Oxidative dehydrogenation of propane on NixMg1-xAl2O4 and NiCr2O4 spinels

Article abstract of DOI:10.1006/jcat.1999.2626

The Ni_xMg_1-xAl₂O₄, NiCr₂O₄, and MgCr₂O₄ spinels have been synthesized, characterized with the XRD and XPS methods, and tested in the oxidative dehydrogenation of propane. The crystallochemical model of solid surfaces, CMSS, has been used to calculate the oxygen cation's bond energies in the spinels. For the NiMgAl spinels the activity and selectivity to propene increase with the increase in the Ni content. The Ni ions surrounded by oxygen in the spinel structure are proposed as active centers for oxidative dehydrogenation to propene. The NiCr spinel is more active but less selective than the NiMgAl spinels; the difference in catalytic behavior has been ascribed to different coordination of Ni ions in the two groups of the spinels and to the lower oxygen cation's bond energy in the NiCr spinel.

Full text of DOI:10.1006/jcat.1999.2626

Journal of Catalysis 187, 410–418 (1999)
Article ID jcat.1999.2626, available online at http://www.idealibrary.com on
Oxidative Dehydrogenation of Propane on NixMg1 xAl2O4
and NiCr2O4 Spinels
Jerzy Sloczyn´ski, Jacek Zio´lkowski, Barbara Grzybowska, , 1 Ryszard Grabowski, Dariusz Jachewicz,
!
!
!
Krzysztof Wcislo, and Leon Gengembre†
Institute of Catalysis and Surface Chemistry, Polish Academy of Sciences, ul. Niezapominajek 30-239 Krako´w, Poland; and †Laboratoire de Catalyse
He´te´roge`ne, URA CNRS 402, Universite´ des Sciences et Technologies de Lille, 59 655 Villeneuve d’Ascq Ce´dex, France
Received March 16, 1999; revised July 5, 1999; accepted July 6, 1999
The series of the NixMg1 xAl2O4 spinels of different
Ni content and the NiCr2O4 and MgCr2O4 spinels were
synthesized. Their structure and composition, both in the
bulk and on the surface, have been verified with XRD and
XPS techniques, respectively. The crystallochemical model
of solid surfaces (CMSS) (8–11), described briefly in the
next paragraphs, has been also applied to evaluate the oxy-
gen binding energies in different coordination of the ions,
both in the bulk and on the surface of the studied systems.
The NixMg1 xAl2O4, NiCr2O4, and MgCr2O4 spinels have been
synthesized, characterized with the XRD and XPS methods, and
tested in the oxidative dehydrogenation of propane. The crystallo-
chemical model of solid surfaces, CMSS, has been used to calculate
the oxygen cation’s bond energies in the spinels. For the NiMgAl
spinels the activity and selectivity to propene increase with the in-
crease in the Ni content. The Ni ions surrounded by oxygen in the
spinel structure are proposed as active centers for oxidative dehy-
drogenation to propene. The NiCr spinel is more active but less
selective than the NiMgAl spinels; the difference in catalytic behav-
ior has been ascribed to different coordination of Ni ions in the two
groups of the spinels and to the lower oxygen cation’s bond energy
c
EXPERIMENTAL
in the NiCr spinel.
1999 Academic Press
Preparation of the Catalysts
Key Words: oxidative dehydrogenation of propane; nickel-chrom
spinels; nickel-magnesium-aluminium spinels.
The NixMg1 xAl2O4 (0
x
1) and MgCr2O4 spinels
were prepared with the citrate method using solutions of
nitrates of the respective metals (0.5–2 M) and 2 M citric
acid as precursors. All the starting materials were of high
purity (at least proanalysis grade).
INTRODUCTION
The desired amounts of the nitrates were introduced
drop-wise to the solution of citric acid in 2–3% excess with
respect to the stoichiometry of the given citrates. The so-
lutions were evaporated in a vacuum rotary evaporator at
363 K for 3 h. The final drying was effectuated at 673 K
for 6 h. The vitrous precursors thus obtained were carefully
ground and calcined in an alundum crucible in a stream
of air at 823 K for 24 h and then again in air at 1173 K
for 144 h. In the course of the last calcination the samples
were cooled down every 48 h, ground in a mortar, and pul-
verized in an ultrasonic disperser. The NiCr2O4 spinel was
obtained with a coprecipitation method. The NH3aq (14% )
was added step-wise to a solution of nickel and chromium
nitrates kept at 353 K under continous stirring till pH 6.5
was reached. The precipitate was aged on a water bath for
2 h, filtered, washed with water, and dried at 373 K. The
coprecipitated Ni [II] and Cr [III] hydroxides were decom-
posed in air at 773 K for 5 h; the sample was then ground,
tableted, and calcined at 1173 for 40 h. The specific surface
areas of the preparations are given in Table 5.
