JOURNAL OF CATALYSIS 4, 212-225 (1975)

Kinetics of NH3 Oxidation on Pt, Rh, and Pdl

T. PIGNET~ AND L. D. SCHMIDT

Department of Chemical Engineering and Materials Science,
University of Minnesota, Minneapolis, Minnesota 55455
Received April 15, 1975

The kinetics of the reactions involved in the catalytic oxidation of NH, have been studied
as a function of temperature over polycrystalline Pt, Rh, and Pd wires in a steady state flow
system using a differentially pumped mass spectrometer to monitor species concentrations.
Unique and reproducible steady state rates were observed in all experiments for tempera-
tures up to 1400°C at ah compositions. Rates of individual reactions could be fit by Lang-
muir-Hinshelwood rate expressions with constant parameters, and comparisons of these
parameters on different metals give reasonable qualitative interpretations for observed dif-
ferences between metals. The competition of NO and NH, for adsorption sites appears to
be important in determining the rates of various reactions. Differences in selectivity
between metals are caused by the different temperature dependences of rates of individual
reactions rather than by uniform differences in rates. Kinetics measured at low pressures
(0.1-I Torr) are used to estimate selectivities of NO and NI production at conditions of
industrial nitric acid production. The predicted selectivity and optimum temperature and
composition on Pt assuming simple models of flow patterns over the catalyst gauze match
closely those observed in the high pressure process.

INTRODUCTION In this paper we examine the kinetics of

Kinetics of catalytic reactions are the reactions involved in NH3 oxidation on
seldom measured with a precision suitable the transition metals Pt, Pd, and Rh
for predictions of rates at conditions far between room temperature and tempera-
from those of actual measurements. This is tures near the melting points of the metals.
largely because of the uncertainties in de- Surfaces are polycrystalline wires which
termining active surface areas of catalysts duplicate closely the wire gauzes used in
and because of constraints on operating the industrial processes. We shall fit data
pressure and temperature set by system for all reactions to Langmuir-
and catalyst stability and the need to avoid Hinshelwood (LH) rate expressions and
side reactions which may become impor- show that such expressions give generally
tant outside of the conditions studied. The good agreement with data at all tempera-
lack of generality in rate expressions ex- tures and gas compositions. The expres-
pressed as powers of concentration and an sions obtained from measurements
effective activation energy and preex- between 0.1 and 1.OTorr will then be used
ponential factor frequently prohibits the to predict selectivities under high pressure
testing of reaction mechanisms in more conditions. A preliminary account of this
than a qualitative manner. Such expres- work on Pt has been given previously (I).
sions are also not suitable for scaleup to In the present paper rates are compared on
realistic operating conditions. all three metals, analysis with LH kinetics
is extended to the simultaneous reactions
’ This research partially supported by N.S.F. under
Grant No. GK16241. in NH3 oxidation, and scaleup is included.
’ Present address: Institut fdr Physikalische Ammonia oxidation is an ideal prototype
Chemie, Universitat Miinchen, Munich, Germany. reaction for such studies because it in-
212
Copy&bt @ 1975 by Academic Press. Inc.
All rigtds of repmdmicm in any form resel-ved.
 KINETICS OF NH3 OXIDATION 213

