Applied Catalysis, 30 (1987) 277-301 Elsevier Science Publishers B.V., Amsterdam - Printed in The Netherlands 277 SOLID ELECTROLYTE POTENTIOMETRY STUDY OF BUTENE OXIDATION OVER VANADIUM PHOSPHATE CATALYSTS E.M. BRECKNER, S. SUNDARESAN and J.B. BENZIGER Department of Chemical Engineering, Princeton University, Princeton, NJ 08544, U.S.A. (Received 29 July 1986, accepted 22 December 1986) ABSTRACT The technique of solid electrolyte potentiometry (SEP) has been used to follow the oxygen activity in vanadium phosphate (VPO) catalysts during the catalytic oxidation of butene. Reaction products and oxygen activity were monitored con- currentlv in an effort to correlate selectivity for partial oxidation of butene to malei: anhydride, furan and crotonaldehyde with oxygen activity in the catalyst. It was found that partial oxidation selectivity increased with decreased oxygen activity in the catalyst. Additionally, the activity for total oxidation and oxygen activity were found to be dependent on the past history of the catalyst for a catalyst with a P/V ratio of 1.07. A high temperature reduction of the catalyst was able to freeze the catalyst into a form that was selective for partial oxidation. Models to account for the rate and oxygen activity data from this study suggest that subsurface oxygen is involved in the oxidation of butene over VP0 catalysts and causes the non-selective total oxidation reaction. INTRODUCTION Vanadium oxide catalysts are widely used to carry out partial oxidation reactions. It is commonly accepted that oxygen incorporated into the metal oxide lattice participates in the reaction [l-3]. Oxygen is exchanged between the surface and the bulk by lattice diffusion. Changes in the surface layer with gas phase com- position and temperature are then mirrored in corresponding changes in the bulk of the vanadium oxide. Variations in catalyst activity and selectivity may then be expected to be correlated with the structure of the metal oxide. Several studies of vanadium phosphates (VPO) have attempted to establish correlations between activity and selectivity characteristics and various solid state properties of the catalyst, such as the crystalline phases present, the average oxidation number of the vanadium cation, and the degree of aggregation of the vanadium cations. Bordes and Courtine [4] employed a variety of techniques including X-ray and electron diffraction, infrared and uv-vis spectroscopies, and thermal gravimetric analysis to investigate a series of VP0 catalysts used for butene oxidation to maleic anhydride. They found that significant selectivity for partial oxidation was observed only when VOP04 and (VO),P,O7 were present. Nakamura and co-workers [5] suggested that a highsdegree of aggregation of vanadium ions with an average oxidation number of +4 gave a high selectivity to maleic anhydride. Ai and others 0166-9834/87/$03.50 0 1987 Elsevier Science Publishers B.V.
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Applied Catalysis, 30 (1987) 277-301 Elsevier Science Publishers B.V., Amsterdam - Printed in The Netherlands
277
SOLID ELECTROLYTE POTENTIOMETRY STUDY OF BUTENE OXIDATION OVER VANADIUM PHOSPHATE
CATALYSTS
E.M. BRECKNER, S. SUNDARESAN and J.B. BENZIGER
Department of Chemical Engineering, Princeton University, Princeton, NJ 08544, U.S.A.
(Received 29 July 1986, accepted 22 December 1986)
ABSTRACT
The technique of solid electrolyte potentiometry (SEP) has been used to follow the oxygen activity in vanadium phosphate (VPO) catalysts during the catalytic oxidation of butene. Reaction products and oxygen activity were monitored con- currentlv in an effort to correlate selectivity for partial oxidation of butene to malei: anhydride, furan and crotonaldehyde with oxygen activity in the catalyst. It was found that partial oxidation selectivity increased with decreased oxygen activity in the catalyst. Additionally, the activity for total oxidation and oxygen activity were found to be dependent on the past history of the catalyst for a catalyst with a P/V ratio of 1.07. A high temperature reduction of the catalyst was able to freeze the catalyst into a form that was selective for partial oxidation. Models to account for the rate and oxygen activity data from this study suggest that subsurface oxygen is involved in the oxidation of butene over VP0 catalysts and causes the non-selective total oxidation reaction.
