Reaction Kinetics of Ammonia & Nitric Acid

.
o , f'
Â't^^.,
.."'- J-
KINETICS OF LOW TEMPERATURE OXIDATION

OF AMMONIA TRACE QUANTITIES
by
PETER CRAIG BENTSEN, B.S. in Ch.E., M.S. in Ch.E
A DISSERTATION
IN
CHEMICAL ENGINEERING
Submitted to the Graduate Faculty
of Texas Tech University in
Partial Fulfillment of
the Requirement for
the Degree of
DOCTOR OF PHILOSOPHY
Approyad
August, 1973
f\
fiit-5057
r3
Cop. a
ACKNOWLEDGEMENTS
The author wishes to express his appreciation to

the members of his Committee for their advice and
constructive criticism throughout both the project
and the writing of this dissertation. Special
acknowledgement is given to Dr. R. R. Graham for his
timely advice, guidance, and encouragement.
Also, the author acknowledges the National
Aeronautics and Space Administration for financial
support of the project through contract NAS 1-9506.
11
TABLE OF CONTENTS
ACKNOWLEDGEMENTS ii
LIST OF TABLES V
LIST OF FIGURES vi
CHAPTER
I. INTRODUCTION AND PRELIMINARY INVESTIGATIONS 1
II. LITERATURE SURVEY 10
III. THEORY 26
IV. EXPERIMENTAL SYSTEMS 39
The Backmix Reactor System 39
Backmix Reactor Experimental Procedure 47
Fixed-Bed Reactor System 48
Fixed-Bed Reactor Experimental Procedure 51
Chemical Analysis Methods 53
Ammonia Analysis 53
Chromatographic Ammonia Analysis 54
Colorimetric Ammonia Analysis and 55
Calibration
Nitrogen Dioxide Analysis 61
Nitrous Oxide Calibration and Analysis 62
V. EXPERIMENTAL DATA AND NUMERICAL PROCEDURES 65
VI. RESULTS OF EXPERIMENTAL WORK 81
Long-Time Data 81
Short-Time Data 86
Backmix Reactor 98

111
AM^
IV
VII. CONCLUSIONS 101

LIST OF REFERENCES 102
NOMENCLATURE 104
APPENDIXES 106
A. Platinum Catalyst Data 107
B. Calculational Procedure for Effectiveness 108
Factor
LIST OF TABLES
1. Results of Thermodynamic Calculations 5

2. Catalysts Included in Screening Tests 6
3. The Effect of Hydrogen on Activation Energy 14
4. Effect of External Mass Transfer 30
5. Gas Chromatograph Parameters 54
6. Backmix Reactor Data at 350**F 67
7. Kinetic Data for Fixed-Bed Reactor at 300**F 70
10. Kinetic Data for Fixed-Bed Reactor at 400F 73
11. Platinum Catalyst Regeneration Data - 75
Regeneration I
Regeneration II
Regeneration III
Regeneration V
Regeneration IV
16. Rate Constants for the Long-Tirae Model 84
LIST OF FIGURES
1. Catalytic Activity Comparison 8
2. Experimental Apparatus - Backmix Reactor 40
3. Backmix Reactor Detail 43
4. Backmix Reactor Catalyst Baskets 46
5. Fixed-Bed Reactor System 50
6. Fixed-Bed Reactor Detail 52
7. Calibration Apparatus 59
8. Nessler Method Ammonia Calibration Curve 60
9. Saltzman Calibration Curve 64
10. Test of Reaction Mechanism at 350**F 69
11. Arrhenius Plot for Constants K~ and K. 85

3 4
12. Fixed-Bed Reactor Performance at 300F 87
16. Calculated vs. Experimental at 300F 91

19. Regeneration I 95
20. Regeneration II 96
21. Regeneration V 99
22. Model Comparison 100
VI
CHAPTER I
INTRODUCTION AND PRELIMINARY INVESTIGATIONS
For many years, scientists and engineers have

been looking for a profitable way to combine hydrogen
and nitrogen to produce ammonia (NH^). However,
ammonia under certain circumstances can be considered
a pollutant and requires some method of elimination.
One method of elimination would be to oxidize the
ammonia catalytically to harmless products such as
nitrogen and water.
The catalytic oxidation of ammonia is particularly
adaptable to closed environment systems such as space
capsules. The source of ammonia in these systems is
human metabolism. The conversion of trace amounts
of ammonia to harmless products is especially
important in this application, since the products
of the oxidation would be returned to the capsule.
This requires that the products not create an
additional or perhaps an even greater hazard than
that caused by the ammonia.
Much current emphasis has been placed on elimi-
nation of the oxides of nitrogen caused by automobile
emissions. Current technology is able to convert
easily these oxides to ammonia (1). If a scheme were
1
available so the ammonia produced could be converted
to harmless products such as nitrogen and water, it
would be of considerable importance to the automobile
industry.
Both of these applications require that the

products of the elimination method be safe and pose
no additional hazard or pollution. They also require
a method that would be relatively low in cost and
power consumption. The use of catalytic oxidation
would be acceptable if a catalyst were available
that would oxidize the ammonia at low temperatures.
Low temperatures favor the production of nitrogen
and water as well as minimize operating problems.
Lower temperatures would permit lower capital
requirements since less heat would be required, less
severe operating conditions and fewer operating
problems encountered.
The oxidation of ammonia has been used for many
years to produce nitric acid. The reaction is
carried out over a platinum wire gauze at temperatures
between 1650 and 1850F. At these temperatures, the
major product is nitric oxide which is then easily
converted to nitrogen dioxide. The nitrogen dioxide
is then absorbed in water. Any nitrogen dioxide not
absorbed in the water then has to be eliminated.
In nitric acid plants this elimination is very
complicated and quite expensive. Unfortunately,
the oxidation of ammonia at these conditions produces
nitrogen dioxide which is as harmful as the ammonia.
The literature reveals considerable information
concerning the production of nitric acid (2).
Catalytic oxidation of ammonia for non-nitric acid
production is quite limited, in fact no information
was located concerning the oxidation of ammonia
in the parts-per-million (ppm) range. The units
for ppm will be moles per million liters throughout
this dissertation. The ppm range is becoming
considerably more important when one considers
the increased emphasis on air-pollution control by
organizations such as the Environmental Protection
Agency.
The oxidation of ammonia could proceed by any
one or any combination of four overall reactions.
Each reaction leads to either an oxide of nitrogen
and water or nitrogen and water. The overall
reactions possible are
4NH2 + 5O2 ;::;=^ 4N0 + 6H2O (1-1)
4NH2 + 7O2 ^:^=^ 4NO2 + 6H2O (1-2)

2NH3 + 202 ^^^^^ ^ 2 ^ "^ ^"2 ^-^""^^
4NH3 + 3O2 ^:;=r 2N2 + 6H2O. (1-4)
Prior to any actual data generation, the heat of

reaction and the equilibrium constant were calculated
for each overall reaction. These calculations
show that all of the reactions are highly exothermic
and irreversible in the temperature range of investi-
gation. These data as well as the equilibrium
constants are given in Table 1.
Based on the values of the equilibrium constants
calculated, the reactions are all thermodynamically
feasible at the temperatures proposed for the study
and all reactions are essentially irreversible.
After consideration of the thermodynamic data,
it was necessary to pick a catalyst that would
oxidize ammonia in the temperature range of interest
at a respectable rate. The catalyst should be such
that the products of the reaction were in accordance
with our goals (Reaction 1-4).
Ammonia oxidation data were obtained during NASA
Project NASA CR-2132 (3) using thirteen different
catalyst materials. Shown in Table 2 are the
catalysts which were included in this preliminary
TABLE 1
Results of Thermodynamic Calculations
kcal/mole
Temp. ** R Heat of Reaction K
Reaction I
760 -215.5 > 0.723 X 10^^*

780 -215.5 > 0.723 X 10^^
810 -215.4 > 0.723 X 10^^
860 -215.3 > 0.723 X 10^^
Reaction II
760 -270.5 > 0.723 X 10^^

780 -270.5 > 0.723 X 10^^
810 -270.5 > 0.723 X 10^^
860 -270.5 > 0.723 X 10^^
Reaction III
760 -131.6 0.635 X 10^^

780 -131.6 0.114 X 10^^
810 -131.6 0 . 3 9 3 X 10^"^
860 -131.5 0.754 X 10^^
Reaction IV
760 -302.0 > 0.723 X 10^^

780 -302.0 > 0.723 X 10^^
810 -301.9 > 0.723 X 10^^
860 -301.8 > 0.723 X 10^^
*0.723 X 10'^^ is the largest number available on the

IBM 360-50 used to perform these calculations.
TABLE 2
Catalysts Included in Screening Tests
Surface Area
Catalyst Material Company Designation
2% manganese dioxide Harshaw Mn-0501 258
supported on silica
0.3% palladium Harshaw Pd-0501 186
supported on high
activity alumina
0.5% platinum Englehard 0.5% Pt 93
supported on alumina supported on alumina
0.5% ruthenium Englehard 0.5% Pt 74
supported on aliomina supported on alumina
4% nickel oxide, 4% Harshaw Ni-1601 78
cobalt oxide, 4% iron
oxide supported on
activated alumina
14% nickel oxide Harshaw Ni-0707 140
supported on high
activity alumina
10% cuprous oxide Harshaw Cu-08 03 137
supported on high
activity alimiina
19% manganese dioxide Harshaw Mn-0201 69
supported on activated
alumina
20% ferric oxide Harshaw Fe-0301 41
supported on activated
alumina
3.5% silver supported Harshaw Ag-0101
on inert alumina
5% cobalt oxide, 5% Harshaw Co-0901 59
copper oxide supported
on activated alumina
Not reported General American Not reported
Transportation
"Gatalyst"
Mixed oxides of Mine Safety Appliance 156
manganese and copper "Hopcalite"
screen matrix. All of the materials tested showed
some activity for the removal of ammonia; however,
platinum, rutheniiim, and Hopcalite seemed most
promising.
Both a fixed-bed reactor and a backmix reactor

were used in the preliminary catalyst screening
tests. In order to make a relatively easily
interpretable comparison of the low-temperature
ammonia oxidation activity of these catalysts, a
first-order reaction rate approximation was used and
the resulting rate constant was plotted versus a
reciprocal absolute temperature scale.
Although this procedure may not fully compensate
for concentration effects, the procedure does permit
the ranking of catalysts in terms of their activity.
Shown in Figure 1 are the results of the preliminary
screen tests. These tests were made using air with
less than 500 ppm water. A high first-order rate
constant at a low temperature is, of course, the
most desirable situation. As previously mentioned,
ruthenium, Hopcalite, and platinum appear to be the
most active catalysts. The two cobalt-containing
catalysts (Co 0901 and Ni 1601) were both relatively
active; however, they both tend to produce a relatively
8
4J
4J
c
d
p
(0
o
u
<u
0)
o
p
CO
u
H
+J
M
Id
04
<
1 I I
700 600 500 400 300 350
Temperature (** F)
Figure 1 - Catalytic Activity Comparison
large fraction of highly undesirable nitrogen dioxide
(NO2) product. "Gatalyst", copper oxide, and nickel
oxide all were considerably less active than the
better materials. The iron oxide, silver and
manganese dioxide were not effective showing only
a small amount of reaction at the highest temperatures
used.
While platinum, ruthenium, and Hopcalite all
performed well under dry conditions, only the platinum
was active under humid conditions (4). The activity
of ruthenium was reported by Gully et al^. (4) to be
"poor" in comparison to the activity of the platinum
catalyst. Hopcalite had also performed poorly under
high humidity operation (3). Based on these results,
platinum was chosen for this study.
The purpose of this study is actually threefold.
The first is to provide the experimental data for
the oxidation of ammonia over a platinum catalyst
in the ppm range which is so obviously missing from
the literature. The second is to use these data to
obtain a quantitative mathematical description or
model of the rate of ammonia oxidation with emphasis
on the utilization of this model for design purposes.
The third is to study catalyst stability and be able
to quantify the deactivation.
CHAPTER II
LITERATURE SURVEY
For many years, the oxidation of ammonia to

nitric oxide (NO) has been of considerable importance
for the manufacturer of nitric acid (2). Consequently,
a considerable amount of information exists concerning
the reaction
4 NH3 + 5 O2 ^^ > 4 NO + 6 H2O. (2-1)
At temperatures above 1470*'F, ammonia can be oxidized

