Вы находитесь на странице: 1из 27

ELSEVIER Applied Catalysis A: General 120 (1994) 17-43

Kinetic modeling of the catalytic oxidation of 0-


xylene over an industrial V205-Ti0
2
(anatase)
catalyst
J.N. Papageorgiou, M.e. Abello, G.F. Froment *
Laboratorium voor Petrochemische Techniek, Universiteit Gent Krijgslaan 281 B-9ooo Ghent Belgium
Received 21 December 1993; revised 3 June 1994; accepted 10 June 1994
Abstract
Experiments carried out in an integral isothermal fixed-bed reactor revealed that the selectivities of
some of the products of the catalytic oxidation of o-xylene (mainly o-tolualdehyde, phthalide and
phthalic anhydride) were strongly influenced by the partial pressure of oxygen. A kinetic model
which accounts for the dependence of the selectivities on the composition of the reaction mixture
(partial pressure of oxygen and aromatic compounds) was developed and tested. The experimental
observations and the predictions of the kinetic model are in excellent agreement.
Keywords: Kinetics; Phthalic anhydride; Vanadium oxide-titanium oxide; Xylene oxidation
1. Introduction
Phthalic anhydride, a chemical of considerable importance, is mainly produced
by the catalytic oxidation of o-xylene over V
20s-Ti02
(anatase) catalysts. In
industrial plants, conversions higher than 99% and selectivities up to 77% are
achieved. The most important side-products are o-tolualdehyde, phthalide, maleic
anhydride, and carbon oxides. Some other side-products mentioned in the literature
are mono- and dimethyl maleic anhydride, benzoic acid, toluic acid, benzene, and
anthraquinone [1,21].
The present paper reports on work carried out in an isothermal fixed-bed reactor.
It was observed that the product distribution not only depends on the conversion of
o-xylene, but also on the partial pressure of oxygen. This behavior of the V-Ti-Q
catalyst used in our kinetic studies could not be predicted by any of the kinetic
* Corresponding author. Fax. (+32-9) 2217455.
0926-860X/94/$07.00 1994 Elsevier Science B.V. All rights reserved
SSDI 0926-860X(94) 00 138-H
18 J.N. Papageorgiou et al. / Applied Catalysis A: General 120 (1994) 17-43
models developed up to now. Therefore, a kinetic model which accounts for the
dependence of the selectivities on the composition of the reaction mixture (partial
pressure of oxygen and aromatic compounds) was developed and tested. Inagree-
ment with recent trends in kinetic analysis, the derivation of the model equations
was based upon insight into the physicochemical nature of the process, rather than
upon purely experimental fitting.
Inrecent years, a considerable number of papers dealt with the reaction mecha-
nism, the kinetics and the composition, properties, structure and reactivity of the
catalysts (ref. [16] and references therein). These nowlead to a clearer understand-
ing of the mechanism of o-xylene oxidation over V-Ti-O catalysts, which in turn
proves to be of particular importance for the derivation of both mathematically and
physicochemically adequate kinetic models.
Important aspects to be addressed in the development of the reaction mechanism
are:
(a) the nature of the catalytic active species,
(b) the process by which molecular oxygen adsorbs on the catalyst surface,
(c) the mechanism by which the evolution of lattice oxygen is promoted by the
interaction between VZ05 and TiO
z,
(d) the initiation of the oxidation reaction: how are the aromatic compounds
activated in the vicinity of the catalyst surface and then chemisorbed on the
active sites?
In many surface characterization studies of V- Ti-O catalysts it was observed
that the surface of the catalyst adsorbed moisture which governs the molecular
structure of the surface vanadium oxide species [4-7,11,12,14]. Bond [2] pro-
posed that under reaction conditions the catalyst surface is in a hydrated form
(tetrahedral oxohydroxy vanadium complex), since ample amounts of water from
the catalytic combustion of o-xylene are present during the oxidation reaction. On
the other hand, Deo et al. [4] pointed out that in the temperature range 200 to
500Cthe moisture desorbs fromthe surface of the catalyst, resulting in a dehydrated
environment for the surface vanadium oxide species. As a result of dehydration,
the surface vanadium oxide species directly links to TiO
z
via V-O-support bonds
and also forms terminal V=0 bonds.
Temperature-programmed reduction/ oxidation (TPR/ 0) [22] and isotope
exchange experiments [13] have been conducted to provide insight into the role
of oxygen in V-Ti-O catalysts. In their TPR/O investigation, Went et al. [22]
reported that the terminal oxygens are much more labile (during TPR) than the
bridging oxygens. Upon combining TPR/O and laser raman spectroscopy (LRS)
experiments, they concluded that a maximum of one terminal oxygen per vanadium
is removed during reduction at temperatures up to 500C, while the vanadyl centers
are restored by adsorption of gas-phase oxygen during the TPO process.
Kopinke et al. [13] studied the reoxidation of a V- Ti-O catalyst by means of
the TAP reactor. Isotope exchange experiments revealed that when o-xylene oxi-
dation takes place over a vanadium oxide catalyst in the absence of gas-phase
J.N. Papageorgiou et al. / Applied Catalysis A: General no(1994) 17-43 19
1