INTRODUCTION
Vanadium oxide catalysts are widely used to carry out partial oxidation reactions.
It is commonly accepted that oxygen incorporated into the metal oxide lattice
participates in the reaction [l-3]. Oxygen is exchanged between the surface and
the bulk by lattice diffusion. Changes in the surface layer with gas phase com-
position and temperature are then mirrored in corresponding changes in the bulk
of the vanadium oxide. Variations in catalyst activity and selectivity may then
be expected to be correlated with the structure of the metal oxide. Several studies
of vanadium phosphates (VPO) have attempted to establish correlations between
activity and selectivity characteristics and various solid state properties of the
catalyst, such as the crystalline phases present, the average oxidation number
of the vanadium cation, and the degree of aggregation of the vanadium cations.
Bordes and Courtine [4] employed a variety of techniques including X-ray and
electron diffraction, infrared and uv-vis spectroscopies, and thermal gravimetric
analysis to investigate a series of VP0 catalysts used for butene oxidation to
maleic anhydride. They found that significant selectivity for partial oxidation
was observed only when VOP04 and (VO),P,O7 were present. Nakamura and co-workers
[5] suggested that a highsdegree of aggregation of vanadium ions with an average
oxidation number of +4 gave a high selectivity to maleic anhydride. Ai and others
proceed through the formation of surface hydroxyl groups, which are in equilibrium
with gas phase water vapor. The low conversions of butene result in low water
concentrations so that this equilibrium lies far to the right.
The conservation equations corresponding to the above model, which must be
solved to determine the rates of production of butadiene, selective oxidation
products and unselective oxidation products are presented in Table 2. The cell
voltage then follows from equation 4.
Several observations from our butene oxidation data were useful to note in
developing the kinetic mechanism shown above. It was found that butadiene pro-
duction and partial oxidation proceed by parallel reaction pathways. A plot of
the rate of partial oxidation (Figure 5d) versus the rate of butadiene production
(Figure 5c) for the P/V = 1.07 catalyst yields a straight line with slope 0.325.
The same trend was observed for the P/V = 1.2 catalyst as well, but with a slightly
different slope. It appears that the slope is dependent on catalyst age and
composition. It is generally believed that there is a slow loss of phosphorus from
VP0 catalysts under typical reaction conditions. Thus the dependence on catalyst
age is probably due to a gradual change in the P/V ratio. The fact that both
butadiene production and partial oxidation show the same functional dependence
on butene concentration but with different magnitudes indicates these two reactions
proceed in parallel.
The products of butene oxidation were butadiene, furan, crotonaldehyde, maleic
anhydride, acetic acid, carbon oxides and water vapor, while oxidation of butadiene
led to the formation of furan, crotonaldehyde, maleic anhydride, carbon oxides
and water vapor. The observation that acetic acid was produced only with a butene
feed and not a butadiene feed suggests that there must be at least two different
reaction paths through which the unselective products can be formed, with acetic
acid being formed from one of these paths and carbon oxides possibly arising from
both paths. Over the entire composition range of our butene oxidation experiments
the gas phase concentrations of butadiene and the partial oxidation products were
at least two orders of magnitude less than that of butene. Therefore, secondary
reactions of butadiene and partial oxidation products have been neglected in our
efforts to model the observed results, although these reactions can become
significant at higher concentration levels of these species.
The maximum in the oxidation rate as a function of butene pressure for the P/V =
1.07 catalyst suggests that the rate limiting step.is the reaction between adsorbed
hydrocarbon intermediates and an adsorbed oxygen complex. The P/V = 1.2 catalyst
differed in its behavior from the P/V = 1.07 catalyst and did not show a maximum
in the butene consumption rate. It appears unlikely that a slight excess of phos-
phorus will change the reaction mechanism. What is more likely is that changing
the P/V ratio alters the rates of the various steps involved in the mechanism to
different degrees so that the rate-determining step may change with P/V ratio.
Figure 3b suggests that the oxidation of butene over the P/V = 1.2 catalyst is
limited by the rates of adsorption of butene and/or oxygen, and not the surface
reactions.
