A transient kinetic study of the mechanism of the NO + H2 reaction over Pt/SiO2 catalysts: 2. Characteristic features of SSITKA profiles

Article abstract of DOI:10.1006/jcat.1999.2567

SSITKA profiles obtained for the NO + H₂ reaction on a 5% Pt/SiO₂ catalyst at low temperatures are analysed in an attempt to gain insight into the reaction mechanism at the molecular level, specifically for the formation of N₂, N₂O, and NH₃. Three different analyses are introduced which give information about the nature of the reaction network and the final intermediates on the Pt surface under reaction conditions. These are (a) the initial distribution of isotopic molecules of product (IDIMP), (b) the temporal redistribution of isotopic molecules of product (TRIMP), and (c) the semilogarithmic plot of (a) versus time. The first two analyses (IDIMP and TRIMP) are discussed in terms of types of production of two atom labelled product molecules. We also introduce the integral characteristics of a reaction mechanism, i.e., the coefficient of percolation (C_PE) and the coefficient of production (C_PR). The conclusions, along with the experimental data, are used to develop a reaction mechanism for the formation of N₂, N₂O, and NH₃ from Pt catalysts in the NO + H₂ reaction.

Full text of DOI:10.1006/jcat.1999.2567

Journal of Catalysis 186, 362–372 (1999)
Article ID jcat.1999.2567, available online at http://www.idealibrary.com on
A Transient Kinetic Study of the Mechanism of the NO + H2 Reaction
over Pt/SiO2 Catalysts
2. Characteristic Features of SSITKA Profiles
A. A. Shestov,1 R. Burch, and J. A. Sullivan
Catalysis Research Centre, Chemistry Department, University of Reading, Whiteknights, Reading, RG6 6AD, United Kingdom
Received February 4, 1999; revised May 11, 1999; accepted May 11, 1999
intermediates in the formation of N2 and N2O. The previ-
ous paper (13) considered “classical” quantitative analysis
SSITKA profiles obtained for the NO + H2 reaction on a 5%
Pt/SiO2 catalyst at low temperatures are analysed in an attempt
to gain insight into the reaction mechanism at the molecular level,
specifically for the formation of N2, N2O, and NH3. Three differ-
ent analyses are introduced which give information about the na-
ture of the reaction network and the final intermediates on the Pt
surface under reaction conditions. These are (a) the initial distri-
bution of isotopic molecules of product (IDIMP), (b) the temporal
redistribution of isotopic molecules of product (TRIMP), and (c)
the semilogarithmic plot of ( ¯ ) versus time. The first two analyses
(IDIMP and TRIMP) are discussed in terms of “types of produc-
tion” of two atom labelled product molecules. We also introduce the
integral characteristics of a reaction mechanism, i.e., the coefficient
of percolation (CPE) and the coefficient of production (CPR). The
conclusions, along with the experimental data, are used to develop
a reaction mechanism for the formation of N2, N2O, and NH3 from
c
of SSITKA profiles and resulted in the generation of values
for such parameters as surface coverages and reactivity of
intermediates.
Since the development of the SSITKA technique, by
Happel (14), Bennett (15), and Biloen (16), there have been
many articles devoted to the numerical analysis of such pro-
files with the aim of generating information about reac-
tion mechanisms. These numerical analyses have generally
taken the form of modelling the transient data generated
from either continuously stirred tank (17, 18) or plug flow
(19) reactors. Comprehensive reviewsofthe technique have
been published (20, 21). However, it is generally the case
that most workers have analysed isotope transfer only, i.e.,
the fraction of isotopes within the components with time.
Distribution of isotopic molecules has been disregarded or
only partially considered without detailed qualitative anal-
ysis of experimental data (4).
Pt catalysts in the NO + H2 reaction.
1999 Academic Press
Key Words: NO reduction; SSITKA; modelling; reaction mecha-
nism; Pt catalysts.
In this paper we introduce and develop several methods
of analysing the data obtained from isotopic transients in or-
der to try to suggest an unambiguous reaction mechanism.
Some features of these methods originally come from clas-
sical isotopic exchange analysis in closed systems without
the occurrence of chemical changes in the system (22–29).
Specifically our approach will be for situations where a
reactant molecule (RAn) containing labelled atoms (where
there are two possible isotopes for the A element) is con-
verted, via intermediates (IAx), into a product molecule
that contains two such labelled atoms (PA2) for a steady
state reaction in a plug flow reactor, e.g., NO +H2 → N2/
N2O, NH3 +O2 → N2/ N2O, CH4 +O2 → C2H4/ C2H6,
O2 + Ph–CH3 → Ph–C O OH, where an asterisk repre-
sents a labelled atom. A full mathematical description of
the methodology used in the interpretation and modelling
of these results is presented elsewhere (30). Only the
essential details are given here.
INTRODUCTION
Steady state isotope transient kinetic analysis (SSITKA)
has previously been used to clarify the mechanisms of sev-
eral reaction types, e.g., Fischer–Tropsch synthesis (1), CO
hydrogenation (2), NH3 oxidation (3), oxidative coupling
of CH4 (OCM) (3, 4), methanol synthesis (5), CO oxida-
tion and CO/NO reaction (6, 7), NO/NH3 reactions (8, 9),
NO/CH4 (10) reactions, and NO/C3H6/O2 reactions (11, 12).
As a general rule the reactions which provide the clear-
est information are those in which the surface contains
large reservoirs of adsorbed intermediates which are slowly
transformed to products.
In the present work, SSITKA is used to study the
NO + H2 reaction over one such supported Pt catalyst (5%
Pt/SiO2) in an attempt investigate the nature of the surface
1 On leave from Boreskov Institute of Catalysis, Russian Academy of
Sciences, Pr. Akademika Lavrentieva, 5, Novosibirsk, 630090, Russia.
Three different analyses (and transformations) of the
experimental profiles for the NO + H2 reaction will be
362
0021-9517/99 $30.00
c
Copyright
1999 by Academic Press
All rights of reproduction in any form reserved.