volves relatively few species and stoichio- we did not examine the hydrogen con-
metrically independent reactions and be- taming species in detail. The objective of
cause the industrial reaction (to produce our work is to describe the kinetics of the
nitric acid) has been carried out for 50 yr above reactions by measuring rates of N,
under only slightly changing conditions and NO production with various mixtures
(2-4). While operating conditions are well of NH3, O2 and NO. We shall show that
established and highly efficient (conver- our results agree with the suggestion of
sions and selectivity to NO approaching Fogel er al. (8) that the series process,
99%), some questions of catalyst stability reactions (1) and (2), adequately describes
remain (2). the production of N, for temperatures
The reaction between NH3 and O2 below 1200°K and at pressures above 0.1
produces some N,O at low temperatures, Torr. However, we shall not speculate on
but above 400°C the only nitrogen con- reaction “mechanisms” (specification of
taining species observed were NO and N,. adsorbed intermediate species) except as
In measurements at a total pressure of 1 they are implied by measured kinetics.
atm Andrussow and Bodenstein (5) re-
ported traces of NH20H, but recent EXPERIMENTAL METHODS
kinetic studies confirm the absence of The apparatus and procedure have been
other than the above products except in described previously (1,10,12). Gases
trace amounts. The older mechanisms for were passed over heated 0.010 in. diame-
this reaction which involve hydroxylamine ter high purity (> 99.9%) wires in a Pyrex
or imide intermediates cannot be tested flow system. Pressures and partial pres-
(5-7) but from more recent experiments sures in the reactor chamber were mea-
(8,9) it has been suggested that N%produc- sured with a thermocouple gauge and by
tion at low temperatures can be interpreted flows through calibrated leaks. Total pres-
more simply as a bimolecular reaction sures were between 0.1 and 1.OTorr. This
between NO and NH, to produce N,. range was chosen because it is sufficiently
However at high temperatures the unimo- low that the surface reactions are kinet-
lecular decompositions of NH, and NO ically controlled and the gas composition
also produce Nz. Thus one can write the and temperature in the reactor chamber
reactions formally as are uniform. The diffusion time across the
NH, + 02 + NO + HzO, 250 ml spherical chamber is s 0.1 set
(1)
while the residence time was 5 1 sec. A
NH, + NO + N, + HzO, (2) gas molecule therefore makes many more
NH, + Nz + Hz, (3) collisions with cold walls than with the
heated filament. Under these conditions
and
(II) the system may be approximated as a
NO-NzfOz, (4) well-mixed reactor with the gas at room
temperature,
Reaction (1) produces, NO, and reactions
(2) and (4) represent series processes qA Pi
ri=m,
leading to N, through NO intermediates,
and reaction (3) a parallel process (not in- where ri is the rate of production of
volving NO) leading to N,. species i per unit catalyst area A and q is
These are the only reactions necessary the volumetric flow rate. Reaction rates
to describe the nitrogen containing prod- are thus simply obtained by measuring ini-
ucts except for NzO. We assume that H, tial and final partial pressures, surface
from reaction (3) would be quickly ox- area, and system volume and pumping
idized to Hz0 in the presence of 02, but speed. The above equation assumes that
214 PIGNET AND SCHMIDT

the mixed reactor equation is applicable. A boundaries spaced -0.05 cm apart. How-
more stringent requirement is that the gas ever extensive etching occurred on Pd
composition at the surface be the same as which formed a structure permeated with
that in the entire chamber. Calculations in- small pits - 2 pm in diameter. Therefore,
dicate (II) that this should be valid for while surface areas of Pt and Rh were
P = 0.1 Tort-, but possibly not for pres- close to the geometrical areas, comparison
sures greater than 1 Torr. of reaction rates on Pd with those on Pt
Surface temperatures were maintained and Rh may be inaccurate. However, it is
by resistive heating and were measured by possible that reaction occurs mostly on ex-
a four lead potential network from the ternal surfaces, and rates will be reported
known resistivity versus temperature of on that basis. Examination of wires and
the three wires. Short lengths of 0.003 in ribbons of Pt which had been heated in
diameter Pt were welded between the cur- O2 and in NHs/02 mixtures by Auger elec-
rent leads and the wires to obtain uniform tron spectroscopy (AES) and sputtering
temperatures. For high temperatures techniques showed that the surfaces were
where an optical pyrometer could be used, generally free of contaminants and con-
temperatures were observed to vary by no tained only monolayers of oxygen or
more than 100°C over the length of the carbon (12). Most of the results reported
wires and these deviations occurred only here were obtained at a total pressure of
over short segments near the ends. 0.115 Tot-r. Data were generally consistent
Gas compositions were determined by with results reported previously (I) at
leaking a small fraction of the gases in the pressures of approximately 1 Torr,
reaction vessel through a pinhole in Al foil although present results are considered
into a differentially pumped quadrupole more reliable.
mass spectrometer, and signals were dis- We shall fit our rate expressions using
played on an electrometer or strip chart LH type expressions (7), i.e., for a unimo-
recorder. Sensitivities for various gases at lecular surface reaction A + B,
operating conditions were calibrated with
known gas mixtures. All species concen- ,. s kR(jA = kRKAPA
1 + KAPA’ (6)
trations could be measured readily except
for NH3 and Hz0 which adsorbed on the and for a second order surface reaction
walls of the mass spectrometer chamber A + B + products,
and exhibited time dependent signals.
Compositions were determined mainly
from the Nz, NO, and O2 signals, although
overall stoichiometric balances were or
checked occasionally. Data were generally
reproducible to within 510% ‘over long
periods and for different wires of the same
material. These are the standard forms usually as-
For most experiments all three wires sumed for these reactions with competitive
were mounted in the same reactor in order [Eq. (7)] or noncompetitive adsorption
to compare relative rates on the three [Eq. (S)] . These equations would be
metals. Wires which had been used many strictly valid only for Langmuir isotherms
hours under all conditions of temperature (s - 1 - (3) and for a homogeneous sur-
and composition were examined by scan- face. Further, all of the k’s should in gen-
ning electron microscopy. Pt and Rh were eral be functions of coverage and therefore
quite smooth and exhibited only grain of pressure.
 KINETICS OF NH3 OXIDATION 215

