Applied Catalysis A: General 270 (2004) 37–47

Kinetics of the oxidative dehydrogenation of propane on vanadia/titania

catalysts, pure and doped with rubidium
R. Grabowski∗
Institute of Catalysis and Surface Chemistry, Polish Academy of Sciences, 30-239 Krakow, Poland
Received in revised form 29 March 2004; accepted 7 April 2004

Available online 4 June 2004

Abstract

Oxidative dehydrogenation (ODH) of propane over VTiO and VRbTiO catalysts has been studied by steady-state kinetic method in the fixed
bed flow reactor. The experiments were carried out by measuring conversion and selectivities for various feed compositions including C3 H8 ,
O2 and He, and for various contact times and temperatures. The experimental results obtained for both catalyst were described by a simple
sequence of steps. Two kinetic mechanistic models: Langmuir–Hinshelwood (LH) and Eley–Rideal (ER)-Steady State Adsorption Model
(SSAM) were tried and the analysis of the calculation results leads to the conclusion that Eley–Rideal-SSAM model is more appropriate
for the description of the propane ODH reaction for both studied catalysts. In the frame of this latter model the role and the influence of
Rb-additive on different steps of propane ODH reaction was discussed.
© 2004 Elsevier B.V. All rights reserved.

Keywords: Propane oxidative dehydrogenation; Kinetics models; Vanadia–titania catalyst

1. Introduction too simplified. Though it is well known that the kinetic stud-
ies are not able to determine the molecular mechanism of the
Kinetic models, which provide detailed quantitative de- reaction, they allow to exclude some of the possible paths
scription of the reaction systems, play an important role in and to determine the reaction network. Therefore, the kinetic
catalysis R&D. Such models are applied for the develop- studies have been used to provide useful information about
ment and optimisation of the industrial catalytic processes. the studied reaction and the catalyst operation. On the basis
Kinetic modeling of the reactions have been successfully of these studies, one can identify the intermediate products,
used to generate new quantitative data on the reaction mech- the nature and quantitative participation of particular reac-
anisms and the role the different catalytic sites play in these tion routes (parallel and consecutive reaction paths, branch-
mechanisms [1,2]. In the latter case, in order to obtain reli- ing of the reactions, etc.) which decide about the selectivity
able information, it is essential to combine kinetic modeling of the process. One can also identify the rate determining
with the experimental reaction studies and catalyst charac- step in a sequence of consecutive reactions.
terization. So far the Mars–van-Krevelen (MK) or steady state ad-
Kinetic study of the oxidative dehydrogenation (ODH) of sorption (SSAM) models have been used, connected or not
alkanes is one of the ways to elucidate the reaction mech- with the Eley–Rideal (ER) or Langmuir–Hinshelwood (LH)
anism and to facilitate selection of the appropriate catalyst mechanisms. The calculations were done for the conversions
for this type of the reactions. Previous studies [3–13] con- of propane which varied in comparatively small ranges. The
cerning the ODH of propane over oxide catalysts, have not aim of this work was comparison of the applicability of the
explained entirely the mechanism of the reaction and some models and the determination of rate constants for particu-
controversies in determination of macroscopic steps of the lar steps of the ODH reaction for a wider range of the ex-
reaction still exist—the models applied to this reaction were perimental data such as conversion of propane and the se-
lectivities to propene and COx measured as a function of
∗ Tel.: +48 12 6395160; fax: +48 12 4251923. contact time, temperature and the initial concentration of the
E-mail address: ncgrabow@[Link] (R. Grabowski). reagents.

0926-860X/$ – see front matter © 2004 Elsevier B.V. All rights reserved.
doi:10.1016/[Link].2004.04.021
38 R. Grabowski / Applied Catalysis A: General 270 (2004) 37–47

coupled directly to a series of gas chromatographs. Propene,

Nomenclature carbon monoxide and dioxide were found to be the main
c concentration (mole fraction) reaction products. Details of the experimental set-up can be
dc/dτ rate of the reaction (s−1 ) found in [8]. The flow rate of the reaction mixture was con-
E activation energy (kJ/mol) trolled with a mass flow meter type ERG-1MKZ. 0.4–0.5 g
k rate constant (s−1 ) samples of a catalyst of the grain size of 0.6–1 mm, diluted
K equilibrium constant with glass beads, were used. It has been checked in the pre-
SSlf sum of squares for lack of fit liminary experiments, in which the mass and the grain size
SSpe sum of squares for pure error of a catalyst sample were varied, that under these conditions
the transport phenomena do not limit the reaction rate. The
Greek letters reaction mixture contained propane, oxygen and helium as
␲ alkene (propene) a balance gas. The kinetic measurements were performed
θ surface coverage (dimensionless) by measuring conversion of propane (X) and selectivities to
τ contact time (s) propene and CO, CO2 (Sj ) at three temperatures as a func-
tion of contact time and composition of the reaction mix-
Subscripts ture. The results of the kinetic experiments for the 1VTiO
1, 2, . . . , OS particular steps of the reaction and 1VRbTiO catalysts are collected in Tables 1 and 2, re-
(e.g. k1 , k2 , . . . , kOS ) spectively. Conversion of propane and a selectivity to
an
A adsorption ith product are defined as: X = (ci −cf )/ci , Sj = cj / cj ,
(COx ) CO + CO2 where ci and cf are the concentrations of the respective
D desorption alkane at the entrance and the exit of the reactor, respec-
(g) gas tively, and cj is the concentration of the product j in the exit
O oxygen gas. For each temperature four different compositions of the
R alkane (propane) reaction mixture—with the propane/oxygen ratio = 2/1, 2/3,
s surface 1/1 and 1/3 were used. The contact time τ was varied be-
tween 0.1 and 3 s by changing the flow rate of the reaction
mixture.