THE NO + H2 REACTION OVER Pt/SiO2 CATALYSTS, 2
363
discussed in relation to the information they yield about in a total flow of 62 cm3 min 1. The catalyst was the same
the mechanism of the N2 and N2O forming reactions. These as used previously, i.e., 5% Pt/SiO2. Data analysis and the
analyses are (a) the initial distribution of isotopic molecules generation of the various functions was performed on a
of product (IDIMP), (b) the temporal redistribution of bench-top PC using a commercial spreadsheet program.
the isotopic molecules of product (TRIMP) (i.e., the func-
tion y(i), the deviation from statistical isotopic distribution
within the product molecules), and (c) a semilogarithmic
TYPES OF PRODUCTION OF TWO-ATOM
LABELLED MOLECULES
plot of the ¯ function (where ¯ represents the (1
) pro-
15
file in the case of a 14NO → NO switch and the profile in
In order to discuss the first and second transformations
14
the case of the reverse 15NO → NO switch) versus time. of the data, we have to first introduce the concept of “types
As detailed previously, is the fraction of heavy atoms in of production.” This involves differentiating steady state
the gas phase (13), e.g., for N2 can be calculated using the reaction mechanisms on the basis of the final reaction step
equation
leading to the generation of the product molecule (PA2)
from the reactant (RAn) and any final intermediates on the
surface (IAx or I0Ax) which go on to form PA2.
[15N2] + 0.5[15N14N]
(N2) =
.
[1]
[15N2] + [14N15N] + [14N2]
Detailed analysis (30) of the modes in which a molecule
of product (PA2) containing two “labellable” A atoms can
be formed from a labelled reactant molecule (RAn) and
intermediates IAx has shown that there are basically five
“pure” types of production of such molecules. These are
shown in Table 1 and briefly discussed below. In the fol-
lowing discussion, A represents a labelled atom (with two
possible isotopes), RAn represents a reactant containing
“n” such atoms, PA2 represents a product containing two
such A atoms, and IAx represents an intermediate contain-
ing “x” A atoms. R, P, and I are parts of the molecule that
do not contain any A atoms (or contain A atoms that re-
main unreactive). We willconsider a singlylabelled reactant
(RA) and either singlyor doublylabelled intermediates(IA
and IA2) to simplify the types of production which we will
discuss.
Plot (a) is a “snapshot” of the initial redistribution imme-
diately following a 14NO + H2 → NO + H2 switch while
15
plots (b) and (c) are presented as a function of time. Note
that IDIMP has been used previously for studying re-
actions such as benzene hydrogenation in open systems
(31, 32).
Profiles (a) and (b) are discussed in terms of the concept
of “types of production” of two atom labelled molecules
(PA2) as discussed later. The first set of plots are also
used to introduce the concept of the integral character-
istics of any network mechanism of the form RAn →
IAx
→ PA2, i.e., where a labelled reactant (RAn) pro-
gresses through one or more surface intermediates (IAx) to
yield a 2-atom labelled molecule (PA2).
The first type of production (labelled 0) is one in which
the surface plays no discernible role and the production of
PA2 involves no surface A atoms, i.e.,
EXPERIMENTAL
The SSITKA switches are performed at 60 C under sim-
ilar conditions to those used in the previous paper (13). The
conditions are altered somewhat to enhance the resolution
and to ensure that the pressure changes on switching are
minimised. The gas composition used was 3% NO, 3% H2
type 0 2RA → PA2 .
[2]
The second type (labelled 1) involves, for example, an
impact mechanism in which a molecule of RA interacts with
TABLE 1
Designation of the Five Possible Types of Production of a Product Molecule PA2 from a Reactant Molecule RA
and Reaction Intermediates (IA and I0A), Showing the Final Reaction Steps
Temporal
redistribution
yi = xi xieq
Initial
Production
coefficient
CPR
Percolation
coefficient
CPE
Type of
production
distribution
{x00, x10, x20}
Mechanism
0
1
2a
2b
2c
2RA → PA2
RA + IA → PA2
2IA → PA2
y1 = y2 = 0
y1 > 0, y2 < 0
y1 = y2 = 0
y1 > 0, y2 < 0
y1 < 0, y2 > 0
{0, 0, 1}
{0, 1, 0}
{1, 0, 0}
{1, 0, 0}
{1, 0, 0}
0
1
2
2
2
2
1
0
0
0
IA + I0A → PA2
IA2 → PA2
Note. Also shown are the predicted temporal redistributions of the isotopically labelled molecules ((PA A and PAA ))
the expected initial isotopic distribution following the same switch, and the expected production (CPR) and percolation
(CPE) coefficients following a step change RA → RA (see text for details).

364
SHESTOV, BURCH, AND SULLIVAN
a surface intermediate which gives one A atom to produce with different sites on the surface. In these cases the situa-
a PA2 molecule. This involves one surface A atom in the tion becomes more complex, and these will not be consid-
elementary act of production of PA2, i.e.,
ered further here (see (30)). For many experimental cases
regarding molecules such as N2 and N2O these exchange
reactions are not important.
type 1 RA + IA → PA2 .
[3]
In terms of the types of productions we can also say that
the overall rate of production of the two-atom molecule
equals the sum of the rates of the five types of production
(in a normalised form at a value of 1),
The third, fourth, and fifth methods of formation of PA2
all involve two surface A atoms in the act of production.
These are thus labelled “2” and are differentiated on the
basis of the types of 2-atom interactions. Production type
2a (a for alone) represents the formation of a PA2 molecule
from two equivalent species on the surface both of which
give one A atom to the final product, i.e.,
+
+
+
+
= 1,
[7]
0
1
2a
2b
2c
where
represents the contribution of type k (k = 0, 1,
k
2a, 2b, 2c), with a rate Rk, in the overall production (with a
rate R6); therefore, k = Rk/R6. The strict method of cal-
type 2a 2IA → PA2 .
[4]
culating
on the basis of isotopic kinetic equations will
Production type 2b (b for binary) represents the forma-
tion of PA2 from the interaction of two different species
on the surface, both of which give one A atom to the final
product, i.e.,
k
be presented later (30). A second, less accurate but simpler
method, based on IDIMP, will be discussed below.