Nickel-containing systems have been found in the past
decade to be active and selective catalysts in oxidative de-
hydrogenation (ODH) of alkanes, a reaction proposed re-
cently as an alternative to classical dehydrogenation for the
production of cheap olefins. Nickel molybdate has been re-
ported as efficient in the ODH of propane (1–4), whereas
nickel phosphate Ni2P2O7 (5, 6) and NiO dispersed on dif-
ferent supports (7) are shown to be promising catalysts in
the respective reaction of isobutane.
There is practically no information about the structure
of the active centers and the role of the environment of
nickel ions in the ODH reactions. In an attempt to ap-
proach this problem in the present work the catalytic per-
formance in the ODH of propane has been studied for the
NiMgAl and NiCr spinels characterized, respectively, by
octa- and tetrahedral coordination of Ni ions in the Ni–O
polyhedra.
1 To whom correspondence should be addressed.
410
0021-9517/99 $30.00
c
Copyright
1999 by Academic Press
All rights of reproduction in any form reserved.

PROPANE OXIDATIVE DEHYDROGENATION ON NiMgAl(Cr) SPINELS
411
XPS
XPS spectra of the samples were recorded with a Leybold
Heraeus LHS 10 spectrometer using an Al K X-ray source
with incitent radiation at 1486.6 eV. The base pressure was
10 8 T. The surface stoichiometry was calculated from the
intensity ratios IA/IB of the elements using the formula
different products were calculated from the formula Si =
P
ci/ ci, where ci are concentrations of products i. The total
conversion values Xp were calculated as
cp0 cp
Xp =
,
[2]
cp0
EkAin/EkBin 1.77nA/nB,
[1]
where cp0 and cp are the concentrations of propane at the
inlet and outlet of the reactor, respectively.
IA/IB =
/
A
B
where are the cross-sections taken from (12) and nA and
nB are the numbers of atoms on the surface.
RESULTS AND DISCUSSION
XRD
XPS
The XRD diffractograms were recorded with a DRON-2
diffractometer (Cu K , a Ni filter, Al as an internal stan-
dard) equipped with a DAS interface. The spectra were
interpreted using the SMOK (Elector Co., Krako´w) and
LATCON (CERN library) programs. The simulation of
diffractograms of the spinels of different inversion degree
was performed with the use of Rietveld’s program DBWS-
9006 PC (A. Sakthivel and R. A. Young, Atlanta).
Figure 1 shows the XPS Ni 2p spectra for the sam-
ples of NiAl2O4, and NiCr2O4. The spectra of other
NixMg1 xAl2O4 samples resemble that of the NiAl2O4 spec-
trum. Table 1 lists the binding energies (BE) and the full
width at half-maximum (FWHM) of Ni 2p, O 1s, Al 2p,
Mg 1s, and Cr 2p levels in the studied samples. The shift be-
tween the Ni 2p3/2 signal and its satellite, 1Esat, and the ratio
of intensities (the peak heights) of the satellite to principal
peaks of the Ni 2p3/2 spectrum Isat/Ip are also reported. The
error in the estimation of the surface content of a given
Catalytic Activity Measurements
element is 5% , the sum of positive (Ni2+, Mg2+, Al3+
,
The activity of the catalysts in oxidative dehydrogenation
of propane was measured in a fixed bed flow apparatus in
the temperature range 673–773 K. A stainless steel reac-
tor (120 mm long, internal diameter 13 mm) was coupled
directly to a series of gas chromatographs. Propene and car-
bon dioxide were found to be main reaction products and
small amounts of carbon monoxide were also observed.
The amounts of the degradation, C2 products, and oxy-
genates (acrolein and acrylic acids) were below 1% of the
total amount of products. The reaction mixture contained
7.1 vol.% of propane in air. The flow rate of the reaction
mixture was controlled with a flow meter type ERG-IMKZ.
One to 2 g samples of a catalyst of grain size 0.63–1 mm
were used. It has been checked in the preliminary mea-
Cr3+) and negative (O2 ) charges is balanced within 4% .
Following the literature data (13–16) the values of BE for
the Ni 2p level and of the satellite shift 1E as well as the
I
sat/Ip ratio permit identification of the Ni ions as divalent in
octahedral (NiAl2O4 and NixMg1 xAl2O4) and tetrahedral
(NiCr2O4) environments. The last column gives the surface
stoichiometry of the samples calculated from the XPS data.
As seen the surface stoichiometry of the samples is close
to that ofthe bulk ofthe samplesassumed at the preparation
of the respective spinels.
