In situ electrochemical promotion by sodium of the platinum-catalyzed reduction of NO by propene

Article abstract of DOI:10.1021/jp963052c

The Pt-catalyzed reduction of NO by propene exhibits strong electrochemical promotion by spillover Na supplied from a ?2a?3-alumina solid electrolyte. In the promoted regime, rate increases by an order of magnitude are achievable. At sufficiently high loadings of Na the system exhibits poisoning, and excursions between the promoted and poisoned regimes are fully reversible. Reaction kinetic data obtained as a function of catalyst potential, temperature, and gas composition indicate that Na increases the strength of NO chemisorption relative to propene. This is accompanied by weakening of the N-O bond, thus facilitating NO dissociation, which is proposed as the critical reaction-initiating step. The dependence of N₂/N₂O selectivity on catalyst potential is in accord with this view: Na pumping to the Pt catalyst favors N₂ production at the expense of N₂O. X-ray photoelectron spectroscopic (XPS) data confirm that electrochemical promotion of the Pt film does indeed involve reversible pumping of Na to or from the solid electrolyte. They also show that under reaction conditions the promoter phase consists of a mixture of sodium nitrite and sodium nitrate and that the promoted and poisoned conditions of the catalyst correspond to low and very high loadings of these sodium compounds. Under all reaction conditions, a substantial fraction of the promoter phase is present as 3D crystallites.

Full text of DOI:10.1021/jp963052c

J. Phys. Chem. B 1997, 101, 3759-3768
3759
In Situ Electrochemical Promotion by Sodium of the Platinum-Catalyzed Reduction of NO
by Propene
†
‡
Ioannis V. Yentekakis, Alejandra Palermo, Neil C. Filkin, Mintcho S. Tikhov, and
Richard M. Lambert*
Department of Chemistry, UniVersity of Cambridge, Cambridge CB2 1EW, England
X
ReceiVed: October 3, 1996; In Final Form: February 19, 1997
The Pt-catalyzed reduction of NO by propene exhibits strong electrochemical promotion by spillover Na
supplied from a â′′-alumina solid electrolyte. In the promoted regime, rate increases by an order of magnitude
are achievable. At sufficiently high loadings of Na the system exhibits poisoning, and excursions between
the promoted and poisoned regimes are fully reversible. Reaction kinetic data obtained as a function of
catalyst potential, temperature, and gas composition indicate that Na increases the strength of NO chemisorption
relative to propene. This is accompanied by weakening of the N-O bond, thus facilitating NO dissociation,
which is proposed as the critical reaction-initiating step. The dependence of N
potential is in accord with this view: Na pumping to the Pt catalyst favors N production at the expense of
O. X-ray photoelectron spectroscopic (XPS) data confirm that electrochemical promotion of the Pt film
2 2
/N O selectivity on catalyst
2
N
2
does indeed involve reversible pumping of Na to or from the solid electrolyte. They also show that under
reaction conditions the promoter phase consists of a mixture of sodium nitrite and sodium nitrate and that the
promoted and poisoned conditions of the catalyst correspond to low and very high loadings of these sodium
compounds. Under all reaction conditions, a substantial fraction of the promoter phase is present as 3D
crystallites.
Introduction
this connection, Rh plays a key role in current three-way
converters because it is highly effective for the dissociative
chemisorption and reduction of NO. In a recent study of
electrochemical promotion of the CO + NO reaction over Pt,8
we showed that the selectivity, and especially the activity, of a
Electrochemical promotion (EP), discovered and developed
by Vayenas et al., is an entirely new way of controlling catalyst
1
performance. Reference 1 provides a detailed account of the
phenomenology and outlines a theoretical basis in terms of
which the EP effect may be interpreted. The method entails
electrochemical pumping of ions from a solid electrolyte to the
surface of a porous, catalytically active metal film: the resulting
changes in work function (∆Φ) of the catalyst alter the
adsorption enthalpies of adsorbed species and the activation
Pt film electrochemically fed with sodium from a â′′-alumina
+
(
a Na ion conductor) support could be dramatically en-
hanced: in effect, Pt could be induced to behave like Rh. The
results indicated that Na induces NO dissociation which is
thought to be the crucial reaction-initiating step. Here we
present an extension of this work to a potentially more important
system: EP of the reduction of NO by propene over Pt/â′′-
alumina. (Propene is the major constituent of the hydrocarbon
component in automotive exhaust and is the industry standard
for catalyst testing.) Very large rate accelerations and substantial
gains in selectivity towards N2 production were achieved, while
XP spectra gave insight into the nature of the promoting species
and the origin of the poisoning effects observed at high promoter
loadings.
1
energies of reactions involving these species. It has been shown
that the electrochemically induced changes in catalyst potential
measured with respect to a reference electrode (∆VWR) are
identical to ∆Φ. The effects on activity and selectivity are
reversible, and the phenomenon provides a uniquely efficacious
and controllable method for the in situ tuning of working
catalytic systems. Thus, in addition to its potential practical
utility, EP also provides a means for the systematic investigation
of certain aspects of promoter action in a manner that was
hitherto not possible. Note that EP is strongly non-faradaic:
one is not dealing with conventional electrocatalysis. The
electrochemically induced catalytic rate changes are typically
Methods
The platinum catalyst (working electrode, W) consisted of a
3
5
1
0 -10 times greater than the rate of supply of promoter
2
porous continuous thin film (∼2 cm geometric area) deposited
species. The EP phenomenon has been reported for over 40
catalytic reactions on Pt, Rh, Pd, Ag, Ni and IrO2 catalyst films
on one face of a 20 mm diameter â′′-Al2O3 disc (Ceramatec)
5
as described in detail elsewhere. Au reference (R) and counter
2
-4
with a variety of solid electrolytes including oxygen,
(
C) electrodes were attached to the other face of the solid
1
,5-8
9
10
sodium,
fluorine, and hydrogen ion conductors.