2. Experimental 2.3. Choice and description of the applied kinetic models

2.1. Catalysts The experimental data used for calculations were obtained
for the 1VTiO and 1VRbTiO catalysts in the integral re-
The studies comprised two catalysts, denoted further in actor (0 < X < 0.7), which made possible the calculation
the text as 1VTiO and 1VRbTiO. Both catalysts contained of the particular rate constants using a broad range of the
one theoretical monolayer (mnl) of vanadia phase deposited independent variables (concentrations of propane, tempera-
on TiO2 -anatase (commercial support, product of Tioxide ture of the reaction and contact time). The data obtained in
Ltd., 48 m2 /g). [7,9] for the 1VTiO catalysts, pure and promoted with al-
One-monolayer of the VOx phase, calculated from crystal- kali metals, have shown that the increase in the conversion
lographic data of V2 O5 as 10 V/nm2 , corresponded to 6.7% of propane with the increasing temperature is accompanied
V2 O5 for a given specific surface area of TiO2 . The cata- by a considerable decrease in the selectivity to propene S␲
lyst was prepared by impregnation of TiO2 with the aque- and, at the same time, a parallel increase in the selectiv-
ous solution of ammonium methavanadate at pH 6, followed ity to CO, SCO , and a smaller increase to CO2 . Carbon ox-
by solvent evaporation, drying for 18 h at 120 ◦ C and cal- ides (COx = CO + CO2 ) are, however, observed in some
cination in a stream of air at 450 ◦ C for 5 h. The promoted amounts also at low conversions. Such a course of changes
1VRbTiO was obtained by adding the appropriate amount of implies a parallel-consecutive mechanism of the formation
rubidium nitrate to the methavanadate solution. The specific of carbon oxides from propane, a behaviour typical of selec-
surface areas of the samples, after the calcination, were 44.4 tive oxidation reactions of hydrocarbons on oxide catalysts.
and 42.7 m2 /g for 1VTiO and 1VRbTiO, respectively. The The different aspects of the mechanism and kinetics of
detailed description of the preparation method was given in the ODH of propane were studied for different mixed ox-
[4,8]. ide systems like V2 O5 /AlPO4 [3], V2 O5 /ZrO2 V2 O5 /TiO2
[11] and Mg–V–Sb–O [13,14]. On the basis of the ki-
2.2. Catalytic activity measurements netic isotopic effect it has been shown that activation of
methylene—C–H bond is the rate determining step for the
The steady-state activity of the catalyst in the oxidative propane dehydrogenation [10]. The kinetic studies of the
dehydrogenation was measured in a fixed bed flow appara- ODH of propane on the Mg–V–Sb–O catalyst have indi-
tus. A stainless steel reactor (120 mm long, i.d. 13 mm) was cated that nucleophyllic oxygen of the catalyst is involved
 R. Grabowski / Applied Catalysis A: General 270 (2004) 37–47 39

Table 1
Experimental and calculated data for ER-SSAM-2 model for the 1VTiO catalyst
Contact Temperature Propane/O2 Experimental Calculated Experimental Calculated Experimental Calculated
time (s) (K) ratio conversion of conversion of selectivity to selectivity to selectivity to selectivity
propane propane propene propene COx to COx
0.1 498 1/1 0.012 0.013 0.620 0.569 0.42 0.431
0.25 0.029 0.028 0.580 0.454 0.536 0.546
0.50 0.048 0.048 0.464 0.338 0.650 0.662
1.0 0.077 0.085 0.350 0.217 0.774 0.783
1.2 0.093 0.098 0.226 0.189 0.812 0.811
1.5 0.109 0.117 0.154 0.157 0.846 0.843
0.1 2/1 0.009 0.009 0.590 0.612 0.410 0.388
0.3 0.026 0.022 0.507 0.500 0.493 0.500
0.5 0.037 0.032 0.432 0.426 0.568 0.574
1.0 0.057 0.054 0.324 0.313 0.676 0.687
1.2 0.063 0.062 0.294 0.283 0.706 0.717
1.5 0.068 0.073 0.263 0.247 0.737 0.753
0.1 1/3 0.016 0.017 0.510 0.536 0.490 0.464
0.25 0.039 0.036 0.366 0.400 0.634 0.600
0.50 0.063 0.065 0.261 0.273 0.739 0.727
1.0 0.099 0.117 0.166 0.157 0.834 0.843
1.2 0.112 0.137 0.142 0.132 0.858 0.868
1.5 0.165 0.167 0.108 0.106 0.892 0.894
0.1 2/3 0.010 0.012 0.560 0.579 0.440 0.421
0.3 0.030 0.030 0.433 0.440 0.567 0.560
0.5 0.045 0.045 0.370 0.354 0.630 0.646
1.0 0.066 0.079 0.263 0.232 0.737 0.768
1.2 0.074 0.091 0.208 0.202 0.792 0.798
1.5 0.102 0.110 0.180 0.168 0.820 0.832
0.1 523 1/1 0.024 0.022 0.420 0.436 0.580 0.564
0.25 0.061 0.049 0.310 0.318 0.690 0.682
0.50 0.104 0.087 0.194 0.213 0.806 0.787
1.0 0.168 0.152 0.111 0.122 0.889 0.878
1.2 0.187 0.176 0.093 0.104 0.907 0.896
1.5 0.212 0.207 0.075 0.085 0.925 0.915
0.1 2/1 0.017 0.015 0.470 0.477 0.530 0.523
0.3 0.052 0.038 0.359 0.361 0.641 0.639
0.5 0.068 0.057 0.318 0.290 0.648 0.710
1.0 0.106 0.096 0.205 0.195 0.795 0.805
1.2 0.106 0.109 0.183 0.174 0.817 0.826
0.1 1/3 0.027 0.028 0.380 0.406 0.620 0.594
0.25 0.069 0.063 0.264 0.271 0.736 0.729
0.50 0.120 0.116 0.158 0.163 0.842 0.837
1.0 0.196 0.213 0.083 0.082 0.917 0.918
1.2 0.226 0.249 0.067 0.067 0.933 0.933
1.5 0.293 0.301 0.052 0.052 0.948 0.948
0.1 2/3 0.021 0.021 0.430 0.444 0.570 0.556
0.3 0.061 0.053 0.307 0.303 0.693 0.697
0.5 0.088 0.081 0.235 0.224 0.765 0.776
1.0 0.148 0.145 0.130 0.129 0.870 0.871
1.2 0.161 0.169 0.114 0.109 0.886 0.891
1.5 0.197 0.203 0.094 0.087 0.906 0.913
0.1 558 1/1 0.052 0.047 0.280 0.277 0.720 0.723
0.25 0.129 0.104 0.185 0.175 0.815 0.825
0.50 0.205 0.184 0.095 0.103 0.905 0.897
1.0 0.265 0.288 0.065 0.059 0.935 0.941
0.1 2/1 0.03 0.032 0.320 0.317 0.680 0.683
0.3 0.106 0.079 0.226 0.211 0.774 0.789
0.5 0.127 0.116 0.179 0.161 0.821 0.839
40 R. Grabowski / Applied Catalysis A: General 270 (2004) 37–47