A key.feature which emerged from this study is that in order to reproduce
the cell voltage data, the model must allow for a more active participation of
the catalyst bulk (i.e., subsurface oxygen) in the reaction network than the
simple oxygen exchange described by reaction R4. In the kinetic model the reaction
of adsorbed hydrocarbon intermediates with subsurface oxygen (MO) as well as with
surface oxygen (SO) was able to reproduce the high order functional dependence
of oxygen activity on butene concentration. To the best of our knowledge such a
participation of the subsurface oxygen in oxidation reactions over oxide catalysts
has not been considered in any previous study, although it seems plausible from
a mechanistic point of view.
We speculate that the interaction of adsorbed hydrocarbon intermediates with
subsurface oxygen results in the scisson of C-C bonds leading to unselective
oxidation products, and that the interaction of adsorbed hydrocarbons with surface
oxygen results in C-H bond scisson and selective oxidation products. Although
these are only speculations one can cite several pieces of circumstantial evidence
to support them.
(i) As the catalyst became more reduced (i.e., as [MO] decreases) we observed
a shift from carbon oxides to acetic acid as the dominant total oxidation
product, consistent within a decrease in C-C bond scisson with reduction
of the catalyst.
(ii) Although a pronounced hysteresis was observed in the cell voltage (and hence
[MO]) as a function of butene concentration for the P/V = 1.07 catalyst
(Figure 5a), no such hysteresis was detected in the rate of butadiene
production (Figure 5c) or the rate of partial oxidation (Figure 5d),
suggesting that the pathway leading to these products should not depend on
[MO]. On the other hand, a pronounced hysteresis was evident in the rate
of total oxidation, consistent with a decrease in the C-C bond scisson
with reduction of the catalyst.
(iii) Vanadium phosphate catalysts prepared in aqueous medium typically show
significant increases in selectivity toward partial oxidation as the P/V
ratio is increased above 1.0, for both butene oxidation [4,5] and butane
oxidation [16]. Hodnett and Delmon found that catalysts prepared from an
aqueous medium were quite prone to bulk oxidation and reduction [31], and
that increasing the phosphorus content in the catalysts prepared from
aqueous media restricted bulk oxidation and reduction [15]. They suggest
that this restriction of oxygen mobility in the bulk catalyst was a means
by which excess phosphorus increases selectivity [15], although the
mechanism by which the oxygen mobility is related to selectivity was not
discussed. We argue that the oxygen mobility in the bulk is a measure of
the reactivity of bulk (subsurface) oxygen with adsorbed hydrocarbon inter-
mediates. The excess phosphorus in catalysts prepared from aqueous media
will therefore decrease the reactivity of bulk oxygen with adsorbed hydro-
297
carbon intermediates. Our postulate that the reaction of bulk oxygen with
the adsorbed intermediates leads to unselective products then implies that
the selectivity towards partial oxidation should increase with increasing
phosphorus content, consistent with the experimental trends [4,5,16].
(iv) It appears from recent studies by Pepera et al. [9], Hodnett and Delmon
[15] and Buchanan and Sundaresan [I71 that vanadium phosphate catalysts
prepared in organic media have a more limited participation of the sub-
surface oxygen in the reaction that those prepared in aqueous media, but
have better selectivities toward partial oxidation than the latter. Our
postulate that the interaction of subsurface (bulk) oxygen with adsorbed
intermediates leads to unselective products is consistent with these
experimental trends as well.
The results for the more reactive metastable state of the P/V = 1.07 catalyst
in Figures 5 a-d (i.e., the results obtained in the sequence of experiments where
the butene concentration was monotonically increased) were used to estimate a
set of parameter values for the model described in Table 2. The best-fit parameter
values are presented in Table 3. The model predictions are compared with the experi-
mental data in Figures 8 a-d. Considering the complexity of the reaction system,
the agreement is excellent. The model does not predict the hysteresis displayed
by the P/V = 1.07 catalyst. This is not surprising, as the hysteresis, in our
opinion, is associated with a phase transition in the vanadium phosphate catalyst
and not a competition between the adsorption and reaction processes. The model
does not include the possibility of such phase transitions.