0:1-
Scheme 1. Interaction between activated o-xylene and catalyst oxygen [23].
oxygen the products contain both vanadyl and bridging oxygen. The oxidized
catalyst did not show any measurable exchange with gas-phase oxygen, which is
not directly involved in the selective oxidation process. Moreover, it was observed
that the reoxidation of the reduced catalyst by oxygen from the gas phase was a
very fast reaction at elevated temperatures.
The interaction of V-O, with TiO
z
and the evolution of lattice oxygen in V-Ti-
catalysts were studied by Liu et al. [15]. The results of this study indicate that
the evolution of lattice oxygen takes place mainly near the V
z05/TiOz
interface.
The bending of the energy band at the interface of VZ05 and TiO
z
draws the defect
surface state of VZ05 closer to the valence band of VZ05' As a result, the probability
of electron exchange between the defect surface state and the valence band of VZ05
increases and the negative charges on the lattice oxygen ions become easy to
remove, thus promoting the migration of lattice oxygen.
The adsorption and activation of aromatic molecules on the surface of the oxi-
dation catalyst is at least as important as the adsorption and activation of oxygen.
Witko [23], in his review on oxidation of hydrocarbons over transition metal
oxides, is of the opinion that hydrocarbons are activated by H-atom abstraction.
Hydrogen abstraction takes place when reactive catalyst oxygen is available. Pre-
sumably, an oxidized site (V
5
+) is required. for the initiation of the oxidation
reaction through chemisorption of the hydrocarbon by hydrogen abstraction. In the
case of o-xylene oxidation, however, a clear distinction should be made between
hydrogen atoms of the aromatic ring and those of the substituent methyl groups. It
is believed that o-xylene adsorbs irreversibly on the catalyst surface via the methyl
group(s) and not via the ring [20]. Moreover, Witko's quantum chemical calcu-
lations [23] suggest that in o-xylene oxidation the plane of the aromatic ring is
parallel to the catalyst surface and the attack of catalyst oxygen on the -CH
z
group
of activated o-xylene occurs along the axis passing through the -CH
z
group per-
pendicular to the ring (the attack from below the plane) . The complex thus produced
is a tolualdehyde-like product with a c-o bond distance typical of aldehydes
(Scheme 1).
20 J.N. Papageorgiou et al.r Applied Catalysis A: General 120 (1994) 17-43
Van Hengstum et al. [20] studied the oxidation of o-xylene by means of in situ
Fourier transform IR (Ff-IR). On the basis of the surface intermediate species that
were identified on a V-Ti-Q monolayer catalyst after adsorption/oxidation of 0-
xylene, the authors reached the following conclusions:
(a) The presence of adsorbed water, probably in relatively small amounts on the
surface of the catalyst, has no effect on the adsorption behavior of the aromatic
compounds.
(b) During the selective oxidation, the aromatic nucleus remains unaffected.
(c) Adsorption of o-tolualdehyde does not lead to bands associated with the C=O
stretching vibration of the free aldehyde.
(d) Both the carbonyl group and the hydrogen atom of the aldehyde interact with
the surface of the vanadium oxide catalyst.
(e) Since an aldehyde molecule already contains an oxygen atom, the initiation
of the oxidation of such a species requires only a single vanadium-oxygen
site.
(f) Cyclic complexes (i.e. phthalide, phthalic anhydride, maleic anhydride) are
bonded to the surface of the catalyst through one of their carbonyl groups
(analogous to the coordinatively adsorbed aldehyde species).
2. Experimental
The catalyst was an industrial V205-Ti0
2
(anatase) on SiC catalyst of the
Japanese type with a V
205
content of 3.5 wt.-%!n the active phase.
The kinetics of the catalytic oxidation of o-xylene were studied in an integral
fixed-bed reactor operated at nearly atmospheric pressure and in the temperature
range of 335-365C. The reactor was made of a stainless steel tube, 1 inch in
internal diameter and 0.7 m in length. In order to avoid too much channeling along
the reactor wall, the size of the catalyst particles was reduced to a diameter of 0.1
inch. Isothermal reactor operation was achieved by mixing the catalyst particles
with SiC beads of the same size and immersing the reactor in a salt bath provided
with electrical heating rods and a stirrer. Several packing layers were prepared by
diluting the catalyst with SiC in different dilution ratios satisfying the dilution
criterion of Van den Bleek et al. [19]. Since the amount of heat released decreases
with conversion (i.e. with bed depth), the reactor was loaded with the packing
layers arranged in a decreasing order of dilution along the reactor length (e.g. 80%,
75%, 70%). The axial temperature of the reactor was measured by a thermocouple,
sliding in a coaxial thermowell of 0.08 inch external diameter. Preliminary exper-
iments revealed that flow rates beyond 100 Nl/h ensure global rates entirely deter-
mined by the rate of reaction and not by the rate of mass transfer.
o-Xylene was fed at flow-rates in the range of 3.5 to 14 ml/h by means of a
Spectra Physics Iso Chrom LC pump. The feed gas consisted of mixtures of air,
nitrogen and oxygen. The flow of the individual components was measured and
I.N. Papageorgiou et al. / Applied Catalysis A: General 120 (1994) 17--43 21
controlled by Brooks mass flow controllers. The total flow-rate through the reactor
was varied between 150 and 400 Nl/h.
The reactor effluent was analyzed on-line in a Packard 438A gas chromatograph
equipped with a fused silica WCOT capillary column and a flame ionization detec-
tor. Benzoic acid, citraconic anhydride, maleic anhydride, phthalide, phthalic anhy-
dride, tolualdehyde and toluic acid were identified. The selectivities of benzoic
acid, toluic acid and citraconic anhydride always remained well below 1 mol-%.
The concentrations of carbon monoxide and carbon dioxide in the effluent were
measured by a Teledyne IR detector.
Some 180 experiments were carried out at 3 temperatures (339, 350 and 361C),
4 inlet concentrations of o-xylene (0.4, 0.6, 0.8 and 1.0 mol-%) and 3 partial
pressures of oxygen (0.04,0.20 and 0.37 atm). The overall conversion of o-xylene
varied between 15 and 85 mol-%. The fresh batches of catalyst were equilibrated
by performing the same experiment until a stable composition ofthe reactor effluent
was reached. For the flow-rates under consideration some 50 h of pretreatment were
required. Between experiments the catalyst was kept in an air flow at 340-360C.
3. A kinetic model for PA synthesis considering both oxidized and reduced
sites (ORES)
The development of the ORES model was based on the reaction scheme given
in Scheme 2.
Two types of catalyst sites were considered: reduced (I) and oxidized (1
o
) sites,
with the latter (i.e, 10 sites) being active for oxidation of the adsorbed species. The
reduced catalyst sites are reoxidized by adsorption of gas-phase oxygen or by lattice
oxygen (0*) migration. Phthalic anhydride synthesis from o-xylene proceeds
according to a 'rake' mechanism with o-tolualdehyde and phthalide as major inter-
mediates [2]. o-Xylene (X) is activated by hydrogen abstraction from one of its
methyl groups and chemisorbs on an oxidized catalyst site (1
o
) ' The resulting
surface intermediate X-1
o
interacts with a neighboring 10 site and yields tolualdehyde
and water through a second hydrogen-abstraction and oxygen-insertion steps. Tolu-
aldehyde (T) is then chemisorbedas a T-I surface intermediate which is transformed
into phthalide upon interacting with an adjacent 10 site. Through similar adsorption,
surface oxidation and desorption steps, phthalide is oxidized towards phthalic
anhydride. Adsorption of tolualdehyde on 10 sites, in a way presumably similar to
that of o-xylene, was also considered. Oxidation of the surface intermediate T-I
o
thus produced leads to phthalic anhydride. For the total oxidation products (carbon
oxides and maleic anhydride) and on the basis of the experimental observations, it
was assumed that they are formed exclusively from o-xylene, the total oxidation
reaction(s) occurring in parallel with the selective oxidation paths. The decom-
position of tolualdehyde, phthalide and phthalic anhydride is neglected under the
operating conditions investigated in this work.
22 J.N. Papageorgiou et al. / Applied Catalysis A: General 120 (1994) 17-43
rr=T-Io
+10
+10
X+ lo l + T P + 1 PA + 1
3
11
II 11
j1
X-lo
1
T-l
2
pol
4
PA-l
- -
+10 +10 +10
+
10
1
5
Col
-
C + l
-
2l-0
d + I I
o
Scheme 2. Reaction scheme of o-xylene oxidation on a V-Ti-Qcatalyst.
The model includes the following steps [9]:
Dissociative chemisorption of oxygen:
leo
I + !Oz ~
kor
Non-dissociative chemisorption of oxygen:
ka
AC + 1 ~ AC-I
Adsorption of aromatic compounds AC = X, Ton 1
0
sites:
K
A
= (AC-/o )
PAC(lo)
Adsorption of aromatic compounds AC = T, P, PA on I sites:
K
B
= (AC-/)
PAdl)
J.N. Papageorgiou et al. / Applied Catalysis A: General 120 (1994) 17-43
Oxidation reactions of adsorbed compounds
k{
X-/o + 1
0
T-I + I rl =