CHARACTERISTIC FEATURES OF SSITKA PROFILES
type 2b IA + I0A → PA2 .
[5]
This section will be divided into the three subsections.
Each section will be introduced separately. The results will
be discussed, both in general and more specifically for our
experimental system, in terms of the information yielded
about the nature of the surface intermediates and the reac-
tion network.
The final type of production of PA2 is type 2c (c for cou-
pled) and this involves the presence of a “coupled” inter-
mediate on the surface which contains, as a minimum, two
A atoms. Both atoms that go on to form PA2 originate from
the same species, i.e.,
type 2c IA2 → PA2 .
[6]
a. Initial Distribution of Isotopic Molecules of Product
(IDIMP) and Integral Characteristics
of the Network Mechanism
These different types of production lead to different re-
sponses in the above-mentioned first and second features
of the obtained profiles, i.e., initial distribution of isotopic
molecules of product (IDIMP) and temporal redistribution
of isotopic molecules of product (TRIMP). The expected
outcomes of the above-mentioned transformations from
any “pure” type of production (0, 1, 2a, 2b, or 2c) are also
detailed in Table 1 and are further discussed below. It must
be remembered that these formalisms do not necessarily
represent actual reactions but rather just “modes” of reac-
tion and that the action of any type can lead to different
profiles which depend on factors such as surface concen-
trations, rates, and reversibility of steps. However, the tem-
poral redistribution and initial distribution plots developed
for every type of production must have the same qualitative
individual features as are discussed below.
Here we will re-introduce and revise a method of anal-
ysis previously proposed by Kemball (22) for the analysis
of isotopic heteroexchange CnHm–D2 exchange and also
introduce the integral characteristics of a network mecha-
nism.
If we consider the overall production of an isotopic prod-
uct molecule containing two “labellable” atoms following
the RA + () → RA + () switch within the SSITKA analysis,
there are normalised connections for the initial distribution
of the isotopic molecules of product, e.g.,
x00 + x10 + x20 = 1
(where xi0 represents the initial fraction of isotopic mole-
[8]
It must also be noted here that for a considered type of cules which contain i heavy atoms, e.g., 14N2 15Ni). The
i
production the atomicity of the surface intermediates (IAx) overall production of N2 remains constant. Therefore, the
is unimportant and the individual features of IDIMP and sum of the fractions of all isotopic molecules must, at all
the qualitative features of TRIMP will remain the same; times, equal a normalised value of 1.
e.g., IA2 + I0A2 → PA2 + IA + I0A also represents a type 2b
The initial isotopic product distribution following the
production and IA3 → PA2 + IA represents type 2c produc- RA + () → R A + () switch (in our case the 14NO +
15
tion and will give the corresponding IDIMP and qualitative H2 → NO + H2) is different for some of the types of pro-
TRIMP features. Additionally, we must allow for isotopic duction ofPA2. Type 0resultsin the initialdistribution being
exchange of the product molecules (through adsorption/ fully doubly labelled, while type 1 production would lead
desorption processes) with either the final intermediates or to fully singly labelled, and all the type 2 productions (2a,

THE NO + H2 REACTION OVER Pt/SiO2 CATALYSTS, 2
365
2b, and 2c) would lead to unlabelled product molecules in
the initial isotopic distribution (see Table 1).
There are connections between the initial isotopic dis-
tribution and the contributions of the different types of
production. If the probability (or contribution) of type of
production k is
and the initial isotopic distribution is
k
represented as {x00, x10, x20}, then it can be shown that
x00 =
x10 =
x20 =
+
+
[9]
[10]
[11]
2a
2b
2c
1
,
0
that isto saythat the initialdistribution givesdirect informa-
tion about the contributions of various types of production
to the overall production of PA2.
Obviously these are theoretical calculations, and it can
be shown (30) that the imposition of an experimental
switch rather than an ideal switch from 14NO to 15NO does
change this distribution somewhat. In the specific case of
the NO + H2 reaction over Pt/SiO2 this makes any IDIMP
analysis of the N2O distribution more prone to errors as this
FIG. 1. Initial distribution of isotopic molecules of product (IDIMP)
following the 14NO + H2 → 15NO + H2 switch over 5% Pt/SiO2 at 60 C.
The relative productions of each isotopic species (14N14N, 14N15N and
15N15N, ᭿, as well as 14N2O, 14N15NO (15N14NO) and 15N2O, ᮀ) are repre-
is very rapid relative to the N2 redistribution and thus more sented.
prone to effects of the imposed “experimental” switch.
production of N2O is relatively rapid and that the choice of
time for the “initial” distribution might be too far from the
true initial distribution to be of any real value.
Generally then, with respect to the gaseous holdup of the
system, the “shorter” the holdup the more accurate the
IDIMP.
In the case of the initial distribution within the N2
molecules it can be seen that the production of the doubly
labelled 15N2 molecule is hardly seen at all while the produc-
tion of the mixed labelled molecule is about twice as preva-
lent as that of the unlabelled one {0.32 0.05, 0.66 0.05,
0.01 0.05}. In terms of the equations presented above, this
can be written as
IDIMP plots are constructed, for our experimental sys-
tem, by taking the normalised values for all the isotopic
molecules of N2O and N2 at a short time following the
isotopic switch. Obviously, for experimental reasons, it is
not possible to get readings directly after the switch (due
to perturbations in the gas stream) and thus readings are
taken approximately 3 s after the beginning of the switch.
At this time the Ar has reached a normalised value of 1,
the mixed-labelled products (14N15N and 14N15NO) have
passed through maxima, and the product redistribution is
in process. It should be emphasised here that it follows from
the mathematical modelling (30) that taking the readings as
these mixed species are at their maxima can most accurately
x00 =
x10 =
x20 =
+
= 0.7
= 0.
