Download as pdf or txt
Download as pdf or txt
You are on page 1of 16

Applied Catalysis, 26 (1986) 211-226 211

Elsevier Science Publishers B.V., Amsterdam - Printed in The Netherlands

KINETICS AND REDOX PROPERTIES OF VANADIUM


PHOSPHATE CATALYSTS FOR BUTANE OXIDATION

J.S. BUCHANAN and S. SUNDARESAN


Department of Chemical Engineering, Princeton University, Princeton, NJ 08544, U.S.A.

(Received 17 December 1985, accepted 5 June 1986)

ABSTRACT
The kinetics of butane oxidation over two vanadium phosphate catalysts of
differing phosphorous to vanadium ratios (P/V = 1.0 and 1.1) were studied in order
to assess the effects of phosphorous content on butane oxidation kinetics. These
catalysts were prepared in an organic medium and demonstrated yields of maleic
anhydride (MA) comparable to those claimed in commercial catalysts. The primary
reaction products were MA and carbon oxides. The kinetic data were correlated by
an eight parameter model which accounts for reaction inhibition by butane and MA.
The PV1.0 catalyst was approximately twice as active as the PV1.1 catalyst, while
the selectivities to MA were similar. Transient reaction experiments and redox
titrations indicate that the redox processes taking place under reaction conditions
are restricted to a near-surface region.

INTRODUCTION
Aromatics or olefins comprise the feedstocks for most selective oxidation
processes. There has been interest in replacing these unsaturated hydrocarbons
with paraffins. The saturated hydrocarbons tend to be less reactive but are often
less expensive, and can sometimes pose less of an environmental risk.
In the past decade, n-butane has replaced benzene as the feedstock of choice
for the production of maleic anhydride (MA). This has been made possible by the
development of highly active and selective catalyts. These catalysts are un-
supported vanadium phosphates, often promoted with other substances. Hodnett [1]
has recently reviewed the use of vanadium phosphate catalysts for MA production.
Many modern patented catalyst preparations involve precipitation of the pre-
cursor from an alcoholic medium in order to obtain high intrinsic surface areas
[2-6]. Successful catalyst preparations using an aqueous medium seem to require
unusual processing such as precipitation at elevated temperatures or washing with
acid to remove undesired phases [7-9]. The catalyst precursor is usually transformed
into vanadyl pyrophosphate, (VO)2P207' upon heating to 623-673 K. Further acti-
vation is required to bring the catalyst to a state of high activity and selectivity
[10] •
Although fluidized catalyst beds offer some advantages [11], most commercial
butane oxidation reactors consist of many packed tubes immersed in a molten salt
bath. A survey of the patent literature [2-9] and other sources [12,13] indicates

0166-9834/86/$03.50 © 1986 Elsevier Science Publishers B.V.


212

that the salt bath temperature is in the range 633-693 K. while the reactor hot
spot may be 20-40 K nigher. The feed concentration of butane in air is typically
kept slightly below the flammability limit of about 1.8%. Molar selectivities
to maleic anhydride (MA) are claimed to be 60-70%. and butane conversions (80-90%)
are chosen to maximize the yield of MA.
The mechanism of selective butane oxidation has not been thoroughly elucidated.
It has been shown in transient experiments that vanadium phosphate catalysts can
supply oxygen for the oxidation of butane even in the absence of oxygen in the feed
mixture [14-16]. Thus, the redox model proposed by Mars and van Krevelen [17]
for oxidation over metal oxide catalysts seems to apply to butane oxidation cata-
lysts. In this model the hydrocarbon reacts with oxygen supplied by the catalyst
lattice, leaving the catalyst site in a reduced state. The catalyst is then
reoxidized in a separate step by gaseous oxygen. Studies using isotopically labelled
oxygen lend further support to the redox model [15,18].
The literature does not reveal any detailed study of the kinetics of butane
oxidation over a well-characterized modern vanadium phosphate catalyst in the
temperature range of commercial operation, although the results of Sharma and
Cresswell [13] might be used in this region. In the present study, the rates of
steady-state butane oxidation over a range of gas-phase compositions and tempera-
tures encountered in an industrial reactor have been measured and correlated for
two catalysts with differing PIV ratios, in order to assess the effects of phos-
phorous content on butane oxidation kinetics. Transient reaction rate measurements
and redox titrations were performed to determine the extent to which oxygen in
the catalyst bulk participates in the redox processes taking place at the surface.