T-I + 1
0
P-I + I r2 = k;(T-/)(/
o)

T-/
o
+ 1
0
PA-I+I r3 = k;(T-Io)(/
o)

P-I + 1
0
PA-I + I r
4
= (/0)

X-/o + 1
0
C-I + I rs = k5(X-/
o)(/o)
Reoxidation of the reduced sites by lattice oxygen:
23
I + O' 1
0
kmr
The total concentration of active sites is:
C
t
= (/) + (/0) + (X-/
o)
+ (T-/
o)
+ (T-l) + (P-/) + (PA-/)
The concentration of C-I intermediates is assumed to be negligible, due to fast
desorption into the bulk gas phase.
Considering various rate determining steps evidently results in various model
equations. Two cases will be developed here:
(i) the reoxidation of the reduced sites is rate controlling (ORES-ROS),
(ii) the aromatic-compound oxidation reactions are the rate determining steps
(ORES-AOR).
3.1. Rate determining step: reoxidation ofsites (ORES-ROS)
The catalyst sites are reoxidized by:
(a) dissociative chemisorption of gas-phase oxygen,
(b) non-dissociative chemisorption of gas-phase oxygen, and by
(c ) interaction of lattice oxygen with the reduced sites.
Each of the above types of reoxidation could be the rate determining step, so that
three different sets of rate equations could be derived. Here, only the derivation of
the rate equations whereby the reoxidation by lattice oxygen is rate controlling is
presented. Choosing another rate determining step [i.e. (a) or (b)] results in rate
equations similar to the ones to be presented here. The degree of similarity is such
that the resulting models cannot be discriminated unless the concentrations of I and
1
0
catalyst sites, lattice oxygen and relevant surface intermediates were available.
From a mechanistic point of view, the case to be presented here is the most likely.
Ifthe reoxidation ofthe catalyst I sites is assumed to be rate controlling, additional
attention should be paid to the following aspects:
24 J.N. Papageorgiou et al. / Applied Catalysis A: General 120 (1994) 17-43
The reoxidation ofthe catalyst and the surface oxidation reactions occur in series.
The former is a common elementary step for the various oxidation reactions which
proceed at different rates. Therefore, the reoxidation of l sites could be rate deter-
mining only if its rate would be different for active sites involved in different
surface oxidation reactions. When the reoxidation by lattice oxygen is rate con-
trolling:
kmi
l + 0* /0
kmri
K
k
m i
I 5
mi = k.' I = ,... ,
mn
Obviously, in the opposite case it would be:
rl = rz = r3 = r4 = r5 = km(l)(O*)
and presumably rp = rZ-r4 = 0, so that phthalide would not be isolated as an
intermediate.
A steady-state balance on lattice oxygen results in:
kaP02 + JP02 r}mri
(0*) (1)
Evidently, in the steady state, the dissociative and the nondissociative chemi-
sorption of molecular oxygen reach a pseudo-equilibrium. The concentration of
lattice oxygen is then obtained from:
(
0 * ) = K
a
rp
Vr0
2
o
(2)
Eq. (I) and Eq. (2) are not incompatible; since the reoxidation of the catalyst
by lattice oxygen is the slow step,
k; "kmri and ; kmi
and, therefore, Eq. (1) can be safely approximated by Eq. (2).
With the reoxidation of the catalyst by 0 * rate controlling, the rates of con-
sumption and/or production of the aromatic compounds are given by the formula:
i = 1,... ,5 (3)
Eq. (3) gives the rate equations of the ORES-ROS model. For a reactor with
plug flow, the selectivities of the products are given by the relations:
(5)
J.N. Papageorgiou et al. / Applied Catalysis A: Generall20 (1994) 17-43 25
The overall selectivity relations (Eqs. 4a-d) imply that the selectivities are
constant, irrespective of the level of conversion and the partial pressures of the
reactants. Point selectivities are obtained from the ratios of the rates of production
of T, P, PA and C to the rate of consumption of o-xylene. It can be seen form Eq.
(3) that the kinetic equations of the ORES-ROS model all have the same driving
force group: the square root of the partial pressure of oxygen. This is not surprising
since in all reaction paths i = I, ... ,5 the same elementary step, i.e. the reoxidation
of the catalyst, was assumed to be rate controlling. Also, since the adsorbedaromatic
compounds are not involved in the rate determining steps, their partial pressures
are evidently not included in the driving force group. Furthermore, the adsorption
group (denominator of Eq. 3) is the same for all rate expressions, because it is
related to the total number of active sites. In conclusion, the point selectivities (Spi'
i = T, P, PA, C) are functions of the kinetic groups only, so that they are inde-
pendent of the conversion. Since oxygen is present in excess, its partial pressure is
considered to remain nearly constant. Therefore, the overall selectivities (Si' i =
T, P, PA, C) and the point selectivities (Spi' i = T, P, PA, C) should be identical.
It was experimentally observed, however, that the selectivities of tolualdehyde,
phthalide and phthalic anhydride depend on both the conversion and the partial
pressure of oxygen. Therefore, the ORES-ROS model presented here is not ade-
quate.
3.2. Rate determining step: aromatic-compound oxidation reactions (ORES-
AOR)
If the aromatic-compound oxidation reactions are assumed to be rate controlling,
the rates of consumption of o-xylene and production of tolualdehyde, phthalide,
phthalic anhydride and total oxidation products are given by:
(k
1
+ k
S)PXP0 2
r -
x- DEN
26 J.N. Papageorgiou et al. / Applied Catalysis A: General 120 (1994) 17-43
(6)
(7)
(8)
(9)
with:
DEN= [1 +K
oVP02
+K
A(PX+PT)VP02
+KB(PT+Pp+PPA)]Z
k
1
=k1'K
A'
(mol g-l h-
I
atm-
z)
k
z
=k
z
' K
B
' (mol g-I h-1 atm-312)
(mol g-I h-
I
atm-
z)
(molg-
1
h-
1
atm-
312
)
(mol g-l h-
1
atm-
z)
(atm-
3
/
Z
)
KB=K
B'
(atm"")
(atm-
1
/
Z
)
From a comparison of the rate equations of the ORES-ROS model (Eq. 3) with
the ones of the ORES-AOR model (Eq. 5-Eq. 9), the following is deduced:
(a) The driving force group in the kinetic relations of the ORES-ROS model is
the square root of the partial pressure of oxygen, whereas, in the ORES-AOR
model, the driving force group is the product of the partial pressure of oxygen (to
either the 0.5 or the 1.0 power) and the partial pressure of the aromatic compound
which is involved in the respective surface oxidation reaction. This difference in
the driving force groups results from the difference in the rate determining steps.
The power of the partial pressure of oxygen in the driving force groups is not the
same for the two models. In the ORES-ROS model, the concentration of the lattice
oxygen is obtained from (Eq. 2) so that the square root of the partial pressure of
oxygen appears as the driving force group in the model equations. In the case of
the ORES-AOR model, however, the driving force group is either of half or first
power with respect to the partial pressure of oxygen, since adsorption of aromatic
compounds on I and/or 1
0
sites were considered (Scheme 2) . For a surface reaction
that proceeds by interaction between an aromatic compound adsorbed on an oxi-
dized site and an adjacent oxidized site (i.e. paths 1,3,5), the partial pressure of
J.N. Papageorgiou et al. / Applied Catalysis A: General 120 (1994) 17-43 27
oxygen appears in the driving force groups of the respective rate equations (i.e. rl'
r3' rs). Aromatic-compound oxidation reactions 2 and 4 on the other hand, involve
an aromatic compound adsorbed on a reduced site (T-I, P-I) which interacts with
an adjacent oxidized site (1
0
) , Therefore, in this case, the square root of the partial
pressure of oxygen appears in the driving force group of rz and r4'
(b) The adsorption group is not the same in the two models. Inthe ORES-ROS
model, the rate determining step is the reoxidation of the reduced catalyst sites 1by
lattice oxygen. The concentration of 1sites is obtained from the balance on the total
number of catalyst sites and the concentration of lattice oxygen is obtained from
Eq. (2). Inthe case of the ORES-AOR model, however, the rate determining step
is the interaction between an adsorbed aromatic compound and an oxidized catalyst
site (1
0
) , Here, the balance on the total number of catalyst sites is used twice to
derive the final rate expressions. Therefore, the denominator in the ORES-AOR
model equations is raised to the power two.
With the ORES-AORmodel, the selectivity relations for a reactor with plug flow
are given by:
1 kzJ(PT/DEN) dr- k4J(Pp/DEN)dr
Sp=----------------
VP
oz
k3J(PT/DEN)dr+ J(Pp/DEN)dr
vPaz
SPA =--------------
(lOa)
(lOb)
( lOc)
r=:; (space time) (lOd)
The ORES-AOR model accounts for the dependence of the selectivities of tolu-
aldehyde, phthalide and phthalic anhydride on the partial pressure of oxygen. Eq.
( lOa) clearly reveals that the selectivity of tolualdehyde increases with the partial
pressure of oxygen. Eq. (lOc) also reflects that the selectivity to phthalic anhydride
decreases with the partial pressure of oxygen. Eq. (lOd) implies that the selectivity
for combustion products is independent of the composition of the reaction mixture.
This results from the fact that the driving force groups for the overall consumption
28 J.N. Papageorgiou et al. / Applied Catalysis A: General 120 (1994) 17-43
1