5
electrolyte wafer (see Figure 1, ref ) by vacuum evaporation, a
procedure that produced electrodes with good adhesion and low
electrical resistance. The actual active surface area of the Pt
catalyst electrode (W) was estimated by comparing the observed
rate of CO oxidation under standard conditions (6 kPa of O2,
1.8 kPa of CO, 623 K, and VWR ) +800 mV) with those
It is well-known that reactions involving the catalytic
reduction of NO are of major environmental importance. In
1
1
*
Corresponding author. Telephone 44 1223 336467; FAX 44 1223
3
36362; e-mail rml1@cam.ac.uk.
†
Institute of Chemical Engineering and High Temperature Chemical
5
Processes and Department of Chemical Engineering, University of Patras,
Patras, GR-26500, Greece.
obtained using samples whose surface area had been directly
determined using an O2/CO titration technique. This gave a
12
‡
Institute of Materials Science and Technology (INTEMA), UNMDP-
-
7
2
value of 5 × 10 mol of Pt (∼200 cm ) for the active metal
CONICET, J. B. Justo 4302, (7600) Mar del Plata, Argentina.
X
Abstract published in AdVance ACS Abstracts, April 1, 1997.
area. The Pt/â′′-Al2O3/Au sample was suspended in a quartz,
S1089-5647(96)03052-0 CCC: $14.00 © 1997 American Chemical Society

3760 J. Phys. Chem. B, Vol. 101, No. 19, 1997
Yentekakis et al.
atmospheric-pressure well-mixed reactor with all three electrodes
7,8
exposed to the reactant gas mixture. The reactor volume was
3
1
15 cm , and the system behaved as a single pellet, continuous
stirred tank reactor (CSTR) as described and discussed else-
4
,5,7,13,14
where.
Inlet and exit gas analysis was carried out by a combination
of on-line gas chromatography (Shimadzu-14B; molecular sieve
5
A and Porapak-N columns) and on-line mass spectrometry
(Balzers QMG 064). N2, N2O, CO, CO2, and C3H6 were
measured by gas chromatography, and NO was measured
continuously by mass spectrometry after performing the required
calibrations. The mass spectrometer was also used to monitor
continuously m/z ) 28 (N2 + CO), 44 (CO2 + N2O), 16 (O),
41 (propene), and 17 (NH3). Reactants were pure NO (Distillers
MG) and propene (Matheson) diluted in ultrapure He (99.996%)
and fed to the reactor by mass-flow controllers (Brooks 5850
TR). The total flow rate was kept essentially constant in all
-
4
-1
3
experiments at 1.3 × 10 mol s (190 cm (STP)/min), with
partial pressures PNO, and PC3H6 varying between 0-6.5 and
-0.4 kPa, respectively. Conversion of the reactants was
0
restricted to <15% in order to avoid mass transfer limitations.
An AMEL 553 galvanostat-potentiostat was used to carry
out measurements in both galvanostatic and potentiostatic
modes. In the galvanostatic mode, constant current, I, was
applied between the catalyst (W) and the counter electrode (C)
while monitoring the potential VWR between the catalyst and
the reference electrode (R).8 This gave information about the
time-dependent behavior of the system. The galvanostatic
transient behavior of VWR was also used to calibrate the Na
coverage scale.5 As noted above, changes in catalyst potential,
VWR, are related to concomitant changes in catalyst work
function according to e∆VWR ) ∆(eΦ) as predicted theoreti-
,15
Figure 1. Galvanostatic transients. Showing response of mass 44 rate
(N
2
O + CO
2 2
) (solid line), N rate (broken line), and catalyst potential
1
16
cally and confirmed experimentally.
(dotted line), in response to a step change in constant applied current
0
0
.
Conditions: (a) inlet partial pressures P NO ) 1.3 kPa, P C3H ) 0.6
In the potentiostatic mode, a constant potential VWR was
maintained between the catalyst and the reference electrodesthis
gave information about activity, selectivity, and kinetics under
steady-state conditions. Most experiments were carried out in
the potentiostatic mode by following the effect of the applied
catalyst potential (VWR) on reaction rates. Periodic reversal of
6
kPa, T ) 648 K; (b) P NO ) 0.8 kPa, P0C3H6 ) 0.4 kPa, T ) 648 K.
0
within 5% were found by a combination of GC and mass
spectroscopic analyses.
Transient Electrochemical Promotion. In order to obtain
an estimate of the Na coverage corresponding to any given value
of VWR, galvanostatic transients were obtained and analyzed as
+
the direction of Na pumping (required in order to regenerate
a clean Pt surface on the working electrode) appeared to avoid
possible difficulties due to Na depletion of the electrolyte. In
this connection, note that an amount of sodium equivalent to
θNa ∼ 3 corresponds to 0.04% of the total amount of Na in the
electrolyte wafer.
1
5
described previously by Vayenas et al. Such transients also
provide a useful means of rapidly mapping out the dependence
of reaction rates on Na coverage in advance of the much more
detailed and laborious steady-state measurements.
XPS measurements were carried out in a VG ADES 400
ultrahigh-vacuum spectrometer system. The EP sample was
mounted on a machinable ceramic block resistively heated by
embedded, electrically insulated tungsten filaments. XP spectra
were acquired with Mg KR radiation with the Pt catalyst
electrode always at ground potential; appropriate electrochemical
potentials (VWC) were imposed between the Pt working electrode
and the Au counter electrode by applying voltage bias to the
latter. The potential (VWR) of the Pt working electrode with
respect to the Au reference electrode was also monitored.
Quoted binding energies are referred to the Au 4f7/2 emission
at 83.8 eV, and Au reference spectra were provided by the Au
wire which formed the electrical connection to the working
electrode. The EP catalyst assembly used for XPS study
measurements was first tested in the EP reactor to ensure that
it exhibited the same catalytic behavior as the samples used to
acquire the reactor data.