KINETICS OF BUTANE OXIDATION - LITERATURE SURVEY


Pepera et al. [15] have demonstrated a kinetic isotope effect for butane oxida-
tion over vanadium phosphate, using deuterated butane. Their results indicate
that the rate-determining step in butane oxidation is the initial activation of
the hydrocarbon. This irreversible adsorption step involves scission of a methylene
C-H bond.
The final products of the reaction are MA and carbon oxides. with trace amounts
of ethylene, furan and C2-C3 aldehydes and acids. Possible intermediates such as
butenes or butadiene are not generally observed, perhaps because they react much
faster than does butane. Olefins can appear, however, under fuel-rich or oxygen-
poor conditions [19].
Escardino et al. [20] studied the kinetics of butane oxidation over a vanadium
phosphate catalyst (P/V = 0.8) at 673-753 K. A triangular reaction network was
proposed:
213

r
1
butane ) MA
r~/r3

MA and carbon oxides are formed directly from butane (at rates r and r 2 respecti-
1
vely), and MA is also oxidized to carbon oxides (at rate r 3 ) . At butane concent-
rations typical of industrial reactors, the rate of butane oxidation was controlled
by the reaction between butane gas and surface oxygen.
Wohlfahrt and Hofmann [21J investigated butane oxidation kinetics over a wide
range of butane and oxygen concentrations at 719-777 K. The vanadium phosphate
catalyst was prepared using an aqueous medium. The reaction scheme was similar to
that described above, except that the production of CO and CO 2 were considered
separately.
Centi et al. [22J used a very active catalyst prepared in an organic medium
which permitted low reaction temperatures (573-613 K) to be employed. Under these
conditions it was found that the rate of reaction of butane to carbon dioxide
did not depend on the hydrocarbon concentration, but only on the concentration
of oxygen.
Sharma and Cresswell [12J studied the kinetics of butane oxidation between
573-653 K over a commercial catalyst, described as a promoted vanadium phosphate
catalyst. The triangular reaction model described earlier was employed to correlate
the res ults.

EXPERIMENTAL
Two catalysts designated PV1.0 and PV1.1 were used in this study. The method
of catalyst preparation generally followed·the procedure disclosed in reference
[4J, and has already been described [10J. Briefly, a slurry of 91 g V205(1 mol V)
in 750 ml methanol was reduced by HCl gas. A mixture of 67.5 9 85% H3P04 and 29.3 g
P205 (for P/V = 1.0) or 74.4 g 85% H3P04 and 32.4 g P205 (for P/V = 1.1) was
added along with 250 ml benzene. After the mixture had refluxed overnight, solvent
was removed using a Dean-Stark trap. The resulting syrup was dried to a porous
cake, which was ground, pressed, broken and sieved to 25-35 mesh. Calculations
indicated that particles of this size should be free of pore diffusion limitation.
The catalyst granules were calcined in air at 663 K for 2 hours, washed in
methanol for an hour and then activated for at least four days at 713 K under a
1.55% butane in air feed. A prior study using this type of catalyst showed that
this pretreatment improved MA yields [10J. Catalysts were stored under nitrogen
in a desiccator to minimize damage to the catalyst by atmospheric water vapor.
The reactor used in this study was a 7 mm 10 glass U-tube in an aluminium
split block. The axial variation of the temperature of the block in the reaction
214

zone was found to be less than 1 K under nonreactive conditions. The catalyst was
diluted with glass granules and a 4 mm diameter axial glass rod as necessary to
keep the axial temperature at the end of the bed within 1-2 K of the block tempera-
ture over the whole range of conversions. Experiments revealed significant blank
reaction of MA in the preheat section of the reactor when an external recycle pump
was used. Therefore, the kinetic experiments were carried out in a once-through,
integra 1 mode.
The feed to the reactor was a mixture of CP grade n-butane in dry air. The feed
and effluent streams were analyzed by gas chromatography, as described earlier
[1oJ. MA was collected in a water bubbler and titrated with a 0.1 N NaOH solution.
Carbon balances were performed for all runs.
Preliminary experiments showed that prolonged runs at 723 K and high conversion
could cause an irreversible loss in selectivity for the catalyst with PjV = 1.0
(henceforth referred to as PV1.o), so all further work was done at lower tempera-
tures. For each catalyst, experiments were performed at 663 K, 688 K and 713 K.
At each temperature, feeds of 0.62,0.94 and 1.55% butane in air were employed.
Residence times were varied to obtain butane conversions from about 18% to about
94%. A total of 68 runs were used in determining the reaction kinetics for the
PV1.0 catalyst, and 59 runs for a catalyst with PjV = 1.1 (henceforth referred
to as PV1.1).
Catalysts were characterized by powder X-ray diffraction and BET surface area
measurements after reaction. X-ray diffractograms were recorded by a Rigaku 5036V2
diffractometer using CUK a radiation with a graphite monochromator. A Quantochrome
Quantasorb system was used for the surface area measurements with nitrogen as the
adsorbate.
Catalyst samples for redox titration were prepared by carrying out oxidation
of butane in air over small quantities (0.1-0.2 g) of catalyst under differential
reactor conditions. Butane concentrations covered the range 0-1.6%. The PV1.0
catalyst samples were run at 688 K, and PV1.1 at 713 K. After 20-24 hours at the
specified temperature, the reactor was cooled to room temperature at a rate of
approximately 20 K min- 1.
The redox titration involves dissolving the catalyst in phosphoric acid,
titrating first with 0.05 N KMn0 4 and then with 0.05 N Fe(NH 4)2(S04)2 to determine
an average oxidation number for vanadium, n [10].
V
Transient reaction experiments were performed using the same reactor as for
the steady-state experiments. A valve was installed immediately upstream of the
reactor to allow switching between different feeds. Experiments were performed
under non-reactive conditions to determine the transport lag in the effluent line.
As the time required for the analysis of a gas sample was typically greater than
the desired time between sample injections, it was necessary to repeat each experi-
ment several times in order to obtain all of the desired data.
Two types of transient experiments were performed using 1 9 of the PV1.0 catalyst.
215