almost exclusively to nitric oxide using a platinum
gauze catalyst.
Griffiths (5) and co-workers oxidized ammonia in
a high-temperature infrared cell at a pressure of
10 torr using a platinum catalyst which was supported
on silica. This arrangement allowed them constantly
to monitor the reaction in the infrared cell. The
ammonia-to-air molar ratio was 1:12, which the authors
assert was the optinum chemical condition for nitrous
oxide (N2O) production. The temperature range
investigated was from 77 to 400F in intervals of 45*'F.
At each temperature, the reaction mixture was analyzed
with the infrared instrioment. The infrared scans
obtained allowed the authors to determine what changes
had taken place. At temperatures below 350F, the
10
11
differences between spectra obtained for ammonia only

and those obtained for mixtures of ammonia and oxygen
were undetectable. At both 35 0F and 375*F, a doublet
was observed which was attributed to the production
of nitrous oxide. Also upon cooling the reaction
mixture to room temperature, a broad band was
observed which was clearly identified as water pro-
duced by the ammonia oxidation. Furthermore, the
authors were able to detect bonds between the catalyst
and reactant species. Among those complexes reported
were SiOH NH^ and Pt NH^. Uniquely, the
catalyst was formed by producing the silica-platinum
catalyst using a conventional infrared solid scimple
press. This allowed very thin discs to be formed
that were approximately one inch in diameter which
would allow the infrared beam to be passed through
the pellet since silica is transparent to infrared
for the wavelengths being used. By being able to
pass the beam through the pellet, the authors were
able to see the bonds between the chemical species
and the catalyst surface. The appearance of nitric
oxide or nitrous oxide was not reported. The
appearance of nitrogen was not reported and it is
doubtful that it could have been detected with their
apparatus.
12
Attempts have been made to oxidize ammonia over

non-noble metal catalysts with limited success.
Novak et al. (6) were able to obtain their highest
reaction rates using ferric oxide-based catalyst
with 5 per cent bismuth oxide. They were able to
increase catalyst activities with the addition of
magnesium oxide and chromic oxide. Other attempts
to alter catalyst activities involved impregnating
the catalyst supports with molar solutions of nitrates.
These attempts resulted in lower catalyst activities;
however, if the impregnating solutions were concentrated
the catalyst activities were as high as the crushed
catalyst. Maximum conversions were realized for
feeds containing 11 per cent ammonia. Feed rates
were reported to be 400 1/hr which corresponds to
2
54 kg ammonia/m hr. The maximum conversion was
obtained at 1410-1635F for the ferric oxide-based
catalyst.
Further ammonia oxidation work was conducted
by Zasorin (7) and involved using a complex catalyst
arrangement. Oxidation was accomplished by first
preheating the ammonia-air mixture to 48 0-520F and
passing it over seven platinum gauze grids. Next
the reaction mixture was passed through a column
13
packed with 9 3 per cent granular ferric oxide and

7 per cent granular chromic oxide. The experiments
were conducted at 5 and 10 atmospheres. Catalyst
2 7
contact times were 2.45 x 10 and 2.79 x 10

seconds. The temperature of the oxide packed beds
was 1625*'F and ammonia conversions were 97 per cent
at 5 atmospheres and 94.5 per cent at 10 atmospheres.
Amano and Taylor (4) investigated the decom-
position of ammonia in atmospheres of nitrogen,
hydrogen, and nitrogen-hydrogen mixtures over the
alumina-supported noble metal catalysts: ruthenium,
rhodium, and palladium. The decomposition reaction is
2 NH^ i N2 + 3H2. (2-2)
They found that the nitrogen content had no effect
on the decomposition but that the hydrogen concen-
tration more pronouncedly inhibited the reaction
over palladium than over ruthenium. Temperature
ranges were selected for the catalyst studies such
that conversions would not exceed 10 per cent. From
the data collected, it is evident that ruthenium is
significantly more active than either rhodium or
palladium since ammonia decomposition was reported
for ruthenium between 645 and 750F, while reaction
temperatures for palladium and rhodium were 950 to
14
1060**F and 750 to 840F, respectively. Activation

energies were reported from 25 to 27 kcal/mole for
non-hydrogen systems on both rhodium and palladium
(no data reported for non-hydrogen ruthenium systems).
With excess hydrogen present in the reaction mixtures,
activation energies increased to 30-32 kcal/mole.
An apparent activation energy increase with an
increase in hydrogen concentration is shown by Amano
and Taylor for palladium catalyst. These data are
shown in Table 3.
TABLE 3
The Effect of Hydrogen on Activation Energy
Inlet Hydrogen kcal/mole

Mole Fraction âct^ kcal/mole
0.0 27.82
0.0834 31.10
0.125 32.15
Amano and Taylor reported a kinetic expression

for the decomposition of ammonia over a ruthenium
catalyst. The rate expression is
d[NH^] f. f- _o q
^ = k[NH3]-^ [H2] ^ (2-3)
Zawadzki (2) points out several observations
made by him and other authors concerning the oxidation
15
of ammonia over platinum at temperatures above 930F.

Among these are that the reaction of ammonia to
nitrogen and hydrogen is almost as fast as the
oxidation of ammonia to nitric oxide. It is further
pointed out that to maximize the yield to nitric
oxide, temperature is the critical variable. At
temperatures below 930F, Zawadzki maintains that
nitrous oxide is present in large proportions in
the reaction products. The author proposes a set of
reactions leading to nitric oxide, nitrous oxide, and
nitrogen, which are reported to agree with thermo-
dynamic calculations. The primary reaction is
NH3 + O -> NH + H2O. (2-4)
Those reactions that lead to nitric oxide are:
NH + O2 ^ HNO2 (2-5)
HNO2 ^ NO + OH (2-6)
20H -^^ H2 + O2. (2-7)
Those reactions leading to nitrous oxide are:
NH + O -> HNO (2-8)
2HN0 ->- N2O + H2O. (2-9)
And those leading to nitrogen are:
NH + NH -> N2 + H2 (2-10)
16
NH + NH3 -> N2H^ "^ ^2 + 2H2 (2-11)
NH + HNO -^ N2 + H2O (2-12)
NH3 + HNO2 -> N2 + 2H2O (2-13)
2N0 ->- N2 + O2 (2-14)
2N2O ^ 2N2 + O2 (2-15)
as well as other possibilities. Certain of these

reactions, specifically, 2-14 and 2-15, have not been
observed at low temperatures (i.e., below 930F).
Giordano et a_l. (9) reported that ammonia could
be oxidized at temperatures between 570 and 840''F
with the products being chiefly nitrogen and nitrous
oxide. Their objective was to alter the reaction
rate and selectivity by doping nickel oxide catalyst.
Previously, they had speculated that the selectivity
for nitrous oxide of the oxidation reaction could be
increased by increasing the p-typeness of the catalyst.
P-typeness is a characteristic of semi-conductors and
is associated with a deficiency of electrons. In
order to vary the p-typeness of the nickel oxide
catalyst, it was doped with various metal oxides.
Among those selected to increase the p-typeness and,
consequently, produce chiefly nitrous oxide, were
the monovalent oxides of lithium, sodium, and potassium,
17
Those that were used to decrease the p-typeness were

the trivalent oxides of chromium, iron, and aluminum.
Results from doping the catalyst with the trivalent
oxides indicate that significantly higher nitrogen
yields could be realized. This is especially true
for ferric and aluminum oxides.
When the nickel oxide catalyst was doped with
lithium oxide, the per cent conversion and the
selectivity to nitrous oxide were greatly altered.
At 825**F, the per cent conversion of ammonia and the
selectivity to nitrous oxide were both essentially
100 per cent. The author summarizes the results
of the study saying that both sodium and lithium ions,
which increase the p-typeness, also increase the
activity and selectivity for nitrous oxide of the
nickel oxide catalyst. Conversely, those ions of
chromium, iron, and aluminum, which decrease the
p-typeness, also decrease the selectivity to nitrous
oxide. The decrease in selectivity to nitrous oxide
results in an increase in selectivity for nitrogen.
Unfortunately, any doping procedure with the trivalent
oxides to increase the selectivity to nitrogen results
in a significant decrease in conversion. Catalysts
doped with approximately 4 per cent monovalent
18
lithium oxide were able to obtain essentially 100

per cent conversion and selectivity to nitrous oxide
at 8 25**F. Conversely, when nickel oxide catalyst was
doped with 3 per cent trivalent aluminum oxide and
operated at identical conditions, the selectivity
to nitrous oxide was reduced to less than 20 per cent
or to 80 per cent selectivity to nitrogen, but the
conversion was reduced to less than 5 per cent.
Maximum conversion realized for trivalent oxide
doped catalyst was around 40 per cent at 8 25**F for
the chromic oxide doped catalyst but selectivity
to nitrogen was close to 30 per cent.
The authors propose a tentative mechanism for
the reaction
(NH3)ads + ' " (NHjO)^^^ + e (2-16)
(NHsO^ds + 0' " <)ads + H2O + e (2-17)
'^3>ads -^ <>ads " "^2 + 2H2O (2-18)
(HNO) ads + ('ads " ^^2 "^ ''2- '^-19)

This reaction scheme leads one to believe that when
Step 2-18 prevails over Step 2-19, the major product
is nitrogen. On the other hand, when the HNO concen-
tration is higher than the NH-^0 concentration. Step
2-19 prevails and nitrous oxide is the major product.
19
The authors attempt to explain that when the catalyst

is doped with the monovalent oxides, an increase in
the extralattice oxygen concentration is achieved.
Conversely, doping with the monovalent oxides
decreases the extralattice concentration of oxygen.
They conclude that Reaction 2-17 is favored by high
oxygen concentrations attainable by monovalent doping.
This then leads to an increased probability of
Reaction 2-19 occurring. The inverse of this argument
leads to Reaction 2-18 being favored and can even be
extended to lower observed catalyst activities,
since the concentration of HNO is reduced thus
causing both Reactions 2-18 and 2-19 to be slower.
Schriber and Parravano (10) investigated the
oxidation of ammonia over a supported ruthenium
catalyst between 475 and 655F. Reaction pressures
were from 176 to 280 cm Hg with feed compositions at
5.3, 9.7, and 16.5 per cent (vol.). The diluent
in every case was oxygen.
At the conditions investigated the products of
the reactions were nitrogen, nitrous oxide, and water.
Of particular importance was the fact that they con-
cluded that the rates of formation were dependent
on the concentration of ammonia, oxygen, and water.
20
The rates were increased with an increase in the

concentration of both ammonia and oxygen. Water
concentration was found to be detrimental to the
formation of both nitrogen and nitrous oxide. The
authors proposed empirical relations for the production
of both nitrogen and nitrous oxide. These expressions
are
\ = % ) N 2 ^NH3 \ ' ' ' ' V ' * ' ' ^'^''^

R
^2^ = ^
(K0^N20
) P
^NH3 P ^'^^
^-^^ ^02 PH2O ~^'^^ - (2-21)
^^ ^^)
They also concluded that the oxidation of ammonia

was accomplished by two competing reactions on the
surface of the catalyst, the first reaction leading
to nitrogen and the second leading to nitrous oxide.
These overall reactions are
2NH3 + 2O2 -^ N2O + 3H2O (2-22)
4NH3 + 3O2 -^ 2N2 + 6H2O. (2-23)
The overall reactions result from a sequence of

elementary steps which involve not only surface
reaction but adsorption and desorption. The authors
assume that the first steps of the reaction involve
forming the nitroxyl and imide radicals on the
catalyst surface. The reaction scheme is as follows.
21
NHO(a) + H20(g)
NH3(g)
02(g)
NH(a) + H20(g) . (2-24)
The (a) shown behind the HNO and NH indicates adsorbed

species. They further assume that nitrous oxide is
formed from
NH3(g) + NHO(a) ^ N20(g) + 4H(a) (2-25)
where the H(a) is desorbed as water. The formation of

nitrogen results from the reaction
2NH ^ N2 + H2. (2-26)
Rate expressions for nitrous oxide and nitrogen

formation derived from these reaction schemes and
expressed in terms of the partial pressure of
ammonia, oxygen, and water yield exponents of the
following magnitude:
NH3 ^2 H2O
N20 1.5 0.5 -0.5
N2 1 0.5 -1.0
22
These exponents are very close to the empirical