.. .
I
i
k.
k,
k.
~
---...
a
b
K.
l
I
(1fT)' l e' . T
l ~ t ~ ~ , e Ie' l ~ 1 ~ l ~
(1fll l O'. T(
Fig. J. o-Xylene conversion vs. space time at three different partial pressures of oxygen (isothermal operation,
plug flow). Curves: kinetic model, points: experimental data. (T = 339C, inlet concentration of 0-xylene = 0.4
mol-%).
of a-xylene and the production of combustion products are identical and, therefore,
are eliminated from the selectivity relations. Interms of reaction mechanism, this
behavior of the catalyst is attributed to the assumption that only a-xylene is totally
oxidized on the catalyst surface. If tolualdehyde and/ or phthalide and/ or phthalic
anhydride were also converted to carbon oxides and maleic anhydride, the selec-
tivity relation of the combustion products would be a function of the partial pres-
sures of these aromatic compounds and consequently of the conversion. The
selectivity to total oxidation products would then increase with the conversion.
Moreover, at high levels of conversion, phthalic anhydride selectivity would
decrease because of its total oxidation, in contrast with the experimental observa-
tions. Therefore, tolualdehyde, phthalide and phthalic anhydride were assumed to
be stable and not subject to total oxidation under usual operating conditions.
The ORES-AOR model predicts the influence of the partial pressure of oxygen
on the selectivities which is in agreement with the experimental results. The situ-
ation is not as clear for the selectivity of phthalide. Nevertheless, it is established
that the maximum in the selectivity of phthalide decreases upon increasing the
partial pressure of oxygen.
4. Results and discussion
Since the ORES-AOR model revealed trends similar to those experimentally
observed, the adequacy of this model was tested by fitting the model equations Eqs.
J.N. Papageorgiou et al. / Applied Catalysis A: General 120 (1994) 17-43 29
(5)-(8) to the experimental data. The parameter estimation procedure [8] also
involved the temperature dependence of the model parameters (5 rate constants
and 3 adsorption equilibrium coefficients). Therefore, parameter estimation per
temperature and parameter estimation based on all temperatures [10,18] were
carried out, so as to obtain the pre-exponential factors, the activation energies of
the oxidation reactions and the heats of adsorption of aromatic compounds and
oxygen on the catalyst surface.
Figs. 1 to 5 are experimental and calculated conversion versus space time plots
for various partial pressures of oxygen. All the conversions increase with increasing
partial pressure of oxygen. Indeed, at higher partial pressures of oxygen, the con-
centration of the oxidized catalyst sites (1
0
) is higher. Since tolualdehyde and
phthalide are intermediates, their conversions exhibit a maximum with respect to
space time. The maximum of the conversion to tolualdehyde is greater than the one
for phthalide and appears at relatively lower xylene conversions (or smaller space
times), thus, indicating that tolualdehyde is the first intermediate in the reaction
network of Scheme 2. Phthalic anhydride and total oxidation products (carbon
PC\- O.37111m
- - - oQo- --
PO, - O.04111m
---8-
PO,- O.20111m
o
Ql
e "C
>-
J:.


