RESEARCH TRIANGLE INSTITUTE
RTI Project No. 475U-3409
June 1988
Mixture Effects in the Catalytic
Oxidation of VOCs in Air
Prepared by
Research Triangle Institute
Research Triangle Park, NC 27709
Prepared for
U.S. Environmental Protection Agency
Air and Energy Engineering Research Laboratory
Research Triangle Park, NC 27711
and
U.S. Air Force Engineering and Services Center
Tyndall Air Force Base, FL 32403
under EPA Cooperative Agreement CR812522
and
U.S.A.F. Interagency Agreement RN57931254
POST OFFICE BOX 121 94 RESEARCH TRIANGLE PARK, NORTH CAROLINA 27709-21 94
-------
RTI Project No 475U-3409 June 1988
Mixture Effects in the Catalytic
Oxidation of VOCs i
Prepared by
S Gangwal, K Ramanathan, P Caffrey,
M Mullms, and J. Spivey
Research Triangle Institute
P O Box 12194
Research Triangle Park, NC 27709
EPA Cooperative Agreement CR812522
and
US A F Interagency Agreement RN57931254
EPA Project Officer Michael Kosusko
Air and Energy Engineering Research Laboratory
Research Triangle Park, NC 27711
Prepared for
U S Air Force U S Environmental Protection Agency
Engineering and Services Center Office of Research and Development
Tyndall Air Force Base. FL 32403 Washington, DC 20460
-------
TABLE OF CONTENTS
Section Page
1.0 Executive Summary 1
2.0 Introduction 2
3.0 Experimental 7
4.0 Results 15
5.0 Development of a Reactor Model for Predicting Deep Oxidation
Behavior of Hydrocarbon Mixtures 31
6.0 Autocatalysis in Deep Oxidation of Ethyl Acetate 62
7.0 Conclusions and Recommendations 68
8.0 Data Qua 1 i ty 69
9.0 Nomenc lature 73
10.0 References 76
-------
LIST OF FIGURES
Number Page
3-1. Microreactor system 10
3-2. Sequential analysis of reactor inlet (bypass)
and reactor outlet with a 10-port "zero" volume valve 12
4-1. Rate controlling regimes in a catalytic combustor 16
4-2. Effect of multicomponent mixtures on n-hexane conversion 17
4-3. Effect of multicomponent mixtures on benzene conversion 18
4-4. Effect of multicomponent mixtures on ethyl acetate
conversion 19
4-5. Effect of multicomponent mixtures on MEK conversion 20
4-6. Comparison of the effect of benzene on ethyl acetate
and MEK conversion 22
4-7. First order rate constants for the oxidation of single
components 24
4-8. First order rate constants for oxidation in a ternary
mixture 25
4-9. Differential reaction rate for n-hexane oxidation 27
4-10. Differential reaction rate for benzene oxidation 28
4-11. Differential reaction rate for ethyl acetate oxidation 29
4-12. Differential reaction rate for MEK oxidation 30
5-1. Inverse reaction rate versus inverse concentration
for n-hexane oxidation 33
5-2. Inverse reaction rate versus inverse concentration for
ethyl acetate oxidation 34
5-3. Inverse reaction rate versus inverse concentration for
MEK oxidation 35
5-4. Mars/van Krevelen rate constants for n-hexane oxidation 37
5-5. Mars/van Krevelen rate constants for ethyl acetate 38
5-6. Mars/van Krevelen rate constants for MEK oxidation 39
5-7. Mars/van Krevelen rate constants for benzene oxidation 40
iii
-------
5-8. Comparison of predicted and actual conversions for
benzene and n-hexane in a binary mixture 49
5-9. Comparison of predicted and actual conversions for
ethyl acetate and n-hexane in a binary mixture 50
5-10. Evaluation of collision integral for modeling purposes
in the temperature range of interest 55
5-11. Comparison of model predictions and experimental data for
n-hexane conversion as a single component 57
5-12. Comparison of model predictions and experimental data for
benzene conversion as a single component 58
5-13. Comparison of model predictions and experimental data for
binary benzene/n-hexane mixture 59
5-14. Comparison of model predictions and experimental data for
ethyl acetate conversion as a single component 60
6-1. Comparison of ethyl acetate conversions in dry and
humidified air streams 64
6-2. Comparison of MEK conversions in dry and humidified air
streams 67
IV
-------
LIST OF TABLES
Number Page
2-1. Approximate range of independent variables for this
research 4
3-1. Test gas mixtures 8
3-2 Catalyst characteristics 9
3-3. Experimental conditions 14
5-1. Mars/van Krevelen rate equation parameters 41
5-2. MuIticomponent oxidation model with internal diffusion 46
5-3. Multicomponent oxidation model in non-dimensional form 47
5-4. Catalyst physical properties for modeling and Deff(,)
calculation 52
8-1. Precision of data 70
8-2. Accuracy of data 72
-------
1 0 EXECUTIVE SUMMARY
Most volatile organic compound (VOC) releases into the environment are
mixtures, and rarely consist of a single component. However, the bulk of
fundamental studies of the catalytic deep oxidation of such compounds are
usually confined to single components This study examines the deep oxidation
of organic mixtures over a heterogeneous catalyst in an attempt to explain
e.^lier observations by Tichenor. et al (1S87) concerning the apparent
inhibition or enhancement of oxidation of some components to establish a
scientific basis for the design and operation of catalytic incineration
systems for VOC control. To elucidate these effects, the oxidation kinetics
of n-hexane. benzene, ethyl acetate, and methyl ethyl ketone in air were
examined over a commercial catalyst (0 1 percent Pt/ 3 percent Ni on v-
alumina.) Reaction rates of these components individually were determined at
temperatures of 150°C to 360°C from differential reactor studies. When these
were compared to overall destruction efficiencies from integral reactor
studies for both individual compounds and mixtures, the Mars/van Krevelen
(MVK) reaction rate model satisfactorily represented the results for some
single organic compounds at lower temperatures By incorporating pore
diffusion effects, the MVK model adequately explains the single component data
over the entire temperature range for some of the compounds A multicomponent
MVK model incorporating competitive adsorption effects is moderately
successful in predicting the observed behavior for a binary mixture of benzene
and n-hexane. however, this model can not predict the apparently enhanced
reaction rate observed for ethyl acetate at higher temperatures (>220°C)
Other reaction pathways available for compounds with carbon-oxygen linkages
and/or the advent of catalytically supported homogeneous combustion with free
radical precursors may explain this phenomenon The enhancement of ethyl
acetate conversions in humidified air streams suggests that autocatalysis by
product water may be a possible mechanism
-------
2 0 INTRODUCTION
The goal of this Cooperative Agreement is to establish a scientific basis
for the selection and evaluation of heterogeneous catalysts and operating
conditions for the control of gas streams containing mixtures of volatile
organic compounds (VOCs) The research has been devoted to both experimental
evaluation of the catalytic oxidation of VOC-contaimng mixtures and to the
kinetic interpretation/modeling of the results
Heterogeneous catalytic oxidation of organic compounds is an important and
intensely studied area Much research has been reported on industrially
important partial oxidation reactions and catalytic oxidation of automotive
exhaust Catalytic oxidation is often used instead of direct thermal oxidation
for control of VOC emissions from a number of sources due to its lower
operating temperature which results in lower NOX emissions and lower operating
cost (primarily due to lower fuel consumption). However, fundamental studies
of the behavior of complex and sometimes incompletely characterized mixtures of
VOCs from operations such as air stripping of contaminated groundwater are
relatively scarce It is the purpose of this Cooperative Agreement to address
this need
Of particular interest is the interaction of one VOC with others in a
multicomponent gas phase mixture. As such a mixture passes over a catalyst
designed to achieve some level of environmental control, this interaction has
extremely practical implications For example, in the oxidation of chlorinated
VOCs. it is possible in principle to produce a gas with higher toxicity than
the inlet gas by incomplete oxidation of the VOC Also of interest are the
relative interactions of classes of VOCs that might be encountered in
environmental control situations. For instance, if some generalization could
be made that aromatics usually enhance the oxidation of ketones, or that the
presence of chlorinated solvents usually inhibits the oxidation of alkanes but
not esters, this would allow the Agency or the US. Air Force to make some
preliminary judgements when faced with the problem of controlling VOC-
-------
containing gas streams. It is likely that any such generalizations would be
highly dependent on the catalyst and reaction conditions. While it is not
within the scope of this research to characterize the behavior of all likely
VOCs on all applicable catalysts at all typical conditions, we hope to describe
quantitatively, through experiments and kinetic modeling, the behavior of a few
well-defined VOC mixtures containing different classes of compounds
Experimental conditions are shown in Table 2-1 We have taken care in
selecting these compounds, catalysts, and reaction conditions to help ensure
that the results are of the most direct use for the Agency and the U S Air
Force
From the literature pertinent to this study (Spivey, 1987), several
conclusions can be drawn
• Catalysts which have been shown to be effective for VOC control at
the conditions of Table 2-1 include supported noble metals (primarily
Pt, with and without base metal oxides such as NiO), single metal
oxides (e g.. C^O^ or ^205) and mixed metal oxides (e g .
hopcalite). As a very general rule, supported noble metals are more
active, more easily poisoned (especially by halogenated VOCs), and
more widely used.
• Applications of catalytic oxidation for VOC control have primarily
involved direct transfer of technology from related applications.
such as the use of an automotive catalyst downstream of a burner.
Little has been done to develop catalysts specifically for this need.