Equations 2-20 and 2-21, previously presented with
the exception of the water exponent for the nitrogen
reaction. Attempts by Schriber and Parravano to
locate an alternate route proved fruitless.
They also conclude that neither mass transfer
nor pore diffusion were significant resistances in
the overall oxidation of ammonia for their particular
reaction conditions. The activation energies for
nitrogen and nitrous oxide were reported to be 12.6
and 35.0 kcal/mole, respectively.
Catalytic oxidation of ammonia was investigated
at Texas Tech University under the National Aeronautics
and Space Administration, Contract No. NAS 1-9506.
Work conducted by Johnson (11) involved the low-
temperature oxidation of ammonia over a supported
ruthenium catalyst. These studies were conducted in
a backmix reactor for inlet ammonia concentration
ranging from 50 to 412 ppm. The reactor temperature
was varied from 242 to 442*'F. From the data collected,
Johnson proposed an empirical rate expression of the form
r = kc^ (2-27)
where r = reaction rate of the particular specie

23
c. = concentration of that specie

k,a = empirically determined constants.
The constant k can usually be correlated as a function
of temperature. The traditional correlation method
is to use the Arrhenius expression
k = A exp C^/RT) (2-28)
where A = the pre-exponential constant
E = the activation energy for the specie
R = the universal gas constant
T = the absolute temperature.
Substitution of the Arrhenius expression into the
rate expression leads to
r = A exp C^/RT) c?. (2-29)
Johnson used a least-squares technique to determine

the empirical constants A, E, and a. By using the
least-squares technique, he was able to correlate
all data simultaneously. The empirical constants
were determined for both the rate of ammonia oxidation
and the rate of nitrous oxide formation. The data
analysis procedure yielded rate expressions of the form
r^^ = (2.051 X 10"^) exp ("^Q^g^^)C^^ ^'^^ (2-30)
r^ Q = (4.926 X lO"^) exp {lM^l)c^^ ^-^2(2-31)

24
From comparisons of the empirical rate Equations 2-30
and 2-31 to the experimental data, average deviations
of 15.7 and 15.9 per cent were calculated for the
ammonia expression and the nitrous oxide expression,
respectively.
Johnson also attempted to detect the effect,

if any, of water and oxygen concentration on the
oxidation rates of ammonia or on the product formation
rates. This effect was not the primary objective
of his study; consequently, the detection effort was
based on very little data. In fact, data were taken
at only one temperature and one water concentration.
Tentative conclusions were that oxygen and water
concentration had little or no effect on the oxidation
rate of ammonia at 400**F.
Additional work conducted at Texas Tech involved
attempts to oxidize ammonia catalytically in a vapor
stream containing predominately steam. This work
was conducted by Graham (4). Experimentally, this
involved vaporizing the ammonia containing water and
passing the steam-ammonia mixture over the catalyst
contained in a tubular reactor. Reaction temperatures
ranged from 295** to 400**F. When ruthenium was used
as the catalyst, no ammonia oxidation was detected.
25
When platinum was used as the catalyst, oxidation of

the ammonia was observed. At 295*F, approximately
97 per cent of the ammonia was oxidized to nitrogen
and water with no significant formation of nitrate.
At temperatures above 295F, essentially all of the
ammonia passed through the reactor was oxidized.
Reactor feed was 5.8 ml/min of a 14 ppm solution of
ammonia in distilled water and an oxygen flow of
500 standard cc/min. The reactor contained a 25 gm
charge of 0.5 per cent platinum. The activity of
ruthenium under similar conditions was at least
two orders of magnitude poorer.
CHAPTER III
THEORY
The phenomena occurring during catalytic reactions

are usually described by one of two methods. One
either (a) choses to disregard the individual steps
and propose an overall expression such as the fractional-
order rate expressions, or (b) one attempts to use a
more fundamental approach such as proposed by Langmuir
and Hinshelwood. The Langmuir-Hinshelwood expressions
attempt to describe the reaction in terms of the
elementary processes that occur at or near the surface
of the catalyst.
Many authors (12, 13) propose a series of seven
major steps that are involved in the catalytic
conversion of reactants to products. This sequence
of steps as given by Smith (12) is as follows:
1. transport of reactants from the bulk fluid to
the fluid-solid interface (external surface of
the catalyst)
2. intraparticle transport of reactants into the
catalyst particle (if it is porous)
3. adsorption of reactants on the catalyst surface
4. chemical reaction of the adsorbed reactants to
the adsorbed products (surface reaction)
26
27
5. desorption of the adsorbed products
6. transport of the products from the interior
sites to the outer surface of the catalyst
particle
7. transport of products from the fluid-solid
interface into the bulk fluid stream.
During catalytic conversion of a reactant, all
of these steps occur either in series or parallel.
The Langmuir-Hinshelwood approach is to postulate
that Step 3, 4, or 5 is the rate-controlling step
and then assume that the others are at equilibrium.
For this study. Steps 1 and 7 are governed by
the mass-transfer rates for the reactants and products.
Since during this study oxygen was in great abundance,
any adsorption steps involving oxygen can be assumed
to be essentially in equilibrium and constitute no
rate limitations.
Techniques are available for determining the
effects of external mass transfer in fixed-beds (14).
If one calculates the difference between the concen-
tration of the reactant in the bulk gas phase and
the concentration of the reactant at the catalyst
surface, the effect of mass transfer on the reaction
becomes apparent. If the concentration difference
28
is negligible, then external mass transfer can be
eliminated from further consideration. This concen-
tration difference can be calculated as follows:
(^b - ^s> = ? r (3-1)

g
where P, = the partial pressure of the reactant in
the bulk phase
P = the partial pressure of the reactant on
the catalyst surface
r, = rate of diffusional process, mole/hr
2
K = mass-transfer coefficient, mole/hr/atm/cm
a = mass-transfer area, cm .
The correlation presented by Satterfield (14)
for the calculation of the mass-transfer coefficient
in packed beds is
^9 = ^ 0 ^ 2/3 (3-2)
Sc
where Gj^ = superficial molar velocity of bulk gas
2
mixture, moles/sec/cm
N^ = Schmidt number, y/pD
Sc
P = total pressure in atmospheres
e - porosity of bed
3D = 0.357/(Nj^^-359e)
D = diffusivity of reactant in boundary layer
29
N = Reynolds number d G/y

G = superficial mass velocity
y = viscosity of bulk fluid
d = diameter of catalyst particle.
By placing the expression for j in the above equation,
the following expression for the mass-transfer
coefficient is obtained:
0.357 ^M
\ = s N, 0.359 2/3. (3-3)
^ Re Sc
This correlation has been found to be valid over the
range 3 "^ N_Re '^ 2000. From Equation 3-3, a value for
the mass-transfer coefficient is calculated. Then if

the reaction is at steady state, the rate of mass
transfer from the bulk phase to the catalyst surface
is equal to the rate of reaction. Knowledge of the
reaction rate permits calculation of the quantity
(P, - P ) and gives an indication of the relative
^b s' ^
importance of external mass transfer. As can be
seen from the sample calculation shown in Table 4,
external mass-transfer resistance can be assumed
negligible for this study. Since external mass
transfer is negligible. Steps 1 and 7 can be ruled
out as rate-limiting steps.
30
TABLE 4
Effect of External Mass Transfer
Physical Properties:
Temperature = 350F
Pressure = atmospheric
e = 0.3
d = 0.3175 cm
P
2
D = 0.44 cm /sec
G^ = 1.9 7 g mol/hr/cm^
y = 0.024 cp
p = 0.79 X lO"^ g/cm*^
2
a = 10.66 cm /g
^NH = 25 X 10 g mol/hr/g catalyst
Calculated Quantities:
^Re = 21
N^
Sc = 0.437
JD = 0.40
2
K = 1.53 3 g mol/hr/cm /atm
(P,
b - Ps ) = 1.53 X 10"^ atm
P - P
^2- s ^ 0.005
^b
31
Steps 2 and 6 involve intraparticle diffusion.
Physical examination of the catalyst particles reveals
a thin black layer of material on the outer surface
of the catalyst particle. Initial observation leads
one to suspect that this dark ring contains essentially
all of the platinum. Correspondence with the
manufacturer verified that the platinum was deposited
in the thin outer layer (15). Additional reinforcement
is provided from an effectiveness factor of 0.95
calculated for the 0.5 per cent platinum catalyst.
The effectiveness factor is defined as
ef f
where ^^ ^^ is the reaction rate in a porous catalyst
when there is no internal diffusion resistance and
r is the observed rate. The calculation and the
procedure are given in Appendix B.
The high value of the effectiveness factor is
expected since all catalyst was reported to be
essentially on the catalyst surface. In addition,
experimental runs were conducted to see if alumina
could catalyze the reaction. Within the temperature
range of this study, no oxidation of auranonia was
observed when alumina was used as the catalyst. Since
32
alumina cannot catalyze the reaction and all the
platinum is essentially on the catalyst surface,
internal diffusion is judged not to be the rate-
limiting step.
As is frequently done, a kinetic expression of
the Langmuir-type was chosen on a trial-and-error
basis as a possible description of the oxidation
reaction. This Langmuir form was found to fit the
data quite well. Verification of this fit is given
in Chapter VI. However, if the mechanistic steps
given by Reactions 3-5 through 3-10 are assumed the
Langmuir-type equation can be derived.
^6
Un + Pt ^==: (NH-.Pt) (3-5)
^6
(NH3*Pt) + ^02 V "^ (NH'Pt) + H2O (3-6)
HÔ + Pt ^ ^7' ^ (H^Ô-Pt) (3-7)

^7
(NH-Pt) + hO^ , ^ (HNO-Pt) (3-8)

k'
^3
(NH-Pt) + (HNO-Pt) ^5
^ N 2 + (H20-Pt) + Pt (3-9)
k.
(HNO-Pt) + (HNO-Pt) ^ N2O + (H20-Pt) + Pt (3-10)
33
Making the plausible assumption that the surface
decomposition of the adsorbed ammonia is the rate-
controlling step, then mechanistic Step 3-6 can be
used to express the reaction rate as
^ = "^1 St.NH3 \''- (3-1^)
This equation may be rewritten using Reaction 3-5 to

obtain Cp^^^^^j yielding
^ = ^6 ^1 ^NH3 ^^ ''0^''/ê' ^^-^2)
During this study, the catalyst activity was observed

to decay. However, at any point of constant catalyst
activity, there are Pt^ total active sites per gram
of catalyst on which reaction can occur. Those sites
available for ammonia adsorption are those active
sites not occupied by water, ammonia, and transition
complexes. The concentration of a transition complex
is usually very small and is not considered to occupy
a significant number of sites (16). The number of
sites occupied by ammonia and water can be obtained
from Reactions 3-5 and 3-7. From this the expression
for the total nxomber of unoccupied sites at steady
state is
^* = " T - kf S H , Pt - ^ C Pt. (3-13)

6 3 1 1
34
Experimental runs demonstrated that water concen-
trations up to 6000 ppm had no effect on the rate of
ammonia oxidation. This implies that k-/kl is a very
small quantity and the term containing it may be
dropped from Equation 3-13. Equation 3-13 is then
solved for Pt and rearranged to yield
This expression is then substituted into Equation 3-12

for the number of platinum sites Pt yielding
^6 (1 + H ""nn/H^
Since the air used during this study contained
approximately 21 per cent oxygen, the difference
between the inlet and outlet oxygen concentration
caused by the ammonia oxidation can be considered
negligible. Consequently, the term C^ can be
"2
considered a constant. Combination of the constants
yields a rate expression of the form
^ ^ ^2 S H 3
During this study the activity of the platinum catalyst

was observed to decay. This phenomenon has been
observed by many others (12, 13, 14, 16). If this
35
decay was due to some specie reacting or combining
with an active site so that the total active sites,
Ptrnf decreased with time, a mechanism could be proposed.
Assuming that some specie, X, combines reversibly
with an active site to render it inactive, then such
a reaction could be described as
Pt + X :;=^ Pt-X. (3-17)

k'
^8
This reaction is assumed to be independent of those
proposed for ammonia oxidation. The total platinum
sites available at time zero is Pt , so at any time
those sites covered with specie X can be expressed as
Pt-X = Pt^ - Pt^. (3-18)
However, those sites expressed as Pt are the sum of

the unoccupied sites (Pt) and those sites (Pt.)
occupied by reactants, products, and transition
complexes. The unoccupied sites or those available
to react with specie X can be expressed as
Pt = Pt^ - Pt^. (3-19)
The rate of disappearance of platinum can be

obtained from Reaction 3-17 as
- ^g|i = kg C^ Pt - k^ P f X . (3-20)
36
Substitution for Pt and Pt-X with the previously

developed relationships yields
d Pt
"^ = ko C^ (Pt^ - Pt.) - kl (Pt^ - Pt). (3-21)
dt ^"8 "'X ' "-T ''î' "^8 ^ o T
Rearrangement of Equation 3-21 yields
All experimental runs to obtain data to quantify

the deactivation were conducted at essentially
identical conditions. Since feed conditions were
constant, the concentration of X is a constant if
it was carried in the feed. The quantity Pt- can
also be considered a constant since
1. the concentration of transition complexes is low
2. the number of sites occupied by water is low
as shown from the small value of k_/kl
3. the rate of change of the ammonia-covered
sites is assumed negligible.
Combination of all constant quantities and separation
of the variables yields

d Pt
- dt = . p. _ ,- . (3-23)
^9 ^^T ^10
After integration of this expression from t=0 to t=t
and from Pt to Pt^, the following expression is obtained