However, we shall show that our rate eters. However, the lumped parameters
expressions can be fit fairly well with ex- should be roughly representative of these
pressions of this form with all parameters quantities and should permit one to assess
assumed to be functions of temperature the validity of various rate expressions and
alone. The parameters certainly contain to compare behavior on different metals.
coverage and crystal plane dependences,
but we suggest that these may be effec- RESULTS AND DISCUSSION
tively lumped into constants which are
functions only of temperature. individual Reaction Rates
Such fits should only be given signifi- Decomposition of NH, and NO. Figure
cance if they reproduce data over a wide 1 shows the rate of N, production on Pt,
range of composition and temperature. Rh, and Pd from NH, at a pressure of
The present results are obtained between 0.115 Torr. The rates do not appear to go
10 and 90% of each reactant and between through maxima as was previously re-
200 and 1400°C. The major emphasis in ported for Pt at higher pressures. It is pos-
this work will be on the temperature de- sible that the rates would attain maxima at
pendences predicted by these equations. If higher temperatures or that EH > ENH3, in
the reaction rate constant is of the form, which case no maximum would be pre-
dicted.
kR= kORexp (-EdRT) , (9) NO decomposition is even slower than
and the adsorption constant is of the form, NH, decomposition, and it was difficult to
obtain accurate rate measurements with
KA = GA exp (EART), (10) the residence times of our experiments. On
then it is easy to show (II) that Eqs. Pt and Pd at an NO pressure of 1.0 Torr
(6)-(g) predict rate maxima with tempera- the rate of Nz formation was - 1 x lo+
ture if the heats of adsorption EA and EB moles cme2 set-’ for temperatures
are sufficiently large. between 1200 and 1500°C. This is over
We fit our experimental rates to expres- 100 times slower than either bimolecular
sions similar to Eqs. (6)-( 10) with a non- reaction involving NH,.
linear least squares program which used a
Gauss-Newton procedure to search for
optimum choices of parameters to fit data.
Points were weighted equally, and no
curves or particular points were excluded 6-
from the fitting program. Our parameters
are those obtained directly by least
squares fits to data; they are not adjusted
to agree with expected magnitudes or even
signs of any parameters. There are mul- c
tiple extrema in a nonlinear fit of this type, 0 3-
X
but for fits of a single reaction the “best”
fits were always independent of initial val- Lz” 2-
ues used in search routines.
We emphasize that the preexponential I -

factors and activation energies obtained by 0 ,P , I 1 I 1

such a procedure should not be expected 400 6ca 600 Dcxl 1200 1400 1600
to be interpretable quantitatively in terms Temperature PC)
of actual reaction or adsorption param- FIGURE 1.
 PIGNET AND SCHMIDT

16 -

60% NH3

i-l
c ‘4-
P
; IZ-

x
E IO-
Y
i
50 0-
2
.-0’ 6 _
‘i
0
a0 400 600
5
800 1000 1200 1400 z0 4-
TEMPERATURE PC)
”
FIGURE 2.
4 2-
On Rh the rate was as high as 1 X lo-’
moles cme2 see-’ at 0.115 Torr, but the Ol-
200
rate did not go through a maximum up to Temperature PC)
1600°C. Rates of NO decomposition on FIGURE 4.
Rh could be fit by the expression
r,, (moles/cm2 set Torr These data were fit assuming competi-
tive adsorption of NO and NHB, Eq. (7).
= 2.2 X 10e3PNGexp(-w). Calculated rates at these compositions are
shown in Figs. 3, 5 and 7, and corre-
The higher rate of this reaction is probably sponding rate parameters are given in
due to the stronger adsorption of NO on Table 1. For Pt and Pd peak temperatures
Rh. As considered below, similar behavior are lower for low NH, concentrations, in-
is also observed in bimolecular reactions dicating that NH3 adsorption was stronger
involving NO. than NO adsorption. Therefore fits for
Reaction between NH, and NO. Fig- these metals were made assuming KNO to
ures 2, 4 and 6 show measured rates of
NO consumption versus temperature on 16 -
Pt, Rh, and Pd, respectively. Data were
obtained with a constant total pressure of 14 -
0.115 Tot-r, and NH, initial compositions
are indicated in Figs. 2, 4 and 6.