• The use of catalytic oxidation as a VOC control technique is more
widespread in Europe than in the United States, perhaps because of
higher energy costs and more stringent environmental regulations
(Acres. 1970)
• Very little has been reported on the systematic scientific study of
VOC mixtures. Reported work tends to be either well characterized
research on pure components in air or anecdotal studies of "real"
streams where only gross performance is reported
-------
TABLE 2-1. APPROXIMATE RANGE OF INDEPENDENT VARIABLES FOR THIS RESEARCH
Temperature 150 - 400°C
Pressure 1 atm
Space velocity 104 - 106 hr"1
Reactant concentration 101 - 10^ ppm
Reactants Chlorinated solvents, alkanes,
aromatics, and oxygenated hydro-
carbons (ketones. acetates, esters)
Humidities 0 - 80% RH
Oxidant Air
-------
• Modeling of the catalytic oxidation of VOCs at conditions of interest
may be complex, especially for mixtures, because of both surface
kinetic and mass transfer effects that can vary with exoenmental
conditions These effects may result in inhibition or enhancement of
the oxidation of a given compound
Most research on catalytic oxidation of VOCs tends to involve either
single pure components, with relatively complete analysis of reaction
products, or field studies, with only the overall performance (e g., "percent
removal") being reported From the relatively few systematic studies of
mixture effects in catalytic oxidation of VOCs. several conclusions can be
drawn.
Though some evidence exists that partial catalytic oxidation of mixtures
of closely related compounds can be predicted directly and linearly from
single component behavior (Emmett, 1960). most studies show inhibition of the
oxidation of one or all of the compounds in a mixture over their behavior as
single components. Satterfield (1980) states that this inhibition can De-
accounted for by competition among the reactants (VOCs in this case) for
adsorption sites on the catalyst. Yao (1973), in studying primarily C2 and C3
hydrocarbons over «-Cr203 further suggests that the observed inhibition in
hydrocarbon mixtures can be explained by competition among the hydrocarbons
for adsorbed 02 or for undefined "adsorption sites" (presumably lattice oxygen
from the «-Cr203)
Cullis et al. (1970) found another type of mixture effect in studying
methane oxidation over Pd catalysts. Propane, when added to methane.
significantly retarded methane oxidation but was itself oxidized very little
Perhaps some trace level of a compound generated by the oxidation of a very
small amount of propane was sufficient to retard methane oxidation. A similar
effect might take place in a VOC mixture
Much work out of the Warren Springs Lab in England (Pope et al. 1976 and
1978. Heyes et al, 1982) has examined the catalytic oxidation of hydrocarbons
in mixtures with sulfur-containing species on some single metal oxides and
supported Pt The general conclusion from this work is that the sulfur-
containing compounds suppressed hydrocarbon oxidation and that small
concentrations of products of incomplete combustion were almost always
observed, though not identified
-------
Tichenor and Palazzolo (1987) also studied mixture effects and observed
enhancement, inhibition, and no effect on some VOCs in mixtures compared to
their oxidation as single components They also identified quite a number of
the incomplete combustion products
The conclusion from these studies can be no more specific than that
mixture effects are typically observed and. at least to date, are fairly
unpredictable It is the goal of this research to extend the current
understanding
-------
3 0 EXPERIMENTAL
3 1 Chemicals
Benzene, n-hexane, ethyl acetate, and methyl ethyl ketone (MEK) were
chosen for this study as representing the classes of aromatic, aliphatic and
oxygenated species respectively. The gases were obtained premixed with
hydrocarbon free, dry air in gas cylinders at concentrations ranging from 10
to 600 ppmv hydrocarbon from Scott Specialty Gases, Plumsteadvilie, PA. A
number of gases containing single components, binary mixtures, and ternary
mixtures were obtained as shown in Table 3-1. The gases were used one at a
time in the deep-oxidation experiments.
3.2 Catalysts
The catalyst was obtained from United Catalyst, Inc . Louisville. KY. as
1/4-in (6.35 mm) cylinders. The commercial designation of the catalyst was
G-43, and its nominal composition as supplied by the vendor was 0.1 percent
Pt, 3 percent Ni/r-A^C^. The catalyst was crushed and screened to 120 x 170
mesh prior to use in the oxidation experiments. The physical properties of
the catalyst, as specified by the vendor and as measured in RTI's laboratory.
are given in Table 3-2.
A laboratory-scale microreactor system, shown in Figure 3-1. was used for
all the experiments The reactor consisted of a 60-cm-long Pyrex tube (6 mm
0 D . 4 mm I.D for the 60 x 80 mesh catalyst: 3 mm CD. 1 5 mm I D for the-
120 x 170 mesh catalyst), and the catalyst was positioned at the center of the
tube using fine Pyrex wool plugs The tube was surrounded by a three-zone 45-
cm-long split tube furnace. The end zones were each 7.5-cm-long and the
central zone was 30-cm-long The temperature of each heating zone was
controlled separately using proportional temperature controllers. A single
zone furnace was used for the experiments involving MEK. which were earned
out after a redesign of the experimental setup Four 0.5-mm 0 D measurement
-------
TABLE 3-1. TEST GAS MIXTURES
Concentration in Air (ppmv)
Single Hydrocarbon benzene: 9, 69, 163. 375, 525
ethyl acetate: 53, 109, 238. 450
n-hexane; 201, 410, 566
methyl ethyl ketone (MEK): 25, 50, 70
135, 190. 298
Binary Mixtures
193 ppm benzene + 172 ppm ethyl acetate
189 ppm benzene + 190 ppm n-hexane
184 ppm ethyl acetate + 190 ppm n-hexane
149 ppm n-hexane + 135 ppm MEK
160 ppm benzene + 135 ppm MEK
Ternary Mixtures 143 ppm benzene + 174 ppm ethyl acetate +
90 ppm n-hexane: 103 ppm n-hexane + 104 ppm
benzene + 93 ppm MEK
-------
TABLE 3-2. CATALYST CHARACTERISTICS
Vendor
Measured
Composition: 0.1% Pt, 3% Ni/Y-
Form:
1/4-in. (6.35 mm) extrudates
(crushed to 120 x 170 mesh and
60 x 80 mesh prior to use)
BET Surface Area: 212 m2/g
Pore Volume:
0-150 nm (Nitrogen BET Method)
1.4 to 14 nm (CC14 Method): 0.26 cm3/g
1.4 to 80 nm (CC14 Method): 0.28 cm3/g
Mean Pore Radius (estimated)
Particle Density
Pore Volume Distribution (Mercury
Porosimetry. 60.000 psi [414 MPa]):
175 to 12um
12 urn to 80 nm
80 to 14 nm
14 to 2.9 nm
0.00 cm3/g
0.08 cm3/g
0.01 cm3/g
0.23 cm3/g
3.18% Ni, 48.2% Al,
46.3% Oxygen*
161 m2/g
0.364 cm3/g
3.6 nm
1.46 g/cm3
* Pt below detection limit of ESCA [electron spectroscopy for chemical
analysis]
-------
Control Thermocouple^^ y Measurement Thermocouple!
GC Garner Gas
Valve-Oven with 10-Port Valve
(Figure 2)
FC: Differential Flow Controller
PR: Two-stage Pressure Regulator
BFMV: Soap Bubble Flow Meter and Vent
•> Process lines
Electrical
Integrator
Actuator
Controller and
Timer
Figure 3-1 .. Microreactor system.
-------
thermocouples (type K) were mounted, equally spaced, along the small catalyst
bed to ensure temperature uniformity The temperature control was oetter chan
±0 5°C in each zone The premixed gas was delivered using a two-stage
pressure regulator with the gas-flow split into two equal flows using two
differential flow controllers as shown One flow stream was passed through
the imcroreactor. whereas the other was passed through a byoass line. After
the two split gas streams flowed through the reactor and the bypass line
respectively, they entered a thermostatted Valco 10-port gas sampling valve
maintained at 100°C The 10-port valve is a unique device that allows
sequential sampling of the reactor outlet stream and the bypass stream (i.e.,
representative of the reactor inlet stream) using two identical 0.25-cm3
sample loops
Sequential sampling using this valve is illustrated in Figure 3-2. The
samples were injected automatically, using a Carle Valve Minder consisting of
a valve actuator and a valve timer, into a Varian 3700 gas chromatograph (GC)
equipped with a flame-ionization detector (FID) A 5 ft (1.5 m) long. 1/8-in
(3 mm) stainless-steel column consisting of 3 percent SP 1500 on 80/120 rnesh
Carbopack B was used for the separation and analysis of the hydrocarbons
isothermally at 140°C A Carle 111 GC with an FID was used in the later
experiments. A complete analysis of the three hydrocarbons of interest could
be performed in 4-1/2 minutes The valve timer was placed on a 9-mmute cycle
to allow the measurement of the hydrocarbons in the reactor inlet and in the
reactor outlet streams sequentially Because the inlet concentration of the
gas stream was a known constant, the outlet concentration was obtained
assuming a linear response of the FID with respect to the inlet concentration
for the hydrocarbons of interest The analysis of the inlet during every
cycle is extremely important because of variation of the FID response during
the course of a typical day of the experiment Even though the FID response
may drift or cycle during the day, it would affect the outlet and the inlet
stream concentrations in a fixed ratio Thus, the sampling procedure allowed
accurate measurement of the reactor outlet concentrations, irrespective of the
drift in the FID response The peak areas were measured electronically using
a Perkin-Elmer Sigma 10B Data Station. This data station was also used to
time the automatic injections in the later experiments The bypass and
reactor outlet gas flows were measured using a soap bubble flow meter to
11
-------
GC Column
GC Carrier Gas
Reactor
Out
In
Detector
Out
In
Bypass
Valve Position A: Analyze reactor outlet
GC Carrier Gas
Reactor
Out
In
GC Column
Detector
Out
In
Bypass
Valve Position B: Analyze reactor inlet (bypass]
Figure 3-2. Sequential analysis of reactor inlet (bypass) and reactor
outlet gases with a 10-port "zero" volume valve.