Table 1 (Continued )
Contact Temperature Propane/O2 Experimental Calculated Experimental Calculated Experimental Calculated
time (s) (K) ratio conversion of conversion of selectivity to selectivity to selectivity to selectivity
propane propane propene propene COx to COx
0.1 1/3 0.059 0.059 0.250 0.250 0.750 0.750
0.25 0.148 0.136 0.146 0.140 0.854 0.860
0.50 0.257 0.250 0.070 0.071 0.930 0.929
1.0 0.381 0.440 0.033 0.030 0.967 0.970
1.2 0.426 0.504 0.025 0.024 0.975 0.976
1.5 0.553 0.587 0.019 0.017 0.981 0.983
0.1 2/3 0.051 0.044 0.280 0.279 0.720 0.721
0.3 0.128 0.116 0.164 0.161 0.836 0.839
0.5 0.185 0.178 0.101 0.107 0.899 0.893
1.0 0.281 0.307 0.048 0.054 0.952 0.946
1.2 0.302 0.348 0.039 0.045 0.961 0.955
1.5 0.369 0.398 0.034 0.036 0.966 0.964

Table 2
Experimental and calculated data for ER-SSAM-2 model for the 1VRbTiO catalyst
Contact Temperature Propane/O2 Experimental Calculated Experimental Calculated Experimental Calculated
time (s) (K) ratio conversion of conversion of selectivity to selectivity to selectivity to selectivity to
propane propane propene propene COx COx
0.1 603 1/1 0.030 0.026 0.624 0.623 0.376 0.377
0.25 0.086 0.066 0.525 0.512 0.475 0.488
0.5 0.140 0.119 0.385 0.384 0.615 0.616
1.0 0.219 0.219 0.240 0.240 0.760 0.760
1.2 0.222 0.255 0.232 0.206 0.768 0.794
0.1 2/1 0.022 0.025 0.651 0.644 0.349 0.356
0.3 0.072 0.070 0.545 0.508 0.455 0.492
0.5 0.103 0.111 0.413 0.419 0.587 0.581
1.0 0.145 0.192 0.267 0.311 0.733 0.689
1.2 0.147 0.215 0.289 0.295 0.711 0.705
0.1 1/3 0.028 0.027 0.620 0.618 0.380 0.382
0.25 0.059 0.064 0.572 0.503 0.428 0.497
0.5 0.114 0.123 0.384 0.369 0.616 0.631
1.0 0.194 0.228 0.232 0.221 0.768 0.779
1.2 0.224 0.267 0.193 0.185 0.807 0.815
1.5 0.267 0.321 0.157 0.146 0.843 0.854
3.0 0.436 0.539 0.077 0.060 0.923 0.940
0.1 2/3 0.030 0.026 0.625 0.623 0.375 0.377
0.3 0.074 0.074 0.486 0.486 0.514 0.514
0.5 0.116 0.118 0.371 0.387 0.629 0.613
1.0 0.197 0.218 0.224 0.241 0.776 0.759
1.2 0.233 0.255 0.174 0.205 0.826 0.795
1.5 0.291 0.306 0.144 0.165 0.856 0.835
3.0 0.487 0.496 0.077 0.087 0.923 0.913
0.1 633 1/1 0.035 0.035 0.550 0.553 0.450 0.447
0.25 0.093 0.085 0.460 0.426 0.540 0.574
0.5 0.172 0.160 0.276 0.293 0.724 0.707
1.0 0.276 0.289 0.185 0.166 0.815 0.834
0.1 2/1 0.038 0.034 0.580 0.567 0.420 0.433
0.3 0.108 0.118 0.378 0.415 0.622 0.585
0.5 0.150 0.151 0.325 0.325 0.675 0.675
1.0 0.245 0.250 0.305 0.222 0.695 0.778
0.1 1/3 0.038 0.036 0.550 0.543 0.450 0.457
0.25 0.118 0.086 0.405 0.416 0.595 0.584
0.5 0.182 0.164 0.285 0.281 0.715 0.719
1.0 0.330 0.300 0.152 0.150 0.848 0.850
1.2 0.402 0.348 0.111 0.122 0.889 0.878
1.5 0.455 0.414 0.089 0.093 0.911 0.907
3.0 0.645 0.657 0.042 0.035 0.958 0.965
 R. Grabowski / Applied Catalysis A: General 270 (2004) 37–47 41