O200 400 600 000 1000 1200 1400

Temperature PC)
FIGURE 5.
 KINETICS OF NH, OXIDATION 217

16 TABLE 1
r RATE PARAMETERS FOR NO + NH3 REACTIONS

ER &HI
(kc& (kcali
Metal k.8 mole) L\H, mole) k,w Em

Pt 5.3 x 102 28 1.3 x 10-10 45 -

Rh 2.8 x 102 9.5 2.4 x 10-x 34 - -

Pd 3.2 x 1O-4 15.3 2.8 x lo-‘0 11.1 2.6 x lo-$ 20.7

n Calculated by fitting rate data of Figs. 2.4. and 6 to Eqs. (7). (9), and
(LO). Preexponential factors in units of moles/cm2 set Torr.

concentrations; otherwise there are no sys-

tematic differences.
20%
0 Ammonia Oxidation Kinetics
0 200 400 600 800 1000 1200
Temperature PC) Figures, 8, 9, and 10 show rates of NO
FIGURE 6. and Nz formation versus temperature on
Pt, Rh, and Pd, respectively. The total
be zero (a four parameter fit). In contrast pressure in the NHJO, feed was 0.115
to Pt and Pd, the peak temperatures on Rh Torr in all experiments, and the initial per-
are higher for low NH3 concentrations, in- centages of NH, are indicated in Figs.
dicating that NO is more strongly ad- 8-10. These curves are generally similar to
sorbed than NH3. A six parameter fit in- those reported previously on Pt at a total
cluding KNOand KNHswas found to give pressure of 0.75 Torr, except for some ad-
best agreement with the broader rate ditional structure in the previous curves.
curves on Rh. The major discrepancies We now know that the apparently complex
between experimental and calculated behavior of some of the previous curves
curves on all metals occur at low NH, was caused by inaccurate measurement of
NO and Nz concentrations in the mass
‘-0-
16
r spectrometer.
The only region in which the accuracy
of measurements was poor was between
200 and 400°C. Here, as in the NO + NH3
reaction, the onset of N, production was
frequently too abrupt to follow closely
and, while results appeared to be identical
for increasing and decreasing tempera-
tures, it was not possible to fill in complete
u 6- curves. Data in this region are therefore
z” shown as essentially discontinuous rises in
4-
most cases. There are several possible
2-
causes for such behavior. First, it is pos-
sible to have multiple steady states in an
0, I isothermal well-mixed reactor with Lang-
0 200 400 600 800 1000 1200 muir-Hinshelwood kinetics (13). This
Temperature PC) would lead to truly discontinuous changes
FIGURE 7. in steady state rates with temperature,
218 PIGNET AND SCHMIDT

6
“0
X
E
$! 6
E
Y
b:
54
E

.-s
t2
4
e
a

z” 0
0 200 400 600 600 loo0 1200 1400

-
200
Temperatue PC)
FIGURE 8.

although we caclulate that steady states Eqs. (l)-(4), even if one ignores product
are probably unique for our kinetics and species other than NO and N,. Also, be-
residence times. A second possibility is cause in a sequence of reactions a product
multiple states caused by heat generated species may continue to react without de-
by the reaction. In most of these experi- sorption and readsorption, surface concen-
ments we did not use an automatic temper- trations and partial pressures may not be
ature controller, and wire temperatures related through adsorption isotherms for
were difficult to control precisely in this such species. Nevertheless, we shall at-
region because of reaction heating. A third tempt to see whether simple reaction rate
possibility is that contaminants may be on expressions are useful in analyzing the
the surface at low temperatures which de- rather complex behavior apparent in Figs.
sorb or are oxidized at these temperatures. 8-l 1 and also to obtain working expres-
The reversibility and reproducibility of our sions for prediction of reaction selectiv-
data argue against these effects, but they ities at high pressures.
cannot be completely excluded from A qualitative examination of the oxida-
present results. tion rates and the individual unimolecular
We next attempt to fit the results of and bimolecular reaction rates shows that
Figs. 8-l 1 to LH rate expressions. We the series-parallel processes [Eqs. (l)-(4)]
note first that this is a very complex seem to describe the NH, oxidation
system of reactions. At least four kinet- process fairly well. Figure 12 shows rates
ically independent reactions are involved. of NO and N2 production in NH3 oxida-
 KINETICS OF NH, OXIDATION 219