I
I
B
I
0
I
0
-
I
.Q!
4 --i

6
'0
c
.Q
I!!
Ql
>
c
8
2
o -f-----,--- --r-- -
o 500 1000 1500 2000 2500 3000 3500
Space Time, W/FO (gr cat*hr/mo l o-xylene)
x
Fig. 2. Conversion of o-xylene to tolualdehyde vs. space time. (T = 339C, inlet concentration of o-xylene =
0.4 mol-%).
30 J.N. Papageorgiou et al. / Applied Catalysis A: Generall20 (1994) 17-43
5 -.,..---
4
o
Pc; - 0.04alm
- -6
Pc; - 0.2OeIm
P,\ - 0.37alm
---A---
O+ - - --,-- - --,-- - - r-- - ,...-- - -.-- - -,-- - -
o 500 1000 1500 2000 2500 3000 3500
Space Time, W/r:. (gr cat*hr/mol o-xylene)
Fig. 3. Conversion of a-xylene to phthalide vs. space time. (T = 339C, inlet concentration of a-xylene = 0.4
mol-%).
oxides and maleic anhydride) are final products and, therefore, their conversion
monotonically increases with space time. The fitting shown in Figs. I to 5 is
excellent.
The selectivity versus conversion plots of Figs. 6 to 9 show how the proposed
model accounts for the dependence of the selectivities on the partial pressure of
oxygen. The toluadehyde selectivity increases with the partial pressure of oxygen.
The oxidation of 0-xylene to tolualdehyde proceeds by interactionbetween0-xylene
adsorbed on an oxidized site (X-/
o
) and an oxidized catalyst site (10)' Evidently,
by increasing the partial pressure of oxygen, the concentration of /0 increases and
more o-xylene is converted to tolualdehyde. After adsorption on either 1or 1
0
sites,
tolualdehyde is converted to phthalide and phthalic anhydride according to reaction
Scheme 2. At elevated partial pressures of oxygen, less 1 sites are available for
adsorption and the oxidation of tolualdehyde into phthalide is suppressed. On the
other hand, the direct oxidation of tolualdehyde to phthalic anhydride (path 3) and
the oxidation of o-xylene along paths I and 5 have the same driving force groups
with respect to the partial pressure of oxygen. Therefore, the partial pressure of
J.N. Papageorgiou et al. / Applied Catalysis A: General 120 (1994) 17-43
eo -,-----------------------,
Poa- O.c.1m
---8---
31
50
10
Poa- 0.2OIIIm
-_-G- --

-._ .0... -
o
1Po,

a 500 1000 1500 2000 2500 3000 3500
SpaceTime, W/Ff (grcat*hr/mol o-xylene)
Fig. 4. Conversion of a-xylene to phthalic anhydride vs. space time. (T = 339C. inlet concentration of o-xylene
= OAmol-%).
oxygen has no effect on the selectivity of path 3. Inconclusion, the oxidation of 0-
xylene to tolualdehyde is favored at higher partial pressures of oxygen, whereas
the oxidation of tolualdehyde is not. As a result of this, the selectivity for tolualde-
hyde increases with the partial pressure of oxygen (Fig. 6). Since tolualdehyde is
the first intermediate, its selectivity reaches a maximum at very low conversions.
Fig. 7 reveals that the phthalide selectivity maxima at partial pressures of oxygen
of 0.20 atm and 0.37 atm are reproduced by the model, while the maximum in
phthalide selectivity decreases with the partial pressure of oxygen and appears at
higher conversions. Indeed, at higher partial pressures of oxygen, the concentration
of reduced catalyst sites decreases. Since these sites are involved in the paths that
are responsible for the net production of phthalide (i.e. paths 2 and 4) , the produc-
tion of phthalide is not favored by an increase of the partial pressure of oxygen. In
addition, the trend indicated by the phthalide selectivity curves is determined by
the combined influence ofthe partial pressure of oxygen and the level ofconversion.
Depending upon the concentrations of tolualdehyde and phthalide, the selectivity
for phthalide either increases or decreases with the partial pressure of oxygen. At
32 J.N. Papageorgiou et al.1Applied Catalysis A: General 120 (1994) 17-43
20---.-------------------------,
Poa- 0.004elm
---B---
Po,- 0.2OeIm
:
~ -
~
"
/
Poa- 0.37etm
"
15
--._,A,_.. -
e
0
Q.
0
c: 0
.2
~
1
.s(
0
I 10
S
c:
,S!
II!
~
0
8
!
0
~
5
,
0
O-+-----,------,r-----,----,-----,---,-----j
o 500 1000 1500 2000 2500 3000
SpaceTime, Wlfx(gr cat*hr/rnol o-xylene)
Fig. 5. Conversion of o-xylene to total oxidation products vs. space time. (T = 339C, inlet concentration of 0-
xylene = 0.4 mol-%).
lower conversions, the concentration of tolualdehyde is higher. In this region, the
production of phthalide along path 2 (rate rz) dominates over its consumption
along path 4 (rate r4) and, therefore, the phthalide selectivity decreases with the
partial pressure of oxygen (Eq. lOb and Fig. 7). At higher conversions, the partial
pressure of phthalide still increases, whereas the tolualdehyde concentration
decreases. In this region, rate r4 dominates and, therefore, the selectivity for phthal-
ide increases with the partial pressure of oxygen (Eq. lOb and Fig. 7).
Fig. 8 illustrates the influence of the partial pressure of oxygen on the selectivity
for phthalic anhydride. An increase of the partial pressure of oxygen does not favor
the production of phthalic anhydride from phthalide (path 4 in Scheme 2), because
the catalyst has less reduced sites 1available for phthalide to be adsorbed on. Also,
since phthalic anhydride is a final product in reaction Scheme 2, its selectivity keeps
on increasing with the conversion.
Fig. 9 shows the selectivity of the total oxidation products versus the o-xylene
conversion. In this case, the partial pressure of oxygen and the conversion have no
effect on the shape of the selectivity curve, as already explained. The horizontal
' 0
J.N. Papageorgiou et al. / Applied Catalysis A: General 120 (1994) 17-43 33
20 10 100
o-Xylene conversion
Fig. 6. Tolualdehyde selectivity vs. o-xylene conversion. (T = 339C, inlet concentration of o-xylene = 0.4 mol-
%).
line reveals that a fixed percentage (23.5%) of o-xylene is converted into carbon
oxides (CO, CO
2
) and maleic anhydride.
The influence of the partial pressure of oxygen on the selectivities of the products
obtained from the catalytic oxidation of o-xylene (T, P, PA, C) could be summa-
rized as follows. The oxidation of the aromatic compounds of the reaction mixture
proceeds after adsorption on either I and/or 1
0
sites. The surface intermediates
resulting from the adsorption of the aromatic compounds on the catalyst surface
are oxidized by interaction with a neighboring oxidized catalyst site 1
0
, By increas-
ing the partial pressure of oxygen, a larger fraction of the catalyst sites is oxidized.
Therefore, the surface reactions proceeding after adsorption on I sites (i.e. paths 2
and 4 in Scheme 2) are disadvantaged.
Fig. 10 reveals a weak influence of the inlet concentration of o-xylene on the
conversion. This leads to an excess of catalyst sites with respect to the number of
adsorbed aromatic molecules. To saturate the catalyst surface with aromatic mol-
ecules the inlet concentration of o-xylene would have to be in the explosion region.
34 J.N. Papageorgiou et al.1Applied Catalysis A: General 120 (1994) 17-43
20
"",", 004 _
0 -
I
I
" ~
o
"",", 037 _
I
I I 0
!.I ~
ill
II
~
.il':
U
~
Ql
Ul
~ '0
~
a..
o
20 10 10 '00
a-Xylene conversion
Fig. 7. Phthal ide selectivity vs. o-xylene conversion. (T=339C, inlet concentration of o-xylene = 0.4 mol-%).
It follows from Figs. 1 to 10 that the fit of the ORES-AOR model (curves) with
respect to the experimental observations (points) is excellent. Table I shows the
F- and t-values at 339C, indicating the adequacy of the model and the significance
of its parameters, respectively. After parameter estimation at 339, 350 and 361C,
the temperature dependence of the model parameters was determined. The Arrhe-
nius plots for the rate constants and the Van't Hoff plots for the adsorption equilib-
rium coefficients of the model are presented in Figs. lla and b, respectively. From
these plots, the frequency factors, the activation energies and the heats of adsorption
were calculated. The results are given in Table 2.
Parameter estimation based on the data at all temperatures was also carried out.
By reparameterizing the pre-exponential factor as outlined in [10,18], the activa-
tion energies, the heats of adsorption and the frequency factors were estimated in
one step. The results are presented in Table 3 and they are in agreement with those
given in Table 2. The statistical tests resulted in satisfactory F- and r-values, Finally,
Fig. 12 illustrates the parity plot of the overall conversion of a-xylene after param-
eter estimation based on all temperatures.
J.N. Papageorgiou et al. / Applied Catalysis A: General 120 (1994) 17--43
eo -,------
35