Figure 1a shows a typical galvanostatic transient: it depicts
the effect of applying a constant negative current (Na supply
to the catalyst) on the catalyst potential, VWR, and on the m/z )
2
8 (N2) and 44 (CO2 + N2O) signals. The experimental
procedure was as follows: first (t e 0), the surface was
electrochemically cleaned of Na by application of a positive
potential (VWR) of the order of 300 mV until the current between
the catalyst and counter electrode vanished. This current
corresponds to the reaction
+
-
Na(Pt) f Na (â′′-Al O ) + e
(1)
2
3
The potentiostat was then disconnected (I ) 0 at t ) -1 min)
and VWR relaxed to the value imposed by the gaseous composi-
tion. Subsequently, the galvanostat was used to impose a
constant cathodic current I ) -100 µA at t ) 0, thus pumping
-
9
Na to the catalyst surface at a rate I/F ) 1.04 × 10 mol of
Na/s. In order to construct an abscissa corresponding to the
Na coverage on the Pt surface (θNa, see Figure 1), θNa was
computed from Faraday’s law according to
Results
The only detectable reaction products were CO2, N2, N2O
and H2O. Overall carbon and nitrogen mass balance closures

Electrochemical Promotion of the Pt/Propene + NO Reaction
J. Phys. Chem. B, Vol. 101, No. 19, 1997 3761
dθNa
-I
FN
)
(2)
dt
-
7
where N ()5.0 × 10 mol) is the number of available Pt sites
independently measured as described above. Under these
conditions (Na pumping to the catalyst surface), a pronounced
decrease in the catalyst potential VWR, and consequently in the
catalyst work function (e∆VWR ) ∆eΦ) occurred. This was
accompanied by a very marked increase (by a factor of 7) in
the production rate of N2. As is apparent from Figure 1, the
reaction rates (r) exhibit maxima for θNa ∼ 0.3-0.4 at VWR ∼
-300 mV, and continued Na pumping eventually leads to
poisoning of the system. The m/z ) 44 (CO2 + N2O) signal is
given in arbitrary units because the separate contributions of
CO2 and N2O could not be measured by mass spectrometry
alone. Potentiostatic imposition of the initial VWR restored the
rates to their initial unpromoted values, corresponding to clean
Pt, thus demonstrating that the system was perfectly reversible.
In general, the relationship between VWR and θNa depends
5
,8
on the gaseous composition as discussed previously.
For
present purposes, it is sufficient to note that the VWR/θNa
calibration (Figure 1a) was performed in the identical gas
atmosphere to that used in the steady-state measurements to be
described below. Figure 1b shows a transient obtained using
the same reactant partial pressure ratio as that in Figure 1a, but
with a higher current and smaller reactant pressures. In line
with expectation the maximum is more pronounced in this case,
and it is interesting to note that it occurs at essentially the same
value of θNa as that found in Figure 1a.
It should be emphasized that the “θNa” scales in Figure 1a,b
are computed on the basis of uniform coverage of the entire Pt
surface by the relevant amount of electrochemically supplied
Na. That is, “θNa” is proportional to the total amount of pumped
Na. If the morphology of the promoter phase is nonuniform,
e.g., 3D islands of Na compound(s) coexisting with a submono-
layer film of promoter species on the Pt surface, θNa > 1 would
still correspond to a catalytically active system. The data in
Figure 1a,b suggest that this is indeed the casesnominal
coverages of greater than unity exhibit activity well in excess
of clean Pt because a substantial amount of the promoter material
is present as 3D crystallites. XPS data (see below) are in accord
with this view. A useful check is provided by ∆VWR transients
measured in He as opposed to the reaction gas. These show
monotonic behavior similar to that shown in Figures 1a,b.
Additionally, the behavior of ∆VWR () ∆Φ) as a function of
computed θNa is in good quantitative agreement with results
17
obtained for Pt{111}/Na in ultrahigh vacuum. This indicates
that in a He atmosphere one is dealing with metallic Na
uniformly distributed on the Pt surface.
A convenient parameter for expressing electrochemically
induced promoting or poisoning effects is the promotion index,
PNa, defined18 as
PNa ) (∆r/r )/∆θ
(3)
0
Na
This takes positive or negative values for promotion or
poisoning, respectively. Here, PNa values up to 20 were found
for the rate of N2 production in the promoted regime. This is
followed by a poisoned regime for θNa > 0.5. This “volcano
type” behavior is also apparent in the potentiostatic, steady-
state data to be presented in the following section. Similar
promoting-poisoning behavior of Na has been also observed
Figure 2. Effect of catalyst potential, VWR on (a) CO , (b) N , (c)
2
2
0
N
0
2
O production rates and on N
.6 kPa; P0NO ) 1.3 kPa, T ) 648 K, total flow rate F
mol/s.
2
selectivity (d). Conditions: P C3H6 )
-4
t
) 1.7 × 10
data for CO2, N2, and N2O production as a function of VWR at
5
13
0
0
in the cases of CO and ethylene oxidation and NO reduction
648 K for constant inlet pressures (P NO, P C H ) of NO and
3 6
8
by CO over Pt/â′′-alumina under appropriate conditions.
C3H6.
Steady-State Effect of Catalyst Potential (VWR) on Reac-
Figure 2d depicts the VWR dependence of the selectivity
tion Rates. Parts a, b, and c of Figure 2 show steady-state rate
toward N2 where the latter quantity is defined as

3762 J. Phys. Chem. B, Vol. 101, No. 19, 1997
Yentekakis et al.
SN2 ) r /(r + r
)
N O
2
(4)
N
N2
2
The data presented in Figure 2a,b show that, for both the
CO2 and N2 rates, a regime of strong promotion is followed by
strong poisoning. Relatively low rates are observed at both very
positive (clean Pt) and very negative (high Na loading) catalyst
potentials. In particular, both rates exhibit an exponential
increase with decreasing VWR in the interval ∼0 to -300 mV:
this is made more apparent by a logarithmic plot, e.g., as shown
in the inset to Figure 2b. The rate of N2O production also
exhibits “volcano-type” behavior, though the effect is not as
pronounced (Figure 2c)sthe point is of significance with respect
to our subsequent discussion of the reaction mechanism. The
maximum enhancements in CO2, and N2 rates over the clean
surface rates are ∼7, and for N2O ∼3. Note also that the
selectivity toward N2 increases from 60% on the clean Pt surface
to 80% on the Na-promoted Pt surface at VWR ∼-300 mV.