XRD
All of the prepared samples gave diffractograms char-
surements in which the mass and the grain size of a catalyst acteristic of pure spinels with sharp, easily indexed re-
sample were varied, that under these conditions the trans- flexions. The values of lattice constants were in good agr-
port phenomena do not limit the reaction rate. For each reement with the literature data (17) or, in the case of
sample the activity measured on increasing and decreas- NixMg1 xAl2O4, with those obtained by the interpolation.
ing the temperature was reproducible within the limits of
The MgNiAl spinels showed a cubic X-ray pattern while
experimental error (5%). A sample was kept at each re- in the case of NiCr2O4 tetragonal distortion connected with
action temperature for about 20 h; no deactivation of the the Jahn–Teller effect was observed. Tetragonal form trans-
catalysts was observed during this period. The kinetic mea- forms into cubic at 310–330 K (18, 19).
surements were performed measuring at each temperature
If we describe AIIBIIIO4 spinels by the general formulae
concentrations of unreacted propane and of reaction prod- accountingfor the tetrahedral( ) and octahedral [ ]distribu-
ucts as a function of a contact time . The latter was varied tion of A and B cations we have: (A)[B2]O4 (normal spinel),
between 0.1 and 3 s by changing the flow rate of the reaction (B)[AB]O4 (inversed spinel), and (A1 zBz)[AzB2 z]O4
mixture. The values of the total conversion of propane did (mixed spinel), where z is the inversion degree.
not exceed 30% . The carbon balance for conversions higher
Figure 2a presents a typical diffractogram of the NiAl2O4
than about 10% was better than 97 2% . At lower conver- spinel, whereas in Figs. 2b and 2c, the diffractograms are
sions the balance was poorer and hence the selectivities to simulated with the DBWS program for z = 1 and 0,

!
´
412
SLOCZYNSKI ET AL.
FIG. 1. XPS spectra of NiAl2O4 and NiCr2O4 in the Ni 2p region.
respectively. It is seen that the diffractogram of NiAl2O4 with the Bertaut–Greenwald method (20–21). The method
strongly resembles the simulated pattern of the inversed consists of comparison of the ratios of experimental inten-
spinel.
sities of reflexions I (400)/I (220) and I (400)/I (224) with
Table 2 lists lattice constants calculated from the experi- the theoretical values simulated with the DBWS-9006 PC
mental data, packing coefficient of the oxygen ions, u (liter- program. For most of the studied samples the coefficients
ature or interpolated data), and the distribution of cations ( )/[ ] were determined with the accuracy 0.01–0.04; for the
between the tetra- and octahedral positions determined NiCr sample it was lower, amounting to 0.1. The results
TABLE 1
XPS Data for Nickel Spinels
BE, eV
hFWHMi
Surface
Spinel
NiAl2O4
Ni 2p3/2
O1s
Me
Ni (1Esat
)
Is/Ip
stoichiometry
856.2
h3.2i
531.1
h2.5i
Al 2p.
74.2
2.4i
6.2
0.51
NiAl2O4.1
h
Ni0.1Mg0.9Al2O4
Ni0.5Mg0.5Al2O4
NiCr2O4
856.4
h3.3i
531.2
h2.9i
Mg 1s
1304.2
h2.7i
6.3
6.3
5.2
0.51
0.49
0.74
Ni0.14Mg1Al2O4.1
Ni0.5Mg0.58Al1.85O4
NiCr2.2O4.1
Al 2p.
74.2
h
2.3i
856.2
h3.2i
531.1
h2.7i
Mg 1s
1304.0
h2.5i
Al 2p.
74.2
h
2.4i
Cr 2p3/2
576.7
h3.5i
856.5
530.2
h1.9i
h
4.3i

PROPANE OXIDATIVE DEHYDROGENATION ON NiMgAl(Cr) SPINELS
413
surface of the spinels with the CMSS (8–11). The model is
based on a new empirical correlation proposed in (8) be-
tween the bond strength, s, defined first by Pauling as the
valence of a cation divided by its coordination number in a
crystal, and the bond length R. Multiplication of s derived
from this correlation by a coefficient J normalized to the
enthalpy of atomization of a simple oxide of a considered
cation gives the bond energy E expressed in the energy
unit as a function of R (9). The bond energy will then de-
pend on structural factors (reflected in the bond length, R,
different for different coordinations, not necessarily sym-
metrical) and on intrinsic properties of a cation included in
the value of J. It should be noted that the coefficient J in-
volves the enthalpy of atomization of an oxide and not the
lattice energy which is the formation energy of the oxide
from ions. The values of E are only approximate since the
model does not take into account the long-range coulombic
interactions as was done with more sophisticated methods;
see, for example, Refs. (22, 23). In spite of these shortcom-
ings the model has been successfully applied to explain the
catalytic properties of several oxide catalysts in oxidation
reactions, as a function of the calculated Me–O bond ener-
gies (10, 24, 25). In calculations of the oxide surface, the sum
of energies of local missing bonds 6Ec per surface unit cell
area has been assumed to be the surface energy, Esurf (11).