Table 2 (Continued )
Contact Temperature Propane/O2 Experimental Calculated Experimental Calculated Experimental Calculated
time (s) (K) ratio conversion of conversion of selectivity to selectivity to selectivity to selectivity to
propane propane propene propene COx COx
0.1 2/3 0.037 0.035 0.560 0.553 0.440 0.447
0.3 0.122 0.100 0.370 0.370 0.630 0.630
0.5 0.193 0.159 0.254 0.265 0.746 0.735
1.0 0.301 0.289 0.136 0.157 0.864 0.843
1.2 0.340 0.335 0.103 0.135 0.897 0.865
0.1 1/1 0.054 0.052 0.480 0.466 0.520 0.534
0.25 0.131 0.122 0.272 0.323 0.728 0.677
0.5 0.234 0.228 0.203 0.198 0.797 0.802
0.1 0.052 0.052 0.445 0.452 0.555 0.548
0.25 673 2/1 0.210 0.147 0.296 0.282 0.704 0.718
0.5 0.275 0.253 0.174 0.173 0.826 0.827
1.0 0.442 0.426 0.090 0.083 0.910 0.917
1.2 0.489 0.483 0.064 0.066 0.936 0.934
1.5 0.578 0.559 0.049 0.049 0.951 0.951
3.0 0.773 0.801 0.017 0.015 0.983 0.985
0.1 1/3 0.053 0.051 0.450 0.454 0.550 0.546
0.25 0.212 0.144 0.249 0.290 0.751 0.710
0.5 0.242 0.227 0.159 0.197 0.841 0.803
1.0 0.368 0.399 0.100 0.097 0.900 0.903
0.1 2/3 0.049 0.052 0.470 0.462 0.530 0.538
0.3 0.186 0.145 0.312 0.289 0.688 0.711
0.5 0.271 0.229 0.199 0.195 0.801 0.805
1.0 0.392 0.403 0.099 0.093 0.901 0.907
1.2 0.457 0.461 0.082 0.074 0.918 0.926

in the formation of propene and adsorbed electrophyllic gen concentration. It is, however, an approximation, because
oxygen is responsible for the total oxidation of propane and it is evident from the experimental data that conversion of
propene. propane and selectivities depend to some extent on oxygen
Similar studies were done for the V–Mg–O catalyst concentration. The power model (PM) [9] was also used as
[14]. The authors tried three different models of the ODH a first approach and gives a better description of the exper-
of propane reaction (the power model and the models of imental data for the VTiO catalyst (matching mean error
Langmuir–Hinshelwood and of Mars–van-Krevelen, MK). 7.6%) but a lack of a physical meaning is its serious disad-
However the fitting errors for all these models were practi- vantage in this case.
cally the same (e.g. 7.5%) and it was difficult to distinguish For the 1VTiO catalyst, the two models mentioned above
which was the best one. In my opinion there are two reasons were compared with the steady state adsorption model
for such results. The first is a small range of the propane (SSAM) [9] which was based on the following physical as-
conversions and the second is a simplified (only consecu- sumptions: (1) a steady state of the surface oxygen coverage
tive) kinetic network, used by the authors in their calcu- is established on the catalyst surface, (2) the oxygen adsorp-
lations. Our models are based on the parallel-consecutive tion has a dissociative character, (3) the reaction of propane
kinetic network and the kinetic data cover a broader range and propene on the catalyst surface proceeds through the
of the propane conversions and selectivities to propene, CO Eley–Rideal mechanism, which means that the hydrocarbon
and CO2 . The difference in the kinetic networks becomes molecule reacts directly from the gaseous phase without
important especially for high conversions of propane, which adsorption, (4) the propene formation and combustion of
is justified by a relatively high activation energy of the hydrocarbon requires only one type of the surface oxygen.
parallel path. On the basis of the above assumptions, this model allowed
Several different models [4,7,9] have been proposed in us to determine the rate constants for all the steps of the
our previous papers for explanation of experimental results. propane ODH. This model describes the experimental data
They are based on the parallel-consecutive scheme of the for the propane ODH on the 1VTiO catalyst with the least
propane ODH reaction. In the paper [4], the pseudohomoge- mean error (4.9%) and it indicates that propene is the only
nous, first zero-order model (FZO) has been proposed and useful product primary produced, and that CO and CO2 are
applied to the 1VTiO and 1VRbTiO catalysts. In this model produced largely by the sequential oxidation of the in situ
it was assumed that the rate of the reaction is proportional produced propene, and to a lesser extent by a parallel route
to the concentration of propane and independent of the oxy- of the direct deep oxidation of propane.
42 R. Grabowski / Applied Catalysis A: General 270 (2004) 37–47

In this paper the SSAM-Eley–Rideal (ER-SSAM) and reactions the following two models were considered. In the
Langmuir–Hinshelwood (LH) models were compared for the first model (LH-1) orders with respect to the surface oxy-
1VTiO and 1VRbTiO catalysts. Ample sets of experimen- gen coverage in the reactions of propene and COx forma-
tal data on propane conversion and selectivities to oxidation tion (Eqs. (11)–(13)) were taken as 1. In the second model
products as a function of the reagent mixture composition, (LH-2) orders with respect to θ s in the reactions of the COx
temperature and contact time were used in the calculations. formation were taken as 2 to take into account that—in to-
The experimental set of data for the 1VTiO catalyst was ex- tal oxidation reaction—breaking of the C–C bond requires
tended to lower contact times (Table 1) when compared with a concurrent interaction of several oxygen atoms [20]. This
the data used earlier [7]. A similar set of data was obtained assumption is also in accordance with results obtained for
for the 1VRbTiO catalyst (Table 2). the VTiO catalyst in anerobic conditions in transient experi-
As a first, the LH model was tried for these catalysts. In ments [7] where concentration of COx decreased faster than
this model it is assumed that the ODH of propane proceeds that of propene.
through the following individual steps:
Adsorption/desorption of propane
R
 