-1 ^9 ^^T " ^10
^ = ir ^^ ^1c Pi- - k ^ (3-24)
^9 ^9 ^ o ^10
37
Equation 3-24 can then be solved for the number of
total active platinum sites, Pt ,
Pt^ = Pt^ - (Pt^ - kô/^9^ (^ " e"^9^). (3-25)

This expression is then substituted into Equation 3-16
for Ptrp. For simplicity all constant combinations
derived to this point and those obtained by the
substitution will be represented by upper case K's.
(Kê-^2^ ^ K3) C^
"^ - ^. (3-26)
4 NH3
Notice that Equation 3-26 reduces to a time-independent

form when the product of K^t is equal to or greater
than 4. The time-independent form of the rate
expression is v r
^3 NH
^ = 1 4. y r ' (3-27)
^ ^ ^4 S H 3
Application of these rate expressions to fixed-bed
reactors requires that the design equation
W
F I .. (3-28)
be evaluated where W is the catalyst weight in grams,

F is the molar flow rate of reactants in moles/hour
and r is the rate of reaction which can be expressed
either by Equation 3-26 or 3-27. From this point, all
C's refer to ammonia concentrations. Substitution
38
of Equation 3-26 and integration from C to C- (inlet
and outlet concentrations) yields
W -1
F in ^ + K4 (C^ - CQ) K3 + Kê"^2^ (3-29)
As before when the quantity K2t approaches 4, the

equation can be reduced to a time-independent form
W (3-30)
F = Kj 1^ C^ -^ ^7 (Î - ^0^
Equation 3-30 can also be obtained by substituting

Equation 3-27 into Equation 3-28 and performing the
integration.
Evaluation of the constants in Equations 3-29
and 3-30 and verification of their applicability is
described in detail in Chapter V.
CHAPTER IV
EXPERIMENTAL SYSTEMS
In order to carry out the experimental portions

of the ammonia oxidation studies, two reaction
systems were utilized. These systems, one a backmix
(or stirred-tank) reactor and the other a fixed-bed
reactor, are separately discussed below. The two
different systems were selected in order to take
advantage of the complementing features of having
both differential and integral type reaction rate
data available for analysis.
The Backmix Reactor System

A schematic diagram of the backmix reactor
system is shown in Figure 2. The air used for the
oxidation studies was originally supplied in
commercial steel cylinders. The fixed-bed reactor
had always operated on air supplied to the building
by the university air compressor, and approximately
midway through the experimental program the backmix
reactor was converted to use the same source. To
insure that the air was dry and oil free, it was
passed through a series of beds. These beds were
operated at 20 psig maintained by a pressure regulator
39
40
0)
o
H
u
> }-4
<u
01 0)
[o}- (0 i
rd 5
s o
CO
:3
<d
a
/ ^ ^ a
0) (U
M >H o
0) (U r H 4->
u
rd (U
I^ <d o
-P (d
Clj u iS^-l O TS C 0)
H X
c CM c J-l -H
fd
(d en rd
to ^^ <u g
(0 E-i X u
O w Ky u <d w fd
CQ
(U
^ - 1 I
CN
(U
u
H
U4
<d Cu
6
u
o
4-)
rd
>1 rH
41
between the source and the purification system. The

first was a bed of silica gel contained in 18 inches
of 1-inch stainless-steel tubing. Following the
silica gel bed was a bed consisting of 25 per cent
activated charcoal and 7 5 per cent magnesium
perchlorate. This bed was constructed of 2-inch
steel pipe 1 foot long. Chromatographic analysis
of the air showed no compounds other than oxygen,
nitrogen and carbon dioxide. From this point, the
air flowed through a Hastings Model LF 3K mass flow-
meter transducer. The transducer was connected to
a meter which provided a visual display of the air
flow rate to the reactor. After passing through
the Hastings meter the air passed through a 1/4-inch
Whitey needle valve, mixed with a small stream of
ammonia and was fed into the reactor. Air flow out
of the reactor occurred through leaks around the
reactor end plates and around the agitator shaft.
Gaseous ammonia was fed to the reactor from a

cylinder containing liquid ammonia. Ammonia vaporized
from the cylinder passed through a pressure regulator
and into a trap constructed of 1-inch stainless steel
tubing containing potassium hydroxide to insure removal
of all water from the ammonia. It then flowed into a
42
1/4-inch stainless steel line which contained a

1/4-inch Whitey needle valve and was connected to a
high-resistance tube. The resistance tube consisted
of a 3/8-inch outside diameter (O.D.) glass tube
packed with crushed glass to provide high resistamce
to flow.
A detailed schematic of the backmix reactor is
shown in Figure 3. The reactor consisted of a 4-1/2-
inch diameter cylindrical shell, 7 inches in length
with circular plates at each end. A shaft, which
held the catalyst baskets and agitator blades,
extended through the centerline of the reactor.
Thermov/ells, sample ports, and feed ports were
located in both end plates. The reactor housing,
end plates, agitator shaft, and agitator blades
were all constructed of stainless steel. The outside
of the reactor housing was wrapped with electrical
resistance heaters which were connected to a variable
voltage transformer. The outside of the reactor
and both end plates were insulated to minimize
heat losses. The entire reactor including insulation
was covered with a poly-(methyl methacrylate) box
slightly longer and wider than the reactor that provided
a dead air space around the reactor. This box helped
43
H
fd
4-)
0)
Q
u
o
4J
U
<d
O
fd
CQ
ro
H
44
considerably in eliminating any horizontal temperature

variations in the reactor.
The temperature within the reactor was monitored
with a pair of bare iron-constantan thermocouples
which passed through the front end plate and followed
the reactor walls to locations just below the catalyst
baskets. They were connected to a Honeywell
Electronik 15 recorder. With this arrangement, the
temperature could be monitored and controlled anywhere
between ambient temperature and 900F. The agitator
shaft was driven by a variable-speed Minarik Model SL14
electric motor which was mounted on a wooden support
which aligns the motor shaft and the agitator shaft.
The manner in which the catalyst pellets were
stacked in the catalyst baskets was of concern. The
catalyst pellets were cylindrical in shape with a
length and diameter of 1/8 inch. The first baskets
which were made had the catalyst stacked one behind
the other approximately eight layers deep in the
direction of rotation. It was suspected for this
arrangement that the rear pellets were shielded from
the gas by the front pellets, thereby reducing the
effectiveness of the catalyst charge. In order to
have adequate exposure of all the catalyst pellets to
45
the gas, it was necessary to have the catalyst spread

out so that resistance to flow through the catalyst
region was minimized. This was accomplished using the
baskets pictured in Figure 4. The baskets were made
from a stainless-steel wire mesh.
Another area of concern was the degree of mixing
within the reactor. In order for a backmix-type
reactor to provide meaningful data, it is essential
that there be no concentration gradients within the
reactor. In order to check this, runs were made
in which samples were taken from both ends of the
reactor. Good agreement was obtained, at least
within the accuracy of the analysis, indicating that
the mixing was good. As another check on mixing,
the temperature at both ends of the reactor was
monitored. Here again good agreement was obtained.
The fact that the reactor itself was not
catalyzing the reaction was established. With no
catalyst in the reactor, samples were taken from
the inlet feed line sample port and reactor end plate
sample ports. Good agreement was obtained at
temperatures as high as 450**F. At temperatures of
600**F and higher, however, some background oxidation
of ammonia to nitrogen dioxide was noticed.
46
Figure 4 - Backmix Reactor Catalyst Baskets

47
Backmix Reactor Experimental Procedure

Prior to making each run, the reactor temperature
was adjusted to the desired level by manually adjusting
the setting of the variable voltage transformer. The
air flow to the reactor was set to the desired rate
by adjusting the needle valve downstream of the mass
flowmeter. The ammonia flow rate was set by adjusting
the needle valve between the ammonia cylinder and
the resistance tube and observing the pressure on
the upstream side of the resistance tube. The purpose
of this observation was to get a qualitative idea of
what the ammonia concentration at the inlet to the
reactor should be. For calculational purposes, the
inlet ammonia concentration was deteimiined by colori-
metric analysis. The agitator was set to the desired
speed by adjusting the Minarik motor controller.
After the temperature and flow rates had been
at their steady state values for approximately one
hour, sampling was begun. The reactor contents were
sampled first. The desired number of samples (usually
three each) for ammonia, nitrous oxide, and nitrogen
dioxide analysis were taken via syringe at the sample
port in the reactor end plate. It was important that
the rate of sample withdrawal did not exceed the feed
48
flow rate into the reactor. This was important

because a rapid sample withdrawal rate could result
in room air leaking into the reactor, thereby
diluting its contents. After sampling of the reactor
contents was completed, the air feed stream to the
reactor was sampled for ammonia via a septum in the
air feed line downstream of the air-ammonia mixing
point.
Fixed-Bed Reactor System

A fixed-bed reactor was constructed to provide
a system for obtaining integral data for the ammonia
oxidation system. The equipment consisted of an
isothermal oven, a feed-preparation system, and the
reactor proper.
The oven was constructed of 18 gauge 316 stainless
steel and insulated with one inch of glass fiber
insulating material. Oven temperature was manually
controlled by adjustment of the applied voltage to
a 500 watt electrical resistance. The oven temperature
was monitored using three thermocouples placed inside
of the oven. A long-shafted 1/6-horsepower electric
motor was used to provide air circulation inside of
the oven and thus insured a uniform temperature distri-
bution.
49
A schematic flow diagram of the fixed-bed

apparatus is shown in Figure 5. The feed-preparation
system consists of an ammonia cylinder connected
through a high-resistance glass tube to the air-feed
line. This tube provided sufficient resistance to
flow to allow very small amounts of ammonia to be
continuously mixed with the air stream. The air
stream was pre-treated as described in the backmix
reactor system section. The fixed-bed system allowed
the air to be humidified if desired. If humidification
was necessary, the air was passed through two fritted
glass bubblers in parallel. These bubblers were
immersed in a temperature-controlled bath to control
the saturation temperature of the air. If humidifi-
cation was not desired, this bath could be bypassed.
The ammonia-and-air mixture was then passed by a
sample port and into the oven. The oven contained a
10-foot by 1/4-inch stainless-steel preheater coil
which brought the feed up to the reaction temperature.
The stream then passed upward through the reaction
section and out of the oven. An exit sample port
was located on the gas stream following the reactor.
The stream then passed through a wet-test meter and
was vented.
50
B
0)
4->
(0
>1
w
o
-p
u
fd
<u
0^
(U
CQ
I
0)
X
I
in
0)
H
51
The reactor detail is shown in Figure 6. The
reactor was constructed of a 1-inch O.D. type 347
stainless-steel tube approximately 6 inches long.
The inside diameter of the tube was 0.76 5 inches.
All lines, fittings, and valves used in both
reactor systems were constructed of 316 stainless
steel except those specifically mentioned.
Fixed-Bed Reactor Experimental Procedure

Prior to a run, the fixed-bed air flow rate
was manually adjusted to the desired level using a
1/4-inch IVhitey metering valve located immediately
before the humidifiers. The oven temperature was
set to the desired value using the Variac connected
to the oven heater. Finally, the ammonia feed
rate was regulated by altering the pressure on the
upstream side of the high-resistance tubing. A
pressure gauge was available on the upstream side of
the high-resistance tubing to give the operator an
idea of the concentration levels to expect. After
the reactor reached thermal equilibrium, usually 30
minutes, sampling was initiated. Inlet and outlet
ammonia and nitrogen dioxide concentrations were
determined colorimetrically. Gas chromatography was
52
V"
% Stainless-Steel Tubing
h" Swagelok
h" Pipe to k" Tubing
1" Stainless Steel
Swagelok Cap
1" Stainless Steel Tubing
Catalyst
Alumina Pellets
roooi
Feed
F i g u r e 6 - Fixed-Bed R e a c t o r D(>tail
53
used for nitrous oxide analysis. After data
collection was complete at one condition, the
temperature, ammonia concentration, or air flow
rate could be altered to provide additional data.
Chemical Analysis Methods

This study of the removal of ammonia required
chemical analysis techniques for the quantitative
measurement of ammonia, nitrogen dioxide, and nitrous
oxide in the range from about 5 ppm to 500 ppm in a
background of humidified air. In order to be most
useful the methods needed to be accurate, simple,
rapid, harmless, and inexpensive. The gas sample
size available for analysis was generally not more
than about 100 cm at ambient temperature and pressure.
After considerable investigation, the analysis
methods chosen were colorimetric in the cases of
ammonia and nitrogen dioxide, and chromatographic
for nitrous oxide.
Ammonia Analysis
A relatively extensive effort was made in the
early stages of this investigation to develop a
reliable method of analysis for ammonia using a helium-
ionization gas chromatograph (Varian Model 1532-B).
54
This work is summarized in the following section.