+
= 0.3
[12]
[13]
[14]
2a
1
2b
2c
0
This indicates that type 1 production operates here (as
compensate for the “experimental nature” of the switch this produces the initial amount of mixed N2) as does some
(and its effects on the accuracy of the IDIMP analysis).
Figure 1 shows these plots for N2 and N2O. The results
type 2 (a, b, or c), as this leads to the unlabelled product.
Type 1 production can occur through an impact mech-
are presented as bar charts showing the normalised produc- anism in which gaseous NO interacts with a species con-
tion on the ordinate and the number of labelled atoms on taining one N atom on the surface. Another possibility for
the abscissa. In the case of N2O production it is seen that this type is that the reaction might involve the interaction
the initial distribution of the isotopic labels within the N2O of a weakly adsorbed, pre-adsorbed or physisorbed NO
molecules is {0.1, 0.1, 0.8}; i.e., almost all of the N2O be- molecule with a species containing one N atom. This would
ing formed is of the doubly labelled variety. From Eq. [10] be equivalent to a situation where the reacting NO is in the
this indicates a high occurrence of the mode 0 type of pro- gaseous state, as the fraction of the labelled pre-adsorbed
duction. This result is in contrast with the deviation from state would equal that of the gas phase given a fast equi-
equilibrium result (presented later) in which it is seen that librium between the two. One possible way to envisage this
mode 2c is prevalent, and it would be expected that a high is through the interaction of an NO “pre-adsorbed” state
level of unlabelled N2O would be produced from the initial with some reduced N species on the surface (NHx) where
redistribution. However, it must be remembered that the x = 0 or 1. Theoretical considerations of where it is possible

366
SHESTOV, BURCH, AND SULLIVAN
to distinguish between these situations are discussed else- where,
where (30).
2
2!
=
for i = 0, 1, 2.
[19]
Integral characteristics of surface participation in the pro-
i
(2 i)!i!
duction of PA2. We can now introduce two integral char-
The most useful probe function for the extent of isotopic
acteristics which reflect the degrees of surface intermediate
(or lattice) participation in the formation of doubly labelled
product molecules (PA2) from labelled molecules (RAn)
and their connection with the above-introduced initial dis-
tribution of isotopic molecules of product.
equilibrium in an flow system is the deviation (yi ) of the
14
fraction of isotopic molecules
N
15Ni (xi ) from their
(2 i)
equilibrium fraction xieq. This function was first used by
Muzykantov et al. (24) for the analysis of isotopic exchange
of O2 in a closed system.
First is the coefficient of surface production (CPR)—this is
the average number of surface A atoms which participate
in one act of production (desorption) of a molecule PA2.
Second is the coefficient of percolation (CPE), which is the
average number of A atoms of the reactant RA which par-
ticipate in one elementary act of formation of the product
molecule PA2.
yi = xi xieq for i = 0, 1, 2.
For example, for the 14N15N molecule, which contains one
[20]
heavy atom,
y1 = x1 2 (1
).
[21]
When the isotopic molecules PA2 Ai are statistically
i
In the production ofthe molecule PA2 there isthe obvious
connection
mixed, then yi = 0 for i = 0, 1, 2. TRIMP therefore shows
the yi variable for the isotopic molecules of product as a
function of time following an isotopic step change, e.g., in
CPR + CPE = 2.
[15]
15
our case the 14NO/H2 → NO/H2 switch. For a complete
With regard to the five types of production, the percolation
and surface production coefficients for each type of pro-
duction are presented in Table 1. These are derived using
IDIMP and the simple equations
description of the composition of isotopic molecules, e.g.,
PA2 Ai , we only need to examine one isotopic species,
i
i.e., PA2, PA A, or P A2, in conjunction with the profile.
This is so because the three deviations from equilibrium (y0,
y1, and y2) are related (at any time (t)) by the relationships
CPR = 2x00 + x10
CPE = x10 + 2x20.
[16]
[17]
y0(t) + y1(t) + y2(t) 0, y0(t) y2(t), and
Here, the values x00, x10, and x20 represent the initial nor-
malised fractions of the fully unlabelled, mixed labelled,
and fully labelled initial product species, respectively.
In our case (the NO + H2 reaction over 5% Pt/SiO2 at
60 C under steady state conditions) for N2, CPR = 1.3 0.1
and CPE = 0.7 0.1. This simply means that every molecule
of N2 produced takes an average of 1.3 N atoms from the
surface and an average of 0.7 atoms from the gas phase. It is
not feasible to carry out this analysis for the N2O production
because, as has been mentioned, the switch is over too fast
for any accuracy to be ensured.
y1(t)
2 y0(t)(or 2 y2(t)). [22]
This indicates that the largest deviation will be seen for
the y1(t) profile (representing the deviation from equilib-
rium for the 14N15N species) and that the y2(t) and y0(t)
profiles (which represent the deviations for the 14N14N and
the 15N15N species) will be mirror images of this profile but
will be one half as intense. In this work we show both the
y1(t) and the y2(t) profiles along with the profile.
Table 1 shows how any variation in the “type of pro-
duction” is manifested in a change in the TRIMP profiles
for a plug flow reactor (30). For types 0 and 2a the iso-
topic molecules of product are always statistically mixed
(y1 = y2 = 0); for types 1 and 2b there is superproduction
of the mixed-labelled species for a time following the switch
(y1 > 0 > y2). Conversely, for type 2c there is subproduction
of the mixed-labelled species for a time following the switch
(y1 < 0 < y2). These only apply in the case of “pure” types
of production, and any concurrent operation of more than
one type of production leads to situations that are more
complex. The operation of more than one “type” (for mul-
tiroute mechanisms) does not generally yield a simple pro-
file resulting from the average of the profiles of the types
involved. The imposition of an experimental step change
rather than an ideal one here also causes some deviation
from “true” yi profiles. This deviation is only seen at the
beginning of the switch when the inverted Ar profile has
b. Temporal Redistribution of Isotopically Labelled
Molecules of Product (TRIMP)
An important characteristic of isotopically labelled
molecules with more than one labelled atom of the same
element is the extent of the redistribution of the isotopic
equilibrium present. This is the situation in our case regard-
ing the product molecules (N2 and N2O). If these molecules
are in a state of isotopic equilibrium when the fraction of
heavy isotopes (15N) in the gas phase is , the equilibrium
fraction (xieq) of the isotopic molecules, e.g., 15Ni15N(2 i), can
be calculated using the binomial distribution (24, 27)
2
xieq
=
i(1
)
,
[18]
2
i
i

THE NO + H2 REACTION OVER Pt/SiO2 CATALYSTS, 2
367
not reached a steady value (0 in our case as 14NO/Ar is
removed from the stream).