In one experiment the reactor feed was switched from 1.55% butane/air to 1.55%
butane/nitrogen. The reactor temperature was 688 K, and the flow rate at 1 atm
and 298 K was 83 cm 3 min-1• The amount of oxygen supplied for reaction by the
catalyst lattice could be calculated from the resulting data. In the other type
of experiment, the reactor feed was switched to 1.55% butane/air from feeds con-
taining 0,0.09 and 0.64% butane in air. The reactor temperature was 663 K with
flowrates of about 73 cm 3 min-1.

RESULTS AND DISCUSSION


After running .under reaction conditions, the surface area of the PV1. 0 catalyst
was 26 m2 g-1 and that of the PV1.1 catalyst was 14 mg.
2 -1
Both catalysts displayed
only the X-ray diffraction pattern of (VO)2P207 [23].

3.0 PV 1.0 catalyst


O. a, 6 • experimental
_ • calculated
)(

82.0 MA
~
E

.2
~ 1.0 COx
c:
sc:
o butane
o
1.5 2.0

3.0 PV 1.1 catalyst

....
r-.
o
)(

82.0
::0
MA
~
~
~ 1.0
c:

§ butane

0.0 0.5 1.0 1.5 2.0 2.5


'to s-g catlcc

FIGURE Effluent concentrations versus time factor for 1.55% butane feed and
688 K.
216

selectivity

oL-__ .J-.._ _......._ _...L-_ _....... _ - ~

o 20 80 100

FIGURE 2 Selectivity and yield of MA versus butane conversion for PV1.0. The
reactor temperature was 688 K and feed contained 1.55% butane in air.

Reaction rate data


Some representative reaction data appear in Figure 1, where measured effluent
concentrations (along with concentrations calculated from the kinetic model,
discussed below) are plotted against time factor. The time factor, T, is defined
as grams of catalyst divided by the volumetric flowrate in cm 3 s-1 under reaction
conditions.
The activity toward butane oxidation per gram of catalyst was about 1.5-2 times
higher for the PV1.0 catalyst than for the PV1.1 catalyst; per unit surface area,
the activities were roughly equivalent. The selectivity of the PV1.0 catalyst
was slightly higher (2-4%) than that for the PV1.1 catalyst at the lower tempera-
tures, while at 713 K they are equivalent. The selectivities of these catalysts
were comparable to that of a similar catalyst studied earlier [10J which had P/V =
1. 2.
For all temperatures and feeds, the selectivity declined with increasing butane
conversion. This trend is illustrated in Figure 2. A maximum in MA yield was
observed in the neighborhood of 90% conversion. Subsequent decline in MA yield
(with increasing butane conversion) is indicative of combustion of the MA. The
mean carbon balance for all the runs was 98.6% with a standard deviation of 1.8%.
The major products besides MA were CO, CO 2 and H20. Trace amounts of acrylic
acid, acetic acid and ethylene were also observed. The CO/C0 2 ratio was 1.5-2.0
for the PV1.0 catalyst and 1.7-2.3 for PV1.1. This ratio tended to decrease with
increasing conversion and increase with increasing temperature. A CO/C0 ratio
2
greater than 1 was also observed by Wohlfahrt and Hofmann [21J and by Sharma and
Cresswell [13J. Moser and Schrader [24J, however, saw no CO when oxidizing butane
over vanadium phosphate catalysts prepared by solid state reaction, and Centi et
al. [22J observed CO only at high conversions. There does not seem to be a strong
correlation between CO/C0 2 ratio and selectivity to MA in comparing these various
studies.
217