200 400 600 800 1000 1200 1400 1600

s
x 8r I

Temperoiure PC)

FIGURE 9.
P 0 200 400 600 600 1000 1200 1400 1600
4 Temveroture f*C)

FIGURE Il.

tion (solid curves) and in the individual

reactions (broken lines). The curves shown
here are from experimental data at a total
pressure of 0.115 Torr and for reactant
compositions of 50% NH3 in NO + NH,
and O2 + NH3 mixtures. The individual
reaction rates are measured at different
compositions than those in the oxidation
200 400 600 800 1000 1200
reactions and therefore comparisons
should be expected to have only qualita-
tive significance. For example, all single
reactions were measured in the complete
absence of 02, and consequently influ-
ences of adsorbed oxygen on reaction
rates would not be apparent in these mea-
surements. In spite of these differences,
the production of NZ in NH3 oxidation is
clearly indicated as arising mostly from the
NO + NH3 reaction at low temperatures
(T < 1000°K) and from the decomposi-
tion of NH, at high temperatures
( T > 1200OK). Unimolecular decomposi-
Temperature (’ C) tion of NO appears to be quite unimpor-
FIGURE 10. tant except on Rh at high temperatures.
220 PIGNET AND SCHMIDT

,---.
lo Rh NO + NH,
\ ,/’
F

r x10’

Cm
Pet>

T PC)
FIGURE 12.

Several more qualitative features of the klPNH3Pg2

rNo = (1 + KNHaPNe) (1 + Kg2Pg2) (12)
oxidation kinetics are evident from the
general behavior at different compositions. and
The onset of N$ production occurs at
lower temperature at lower NH, composi- rN2 =
k&&o
tions. The rate is therefore of negative ( 1 + &Pa + &xJ'NH~~

order with respect to NHs in this region, k&m

+ 1 + K4Po, + K~.&E, ’
(13)
and this suggests competitive adsorption of
NHs and some species, in agreement with
the kinetics previously shown to fit the The term K4P02 has been added in the
NO + NH3 reaction [Eq. (7)]. On the unimolecular and bimolecular reactions of
other hand the NO production rate from NH3 which were not present in previmus
NH3 and O2 (Figs. 8-10) is independent of fits of the reactions individually because
the NH3 composition in excess O2 at low now the reactions occur in a high partial
temperature. Zeroth order kinetics at low pressure of 02. These are obviously
temperatures suggest noncompetitive ad- lumped parameter equations. It was not
sorption of NH3 and oxygen [ Eq. (S)] . possible to fit the more rigorous equations
Further, one expects that oxygen should to the experimental data because of insta-
adsorb dissociatively to react, and there- bilities resulting from the larger number of
fore its isotherm should be of the form adjustable parameters. The 14 parameters
in Eqs. (12) and (13) should contain the es-
e. = Kg2Pg2/( 1 + Kg2Pg2). ( 11)
sential features of the rate process as
From these simple arguments one would sketched above.
expect that the rates of NO and N2 Figure 12 shows calculated rates of NO
production might be represented by ex- and NP formation versus temperature on
pressions of the form Pt using Eqs. (12) and (13) to fit the data
 KINETICS OF NH, OXIDATION 221

of Fig. 8. The rate parameters in Eqs. (12) simply as being lower or higher on a par-
and (13) were titular metal but must be described