In the present case, Fig. 2 (a and b), neither the N2
nor the N2O products are in statistical isotopic equilibrium
throughout the course of the whole switch.
In the case of the N2 profiles (Fig. 2a) the deviations ap-
pear to be superpositions of more than one mode. There
is superproduction of the mixed labelled species directly
following the switch (with a maximum value of y1 approx-
imately 20% ). This becomes a subproduction roughly 55 s
after the switch (with a minimum value of y1 of approxi-
mately 3% 175 s after the switch), before statistical redis-
tribution is reached 360 s after the switch (at an value of
0.9). We can therefore say that there is not a “pure” type
of production operating here. It should also be noted that
this generated profile is reproducible, having been repeated
15
several times following the 14NO + H2 → NO + H2 switch
for these and other experimental conditions.
The mechanism proposed for N2 formation in the
NO/C3H6/O2 reaction (33) envisages the combination of
two N atoms on the surface. This is analogous to mech-
anism 2a above, in which two equivalent surface species
interact to yield the product molecule N2. Theoretical con-
sideration (30) predicts that this situation would lead to
no deviation from statistical isotopic equilibrium following
15
the 14NO + () → NO + () switch. Thus, we can discount
this mechanism as the sole source of N2 for the NO + H2
reaction on Pt/SiO2.
As mentioned, it appears that the observed profiles, in
which there seem to be two controlling factors governing
the sign of the y1 profile, correspond to superpositions of
at least two types of production of N2 from NO. We know
from the IDIMP analysis that this is the case: = 0.7,
+
1
2a
+
= 0.3.
2b
2c
Extensive modelling (30) has shown that there are only
two “natural” cases where this change of sign (from positive
to negative) in the y1 profile can take place. In both cases
we must assume that there are two routes with two types of
production of PA2 and that the production of N2 follows an
“isotopically first/isotopically second”-type series of pro-
cesses as seen before with respect to the production of N2
and N2O (Scheme 1).
There is also a connection between the contribution of a
route r and the type of production . If, for a multiple-route
mechanism, a route (i) gives only a type of production k,
FIG. 2. TRIMP plots (in the y1-time coordinate) for (a) 15N2 (y2)(4)
and 14N15N (y1)(᭞) and for (b) 15N2O (y2)(4) and 14N15NO (and 15N14NO)
(y1) (᭞) following the same experiment shown in Fig. 1. Profiles for the
respective function (᭺) and the inverted Ar trace (ᮀ) are also shown.
SCHEME 1

368
SHESTOV, BURCH, AND SULLIVAN
then the contribution of this route ri = i /6
=
k, where (see Fig. 3). This is interesting since “pure” type 1 leads to
i
i = 1, 2, . . . N (number of routes) and k = 0, 1, 2a, 2b, 2c an overproduction of 14N15N and “pure” type 2a leads to
(number of types of production).
a statistical mixing of isotopic molecules of product. How-
Processes that can lead to a superproduction of 14N15N ever, when both types are operating together, the y1 profile
(as seen in the isotopically first N2 produced) can be type 1 changes sign and there is a subproduction of 14N15N seen
or type 2b production. However, we know that we have a for a time after the switch. Scheme 1 shows these two com-
contribution of type 1 production from the IDIMP results. binations of types of production.
The only “pure” type of production that leads to a sub-
Briefly, this refers to a situation where RA adsorbs on the
production of 14N15N is that which comes from a coupled catalyst surface to form a physisorbed species (IA) (step 1).
intermediate on the surface (type 2c). Thus it is possible that This then fully adsorbs on the surface (step 2 to form I0A).
the formation of N2 takes place through two routes, one be- Now it can interact with a gaseous RA molecule (or the
ing type 1 and one being type 2c, with the former leading to physisorbed IA species) to form PA2 (step 4—producing
isotopically first N2 and the latter leading to isotopically sec- the “isotopically first” component of PA2 with a type 1 pro-
ond. Our modelling shows (30) that the interaction of these duction, as discussed earlier). It can also go on (step 3) to
two types of production leads to a y1 profile that changes form a second type of intermediate I0Ax (x = 1 or 2). If x = 1,
sign in a qualitatively similar manner to that seen here then this species can then interact with another I0Ax species
(see Fig. 3).
to form PA2 (isotopically second in a type 2a production—
Also from modelling it follows that the interaction of step 5). If x = 2, then this species can itself form PA2 (iso-
routes with a type 1 production (in which RA + IA → PA2) topically second in a type 2c production—step 5).
and a type 2a production (in which 2IA → PA2) can also
Thus, we will need to discriminate between these two sit-
produce a profile qualitatively similar to that seen in Fig. 2a uations (superpositions of types 1 + 2a and types 1 + 2c) in
order to get a picture of the actual mechanism of N2 forma-
tion under reaction conditions. Figure 3 shows modelling
results for the y1, y2, and profiles expected from a com-
bination of 1 + 2a (y1, ᭝; y2 ᭡) and 1 + 2c (y1, ᭺; y2 ᭹).
These are modelled on the assumption that the contribu-
tion of type 1 is
= 0.7 (from the IDIMP analysis), and
1
in both cases the profile (ᮀ) is the same. Figure 4 shows
these modelled data presented as y1 (y2) as a function of .