3.0 . . . - - - - - - - - - - - - - - . . . ,

2.0

0.0 L-_--JL....-_----I._ _~_ _........._ _.......

o 20 40 60 80 100
Conversion, %

FIGURE 3 Pseudo-first order rate constant for PV1.n at 688 K. The feed contained
1.55% butane in air.

Reaction rate expressions


Graphical manipulation of the data indicated that the rate of butane oxidation
was less than first order in butane concentration. For instance, the values of
a pseudo-first order rate constant for butane conversion were larger at high con-
versions (Figure 3). The psueudo-first order rate constant, kps' was calculated
as k = -£n(1 - X)/T, where x is the fractional butane conversion.
ps
Differentiation of the reaction data showed that the rate of butane oxidation
cannot be correlated satisfactorily in terms of butane concentration alone. The
oxygen concentration in a reactor with 1.5% butane feed and 90% conversion ranges
from about 21 near the entrance to about 15% near the exit of the reactor.
Separate experiments in which air in the feed was diluted with nitrogen indicated
that a change in oxygen concentration of this magnitude resulted in a decrease
in the reaction rate of approximately 8%. Explicit dependence of reaction rate
on oxygen concentration was therefore included in the kinetic model.
Substantial inhibition of the oxidation reactions by MA was found by Sharma
and Cresswell [13J. Preliminary experiments with MA and water in the feed showed
product inhibition effects with our catalyst. The mechanism of such inhibition
has not been thoroughly investigated. Since the water/MA ratio does not vary
greatly in the course of reaction, it would be difficult to estimate separate
inhibition parameters for the two product species. Product inhibition was there-
fore represented by a single lumped term involving only the MA concentration.
This approach led to a satisfactory fit of the experimental data while economizing
on adjustable parameters.
A simple two-step redox reaction model leads to a rate expression for butane
oxidation of the form:
218

TABLE
Estimated kinetic parameters and standard deviations of errors

PV1 .0 PV1.1

Pre-exponential factors
3 -1 -1
/cm s g cat
1.96 x 10
10 1.16 x 10
9
k
10
k20 3.40 x 10
11
7.50 x 109
k 1.70 x 10
13
4.80 x 109
30
Activation energies
/kJ mol- 1
E 125 116
1
E
2
145 130
E 180 138
3

Inhibition factors
K 59 20
1
K 26 12
2
Standard deviation of
errors for species
divided by range of
species x 100/1;
Butane 2.0 1.4
MA 2.9 1.9
Carbon oxides 3.1 1.6

where CB and Co denote the concentrations of butane and oxygen respectively. The
factor 1/[1 + KCs/ConJ is representative of the oxidation state of the catalyst
surface. The redox model may yield values of 1/2 or 1 for the exponent n of the
oxygen concentration, depending on the details of model formulation. Which of these
values were chosen made little difference in fitting our rate data. The above
rate expression can be modified to include the effect of product inhibition. We
found that our reaction rate data were satisfactorily correlated by the following
empirical rate expressions for the triangular network described in section 2:

where CM denotes the concentration of MA. For both catalysts, the selectivity at
219

low conversions (where reaction r 3 is relatively unimportant) decreased with


increasing temperature. This implies that the activation energy for r is greater
2
than that for r Therefore, the values of E1 and E2 were allowed to be different,
1.
even though the rate-controlling step appears to be similar in these two reactions
USJ.
Estimates of kinetic parameters were first obtained from plots of the reaction
rate data. Further refinement was obtained by performing nonlinear least-squares
parameter estimation, first at each temperature and finally for all three tempera-
tures. A modified Hooke-Jeeves search routine [25J was coupled with a set of pro-
grams for integrating initial-value ordinary differential equations [26J in order
to minimize the sum of the squares of the differences between calculated and
experimental effluent concentrations. Estimated values of the kinetic parameters
are listed in Table 1. Graphical comparison of some calculated and experimental
effluent concentrations are shown in Figure 1. The sets of estimated parameters
for the two catalysts had largely similar features. The estimated rate of reaction
r 1 is roughly twice that of r 2, for both catalysts. The rate of r is much smaller
3
than that of r 2, except at very high butane conversions. Thus, carbon oxides are
formed primarily from the unselective oxidation of butane, rather than by further
reaction of MA. The large estimated values of E3 indicate that oxidation of MA
becomes more significant at higher temperatures. The activation energies for
PV1;O are somewhat larger than those for PV1.1. The reaction inhibition factors
K1 and K2 are 2-3 times larger for PV1.0 than for PV1.1.
The activation energies obtained in the present study are somewhat higher than
those reported by Sharma and Cresswell [13J. The activation energies estimated
by Escardino et al. [20J, Wohlfahrt and Hofmann [21J and Centi et al. [22J were
generally lower than those reported by Sharma and Cresswell [13J and obtained
in our study.