>

QI
Qj
Ul
QI
'0
' 1:
'0

I:
III
I
o,
20 l

I o.04 81m
---9-
I 0.37 aim

to
to
0;>
0
"v
to
,P
to
"
6
o -l-- - - -.,.- - - - r-- - - -.---- - - -,- - - ----j
o 20 eo 100
o-Xylene converslon
Fig. 8. Phthalic anhydride selectivity vs. o-xylene conversion. (T = 339C. inlet concentration of o-xylene = 0.4
mol-%).
For the activation energies obtained after parameter estimation per temperature
(Table 2) and those based on all temperatures (Table 3), the following may be
concluded:
(a) The highest activation energy is that of the total oxidation of o-xylene into
carbon oxides and maleic anhydride. It can indeed be expected that the energy
barrier for the destruction of the aromatic ring is higher than that for the selective
oxidation paths 1, 2, 3, 4 of reaction Scheme 2.
(b) Of the selective oxidation reactions, the one with the highest activation
energy is the oxidation of o-xylene to tolualdehyde (path 1). Since tolualdehyde
and phthalide already include an oxygen atom in their molecule, their further
oxidation by insertion of another oxygen atom is likely to be easier. Furthermore,
the direct oxidation of tolualdehyde to phthalic anhydride (path 3) and the oxidation
of phthalide to phthalic anhydride (path 4) have comparable activation energies,
presumably due to a phthalide-like intermediate formed along both these oxidations
routes. The lowest activation energy was obtained for the oxidation of tolualdehyde
to phthalide (path 2), thus implying that, after cyclization to a five-membered ring,
oxygen insertion in the carbon atom of the -CH
2
- group of this ring (path 4) is
more difficult.
36
J.N. Papageorgiou et al. / Applied Catalysis A: General 120 (1994) 17-43
50
r-
- ---
1

020_
40
037_

..
"E
Q.
C
30
a
.."
s
.c
0
" " " E
"
" 0"
a
0
"
0
u 0
0
6
0
... 0 CO
a
20
0
-

.i!
U
<ll
(jj
en
10
0 -+-- - - - - .,.-- - - - --,,.-- - - - ---,-- - - - - -
o 20 eo eo
a-Xylene conversion
Fig. 9. Selectivity for combustion products vs. o-xylene conversion. (T = 339C, inlet concentration of o-xylene
= 0.4 mol-% ).
A kinetic model which is consistent with the reaction mechanism should satisfy
well-established physicochemical laws. Boudart et al. [3] derived constraints on
the adsorption enthalpies and entropies. Since adsorption is exothermic, the adsorp-
tion enthalpy has to satisfy the inequality:
and the adsorption entropy has to satisfy

( lla)
(lIb)
is the standard entropy of adsorption and is the standard entropy of
the molecule in the gas phase. Indeed, for a non-dissociative chemisorption, a
reduction in entropy occurs when a gaseous molecule transfers from a three-dimen-
sional phase (the gas phase) to a two-dimensional phase (the catalyst surface).
Another criterion, which is based on the change in volume that occurs when a
gaseous molecule is adsorbed, is the following:
41.8 < < 51.04 + 0.OOI4( (kJkmol-
1
K-
1
) ( lIc)
J.N. Papageorgiou et al. / Applied Catalysis A: Generall20 (1994) 17-43 37
70
Px0. O.llmol%
......
Px. 0.4mo1%
. s .
Px. O.llmol%
Px. O.llmol%
Px. 0.8mo1%
Px. 0.4mol%
0
110 - :
~ ~
40
0
20
10
o
c
o
'j!!
~
8
~
QI
~ 3
6
o 500 1000 1500 2000 2500 3000
Space Time, W ~ (gr cat*hr/mol o-xylene)
Fig. 10. a-Xylene conversion vs. space time at three different inlet concentrations of o-xylene. Curves: kinetic
model, points: experimental data. (T = 339C, partial pressure of oxygen = 0.20 atm) .
Table 1
ORES-AOR model; parameter estimation by non-linear regression; T = 339C, F-value = 6102.18, correlation
coefficient = 0.9935
Parameter Estimated values Approx. 95% confid. limits t-value
Lower Upper
k, ( mol g-I h-
'
atm-
2
)
7.23 6.57 7.88 22.12
k
2
(molg-
t
h-
t
atm-I.' )
17.02 15.04 19.01 17.19
k
3
(mol g" ' h-
'
atm-
2
) 41.80 35.62 47.97 13.54
k
4
(molg -I h-
I
atm- I.' )
23.85 20.54 27.15 14.42
k, (molg-
t
h-
'
atm" ) 2.25 2.04 2.46 21.58
s; (atm -
a
., )
3.38 2.89 3.87 13.78
K
A
( atm" !" ) 1026 903.1 1148 16.75
K
B
(atm" ') 860.3 661.4 1059 8.65
38 J.N. Papageorgiou et al. / Applied Catalysis A: General 120 (1994) 17-43
Hi al
4 .....
a
k
3
0
3.5
- ~ -
k.
- - -- --A___
k
2
a
- -
3 -i
0 0
0
-
~
2.5 -i
c:
, J
..
k
1.5
. _. 5
..

