT User’s Instruction
(3) Weckhuysen, B. M.; Schoonheydt, R. A. Alkane Dehydro- Manual; Neste Oy: Porvoo, Finland, 1999.
genation over Supported Chromium Oxide Catalysts. Catal. Today (27) Fogler, H. S. Elements of Chemical Reaction Engineering;
1999, 51, 223. Prentice Hall International: New York, 1992.
(4) Hakuli, A.; Kytökivi, A.; Krause, A. O. I.; Suntola, T. Initial (28) van Santen, R. A.; Niemantsverdriet, J. W. Chemical
Activity of Reduced Chromia/Alumina Catalyst in n-Butane De- Kinetics and Catalysis; Plenum Press: New York, 1995.
hydrogenation Monitored by On-Line FT-IR Gas Analysis. J. (29) Calculated with the FLOWBAT program, Neste Oy,
Catal. 1996, 161, 393. Porvoo, Finland, 1999.
(5) Hakuli, A.; Kytökivi, A.; Krause, A. O. I. Dehydrogenation (30) Hakuli, A. Ph.D. Thesis, Helsinki University of Technology,
of i-Butane on CrOx/Al2O3 Catalysts Prepared by ALE and Espoo, Finland, 1999.
Impregnation Techniques. Appl. Catal. A 2000, 190, 219. (31) Gorriz, O. F.; Cortés Corberán, V.; Fierro, J. L. G. Propane
(6) Cavani, F.; Koutyrev, M.; Trifirò, F.; Bartolini, A.; Ghisletti, Dehydrogenation and Coke Formation on Chromia-Alumina
D.; Iezzi, R.; Santucci, A.; Del Piero, G. Chemical and Physical Catalysts: Effect of Reductive Pretreatments. Ind. Eng. Chem.
Characterization of Alumina-Supported Chromia-Based Catalysts Res. 1992, 31, 2670.
and Their Activity in Dehydrogenation of Isobutane. J. Catal. (32) Liu, B.; Terano, M. Investigation of the Physico-Chemical
1996, 158, 236. State and Aggregation Mechanism of Surface Cr Species on a
(7) Weckhuysen, B. M.; Bensalem, A.; Schoonheydt, R. A. In Phillips CrOx/SiO2 Catalyst by XPS and EPMA. J. Mol. Catal. A
Situ UV-Vis Diffuse Reflectance Spectroscopy-On-line Activity 2001, 172, 227.
Measurements. Significance of Crn+ species (n ) 2, 3 and 6) in (33) Cortright, R. D.; Hill, J. M.; Dumesic, J. A. Selective
n-Butane Dehydrogenation Catalyzed by Supported Chromium Dehydrogenation of Isobutane over Supported Pt/Sn Catalysts.
Oxide Catalysts. J. Chem. Soc., Faraday Trans. 1998, 94, 2011. Catal. Today 2000, 55, 213.
(8) De Rossi, S.; Casaletto, M. P.; Ferraris, G.; Cimino, A.; (34) Wan, B.-Z.; Chu, H. M. Reaction Kinetics of Propane
Minelli, G. Chromia/Zirconia Catalysts with Cr Content Exceeding Dehydrogenation over Partially Reduced Zinc Oxide Supported on
the Monolayer. A Comparison with Chromia/Alumina and Chro- Silicalite. J. Chem. Soc., Faraday Trans. 1992, 88, 2943.
mia/Silica for Isobutane Dehydrogenation. Appl. Catal. A 1998,
167, 257. Received for review May 20, 2002
(9) De Rossi, S.; Ferraris, G.; Fremiotti, S.; Garrone, E.; Ghiotti, Revised manuscript received September 11, 2002
G.; Campa, M. C.; Indovina, V. Propane Dehydrogenation on Accepted September 13, 2002
Chromia/Silica and Chromia/Alumina Catalysts. J. Catal. 1994,
148, 36. IE020371J

Potrebbero piacerti anche