4.1 PV 1.1
•• • • . .• •

FIGURE 4 Average oxidation number of vanadium versus butane concentration over


the catalyst.
220

Catalyst oxidation state


The magnitudes of the estimated K values indicate significant reaction inhibition
1
by butane, especially for PV1.o. In the redox model this is attributed to reduction
of the catalyst. Redox titration experiments were performed to determine how changes
in butane concentration affect the oxidation state of the catalyst. The results
are presented in Figure 4.
The vanadium of the PV1.o catalyst is generally in a higher oxidation state
than in the PV1.1 catalyst. For both catalysts, there is a large jump in nV values
+4 +5
between 0% butane and 0.1 butane. The interchange between V and V appears
to be an important part of the redox process. When little butane is available to
sustain the redox cycle, the catalysts become heavily oxidized. The pentavalent
a-VOP0 phase [23] was detected in the catalysts treated in just air and their
4
(VO)2P207 content decreased.
The PV1.1 catalyst shows no clear change in nV throughout the range of butane
concentrations likely to be encountered in a reactor (0.15-1.5%). While the change
in nV of the PV1.0 catalyst for the same range of butane concentrations is small,
it appears that this catalyst is somewhat more subject to oxidation or reduction
under these conditions than PV1.1. It is therefore not surprising that the value
of K for the PV1.0 catalys is larger than that for PV1.1. This is also in accord
1
with other studies where excess phosphorous has been shown to stabilize the +4
state of vanadium [27,28]. It must also be noted that the variations in nV with
butane concentration for the two catalysts are very small and this appears at
first to be inconsistent with large K1 values. A plausible explanation is that
the changes in the oxidation states are confined to a thin region near the surface
and that substantial changes can occur in the extent to which this surface layer
can be oxidized or reduced. This would be consistent with large K1 values and also
only small changes in nv with butane concentration. This will be discussed in
greater detail in the following section.

Transient kinetic results


From the significant reaction inhibition by butane it might be expected that
the catalyst oxidation state would change substantially as the butane concentration
was varied from 0.15% to 1.5%. The extent of bulk oxidation/reduction measured
by redox titration as the butane concentration was varied in this range was,
however, quite small for PV1.0 and not detectable for PV1.1. A possible explanation
of these observations is that much of the catalyst oxidation or reduction is con-
fined to the near-surface region, such that the bulk of the catalyst is largely
unaffected unless the reaction conditions become severe.
Transient reaction experiments can shed some light on the extent of bulk cata-
lyst participation in the redox process. The results of switching from a butane/air
to a butane/nitrogen feed appear in Figure 5. The butane concentration was 1.55%
in both feeds. The conversion dropped very rapidly in the first two minutes, then
221

50

5 10 15 20
Time, minutes

FIGURE 5 Conversion after switching from 1.55% butane/air to 1.55% butane/nitrogen.

decreased slowly until becoming undetectable after about 40 minutes. When pure
butane is used for catalyst reduction, substantial carbon deposition can result
[10,15]. Under the conditions used in this experiment, however, a carbon balance
indicated that virtually all the butane consumed during the period of butane/
nitrogen feed reached to MA, CO and COr The selectivity to MA dec1i ned from about
50% initially to about 40% after 15 minutes, then declined towards zero.
The amount of oxygen consumed by reaction during this experiment was calculated
to be 6.5 x 10- 4 moles, of which about one-third was consumed in the first two
minutes. This catalyst sample had a surface area of about 17 m2 g-1. Using the
approach of Pepera et al. [15], it was estimated that about 2.9% or 9.3 x 10-5
moles of the (VD)2P207 units in the 1 gram of catalyst were at the surface. If
only the vanadyl oxygens are considered to participate in the redox process, the
surface layer contains approximately 1.9 x 10-4 moles of available oxygen. Thus
the total amount of oxygen consumed during this experiment is equivalent to the
oxygen available in the first 3-4 layers of the catalyst. The initial period of
very steep decline in reaction rate probably corresponds to the rapid depletion
of oxygen in an active surface layer, while the subsequent period of gradual
decline corresponds to the much slower diffusion of oxygen from the subsurface
layers. These results support the idea that the near-surface layer of the catalyst
may undergo substantial oxidation or reduction with relatively little effect on
the bulk.
Studies using different conditions and other catalysts have likewise indicated
limited subsurface involvement of butane oxidation catalysts [15,18], Using a
'pulse reactor, Pepera et al , [15J determined that the reversible redox capacity
of a vanadium phosphate catalyst prepared by precipitation from isobutanol is about
one oxygen per exposed vanadium atom and that the exchange of oxygen between the
222
60 r----------------,
butane concentration
in initial feed : c 0.00 %
• 0.09%
o 0.64%