Control experiments were carried out in which the total flow
rate was varied by a factor of 2 in order to verify that the
observed changes in activity were due to a true increase in
surface reaction rates and not due to mass transfer limitations
in the reactor. As found in previous EP studies with Pt/â′′-
alumina5 for any given reaction conditions, the promotional
effect is entirely controlled by the applied potential and the
resulting Na coverage. After current interruption the promoted
rates remained practically constant over periods of many
minutessthe slow decline that occurred under open circuit
conditions is associated with various possible loss mechanisms
for the surface Na species, e.g., slow evaporation at reaction
temperature. In order to restore the initial unpromoted catalytic
rate, potentiostatic imposition of the initial potential of +300
mV was necessary.5 It is significant that in a reaction gas
environment the total amount of Na pumped in order to attain
a given value of VWR was about an order of magnitude higher
than in a He environment. This is consistent with the ac-
cumulation of the Na into 3D islands of a surface compound in
the presence of a reactive atmosphere.
,13
Effect of PC H and PNO on Promoted and Unpromoted
3
6
Steady-State Rates. Parts a, b, and c of Figure 3 show 648 K
potentiostatic data for the rates and turnover frequencies (TOFs)
of CO2, N2, and N2O taken at four different fixed values of
catalyst potential as a function of the outlet pressure of propene,
for a fixed NO outlet pressure of 1.4 kPa. Figure 3d shows the
corresponding results for the N2 selectivity, SN . In every case
2
the system was perfectly reversible and exhibited no hysteresis
when switching between negative and positive catalyst poten-
tials.
The observed rate variations show that the reaction obeys
Langmuir-Hinshelwood (LH) type kinetics with the charac-
teristic rate maximum reflecting competitive adsorption of the
two reactants. It is also apparent that as the catalyst potential
is decreased to more negative values (increasing θNa), there is
a systematic shift of the rate maxima to higher propene partial
*
pressure (P C H ). Figure 4a-d depicts analogous results for
3
6
the effect of NO at four different catalyst potentials for fixed
PC H ) 0.27 kPa. The CO2 and N2 rates show essentially
3
6
similar behavior as PNO is variedsa pronounced enhancement
in catalytic activity occurs upon decreasing the catalyst potential
from 300 to -300 mV. In this case the rate maxima are
inaccessible within the constraints imposed by our experimental
conditionsshere, one expects the opposite effect, namely that
increasing levels of Na should shift the (inaccessible) value of
P NO to lower values. A comparison of the trends shown in
Figures 3 and 4 indicates that under these conditions the
adsorption of propene on Pt is stronger than that of NO.
Figure 3. Effect of C H partial pressure (PC3H6) on the rate of (a)
3
6
2 2 2 2
CO , (b) N , (c) N O production and on N selectivity SN2, (d) at
different catalyst potentials, VWR. Conditions: PNO ) 1.4 kPa, T )
-
4
6
48 K, total flow rate F
t
) 1.3 × 10 mol/s.
Effect of Temperature on Promoted and Unpromoted
*
Steady-State Rates. It is of interest to compare the temperature
dependent-behavior of the chemically complex NO + propene
system with the simpler NO + CO system and the even simpler

Electrochemical Promotion of the Pt/Propene + NO Reaction
J. Phys. Chem. B, Vol. 101, No. 19, 1997 3763
Figure 5. Effect of catalyst potential on the apparent activation energy
for (a) CO , (b) N , and (c) N O production.
2 2 2
in the data, two features are apparent: (i) the activation energy
of all the three reactions increases on going from high positive
to high negative potential values; (ii) the activation energy
associated with the promoted reaction regime increases mono-
tonically with the Na loading. However, the variation in
activation energy with increasing Na coverage is neither abrupt
5
nor pronounced as it is in the case of the EP CO + O2 and EP
8
CO + NO systems.
XPS Studies of the EP Catalyst. XP spectra were obtained
in order to shed light on several important issues. (i) What
evidence is there for the reversible transport of Na between
electrolyte and catalyst under vacuum conditions? (ii) Does
the chemical species transported differ in behavior from vacuum-
deposited Na? (iii) What surface species are formed by Na
pumping to Pt at reaction temperature and in the presence of
the releVant partial pressures of the reactant gases? (iv) What
differences are there between the promoted and poisoned
regimes? (v) What is temperature stability of the promoter
phase?
Figure 6A shows Na 1s spectra acquired at 600 K in ultrahigh
vacuum (UHV). The electrochemically cleaned surface (VWR
)
+500 mV) exhibits a feature at 1074 eV BE which is due to
Na in the â′′-alumina visible through cracks in the porous Pt
film; this is confirmed by the corresponding O 1s spectra, and
similar spectra have been reported for electrochemically cleaned
1
9
Pt films on yttria-stabilized zirconia. At VWR ) -600 mV,
two features are now present, that at 1073.5 eV being due to
Na pumped to the Pt surface. Note that the â′′-alumina-derived
peak has shifted to ∼1075.5 eV, i.e., by an amount about equal
to the change in VWR (1.1 eV). In measurements such as these,
one expects ∆(apparent BE of Na in electrolyte) ) ∆VWR,
because the latter is the change in overpotential between the
electrolyte and the electrode due to the electrochemical double
layer. The Pt film is at ground potential, and Na 1s photo-
emission from the Pt should of course appear at fixed kinetic
energy, independent of VWR. Thus, our XP spectra clearly reveal
the location of and relationship between the two types of Na
sampled by photoemission. For intermediate values of VWR the
amount of Na on Pt varies monotonically with catalyst potential,
and the (constant intensity) Na 1s emission from the â′′-alumina
shifts with VWR. This spectral behavior was reproducible and
reversible as a function of VWR.
Figure 4. Effect of NO partial pressure (PNO) on the rate of (a) CO
b) N , (c) N O production and on N selectivity S 2 (d) at different
catalyst potentials VWR. Conditions: PC3H ) 0.27 kPa, T ) 648 K,
2
,
(
2
2
2
N
6
-
4
total flow rate, F
t
) 1.3 × 10 mol/s.