One difference between these profiles (Fig. 3) is that the
minimum value of the y1 profile is lower in the second case
(superposition of 1 and 2c) than in the first (superposition
of 1 and 2a). Another difference is that in the former case
(1 + 2a) statistical equilibrium (y1 = y2 = 0) is reached when
the value is 0.95, while in the latter case (1 + 2c) the
molecules of product are not statistically mixed until the
value approaches 1. This is best seen in Fig. 4. Another
way of pointing this out is to observe that in the 1 + 2c case
the y1 (y2) profile, upon changing sign, is approximately
symmetrical between this value of and = 1. The y1 (y2)
profile is not symmetrical (between the value of at which it
changes sign and reaches a value of = 1) in the case where
1 + 2a operates as it reaches a value of zero before = 1.
Note here that the approximately symmetrical form for
yi in the yi
graphical presentation is a common feature
for the superposition of 1 + 2c. This difference in symmetry
(between 1 + 2c and 1 + 2a) will be more pronounced when
the concentration of the I0Ax is relatively low. In both cases
the absolute value of the minimum for y1 (or the maximum
FIG. 3. Modelled (PA2), y1, and y2 profiles from the superposition for y2) will be lower when the I0Ax concentration is low.
of types of production 1 + 2a and 1 + 2c (Scheme 1): profile (ᮀ), y1
profile (1 + 2a (4) 1 + 2c (᭺)), y2 profile (1 + 2a (᭡) 1 + 2c (᭹)). Results
are modelled setting the contribution of type 1 process at 0.7. All steps are
irreversible, the rate of adsorption is set at 1, C(IA)preads is small, CI0A = 1,
CI00Ax = 0.25 (with respect to the number of A atoms), and all units are
unidimensional.
The yi versus profiles derived from the present data
are shown in Fig. 5. The y1 (y2) profiles are not sym-
metrical following the change in sign. Indeed isotopically
statistically mixed N2 molecules of product are formed
when (N2) 0.95. Thus our experimental data supports

THE NO + H2 REACTION OVER Pt/SiO2 CATALYSTS, 2
369
14
profile in the case of the reverse 15NO → NO switch).
In this case ¯ refers to 1
as the switch considered is
NO + () → NO + (). This gives information about the re-
action network leading to products. This presentation is de-
rived from classical isotopic exchange analysis and is used
to reflect the heterogeneity of the surface (or lattice) inter-
mediates (30).
Specifically, at steady state, it gives information regarding
the presence of a buffer step (in which a pool of inactive in-
termediates could be formed, in a reversible process, from
a pool of active intermediates) or the presence of a consec-
utive mechanism (in which one pool of intermediates goes
on to form another pool of intermediates before forming a
product molecule), or, for example, the presence of a mech-
anism in which there is only one intermediate.
In each case the semilogarithmic plot of ¯ versus time has
a different shape. In the first case the curve is convex, in the
second it is concave, and in the third a straight line is ob-
tained. A mechanism in which there are parallelrouteslead-
ing to product and both show reversibility also yields a con-
vex plot. Details of these different possibilities can be found
in Table 2. The following shapes are expected. In the case of
irreversible adsorption of reactant and desorption of prod-
uct, for a one-pool mechanism (t) is a function of one ex-
ponent (19) (therefore the semilogarithmic plot of ¯ versus
FIG. 4. The data in Fig. 3 plotted as a function of yi against : y1 profile
(1 + 2a (4) 1 + 2c (᭺)), y2 profile (1 + 2a (᭡) 1 + 2c (᭹)).
a two-routes mechanism for the formation of N2. One of
these routes gives a type 1 production while the second
gives a type 2a production.
The temporal redistribution profile for the mixed-
labelled N2O species (Fig. 2a) is lower than would be ex-
pected from the above-mentioned statistical equilibrium
calculation. Conversely, the unlabelled and doubly labelled
N2O species obviously show the reverse behaviour, with
production profiles consistently higher than would be ex-
pected from a statistical equilibrium mixture. Once the
switch is over (regarding the product molecule of interest)
then statistical equilibrium is restored; i.e., all the molecules
formed are doubly labelled and the value for the product
molecule is equal to 1.
This is not what would be expected from the proposed
mechanism of N2O formation over supported Pt catalysts
from the NO/C3H6/O2 reaction (33). Therefore, we can say
that a different mechanism of N2O formation operates in
the NO + H2 reaction over these catalysts. This is indicative
of a type 2c production.
c. Semilogarithmic Plots of ¯ versus Time
These plots involve plotting the function |ln|( ¯ )|| against
time following the isotopic switch (where ¯ represents the
(1
FIG. 5. Experimental data plotted as shown in Fig. 4, i.e., yi as a func-
15
) profile in the case of a 14NO → NO switch and the tion of , y1 (᭺), y2 (ᮀ).

370
SHESTOV, BURCH, AND SULLIVAN
time is linear). Buffer (19) and parallel (20) pools result
in (t) being the sum of two exponents (semilogarithmic
plot of ¯ versus time is convex), and in the case of consecu-
tive pools then (t) is the difference of two exponents (19)
(semilogarithmic plot of ¯ versus time is concave). Combi-
nations of these mechanisms lead to semilogarithmic plots
of ¯ versus time that are superpositions of the former plots.
Our results are generated from modelling reversible
cases of adsorption of RA and desorption of PA2 (when
(t) cannot be represented in exponential form), and these
show qualitatively similar shaped features to the situations
discussed above (34). In the case of reversible adsorption
of reactant, a situation that has heretofore been ignored,
then the same transformation of the profile of the unre-
acted reactant (RA) can also yield information regarding
the reaction network (34). As was the case in the IDIMP
and TRIMP studies, the nature of the “gas-phase holdup”
effects the shape of these profiles. However, in this case the
effect is far less severe. The relationship between the “gas-
phase holdup” and the (product) response (in the case
of irreversible adsorption/desorption) has been previously
derived (35).