60 120 180
Time. seconds

FIGURE 6 Conversion of 1.55% butane feed after switching from lower feed concent-
rations .

......
'0

x
3.0 -
1;;
o
9'2.0
!!?
"0
E 1.0
~
a: 0.0 L- . . - _......_ _....... ......_ _....... ...................

o 30 60 90 120
Time.seconds

FIGURE 7 Average rate of butane oxidation. Feed switched from 0.09% to 1.55%
butane at time = O.

catalyst bulk and the surface layer is very slow. Kruchinin et al. [18] found
the pool of exchangeable oxygen in a cobalt-promoted vanadium phosphate catalyst
to be restricted to the oxygen in a near-surface layer.
Many catalyst characterization techniques mainly furnish information about
the catalyst bulk rather than the surface layer. The surface properties, which
are crucial in determining the reactivity of the catalyst, may not be readily
correlated with the bulk properties. A number of examples can be cited which
substantiate this lack of correlation for the vanadium phosphate catalyst. A study
on the effects of pretreatments on a vanadium phosphate catalyst found no con-
sistent relationship between reactivity in butane oxidation and the bulk catalyst
properties [10]. The catalysts examined in the present study seem to undergo
substantial oxidation or reduction of the near-surface layers, with relatively
small changes appearing in the bulk oxidation state. Similar results were found
by Pepera et al. [15]: there was essentially no difference in the redox state of
223

vanadium in the bulk, as monitored by ESR, between a catalyst which had been reduced
under pulses of butane and one which had been oxidized under pulses of oxygen.
Moser and Schrader [24J found that the Raman spectrum of a NO)2P207 catalyst
hardly changed over a 5-8 hour activation period, even though the selectivity to
MA more than doubled during this period, showing that significant changes had
occurred in the near-surface layer. Clearly the need exists for greater application
to butane oxidation catalysts of characterization techniques that are specifically
surface-sensitive, in order to gain a better understanding of the relationships
between catalyst structure and reactivity.
Transient experiments were done where the feed mixture was switched to 1.55
butane/air from mixtures containing less butane. The butane conversions were high
immediately after the feed switch, and declined to the steady value of about 37
after about 5 minutes (Figure 6). Selectivity to MA did not vary appreciably during
this time. The more oxidizing prior feed mixtures (i.e., those with less butane)
led to higher reaction rates after switching. Apparently the catalyst (or at least
the surface layer) became oxidized under the prior feeds, which led to an initially
higher reaction rate after switching feeds. The changes in the catalyst appear
to be reversible, as evidenced by the fact that the catalyst activity returned
to the same steady level after being oxidized several times.
The results for the 0.09" butane feed are shown in more detail in Figure 7,
where the average reaction rate in the reactor is plotted against time. The average
rate was calculated as (butane concentration in the feed - butane concentration
in the effluent) x (flowrate)/(grams of catalyst). The reaction rate naturally
rises when the feed concentration of butane is increased from 0.09~ to 1.55", but
the rate overshoots the eventual level by about 18:. To our knowledge, the con-
sequences of this type of rate-overshoot phenomenon for the behavior of large-
scale reactors has not been assessed. It seems possible that a reactor which is
stable at steady state under a variety of feed conditions could be vulnerable to
overheating if the feed concentration of the hydrocarbon or feed flowrate were
increased too quickly or if the feed temperature were decreased rapidly, such that
a region of oxidized catalyst were suddenly contacted by a hydrocarbon-rich gas
phase.