0.5
1.57 1.511 1.58 i .e l. e1 1.82 1.83 1.84
(1fT)*10
3
, T(I<)
Boudart's criteria (Eq. l l a-c) refer to the non-dissociative chemisorption of a
gaseous molecule on the catalyst surface. Before applying them to the adsorption
equilibrium coefficients of the ORES-AOR model, additional attention should be
paid to the following aspects:
( a) K; was defined as the adsorption equilibrium constant of the dissociative
chemisorption of oxygen on the catalyst surface. During dissociative chemisorption
the structure of the diatomic oxygen molecule is disturbed. This could lead to a
positive change in the entropy, in contrast with Boudart 's criterion of Eq. (Llb).
Data concerning the bond strengths between the oxygen atoms of the oxygen
molecule and between the vanadia and oxygen atoms of the catalyst lattice are
required, so as to provide an indication for the change in the entropy of this
adsorption.
(b) K
A
is the product of the adsorption equilibrium constant of xylene and
tolualdehyde K
A
' (adsorption on /0 sites) and the equilibrium coefficient of the
dissociative chemisorption of oxygen K
o
Therefore:
L1 ~ A = L1 SOaN +L1 ~ . o
L1 H?.A = L1H?.AI +L1H?,0
J.N. Papageorgiou et al. / Applied Catalysis A: General 120 (1994) 17-43 39
8
-- ----- ---
I
I
I
71
K
A

K
B
6
~
E
2 I-
-- -----
b

1.64 1.63 1.62 1.61 1.6


I
1.sg 1.58 1.57
O ~ r _ _ r r _ r
(1fT)*101, T(K)
Fig. 11. (a) Arrhenius plot for the rate constants of the ORES-AORmodel. (b) Van't Hoff plot for the adsorption
equilibrium coefficients of the ORES-AOR model.
Table 2
ORESAOR model; temperature dependence of rate coefficients and adsorption equilibrium constants
Parameter Frequency factor E, (kJkmo1-
1
)
k, (mol g-l h-
1
atm-
2
) 185.4.10
6
86.81,10
3
k
2
(mol g-l h-
1
atm "!") 0.407,10
3
16.06,10
3
k
3
(mol g-l h-
1
atm-
2
) 40.40,10
3
34.93 .10
3
k
4
(mol g-l h-
1
atm-1.5)
50.3010' 38.89'10
3
k
5
(mo1g-
1
h-
1
atm-
2
) 43.10,10
9
120.6.10
3
Parameter Pre-exponential factor -.i!Jt;. (kJ kmol"")
s; (atm-
O
.
5)
0.638 8.502,10
3
K
A
(atm-1.5)
2.669 30.37 .10
3
K
B
(atm-
1
) 0.188 42.95,10
3
40 J.N. Papageorgiou et al.1Applied Catalysis A: General 120 (1994) 17-43
Table 3
ORES.AOR model. Estimation of frequency factors. activation energies and adsorption enthalpies from the data
at all temperatures. F-value = 9927.31. Correlation coefficient = 0.9928
Parameter Estimated values Approx. 95% confid. limits t-value
A
Lower Upper
k, 14 )(f 13 )(f 15.10
6
41.59
k
2
80 74 85 29.16
k
3
8.6W 7.9W 9.3 .10
3
24.09
k. 19'10
3
17.10
3
21'10
3
25.39
k
s
19.10
9
18'10
9
20.10
9
40.69
s, 2.2 2.0 2.4 23.21
K
A
12 II 13 30.75
K
B
40.10-
2
37.10-
2
43.10-
2
25.12
k,
72.1,10
3
66.8.10
3
77.4.10
3
27.19
k
2
6.5610' 5.89W 7.23 .10
3
19.84
k
3
25.5 .10
3
22.6.10
3
28.3 10' 17.79
k. 32.410' 25.1.10
3
39.8W 8.79
k
s
11510' 107 '10
3
123.10
3
28.01
s; 3.32,10
3
2.73 10' 3.91,10
3
11.26
K
A
23.5 .10
3
19.0.10
3
27.9,10
3
10.51
K
B
39.8 10' 32.3 .10
3
47.3.10
3
10.58
where subscript 'A' refers to K
A
, whereas, 'A" and '0' refer to K
A
' and K; respec-
tively. As already mentioned, there may be a significant level of uncertainty on
L 1 ~ . 0 Moreover, the adsorption of o-xylene and tolualdehyde on 1
0
sites may
involve hydrogen abstraction from the substituent group( s) of the aromatic com-
pounds. This type of chemisorption has a 'dissociative' character. Data on the bond
strengths of the groups formed after adsorption (presumably C-O-V and V-o-H)
are not available for the catalyst of this study. In conclusion, Boudart's criteria may
not be strictly applicable to K
A
Nevertheless, the temperature dependence of K;
and K
A
after parameter estimation based on all temperatures (Table 3) resulted in
positive entropy changes for the adsorption of oxygen, xylene and tolualdehyde.
(c) For the adsorptionconstant oftolualdehyde, phthalide and phthalic anhydride
(K
B
) , Boudart's criterion (Eq. l lb) is satisfied. Based on the results of Table 3,
the Van't Hoff equation for K
B
is:
InK = -0916+4787.10
B T(K)
so that the value for L 1 ~ B is - 7.61 kJ kmol " 1 K- 1. The standard entropies in the
gas phase for tolualdehyde, phthalide and phthalic anhydride were calculated at the
mean temperature (ca. 350C) according to Benson's method [17]. ~ for these
compounds amounts on the average to 502.1 kJ kmol " 1 K- 1. Criterion Eq. (Llb)
is satisfied. The narrower, but less stringent, constraints of Eq. (11c) are not
satisfied for K
B