The surface slices are selected according to the minimum of
Esurf. In accordance with Esurf the equilibrium shape of the
grains has been determined with the Curie–Wulff construc-
tion and the impact of various (hkl) faces to the external
surface has been calculated (11). The rigid surface model
neglecting the surface relaxation phenomena has been
adapted in the calculations. It is, however, assumed that
a molecule reacting on the surface plays the role of an ad-
ditional ligand neutralizing (absorbing) the local relaxation
energy (10, 11).
FIG. 2. Diffractograms of NiAl2O4 spinel. Experimental (a), simu-
lated with the DBWS program; inverse spinel (b); normal spinel (c).
confirm the structure of the spinels reported in the litera-
ture: NiAl preparation is pure inverse spinel and MgCr and
NiCr are normal spinels, whereas the MgNiAl preparations
are almost totally inverse spinels.
Oxygen Cations and Cation–Oxygen Binding Energies
Table 3 gives the values of the oxygen cation’s binding
energies 6E in NiAl2O4, NiCr2O4, and MgCr2O4, calcu-
lated for different coordinations in the bulk and on the
In the structure of spinels each oxygen ion coordi-
nates four cations, one of which is tetra- and three octa-
coordinated: 1
O
( ) + 3
O
[ ]. 6E is a sum of the
TABLE 2
Structural Data of the NiMgAl, NiCr, and MgCr Spinels
a
˚
Spinel
MgAl2O4
Mg0.9Ni0.1Al2O4
Mg0.5Ni0.5Al2O4
NiAl2O4
a (A)
ub
Cation distributionc
8.0833(10)
8.0790(10)
8.0617(13)
8.0426(17)
8.3374(25)
0.2624
0.2618
0.2594
0.2560
0.2612
(Mg0.77Al0.23
(Mg0.90Al0.10
)
)
[Mg0.23Al1.77
[Ni0.10Al1.90]
]
(Mg0.50Ni0.03Al0.47
(Al1.00
(Mg1.00
)
[Ni0.47Al1.53
[Ni1.0Al1.00
[Cr2.00
]
)
)
]
MgCr2O4
]
NiCr2O4
a = 8.2545(10)
d
d
c = 8.4269(15)
(Ni1.00
)
[Cr2.00
]
a Experimental.
b Literature or interpolated data.
c Determined with the Bertaut–Greenwald method (typical error for the coefficient of the pre-
dominating ion in the indicated position is 5% ).
d According to (17).

!
´
414
SLOCZYNSKI ET AL.
TABLE 3
Oxygen Cation’s and Metal–Oxygen Binding Energies 6E and E(Me-O) (kcal mol 1) in
Various Environments in the Bulk and at the Surface of NiAl, NiCr and MgCr Spinels
NiAl2O4
NiCr2O4
MgCr2O4
E(Me–O)
E(Me–O)
E(Me–O)
Ni Oc Al Td Al Oc
36.60 91.58 61.05
Ni Td Ni Oc Cr Oc
54.90 36.60 53.30
Mg Td Mg Oc Cr Oc
59.55 39.70 53.30
Coordination
6E
Coordination
6E
Coordination
6E
Bulk
(Al) [AlAlAl]
(Al) [AlAlNi]
(Al) [AlNiNi]
(Al) [NiNiNi]
Bulk
(Ni) [CrCrCr]
Bulk
(Mg) [CrCrCr]
274.73
250.28
225.83
201.38
214.80
219.45
Surface
Surface
Surface
1 Td
1 Td
1 Td
[AlAlAl]
[AlAlNi]
[AlNiNi]
183.15
158.70
134.25
[CrCrCr]
159.90
161.50
108.20
[CrCrCr]
159.90
166.15
112.82
1 Oc
1 Oc
(Ni) [CrCr]
1 Oc
(Mg) [CrCr]
(Al) [AlAl]
(Al) [AlNi]
(Al) [NiNi]
213.68
189.23
164.78
2 Oc
(Al) [Al]
(Al) [Ni]
2 Oc
(Ni) [Cr]
2 Oc
(Mg) [Cr]
152.63
128.18
Note. Td, tetrahedral sites; Oc, octahedral sites; , the most probable surface sites.
energies of all O-cation bonds existing in a given oxygen in NiCr2O4, which may arise from the stronger Mg–O bond
coordination, which determines the ability of abstraction as compared with the Ni–O bond.
of an oxygen ion from the catalysts.
It has been shown in previous works (24, 26) that the most
frequently exposed faces of the spinels (100), (110), and
(111) exhibit unsaturated oxygen atoms of the following
Catalytic Performance and Its Relation to the Composition
and Structure of the Catalysts
Figure 3 presents the changes of the selectivities to
the main reaction products, propene and CO2, with total
propane conversion at 723 K for the NiMgAl spinels. The
selectivity to propene increases and that to CO2 decreases
with the increase in the Ni content. In Fig. 4 the curves of the
propene selectivity vs conversion for the NiAl and MgAl
coordinations:
3
O
[ ] (= 1 Td)
1
1
O
O
( ) + 2
O
O
[ ] (= 1 Oc)
( ) (= 2 Oc).