k1A
C3 H8(g) ⇔ C3 H8(a) = K1 (1)
k1D
Formation of adsorbed propene
C3 H8(a) + O(s) → C3 H6(a) + H2 O + VO k2 (2)
In LH model it is assumed: (1) that the rates of propane
Adsorption/desorption of propene consumption and the formation of COx and propene, de-
II pend on the rates of the surface reactions of the appropriate
 
k3A species and (2) that the equilibrium between adsorption and
C3 H6(g) ⇔ C3 H6(a) = K3 (3)
k3D desorption of propane, propene and oxygen is established
immediately. This assumption is similar to that proposed
Dissociative chemisorption/desorption of oxygen for the ODH of propane on V–Mg–O [14] and in [16]. The
 
k4A cases in which the adsorption of propane or the adsorption
0.5O2 + VO ⇔ O(s) = K4 (4)
k4D of oxygen on the surface of the catalyst are the rate deter-
mining steps in the propane ODH reaction were also tried
Parallel combustion of propane but without success. Assuming that a Langmuir isotherm
adequately describes the adsorption equilibrium of propane,
C3 H8(a) + (3x + 4)O(s) propene and oxygen the following set of algebraic equations
→ 3COx + 4H2 O + (3x + 4)VO k5 (5) can be written:
Consecutive combustion of propene k1A cR (1 − θR − θ␲ − θS ) − k1D θR = 0 (8)

C3 H6(a) + (3x + 3)O(s) k3A c␲ (1 − θR − θ␲ − θS ) − k3D θ␲ = 0 (9)

→ 3COx + 3H2 O + (3x + 3)VO k6 (6) k4A cO (1 − θR − θ␲ − θS ) + k4D θS = 0 (10)

where VO and O(s) denote the oxygen vacancy and the sur- where cR , c␲ , cO are concentrations of propane, propene
face oxygen species on the surface of the catalyst, and in- and oxygen in the gas phase and θ R , θ S and θ ␲ are the
dexes A and D denote adsorption and desorption, respec- surface coverages by propane, oxygen and propene respec-
tively and x is a mean value of oxygen/carbon ratio in the tively which can be calculated by solving the above set of
combustion products (CO + CO2 ). It is obvious, that none the equations. In this model, CO and CO2 are formed in
of the above reactions (Eqs. (2), (5) and (6)) is an elemen- the reaction of the total oxidation of adsorbed propane and
tary reaction, so one must, on the experimental basis, as- propene. Additionally it is assumed that these reactions are
sume the appropriate reaction orders. It is also assumed that strongly shifted to the formation of carbon oxides and that
only one type of the catalyst oxygen is required for both se- the carbon oxides formed desorb immediately from the sur-
lective and combustion reactions and oxygen of the catalyst face of the catalyst and do not block it. Taking into account
is replenished by the dissociative adsorption from the gas that the propane and propene adsorption is a fast process in
phase on the vacancies (Eq. (4)). The rate of the reaction is comparison with the surface reactions, one can write equa-
taken as proportional to cO0.5 where c is the concentration
O tions for the formation of propene, COx and consumption
of oxygen in the gas phase, and of the first order in the sur- of propane in the gaseous phase. The analysis of the evo-
face coverage with the vacancies. On the basis of the above lution of the selectivities to propene and to COx and of the
 R. Grabowski / Applied Catalysis A: General 270 (2004) 37–47 43

conversion of propane was performed using the method pro- Concentration of propane in the reaction mixture
posed by Ai et al. [15]. For the model LH-1 it has led to an dcR
ordinary differential equations which can be written as: = −cR (k1 θs + k3 θs2 ) (19)
dτ
dcR
= −(k2 + k5 )θR θS (11) Concentration of propene
dτ
dcπ
dc␲ = k1 cR θs − k2 c␲ θs2 (20)
= k2 θR θS − k6 θ␲ θS (12) dτ
dτ
Concentration of COx
dcCOx
= (k5 + k6 )θS θR (13) dcCOx
dτ = 3(k2 c␲ + k3 cR )θs2 (21)
dτ
where τ is the contact time. In model LH-2 oxygen cov-
erage θ s for the propane and propene combustion reactions For the ER-SSAM-1 model all θs have the same exponent
(Eqs. (12) and (13)) is squared. All these reactions are as- of 1.
sumed to be of the first order with respect to the hydrocarbon
concentration in the gas phase.
In ER-SSAM model applied to the ODH reaction of 3. Results and discussion
propane on the 1VTiO and 1VRbTiO catalysts, only one
type of the oxygen species is assumed [7]. This oxygen is To verify two above mentioned models in the kinetic
responsible both for the selective oxydehydrogenation of way, 12 sets of data, described in Section 2, were used
propane and for the reactions of the total oxidation. Simi- for non-doped (Table 1) and the rubidium doped catalysts
larly as it had been done for the LH models in the reaction (Table 2) (selectivities to propene and COx as well as the
of the propene and COx formation, the orders with respect conversion of propane are given as a function of the contact
to the oxygen coverage were taken as 1 and 1 or as 1 and 2 time and feed mixture composition at three different temper-
for the ER-SSAM-1 and ER-SSAM-2 models, respectively. atures). The integration of Eqs. (11)–(13) for the LH models
All these reactions are assumed to be of the first order or Eqs. (19)–(21) for the ER-SSAM models was performed
with respect to the hydrocarbon concentration in the gas using Runge–Kutta (fourth order). The fitting procedure was
phase. These assumptions correspond to the following set performed by using the values of the surface coverages ob-
of reactions: tained by the solution of algebraic Eq. (18) for the ER-SSAM
k1
(R) C3 H8(g) + O(S) −
→ () C3 H6(g) + H2 O(g) + V (14) models or solutions of the set of algebraic Eqs. (8)–(10) for
the LH models, and a subsequent minimization of the object
k2
C3 H6(g) +3(1+x)O(S) −
→ 3COx(g) + 3H2 O(g) + 3(1 + x)V function defined as a sum of squared differences of exper-
imental and calculated selectivities and conversions using
(15) Marquardt method in a program written in BASIC.
The difference between the ER-SSAM-1 and ER-SSAM-2
k3 models is insignificant. For the VTiO catalyst the fit errors
C3 H8(g) +(4 + 3x)O(S) −
→ 3COx(g) + 4H2 O(g) + (4 + 3x)V
for both models are almost the same (5.2%) but for the
(16) RbVTiO catalyst the fit error for the ER-SSAM-2 model is
kOS a little lower (6.4%) than that for the ER-SSAM-1 model
O2 + 2V −
→ 2O(S) (17) (6.8%). Differences between the LH-1 and LH-2 models are
more significant and the LH-1 model describes better the
According to this model, propane reacts directly from
experimental data than the LH-2 model. For the VTiO and
the gaseous phase with the surface oxygen of the catalyst,
RbVTiO catalysts the fit errors for the LH-1 model are 14.3
the reaction products desorb quickly and the surface is not
and 21.8%, respectively, while for the LH-2 model they are
blocked. For ER-SSAM-2 model changes of the oxygen sur-
15.4 and 24.1%. On the basis of the obtained results one can
face coverage in the time of the reaction is given by the
state that the ER-SSAM model describes the propane ODH
Eq. (18).
reaction on the VTiO catalyst—both doped by rubidium and
dθ √ undoped—better than the LH model.
= 2kOS cO (1 − θS ) − k1 cR θS Tables 3 and 4 give activation energies and pre-exponential
dt
− 3(1 + x)k2 c␲ θS2 − (4 + 3x)k3 cR θS2 (18) factors for the different reaction steps for the LH-1 and
ER-SSAM-2 models, respectively. Figs. 1 and 2 present
At the steady-state conditions, dθ s /dt is equal to zero the parity plots for the LH-1 model for the non-doped and
which leads to an algebraic equation which can be solved rubidium doped catalysts. In Tables 1 and 2, experimental
in relation to θ s . The set of differential equations describing conversion and selectivity data are compared with values
propane, propene and COx concentrations in the gas phase calculated on the basis of the ER-SSAM-2 model. As can
for the ER-SSAM-2 model can be written as: be seen from Tables 3 and 4, the fitting error is much lower
44 R. Grabowski / Applied Catalysis A: General 270 (2004) 37–47