Method of catalyst preparation


It is quite conceivable that the properties of a complex catalyst such as
vanadium phosphate are dependent on the method of preparation. The experimental
procedure used in the various studies to prepare vanadium phosphate catalysts
can be divided into two broad categories: organic and aqueous. At present the
effects of preparation procedure on the redox properties of the vanadium phosphate
catalyst cannot be resolved with certainty, yet certain observations may be helpful.
Cavani et al. [29J observed that vanadium phsophate catalysts prepared in an
organic medium are more stable towards oxidation than catalysts prepared by eva-
224

poration of an aqueous solution. A variety of crystalline phases have been detected


in catalysts prepared from aqueous solutions, and the crystalline phase compositions
of these catalysts can be modified by changing the gas phase composition (at
reaction temperatures) to which they are exposed [23,30,31]. Relatively few studies
have used catalysts prepared in an organic medium, but they indicate that this
type of catalyst is less liable to form phases other than (VO)2P207 [28,10]. The
catalysts used in the present study and by Pepera et al. [15] were prepared in
organic media and showed little bulk oxidation/reduction. In contrast, the catalysts
prepared from an aqueous medium were found by Hodnett and Delmon [14] to be quite
prone to bulk reduction when exposed to a butane/nitrogen mixture at 698 K. Thus,
although the evidence is scanty, there are grounds for suggesting that catalysts
prepared by evaporation of an aqueous solution may in general be more susceptible
to bulk oxidation/reduction than those prepared from organic medium.
This may also help explain certain trends in vanadium phosphate catalysts
regarding the relationship between selectivity and phosphorous content. Studies
involving catalysts prepared in an aqueous medium typically find that selectivity
to MA increases significantly as the P/V ratio increased above 1.0 for both butene
oxidation [23,32] and butane oxidation [33]. In contrast no increase in selectivity
with catalysts prepared in an organic medium has been found upon increasing the
P/V ratio above 1.0 for either butene oxidation [28] or (in accord with our results)
for butane oxidation [29]. Hodnett and Delmon [27] observed that increasing the
phosphorous content in their catalysts (prepared from aqueous media) restricted
bulk oxidation and reduction, and suggested that this restriction of oxygen
mobility in the bulk catalyst was a means by which excess phosphorous increased
selectivity. It is suggested here that a similar restriction in bulk oxygen
mobility may come about in catalysts prepared in organic media (for reasons that
are not yet clear) by virtue of the method of preparation, and therefore high
selectivity may be obtained in these catalysts without the addition of excess
phosphorous.
Although it may not be required for high selectivity, some excess phosphorous
is probably desirable in practice (even for catalysts prepared in an organic medium)
in order to stabilize the catalyst against long-term, irreversible deactivation
[1]. In a packed bed reactor for butane oxidation, the catalyst is usually diluted
with inert pellets in the vicinity of the hot spot that develops near the entrance
of the reactor, in order to moderate the temperature. In this region, it might
be better to load catalyst with a higher phosphorous content rather than simply
diluting the catalyst. Excess phosphorous would give the lower catalytic activity
desired near the hot spot, and it might also serve to stabilize the catalyst under
the severe conditions there.
225

SUMMARY AND CONCLUSIONS


The kinetics of butane oxidation over two catalysts of differing P/V ratios
were studied. These catalysts were prepared in an organic medium, and demonstrated
yields of MA comparable to those claimed for commercial catalysts. The kinetic
data obtained using these catalysts should be suitable for use in simulating the
performance of a typical industrial butane oxidation reactor. The kinetic data
could be satisfactorily correlated by an eight parameter model, which accounts
for reaction inhibition by butane and maleic anhydride.
The PV1.0 catalyst was approximately twice as active as the PV1.1 catalyst,
while the selectivities to maleic anhydride were similar. For both catalysts,
the activation energies for the non-selective reactions were greater than that for
the formation of maleic anhydride. Therefore, the higher the hot-spot temperature
in a packed bed reactor for butane oxidation, the more adversely will the selecti-
vity towards maleic anhydride be affected.
The oxidation reactions were found to be appreciably inhibited by butane,
indicating that the surface layer could undergo significant reduction by butane.
Yet bulk oxidation number as determined by redox titration changed very little
as butane concentration was varied within the limits found in the reactor. Thus
the redox processes taking place under reaction conditions appeared to be limited
to a near-surface region. Limited subsurface participation was confirmed by transient
reaction experiments. Therefore, catalyst characterization techniques such as
X-ray diffraction, infrared and Raman spectroscopies, ESR and redox titration,
which mainly furnish information about the catalyst bulk rather ,than the surface
layer, are unlikely to lead to a better understanding of the relationship between
reactivity and structure of vanadium phosphate catalysts (particularly those
prepared in organic media) unless they are supplemented by surface-sensitive
characterization techniques.