Figure 6B demonstrates that, under UHV conditions, elec-
tropumped Na is identical in behavior and in chemical state
with Na supplied by vacuum deposition from a Na evaporation
source. Control experiments carried out with a Pt{111} sample
in the same UHV chamber indicated that the amount of Na
deposited on the catalyst film at 300 K was approximately 1015
O2 + CO system, the latter two having also been studied under
EP conditions over Pt/â′′-alumina. Accordingly, Figure 5 shows
data extracted from Arrhenius plots over the temperature interval
6
00-675 K for the CO2, N2, and N2O production rates as a
function of catalyst potential. Although there is some scatter

3764 J. Phys. Chem. B, Vol. 101, No. 19, 1997
Yentekakis et al.
the electrochemically cleaned surface, with the promoted surface
level lying in between. Figure 7c shows the corresponding Na
KLL X-ray-excited Auger spectrasnotice that the difference in
integrated intensities between the promoted and poisoned cases
is much more apparent here. It is important to note that the
absolute intensity of the Pt 4f spectrum was measurably
attenuated on both the promoted and poisoned surfaces relative
to the electrochemically cleaned case (∼6 and ∼15%, respec-
tively), indicating a substantially higher loading of Na surface
compounds in the latter case. Based on our Pt{111}/Na
calibration, the estimated Na loadings of the promoted and
1
5
15
-2
poisoned surfaces are g1.5 × 10 and g4.5 × 10 cm , i.e.,
equivalent to θNa g 1 and g 3, respectively. The former value
again implies that even in the promoted case, a substantial
proportion of the Na compound(s) must be present in 3D islands,
thus leaving a significant amount of uncovered (promoted) Pt
surface on which the reaction takes place. That is, one has
Volmer-Weber growth of the promoter on the metal surface:
the driving force behind agglomeration comes partly from the
3
D Madelung energy of the polar crystals that are formed.
Analogous C 1s spectra are shown in Figure 7d: carbon-
containing species are detectable only on the clean and promoted
surfaces. Finally, Figure 7e shows O 1s results. The behavior
is similar to that of the N 1s spectra: the promoted surface
shows no detectable lines; the poisoned surface shows peaks at
5
31 and 535 eV BE, and these are quenched by 10 h exposure
to the ambient vacuum at 300 K.
As will be argued below, the data in Figure 7 are to varying
extents influenced by the sample handling procedure and
especially by continuing reactions that occurred during reactor
pump-out and sample cool-down, prior to transfer into the
spectrometer. In order to characterize the promoter/poison
phase(s), their temperature stability and their reactivity in more
detail, XP spectra were also acquired after sequential exposure
of the sample to NO followed by propene. The results of this
approach are illustrated in Figure 8. The spectra in Figure 8a
were acquired as a function of temperature under open circuit
conditions after gas pump-out. The spectra in Figure 8b,c were
acquired as a function of gas exposure at 485 K under open
circuit conditions; data were recorded at 485 K, also under open
circuit conditions, so as to avoid any voltage/temperature-
assisted effects. The object of this procedure was to “freeze
out” the catalyst surface in a condition that captured at least
some aspects of its properties under actual EP reaction condi-
tions. Figure 8a shows temperature-dependent effects, as
follows. NO dosing at 600 K leads to the formation of two
distinct sodium nitroxy surface compounds. The 405 eV BE
is assigned to NaNO2 (as above in regard to Figure 7A); the
Figure 6. Na 1s XPS. (a) Showing effect of catalyst potential in
pumping sodium to/from the Pt film under UHV conditions at 600 K.
(b). Demonstrating electrochemical pumping of vacuum-deposited Na
away from the surface under UHV conditions at sufficiently high
temperature and positive catalyst potential.
cm- (spectrum 1). Heating to 400 K under open circuit
conditions caused no change (spectrum 2). At 500 K (when
the â′′-alumina becomes appreciably conducting) and with VWR
2
)
+1000 mV, the vacuum-deposited Na was strongly pumped
away from the surface (spectrum 3).
The XP spectra in Figure 7 were obtained after exposing the
appropriately biased catalyst film to the conditions of temper-
ature and reactant partial pressures typical of those encountered
in the reactor (PC H ) 0.6 kPa, PNO ) 1.2 kPa, T ) 600 K).
21
3
6
409 eV BE is assigned to NaNO3. Heating to 500 K leads to
During exposure of the catalyst to the reaction mixture, the VWR
conditions were such that the Pt film was either (i) electro-
chemically clean, (ii) promoted, or (iii) poisoned. The spectra
were acquired after a pump-down period of 2 h. The sample
temperature was lowered to 420 K before pump-down, and the
imposed voltage bias (Vwc) was switched off when the sample
temperature was below 450 K, i.e., when the Na mobility was
low. It is apparent from the N 1s spectra (Figure 7A) that no
detectable N-containing species were retained on either the clean
or the promoted Pt film. However, the poisoned film shows a
very significant line at 405.6 eV, which we ascribe2 to sodium
nitrite (NaNO2). Note in the figure that exposure of this surface
to the base vacuum for 10 h at 420 K resulted in removal of
the nitrite, presumably due to slow reaction with ambient
reducing gases. Figure 7B (upper) shows corresponding (raw)
Na 1s data from which it is clear that the loading of Na
compounds is highest on the poisoned surface and smallest on
a definite (∼25%) increase in intensity of the nitrite emission
while the nitrate emission is constant within experimental error.
We attribute this to increased dispersion of the nitrite but not
the nitrate (consistent with the melting points of the bulk com-
pounds: 544 and 580 K respectively). After 520 K heating
the nitrite begins to diminish relative to the nitrate, and after
5
30 K heating the nitrite intensity is the smaller of the
twosagain, consistent with the decomposition temperatures of
the bulk compounds (590 and 650 K).
0
Discussion
As noted above, the term “Na coverage” does not imply that
the promoter is thought to be present in the form of chemisorbed
metallic sodium, as it would be in vacuum.