12
-------
ensure that they were equal This equality of flow was necessary for accurate
determination of the reactor outlet concentration
3.4 Procedure
Typical experimental conditions (including reactor size, catalyst parficle
size, temperature range, pressure, and space velocities) are shown in
Table 3-3. The experimental procedure consisted of measuring the reactor
inlet and outlet concentrations at a series of constant temperatures and gas
flow rates for each gas Integral reactor high-conversion experiments were
carried out initially Single-component gases were run first, followed by the
binary mixtures and the ternary mixture. These experiments allowed gross
evaluations of the regimes of surface reaction, pore diffusion, external
diffusion, and catalytically supported thermal reaction These experiments
also allowed us to establish whether or not mixture effects were present to a
significant degree More fundamental differential reactor experiments (less
than 10 percent conversion) were then carried out for each pure component in
the surface reaction controlled region to evaluate its rate expression These
rate expressions are used to explain the results of the integral reactor
experiments and to evaluate mixture effects.
13
-------
TABLE 3-3. EXPERIMENTAL CONDITIONS
Pressure:
Temperature Range:
Space Velocity at Reaction
Temperature:
Catalyst Bed:
Hydrocarbon Concentration
in Air:
Catalyst Particle Size:
Ratio of Reactor Diameter to Particle
Diameter
Atmospheric
140 to 360°C (isothermal
operation)
50.000 to 1.000,000 h-
17 to 20 mg
10 to 600 ppmv
120 to 170 mesh
20:1
14
-------
4.0 RESULTS
During initial exposure of the 0.1-percent Pt. 3-percent
catalyst to an air stream containing n-hexane at approximately 315°C, the
catalyst exhibited a prolonged induction period (up to 72 hours) during which
catalyst activity increased by nearly 2 orders of magnitude. Based on
scanning electron microscopy/energy-dispersive x-ray analysis experiments, we
believe that this induction period was caused primarily by slow redispersion
of the Pt on the surface of the catalyst. The activity increased further at a
higher reaction temperature of 365°C until a stable activity was reached.
Once the steady-state activity was reached, it remained stable throughout the
duration of the experiments. Only the steady-state results are presented.
Before presenting the results, it is instructive to consider the three
operating regions of an industrial catalytic combustor as designated by Prasad
et al. (1984) and shown in Figure 4-1. Generally, the inlet fuel/air mixture
temperature is low, and the rate of combustion (oxidation) is controlled by
intrinsic kinetics (region A). As temperature rises along the bed. the rate
increases exponentially to a point where it is not possible to supply the
reactant to the catalyst surface fast enough to keep up with the rate of
reaction. This marks the beginning of the mass-transfer-controlled region
(region B). The rate in region B may be controlled by external diffusion,
internal (pore) diffusion, or a combination of both, depending on hydrodynamic
conditions and catalyst physical structure. Farther downstream in the bed,
the exothermic oxidation reactions increase the temperature to a point where
gas-phase reactions begin to occur simultaneously with the catalytic
reactions. This region is designated as the catalytically supported
homogeneous reaction region (region C). where the rate again increases
exponentially.
Figures 4-2 through 4-5 show the conversion data for n-hexane. benzene,
ethyl acetate and methyl ethyl ketone (MEK) as a function of temperature in
various mixtures. All these results were obtained at an identical weight
15
-------
2
CO
oc
o
°S
§
o
OC
Catalytically Supported
Homogeneous Reaction
(Region C)
Mass Diffusion
Controlled
(Region B)
Surface Kinetics
Controlled
(Region A)
Temperature
After Prasadetal. (1984)
Figure 4-1. Rate controlling regimes in a catalytic combustor.
(reproduced with permission from Marcel Dekker, Inc., N.Y., N.Y.)
16
-------
o
I
0)
£
110
100 -
90 -
80 -
70
60
50
40
410 ppm n-hexane in air
• 190 ppm n- hexane with
184 ppm ethyl acetate in air
• 190 ppm n- hexane with
189 ppm benzene in air
• 90 ppm n- hexane with
174 ppm ethyl acetate and
143 ppm benzene in air
17 mg. 120 X 170 mesh catalyst
WHSV = 209
140
180
I l i
220 260 300
Temperature (°C)
340
380
Figure ^~2 . Effect of multicomponent mixtures on n-hexane conversion.
17
-------
o
e
01
O
U
**
0)
£
110
100
90
80
70
60
50
40 '
30 -
20 •
10 -
375 ppm benzene in air
* 193 ppm benzene with
172 ppm ethyl acetate in air
T 189 ppm benzene with
190 ppm n-hexane in air
• 143 ppm benzene with
174 ppm ethyl acetate and
90 ppm n-hexane in air
17 mg. 120 X 170 mesh catalyst
WHSV = 209
1 ' I I I 1 1 |—
140 180 220 260 300
Temperature (°C)
1
340
380
Figure 4-3.Effect of multicomponent mixtures on benzene conversion.
18
-------
c
g
0)
c
0
o
c
01
k.
O)
a.
1 IU
100
90 -
80 -
70 -
60 '
50 -
40 -
30 •
20 •
10 -
V 450 ppm ethyl acetate in air ^ -
• 172 ppm ethyl acetate with *
193 ppm benzene in air T
• 184 ppm ethyl acetate with
190 ppm n-hexane in air
• 174 ppm ethyl acetate with
143 ppm benzene and
90 ppm n-hexane in air
^^^
1 7 mg, 1 20 x 1 70 mesh catalyst •*
WHSV = 209
T
*
*
t * T
f. **** T
I I I I 1 I I i i I I
140 180 220 260 300 340 38(
Temperature (°C)
Figure 4-4. Effect of multicomponent mixtures on ethyl acetate conversion.
19
-------
c
o
C
0)
>i
c
0,
o
C;
o>!
1
11U
100
90
80
70
60
50
40
30
20
10
Q
I I *
A
*
—
—
^
A
WHSV = 209
" 20 mg( 120 x 170
mesh catalyst
T
»
- •
1.1,1,1,1,
140 180 220 260 300 340 38
Temperature (°C)
297 ppm MEK in air
135 ppm MEK and 160 ppm benzene in air
135 ppm MEK and 149 ppm hexane in air
T 93 ppm MEK, 103 ppm n-hexane
and 104 ppm benzene in air
Figure 4-5. Effect of multicomponent mixtures on MEK conversion.
20
-------
hourly space velocity (WHSV) of 209. where WHSV is defined as the weight of
the flowing gas per unit weight of catalyst per hour. A direct comparison of
these figures with Figure 4-1 is not possible because the abscissas are
percent conversion in one case and reaction rate in the other. Direct
conclusions regarding the operating regions of our data is thus not possible
without measurement of reaction rates.
As shown in Figure 4-2. the conversion of n-hexane is significantly
inhibited in mixtures versus as a single component. The inhibition at low
temperatures (<260°C) is highest in the presence of benzene. From Figure 4-3,
the conversion of benzene appears to be nearly unaffected in the mixtures.
From Figure 4-4. the conversion of ethyl acetate is significantly enhanced in
the mixtures. (Enhancement is highest in the mixtures containing benzene).
These observations confirm those of Tichenor and Palazzolo (1987). The
enhancement of ethyl acetate conversion led to the selection of another
oxygenated species, MEK, for mixture studies in order to see if enhancement
was a trend for oxygenated species. From Figure 4-5, MEK conversion is
enhanced but not as much as ethyl acetate. The pure compound reactivity of
MEK is comparable to benzene. Figure 4-6 compares mixture effects of ethyl
acetate and MEK in binary mixtures with benzene.
As pointed out earlier, all data in Figures 4-2 through 4-5 are at the
same space velocity. Another way to present the data would be in terms of
reaction-rate constants, which allows data at different space velocities to be
represented concisely as shown in Equation (1) below. For deep oxidation of
trace organics, the most popular assumption in the literature for kinetics is
first order in terms of the organic concentration (Hawthorn. 1974; Morooka et
al., 1967; Sadamori et al., 1981; Prasad et al., 1984). Also, for oxidation
of mixtures, it is usually assumed that each specie is oxidized independently
(Prasad et al., 1984). (It is not implied here that the rate of oxidation of
a specie in a mixture is identical to that as a single component.) Because
all our experiments were carried out in large excesses of oxygen, a first-
order kinetics assumption would enable us to determine a pseudo first-order
rate constant for each oxidation reaction. With the further assumption of
plug flow in the laboratory tubular reactor, a differential mass balance
followed by integration yields the following equation for the pseudo first-
order rate constant:
21
-------
c
o
C
> !
c
3;
*-
c1
0) ;
ll
Q. '
j
i
IIU
100
90
80
70
60
50
40
30
20
* * * sr
•
A
D
-
^
0
A
-
a
•
A
^
A D
tWHSV •= 209
t A Q
n
_io_l ID I i i i . , , ,
140 180
220 260 300 340
Temperature (°C)
380
a 450 ppm ethyl acetate in air
A 172 ppm ethyl acetate and 193 ppm benzene in air
• 297 ppm MEK in air
• 135 ppm MEK and 160 ppm benzene in air
Figure 4-6. Comparison of the effect of benzene on ethyl acetate and MEK
22
conversion.
-------
(WHSV) Cio
kps In (1)
3600 PQJ Ci
where PGT is the density of the gas (air) at temperature T.
kps may then be represented by the familiar Arrhenius equation
kps = A exp [-E/RT] . (2)
It is emphasized here that the above treatment should be regarded only as
a way of representing the data because first-order kinetics may not apply.