The ammonia-analysis method finally chosen was a
colorimetric technique using Nessler's reagent. A
description of the method is given later in this
chapter.
Chromatographic Ammonia Analysis

Many different schemes for the chromatographic
analysis of ammonia were tried during the course of
the project. Ammonia peaks under all conditions had
extremely long retention times and tailed very badly
Summarized in Table 5 are the columns and conditions
that were investigated.
TABLE 5
Gas Chromatograph Parameters
Support Dimensions Temp. Range Column Const.
Porapak Q 1/8" X 20' Ambient to Stainless Steel

300F
300*'F
300*F
Chromo 104 1/8" X 15' Ambient to Teflon-Lined
300F Aluminum
Porapak Q 1/8" X 12' Ambient to Teflon Tubing
300F Encased in
Stainless-Steel
Shield
55
The shape of the ammonia peak is indicative of

physical adsorption of ammonia on the column and
chromatographic hardware. This is further strengthened
by the fact that a 5-foot length of unpacked tubing
for a column gave identically shaped peaks. In order
to minimize the adsorption of ammonia on the chroma-
tograph hardware, approximately 50 per cent of the
tubing and fittings which are outside the oven were
wound with electrical resistance wire and heated to
250F. This also had little effect on the ammonia
peak shape.
Effort was also directed toward using a very

short column. This would allow everything but
ammonia to pass through the column quickly as well
as minimize the column surface area. This effort did
not significantly improve the ammonia peak shape.
Further attempts to improve the ammonia peak shape
were to inject directly into the column and to use
Teflon-coated tubing. Neither of these methods
proved satisfactory.
Colorimetric Ammonia Analysis and Calibration

Gas samples in this investigation were analyzed
for ammonia colorimetrically using Nessler's reagent
56
in a technique similar to that described in Standard

Methods (17). The method consists of contacting
water containing absorbed ammonia with Nessler's
reagent and reading the transmittance of the resulting
yellow-brown solution against an appropriate blank
on a colorimeter. The intensity of the color change
which occurs is proportional to the ammonia concen-
tration.
The step by step analytical procedure is as

follows.
1. 50 cm polypropylene syringe is loaded with
2 cm-^ of distilled and deionized water.
2. The water is contacted with sample gas by
drawing 48 cm-^ of gas into the syringe, being
careful to let the gas bubble through the
water. The syringe is shaken for approximately
one minute to promote amm.onia absorption.
3. The contents of the syringe are transferred to
a cuvette and 1 cm-^ of Betz Nessler Reagent
No. 246 which has been diluted 5 to 1 with
distilled water is added.
4. After allowing approximately ten minutes for
color development, the transmittance of the
sample is read on a Bausch and Lomb Spectronic 20
57
coJorirnGter at 420 nm against a blank containing
2 cm~* of distilled water and 1 cm of diluted
Nessler reagent.
The sample cuvette and the blank cuvette are chosen
such that they yield the same transmittance values
when filled with distilled water.
Approximately midway through the experimentation,
it was found advantageous to modify the syringe
3
sampling technique by loading the syringe with 3 cm
3
of water instead of two and drawing two 47 cm gas
volumes instead of one 48 cm 3 volume. For this case
3
the blank consisted of 1 cm of diluted Nessler
3
reagent and 2 cm of distilled water. The new
procedure had the effect of decreasing the deviation
in per cent transmittance readings between samples.
With modified sampling procedure, a standard
deviation of 3.4 ppm was calculated at 121 ppm versus
5.1 ppm at 135 ppm for the original procedure. Since
there was no interaction between precision and concen-
tration, these numbers seem to indicate an increased
precision for the modified procedure. Three determina-
tions were made of each gas stream sampled.
In order to have a quantitative analysis, it
was necessary to obtain a calibration curve which
58
relates ammonia concentration to per cent transmittance
This was done by preparing known concentrations of
ammonia in air and establishing the per cent trans-
mittanca associated with each concentration. Known
ammonia concentrations were prepared by passing
measured amounts of air over a liquid-ammonia-filled
Teflon permeation tube in the apparatus diagrammed
in Figure 7. The permeation tube, when held at
constant temperature, lost weight linearly with time
and thus allowed one to know the exact concentration
of ammonia when the air flow rate was knov^n. The
permeation tube weight loss was determined by
periodic removal from the apparatus and weighing.
The apparatus in Figure 7 was operated by

passing air from a compressed-air cylinder through a
coil in a constant-temperature bath, over the ammonia-
filled permeation tube and through a calibrated
wet-test meter. The permeation tube was held at
constant temperature in a jacketed glass housing.
Water from a constant-temperature bath was circulated
through the housing via a positive-displacement pump.
From the above calibration procedure, it was
possible to prepare the calibration curve of Figure 8.
59
4->
0)
0 I
0)
3:
(U
e
w
3
-P
fd
fd
a,
<
CU o
o
H
4-1 0)
0^
3
H
-P
fd
fd x> u
Q) 3 H
(U rH
u u fd
3
u
-P
fd
T B
5-1
Q)
Pu
-p
>-l
3
(D
fd en
m H
-P
fd
4J
j
(0
c
o
u
K h M
0 (U
p ^ TJ
fd H d
rH < -H
3 rH
en
Q)
>i
U
iii__._
o 60
1 1 T" 1 1 i/ 4 1 I 1
/ O
1 O
fN
O
o
fd
c
O -H
00
S O
B -P
X
o U
0)
rH
CO
to
/ (U
/ o 00
^
3
/
H
/ o
CN
'
)
1 1 1 1 1 1 1 1
O o o o o
fN
o 00
e)ûe::^:^TlusueJJ, :Û93 jaj
61
The linearity of the line shows that Beer's law is
obeyed in the concentration range investigated.
The only interference problem encountered with

the Nessler reagent was that aldehydes cause the
solution to be clouded and give a false reading.
Careful attention to the purity of the water used
in preparing reagents and avoidance of aldehydes
around the analysis area prevented interference.
Nitrogen Dioxide Analysis

The analysis of nitrogen dioxide was performed
in a manner similar to the ammonia analysis. Sample
analysis was performed colorimetrically using the
modified Saltzman reagent (18) and the syringe
method of Meador and Bethea (19).

3
The analysis is performed by loading 5 cm of
Saltzman reagent into a 50 cm-^ polypropylene syringe

and withdrawing 45 cm-^ of gas sample. The syringe
is shaken vigorously for approximately three minutes
after which time a magenta color completely develops.

The contents of the syringe are emptied into a
cuvette and read on the Bausch and Lomb Spectronic 20

colorimeter at 550 nm against a blank consisting of
5 cm-^ of undeveloped Saltzman reagent. Using this

62
technique Meador and Bethea (19) report a relative

error of 0.43 per cent at 41.3 ppm, 1.03 per cent
at 26.3 ppm, 2.83 per cent at 9.8 ppm, 1.77 per cent
at 5.7 ppm, and 2.29 per cent at 2.6 ppm.
Calibration of the Saltzman reagent for nitrogen

dioxide analysis was done in the same manner as for
the ammonia analysis. Known concentrations of
nitrogen dioxide in air were prepared by passing air
from a compressed-air cylinder over a Teflon
permeation tube filled with liquid nitrogen dioxide.
The apparatus diagrammed in Figure 7 was used for
this purpose, as it was in the ammonia-calibration
work. A typical nitrogen dioxide calibration curve
is shown in Figure 9.
The Saltzman reagent was prepared by mixing

the following chemicals with enough distilled and
deionized water to make one liter of solution:
0.050 g N-(1-napthyl) ethylene-diamine dihydrochloride

0.050 g 2-napthol 3,6 disulfonic acid disodium salt
1.500 g sulfanilamide
15.0 g tartaric acid.
Nitrous Oxide Calibration and Analysis
Nitrous oxide was analyzed chromatographically
using a Varian Model 1532-B trace gas analyzer employinq

63
a helium-ionization detector. The following is a

listing of the pertinent chromatograph operating
conditions which were used:
packing material Porapak Q
column length 20 ft.
column temperature 95F
detector temperature 212 F
detector voltage 360 V
helium flow rate 1 cm-^/sec.
Because the sensitivity of the helium-ionization
detectors decayed with time, it v/as necessary to
calibrate the chromatograph daily. This was easily
done with two prepared sample cylinders purchased
from the Matheson Company containing 33 ppm and
103 ppm nitrous oxide in air. Approximately four
each of these samples were injected daily to provide
calibration peak heights. Nitrous oxide peak heights
from reactor gas samples were then referred to these
standard peak heights in order to determine nitrous
oxide concentration. A standard deviation of 0.59 ppm
at 33 ppm was calculated from typical calibration data
64
100
90
80
70
60
i\

U 50
c N
fd ^
-p
-p 40
H
B
(0
fd N.
u 30
EH
-P
C
(D
U
V
20
\
s
10
0 10 20 30 40
PPM NO2
Figure 9 - Saltzman Calibration Curve
CHAPTER V
EXPERIMENTAL DATA AND NUMERICAL PROCEDURES
A backmix reactor was used to obtain some data

so that preliminary indications of rate-expression
suitability could be obtained. The backmix reactor,
because of its operational method, leads to direct
calculation of reaction rates. This is more clearly
seen if one considers a steady-state material balance
on the reactor shown below.
F. F
1 I 1 o
V ^
Âi Âo
At steady state a material balance on component A
yields
F. C^. - F C^^ + Wr^ = 0 (5-1)
1 Al o Ao A
where F. = total molar flow rate in
1
F
= total molar flow rate out
o
C,. = inlet concentration of A
Al
C, = outlet concentration of A
Ao
r. = rate of production of A
W = weight of catalyst.
This study involved the reaction of ammonia in
the ppm range in air, consequently, the total molar
flow rate in can be assumed equal to the total molar
65
66
flow r a t e o u t . Then l i q u a t i o n 5 - 1 c a n be rewritten
Wr^ = F . (C^ - C^.). (5-2)

A 1 Ao Al
Both the molar flow rate in as well as the inlet and
outlet ammonia concentrations were measured so the
rate can be calculated directly. Since the rate is
for the rate of disappearance of ammonia. Equation 5-2
can be rewritten as
^î - ^0^ ., 3.
^NH3 = WTF ^^"^^
where C^ = inlet ammonia concentration
C^ = outlet ammonia concentration
r,, = rate of ammonia disappearance.
NH3
The data collected on the backmix reactor at

350F and the calculated rates are given in Table 6.
The catalyst used to obtain the backmix reactor
data was installed in the reactor on July 14, 1972, and
during the time between its installation and the
acquisition of the 350F data, it ran continuously.
Because the catalyst had been in service for over 100
hours. Equation 3-27 should be valid if the assumed
surface decomposition is controlling.
If Equation 3-27 is rewritten as
^4 S H - 1 SH^
67
I
O
o
B
fd P vo in CN t^ r^ [^ o CT> cy> in
fd VD ^ 00 r>- 1^ in fsj vo o r* ^
CN X> ^ r^ o CN VD
u
o ro CN fN CO CN ro CO vo in
B
a.
fa
o
O
in
en VD
P o
fd iH
0)
(d -p X
fd
D
s 4J
c fd fH r^ o rH o CN CO in 00 'ii' in cy in
0 in iH a\ CN CO in t^ CN r^ CN r-\ in r^
PQ o o VO
< p +J Cr rH 00 VD 00 'sr CN ro r- o vo i n (T 00 a\
EH o ^' CO rH CVJ CM CN cn CN iH r-{ CN CN
fd u
fd *-%.
CN CN iH
(U 0)
(hr
c
f^
X ffi rH
OUI
H z
;^
o
fd
CQ
CN fvj CN ro *^ ro CN vo <*
in v) vo 00 O 00 CN CM r* iH vo in o rg
rH CM r^ r-
u iH
iH ^ in CN a^ cj> r^ in ro
ro
a in r>j vo VO CN
i n c N ' : : j < i n i n i n o o c N f M i n i n o o o o
CN CN ro
U r o i n < x ) r - r H r i H ' * o o r > - r ^ r ^ L n t ^

r O r H C N C N r O r O C N r H C N C N r O f O i H
CVJ CN rsi CM CN CN rsi CN

CN CN CN CN CN rsi r^ r- r^ r^ l^ r- r* r^
(1> r^ r> r^ r* r^ r- 1 1 1 1 1 1 1 1
-p r 1 1 1 1 1 o o o o O o CN ^
fd in r^ 00 00 00 (T\ fH tH fH rH rH fH fH CN
o 1 1 1 1 1 1 1 1 1 1 1 1 1 1
00 00 00 00 00 00 00 00 00 00 00 00 00 00
8
a plot of Cj^j^ /r vs. C should yield a straight line.