3.4 X 10m8exp(+21,700/RT)PNHIPif
(14)
rNo= [ 1 + 8 X low2 exp(+4400/RT) Pg2] [ 1 i- 1.6 X lop3 exp(+25,5OO/RT) PNHa]
and
4.2 X lo-’ exp(+26,000/RT) PNOPNH3
rNz= [ 1 + 5.4 x 10e2exp( 10,900/RT)Po, + 4.5 X 10m6exp(20,900/RTIPNH312
2.2 X lop6 eXp(-570/RT)PNH,
(15)
+ 1 + 5.4 x low2 exp(+10,900/RT)P, + 4.5 X 10m6eXp(20,900/RT)PNH,’

with pressures in Torr and rates in units of through the dependences on temperature.
moles cmp2 see-‘. Figure 12 shows that, while all reactions
In formulating reaction rates in terms of have comparable rates at their maxima,
Langmuir-Hinshelwood kinetics one as- these occur at temperatures which differ
sumes implicitly that adsorption and de- by several hundred degrees. The rates and
sorption steps are much more rapid than selectivity to be expected at a particular
reaction so that coverages and pressures catalyst temperature and gas composition
may be related by an adsorption isotherm. are therefore dominated by the tempera-
Our reaction rates were always less than ture dependences of various rates. The
1O-2 of the reactant fluxes to the surface, dependences on partial pressure are also
and we have used model calculations to strongly influenced by the shifts in rate
show that LH expressions should be fairly maxima with temperature.
accurate even when adsorption and de- The differences between the individual
sorption rates are comparable to reaction bimolecular reactions appear to be caused
rates. mainly by the stronger bonding of NO on
However, in the case of N2 formation Rh compared to Pt and Pd. This produces
from NO, a reaction intermediate, this a higher rate of NO decomposition on Rh
approximation may be invalid because it is and causes the bimolecular reaction to be
not necessary that NO first evaporate and shifted to higher temperature. At low tem-
then readsorb to react with NH3. We fit peratures (T < 700°C) Rh is a very poor
the experimental rate data on Pt to a catalyst for the NO + NH3 reaction while
model which included this possibility by at high temperatures (T > 900°C) Pt is a
solving the appropriate gas and adsorption poor catalyst but Rh is a good one. The
mass balance equations simultaneously strength of NO adsorption compared to
(II). The least squares program invariably NH3 (Figs. 2, 4 and 6) results in quite dif-
set the parameters for NO evaporation and ferent reaction rate pressure dependences
adsorption equal to their lower bounds. for these gases on the three metals, and
However, this model gave fits to the these pressure dependences shift com-
experimental data which were inferior to pletely within a few hundred degrees as
those using Eqs. (12) and (13). the rates pass through their maxima. This
difference can be seen from the “heats of
Comparisons Between Metals adsorption,” the EN& and EN0quantities in
In this section we discuss qualitatively Table 1, but, while these are frequently
the differences in various reaction rates in similar to reported literature values of
terms of adsorption parameters on the heats of adsorption, we do not regard
three metals. Rates cannot be described either our values or literature values to be
222 PIGNET AND SCHMIDT