J.N. Papageorgiou et al./Applied Catalysis A: Generall20 (1994) 17--43 41
100 I
1
0/
/0
I
I
,
80 t-
o
/
o
en
s 0
"iii
00
>
0
"tl
I
:s
~
u
~
0
i
I
c
0
.2
40
l!!
~
c
8
~
~
~
20 f-
0
f-

o "
0 20 40 60 80 100
o-xylen conversion (mol%), observed value
rig. 12. Parity plot for the conversion of o-xy lene.
Finally, it can be seen from Tables 2 and 3 that all the estimated enthalpies of
adsorption are negative, so that Boudart' s criterion (Eq. IIa) is satisfied. This
implies that the adsorption of oxygen and aromatic compounds on the catalyst
surface is exothermic.
s. Conclusions
The selective oxidation of o-xylene into phthalic anhydride over V-Ti-o cata-
lysts follows a so-called ' rake' mechanism. The major intermediates along this
route are o-tolualdehyde and phthalide. The catalyst sites involved in this process
consist of two types: reduced (1) and oxidized (1
0
) sites. The oxidized sites offer
catalyst oxygen for: (i) the abstraction of hydrogen atoms from the aromatic
molecule and subsequent water formation and (ii ) the insertion of an oxygen atom
42 J.N. Papageorgiou et al. / Applied Catalysis A: Generall20 (1994) 17-43
in the aromatic molecule. In parallel with its selective oxidation, o-xylene is totally
oxidized to carbon oxides and maleic anhydride. The selectivity of this process
remains nearly constant at the level of 23.5 mol-% at 34D-360C.
The proposed kinetic model (ORES-AOR model) is a further development of
Hougen-Watson type rate equations, accounting in detail for the interaction of the
reacting species with the active sites of the catalyst. The reduced catalyst sites are
readily reoxidized either by adsorption of gas-phase oxygen, or by catalyst oxygen
from the lattice. The rate determining steps in the reaction network of the oxidation
of 0-xylene (Scheme 2) are the aromatic-compound oxidation reactions.
According to the proposed reaction scheme and kinetic model, the ratio of
oxidized to reduced sites (1
0
/ 1) and the composition of the reaction mixture deter-
mine the product distribution. Evidently, while the reaction takes place, the ratio
1
0/1
is determined by the partial pressure of oxygen and the partial pressures of the
aromatic compounds that adsorb on the catalyst surface. When the ratio 1
0/1
increases, the oxidation of o-xylene to tolualdehyde is favored, whereas the oxi-
dation of tolualdehyde to phthalide and further to phthalic anhydride are not.
Although the overall conversion of o-xylene increases with the partial pressure of
oxygen, the selectivity for phthalic anhydride is lower at higher partial pressures
of oxygen.
Notation
A frequency factor
C products of the total oxidation of o-xylene (CO, CO
2
and maleic
anhydride)
E; activation energy (kJ/kmol)
~ molar feed rate of o-xylene (mol/h)
11lf1. standard enthalpy of adsorption (kJ/kmol)
k, K symbols used for rate constants and equilibrium coefficients (further
definitions in the text)
1 reduced catalyst sites
1
0
oxidized catalyst sites
0* catalyst lattice oxygen
P, partial pressure of component i (atm)
P phthalide
PA phthalic anhydride
r, rate of reaction i (mol g cat - 1 h- 1)
~ standard entropy of a gaseous molecule (kJ kmol " 1 K- 1)
S, selectivity of product i (mol i/mol o-xylene or mol-%)
~ standard entropy of adsorption (kJ kmol-
1
K-
1
)
T o-tolualdehyde or temperature (K)
w
X
J.N. Papageorgiou et al. / Applied Catalysis A: Generall20 (1994) 17-43
amount of catalyst (g)
o-xylene
43
Acknowledgements
The authors are grateful to the Belgian Ministerie voor de Programmatie van het
Wetenschapsbeleid for the Center of Excellence Award. G. Creten, M. Riebbels
and N. De Vleeschauwer are gratefully acknowledged for their contribution to the
experimental results.
References
[1] S.L.T. Andersson, J. Catal., 98 (1986) 138.
[2] G.C. Bond, J. Catal., 116 (1989) 531.
[3] M. Boudart, D. Mears and M.A. Vannice, Ind. Chim. Belg., 32 (1967) 281.
[4] G. Deo, A.M. Turek, I.E. Wachs, T. Machej, J. Haber, N. Das, H. Eckert and A.M. Hirt, Appl. Catal. A, 91
(1992) 27.
[5] G. Deo and I.E. Wachs, J. Phys. Chern., 95 (1991) 5889.
[6] H. Eckert and I.E. Wachs, 1. Phys. Chern., 93 (1989) 6796.
[7] H. Eckert and I.E. Wachs, Mater. Res. Soc. Symp. Proc., 111 (1988) 455.
[8J G.F. Froment, AIChE J., 21 (1975) 1041.
[9J G.F. Froment, 2nd Symposium on Hydrocarbon Oxidation, Manchester, England (1976).
[IOJ G.F. Froment and K.B. Bischoff, Chemical Reactors Analysis and Design, 2nd ed., Wiley, New York, 1990,
pp. 340--343.
[11] P.J. Gellings, in G.C. Bond and G. Webb (Editors), Specialist Periodical Reports - Catalysis, Vol. 7, Royal
Society of Chemistry, London, 1985, p. 105
[12] G. Busca, G. Genti, L. Marchetti and F. Trifiro, Langmuir, 2 (1986) 568.
[13 J F.D. Kopinke, G. Greten and G.F. Froment, in P. Ruiz and B. Delmon (Editors), Proceedings of the 3rd
European Workshop Meeting on New Developments in Selective Oxidation by Heterogeneous Catalysis,
Louvain-La-Neuve, Belgium, 8-10 April 1991 (Studies in Surface Science and Catalysis, Vol. 72), Elsevier,
Amsterdam, 1992, p. 317.
[14] R Kozlowski, RF. Pettifer and J.M. Thomas, J. Phys. Chern., 87 (1983) 5176.
[15] z.x. Liu, o.x, Bao and N.J. Wu, J. Catal., 113 (1988) 45.
[16] V. Nikolov, D. Klissurski and A. Anastasov, Catal. Rev. Sci. Eng., 33 (1991) 319.
[17] RC. Reid, J.M. Prausnitz and B.E. Poling, The properties of gases and liquids, 4th Ed., McGraw-Hill, 1987,
pp. 173-190.
[18] PJ. Van Damme, S. Narayanan and G.F. Froment, AlChE J., 21 (1975) 1065.
[19] C.M. Van Den Bleek, K. Van Der Wiele and P.J. Van Den Berg, Chern. Eng. Sci., 24 (1969) 681.
[20J A.J. Van Hengstum, J. Pranger, S.M. Van Hengstum-Nijhuis, J.G. Van Ommen and PJ. Gellings, J. Catal.,
101 (1986) 323.
[21] M.S. Wainwright and N.R. Foster, Catal. Rev. Sci. Eng., 19 (1979) 211.
[22] G.T. Went, L. Leu and A.T. Bell, J. Catal., 134 (1992) 479.
[23] M. Witko, J. Mol. Catal., 70 (1991) 277.

Вам также может понравиться