( ) + 1
On the spinel surface one tetrahedral ( 1 Td) or one oc- spinels are compared with those for the Cr-containing com-
tahedral ( 1 Oc) or two octahedrals ( 2 Oc) are missing pounds and with the data for pure oxides Al2O3 and MgO.
with respect to the bulk. The most probable surface coor- As seen from Fig. 4 the NiAl spinel is, at comparable con-
dinations of oxygen are marked with . Table 3 provides versions, much more selective in the propene formation
also the binding energies, E (Me–O) for the Mg–O, Ni–O, than NiCr2O4. The selectivities to propene for NiAl2O4 are
and Cr–O bonds in tetrahedral (Td) and octahedral (Oc) comparable to those reported for the best catalysts in the
sites.
ODH of propane, i.e., NiMoO4 (1–4) and V–Mg–O (27–29)
As seen from the table, for the same type of environ- systems.
ment of the oxygen ions the values of 6E both in the bulk
The MgAl and MgCr spinels, MgO and Al2O3, show
and on the surface of the spinels are distinctly higher for low activity in the propane–oxygen reaction, their selec-
NiAl2O4 than for the Cr-containing compounds. This is due tivities to propene being considerably lower than those for
to the higher Al–O binding energy as compared with that NiAl2O4 and higher than the selectivities for NiCr2O4.
of Mg–O or Ni–O bonds. For the two Cr-containing com-
The decrease of the selectivity to propene accompa-
pounds, oxygen is more strongly bound in MgCr2O4 than nied by the increase in the selectivity to CO2 indicates the

PROPANE OXIDATIVE DEHYDROGENATION ON NiMgAl(Cr) SPINELS
415
efficients are comprised between 0.952 and 0.975. The first
order with respect to propane can be then used as a first ap-
proximation in description ofthe reaction network. The first
order with respect to propane has been also reported for the
ODH propane on other catalysts, e.g., V2O5/TiO2 (30–32)
or CoNi molybdates (33). In the same works (31–33) the 0th
order with respect to oxygen has been found. Our data on
reduction with propane and reoxidation with gaseous oxy-
gen of the V2O5/TiO2 pure and alkali promoted catalysts
(34) indicate, moreover, that the surface reoxidation rate
is considerably higher than that of the reduction. It seems
justified then to assume that in the present measurements,
in which an excess of oxygen in the reaction is used, the
reaction is of the first order with respect to propane and is
independent of the oxygen pressure. The first order with
respect to propene in step 2 is also assumed. With these
assumptions the expressions for the rates of different steps
shown in Scheme 3 are given by
dcp
d
dc
d
= kpcp,
1 dcCOx
= k1cp k2c
[4]
[5]
= k3cp + k2c
3
d
where kp = k1 + k3, cp, and c are concentrations of pro-
pane and propene, respectively.
FIG. 3. Selectivity to main products of the C3H8 + O2 reactions as a
function of propane conversion for NixMg1 xAl2O4 spinels.
consecutive route of the reaction, in which propene formed
in the primary reaction is further oxidized to COx. The fact
that selectivities to propene extrapolated to 0th conversion
are considerably lower than 100% suggests, moreover, that
a fraction of carbon oxides is formed also by a route par-
allel to propene formation. The propane–oxygen reactions
on the studied catalysts can be hence described by the fol-
lowing network:
[3]
In Fig. 5 the propane conversion at 723 K is plotted as a
function of contact time for the NiMgAl spinels of dif-
ferent Ni content. The curves are calculated according to
the equation of the first order with respect to propane; the
points represent the experimental data. The correlation co-
FIG. 4. Comparison of selectivity vs conversion curves for Ni(Mg)Al
and Ni(Mg)Cr spinels.

!
´
416
SLOCZYNSKI ET AL.
gies of steps 1, 2, and 3 for the NixMg1 xAl2O4 spinels are
also given.
Generally, the rates of the three reactions are higher for
the Cr-containing spinels as compared with those contain-
ing Al. The Mg spinels (MgAl and MgCr preparations)
have much lower activity in the three reactions than the
respective Ni spinels, suggesting that the active centers in
the propane–oxygen reactions in NiMgAl and NiCr systems
involve Ni ions. By analogy to oxidation reactions of other
hydrocarbons (35) a Ni2 + –O2 couple can be envisaged as
an active center for the propane reactions, the oxygen ions
participating in abstraction of a hydrogen from a hydrocar-
bon molecule and/or in incorporation into the products. Ni
ions on the other hand facilitate the activation of the C–H
bond and the electron transfer.
MgCr2O4 is considerably more active than MgAl2O4,
which indicates that Cr ions can also catalyze the reactions
under consideration, though to a smaller extent than Ni.
The activity of Cr in chromium oxide dispersed on sup-
ports in the ODH of propane and isobutane has been in
fact observed in earlier works (36–39).