In our case the semilogarithmic plot of ¯ versus time for
the production of both N2 and N2O is shown with a standard
Ar response in Fig. 6. Both of these are convex, and thus
both sets of intermediates (of N2 and N2O) must have buffer
statesor possiblyare formed from parallelmechanismssuch
as those detailed in the fourth row of Table 2. It is seen that
the N2O profile is far sharper than that of the N2. This is as
expected as the transient in N2O is itself far quicker than
that for N2 (N2O is the isotopically first product). A value of
4 in the semilogarithmic plot of ¯ versus time is indicative
of 99% transference of the heavy (15N) isotope into the
product molecules. Thus in the reaction network scheme
we must, with a high probability, allow for the presence of
an intermediate buffer state for N2O and for N2 produc-
FIG. 6. Semilogarithmic plots of the function |ln|( ¯ )|| against time
for the product molecules N2O and N2 as well as an Ar profile follow-
ing the 14NO + H2 → 15NO + H2 switch over 5% Pt/SiO2 at 60 C: N2 (᭿),
N2O (᭹), inverted Ar (ᮀ).
tion, and both buffers cannot be represented by the same
species.
The formation of the buffer can be considered as NO ad-
sorbed on different interconvertible sites, one ofwhich isac-
tive for further reaction/desorption and one of which is not.
This could be a physically different site on the catalyst (e.g.,
at a different metal site or at the metal–support interface,
but not the support itself as SiO2 does not adsorb significant
quantities of NO) or a site that is periodically modified un-
der reaction conditions, e.g., NOads → NO2ads. At steady
state the concentration of the buffer must be a constant;
i.e., its rate of formation must equal its rate of removal.
TABLE 2
The Variations Expected in the Semilogarithmic Plots of the
Function |ln|( ¯ )|| against Time for Various Mechanisms for the
Transformation of Reactant R into Product P
Shape of semilogarithmic
plot (with time) |ln( ¯ (t))|
PROPOSED MECHANISM
Label
Direct
Consecutive
Buffer
Mechanism
From the above analyses and those presented in the pre-
vious paper (13) we can outline a reaction scheme for the
production of N2, N2O, and NH3 during the NO + H2 reac-
tion.
R ⇔ I1 ⇔ P
Straight line
Concave curve
Convex curve
R ⇔ I1 ⇔ I2 ⇔ P
R ⇔ I1 ⇔ P
m
I2
With respect to the formation of N2O, from the present
analysis, we can say:
I1
Parallel
Convex curve
R
P
There is a buffer state present in the reaction network
(as qualitatively determined from the semilogarithmic plots
of ¯ versus time).
I2
Note. See text for details.

THE NO + H2 REACTION OVER Pt/SiO2 CATALYSTS, 2
371
SCHEME 2. Network of NO conversion into N2O, N2, and NH3 over 5% Pt/SiO2 in the NO + H2 reaction as derived from SSITK analysis.
The formation is very rapid on the surface (relative to “equivalent intermediates” mechanism (type 2a) produces
N2 production).
N2 isotopically second.
A final intermediate on the surface in the production of
N2O contains two nitrogen atoms (from the TRIMP data).
Scheme 2 is a general outline of the network of the
N-containing species, following them from gaseous NO
through the various surface intermediates and showing pro-
duction ofN2, N2O, and NH3. The scheme takesinto account
With respect to the formation of N2, from the present
analysis we can say:
There is buffer state on the catalyst for N2 precursors the above-mentioned points, as well as the conclusions from
(from semilogarithmic plots of ¯ versus time). Also we can the previous paper (13).
say that this buffer is not the same one as that seen for the
production of N2O, as this is a far slower process on the adsorbed state (step 1). The pre-adsorbed (weakly bonded)
surface. NO can dimerise to produce dinitrosyl (NO)2-type species
In this scheme the adsorption of NO goes through a pre-
There are at least two types of production of N2 on the (step 2). This can be stabilised—perhaps by the addition
catalyst. One of these is through a type 2a process, i.e., a of an electron from the Pt particles (yielding an (NO)2
route where two equivalent species on the surface interact species). Thiscan be considered asa buffer step (step 3). The
to form N2, and one is through type of production 1—an (NO)2 species could also dissociate to the N2O(g) and leave
“Impact” route, i.e., where gaseous or physisorbed NO in- behind an Oads species (step 4). The overall concentration of
teracts with an NO-derived species on the surface to form surface intermediates (NO)2 and (NO)2 must be quite low.
N2 (from the temporal redistribution plots and the initial
The pre-adsorbed state can also move to a more “fully”
distribution data). The contribution of type 1 is around 70% adsorbed state on the metal surface (step 5). From here it
while that of type 2a is near 30% —as determined from the can either enter a buffer state (step 6) from which it can
IDIMP analysis;
re-emerge later, or it can become more reduced (by the
The concentration of final intermediates which form formation of N, step 7).
N2 with type of production 2a is relatively small. We can
This species (formed in step 7) could, in the presence of
say this as there is only a small negative deviation from excess H2, become further hydrogenated (step 8, forming
equilibrium of the y1 values from the TRIMP analysis; NHx). Further hydrogenation would result in the genera-
The N2 formation through the “impact” type of pro- tion of NH3 (step 12). However, it is considered here that its
duction (type 1) produces N2 “isotopically first” while the primary reaction route is to interact with the pre-adsorbed

372
SHESTOV, BURCH, AND SULLIVAN
(or gaseous) NO to form N2 (in the modified “impact” route,
step 10). Another possible method of forming N2 via the
“impact route,” as detailed in step 9, could be the interac-
tion of the physisorbed NO with the N0 species formed via
NO decomposition. Both of these routes give the type 1
production of N2.
We also need to account for the production of N2 from
two intermediates that are equivalent and stable on the
surface (type of production 2a). This could be formed by
an interaction (step 11) between two of the N0 species
formed in step 7.
REFERENCES
1. Hanssen, K. F., Blekkan, E. A., Schanke, D., and Holmen, A., Stud.
Surf. Sci. Catal. 109, 193 (1997).
2. Bajusz, I. G., Kwik, D. J., and Goodwin, J. G., Catal. Lett. 48(3-4), 151
(1997).
3. Efstathiou, A. M., and Verykios, X. E., Appl. Catal. A General 151(1),
109 (1997).