Table 3
Kinetic constants for the Langmuir–Hinshelwood (LH-1) model for the 1VTiO catalyst undoped/doped with Rb
Catalyst Rate constants Matching error (%)
k2 k5 k6 K1 K3 K4
VTiO 166a (29.5)b 1870 (41. 6) 1200 (36.1) 7500 (31.4) 228100 (31.4) 30500 (41.1) 14.3
SS2lf /SS2pe F(47,7,095) = 3.34 Fcon = 0.652; Fsel␲ = 1.167; FselCOx = 93.048
VRbTiO 70 (25.8) 3330 (50.1) 2200 (42.8) 4280 (32.9) 6610 (26.0) 4210 (32.9) 21.8
SS2lf /SS2pe F(40,11,0.95) = 2.43 Fkon = 4.358; Fsel␲ = 4.76; FselCOx = 143.788
a Pre-exponential factor: ko (s−1 ).
b Activation energy: E (kJ/mol).
Table 4
Kinetic constants for Eley–Rideal (ER-SSAM-2) model for VTiO catalyst undoped/doped by Rb
Catalyst Rate constants Matching error (%)
k1 k2 k3 kOS
VTiO 1.14 × 102a (26. 0)b 2.23 × 103 (22.3) 2.02 × 103 (40.0) 6.5 × 106 (72.5) 5.2
T = 523 K 0.288 13.21 0.204 0.373
SS2lf /SS2pe F(51,7,095) = 3.3 Fcon = 0.431; Fsel␲ = 1.320; FselCOx = 1.325
VRbTiO 28.1 (25.0) 1.27 × 103 (30.0) 1.37 × 103 (49.0) 1.1 × 107 (74.0) 6.7
T = 553 K 0.122 1.861 0.039 1.125
SS2lf /SS2pe F(43,11,0.95) = 2.5 Fkon = 1.871; Fsel␲ = 1.350; FselCOx = 1.417
a Pre-exponential factor: ko (s−1 ).
b Activation energy: E (kJ/mol).
Sel to propene (fractions),

0,6
sel to propene(fractions),

0,5 0,7
Calc. values,

0,6
0,4
Calc. values,

0,5
0,3
0,4
0,2 0,3
0,1 0,2
0,0 0,1
0,0 0,1 0,2 0,3 0,4 0,5 0,6 0,0
Sel to propene(fractions), observed values 0,0 0,1 0,2 0,3 0,4 0,5 0,6 0,7
Sel to propene(fractions), observed values
1,0 1,0
sel to COx (fractions),
Calc. values,

0,8
Calc. values,

0,8
sel to COx,

0,6 0,6
0,4 0,4
0,2
0,2
0,0
0,0 0,1 0,2 0,3 0,4 0,5 0,6 0,7 0,8 0,9 1,0 1,1 0,0
0,0 0,1 0,2 0,3 0,4 0,5 0,6 0,7 0,8 0,9 1,0
Sel to COx (fractions), observed values
Sel to COx(fractions), observed values

0,6
propane conv(fractions)