C^jj is the outlet or tank concentration since the
reactor is assumed to be perfectly mixed. These data
are plotted in Figure 10. The linearity of the plot
indicates that the proposed model fits the data quite
well. The slope of a straight line passed through the
points should equal K./K3 and the intercept should be
equal to I/K3. A linear least-squares fit of the
data yields the slope equal to 0.0296 and the intercept
equal to 0.60. From this the values of K3 and K. are
1.66 mol/(hr)(g cat) and 0.0493 ppm , respectively.
Application of this rate expression to the data
obtained on the integral or fixed-bed reactor requires
that the rate equation be substituted for r in the
design equation and integrated to the form given in
Chapter III.
w 1 1 4
F = iq '" ^ * 4 ^""^ ''o' ''''
These data obtained at 300F, 320F, 350F, and
400**F are given in Tables 7, 8, 9, and 10, respectively.
Using these data obtained from catalyst which had been
in the reactor at least 75 hours, the constants K-.
and K^ were calculated using the least-squares pro-
cedure. Values fpr W the catalyst weight, F the
69
in
'
o
\ in
rH
fa
0
o
in
ro
in -P
(N fd
r-i
B
ro CO
K H
2 3
fd
eCL jr:
O
o
o
ex (U
iH ^ s
\ 3
c
0 0
H H
-P 4J
fd O
>-i fd
\ -p Q)
in
c:
r-
<u IH
u
3 0
0
U -P
to
Ai <u
c t^
fd
H 1
1
o
Figure 10
\ in
\

\
in
\ CN

\
00 VO
o
g_OTXiom/(:;po 6) (J^) (uidd) 'o:^py/ D
70
TABLE 7
Kinetic Data :Eor Fixed -Bed Reactor at 300F
ppm NH3 Air Catalyst Weight
Date l ft3/hr* grams
1-9-72 41.7 30.1 10.0 12.033

1-9-72 74.0 52.8 10.0 12.033
1-9-72 84.0 63.0 10.0 12.033
1-10-72 96.0 75.6 9.8 12.033
1-10-72 123.0 93.5 10.0 12.033
1-12-72 149.3 127.6 10.0 12.033
1-12-72 78.0 61.0 10.0 12.033
1-12-72 109.8 90.0 10.0 12.033
7-15-72 81.6 72.8 6.6 7.9954
7-15-72 85.8 76.2 6.1 7.9954
7-15-72 215.0 195.2 6.1 7.9954
7-17-72 143.6 127.5 6.1 7.9954
7-17-72 196.5 177.4 6.1 7.9954
7-17-72 55.9 50.0 6.1 7.9954
7-18-72 23.2 14.5 6.1 7.9954
7-18-72 123.0 99.2 6.1 7.9954
7-18-72 218.0 197.9 6.1 7.9954
8-3-72 94.6 70.9 9.9 13.0064
8-3-72 197.9 154.8 9.9 13.0064
8-3-72 147.6 114.4 9.9 13.0064
8-3-72 241.0 201.0 9.9 13.0064
8-4-72 264.9 227.4 9.9 13.0064
8-4-72 218.0 171.5 9.9 13.0064
8-15-72 148.6 129.3 10.0 13.0064
8-15-72 271.0 263.0 10.0 13.0064
8-15-72 65.1 51.1 10.0 13.0064
*Air flow measured at ambient conditions

(approximately 72F and 0.9 atmospheres)
71
TABLE 8
Kinetic Data for Fixed-Bed Reactor at 320F
Date i Sa ;^V!ll* g^^"^s
7-21-72 134.9 70.9 10.1 13.0064
7-24-72 115.2 76.2 9.9 13.0064
7-24-72 116.9 80.2 9.9 13.0064
7-25-72 123.0 85.1 9.9 13.0064
7-26-72 118.6 68.9 10.0 13.0064
7-26-72 128.4 83.0 9.9 13.0064
7-27-72 237.3 187.4 9.8 13.0064
8-15-72 59.5 46.1 10.0 13.0064
7-13-72 236.2 187.4 6.1 7.9954
7-13-72 161.3 118.6 6.1 7.9954
7-13-72 69.5 42.0 6.1 7.9954
7-13-72 123.9 88.7 6.1 7.9954
7-14-72 195.2 154.8 6.2 7.9954
7-14-72 46.5 24.2 6.2 7.9954
7-14-72 23.7 4.8 6.2 7.9954

(approximately 7 2F and 0.9 atmospheres)
72
TABLE 9
Kinetic Data for Fixed-Bed Reactor at 350F

Date Cj CQ ft3/hr* grams
1-19-72 47.8 24.6 9.9 12.033

1-19-72 71.4 42.9 9.9 12.033
1-19-72 120.3 85.9 9.9 12.033
1-20-72 162.4 106.0 10.1 12.033

1-20-72 199.0 145.3 10.2 12.033
1-20-72 96.7 68.6 10.2 12.033
1-20-72 208.0 167.0 10.0 12.033
1-21-72 191.0 149.0 10.0 12.033
1-21-72 98.0 63.0 10.0 12.033
1-22-72 141.0 95.0 10.0 12.033
8-14-72 192.5 120.4 10.0 13.0064
8-14-72 227.4 165.8 10.0 13.0064
8-14-72 127.5 57.0 10.0 13.0064
8-14-72 78.9 28.2 10.0 13.0064
8-14-72 43.1 17.3 10.1 13.0064
8-14-72 163.5 96.1 10.1 13.0064
8-17-72 163.5 103.8 10.0 13.0064
8-17-72 200.6 134.9 10.0 13.0064
73
TABLE 10
Kinetic Data for Fixed-Bed Reactor at 4 00F

Date ft3/hr grams
1-13-72 13.9 1.9 9.92 12.033
1-13-72 34.7 3.9 9.90 12.033
1-13-72 56.6 1.9 9.90 12.033
1-14-72 99.9 4.9 9.95 12.033
1-14-72 255.0 24.6 9.95 12.033
1-14-72 193.9 17.1 10.13 12.033
74
molar flow rate of reactants, C^ the inlet concentra-

tion of ammonia, and C^ the outlet ammonia concentration
were known for all data points at all four temperatures.
The constants K3 and K. were evaluated independently
at each temperature. This was required since K3 and
K, are functions of temperature.
During the experimental portion of this study,

it was discovered that the catalyst activity did decay
with time but that the activity could be restored by
passing pure ammonia over the catalyst at 350 to 400F.
-3
The ammonia flow rates were approximately 3.0 x 10
moles per hour. Because regeneration was possible,

the catalyst could be returned to a base or regenerated
state and data taken to determine the effect of time
on the rate of reaction. This was accomplished by
regenerating the catalyst and then putting the
reactor back on line.

During this time the inlet or feed concentration
was held constant and the outlet concentration measured.
Three complete regenerations and one partial regeneration
were conducted at 350F. After each regeneration, a
run was conducted to determine the effect of time on
catalyst activity. The data obtained during these
runs are given in Tables 11, 12, 13, and 14. An

75
TABLE 11
Platinum Catalyst Regeneration Data
Regeneration I
Temperature 350F Humidified Operation

Feed Concentration 195 ppm
Catalyst Weight 13.0064 grams
Regeneration Conditions 16 hours at 350F
Air Flow Rate 10.1 ft /hr at 72F and 0.9 atmospheres
ppm NH3
Date Time Hours Exit
8-6-72 5:15 P 3.25 31.3
8-6-72 11:15 P 9.25 55.8
8-7-72 12:15 P 22.25 74.8
8-7-72 10:30 P 32.50 91.6
8-8-72 11:00 a 45.00 99.2
8-8-72 4:00 P 50.00 107.0
8-9-72 12:15 a 58.25 111.9

76
TABLE 12
Regeneration II

Feed Concentration 189.5 ppm

3
Air Flow Rate 10.1 ft /hr at 7 2F and 0.9 atmospheres
ppm NH3
8-9-72 10:45 P 1 9.8
8-10-72 12:45 a 3 27.2
8-10-72 3:15 a 5.5 42.0
8-10-72 5:15 a 7.5 45.0
8-10-72 3:00 P 17.25 67.0
8-10-72 9:45 P 24.00 78.9
8-11-72 3:45 P 42.00 85.8
8-12-72 1:45 P 64.00 113.6
8-13-72 10:45 P 97.00 125.2
8-14-72 10:00 a 108.25 120.4
8-17-72 3:30 P 185.75 134.9

77
TABLE 13
Regeneration III

Regeneration Conditions 15 minutes at 350F
ppm NH3
8-17-72 5:00 p 1 64.4
8-17-72 11:00 p 7 77.6
8-18-72 1:00 a 9 87.3
8-18-72 Noon 20 100.7
8-18-72 5:00 p 25 105.4

78
TABLE 14
Regeneration V
Temperature 350F Non-Humidified Operation

Air Flow Rate 10.1 ft^/hr at 72F and 0.9 atmospheres
ppm NH3
8-23-72 1:00 P 1 20.2
8-23-72 2:45 P 2.75 30.3
8-23-72 4:45 P 4.75 36.1
8-23-72 10:30 P 10.50 38.8
8-24-72 11:15 a 23.25 46.5
8-25-72 12:01 a 36.0 64.4
8-25-72 11:45 a 47.75 84.4

79
additional run was made at 400F but after 28 hours no
decay in catalyst activity was observed. These data
are given in Table 15. It should also be noted that
Regeneration V gave the only regeneration and decay
data obtained under dry air conditions. All other
decay data were collected using air at 23 per cent
relative humidity at 7 2F.
The 350F data were the only data readily adapt-

able to analysis with a time-independent model,
because they were the only data obtained that had both
a common reference point (completely regenerated)
and the times (hour and minute) recorded. The solution
for the constants K, and K^ required only that the
integrated form of the time-dependent model be rearranged
from that given in Chapter III (Equation 3-29) to
Kê '^2^ = -K3 + w^F + i/F-^- '5-6)
Since everything on the right of the equal sign is

known, including K-. and K. which are the same numerical
value as obtained from Equation 5-5 at identical
conditions, the logarithm of both sides yields a form
that is easily handled by linear least squares. The
slope obtained is -K^ and the intercept is In K,.
80
TABLE 15
Regeneration IV

Feed Concentration - Variable
3
ppm NH3
Date Time Hours Feed Exit
8-21-72 1:15 p 1 184.8 8.9
8-21-72 3:15 p 3 206.2 8.9
8-21-72 9:15 p 9 210.0 8.0
8-22-72 10:30 a 22.25 215.0 10.7
8-22-72 4:15 p 28.0 301.3 22.2

CHAPTER VI
RESULTS OF EXPERIMENTAL WORK
The preliminary results based only on one set

of data obtained at 350F in the backmix reactor
seemed to indicate that the model developed in
Chapter III represented the data quite well. This
conclusion is based on the linearity of the plotted
data as shown in Figure 10.
Long-Time Data
Application of the rate equation to integral or
fixed-bed data requires using the fixed-bed reactor
design equation that has been integrated from C to
C^. This equation as given in Chapter III for the case
where time is no longer a factor is
Application of the least-squares procedure described
in Chapter V to the data obtained at 300, 320, 350,
and 400F resulted in the values of K3 and K. shown
in Table 16.
The constant K. for the 400F data was determined
to be zero based on the least-squares analysis. If K.
is zero, the proposed rate expression reduces to the
81
82
form expected of a first-order reaction. This result
was not unexpected because the 400F data indicated
that the catalyst activity was independent of time
at that temperature which caused some suspicion that
a mechanism change might occur between 350F and 400F.
This point is not the objective of this work, because
it lies outside the 300 to 350F temperature range of
this study. The 400F data were taken only to provide
an additional point for an Arrhenius plot.
In most kinetic rate expressions, the rate

constants are functions of temperature. The temperature
dependency has been found in practically all cases to
be well represented by Arrhenius' law (12):
K = Ae"^/^^ (6-2)
where K = rate constant