In the reaction mixture one expects the formation of surface
compounds of Nasin the present case, carbonate and nitrate

Electrochemical Promotion of the Pt/Propene + NO Reaction
J. Phys. Chem. B, Vol. 101, No. 19, 1997 3765
Figure 7. XP spectra of catalyst film after exposure to reaction mixture for 15 min (PC3H6 ) 0.6 k Pa, PNO) 1.2 k Pa, T ) 600 K). The results refer
to electrochemically cleaned, promoted, and poisoned conditions. (a) N 1s. (b) Na 1s. 1, +600 mV; 2, -250 mV; 3, -250 mV, after exposure to
reaction mixture; 4, -500 mV; 5, -500 mV, after exposure to reaction mixture. (c) Na KLL XAES. 1-5, same conditions as (b). (d) C 1s. (e) O
1
s. (a) and (e) also show the effect of continued reaction with ambient.
are plausible possibilities; the XP spectra provide relevant
information. For present purposes, we note that adsorbed polar
alkali compounds lead to large decreases in work function
relative to the clean metal which are of the same order as those
produced by the pure alkali metal itself. Substantial changes
in catalyst potential are therefore still to be expected under
electropumping in a reactive atmosphere.
promotion under Na pumping to the catalyst (Figure 2). As
discussed below and suggested earlier in connection with our
8
earlier work on EP of the CO + NO reaction, we infer that the
promotional effect is due to enhanced NO chemisorption and
dissociation on the Pt surface induced by electrochemically
pumped Na. Recall that the polycrystalline Pt film consists
mainly of large crystallites (∼1 µm) whose external surfaces
are dominated by low index planes: such low index planes of
NO reduction by propene exhibits strong electrochemical

3766 J. Phys. Chem. B, Vol. 101, No. 19, 1997
Yentekakis et al.
promotion (0 to -350 mV), the activity toward formation of
all products shows an exponential increase with (decreasing)
catalyst potential, as predicted by the model of Vayenas.1
Eventually, at around -350 mV, there is a precipitous fall in
rates as the amount of Na species increases beyond a critical
value; i.e., the regime of electrochemical promotion is followed
by a regime of strong poisoning. This poisoning behavior is
discussed below.
,3
The dependence of activity and selectivity on catalyst
potential may be rationalized by considering the effects of the
Na promoter species on the surface chemistry of propene and
NO. At high positive VWR the Pt surface is free of sodium and
covered mainly with propene: the relative strengths of propene
and NO chemisorption on Pt are apparent from Figures 3 and
4
. We postulate that the low clean surface residual rate is due
to dissociation of NO at defect sites; as noted above, we take
the view that NOa f Na + Oa is the reaction-initiating step.
As the catalyst potential is decreased toward negative VWR
values (pumping Na to the Pt), the electronic effect of the
sodium promoter strengthens the Pt-N bond (increasing NO
coverage) and weakens the N-O bond (facilitating NO dis-
sociation). The dissociation of chemisorbed diatomic molecules
in the field of coadsorbed cations has been discussed in detail
2
3
by Lang et al. More recently, the Na-induced dissociation of
2
4
NO on Pt{111} has been demonstrated directly. Thus, the
effect of the promoter is to activate previously inactive crystal
planes on the catalyst surface toward NO dissociation. This
effect increases progressively with the amount of Na pumped,
and the exponential dependence of rate on VWR in the promoted
regime suggests that changes in catalyst work function are
indeed strongly correlated with the dramatically increased
1
activity. In addition, Na pumping also enhances the selectivity
toward N2 formation from 60 to 80% (Figure 2). This quantity
is determined by the competition between the following
reactions that occur on the Pt surface.
N + N f N
2
a
a
N + NO f N O
(5)
a
a
2
The observed increase in selectivity is a consequence of the
Na-induced decrease and increase, respectively, in the amounts
of molecularly adsorbed NO and atomic N on the surface, thus
favoring the first reaction over the second. Qualitatively similar
effects on SN2 have been found for the electrochemically
8
6
promoted reduction of NO by CO and by H2 with the same
catalyst system.
The systematic effect of VWR on the reaction kinetics (Figures
3
and 4) is also understandable in terms of the effects of the
Na promoter. For the range of VWR values investigated in
Figures 3 and 4 (clean surface, promoted, and further promoted),
EP by Na leads to activity enhancement under all conditions of
partial pressure. This is consistent with the arguments developed
*
above. The systematic shift of P C H to higher propene partial
Figure 8. N 1s spectra (open circuit) after exposure to NO at 600 K
and -500 mV, showing the effect of (a) temperature: 1, clean, 600 K;
3 6
pressures (Figure 3) as the sodium coverage is increased reflects
an increase in chemisorption bond strength of NO relative to
propene with increasing Na coverage. Such behavior is exactly
what one would expect in the case of an electropositive
promoter: the chemisorption strength of electron donors (pro-
pene) should be decreased, whereas the chemisorption of
electron acceptors (NO and its dissociation products) should
be enhanced.
The systematic increase in activation energies with increasing
Na loading deserves some comment. In our earlier EP studies
of the CO + O2 and CO + NO reactions over Pt/â′′-alumina,
a pronounced step increase in Ea was observed as the amount
2
, 470 K; 3, 485 K; 4, 510 K; 5, 530 K; 6, 560 K. (b) differential
reactivity of nitrate and nitrite species toward propene dosed at 485 K:
3
3
6
1
, as 6 in (a); 2, 10 langmuirs; 3, 6 × 10 langmuirs; 4, 3 × 10
langmuirs; (c) O1s spectra corresponding to (b).
