N.W. Cant et al. / Journal of Catalysis 278 (2011) 162–166
163
does not desorb intact, that its dissociation by Eq. (7) is rate limiting
and that the surface is largely covered by bridge-bonded CO.
The model of Uner [12] does not seem very reasonable in that
the nitrogenAnitrogen bond in gaseous N2O is much stronger than
the nitrogenAoxygen one, 474 kJ/mol versus 161 kJ/mol [13].
Nonetheless, Holles et al. [6] concluded, on the basis of in situ
infrared spectroscopy and reaction order data, that the N2O + CO
reaction ‘‘appeared to involve N2O dissociating into N and NO with
subsequent dissociation of NO’’, as per the model of Uner [12]. In
particular, they observed the accumulation of NCO species, the for-
mation of which necessarily required cleavage of the NAN bond.
Recent DFT calculations have also indicated that cleavage of the
nitrogenAnitrogen bond in N2O is feasible on a Rh(1 1 1) surface,
via an intermediate in which N2O is bound through its terminal
atoms, albeit at a rate predicted to be several orders of magnitude
slower than that of nitrogenAoxygen cleavage [14].
The stream leaving the reactor was analysed on-line in three
ways – by quadrupole mass spectrometry (Balzers Thermostar
model GSD 300T) every 20 s, by gas chromatography (MTI model
M200 with molecular sieve and Poraplot U columns) every 3 min
during isotopic substitution and by FTIR spectroscopy (Nicolet
Magna 550 fitted with a 16 cm flow-through cell) accumulating
64 scans at 0.25 cmꢁ1 resolution every 10 min. The infrared spectra
were processed to obtain outlet concentrations using the program,
MALT, developed by Griffith [16] and, in the case of the isotopo-
mers of N2O, using a database attributable to Toth [17].
The 15N14NO was obtained from Cambridge Isotope Laborato-
ries with the chemical and isotopic purity each stated to be
>98%. Mass spectrometric analysis over the parent ion region
showed 1.0% 14N2O and 0.4% 15N2O, with the balance nitrous oxide
containing a single 15N. FTIR analysis showed that the latter was
>95% 15N14NO with 14N15NO below the detection limit of 5%. Unla-
belled and labelled nitrous oxides were each made up as ꢀ5% mix-
tures in helium and supplied from separate electronic mass
controllers with the two streams interchanged using a low volume
four-port valve (Valco Inst., 1/1600). The streams were run in turn
for 20 min each with the changeover between them completed
by the second subsequent cycle of the mass spectrometer (ꢀ40 s).
Kinetic studies subsequent to Uner’s proposal have examined
other models. Angelides and Tzitzios [7] tested their data for Rh/
Al2O3 for conformity to three models – that of McCabe and Wong
[3], a modified version in which N2O dissociation required an adja-
cent vacant site
ꢂN2O þ ꢂ ! N2ðgÞ þ Oꢂ þ ꢂ;
ð10Þ
and
a standard bimolecular Langmuir–Hinshelwood model in
3. Results and discussion
which the rate limiting step was the reaction between adsorbed
N2O and adsorbed CO
3.1. The reduction of N2O by CO
ꢂN2O þ ꢂCO ! N2ðgÞ þ CO2ðgÞ þ 2ꢂ:
ð11Þ
The lines in Fig. 1 show the course of the 14N2O + CO reaction
over the Rh/SiO2 catalyst when ramped up at 2 °C/min. Conver-
sions (X) were calculated from the concentrations (C) of nitrous
oxide and carbon dioxide using the relationships
The bimolecular model gave the best fit. However, the most recent
study, that of Granger et al. [8], favoured the modified (i.e. adjacent-
site requiring) version of the McCabe and Wong model on the basis
that the kinetic parameters obtained were in better conformity with
those for the NO + CO reaction.
XN2O ¼ 100ꢂ½CN2OðinÞ ꢁ CN2OðoutÞꢃ=CN2OðinÞ
and
ð12Þ
Overall, although the model proposed by Uner [12] does not
seem likely, a direct test, as is possible when nitrous oxide contain-
ing a single 15N atom is used, seems worthwhile. If 15N14NO is re-
acted with CO, then, according to the standard McCabe and Wong
scheme, the modified version of it and also the bimolecular model
of Angelides and Tzitzios [7], 14N15N alone should be produced. On
the other hand, Uner’s scheme implies that product nitrogen arises
through random combination of individual nitrogen atoms and
hence 14N2, 14N15N and 15N2 should be formed in the statistical
ratio 1:2:1. In the present work, we have carried out this experi-
ment and the corresponding one between 15N14NO and H2 where
some of the complicating features of the analysis that are present
when CO is the reductant are absent.
XCO ¼ 100ꢂCCO2ðoutÞ=CCOðinÞ:
ð13Þ
The concentration of nitrous oxide was monitored using the signal
from its fragment ion at m/z = 30 (14N16O+) and that of carbon
2. Experimental
The 0.5 wt% Rh/SiO2 sample was made by impregnation of silica
gel (Davison grade 62, BET surface area of 285 m2/g, particle size
180 to 210 lm), with a solution of RhCl3. The dispersion was 38%
as measured by H2 chemisorption following overnight drying at
140 °C and reduction in 10% H2/He on a ramp (at 1 °C/min) ending
with 2 h at 350 °C.
Catalytic measurements were carried out in a flow system de-
scribed in detail previously [15]. In essence, the reactant stream
(nominally 2000 ppm N2O, 2000 ppm CO or H2, balance He with
total flow rate of 100 cm3 (STP)/min) was passed down-flow over
75 mg of the Rh/SiO2 supported between quartz wool plugs in a
Pyrex reactor with an internal diameter of 6 mm. The sample of
Rh/SiO2 had been used previously in studies involving reactions
of mixtures of NO, CO and H2 [15] during which it exhibited con-
stant activity if flushed with He overnight between runs and the
same procedure was followed here.
Fig. 1. Conversions versus temperature for the reaction of 1927 ppm N2O with
1980 ppm CO over 75 mg of 0.5 wt% Rh/SiO2. The continuous curves are calculated
from the MS data for the reaction of 14N2O when the temperature of the catalyst
was ramped up at 2 °C/min. Symbols are for subsequent reaction at the fixed
temperatures used for the runs with 15N14NO. The inset plot shows an Arrhenius
plot for turnover frequencies over the conversion range from 5% to 20% during the
ramp-up.