A = pre-exponential term or frequency factor
E = activation energy
R = universal gas constant
T = absolute temperature.
If this equation is rearranged by taking the natural
logarithm of both sides, the equation is
In K = In A - E/RT. (6-3)
If the natural logarithm of the rate constant is

plotted against the reciprocal of the absolute
83
temperature, the slope of the curve is -i:/R and the
intercept is In A. An Arrhenius-type equation is
frequently attempted for temperature correlation
of empirical constants. The Arrhenius approach was
attempted for temperature dependency correlation of
K3 and K^. Figure 11 is an Arrhenius plot of the
constants given in Table 16. Based on this plot,
the pre-exponential constant and the activation
energy associated with K^ are 3.746 x 10 and 13.556
kcal/mole, respectively. When the same analysis
was applied to K., the values were 7 4 3.5 for the pre-
exponential constant and 9.525 kcal/mole for the activa-
tion energy. The units for K^ and K. are mol/(hr)(g cat)
and ppm , respectively. The long-time fixed-bed
reactor equation given by Equation 6-1 can now be
rewritten as
A3e y CQ A3e 3'
where C = inlet ammonia concentration, ppm

C^ = exit ammonia concentration, ppm.
This equation was used to predict outlet concen-
trations based on inlet concentrations at 300, 320,
350, and 400F for all of the long-time data. Plots
of the inlet concentration versus the outlet concen-
84
TABLE 16
Rate Constants for the Long-Time Model
K3 K4 X 10^
Temperature mol/(hr) (g cat) ppm~l
300F 0.3626 9.0115

320F 0.5263 11.05
350F 1.020 17.97
400F 2.367 0.0
85
30
20
K4 X 102
10
9 V^
8 VX
7 V T ' -
6
5
3 4 -
1
fd
ro
Ui
(0
p \
3 >
fd
4J
(0
3
O
\^K3
u
<u
+3
fd

1
.9
.8
.7
.6 V
.5 v
.4 \
\
#.
.3
K3 = m o l / ( h r ) ( g ca :)
K4 = ppm" 1
20 21 22 23 24 25
1/T, O
OK
v-l X 10'
Figure 11 - T^rhenius Plot for Constants K3 and K4
86
tration are given in Figures 12, 13, 14, and 15 for
the 300, 320, 350, and 400**F data, respectively. The
points on these plots are the observed data points
while the solid lines are the predicted performance
based on the model represented by Equation 6-4. Visual
inspection of these plots indicates that the model
represents the experimental data quite wel4.. This is
further reinforced when the next three plots. Figures
16, 17, and 18, are viewed. These plots slvOW the
calculated reactor exit concentration plotced against
the experimental outlet concentration. If the model
was an exact representation of the experimental data,
all points would fall on a line with a slope of one
and an intercept of zero. The solid lines drawn on
these figures are such lines.
Short-Time Data
During the experimental work, it was observed
that the platinum catalyst could be regenerated by
passing pure ammonia over the deactivated catalyst
at 350-400F. At 350F and with a humidified air feed,
the catalyst activity was observed to decay quite
rapidly to a fairly constant value over a period of
about 75 hours. After this time the decay was
87
o
in
, Model
ro
u
o 3
\ o fd
erform
ro

o p^
in
ctor
(N
^ \
+J (d Pri
o (U (U
o iH o
CN 3 O
\ * H TJ O
(U ro
1
1 4J
ro TJ fd
o ffi (U
\ . in S X
rH H
B fa
\ a. 1
a
o rv)
o f-K
iH
igure
^ k\
O
in
fa

in o o
o
O o o
r- in in o in
(N CN CN
:^^X:^no - ^HN uidd

HH
in
ro
o
o fa
ro 0
o
CN
Model
ro
+J
o fd
in 0)
rformanc
rsi
o -p Q)
o (D PK
CM r-^
3 u
acto
M

1
ro <D
K Pi^
o 2
in TJ
rH B (U

a
Fixed-
a
O
o 1
ro
3
Cn
H
o fa
in
o o O o o
in o in o in
rvj CN
^ST^no - ^HN uidd

o 89
in
ro
o
o fa
ro O
o
in
ro
Model
fd
Q)
CM U
3
fd
B
u
o
u
o -p
o
CN
3 U
H O
+J
I U
fd
ro Q)

o
in 13
B Q)
a PQ

a I
13
\ 0)
X
H
fa
o
o I
\
\ 0)
3
Cn
H
fa
in
o o o o o
in o in o in
CN
râXt^no - ^HN uidd

o 90
in
ro
from Regenerati Dn IV
^ o
o fa
ro 0
from rable 10
Model
o
o
'^
4-)
fd
o
in <u
rformanc
rsj
\ <
<^ o -p 0)
o Q) Cn
rsj rH
\
3 u
acto
< M
1
ro Q)
ffi oi
o 2
I in 13
iH
B 0)
1 Ou CQ
Fixed-
O.
o
o 1
\
in
0)
5-1
3
Cn
H
o fa
in
o o o o o
in o in o in
CM CM
:âx:vno - ^HN uidd
o 91
m
ro
o
o
ro
fa
0
o
o
ro

o 4J
in fd
(N
rH
fd
\ rH 4J
fd 3
4J 0)
3 e
(D H
o B )-l
^ o -H 0)
rsi U a
0) X
X w
a
\ H CO
\ >
^
4-> 13
o 3 Q)
in O -P
fH fd
ro iH
K 3
S U
fH
B fd
a u
O a 1
o
<-^
Figure 16
9
V
o
in
o o o o o
in o in o in
CNJ CNJ rH iH
PB:ÊX1^3T^0 '^1^0 ^HN uidd

92
in
II
'
;
ro
o
o
ro
fa
0
o
i i
CM
ro
I o 4-)
^ ^ in fd
r>j
' i iH
I I fd
i rH 4->
fd 3
-p Q)
3 B
(U H
o B U
r \ \ o H (U
(N U a
Q) X
a.
X
w
w >to
^
4-) 13
\ | o 3 (U
in 4J
rH o fd
ro .H
E 3
2 u
B rH
fd
a u
o
a. 1
o
Figure 17
f-{
o
in
o o o o o
in o in o in
CN rI r1
patêxiôxeo ':^no ^HN uidd

93
in
ro
o
o
ro
fa
0
o
in
ro
o 4->
in fd
CM
rH
fd
r-\ 4-)
! fd 3
4J 0)
3 B
I Q) H
o B U
o -H Q)
CN U
I
a
X
i
o fa
a
X
w to
K
>
4- 13
O 3 (U
-^f^; 1 in
iH
O -P
fd
ro iH
K 3
S U
rH
B (d
Cu u
O
a 1
% \ , .
o
Figure 18
iH
\^
O
in
\
\
v
o o o o o
in o in o in
CN CN
pe:êxnox^D '^no ^HN uidd

94
considerably slower and after 100 hours the activity
was essentially constant. This observation led to
the postulation that some specie, X, must be combining
with an active site. This combination continued until
an equilibrium state existed where the rate of
combination was equal to the reverse rate. At this
point the catalyst activity became constant and time
no longer was a factor. The development of the model,
based on the kinetics arrived at from the long-time
data and the X-active site combination, is given in
Chapter III.
The use of conventional least-squares procedures
and the data obtained for the short-time model in
Tables 11 and 12 resulted in values of 2.723 mol/(hr)
(g cat) and 0.0253 hrs for K, and K2, respectively.
Application of the model
I = [ln(C^/C^) + K4(Cj - C Q ) ] [K3 + Kê'"^2^] (6-5)
where t is the time in hours and all other symbols

have been previously defined, yielded the results shown
in Figures 19 and 20. As before, the points are the
experimental data while the solid lines are the perform-
ance the model predicts.
95
o
3
0
H B fa
4-) CU 0 o
195 ]
rrt O. r^j
35(
Humidified Operc
Temperature
NH3 Inlet
o
o
3
O
H
4J
fd
o CO }^
0)
00 u 3
3 Q)
Mode 1
O
Q)
PC
0)
B I
H
o CT\
0)
V4
\
3
v H
fa
c
N.
\
o
rsj
^ \ ^ ^
*^^^^^^#
o o o o o o
rsj o 00 CM
Uidd ' u o T : ^ e j : : ; u a o u o 3 :âx:^no HN

c 96
o
e
1
a 1
a
1
in
in
3
89.
0 iH
H fa
4J 3 o iH
fd 0 o (U
H in 13
u
Q) 4-> ro 0
Ch fd
O U Q) s o
4J ^ in
13 3 3
(U 0) 4-1
H u fd
4-1 3 S-i
H 0 Q)
13
H u g
a
M
g 4J (1)
in M
3 Q) EH
ffi ,_l CN
ation
3 fH
M
ro 1
l*
ffi
^. to u
U Q)
3 3
O 0 (D
O
a: CP
iH (U

^
<D
B 1
H
EH O
CM
0)
in M
Figu
r--
c
in
in
\ CN
o o o
o o in
CN in o
uidd ' u o T : ^ p j : ; u a o u o D tâx'^^no HN
97
An attempt was made to determine the length of
time required to regenerate completely the catalyst.
A regeneration time of 15 minutes at 350'*F resulted
in the data given in Table 13. Based on the outlet
concentration at one hour for both Regenerations I
and III, the activity seems to be 50 per cent of the
totally regenerated catalyst. The comparison is
valid since operating conditions are essentially
identical for both runs.
The fourth regeneration was conducted at 400F
for a period of 14 hours. The reactor was then returned
to service but at 400F. After 28 hours, the conversion
was still at 93 per cent which was only slightly lower
than the 9 5 per cent conversion at one hour. Based
on these data, the catalyst activity decay was judged
negligible at 400F.
The last regeneration was to determine if the
elimination of water from the air feed stream would
affect the catalyst decay curve. The same analysis
was applied to the data obtained after Regeneration V
as was applied to the data from Regeneration I and II.
The constants K, and K2 obtained were 3.26 mol/(hr)
(g cat) and 0.016 hrs" , respectively.
98
In order to determine if there was a significant
difference between the rate of decay obtained for the
humidified runs and for the non-humidified runs,
constant K2 was statistically tested as described
by Volk (20). The results of the statistical test
show that there is a significant difference between
the slopes at the 99.9 per cent level. These results
indicate that the humidity in the air does affect the
rate of decay of the catalyst.
If the constants obtained for the 350F non-
humidified run are used in the model, the curve given
in Figure 21 is obtained.
Backmix Reactor
The constants K3 and K. obtained for the long-time
data were used to calculate the reaction rate from the
non-integrated rate equation. The curve obtained is
plotted in Figure 22. This shows how the constants
obtained from integral data agree with differential
data. The agreement seems to be fair, but the rates
predicted at high concentrations seem to be slightly
higher than those obtained in the backmix reactor.
This could be due to physical factors such as not
having a perfectly mixed reactor, leaks, or analytical
problems.
99
o
in
3
O o
Non-Humidified Operati
t
!
207 ppm
350F
o
Temperature
Inlet
o
o
NH
i 3
1
1 O
H
4J
fd
o U u
00
3 <D
O 3
(U
CP
0)
O^
B
i c
H I
CN
QJ
U
3
^ Cn
H
fa
c
\
\ ^
o
CN
o
o o o O
X. o
rsi
00
Uidd ' u o T : ^ p j : ^ u a o u o o :âx^no HN

100
in
t^
i
Model
to o
^ 11
in
13 3
0) H
CQ X 0
1 H Cl^
13 B
\ ^ M fd
\ X O 4J
\ ' - ' fd fd
CQ Q in
\ ^ CN
rH
son

B H
ih 5H
Cu fd
O ^ a.
B
o 3 o
-H 0
-H
u
4-) iH
td (U
u 13
4-)
3
o
dJ
\ U 1
\ % in 3
A V
r^ 0 CN
CN
u
ro (D
\ ffi U
Figu
S
\
o
in
\ <
> in
CN
Nv
o o o
ro
O
CN
in
(:^PD 5) ( : [ q ) / x o u i b ' OT ^ ^ ^ ^ H uoT:ôpaH