Pt are known to be relatively ineffective for NO dissociation.22
Na-induced NO dissociation produces the Oa species which are
then responsible for initiating the ensuing oxidation reactions
of adsorbed propene and propene fragments. The maximum
gains in rate over the clean surface rates are of the order of
factors of 13 and 25 for N2 and CO2 production, respectively
5
8
(Figures 3 and 4). In the region of strong electrochemical

Electrochemical Promotion of the Pt/Propene + NO Reaction
J. Phys. Chem. B, Vol. 101, No. 19, 1997 3767
5
of Na increased. There is persuasive evidence that this step
be detected after withdrawal of a poisoned catalyst from the
NO + propene mixture. In this case, the nitrate was reacted
away before XP analysis was possible. With the promoted
sample, on which both compounds are present in much smaller
amounts, neither is detectable after transfer from the reaction
mixture. The Na 1s results (Figure 7b) confirm that the
poisoned surface is more heavily loaded with Na compounds
than the promoted surface; however, the effect is more apparent
in the Na KLL Auger spectra shown in Figure 7c. This reflects
the very different sampling depth in the two cases (electron
kinetic energies ∼180 and ∼990 eV, respectively). The
conclusion is that in the poisoned case the Na compounds are
present as very substantial 3D crystallitessthis “bulk” material,
which presumably covers most of the Pt surface, would
contribute significantly to the total intensity of the KLL Auger
spectrum but would contribute to the Na 1s intensity far less.
The C 1s spectra (Figure 7d) show relatively little carbon
retention on the poisoned surface, presumably due to extensive
coverage by sodium nitrite/sodium nitrate. The electrochemi-
cally cleaned surface accumulated some graphitic carbon, and
on the promoted surface the 286.2 eV BE emission indicates
change is associated with a surface phase transitionsthe
formation or disruption of islands of CO. Although CO has
been detected as a surface intermediate in the Pt-catalyzed
oxidation of ethylene,13 one may speculate that the absence of
a step change in the present case is not unexpected, given the
greater chemical complexity of the adsorbed phase. It is
possible that the systematically increasing Ea values are at least
partly associated with Na-induced increase in chemisorption
strength of the three electron-withdrawing adsorbates (Na, Oa,
NOa), though the complexity of the system precludes detailed
analysis. The increase in Eas is accompanied by large increases
in the preexponential factors (thus giving the net rate accelera-
tion) which may plausibly be associated at least in part with an
increased density of sites for NO dissociation due to the presence
of Na.
The strong poisoning behavior observed at high Na loadings
is understandable in terms of the XPS results (see below) which
indicate the presence of large amounts of Na compounds at very
negative catalyst potentials. In other words, coverage of active
Pt sites by overloading with promoter is a likely major cause
of poisoning. Additionally, as noted above, the chemisorption
bonds of propene and oxygen should be respectively weakened
and strengthened by an electropositive promoter. Therefore,
at high levels of Na the coverage of propene should be markedly
attenuated, while the activation energy for reactions involving
Oa could be increased.
2
1a
the presence of CHx species spresumably resulting from
dissociation and partial oxidation of propene on the promoter-
activated surface. Figure 7e shows that the poisoned surface
exhibits O 1s emission at 531 and 535 eV BE. The latter is
21b
due to the nitrate species; the associated emission is quenched
by standing in ambient vacuum for 10 h, in line with the
behavior of the corresponding N 1s emission (Figure 7a). The
origin of the 531 eV BE signal is less clearsgiven the state of
the surface and the sample handling procedures, it is unlikely
to be due to Oa on Pt. It could possibly be due to oxidic oxygen
associated with Na, although on the basis of the present results
we cannot be sure about this. As noted in the Results section,
Figures 8a,b nicely shows the different thermal stabilities and
reducibilities of the nitrate and nitrite species. These are in
accord with the corresponding properties of the bulk materials
and the detectable redispersion of the nitrite at 520 K suggest
that the nitrite phase might be very mobile at reaction temper-
ature. The O 1s results in Figure 8c confirm that the nitrite
and nitrate are reduced by propene. The 532 eV peak
corresponds to that also observed in the postreaction experiments
The central assumption underlying all of the preceding
discussion is that, under EP conditions, reversible changes in
VWR correspond to the reversible pumping of Na to/from the Pt
from/to the solid electrolyte. Our XPS data (Figure 6a) clearly
demonstrate that such reversible transport of Na between â′′-
alumina and the gas-exposed surface of the Pt film does indeed
occur under the conditions of voltage and temperature that were
used for the reactor studies. This is an important observation
that substantiates our view of the mode of action of the EP
system. Additionally, the identical electrotransport properties
and electron binding energies exhibited by vacuum deposited
Na and by electropumped Na in vacuum clearly establish that
these are the same chemical species. This is an important point
and should help to clarify apparent misunderstandings that
sometimes arise about the chemical nature of alkali promoters
on metal catalyst surfaces. The mode of delivery of pure Na
to the Pt surface is irrelevant: in vacuum it is present as an
(Figure 7e), tentatively assigned above to “oxidic” oxygen;
clearly, it is much more resistant to reduction than the nitrate
and nitrite. The 533 eV BE shoulder observed in spectrum 4
after the largest dose of propene could be assigned to carbonate;
some support for this suggestion comes from the recent
observation that COa is observed as a surface intermediate during
the oxidation of ethylene on Na-promoted Pt{111}.
adatom that has undergone significant charge transfer to the
δ-
(
in this case) Pt surface, Naδ+/Pt . The presence of a reactive
gas atmosphere converts this Na into surface compound(s)
whose composition and degree of dispersion depend on the
temperature and the composition of the gas phase, as discussed
below. We have found that these surface compounds can also
be electrochemically destroyed by pumping Na away from the
Pt film.
Finally, we draw attention to an interesting question. The
observation that the surface nitrate is efficiently reduced by
-
-
propene raises the possibility that NO3 /NO2 may constitute
a redox couple that is involved in the catalytic turnover, thus
The XPS results reveal the identity of the principal surface
chemical compounds formed during Na electropumping in the
reaction gas. It seems clear that both NaNO2 and NaNO3 are
formed, and we accordingly identify nitrite and nitrate as the
main counterions which, with Na, constitute the promoter phase.