The advantages of the above equations are their simplicity and ability to
represent a large amount of data concisely. It should also be noted that
space velocity is incorporated into the first order equation so that data at
different space velocities may be represented. A further advantage is that
the Arrhenius plot may allow determination of temperature ranges for regions
of surface-reaction control and mass-transfer control, as evident from Figure
4-7.
Figure 4-7 shows the first-order rate constants for the pure components.
The order of reactivity of the four compounds is benzene ~ MEK > n-hexane >
ethyl acetate. The inlet concentrations of the gases for these data were 325,
297, 410. and 450 ppmv, respectively, for benzene. MEK, n-hexane. and ethyl
acetate. The n-hexane plot exhibits apparent surface-reaction region below
about 220°C. Also, linearity of the plot in the surface-reaction region
suggests that first-order kinetics may be a reasonable approximation for n-
hexane oxidation. The Arrhenius plot for benzene also shows apparent first-
order behavior at low temperatures, but the plot for ethyl acetate shows a
distinct curvature indicating that first-order kinetics may not be a
reasonable representation. The plot for benzene also appears to show a shift
into the mass-transfer region at a temperature just below 220°C, similar to
the n-hexane plot.
Figure 4-7 for the single components is to be contrasted with the rate
constants of the components in a ternary mixture (Figure 4-8). Ethyl acetate
is now more reactive than n-hexane, and n-hexane's plot shows a curvature.
indicating that first-order kinetics may not be obeyed in the mixture.
Obviously, complex intercomponent interactions are present when mixtures are
oxidized.
The apparent inhibitions and enhancements shown in Figures 4-2 through 4-
6 may be caused by: (1) mixture effects involving competition for active
23
-------
7-
6-
5-
4-
3J
2-
1-
0-
-1 _
-2
10
360 320
1.5
i
1.7
Temperature (°C)
280 240 200
160
1.9 2.1
1,000/T (K)
2.3
A benzene, 375 ppm
a n-hexane, 610 ppm
T MEK, 297 ppm
• ethyl acetate, 450 ppm
Figure 4-7. First order rate constants for the oxidation of single components
24
-------
7-
6-
5.
4.
3-
2.
1 -
0 -
-1 -
-2
1000
100
10
co
u
360 320
Temperature (°C)
280 240 200
160
1.5
1.7
1.9
1,000/T (K)
—i—
2.1
2.3
• benzene
D hexane
A MEK
V ethyl acetate
Figure 4-8. First Order rate Constants for oxidation in a ternary mixture.
25
-------
sites and differences in external and internal mass-transfer rates, and (2)
mixture effects involving surface modification, cocatalysis or inhibition by
one of the reacting species or intermediates (including the products CO2 and
H20). Of the reasons listed above, (1) appears to be mathematically tractable
from surface kinetic data for single component given that a suitable kinetic
model for the reaction of the various species in a mixture may be developed.
Once this is done, and the effects are still not explained, then a combination
of (1) and (2) above or some other unknown complex interactions are probably
involved. To sort this out, it was deemed necessary to obtain intrinsic
kinetic rates for the four components individually. This was done by
operating the reactor as a differential reactor (conversions between 5 and 8
percent) for the four components individually at a series of concentrations
and temperatures. The temperatures chosen were generally below 220°C (as
guided by the previous experiments) to ensure that data were obtained in the
surface kinetic region.
The rate data as a function of temperature and concentration are shown in
Figures 4-9. 4-10. 4-11. and 4-12 for n-hexane, benzene, ethyl acetate, and
MEK. respectively. n-Hexane exhibits close to first-order behavior at
temperatures up to 201°C (Figure 4-9). On the other hand, benzene exhibits
close to zero-order behavior at all temperatures tested, at concentrations
above about 4 x 10~9 mol/cm3. Ethyl acetate and MEK have reaction orders
between 0 and 1. A power law fit gave an exponent of 0.7 in both cases. We
may fit an empirical power law kinetic expression to the differential data for
n-hexane, ethyl acetate and MEK, and assume a zero-order behavior for benzene.
Although these kinetic expressions would represent single-component behavior
accurately in the intrinsic kinetic region, they would not be useful for
explaining the complex interactions observed in the catalytic oxidation of
mixtures. A more rigorous approach is to develop kinetic expressions
involving surface interactions (e.g., Langmuir/Hinshelwood, Eley/Rideal, or
Mars/van Krevelen) because they may allow explanation of the complex
interactions; e.g., via competitive adsorption. This approach has been
adopted and is detailed in section 5.0 of this report.
26
-------
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6
0.0
108 C, (mol/cm3)
Figure 4-9. Differential reaction rate for n-hexane oxidation.
27
-------
00
1 1 \ T
1.0 1.2 1.4
108 C2 (mol/cm3)
Figure 4-10 . Differential reaction rate for benzene oxidation.
28
-------
0.0
0.4 0.6 0.8
108 C3 (mol/cm3)
1.0
I
1.2
Figure 4-11 • Differential reaction rate for ethyl acetate oxidation.
29
-------
r>
O
•5*
"o
r*.
o
0.0*
4 6
109 C4 (mol/cm3)
8
Figure 4-12. Differential reaction rate for MEK oxidation.
30
-------
5.0 DEVELOPMENT OF A REACTOR MODEL FOR PREDICTING
DEEP OXIDATION BEHAVIOR OF HYDROCARBON MIXTURES
As stated earlier, the conversions of n-hexane. benzene, ethyl acetate
and MEK during isothermal oxidation were measured over a range of temperatures
(Figures 4-2 to 4-5). These results show complex interactions as evident by
the fact that single component conversions are different from those in
mixtures for most of the hydrocarbon species studied. Mechanisms of possible
interactions can be postulated in the form of mathematical kinetic models.
The development of a multicomponent kinetic model is described below followed
by its application to the specific reactor and catalyst configuration used in
our experiments.
Development of Multi-Component Kinetic Model
The mechanistic kinetic model that has received the widest support
(Balasubramanian and Vishwanath, 1975: Golodets. 1983: Jaswal et al.. 1969)
for oxidation of organic compounds is due to Mars and van Krevelen (1954) and
is:
*i
Ri + oxidized catalyst »• reduced catalyst + products (1)
s
02 + reduced catalyst *• oxidized catalyst (2)
In this mechanism, the above steps are assumed to be first order in the
respective gaseous species. Thus, if © is the fraction of oxidized catalyst
at any time, then:
r> M If . f* . A / o \
i ~ Ki Li e (3)
and
ro = ko.Co (1-0) «>
31
-------
If stoichiometry for oxidation requires Y.J moles of oxygen per mole of
hydrocarbon, then:
r = Yi r,- (5)
2
from Equation (3), (4) and (5), eliminating ©,
n = (6)
ko. Co * YikiCi
which is the Mars/van Krevelen rate expression for oxidation of hydrocarbon
species i Note that the apparent experimentally observed order of the
hydrocarbon oxidation rate with respect to the hydrocarbon species varies
between the extremes of 0 and 1. When:
k C » Y. k C.
o. o 111
then
ri =* k^ C. (first order with respect to hydrocarbon). (7)
and when
o. o 111
k C
r « (zero order with respect to hydrocarbon). (8)
Y.
The next step is to evaluate the respective k values for the components
of interest, namely, n-hexane, benzene, ethyl acetate and MEK. This may be
facilitated by inverting Equation (6):
ri ^i Cf kQ CQ
Because in our experiments Co is nearly a constant due to the large excess of
oxygen. Equation (9) should be a nearly linear relationship between inverse
reaction rate and inverse concentration if Equation (6) adequately represents
1 1
the oxidation rate data. Plots for — vs. — for n-hexane. ethyl acetate and
ri C1
MEK are shown in Figures 5-1. 5-2 and 5-3 respectively, for several
32
-------
, (cm3/mol)
Figure 5-1. Inverse reaction rate versus inverse concentration
for n-hexane oxidation.
33
-------
450
400
200
400 600
106/C, (cm3/mol)
800
Figure 5-2. Inverse reaction rate versus inverse concentration for
ethyl acetate oxidation.
34
-------
200 400 600 800 1000 1200 1400 1600
106/C4 (cm3/mol)
Figure 5-3. Inverse reaction rate versus inverse concentration
for MEK oxidation.
35
-------
temperatures. The linearity of these plots lends credence to the Mars/van
Krevelen mechanism for these species. Arrhenius relationships and plots for
the corresponding k values are shown in figures 5-4, 5-5 and 3-6.
A problem is encountered if a similar approach for benzene is attempted
because benzene exhibits nearly a zero-order behavior at concentrations
greater than 4 • 10~9 mol/cm3 (Figure 4-10). The limiting zero-order
relationship is given by Equation (8). which may be used to evaluate k for
o.
benzene as a function of temperature from rate data in the nearly horizontal
portions of the plots in Figure 4-10. k, for benzene may be evaluated using
the data at the lowest concentration from the limiting first-order Equation
(7).
Arrhenius plots for k, and kQ evaluated in this manner are shown in
i
Figure 5-7. Only the three lowest temperatures have been used to evaluate k,
because it was not possible to obtain the data at the highest temperature
under true differential reactor conditions due to experimental limitations
The behavior of MEK in the concentration range studied appears to fit equation
(7) which represents the special case of Mars/van Krevelen kinetics where
hydrocarbon oxidation controls.
The complete Mars/van Krevelen formulations for the four components for
oxidation as single components are summarized in Table 5-1. To apply these
relationships to the complex mixture effects shown in Figures 4-2 to 4-5, a
multi-component model needs to be postulated based on the Mars/van Krevelen
mechanism.