FIG. 5. Propane conversion as a function of contact time for
NixMg1 xAl2O4 spinels. Reaction temperature, 723 K.
The solutions of the above equations give the following
expressions for conversion and selectivities:
Another general observation is that the rates of oxida-
tion of propene to COx (step 2) are considerably higher
than those of the propene formation from propane. This
phenomenon, observed previously for other catalytic sys-
tems (30–32), has been explained by lower energy of allylic
C–H bond in a propene molecule as compared with that of
the parafinic bonds in propane.
For the NixMg1 xAl2O4 spinels at all the reaction temper-
atures the specific rate of step 1 increases with the increase
in the Ni content, x, confirming that the Ni ions are in-
volved in active centers for oxidative dehydrogenation of
propane to propene. The rates of steps 2 and 3 show, on
the other hand, a maximum at x = 0.5 (except for step 2
at the highest reaction temperature 773 K). This suggests
that other centers beside Ni ions may contribute to total
oxidation reactions and/or different surface complexes of
adsorbed hydrocarbons are responsible for selective and
Xp = 1 exp( kp ),
[6]
k1 exp( kp
)
exp( k2
)
S =
k2 kp
1
exp( kp
)
k3
=
k1k2
1
1 1 exp( k2
k2 1 exp( kp
)
)
SCOx
+
,
kp
k2 kp kp
[7]
SCOx = 1 S .
The rate constants k1, k2, k3 were calculated by a standard
nonlinear regression method with the Sigma-plot program
(Jandel Scientific).
Table 4 lists the values of the specific rates of reac-
tions 1, 2, and 3 obtained using the calculated values of
ki (i = 1, 2, 3) for the propane and propene initial content
of 7.1 vol.% for all the studied spinels. The activation ener-
TABLE 4
Catalytic Activity of the Ni Contained Spinels in ODH of Propane
Specific catalytic activity
mol C3H8 (C3H6) s 1 m
10 3
2
Activation
Specific
surface
C3H8 → C3H6
C3H8 → COx
C3H6 → COx
energy
(kJ mol
1
Temp., K
Temp., K
Temp., K
)
area
1
Catalyst
MgAl2O4
Ni0.1Mg0.9Al2O4
Ni0.5Mg0.5Al2O4
NiAl2O4
(m2 g
)
673
723
773
673
723
773
673
723
773
E1
E3
E2
20.7
12.0
7.3
9.15
3.6
0
0.26
0.79
5.86
12.15
21.3
3.6
0.90
2.47
12.14
25.42
—
—
1.33
3.59
3.22
—
2.86
4.23
10.37
6.93
33.4
7.0
7.56
11.46
17.6
13.47
—
—
61
92
73
—
—
39
79
135
122
121
64
47
99
188
203
—
115
107
89
69
—
91
94
86
62
—
—
17
25
31
43
—
—
0.21
1.58
5.20
—
NiCr2O4
MgCr2O4
15.5
—
—
—
—
—
—

PROPANE OXIDATIVE DEHYDROGENATION ON NiMgAl(Cr) SPINELS
417
TABLE 5
Turnover Frequency (TOF) for Propene Formation in the Nickel Contained Spinels
Contributions
of crystal faces
TOF
C3H6 molecules (at. Ni) 1 s
103
1
Surface Ni
concentration
2
Catalyst
(100)
(110)
(111)
at. nm
673 K
723 K
773 K
Ni0.1Mg0.9Al2O4
Ni0.5Mg0.5Al2O4
NiAl2O4
0.35
0.37
0.35
0.33
0.01
0.10
0.24
0.33
0.64
0.53
0.40
0.33
0.334
1.82
4.45
3.3
0.38
0.52
0.70
—
0.96
1.84
1.60
3.88
2.84
3.72
3.32
—
NiCr2O4
nonselective paths of oxidation—as has been claimed for that the error in the evaluation of TOF can be high. The
other oxide systems (35). Further studies, now in course, differences between the TOF values of the three NiMgAl
are necessary to elucidate this problem.
spinels become smaller for higher reaction temperatures at
The activation energy of steps 1 and 3 decrease with which the overall activity is higher and hence the error in
the increase in the Ni content, suggesting modification of estimation of catalytic data is smaller. The standard error of
the active centers for these steps. One may speculate that the mean of the TOF values for the three spinels is 17% at
this modification consists in changing the electron den- 673 K and 8% at 773 K. NiCr2O4 shows much higher values
sity around oxygen atoms surrounding the nickel cations of TOF as compared with the NiAl spinels.
to facilitate the hydrogen abstraction from a propane
The data obtained imply higher activity and lower se-
molecule (a rate determining step in the selective oxidation lectivity in the ODH of propane of the Ni ions in the
reactions (35)) and/or the oxygen abstraction from the tetrahedral environment inherent in the NiCr2O4 spinel, as
solids involved in the formation of carbon oxides.