sufficiently accurate to justify detailed (2-4) the maximum selectivity of 9599%

comparisons. The preexponential factors is attained with -10% NH, in air and
in the adsorption constants Kti are ratios gauze temperatures of - 900°C. This is ob-
of sticking coefficients to desorption preex- served in industrial operation at 1-8 as
ponential factors vti, well as in our laboratory studies in
NH,/air mixtures at 1 atm (I). Under
KM= soi these conditions the reactions are mass
vti(2rrmkT)“*’ (16)
transfer controlled in the sense that NH3
and, if the initial sticking were unity and v,, consumption can be calculated from mass
were 1013set-‘, then the “normal” value transfer coefficients to the surface of the Pt
of Kti would be - lo-’ Torr-‘. Some of the gauzes (14). However, since selectivity is
adsorption constants in Table 1 are close the important parameter, the relative rates
to this value, but the results and the fits do of production of NO and N, are still gov-
not have sufficient accuracy to permit one erned by the kinetics of the individual
to associate deviations with low s,, or with reactions even though essentially all NH,
anomalous preexponential factors. is consumed. In this section we shall at-
The interpretation of the oxidation rates tempt to use the rate parameters obtained
is even more difficult because of the in the 0.1 Torr reaction to calculate the
number of independent reactions. How- selectivities at high pressure as functions
ever it seems clear that the maximum in of composition, gauze temperature, and
N2 production is associated with the contact time.
NH, + NO reaction at low temperatures A simple model of the gauze catalyst is
and with NH3 decomposition (along with that of an isothermal stirred tank reactor
NO decomposition on Rh) at high temper- with a mass transfer coefficient calculated
atures. The selectivity of oxidation to NO from the flow characteristics over the
is thus determined by the temperatures at gauze,
which the three rates are high. On all
vi=F[(Pf-PPS], (17)
metals there is a temperature “window”
between 700 and 1000°C where the selec- with Pf the pressure of the limiting reac-
tivity to NO is high, but the maximum tant far from the surface and Pt the pres-
selectivity is larger for Pt (94% conversion sure near the surface. A typical industrial
to NO) than for Pd (90% conversion to throughput at 1 atm of 10% NH3 in air is
NO) or Rh (79% conversion to NO). The 100 lb NH, day-’ oz catalyst-’ with cata-
highest selectivity on Pd occurs at the lyst an 80 mesh gauze of 0.003 in. diame-
highest temperatures where metal evapora- ter wire. From this information we es-
tion is significant, so that at these pres- timated a mass transfer coefficient
sures Pt is by far the best catalyst for NO F = 5.8 X 1O-8 moles cm-* see-’ Tort-‘.
production. The reason that a high selec- This is smaller than the value obtained
tivity is not observed on Rh is because the from the smooth wire area by a factor of
NO + NH, reaction maximum almost - 10, which is reasonable because of sur-
overlaps that for NH3 oxidation, again pre- face roughness in the industrial gauze.
sumably because of strong adsorption of Conversions were calculated by solving
NO on Rh. Eqs. (14), (15) and (17) simultaneously by
eliminating p; between then. Figure 13
Prediction of High
shows the calculated selectivity of conver-
Pressure Selectivity
sion of NH3 to NO on Pt versus tempera-
In the high pressure production of NO ture using the stirred tank model and the
from NH3 and air over Pt-10% Rh gauzes kinetic parameters of Eqs. (14) and (15).
 KINETICS OF NH3 OXIDATION 223

Temperature (‘32

FIGURE 13.
0 .2 .4 .6 .8 1.0
Z/L

FIGURE 14.
This reproduces fairly well the observed
high selectivity up to 10% NH3 and the
rapid falloff with feed composition as one
approaches the stoichiometric ratios single stirred tank reactor because the ini-
( 14.1% for conversion to NO and 2 1% to tial product NO can react with unreacted
NJ. The maximum selectivity at 10% NH, further down the gauze. This explains
NH3 is obtained at - 900°C exactly that the contact time effect, known since An-
observed to be optimum in the industrial drussow’s early experiments, in which the
reactor. It should be noted, however, that NO yield decreases with increasing con-
the selectivity is not a sensitive test of tact time. The profiles shown in Fig. 14 are
choices of all rate parameters because in agreement with the estimates that in
these merely determine the rate limiting industrial reactors the reaction occurs
concentrations versus temperature. At mainly in the first one-half of the bed
high temperatures (T > 1OOoOC) the selec- length (2).
tivity is observed to fall rather rapidly,
while a fairly gradual decrease is calcu- SUMMARY
lated from this model. This is probably The reaction mechanism of the catalytic
caused by homogeneous boundary layer oxidation of ammonia on platinum has
reactions in the reactor which consume been the object of an enormous number of
NH, and produce NZ. There have been a experimental studies over the past 50 yr,
number of reports of heterogeneously in- yet it remains uncertain. We feel that the
duced chain reactions in the gas phase for most useful kinetic model of the reaction is
this reaction at high temperatures (IS). one that avoids detailed “atomistics” and
To test the sensitivity of the calculated describes the overall reaction in terms of
selectivity on the reactor model, we also the relatively simple series-parallel mecha-
made calculations assuming 10 stirred tank nism, Eqs. (l)-(4). The three individual
reactors in series, each with a reactor pa- surface reaction steps may then be mod-
rameter F ten times that assumed pre- eled by one of the various, classical Lang-
viously. Figure 14 shows the calculated muir-Hinshelwood kinetic forms. Our ob-
conversion to NO versus position in the jective in this study was to obtain, com-
gauze z/L for several gauze temperatures. pare, and analyze steady state rate data at
It is interesting that this model gives lower low pressure for all of the individual reac-
yields of NO than does the corresponding tion steps in the same reactor geometry, on
224 PIGNET AND SCHMIDT