9
CHAPTER VII
CONCLUSIONS
The catalytic oxidation of ammonia over a supported

platinum catalyst was studied at atmospheric pressure
(approximately 0.9 atmospheres in Lubbock, Texas). The
temperature range investigated was 300 to 400F. The
major conclusions reached from this investigation are
as listed.
1. The reaction rate of ammonia can be successfully

predicted by use of the Langmuir-Hinshelwood
approach if one assumes the surface reaction is
the rate-limiting step.
2. The reaction rate of ammonia is not only a
function of ammonia concentration and temperature
but is also a function of time. The time
dependency disappears after 75 hours.
3. A time-dependent form of the quantitative
mathematical model is derived that predicts
the performance of the freshly regenerated
catalyst. The model may be simplified to a
time-independent form for those data after
75 hours.
4. The platinum catalyst may be regenerated by
passing pure ammonia over the catalyst at 350-400F

101
LIST OF REFERENCES
1. Klimisch, R. L. and Schlatter, J. C. "The Control

of Automotive Emissions by Catalysts", General
Motors Research Publication GMR-1268, September 15,
1972.
2. Zawadzki, J. "The Mechanism of Ammonia Oxidation

and Certain Analogous Reactions", Trans. Faraday
Soc. , 8^:140 (1950) .
3. Gully, A. J., Graham, R. R., Halligan, J. E., and
Bentsen, P. C. The Catalytic Removal of Ammonia
and Nitrogen Oxides from Space Cabin Atmospheres,
National Aeronautics and Space Administration,
Washington, D.C., February 1973.
4. Gully, A. J., Graham, R. R.,and Bentsen, P. C.
The Catalytic Removal of Ammonia and Nitrogen
Oxides from Space Cabin Atmospheres, Progress
Report No. 22, June 30, 1971.
5. Griffiths, D. W. L., Hallam, H. E., and Thomas, W. J
"Infra-Red Study of Adsorption and Oxidation of
Ammonia on Silica-Supported Platinum and Silica",
Trans. Faraday Soc., 6^:3361-3369 (1968).
6. Novak, Miroslav and Vosolsobe, Jan. "Nonplatinum
Catalyst for Ammonia Oxidation", Min. Prum, CSR,
Prague, (Czech.) Chem. Prum., 20, 316-22 (1970);
Original not seen, condensed in Chem. Abs.,
123942Y, 21i 1970.
7. Zasorin, A. P., Kleschev, N. F., and Atroshchenko,
V. I. "Oxidation of Ammonia on a Complex Catalyst
Under Pressure", (Russian) Khim. Prom. (Moscow),
46 (7), P. 513 (1970); Original not seen, condensed
in Chem. Abs., 122004G, 73, 1970.
8. Amano, A. and Taylor, H. "The Decomposition of

Ammonia on Ruthenium, Rhodium, and Palladium
Catalyst Supported on Alumina", J. ^ . Chem. Soc.,
7_6:4201 (1954)
Giordano, N., Cavaterra, E., Zema, D. "Semi Conduc-
tivity and Catalytic Behavior of Doped Nickel Oxides
in the Low-Temperature Oxidation of Ammonia",
Journal of Catalysis, ^:325-331 (1966).
102
103
10. Schriber, T. J. and Parravano, G. "The Low Tempera-

ture Oxidation of Ammonia Over a Supported Ruthenium
Catalyst", Chem. En. Sci.. , 2^:1067-1078 (1967).
11. Johnson, T. E. "The Oxidation of Ammonia on a

Supported Ruthenium Catalyst", Unpublished M.S.
Thesis, Texas Tech University (1971).
12. Smith, J. M. Chemical Engineering Kinetics,
2nd Ed., P. 274, McGraw-Hill, New York (1956).
13. Hougen, O. A. and Watson, K. Chemical Process
Principles, Part III, Wiley, New York (1950).
14. Satterfield, Charles N. Mass Transfer in
Heterogeneous Catalysis, M.I.T. Press, Cambridge,
P. 79-83 (1970).
15. Kenah, Paula M., Private Communication, Englehard
Industries, July 7, 1972.
16. Levenspiel, Octave. Chemical Reaction Engineering,
2nd Ed., P. 361, Wiley & Sons, New York (1972).
17. Standard Methods for the Examination of Water and
Wastewater, 12th Ed., P. 186, American Public
Health A s s o c , New York (1966).
18. Lyshkow, N. A. "A Rapid and Sensitive Colorimetric
Reagent for Nitrogen Dioxide in Air", J. Air Pell.
Control Assoc. , 15^:481 (1965).
19. Meador, M. C. and Bethea, R. M. "Syringe Sampling
Technique for Individual Colorimetric Analysis
of Reactive Gases", Environ. Sci. Tech., :853-855
(1970) .
20. Volk, William. Applied Statistics for Engineers,
McGraw-Hill, New York (1958).
NOMENCLATURE
A = pre-exponential term in Arrhenius expressioj

2
a = mass transfer area (cm )
Cj = inlet ammonia concentration (ppm)
CQ = outlet ammonia concentration (ppm)
D = diffusivity of reactant in boundary layer
(cm^/sec)
d = diameter of catalyst particle (cm)

P
E = activation energy (calories/g mol)
E^ = effectiveness factor
F = molar flow rate of reactants (moles/hr)
G = superficial mass velocity
G = superficial molar velocity of bulk gas
mixture (moles/sec/cm2)
j^ = 0.357/(Nj^-^^ê)
k = reaction rate constant
k' = reaction rate constant for reverse reaction
K = empirical constant
K = mass transfer coefficient (mole/hr/atm/cm i

g
nm = nanometers
N^ = Schmidt number
Sc
N = Reynolds number
Re
P = total pressure (atmospheres)
P, = partial pressure of reactant in the bulk

phase
104
105
ppm = part-per-million (moles/million liters)

Pg = partial pressure of the reactant on the
catalyst surface
Pt = total number of active sites per gram of

catalyst that are unoccupied
Pt^ = total number of active platinum sites per

gram of catalyst covered with reactants,
products and transition complexes

gram of catalyst at time zero

gram of catalyst at any time
PtX = total number of platinum sites per gram of
catalyst occupied by specie X
r = reaction rate [g mol/(hr)(g cat)]
r^ = rate of diffusion process (mole/hr)
R = universal gas constant
W = catalyst weight (grams)
X = unknown reacting specie
Greek Symbols
a = empirically determined ammonia reaction order

(dimensionless)
e = porosity of bed
y = viscosity (g/cm/sec)
3
p = density (g/cm )
APPENDIXES
106
k.
APPENDIX A
Platinum Catalyst Data
Catalyst Manufacturer Engelhard Industries
Lot Number 18,381
Platinum Content 0.5 per cent

Support Material Alumina
Catalyst Dimensions 1/8 in. x 1/8 in. cylinders
2
Surface Area 97 m /g
3
Apparent Bulk Density 60 lb/ft
2
Catalyst External Area 10.7 cm /g
Pore Volume Distribution
Pore Diameter
Angstroms cc/g
<20 0.01
20-30 0.06
30-40 0.04
40-50 0.05
50-60 0.04
60-70 0.02
70-80 0.005
80-90 0.005
90-100 0.01
100-200 0.01
200-300 0.005
300-500 0.005
500-600 0.02
600-700 0.02
800-1,000 0.01
1,000-1,500 0.005
1,500-2,000 0.015
2,000-2,500 0.01
2,500-3,000 0.01
3,000-4,000 0.015
4,000-5,000 0.01
5,000-10,000 0.01
>10,000 0.005
107
APPENDIX B
Calculational Procedure for Effectiveness Factor
Calculational Procedure
The effectiveness factor is traditionally defined

as
E. = - - ^ (B-1)
^ êff
where r is the observed rate of reaction and r ^^ is
ef f
the rate of reaction that would be observed if all the
active catalyst sites were on the catalyst surface.
Another definition relates effectiveness factor to the
Thiele modulus
g^^tanhjh.)_ (3_2)
The Thiele modulus is defined as

2 V I kT '
X V g K
where V = total volume of a single catalyst particle
S = external area of a single catalyst particle
X
k, = intrinsic first-order rate constant

V = pore volume per unit mass of catalyst
D^ = Knudsen diffusivity, (9700 r^Jr/M)
r = pore radius, cm
T = temperature, K
M = molecular weight.
108
109
From Equations B-1, B-2, and B-3 a trial-and-error
proceudre can be used to calculate E^. Johnson (11)
gives the following procedure.
1. A value of E^ is assumed, and h is calculated
from Equation B-2.
2. A value of k, is calculated from Equation B-3,
using the known values of V , S , V , and D, .

p X g k
3. A new E-- is calculated from
, "^l' obs
Ef = k ^
where (k,) , = observed first-order rate con-
1 obs
stant.
4. If the E^ calculated agrees with the E^ assumed,
then the correct value of E^ has been found. If
it does not agree. Step 1 is repeated. The new
value of E^ assumed is found by adjusting the
old value using a regula-falsi procedure.
Data Used for E^ Calculation
V = 0.0251 cm^
P
S = 0.47 5 cm
X
Vg = 0 . 3 9 cm^/g
2
D, = 0.025 cm /sec
k
(kn)^Ko = 1-02 (l)/(hr)(g cat)
1 ODS
From these data and the trial-and-error procedure de-
scribed, an effectiveness factor of 0.95 was calculated

w^9^tT
lb.fiV

Reaction Kinetics of Ammonia & Nitric Acid

Uploaded by

Copyright:

Available Formats

Reaction Kinetics of Ammonia & Nitric Acid

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Reaction Kinetics of Ammonia & Nitric Acid

Uploaded by

Copyright:

Available Formats

What is the topic of the dissertation?

What is the topic of the dissertation?

What experimental systems are described?

What experimental systems are described?

.

KINETICS OF LOW TEMPERATURE OXIDATION

The author wishes to express his appreciation to

VII. CONCLUSIONS 101

1. Results of Thermodynamic Calculations 5

1. Catalytic Activity Comparison 8

2. Experimental Apparatus - Backmix Reactor 40

3. Backmix Reactor Detail 43

4. Backmix Reactor Catalyst Baskets 46

5. Fixed-Bed Reactor System 50

6. Fixed-Bed Reactor Detail 52

8. Nessler Method Ammonia Calibration Curve 60

9. Saltzman Calibration Curve 64

10. Test of Reaction Mechanism at 350**F 69

11. Arrhenius Plot for Constants K~ and K. 85

12. Fixed-Bed Reactor Performance at 300F 87

13. Fixed-Bed Reactor Performance at 320F 88

14. Fixed-Bed Reactor Performance at 350F 89

15. Fixed-Bed Reactor Performance at 400F 90

16. Calculated vs. Experimental at 300F 91

17. Calculated vs. Experimental at 320F 92

22. Model Comparison 100

For many years, scientists and engineers have

Both of these applications require that the

4NH2 + 7O2 ^:^=^ 4NO2 + 6H2O (1-2)

4NH3 + 3O2 ^:;=r 2N2 + 6H2O. (1-4)

Prior to any actual data generation, the heat of

760 -215.5 > 0.723 X 10^^*

760 -270.5 > 0.723 X 10^^

760 -131.6 0.635 X 10^^

760 -302.0 > 0.723 X 10^^

*0.723 X 10'^^ is the largest number available on the

Both a fixed-bed reactor and a backmix reactor

For many years, the oxidation of ammonia to

4 NH3 + 5 O2 ^^ > 4 NO + 6 H2O. (2-1)

At temperatures above 1470*'F, ammonia can be oxidized

differences between spectra obtained for ammonia only

Attempts have been made to oxidize ammonia over

packed with 9 3 per cent granular ferric oxide and

contact times were 2.45 x 10 and 2.79 x 10

1060**F and 750 to 840F, respectively. Activation

Inlet Hydrogen kcal/mole

Amano and Taylor reported a kinetic expression

of ammonia over platinum at temperatures above 930F.

NH3 + O -> NH + H2O. (2-4)

Those reactions that lead to nitric oxide are:

20H -^^ H2 + O2. (2-7)

Those reactions leading to nitrous oxide are:

NH + O -> HNO (2-8)

2HN0 ->- N2O + H2O. (2-9)

And those leading to nitrogen are:

NH + NH3 -> N2H^ "^ ^2 + 2H2 (2-11)

NH + HNO -^ N2 + H2O (2-12)

NH3 + HNO2 -> N2 + 2H2O (2-13)

2N0 ->- N2 + O2 (2-14)

2N2O ^ 2N2 + O2 (2-15)