compared to the octahedral coordination in NiAl2O4. This
On the other hand, the activation energy for the consecu- is in keeping with the general rule stating that ions in an en-
tive propene combustion increases with the Ni content, im- vironment with a low CN are more apt to form a bond with
plying that a propene molecule is more strongly bound to a reacting molecule (an additional ligand) in the chemisorp-
the surface when Mg ions are being replaced by Ni ions. This tion process involved in a catalytic transformation than the
may be due to stronger interaction of an olefinic molecule ions of high CN. Higher activity of oxide catalysts in which
by donation of electrons in the case of Ni (a transition metal ions acting as active centers are present in tetrahe-
metal ion with unfilled d orbitals) than in the case of Mg dral coordination as compared with octahedral has been
ions.
reported for oxidation of CO to CO2 on CuCr, CuFe, and
Table 5 lists TOF values for the propene formation (TOF CuRh spinels (40). In the ODH of propane on systems con-
is the number of propene molecules formed in one second tainingisolated VOx groupsdispersed on different supports,
per one surface Ni atom) on the NiMgAl spinels (after sub- high selectivityand activityofthe tetrahedral vanadium was
traction of the activity of MgAl2O4) and compares them observed (41, 42).
with that for NiCr2O4. The TOF valueswere calculated from
Another reason (besides the Ni coordination) for the
the initialrate ofpropene formation, obtained from the data higher activity and lower selectivity of the Cr-containing
at low conversions (<5% ) by extrapolation. The number of compounds may be the lower oxygen cation’s bonding en-
Ni ions on the catalyst surface (column 5) was calculated ergy as compared with the NiMgAl spinels, as deduced from
taking into account the contributions of the most probable the calculations with the CMSS model. It is generally rec-
crystal faces to the total surface area of the spinels (columns ognized that the oxide systems of low Me–O bond energy
2–4). Fractions of various crystal faces were obtained with show low selectivity to partial oxidation products in the oxi-
the Curie–Wulff construction, assuming the equilibrium dation reactions, the loosely bound and mobile oxygen ions
shape of the crystal as reviewed in Refs. (11, 25). The condi- attackingan organicmolecule at different carbon atomsand
tions of preparation of the samples (high temperature and forming easily carbon oxides (35).
long time of heating) seem to justify this assumption. The
TOF values at a given reaction temperature for the NiMgAl
spinels are very close for Ni0.5Mg0.5Al2O4 and NiAl2O4 and
are lower (particularly at lower reaction temperatures) for
CONCLUSIONS
1. The NiMgAl and NiCr spinels are active and selective
the spinel of the smallest Ni content, Ni0.1Mg0.9Al2O4. It catalysts in oxidative dehydrogenation of propane.
should be, however, observed that for the latter compound 2. In the NixMg1 xAl2O4 system the activity and selec-
the measured activity (the amount of propene formed) is tivity in the ODH of propane increase with the increase
small and only slightly different from that of MgAl2O4, so in the Ni content. It is suggested that the Ni–O centers

!
´
418
SLOCZYNSKI ET AL.
are active in oxidative dehydrogenation of propane to 14. Lenglet, M., D’Huysser, A., and Jørgensen, C. K., Inorg. Chim. Acta
133, 61 (1987).
propene.
15. Lenglet, M., and D’Huysser, A., C. R. Acad. Sci. Paris II 310, 483
3. NiCr2O4 spinel shows higher activity and lower selec-
tivity to propene (at comparable propane conversions) as
compared with the NiMgAl spinels. The comparison of the
(1990).
16. Venezia, A. M., Bertoncello, R., and Deganello, G., Surf. Interface
Anal. 23, 239 (1995).
catalytic behavior of MgCr2O4 and NiCr2O4 with that of 17. Hill, R. J., Craig, J. M., and Gibbs, G. V., Phys. Chem. Miner. 4, 317
(1979).
MgAl2O4 suggests that Cr–O centers in the spinel structure
also participate (though to the smaller extent than Ni) in
the propane–oxygen reactions.
4. The calculations with the CMSS model of the oxy-
18. Inaba, H., Yagi , H., and Naito, K., J. Solid State Chem. 64, 67 (1986).
19. Atanasov, M., Kesper, U., Ramakrishna, B. L., and Reinen, D., J. Solid
State Chem. 105, 1 (1993).
20. Bertaut, F., C. R. Acad. Sci. Paris 230, 213 (1950).
gen cation’s bond energy show that oxygen ions in the 21. Greenwald, S., Pickart, S. J., and Granis, F. H., J. Chem. Phys. 22, 1597
(1954).
Cr-containing spinels are less strongly bound than in the
MgAl systems.
5. It is suggested that the differences in the activity and
selectivity observed for the NiMgAl and NiCr spinels may
22. Davies, M. J., Parker, S. C., and Watson, G. W., J. Mater. Chem. 4, 813
(1994).