Because the reoxidation rate constants (k 's) for the various hvdro-
°i
carbons are different, it is reasonable to assume that the catalytic specie is
reduced to different oxidation states by the different hydrocarbons, n-
Hexane. benzene, and ethyl acetate as Rj, R2. and R^. respectively, are
designated. Then, for oxidation of a mixture of these hydrocarbons, the
following may be postulated:
kl
R! + oxidized catalyst *- [reduced catalyst]j + products (10)
36
-------
50 T
20 -
10
5 ••
kt = 1.66 ElOexp (-82763/RT)
Corr. Coeff. = 0.995
10
+rf
VI
O
o
0.5 -
0.2 ••
0 1
ko, = 1.75 E8 exp (-77905/RT)
Corr. Coeff. = 0.964
00
220 210
t(°C)
200 190
180 170165.
~T i i i—i—I—i—I—I—i—i—i—r—r-
200 2.04 2.08 212 216 2.20 224 228
1000/T (1/K)
Figure 5-4. Mars/van Krevelen rate constants for n-hexane oxidation,
37
-------
20
10 •
5 •
I 0.5
c
IB
C
o
CJ
01
**
03
(T
0.2 ••
0.05 -
0.02
001
k3 = 3.53 E5 exp (-45.103/RT)
Corr. Coeff. = 0.981
ko3 = 8.14 E2 exp (-40,159/RT)
Corr. Coeff. = 0.973
200
204
208
190
1 1 1 T
212 2 16
1000/T (1/K)
Figure 5-5. Mars/van Krevelen rate constants for ethyl acetate.
38
-------
I 0
O
o
o>
** -1
«o '
OC
c
-2
-3
-4
-5
k4=1.98E7exp[-51.015/RT]
R Sq = 0.986
k04 = 5.94E7 exp[-76,669/RT]
I
R Sq = 0.982
I , I ,
I
I
I
I
2.15 2.17 2.19 2.21 2.23 2.25 2.27 2.29 2.31 2.33 2.35 2.37
1000/T (1/K)
Figure 5-6. Mars/van Krevelen rate constants for MEK oxidation.
39
-------
0.1
7 .07
n
s
3 .05
*
A3
J02
.016
.01
k2 - 6.347 E13 «xp (-96351/RT)
Corr. Co«ff. - 0.998 (thrM loww points)
, - 3.41 £10 exp (-97939/RT)
Coa.Corff.- 0.995
300
200
160
100
70 -p
Oi
ci
60 1
40 •"
30
20
2.24
2^28 2JO 2J2 i34 2J6 2J8 2.40 Z42
1000/T(1/K)
Figure 5-7. Mars/van Krevelen rate constants for benzene
oxidation.
40
-------
TABLE 5-1.
Mars/van Krevelen
Rate Equation Parameters*
* C
n-hexane
benzene
Kci,
cm0
g.S
"s.
cm3
g-s
•M-)
1.75- 108exp[-77,905/RT]
3.41 .1010exp[-97,939/RT]
ethyl acetate 8.14 • 102 exp [ - 40,159/RT]
MEK 5.94 • 107 exp [ - 76,669/RT]
1.660 OO10 exp [ -82,763/RT] 95
5347 «1013 exp [ -96351 /RT] 75
3533 «104 exp [ -45,108/RT] 5.0
138 .107exp[-51,015/RT] 55
*R = 8.314 Jmor1 K
-1 IX-1
41
-------
02 + [reduced catalyst]i *• oxidized catalyst (11)
R2 + oxidized catalyst 2—^. [reduced catalyst]2 + products (12)
ko2
02 + [reduced catalyst]2 *• oxidized catalyst (13)
*3
R3 + oxidized catalyst *. [reduced catalyst^ + products (14)
(15)
S
02 + [reduced catalyst^ *. oxidized catalyst
If at any instant ©j. ©2- 03 are the fractions of the catalyst surface
occupied by [reduced catalyst]^ [reduced catalyst^, and [reduced catalyst^.
respectively, then the following balances may be written:
rl = kl ^1 n~®l ~®2 ~°3) = Yl k CQ ®1 H6)
r2 = k2 C2 (l-©i -©2 -©3) = V2 k C0 ©2 (17)
°2
r3 = k3 C3 t1'©! -°2 ~®3) = r3 k0 co ®3 (18)
3
©1. ©2. and ©3 may be evaluated from Equations (16), (17), and (18) and
substituted in the right-hand sides of these equations to yield the respective
rate equations for oxidation in a mulicomponent mixture:
(19)
ko. Co + Yl kl Cl + v 2 — k2 C2 + Y3 r- k 3C 3
°2 °3
ko, k, C C,
2202 ___ (2Q)
Co + V2 k2 C2 + rl kl Cl + V3 k3 C3
42
-------
S
k C
3 3 l
o, o2
A general rate from for deep oxidation of the ith component in an n-
component mixture may be easily inferred from Equations (19) to (21):
i = 1,2 (22)
ko. k. C C.
1101
n k
k C + Z ' Y.
>i v
kJCJ
Development of Reactor Model
In this section, we describe the application of the multicomponent
intrinsic kinetic rate expression (Equation 22) to the specific reactor and
catalyst used in our experiments. The ultimate objective is to develop a
reactor model which would predict the complex behavior observed in Figures 4-2
through 4-5.
Since a tubular reactor is used, at high conversions typical of higher
temperatures in Figures 4-2 to 4-5. the concentration of the hydrocarbon
species changes considerably along the catalytic reactor bed. Thus the
kinetic rate of reaction changes along the bed and a differential equation is
needed to describe the reactor concentration profile as a function of catalyst
bed depth (or catalyst weight). Additional complication is caused by the
possibility of diffusional limitations into the catalyst porous structure at
high temperatures. Typically this would result in a global rate of reaction
which is slower than the kinetic rate and must be accounted for in the reactor
model. Thus, at any depth along the reactor bed. a concentration profile will
exist within the catalyst particles themselves and this will give rise to a
particle differential equation for concentration within the particle as a
function of particle radius. The reactor differential equation and the
particle differential equation will be coupled through a concentration
boundary condition at the external surface of the catalyst particle.
43
-------
Two types of boundary conditions may be considered. It may be assumed
that the mass transfer across the gas film around the particle is equal to the
diffusive flux into the particle:
-Oeff(i) I 1 = kf(i)
-------
In dimensionless form Equations (35) and (36) may be written as:
doti on
= B . (i = 1.2.3...) (37)
dw n
1 +
OH (0) = 1 . (i = 1.2.3...) (38)
The fractional conversion of the ith species is defined as:
C,-
= 1 -«i (39)
Equations (37) and (38) were solved by a fourth-order Runge-Kutta method to
obtain the fractional conversion in mixtures at the bed exit for a series of
temperatures. Only the results for the binary mixtures of n-hexane and
benzene, and n-hexane and ethyl acetate are presented because similar results
were obtained for other mixtures. The model evaluations for the two binary
mixtures (n-hexane/benzene and n-hexane/ethyl acetate) are compared to the
actual experimental results (from Figures 4-2 through 4-4) in Figures 5-8 and
5-9. respectively.
Considering Figure 5-8 first, the model overpredicts somewhat the
experimentally observed conversions of benzene throughout the temperature
range and is adequate for n-hexane up to approximately 220°C. Above 220°C the
model begins to overpredict the experimentally observed conversions for both
components. The overprediction of n-hexane begins to rapidly increase at the
point where benzene is nearly completely consumed as predicted by the model.
The reason for this is that the magnitude of the denominator (adsorption term)
of Equation (22) strongly affects the rate for either component and is a
function not only of the rate constants but also of the concentrations. Since
benzene is overpredicted in the kinetic region, the magnitude of the
adsorption term for benzene is overpredicted in the kinetic region, the
magnitude of the adsorption term for benzene must be actually somewhat greater
than that given by the proposed multicomponent Mars/van Krevelen model. A
more rigorous model is thus needed which results in a larger adsorption term
for benzene. One possibility is the strong chemisorption of reaction products
45
-------
TABLE 5-2. MULT[COMPONENT OXIDATION MODEL WITH INTERNAL DIFFUSION
Reaction Equations (i = 1.2.3....)
dC,
dW
30eff(i)
vR0Pp
dCik
dR
Inlet conditions: C,- (0) = C,
1 "o
Particle Equations (i = 1.2,3...)
2 dC1k
R dR
(25)
(26)
Deff(i)
(27)
Boundary conditions:
•ik
• ci
(28)
dCik
(29)
dR 0
where rik is given by equation (22), the subscript i represents the
hydrocarbon species, and the subscript k represents conditions within the
particle.
The assumptions made in the development of Table 5-2 equations are:
(1) the tubular reactor flow hydrodynamics may be approximated by plug-
flow behavior
(2) the external mass transfer resistance is negligible
(3) the particles are spherical with center symmetry
(4) the reactor is isothermal and the change in volumetric flow rate (v)
due to reaction is negligible.
46
-------
TABLE 5-3. MULT[COMPONENT OXIDATION MODEL IN NON-DIMENSIONAL FORM
Reactor Equations (i = 1,2,3...)
da.
dw
docik
dn
Inlet conditions: cc^ (0) = 1
Particle Equations (i = 1,2,3...)
, 2 dCC-JL-
dn2 n dn
Boundary
Conditions:
aik
*- i = l
(30)
(31)
(32)
(33)
da1k
dn
(34)
47
-------
(e g. water) which would increase the magnitude of the adsorption term. This
will reduce benzene conversion and thereby correctly predict n-hexane
conversion as well at the higher temperatures. At the higher temperatures.
the increase in overprediction may also be caused due to diffusional effects.