1,0
propane con. (fractions),

0,5
0,8
Calc. values,

Calc. values

0,4
0,6
0,3
0,2 0,4
0,1 0,2
0,0 0,0
0,0 0,1 0,2 0,3 0,4 0,5 0,6 0,0 0,1 0,2 0,3 0,4 0,5 0,6 0,7 0,8
propane conv(fractions),observed values Propane con.(fractions), observed values
Fig. 1. Parity plots for the ODH of propane (the LH-1 model) on the Fig. 2. Parity plots for the ODH of propane (the LH-1 model) on the
1VTiO catalyst. 1VRbTiO catalyst.
 R. Grabowski / Applied Catalysis A: General 270 (2004) 37–47 45

for the ER-SSAM-2 model than for the LH-1 model and been supported in [8,18] where it was found that the rubid-
F-test shows that in the case of the Rb-doped catalyst the ium additive increases the yield to propene in comparison
LH-1 model does not describe adequately the experimental with the non-doped catalyst. Similar tendency was observed
data. Lack of fit for the LH-1 model is clearly visible for for the ODH of propane on the MoO3 catalyst doped with
the selectivity to COx , but also for propane conversion and cesium [17].
selectivity to propene. In the case of the non-doped cata- The rubidium additive practically does not change the
lyst, the LH-1 model does not fit the selectivity to COx , but rate of the reoxidation kOS , but the decrease in the rates of
in contrast, it fits adequately the conversion of propane and the processes which consume oxygen, leads to an increase
the selectivity to propene though the matching error is much of the (k1 + k2 + k3 )/kOS ratio responsible for the surface
higher for the LH-1 than for the ER-SSAM-2 models, re- reduction degree of the catalyst (1 − θ s ). Fig. 3B shows that
spectively (14.3 and 5.2%). Results of the calculations car- this ratio is much lower for the rubidium doped catalyst and
ried out on the basis of the LH-1 model show that the rubid- is practically independent of the temperature, whereas for
ium additive leads to a certain decrease of all rate constants. the non-doped catalyst the degree of the surface reduction
The increase of the selectivity to propene is mainly result of is higher and depends on the temperature.
the decrease (about 10 times) of propene adsorption equilib- As can be seen from Table 4, the rubidium additive
rium constant K3 , thus of the decrease of the adsorption of strongly decreases the k2 value mainly by increasing the
propene from the gaseous phase. It should be born in mind, activation energy. This reflects a stronger basicity of the
however, that the LH-1 model describes the experimental catalyst surface. Propene molecules contain ␲-bond, which
data with a significant error (Table 3). Concluding, one can makes them more basic than propane and leads to their eas-
state that qualitatively LH-1 is correct from the conceptual ier desorption from the catalyst surface doped by rubidium.
point of view, but it must be excluded on the ground of the These changes in the k2 value lead to an increase of the
lack of fit. k1 /k2 ratio (Fig. 3D).
The rate constants obtained for the ER-SSAM-2 model The same increase of the activation energy E3 is ob-
applied to the ODH of propane on the VTiO catalyst us- served for the rubidium doped catalyst. The decrease of the
ing more complete experimental data than in [7] are qual- pre-exponential factor k3 acts in the same way (due to the
itatively similar to those obtained by us for the compara- blocking of the active centers responsible for the interac-
tively high propane conversions (high contact times). There tion of propyl intermediates) leading in consequence to a
exist, however, some quantitative differences, especially for decrease in the rate constants k3 for this catalyst. The effect
the estimation of k1 /k3 ratio, which is higher for the more of Rb additive on k1 /k3 ratio which reflects the relative rate
complete data. of the ODH reaction and the primary propane combustion is
The differential selectivity to propene can be obtained in shown in Fig. 3C. As can be seen from this figure, the k1 /k3
ER-SSAM-2 model by dividing Eq. (20) by Eq. (19) which values increase when rubidium is added, what reflects the
leads to Eq. (22): decrease of undesired combustion of propyl intermediates
dc␲ k1 c R − k 2 c ␲ θ s before their transformation in propene.
S␲
= − = (22) Generally in the frame of ER-SSAM model, rubidium
dcR cR (k1 + k3 θs )
additive leads to a decrease to a different extent of all the
This allows to calculate the limit value of the selectivity pre-exponential factors (Table 4) for rate constants which
to propene when X → 0 describe the reaction of propane ODH. This is the reason of
k1 a lower activity of the rubidium doped catalyst. The same
lim S␲
= (23) effect has been observed for other alkali metal additives (Li
X→0 k 1 + k 3 θs
and K) in the ODH of propane on the VTiO catalyst [18].
For X → 0, the oxygen surface coverage is expected to be Unfortunately the activation energies and the pre-
close to 1. The limit values for the selectivity of propene cal- exponential factors for the calculated rate constants cannot
culated from the Eq. (22) for studied catalysts are reasonable be compared with any literature data for the ODH reaction
and are in agreement with the experimental data. Thus, one on the VTiO catalyst. The only kinetic data available for
can state that the ER-SSAM-2 model describes adequately the ODH of propane concern the V–Mg–O catalyst [14],
the ODH reaction of propane on the 1VTiO and 1RbVTiO which is much less active. Furthermore, the models applied
catalysts in the broad range of the conversions of propane. by the authors to describe the ODH reaction of propane
On the basis of the calculated results collected in Table 4, [14] have been based on different reaction network and
it is evident that the Rb additive increases to some extent the have had different reaction stages. For this reason a quan-
activation energy for the consecutive oxidation of propene, titative comparison of the activation energies and the rate
which is the main route responsible for the total oxidation constants determined in this paper with the data obtained
of propane. As can be seen from Fig. 3A, which shows the by authors of [14] is impossible. Moreover the catalytic ac-
dependence of the ratio of the selective reaction (k1 ) to the tivity of V–Mg–O is considerably lower in comparison with
total oxidation (k2 + k3 ), the rubidium additive leads to the V2 O5 /TiO2 , i.e. for the V–Mg–O catalyst the same catalytic
four time increase of this ratio. Quantitatively this result has activity is attained at the temperature about 200 K higher.
46 R. Grabowski / Applied Catalysis A: General 270 (2004) 37–47