23. Binks, D. J., Grimes, R. W., Rohl, A. L., and Gay, D. H., J. Mater. Sci.
31, 1152 (1996).
!
be due to different coordination of Ni ions (confirmed by 24. Zio´lkowski, J., Barbaux, Y., J. Molec. Catal. 67, 199 (1991).
!
25. Zio´lkowski, J., Bordes, E., and Courtine, P., J. Catal. 122,
the XPS measurements): octahedral in NiMgAl and tetra-
126 (1990).
hedral in NiCr systems. The higher activity and lower se-
lectivity of the NiCr spinels may be also due to the lower
oxygen cation’s bond energy in these systems.
!
26. Zio´lkowski, J., J. Solid State Chem. 121, 388 (1996).
27. Chaar, M. A., Patel, D., and Kung, H. H., J. Catal. 109, 463 (1988).
28. Siew Hew Sam, D., Soenen, V., and Volta, J. C., J. Catal. 123,
417 (1990).
29. Michalakos, P. M., Kung, M. C., Jahan, I., and Kung, H. H., J. Catal.
140, 226 (1993).
30. Sloczyn´ski, J., Grabowski, R., Wcislo, K., and Grzybowska-Swierkosz,
B., Polish J. Chem. 71, 1585 (1997).
31. Andersson, S. L. T., Appl. Catal. A: General 112, 209 (1994).
32. Boisdron, N., Monnier, A., Jalowiecki-Duhamel, L., and Barbaux, Y.,
J. Chem. Soc., Faraday Trans. 91, 2899 (1995).
REFERENCES
!
!
´
1. Mazzocchia, C., Aboumrad, C., Diagne, C., Tempesti, E., Herrman,
J. M., and Thomas, G., Catal. Lett. 10, 181 (1991).
2. Mazzocchia, C., Tempesti, E., and Aboumrad, C., US Patent 5,086,032
(1992).
3. Yoon, Y. S., Fujikawa, N., Ueda, W., Moro-oka, Y., and Lee, K. W.,
Catal. Today 24, 327 (1995).
33. Stern, D. L., and Grasseli, R. K., J. Catal. 167, 560 (1997).
34. Sloczyn´ski, J., Appl. Catal. A: General 146, 401 (1996).
!
4. Stern, D. L., and Grasselli, R. K., J. Catal. 167, 550 (1997).
5. Takita, Y., Kurosaki, K., Mizuhara, Y., and Tshihara, T., Chem. Lett.
335 (1993).
35. Bielan´ski, A., and Haber, J., in “Oxygen in Catalysis.” Marcel Dekker,
New York, 1991, and references therein.
!
36. Grabowski, R., Grzybowska, B., Samson, K., Sloczyn´ski, J., and
!
6. Takita, Y., Sano, K., Kurosaki, K., Kawata, N., Nishiguchi, H., Ito, M.,
and Ishihara, T., Appl. Catal. A 167, 49 (1998).
7. Moriceau, P., Grzybowska, B., and Barbaux, Y., Polish J. Chem. 72,
910 (1998).
Wcislo, K., React. Kin. Catal. Lett. 57, 127 (1996).
37. Grabowski, R., Grzybowska, B., Sloczyn´ski, J., and Wcislo, K., Appl.
Catal. A 144, 335 (1996).
!
!
38. Moriceau, P., Grzybowska, B., Barbaux, Y., Wrobel, G., and Hecquet,
G., Appl. Catal. A 168, 269 (1998).
!
8. Zio´lkowski, J., J. Solid State Chem. 57, 269 (1985).
!
!
!
!
9. Zio´lkowski, J., and Dziembaj, L., J. Solid State Chem. 57, 291
39. Grzybowska, B., Sloczyn´ski, J., Grabowski, R., Wcislo, K., Kozlowska,
A., Stoch, J., and Zielin´ski, J., J. Catal. 178, 687 (1998).
40. Ghose, J., and Murthy, K. S. R. C., J. Catal. 162, 359 (1996).
41. Corma, A., Lo´pez-Nieto, J. M., Parades, N., Pe´rez, M., Shen, Y., Cao,
H., and Suib, S. L., Stud. Surf. Sci. Catal. 72, 213 (1992).
42. Concepcion, P., Lo´pez-Nieto, J. M., and Pe´rez-Pariente, J., Catal. Lett.
19, 333 (1993).
(1985).
!
10. Zio´lkowski, J., J. Catal. 100, 45 (1986).
!
11. Zio´lkowski, J., Surf. Sci. 209, 536 (1989).
12. Scofield, J. H., J. Electron Spectrosc. 8, 129 (1976).
13. Lenglet, M., D’Huysser, A., Arse`ne, J., Bonnelle, J. P., and Jørgensen,
C. K., J. Phys. C 19, L363 (1986).