More severe limitation of the model is seen in Figure 5-9. The model
overpredicts n-hexane conversion nearly throughout the temperature range and
underpredicts ethyl acetate beyond about 210°C. The overprediction of n-
hexane in Figure 5-9 at temperatures below 220°C indicates complex effects in
the intrinsic kinetic region, associated with the presence of an oxygenated
specie (i.e.. ethyl acetate) as opposed to a nonoxygenated specie (i.e.,
benzene in Figure 5-8).
Predictions using the proposed model for other mixtures containing ethyl
acetate are not shown herein but were similar. For example, for benzene/ethyl
acetate mixture, the model overpredicted benzene conversion and severely
underpredicted ethyl acetate conversion past the temperature of about 210°C.
For the ternary benzene/ethyl-acetate/n-hexane mixture, the model over-
predicted the conversions of benzene and n-hexane and underpredicted the
conversion of ethyl acetate past the temperature of 210°C.
The breakdown of the model in the presence of ethyl acetate suggests that
the partially oxygenated structure of ethyl acetate may play a strong role in
multicomponent oxidation kinetics. A more rigorous multicomponent kinetic
model is needed. It is impossible that reaction products such as water may
hinder the oxidation of non-oxygenated species but enhance the oxidation of
oxygenated species via alternative reactions such as hydrolysis. This may
account for the underpredict ion of ethyl acetate between 210 and 220°C and the
overprediction of n-hexane. Beyond 220°C however, the underprediction of
ethyl acetate is quite severe and suggests the advent of simultaneous gas
phase reactions with higher activation energies. This is discussed further
after the modeling results of the pore diffusion model are presented.
We next consider the case where the inhibiting effect of internal
diffusion (pore diffusion) on the rate has been taken into account by
48
-------
o
0>
o
0)
O.
InUt Compotition:
189 ppm B«nz«n«
190 ppm n-H«x*n«
WHSV-209
100
140
180
Temperature (°C)
260
300
Figure 5-8. Companion of predicted and actual conversions for
benzene and n-hexane in a binary mixture.
49
-------
Inlet Conditions:
184 ppm ethyl acetate
190 ppm n-hexane
180
220 260 300
Temperature (°C)
340
380
ethyl acetate exp.
n-hexane exp.
model predictions
Figure 5-9. Comparison of predicted and actual conversions for ethyl acetate
and n-hexane in a binary mixture.
50
-------
considering the reactor equation and particle equation separately (Tables 5-2
and 5-3).
The particle equations in Table 5-3 were solved by first converting the
second order particle differential equation into two first order differential
equations. Equations (32) through (34) were rewritten as:
foik
dn
dm.,-
dn
. (i = 1.2.3...)
2m
n
(40)
1 + y, k^i
(41)
With boundary conditions.
°Hk I = «i •
(i = 1.2.3...)
(42)
and
= 0
(i = 1,2,3...)
(43)
The problem is now to solve Equations (31). (32) and (40) through (43)
simultaneously using a numerical technique. The primary difficulty in using a
straight forward Runge-Kutta method is that the concentration at the center of
the particle is unknown, thus an initial condition required to advance the
Runge-Kutta iteration is not available.
Our approach around this difficulty involved a trial and error solution.
Before presenting the method of solution, however, it is necessary to estimate
the effective diffusivity, Qeff(i) for each hydrocarbon species. This
parameter is a function of both the hydrocarbon type and the complex porous
structure of the catalyst. The catalyst used (G-43, United Catalyst. Inc.)
has a bimodal pore distribution as shown in Table 5-4.
51
-------
TABLE 5-4. CATALYST PHYSICAL PROPERTIES FOR MODELING AND Deff(l) CALCULATION
Pore Size Range
(rim)
Pore Volume Contained
(cm3/g)
1.4 - 2.9
2.9 - 14
14 - 80
80 -150
0.03
0.24
0.00
0.08
Total Pore Volume = 0.35 cmVg
Particle Density. P = 1.46 g/cm3
Surface Area. Sg = 212
Micropore Void Fraction, em = 0.394
Macropore Void Fraction. cu = 0.117
Average Micropore Radius, u = 2.57 nm
Average Macropore Radius, u = 57.5 nm
The random pore model proposed by Wakao and Smith. (1962) was used to
estimate Oeff(i) for each hydrocarbon species. This model is given by
- 2 Eu2 (i + 3 eM) _
Deff(i) = °Mi CM2 + OUT (44)
1 - £M
1 1 1
where = + (45)
1 1 1
- + (46)
OUT DI-
air
52
-------
D-j_a-jr for each hydrocarbon species was estimated from the Chapman-Enskog
formula:
.0018583 T3/2
Oi-air
i (MW)air
ai-air ni-air
(47)
where tri_air = - T, +
-------
The fit of the above equation with actual published values is shown in Figure
5-10.
and (D|()U1. the respective Knudsen diffusion coefficients in the
macro- and micro-pore regions were calculated from
/ T \
I - 1
\ (MW), /
DK = 9.7 x 103 r I - 1 (52)
where r is the average pore radius (Table 5-4). Thus knowing DMi and Oui from
Equations (47) and (48) and CM and eu (from Table 5-4), Deff(i) was calculated
for each hydrocarbon species as a function of temperature.
Once Deff was known, all parameters in Table 5-3 model equations are
available. The method of solution involved first dividing the catalyst bed
into a number of grids.
At the first grid (i.e., at the entrance to the bed) the particle
equations were then solved. The solution was carried out by a combination of
the sixth order Runge-Kutta-Verner method (IMSL Routine DVERK) and a global
constrained function minimizer (IMSL subroutine ZXMWD) on an 1MB 370 mainframe
computer. A guess value is assigned at the particle center for each
hydrocarbon species concentration. The Runge-Kutta is solved and the
concentration at the external surface is compared to bed inlet concentration.
The difference is minimized (by minimizing the sum of the square of the
dimension less concentration differences of the various species) until it is
close to 0. The radial gradient at the surface is then evaluated and
substituted in Equation (30) to advance the bed dimension less concentration to
the next grid. The above operations are repeated until the concentration
profile in the bed is completely determined. The accuracy of the solution can
be increased by increasing the number of grids. We have found 20 grids to be
a reasonable number in our solution.
Comparisons of model predictions to experimental data are presented in
Figures 5-11 to 5-14. The predictions in Figures 5-11 through 5-13 are shown
with and without pore diffusion incorporated into the model. The negligible
difference in the predictions demonstrate the absence of pore diffusion
54
-------
en
ui
_
1.10
1.09-
1.08-
1.07-
1.06-
1.05-
1.04-
1.OS-
S' 1.02-
^ 1.01-
1 0.99-
O
0.98-
0.97-
0.96-
0.95-
0.94-
0.93-
0.92
0.26
= 0.996
i-air
= -.713log
10
-i-air
1.2874
0.3
1 1 1 1 1 1 1 1 1—
0.34 0.38 0.42 0.46 0.5
'°9lOei-air
Figure 5-10. Evaluation of collision integral for modeling purposes in the temperature
range of interest.
-------
inhibitions for the 120-170 mesh particles used in our experiments. The pore
diffusion model, however, is quite useful for larger particles (e.g. 6 mm)
typical of industrial reactors. From Figure 5-11, both models over-predict
simple component n-hexane data above about 260°C indicating that external
diffusion effects may become significant above 260°C. A second possibility is
the retardation of the kinetics by reaction products such as water which would
be more apparent at higher conversions as obtained above 260°C. From Figure
5-12, the effect of external diffusion is not so apparent for benzene. The
pore diffusion model is somewhat superior at higher conversions than the
simpler model without internal diffusion. From Figure 5-13. the benzene data
in the mixture is overpredicted throughout the temperature range, confirming
the need for a larger adsorption term for benzene in the kinetic equation (as
concluded previously using the simpler model). Also, as stated previously.
the presence of strong water chemisorption will lead to a bigger adsorption
term and may allow a superior fit. Experiments at varying inlet humidities
are needed to verify this. Once the benzene is correctly predicted, n-hexane
would be better predicted as well because the overprediction of n-hexane
conversion in the mixture beyond about 210°C is believed to be due to a
smaller adsorption term resulting from benzene overprediction.
Finally, Figure 5-14 shows the comparison of single component ethyl
acetate conversion predicted by the simple model to experimental data. The
fit in the kinetic region (>220°C) is excellent, indicating the suitability of
Mars/van Krevelen kinetics in this region. At >220°C, the conversions
increase much more rapidly with temperature indicating a significantly higher
activation energy than given by the Mars/van Krevelen kinetics. The increased
activation energy may be due to the onset of catalytically supported thermal
reactions and differential data needs to be collected in this higher
temperature region to develop a better model to represent ethyl acetate
oxidation.
Summarizing the modeling work to date, a simple multicomponent kinetic
model based on the single component Mars/van Krevelen mechanism was found to
be adequate for predicting oxidation of non-oxygenated hydrocarbon such as
benzene and n-hexane. At the trace level concentrations used, however, water
formed from reaction may inhibit the kinetics. The pore diffusion model
56
-------
INLET COMPOSITION:
410 ppmv n-hexane in air
WHSV • 209
• experimental
— model without pore diffusion
• model with pore diffusion
100
140
180 220 260
Temperature. °C
Fidure 5-11. Comparison of model predictions and experimental data for n-hexane
conversion as a single component.
57
-------
100
90
80
70
o 60
.o
§ 50
5 40
I
I
s-
Ol
Q.
30
20
10
• INLET COMPOSITION:
375 ppm benzene in air
. WHSV - 209
O experimental
.—model without
pore diffusion
, • model with pore
diffusion
100 140 180 220
Temperature, °C
260
Figure 5-12. Comparison of model predictions and experimental data
for benzene conversion as a single component.