the same catalyst surface, and over the creasing the maximum attainable NO/N,
widest possible temperature range selectivity. It should be noted, though, that
(200-1500°C). This last was most impor- in O2 excess the low temperature
tant for demonstrating the LH character of NO-NH, reaction is suppressed to a much
the kinetics. Finally, as regards activity greater extent on Pd and Rh than on Pt.
and selectivity, platinum was compared Most of the individual reaction rate data
with palladium and rhodium, two metals was subjected to analysis in Langmuir-
often used in similar reactions. Hinshelwood terms -the unimolecular LH
The NO/N, selectivity in the oxidation form for NH, decomposition on Pt and the
reaction on Pt may be described qualita- LH form of competitive bimolecular reac-
tively as follows. The surface reaction tion in the NO-NH, reaction on all three
forming NO from NH3 and O2 is metals. The kinetic constants of Table 1
extremely rapid and proceeds measurably should not, however, be taken literally as
even below 200°C. In the low temperature the adsorption or reaction parameters that
regime, though, the NO formed may be the model implies because all the data fits
bound rather strongly to the surface and are at best only semiquantitative. Also, we
may readily react with NH, to form N,. find certain of the fitted constants changing
With increasing temperature more of the from one reaction to another when they
NO formed escapes the surface and the should correspond, in theory, to the same
selectivity begins to shift toward net NO adsorption parameter. That is, the addition
production. This is a key feature of LH of NO to the system appears to alter the
kinetics -decreasing surface reaction rate adsorption characteristics of NH3 as deter-
due to rapidly decreasing surface coverage mined in the NH, decomposition reaction,
of reactant. and the addition of O2 in the overall oxida-
At some temperature (,> 800(C) virtually tion reaction alters them even more. The
all of the NO formed escapes (in O2 excess change here is noticeable in the qualitative
conditions), but the peak NO rates shift to nature of the data. On Pt and Pd the low
higher temperatures with increasing NH, temperature NO-NH3 reaction proceeds
pressure as NH, hinders the escape. It at lower temperatures when oxygen is
also appears that the NH,-0, reaction to present, the temperature of incipient reac-
NO itself becomes adsorption limited at tion shifting continuously from 200 to
high temperatures and the overall reaction 500°C (the latter the same as in the pure
then degrades to NH, decomposition, the NO-NH, reaction) with decreasing O2
selectivity shifting back to N,. concentration.
Palladium and rhodium were found to
have roughly the same peak activity as REFERENCES
platinum, although the effective palladium 1. Pignet,T., and Schmidt, L. D., Chem. Eng. Sci.
surface area may have been increased by 29, 1123 (1974).
several times over its geometric area and 2. Gillespie, G. R., and Kenson, R. E., Chem.
Tech., Oct., (1974).
the temperature regime of activity for rho- 3. Chilton, T. H., “Strong Water.” M.I.T. Press,
dium was much higher than that for the Cambridge, MA, 1968.
other two meals. A distinctive character- 4. Handforth, S. L., and Tilley, J. N., Ind. Eng.
istic of the NO/N, selectivity on Pt is the Chem. 26, 1287 (1934).
relatively wide separation in temperature 5. Andrussow, L., and Bodenstein, M., Z. Angew.
Chem. 39, 321 (1926); 40, 174 (1927); 41,
of the N, and NO peaks while there is
205, 207, 262 (1968).
much more overlap for Pd and Rh. For Pd 6. Zawadski, V., Rocz. Chem. 22, 220 (1948).
especially it appears that the NO- and N,- 7. Bond, G. C., “Catalysis by Metals.” Academic
forming reactions peak together, de- Press, New York, 1962.
 KINETICS OF NH3 OXIDATION 225

8. Fogel, Y. M. et al., Kinet. Katal. 5, 127 (1964). 12. Pignet, T., Schmidt, L. D., and Jarvis, N. L., J.
9. Nutt, C. W., and Kapur, S., Nature (London) Catal. 31, 145 (1973).
220, 697 (1968); 224, 169 (1969). 13. Matsuura, T., and Kato, M., Chem. Eng. Sci. 22,
10. Hori, G. K., and Schmidt, L. D., J. Catal. 38, 171 (1967).
335 (1975). 14. Nowak, E. J., Chem. Eng. Sci. 24, 421 (1969).
11. Pignet, T., PhD thesis, Univ. of Minnesota, 15. Vladov, D., Dokl. Akad. Nauk, SSSR 109, 561
1974. (1956).