0,07 (A) 40 (B)

35
0,06

RbVTiO 30
0,05

25
(k1+k 2+k 3)/kOS
k1/(k2+k 3)

0,04

20

0,03
15
VTiO
0,02 VTiO
10

0,01
5

RbVTiO
0,00 0
520 540 560 580 600 620 640 520 540 560 580 600 620 640

5,5
0,070
(C) (D)
5,0
0,065

4,5 0,060
RbVTiO
0,055
RbVTiO
4,0
0,050
3,5
0,045

3,0 0,040
k1/k3

k1/k2

0,035
2,5
0,030
2,0
0,025

1,5 0,020
VTiO
VTiO 0,015
1,0
0,010
0,5
0,005

0,0 0,000
520 540 560 580 600 620 640 520 540 560 580 600 620 640

Temperature [K] Temperature [K]

Fig. 3. Temperature dependence of k1 /(k2 + k3 ), (k1 + k2 + k3 )/kOS , k1 /k3 and k1 /k2 for the ER-SSAM-2 model for the non-doped and rubidium doped
1VTiO catalyst.
 R. Grabowski / Applied Catalysis A: General 270 (2004) 37–47 47

The kinetic model similar to ER-SSAM for the ODH propane indicates a blocking of the active centers and
of propane on the VTiO catalyst has been applied for the leads to the decrease of the catalytic activity of the ru-
description of the non-stationary kinetics of the ODH of bidium doped catalyst in comparison with the non-doped
propane in anaerobic conditions by Genser and Pietrzyk catalyst.
[19]. Though it differs in some details from the ER-SSAM
model of the present work, it is possible to compare the
value of the activation energy obtained for the formation of Acknowledgements
propene (k1 ). The activation energy of 26 kJ/mol obtained
in the steady state experiments (Table 4) is close to that of This research was partially supported by the Polish
29.1 kJ/mol given in [19] for the selective ODH reaction. State Committee for Scientific Research, under the project
PBZ/KBN/018/T09/99 4b.

4. Conclusions
References
1. The ODH reaction of propane is described more ade-
quately by the ER-SSAM model than the LH model, both [1] S.W. Weller, Catal. Rev. Sci. Eng. 34 (3) (1992) 227–280.
for the non-doped (1VTiO) catalyst and for the rubidium [2] J.R. Kittrell, Adv. Chem. Eng. 8 (1970) 97.
doped (1RbVTiO) catalysts. [3] S.L.T. Andersson, Appl. Catal. A: Gen. 112 (1994) 209.
[4] J. Słoczyński, R. Grabowski, K. Wcisło, B. Grzybowska–Świerkosz,
2. In the frame of the ER-SSAM model in the ODH reaction Polish J. Chem. 71 (1997) 1585–1593.
of propane, propane from the gaseous phase interacts [5] D.L. Stern, R. Grasselli, J. Catal. 167 (1997) 560–569.
with the catalyst oxygen and next it undergoes a selective [6] S. Pietrzyk, F. Genser, Stud. Surf. Sci. Catal. 130 (2000) 1871.
reaction or is oxidized to CO and CO2 . The desorption [7] R. Grabowski, S. Pietrzyk, J. Słoczyński, F. Genser, K. Wcisło, B.
Grzybowska-Świerkosz, Appl. Catal. A: Gen. 232 (2002) 277–288.
of the reaction products to the gaseous phase is rapid and
[8] R. Grabowski, B. Grzybowska, K. Samson, J. Słoczyński, J. Stoch,
they do not block the surface of the catalyst. A subsequent K. Wcisło, Appl. Catal. A: Gen. 125 (1995) 129.
combustion of propene which is the main source of COx [9] R. Grabowski, J. Słoczyński, M. Grzesik, Appl. Catal. A: Gen.
occurs by the consecutive interaction of propene from the 242 (2) (2003) 297–309.
gaseous phase with the surface of the catalyst. [10] K. Chen, E. Iglesia, A.T. Bell, J. Catal. 192 (2000) 197–203.
[11] N. Boisdron, A. Monnier, L. Jalowiecki-Duhamel, Y. Barbaux, J.
3. Addition of rubidium leads to a decrease of the rate
Chem. Soc., Faraday Trans. 91 (17) (1995) 2899.
constants of the propene formation and a simultaneous [12] J.N. Michaels, D.L. Stern, R.K. Grasseli, Catal. Lett. 42 (1996) 135–
increase of the activation energy of the subsequent ox- 137.
idation of propene, which results in the increase of the [13] J.N. Michaelis, D.L. Stern, R.K. Grasselli, Catal. Lett. 42 (1996) 139.
selectivity to propene. [14] D. Creaser, B. Andersson, Appl. Catal. A: Gen. 141 (1996) 131–152.
[15] M. Ai, P. Boutry, R. Montarnal, G. Thomas, Bull. Soc. Chim. France
4. Rate constant of the parallel formation of COx (k3 ) for
8–9 (1970) 2873.
the catalyst doped with rubidium is decreased in compar- [16] Z. Belohlav, P. Zamostny, Can. J. Chem. Eng. 78 (2000) 513–521.
ison with the non-doped catalyst, which comes out from [17] K. Chen, S. Xie, A.T. Bell, E. Iglesia, J. Catal. 183 (1999) 210–221.
the increase of the activation energy and decrease of the [18] R. Grabowski, B. Grzybowska, A. Kozłowska, J. Słoczyński, K.
pre-exponential factor. Wcisło, Top. Catal. 3 (1996) 277–288.
[19] F. Genser, S. Pietrzyk, Chem. Eng. Sci. 54 (1999) 4315.
5. The decrease of the pre-exponential factors of the rate
[20] H.H. Kung, Adv. Catal. 40 (1994) 1.
constants for the all stages of the ODH reaction of