58
-------
o
"e
I
o
O
OJ
(_>
OJ
o.
100
90
80
INLET COMPOSITION:
- 189 ppm benzene )jn air
190 ppm n-hexane j
WHSV « 209
• benzene experimental
o n-hexane
experimental
— model without
pore diffusion
model with pore
diffusion
100
180 220
Temperature. °C
260
300
Figure 5-13 . Comparison of model predictions and experimental data
for binary benzene/n-hexane mixture.
59
-------
50
o
u
40-
'§ 30
0>
OJ
u
O)
Q.
20-
10-
0
Inlet Composition: 450 ppmv ethyl acetate
WHSV = 209
— model without pore diffusion
D ethyl acetate experimental
180
220
T
260
Temperature (°C)
300
T
340
380
Figure 5-14. Comparison of model predictions and experimental data for ethyl acetate
conversion as a single component.
-------
simulations indicated the absence of pore diffusion hinderances in the 120 to
170 mesh particles, but the model is believed to be necessary for layer
particles typical of industrial reactors. The model may also be extended to
account for external diffusion effects. The Mars/van Krevelen kinetics
reasonably simulate ethyl acetate oxidation at temperatures less than 220°C.
Beyond 220°C, alternative reaction pathways such as thermal reactions (via
free radical intermediates) may be present as seen from the much higher
activation energy.
61
-------
6.0 AUTOCATALYSIS IN DEEP OXIDATION OF ETHYL ACETATE
When the apparent order of a reaction shifts with increasing conversion.
this suggests that the products of reaction are somehow speeding up the rate.
Systems in which this occurs are often referred to as "autocatalytic". This
may be represented in its simplest form for a second order reaction:
A + B -» B + B where TA = kCACg.
This implies, of course, that some of reactant B must be present for the
reaction to proceed.
In principle the equations describing an autocatalytic reaction are
simple; however, as noted earlier, these equations require an initial amount
of compound B in order to be solved. The system under consideration here is a
great deal more complex, in that a network of reactions contributes to the
autocatalytic effect observed. Consider the homogeneous hydrolysis of ethyl
acetate:
H20 -» C2H3OOH + C2H5OH
Here, water must be present for the reaction to proceed to acetic acid and
ethanol. This water may be present initially or generated during the
combustion process:
C2H300C2H5 + 502 -» 4C02 + 4H20
Essentially, once water (W) is initially formed in the reaction process.
a mixture of ethyl acetate (EA). acetic acid (A) and ethanol (E) will be
produced. It has been noted (Morrison and Boyd, 1974) that unlike
hydrocarbons and alcohols with their uniform carbon-carbon linkages, esters
with their oxygen linkages are comparatively unreactive compounds. Since this
oxygen linkage is in fact quite stable towards oxidizing agents, attacks will
preferentially occur at the electron deficient carbonyl carbon, resulting in
the replacement of the 0-C2Hs group by an electron rich moiety such as the
62
-------
catalyst surface or a hydroxyl group. Combination with surface adsorbed
oxygen is much less likely than hydrolysis at lower temperatures, and indeed,
the presence of an acid catalyst makes the carbonyl even more susceptible to
nucleophilic attack. The ethyl alcohol formed is substantially more reactive
towards oxidation than the parent compound. Similarly, the presence of more
reactive hydrocarbons such as benzene would also start and promote the
hydrolysis/oxidation cycle. Thus, the reaction pathway for ethyl acetate
oxidation would employ both homogeneous and heterogeneous steps. A general
reaction scheme including carbon dioxide (C) and oxygen (0) may be written as
follows:
"1
EA + W —» A + E
EA + 0 -» W
A + 0 —» W + C
k4
E + 0 —» W + C
If the relative value of the rate constant kj is greater than k2
hydrolysis would be the original mechanism for decomposition and the
sequential autocatalytic mechanism holds. The methodology for confirming this
mechanism is fairly straight forward - merely humidify the inlet stream. The
integral reactor results for a dry and humidified ethyl acetate stream are
shown in Figure 6-1. A very clear enhancement is seen for all except the
highest and lowest temperatures. For temperatures between 200°C and 260°C,
the total conversion is more than doubled. These results would indicate that
a hydrolysis mechanism is the most logical explanation for the "enhancement"
phenomena observed earlier in the mixture tests. For example, with a reactive
compound such as benzene:
"5
C6H6 + 7.502 -» 6C02 + 3H20
63
-------
100
10
160
Comparison of Ethyl Acetate Conversions
in Dry and Humidified Air Streams
dry
0.01 g water/g air
WHSV = 209
150 ppm ethyl acetate
200
220
240 260
Temperature (C)
280
300
320
340
Figure 6-1. Comparison of ethyl acetate conversions in dry and humidified air streams.
-------
three moles of water are produced per mole of benzene. Since we know from
single component experiments that benzene oxidizes at substantially lower
temperatures than ethyl acetate, water would be available for the hydrolysis
reaction to occur at lower temperatures than with ethyl acetate alone. This
mixed homogeneous/heterogeneous mechanism would also be consistent with the
observation of the enhancement over different catalyst formulations used by
Tichenor and Palazzolo (1987).
In the absence of detailed information on the water concentrations during
the mixed oxidation reactions, one can still propose a kinetic expression for
the hydrolysis/oxidation of ethyl acetate based upon several assumptions.
First, that the hydrolysis step is the rate limiting step in ethyl acetate
destruction. Second, that the contribution of water from the most reactive
compound is most responsible for low-temperature enhancement. Finally, that
the benzene rate expression is insensitive to mixture effects. Here, the rate
of water production may be directly related to the benzene destruction:
k k P P
°CKHfi C6H6 C6H6 °2
0 o
3 "6n6 k0 P + v kc H Pc H
°C6H6 °2 C6H6 C6H6 C6H6
At steady state, the production of water from benzene oxidation is
constant, and may be used in the expression for the homogeneous rate of
hydrolysis of ethyl acetate:
rEA = k CEA Cw
Although this kinetic approach has merit, several other experiments
should be used to confirm this mechanism.
First, since the hydrolysis reaction is first order in water, the
rate of reaction should be proportional to the water concentration in
the inlet gas (for a differential reactor) up to the equilibrium
limit.
Second, the hydrolysis rate could be measured independently in a plug
flow reactor without catalyst present. This rate could be compared
to the rate of complete oxidation of ethyl acetate over a catalyst
with and without water present.
Thirdly, for mixtures the rate of ethyl acetate oxidation could be
examined as a function of the water concentrations produced during
the deep oxidation of the hydrocarbon mixtures. For example, in
65
-------
mixtures of benzene and ethyl acetate one would predict the rate of
ethyl acetate destruction to be directly proportional to the rate of
benzene combustion. Similarly, the n-hexane/ethyl acetate mixtures
the rate should be proportional to the rate of n-hexane combustion.
Since the reactivity of n-hexane is lower than that of benzene, the
"enhancement" should be less, and that is indeed what is observed for
those binary mixtures. If sufficient information on the water
concentrations produced in those tests exists, the mechanism could be
confirmed quantitatively.
Finally, the phenomenon of "auto-catalysis" may explain the
inflection point in the first order Arrhenius plots noted for ethyl
acetate much earlier in this study. Froment and Bishoff (1979)
allude to such inflection points occurring with increasing
conversion, so it may be worthwhile analyzing the results using this
approach. Autocatalysis usually only becomes noticeable at
conversions greater than ten percent. Once this mechanism becomes
dominant, a distinct increase in activation energy should be noted
for the reaction.
Water vapor seems to have only a minimal effect on the other oxygenated
hydrocarbon, methyl ethyl ketone (MEK). As seen in Figure 6-2, some
deviations between the humidified and dry stream reaction rates occur at
temperatures below 240°C. At higher temperatures where autocatalysis should
cause the largest deviation, the results match very closely. The suppressed
rate below 180°C may be due to competitive adsorption of water on the
catalyst, but clearly a different mechanism is in operation. This difference
is not surprising, because MEK does not contain an ester linkage that may
hydrolyze. Obviously, the mere presence of oxygen within a reactant is not
nearly as important as the structure of the compound in determining the
overall reaction rate for oxidation.
66
-------
100
dry
0.01 g water/g air
180
200
220
240 260 280
Temperature (C)
300
320
340
360
Figure 6-2. Comparison of MEK conversions in dry and humidified air streams.
-------
7.0 CONCLUSIONS AND RECOMMENDATIONS
• Conversions of components in a mixture of organics may be significantly
higher or lower than when present by themselves.
• Mars/van Krevelen (MVK) kinetic mechanism is an adequate representation
for the deep oxidation of single organic compounds over the catalyst used in
this study.
• A reactor model incorporating pore diffusion effects and MVK kinetics
adequately explained single component benzene data over the entire temperature
range of interest (150°C to 360°C).
• A multicomponent reactor model incorporating pore diffusion effects and
a simple proposed extension of MVK kinetics was marginally successful in pre-
dicting benzene - n-hexane mixture behavior.
• Conversions of oxygenated species such as ethyl acetate are higher in
mixtures than in single components. The MVK model appears to explain the data
in the kinetic regime. MEK oxidation shows lower apparent enhancement than
ethyl acetate. Other reaction pathways, including thermally enhanced free
radical mechanisms and the interactions of oxygen containing species with
partially reduced metal surfaces, may explain the observed phenomena. Further
experimentation is necessary before a specific model can be postulated for the
oxidation of oxygenated compounds.