ACKNOWLEDGEMENTS
We thank E. Arnold for assistance with the transient kinetic experiments, and
D. Olsen and N. Goecke (Mobil Research and Development Corporation) for obtaining
the X-ray diffractograms. One of us (J.S. Buchanan) received support through a
National Science Foundation graduate fellowship. Financial support for this work
by National Science Foundation (CPE-8405132) is gratefully acknowledged.

REFERENCES
1 B.K. Hodnett, Catal. Rev.-Sci. Eng., 27 (1985) 373.
2 R.A. Schneider, U.S. Patent 3,864,280 (1975).
3 K. Katsumoto and D.M. Marquis, U.S. Patent 4,132,670 (1979).
4 C.A. Udovich and R.J. Bertolacini, U.S. Patent 4,328,126 (1982).
5 E.C.Milberger, N.J. Bremer and D.E. Dria, U.S. Patent 4,333,853 (1982).
6 T.C. Yang, K.K. Rao and 1.0. Huang, U.S.Patent 4,116,868 (1978).
7 R.A. Mount, H. Raffelson and W.O. Robinson, U.S. Patent 4,116,868 (1978).
8 G. Stefani and P. Fontana, U.S. Patent 4,100,106 (1978).
9 R.J. Hutchings and R. Higgins, U.S. Patent 4,288,372 (1981).
226

10 J.S. Buchanan, J. Apostolakis and S. Sundaresan, Appl. Catal., 19 (1985) 65.


11 S.C. Arnold, G.D. Sucin, L. Verde and A. Neri, Hydrocarbon Processing, 64(9)
(1985) 123.
12 F. Budi, A. Neri and G. Stefani, Hydrocarbon Processing, 61(1) (1982) 159.
13 R.K. Sharma and D.L. Cresswell, paper presented at AIChE Annual Meeting, San
Francisco, November 1984.
14 B.K. Hodnett and B. Delmon, Ind. Eng. Chem. Fundam., 23 (1984) 465.
15 M.A. Pepera, J.L. Callahan, M.J. Desmond, E.C.Milberger, P.R. Blum and N.J.
Bremer, J. Amer. Chem. Soc., 107 (1985) 4883.
16 V.A. Zaxhigalov, Yu. P. Zaitsev, V.M. Belousov, N. Wyustnek and H. Wolf,
React. Kinet. Catal. Lett., 24 (1984) 375.
17 P. Mars and D.W. van Krevelen, Chem. Eng. Sci. (Special Suppl.), 3 (1954) 41,
18 Yu. A. Kruchinin, Yu. A. Mischenko, P.P. Nechiporuk and A.I. Gel 'bstein,
Kinet. Katal., 25 (1984) 392.
19 G. Centi, G. Fornasari and F. Trifiro, J. Catal., in press (1985).
20 A. Escardino, C. Sola and F. Ruiz, An. Quim., 69 (1973) 385. Summarized in
Chem. Abstr., 79:35474 and in Ind. Eng. Chem. Prod. Res. Dev., 18 (1979) 7.
21 K. Wohlfahrt and H. Hofmann, Chem. Ing. Tech., 52 (1980) 811,
22 G. Centi, G. Fornasari and F. Trifiro, Ind. Eng. Chem. Prod. Res. Dev., 24
(1985) 32.
23 E. Bordes and P. Courtine, J. Catal., 57 (1979) 236.
24 T.P. Moser and G.L. Schrader, J. Catal., 92 (1985) 216.
25 R. Hooke and T.A. Jeeves, J. Assoc. Camp. Mach., 8 (1961) 212.
26 J. Villadsen and M.L. Michelsen, Solution of Differential Equation Models by
Polynomial Approximation, Prentice-Hall (1978).
27 B.K. Hodnett and B. Delmon, J. Catal., 88 (1984) 43.
28 F. Cavani, G. Centi, I. Manenti and F. Trifiro, Ind. Eng. Chem. Prod. Res.
Dev., 24 (1985) 221,
29 F. Cavani, G. Centi and F. Trifiro, Appl. Catal., 9 (1984) 191.
30 B.K. Hodnett and B. Delmon, Appl. Catal., 9 (1984) 203.
31 G. Centi, C. Galassi, I. Manenti, A. Riva and F. Trifiro, in Preparation of
Catalysts III, G. Poncelet, P. Grange and P.A. Jacobs (Eds.), Elsevier,
Amsterdam (1983) 543.
32 M. Nakamura, K. Kawai and Y. Fujiwara, J. Catal., 34 (1974) 345.
33 B.K. Hodnett, P. Permanne and B. Delmon, Appl. Catal., 6 (1983) 231.

You might also like