TABLE 3
Parameter values for kinetic model.
n 6 = IO n8 = 9.5
k3 = 1.084 x IO-* moles min -1 atm -1
k-3 = 3.101 x IO-' moles min-'
k4 = 6.084 x lO-4 moles min -1
k-4 = 6.084 x 10m4 moles min-'
k5 = 8.11 x lO-2 moles min -1 atm
-1
k6 = 6.5062 moles min -1
k7 = 3.17335 x IO-' moles min -1
kg = 8.52 x 1O-5 moles min -1
k9 = 9.594 x lO-6 moles min -1
k10 = 2.952 x 10e5 moles min-'
4:
Bu
ten
e in
G
as
Ph
ase
> -
1.0
2.0
%
Bu
ten
e in
G
as
Ph
ase
x 10
0 3 a 3
c-8
OE
zi
; E
L?
30
6
CE
o
- 0
4 2 1
.o
2.0
%
Bu
ten
e in
G
as
Ph
ase
1.0
2.0
%
Bu
ten
e in
G
as
Ph
ase
FIGURE 8
Experimental
data used to obtain parameter
estimates
and
model
predictions
for
P/V
= 1.07 catalyst;
T
15.5%
oxygen.
= 748 K,
_ -
,”
_ _
.,,
., -
I I
- ,,.
.
,. _
I..
,,.
r -
m
__
_
^_-^
.^_
I .-
299
The results for the aged P/V = 1.07 catalyst are qualitatively similar to those
modelled, and can be produced by the model presented with appropriate changes in
the rate constants. The results for the P/V = 1.2 catalyst, while qualitatively
different from those for the fresh P/V = 1.07 catalyst, can also be produced by
changing the rate constants so that the adsorption step becomes rate-limiting.
Therefore, by modclling the most complex of the three data sets, we have obtained
a mechanism which qualitatively accounts for all three.
In addition to being sufficiently active and selective, vanadium phosphate
catalysts used in industry to carry out maleic anhydride synthesis must also be
robust. The role played by excess phosphorus in imparting robustness to the VP0
catalysts has been considered by several investigators. For example, Nakamura and
co-workers [5] and Trifiro et al. [32] have shown that when phosphorus is present
at P/V ratios of one or greater the +4 oxidation state of the vanadium is
stabilized. The present study demonstrates the beneficial role of excess phosphorus
in a dramatic way. The P/V = 1.07 catalyst could be oxidized or reduced by a slow
process, which led to slow variation in the activity and selectivity. This gave
rise to the hysteresis in the catalyst behavior where the catalyst could exist
in two different states with the same gas phase composition, as shown in Figures
5 and 6. Although by suitable (either oxidizing or reducing) pretreatment, the
state of the catalyst could be changed reversibly, such was not possible at a
lower temperature where the catalyst remained frozen into either a selective or
unselective state. The P/V = 1.2 catalyst which contains more phosphorus than the
P/V = 1.07 catalyst was apparently frozen into a single state and showed uniform
behavior independent of its history. Industrial VP0 catalysts typically contain
P/V > 1.1 and generally are very robust and do not appear to undergo changes even
under severe conditions [33,34]. The hysteresis characteristics of the P/V = 1.07
catalyst suggest that it may be possible to periodically treat VP0 catalysts to
altered feed conditions to improve overall process performance.
CONCLUSIONS
It has been shown that solid electrolyte potentiometry can be used as an in
situ probe of oxygen activity in vanadium phosphate catalysts under reaction
conditions. This is a very sensitive probe to the presence of hydrocarbons and
shows that the adsorption of oxygen is far from equilibrium during reaction con-
ditions. The SEP data about the oxygen activity could be coupled with reaction
rate data to help elucidate the behavior of VP0 catalysts. The findings of this
study are:
i) The rate limiting step in the oxidation reaction over the P/V = 1.07 catalyst
is the bimolecular surface reaction of adsorbed hydrocarbon intermediates with
surface or subsurface oxygen. On the other hand, the oxidation of butene over the
P/V = 1.2 catalyst is limited by the adsorption of butene and/or oxygen and not
the surface reactions.
300
ii) The interaction of adsorbed hydrocarbon intermediates with subsurface
oxygen results in unselective products.
iii) The selective oxidation step requires the reaction of adsorbed hydrocarbon
intermediates with surface oxygen complexes exclusively.
iv) A more reduced catalyst is more selective for partial oxidation.
v) Oxidation and reduction of the bulk of VP0 catalysts is slow, and metastable
states of the P/V = 1.07 catalyst could be prepared at higher temperatures and
frozen into the catalyst to obtain good or bad selectivity.