This identification depends on the observed N 1s binding
energies and the relative tendencies toward wetting and decom-
position exhibited by the two nitroxy compounds: the nitrate
is more thermally stable than the nitrite (Figure 8A). The
assignment is strengthened by the observed differences in
reactivity towards propene: the nitrate is more chemically
reactive than the nitrite (Figure 8B). This is also consistent
with the observation (Figure 7A) that only nitrite survives to
hydrocarbon + nitrate > oxidation product + nitrite
nitrite + O (from NO dissociation) > nitrate
a
If this were the case, it would mean that we are dealing with a
promoter system in which the cation triggers the primary
chemistry while the anion facilitates subsequent oxidative
reactions. Further work is in progress to investigate this
possibility.
Conclusions
1. The catalytic reduction of NO over Pt by propene exhibits
strong electrochemical promotion when Na from â′′-alumina

3768 J. Phys. Chem. B, Vol. 101, No. 19, 1997
Yentekakis et al.
is pumped to the catalyst surface. The effects are reversible
with catalyst potential and rate gains of about an order of
magnitude over the clean surface rates can be achieved. At
sufficiently high Na loadings the reaction is poisoned.
(3) Bebelis, S.; Vayenas, C. G. J. Catal. 1989, 118, 125.
(
(
4) Yentekakis, I. V.; Bebelis, S. J. Catal. 1992, 137, 278.
5) Yentekakis, I. V.; Moggridge, G. D.; Vayenas, C. G.; Lambert, R.
M. J. Catal. 1994, 146, 292.
6) Marina, O. A.; Yentekakis, I. V.; Vayenas, C. G.; Palermo, A.;
(
2
. All the data are consistent with Na-induced NO dissocia-
tion being the critical reaction-initiating step.
. The selectivity toward N2 formation Versus N2O produc-
Lambert, R. M. J. Catal., in press.
(7) Lambert, R. M.; Tikhov, M.; Palermo, A.; Yentekakis, I. V.;
Vayenas, C. G. Ionics 1995, 5, 366.
3
(8) Palermo, A.; Lambert, R. M.; Harkness, I. R.; Yentekakis, I. V.;
tion always increases with decreasing catalyst potential (increas-
ing Na coverage): this is understandable in terms of point 2.
4
electrochemically promoted Pt film does indeed involve revers-
ible pumping of Na to or from Pt from or to the solid electrolyte.
Under UHV conditions, this electropumped Na is indistinguish-
able from Na put down by vacuum deposition.
Marina, O.; Vayenas, C. G. J. Catal. 1996, 161, 471
(9) Yentekakis, I. V.; Vayenas, C. G. J. Catal. 1994, 149, 238.
(10) Politova, T. I.; Sobyanin, V. A.; Belyaev,V. D. React. Kinet. Catal.
Lett. 1990, 41, 321.
. XPS data confirm that the mode of operation of the
(
11) Taylor, K. C. Catal. ReV.sSci. Eng. 1993, 35, 457.
(12) Yentekakis, I. V.; Neophytides, S.; Vayenas, C. G. J. Catal. 1988,
111, 152.
(13) Harkness, I. R.; Hardacre, C.; Lambert, R. M.; Yentekakis, I. V.;
Vayenas, C. G. J. Catal. 1996, 160, 19.
5
. Under reaction conditions the Na promoter is present
(14) Harkness, I. R.; Lambert, R. M. J. Catal. 1995, 152, 211.
mainly as a mixture of nitrate and nitrite; it is possible that some
form of sodium oxide is also present. The nitrate is thermally
more stable but also more easily reduced than the nitrite. The
loading of promoter compounds is relatively low under pro-
moted conditions and much higher under poisoned conditions.
A substantial fraction of the Na compounds is present as 3D
crystallites under all reaction conditions.
(
15) Vayenas, C. G.; Bebelis, S.; Despotopoulou, M. J. Catal. 1991,
1
28, 415.
(16) Vayenas, C. G.; Bebelis, S.; Ladas, S. Nature 1990, 343, 625.
(17) Bonzel, H. P.; Pirug, G.; Ritke, C. Langmuir 1991, 7, 3006.
(18) Cavalca, C.; Larsen, G.; Vayenas, C. G.; Haller, G. L. J. Phys.Chem.
993, 97, 6115.
1
1
(
19) Ladas, S.; Kennou, S.; Bebelis, S.; Vayenas, C. G. J. Phys.Chem.
993, 97, 8845.
20) (a) Folkesson, B. Acta Chem. Scand. 1973, 27, 287. (b) Fuggle, F.
(
Acknowledgment. A.P. holds a Postdoctoral External Fel-
lowship, CONICET. A.P. and R.M.L. acknowledge support by
the British Council and Fundaci o´ n Antorchas. Financial support
from the UK Engineering and Physical Sciences Research
Council under Grant GR/K45562 is gratefully acknowledged.
G.; Menzel, D. Surf. Sci. 1984, 79, 1. (c) Kiskinova, M.; Pirug, G.; Bonzel,
H. P. Surf. Sci. 1983, 140, 1.
(21) (a) Siegban, K.; Nordling, C.; Fahlman, A.; Nordberg, R.; Hamrin,
K.; Hedman, J.; Johansson, G.; Bergmark, T.; Karlsson, S.-E.; Lindgren,
I.; Lindberg, B. ESCA-atomic, molecular and solid state structure studies
by means of electron spectroscopy; Nova Acta Regiae Societatis Scientarium
Upsaliensis, ser. IV, 1967; Vol. 20. (b) Allen, H. C.; Laux, J. M.; Vogt, R.;
Finlayson-Pitts, B. J.; Hemminger, J. C. J. Phys. Chem. 1996, 100, 6371.
References and Notes
(
22) Masel, R. I. Catal. ReV.sSci. Eng. 1986, 28 (2,3), 335.
(
1) Vayenas, C. G.; Bebelis, S.; Yentekakis, I. V.; Lintz, H.-G. Catal.
Today 1992, 11 (3), 303.
2) Pliangos, C.; Yentekakis, I. V. ; Verykios, X.; Vayenas, C. G. J.
Catal. 1995, 154, 124.
(23) Lang, N. D.; Holloway, S.; Norskov, J. K. Surf. Sci. 1985, 150,
24.
press.
(
(24) Harkness, I. R.; Lambert, R. M. J. Chem. Soc., Faraday Trans., in