• The presence of water vapor in the gas stream increases the conversion
of ethyl acetate significantly. This suggests that autocatalysis by product
water may be responsible for the observed enhancement described above. The
fact that MEK conversions are not significantly affected by the presence of
water vapor lends credence to this theory.
68
-------
8.0 DATA QUALITY
The experiments were conducted in accordance with an approved draft of a
quality assurance project plan.
8.1 Precision
All experiments involved sequential injections of reactor inlet and
outlet to the gas chromatograph. The average of eight to ten injections was
taken as the representative conversion for the conditions of each experiment.
Since the inlet concentration was measured each time before measuring the
outlet concentration, this gave eight to ten independent measurements of
conversion for each experiment. This procedure also corrected for any drift
in detector response over the course of the experiment. The calculated
conversion is based on measured ratios of inlet to outlet peak areas. The
relative percent deviation (RPD) may be calculated as
RPO = 100(ratiomax - ratiomin)/ratioave
where ratiomax is the maximum value of inlet to outlet peak areas. ratiomin is
the minimum value of inlet to outlet peak areas, and ratioave is the average
value of inlet to outlet peak areas for a given experiment.
Table 8-1 shows RPDs calculated for representative experiments with the
four compounds studied. All values lie well within the quality assurance
objective of 10 percent.
Gas flow rates were calibrated against a soap bubble flow meter at the
beginning of each experiment and were not found to vary to any measurable
exent during the course of the experiment.
Catalyst temperature was found to be reproducible to within 1 percent for
a given controller setting.
8.2 Accuracy
All reported gas concentrations were referenced to standards certified by
Scott Specialty Gases. Inc. against Standard Reference Materials (SRMs), which
69
-------
TABLE 8-1. PRECISION OF DATA
Max. Min. Ave.
Compound Concentration Temperature Ratio Ratio Ratio RPD (%)
Benzene 375 ppm 180°C 1.9165 1.893 1.9033 1.23
MEK 300 ppm 200°C 1.6988 1.654 1.6783 2.67
n-Hexane 410 ppm 280°C 2.9953 2.8458 2.9269 5.17
Ethyl Acetate 450 ppm 308°C 1.9014 1.8909 1.8947 0.55
70
-------
are primary standards. Flame ionization detectors have responses that are
linear in concentration, especially in the range of this study. This was
verified in early experiments. Table 8-2 shows the bias in the concentrations
reported by the vendor for uncertified tanks of MEK. In this case, the "true"
value is that obtained by referencing against a certified tank of gas assuming
linear FID response.
Gas flow rates were measured against a soap bubble flow meter, a primary
standard, for each experiment.
Temperature measurements were referenced to an ice bath at 0°C.
8.3 Completeness
Since this was a laboratory study, all invalidated experiments could be
repeated and all necessary experiments were performed.
8.4 External QA Audit
None was performed.
71
-------
TABLE 8-2. ACCURACY OF DATA
Nominal Value (ppm)
39
71.7
135
188.8
300.3
True Value (ppm)(a)
50.4
70.0
141.8
189.4
297.4
Bias (*)
22.61
2.43
4.80
0.32
0.98
Referenced against certified standard assuming linear FID response
72
-------
9.0 NOMENCLATURE
Ci =
cik =
CiQ. =
Co •
°eff(i)=
Di-air =
(Dk)Mi =
(Dk)ui =
°Mi =
Dui •
E
f • =
hi -
KB =
kf(i) =
KH20 =
kf =
Ki =
hydrocarbon [VOC] concentration
intraparticle concentration of species i
inlet hydrocarbon [VOC] concentration
oxygen concentration
effective diffusivity of species i
binary molecular diffusivity
macropore Knudsen diffusivity
micropore Knudsen diffusivity
effective macropore diffusivity
effective micropore diffusivity
activation energy
fractional conversion of species i
/ kipP
Thiele moduli i for species i R0J —
* o
°eff(i)
Boltzmann constant
mass transfer coefficient of species i
adsorption equilibrium constant for H?0
surface reduction rate constant for species i
ki cio
k C0
L °iJ L °J
Units
[mol/cm3]
[mol/cm3]
[mol/cm3]
[mol/cm3]
[cm2/s]
[cm2/s]
[cm2/s]
[cm2/s]
[cm2/s]
[cm2/s]
[J/mol.K)]
(-)
[-]
[cm/s]
[ — 1
[cm3/(g.s)]
I— ]
surface reoxidation rate constant for species i
(daik/dn)
[cm3/(g.s)]
73
-------
(MW)i
P
R
r =
ri =
rik =
t
T
v
w
W
WF
WHSV
molecular weight of species i
pressure
particle radial coordinate
average pore radius
rate of oxidation of species i
hydrocarbon species, i
intraparticle rate of reaction
average macropore radius
particle radius
average micropore radius
rate of oxygen consumption
catalyst surface area
temperature
absolute temperature
volumetric flow rate at reaction temperature
dimension less catalyst weight in reactor
catalyst weight coordinate
total weight of catalyst in reactor
[atm]
[cm]
[cm]
[mol/(g.s)]
[mol/(g.s)]
[cm]
[cm]
[cm]
[mol/(g-s)]
[cn)2/g]
[K]
[cm3/s]
[g]
[g]
[g]
weight hourly space velocity
[g gas/(g catalyst.h)J
GREEK NOMECLATURE
a-i = dimensionless bulk concentration. Cj/Cio
a-jk = dimensionless intraparticle concentration,
8 = kn- WF/v
c = Lennard-Jones parameter
EM = macropore void fraction
[-]
74
-------
eu = micropore void fraction [—]
r, = stoichiometric coefficient for deep oxidation [—]
of species i
n = dimension less particle radial coordinate, R/RO [—]
xi s <3 °eff(i) WF)/(vR02 Dp) [__]
Op = particle density [g/cm3]
-------
10.0 REFERENCES
Acres, G.J.K. 1970. Platinum Catalysts for the Control of Air Pollution
Plat. Met. Rev. 14:2-10.
Balasubramanian, S. and D. S. Viswanath. 1975. Ind. Eng. Chem. Fundam..
14:3, p. 158.
Box. J. M. 1966. Comp. J. pp.67-77.
Cullis. C.F., O.E. Keene. and D.L. Trimm. 1970. Studies of the Partial
Oxidation of Methane over Heterogeneous Catalysts. Journal of Catalysis
19:378-385.
Emmett, P.M. (ed.). 1960. Catalysis. Reinhold, New York.
Froment. G.F.. and K.B. Bischoff, 1979. Chemical Reactor Analysis and Oesian
John Wiley and Sons. '
Golodets. G. I. 1983. Heterogeneous Catalytic Reaction Involving Molecular
Oxygen. Elsevier. New York. Part III. p. 396.
Hawthorn. R.D. 1974. Afterburner Catalysts - Effects of Heat and Mass
Transfer Between Gas and Catalyst Surface. AIChE Symp. Sen/.. Recent
Adv. Air Pol. Control. 70:137, p. 428.
Heyes. C.J.. J.G. Irwin. H.A. Johnson, and R.L. Moss. 1982. The Catalytic
Oxidation of Organic Air Pollutants. Part 2. Cobalt Molybdate and Copper
Chromite Catalysts. J. Chem. Tech. Biotechnol. 32:1034-1041.
Hirschfelder. J. 0.. C. F. Curtiss. and R. B. Bird. "Molecular Theory of Gases
and Liquids." pp.539, 578, John Wiley and Sons, Inc. NY.
Jaswal, I. S., R. F. Mann, J. A. Junsola. and J. Downie, 1969. Can. J. Chem
Eng.. 47. p. 284.
Mars. P.. and D. W. van Krevelen. 1954. Special Supplement to: Chem. Eng
Sc i., 3, p 41.
Moro-Oka. Y.. Y. Morihama. and A. Ozaki. 1967. J. Catal. 7:2.
Morrison, R.T., and R.N. Boyd, 1974. Organic Chemistry (3rd ed.). Allyn and
Bacon. Inc.
Palazzolo. M. A., J. I. Steinmetz. D. L. Lewis, and J. F. Beltz. 1985
Parametric Evaluation of VOC/HAP Destruction Via Catalytic Incineration
Report No. EPA-600/2-85-041 (NTIS No. PB85-191187/A7). U.S. Environmental
Protection Agency.
Pope. D.. D.S. Walker, and R.L. Moss. 1976. Evaluation of Cobalt Oxide
Catalysts for the Oxidation of Low Concentrations of Organic Compounds in
Air. Atmospheric Environment 10:951-956.
76
-------
Prasad. R., L.A. Kennedy, and E. Ruckenstein. 1984. Catalytic Combustion
Catal. Rev. Sci. Eng. 26(l):l-58.
Sadamori. H.. A. Chikazawa. S. Okamura. and T. Noda, 1981. Development of
Low Temperature Catalytic Burner. Gas Research Conference. Gas Research
Institute, p. 1467.
Satterfield. C.N. 1980. Heterogeneous Catalysis in Practice. McGraw-Hill
New York.
Smith, J. M. 1981. Chem. Eng. Kinetics. 3rd ed., p.455. McGraw-Hill. NY
Spivey. J.J. 1987. Complete Catalytic Oxidation of Volatile Organics
I & EC Res., (in press).
Tichenor. B. A. and M. A. Palazzolo. 1987. Destruction of Volatile Organic
Compounds via Catalytic Incineration. Environmental Progress, 6(3), 172-
176.
Wakao. N. and J. M. Smith. 1962. Chem. Eng. Sci., 17, p. 825.
Yao, Y.F.Y. 1973. The Oxidation of Hydrocarbons and CO Over Metal Oxides:
II. «-Cr203. Journal of Catalysis 28:139-149.
77
------- |