ACKNOWLEDGEMENT
We wish to thank Amoco Chemical Corporation, the Petroleum Research Fund,
administered by the American Chemical Society (PRF #13482-AC5), and the Camille
and Henrey Dreyfus Foundation for their financial support of this work. We also
wish to thank Bendix-Autolite for supplying us with the electrolyte thimbels used
in this work.
REFERENCES
1 P. Mars and D.W. van Krevelen, Chem. Eng. Sci., 3 (Special Supplement) (1954) 41.
2 W. Ueda, Y. Moro-oka and T. Ikawa, J. Chem. Sot., Faraday Trans. I, 78 (1982) 495.
3 M.S. Wainwright and T.W. Hoffman, Can. J. Chem. Eng., 55 (1977) 552. 4 E. Bordes and P. Courtine, J. Catal., 57 (1979) 236. 5 M. Nakamura, K. Kawai and Y. Fujiwara, J. Catal., 34 (1974) 345. 6 M. Ai and S. Suzuki, Bull. Chem. Sot. Japan, 47 (1974) 3074. 7 M. Ai, Bull.Chem. Sot. Japan, 43 (1970) 3490. 8 J.S. Buchanan, J. Apostolakis and S. Sundaresan, Appl. Catal., 19 (1985) 65. 9 M.A. Pepera, J.L. Callahan, M.J. Desmond, E.C.Milberger, P.R. Blum and N.J.
Bremer, J. Amer. Chem. Sot., 107 (1985) 4883. 10 T.P. Moser and G.L. Schrader, J. Catal., 92 (1985) 216. 11 G.L. Simard, J.F. Steger, R.J. Arnott and L.A. Siegel, Ind. and Eng. Chem.,
47 (1955) 1424. 12 F. Cavani, G. Centi, I. Manenti, A. Riva and F. Trifiro, Ind. Eng. Chem. Prod.
Res. Dev.: 22 (1983) 565. F. Cavani, G. Centi and F. Trifiro, Ind. Eng. Chem. Prod. Res. Dev., 22 (1983) 570. B.K. Hodnett, Catal. Rev.-Sci. Eng., 27 (1985) 373. B.K. Hodnett and B. Delmon, J. Catal., 88 (1984) 43. B.K. Hodnett, P. Permanne and B. Delmon, Appl. Catal., 6 (1983) 231. J.S. Buchanan and S. Sundaresan, Appl. Catal., in press. A. Kvist, in Physics of Electrolytes, (J. Hladik, ed.), Vol.1 (1972) Academic Press, London. W.D. Kinger, J. Pappis, M.E. Doty and D.C. Hill, J. Am. Cer. Sot., 42 (1959) 393.
20 S. Pancharatnam, R.A. Huggins and D.M. Mason, J. Electrochem. Sot., 122 (1975 869.
21 C.G. Vayenas and H.M. Saltsburg, J. Catal., 5 7 22 M. Stoukides and C.G. Vayenas, J. Catal., 64 ( 23 J.N%* Michaels and C.G. Vavenas. J. Catal., 85 24 H. Okamoto, G. Kawamura and T.-Kudo, J. Catal 25 H. Okamoto; G. Kawamura and T. Kudo, J. Catal 26 H. Okamoto. G. Kawamura and T. Kudo. J. Catal 27 H. Okamoto; G. Kawamura and T.Kudo,- J. Catal 28 I.S. Metcalfe and S. Sundaresan, Chem. Eng. S
29 R.A. Mount and H. Raffelson, U.S. Patent 4,337,174, 1982. 30 E.M. Breckner, Ph. D. Thesis, Princeton University, 1986. 31 B.K. Hodnett and B. Delmon, Ind. Eng. Chem. Fundam., 23 (1984) 465. 32 G. Poli, I. Resta, 0. Ruggeri and F. Trifiro, Appl. Catal., 1 (1981) 935. 33 R.A. Schneider, U.S. Patent 3,864,280, 1975. 34 E.C. Milberger, N.J. Bremer and D.E. Dria, U.S. Patent 4,333,853 (1982).