4. Nibbelke, R. H., Scheerova, J., de Croon, M. H. J. M., and Marin, G. B.,
J. Catal. 156, 106 (1995).
5. Ali, S. H., and Goodwin, J. G., J. Catal. 170(2), 265 (1997).
6. Oukaci, R., Blackmond, D. G., Goodwin, J. G., Jr., and Gallagher,
G. R., in “Catalytic Control of Air Pollution,” Ch. 5, pp. 61–72. Am.
Chem. Soc., Washington, DC, 1992.
7. Frost, J. C., Lafyatis, D. S., Rajaram, R. R., and Walker, A. P., in “4th In-
ternational Congress on Catalysis and Automotive Pollution Control,
Brussels, April 1997,” Vol. 1, p. 129, O14, 1997.
8. Efstathiou, A. M., and Fliatoura, K., Appl. Catal. B Environ. 6(1), 35
(1995).
9. Janssen, F. J. J. G., Van Den Kerkhof, F. M. G., Bosch, H., and Ross,
J. R. H., J. Phys. Chem. 91(27), 6633 (1987).
10. Kumthekar, M. W., and Ozkan, U. S., J. Catal. 171(1), 54 (1997).
11. Burch, R., and Sullivan, J. A., J. Catal. 182(2), 489 (1999).
12. Burch, R., Shestov, A. A., and Sullivan, J. A., J. Catal. 182(2), 497
(1999).
We can also comment on the rates and reversibility of
certain of the steps from the SSITKA and NSSITKA in the
previous paper (13). For example, we can say that the rate
of the N2O formation is of the same order as the net rate of
NO adsorption. This, coupled to the information that the
overall concentration of N2O intermediates is small (13),
would explain how the N2O profile and the 15NO profile
15
are the same following the 14NO → NO switch.
We can also say that step 2 and step 3 are reversible
but that step 5 is irreversible. This we have seen from the
stopped flow switching experiments (13), in which the N2O
precursors could be transformed into N2 precursors by the
removal of the gaseous NO. The representation of the cou-
pled intermediate (NO)2 and its stabilisation by donation
of an electron have been postulated previously (36) and are
chemically feasible.
13. Burch, R., Shestov, A. A., and Sullivan, J. A., J. Catal. 186, 354
(1999).
14. Happel, J., Chem. Eng. Sci. 33, 1567 (1978).
15. Bennett, C. O., in “Catalysis Under Transient Conditions” (A. T. Bell
and L. L. Hegedus, Eds.), ACS Symposium Series, Vol. 178, p. 1. Am.
Chem. Soc. Washington, DC, 1982.
16. Biloen, P., J. Mol. Catal. 21, 17 (1983).
17. Godfrey, K., “Compartmental Models and their Application.” Aca-
demic Press, London, 1983.
CONCLUSIONS
18. Happel, J., Walter, E., and Lecourtier, Y., I&EC Fundam. 25, 704
(1986).
A mechanism has been proposed for the NO/H2 reaction
over Pt/SiO2 catalyststhat explainsallthe observed features
from the SSITKA and temperature programmed analysis.
This involves two routes (with types of production 1 and 2a)
for the formation of N2. The first of these is via an “impact”
mechanism where gaseous (or weakly adsorbed) NO inter-
acts with a surface species, and the second involves interac-
tion of two equivalent species on the surface. The contribu-
tion of the Eley–Rideal-like “impact” route is roughly 70%
while that of the Langmuir–Hinshlewood route is roughly
30% . N2O is formed from the decomposition of a surface
species containing 2 N atoms (with a 2c type of production).
The formation of N2O is far faster on the surface than that
of N2 and its surface intermediates are far less stable than
those of N2.
19. Happel, J., Walter, E., and Lecourtier, Y., J. Catal. 123, 12
(1990).
20. Shannon, S. L., and Goodwin, J. G., Jr., Chem. Rev. 95, 677
(1995).
21. Mirodatos, C., Catal. Today 9, 83 (1991).
22. Kemball, C., Adv. Catal. 11, 223 (1959).
23. Klier, K., Novakova, J., and Jiru, P., J. Catal. 2, 479 (1963).
24. Muzykantov, V. S., Popovskii, V. V., and Boreskov, G. K., Kinet. Catal.
5(N4), 624 (1964).
25. Boreskov, G. K., and Muzykantov, V. S., Ann. N.Y. Acad. Sci. 213, 137
(1973).
26. Muzykantov, V. S., React. Kin. Catal. Lett. 33, 937 (1987).
27. Boreskov, G. K., in “Catalysis, Science and Technology,” Vol. 3, p. 39.
Springer-Verlag, New York, 1982.
28. Winter, E. R. S., J. Chem. Soc. A, 2889 (1968).
29. Ozaki, A., in “Isotopic Studies of Heterogeneous Catalysts.” Aca-
demic Press, New York, 1977.
30. Shestov, A. A., Burch, R., and Sullivan, J. A., to be submitted.
31. Mirodatos, C., J. Phys. Chem. 90, 681 (1986).
32. Mirodatos, C., Dalmon, J. A., and Martin, G. A., J. Catal. 105, 405
(1987).
33. Burch, R., Sullivan, J. A., and Watling, T. C., Catal. Today 41(1-2), 13
(1998).
34. Shestov, A. A., Burch, R., and Sullivan, J. A., “Reaction Kinetics
and the Development of Catalytic Processes,” Belgium, April 19–21,
1999.
At 60 C, under steady state conditions every molecule of
N2 formed during the NO + H2 reaction takes an average
of 1.3 atoms from the surface intermediates and an average
of 0.7 atoms from gas-phase NO (CPR = 1.3, CPE = 0.7).
ACKNOWLEDGMENTS
We are grateful to the EPSRC for supporting this research through
contract GR/K70403. A.A.S. thanks NATO and The Royal Society for
providing a fellowship (NATO/96A).
35. Shannon, S. L., and Goodwin, J. G., Appl. Catal. A General 151, 3
(1997).
36. Acke, F., Ph.D. thesis, University of Go¨teborg, Sweden.