Hydrogen cyanide synthesis on polycrystalline platinum and 90:10 platinum-rhodium surfaces

Article abstract of DOI:10.1039/a704184e

The synthesis of HCN from methane-based feedstocks has been studied at relatively low temperatures (up to 780°C) and pressures (0.02-0.15 Torr) over both pure Pt and Pt-10%Rh alloy model catalysts. Steady-state synthesis of HCN is observed over Pt from CH₄-NH₃-O₂ feed mixtures with a maximum in activity observed using a 1:1:0.5 mixing ratio. The activity of the Pt-10%Rh alloy under comparable conditions is significantly lower, indicating that Rh has an inhibiting effect on the synthesis reaction at these temperatures and pressures. Comparable steady-state levels of activity were not observed over Pt for pure CH₄-NH₃, CH₄-NH₃-CO₂ or CH₄-NO feed mixtures; in particular, for CH₄-NH3 feeds in the absence of oxygen a high initial activity was seen to decay rapidly over time. The effects of various surface pretreatments of the Pt were found to be transitory, with a gradual return of the HCN activity to its normal level in CH₄-NH₃-O₂ (1:1:0.5) mixtures in all cases, and the most beneficial pretreatment effect was observed for a simple physical surface roughening induced by mild ion bombardment. By contrast, pre-exposure of the catalyst surface to ethylene or ammonia at elevated temperatures led to an initial inhibition or decrease in the synthesis activity.

Full text of DOI:10.1039/a704184e

View Article Online / Journal Homepage / Table of Contents for this issue
Hydrogen cyanide synthesis on polycrystalline platinum and 90 : 10
platinum–rhodium surfaces
A. Bockholt, I.S. Harding and R.M. Nix*¤
Department of Chemistry, Queen Mary and W estÐeld College, University of L ondon, L ondon, UK
E1 4NS
The synthesis of HCN from methane-based feedstocks has been studied at relatively low temperatures (up to 780 ¡C) and pres-
sures (0.02È0.15 Torr) over both pure Pt and PtÈ10%Rh alloy model catalysts. Steady-state synthesis of HCN is observed over Pt
from CH ÈNH ÈO feed mixtures with a maximum in activity observed using a 1 : 1 : 0.5 mixing ratio. The activity of the
4
3
2
PtÈ10%Rh alloy under comparable conditions is signiÐcantly lower, indicating that Rh has an inhibiting e†ect on the synthesis
reaction at these temperatures and pressures. Comparable steady-state levels of activity were not observed over Pt for pure
CH ÈNH , CH ÈNH ÈCO or CH ÈNO feed mixtures; in particular, for CH ÈNH feeds in the absence of oxygen a high initial
4
3
4
3
2
4
4
3
activity was seen to decay rapidly over time. The e†ects of various surface pretreatments of the Pt were found to be transitory,
with a gradual return of the HCN activity to its normal level in CH ÈNH ÈO (1 : 1 : 0.5) mixtures in all cases, and the most
4
3
2
beneÐcial pretreatment e†ect was observed for a simple physical surface roughening induced by mild ion bombardment. By
contrast, pre-exposure of the catalyst surface to ethylene or ammonia at elevated temperatures led to an initial inhibition or
decrease in the synthesis activity.
Introduction
Experimental
The Andrussow process1 is
a
long-established catalytic
The apparatus used in these studies consisted of an ultra-high
vacuum (UHV) surface analysis system equipped with a high-
pressure catalytic reaction cell. The analytical chamber itself
was equipped for X-ray photoelectron spectroscopy (XPS),
low-energy electron di†raction (LEED), Auger electron spec-
troscopy (AES) and temperature-programmed desorption
(TPD): the XPS data reported below were obtained using
process for the industrial production of hydrogen cyanide,
which involves the formation of HCN from a near equimolar
mixture of CH , NH and O at 1È2 atm over a 90 : 10
4
3
2
Pt : Rh (PtÈ10%Rh) gauze catalyst under adiabatic conditions
at ca. 1400 K and with very short contact times. The alterna-
tive Degussa process also employs a Pt-based catalyst but, in
this case, the process is operated in the absence of oxygen
using a CH ÈNH feed at yet higher temperatures and exter-
Mg-Ka radiation from
a standard, non-monochromated
X-ray source and a VG-CLAM2 100 mm hemispherical
analyser which was operated in a Ðxed pass-energy mode for a
given spectral region (50 eV for C 1s, O 1s and N 1s; 20 eV for
Pt 4f and Rh 3d). The spectrometer was calibrated using a
CuÈAu reference sample and spectra were acquired at normal
emission angles; quoted binding energies are taken directly
from the intensity maxima of the raw data.
4
3
nal heating provided by methane combustion.
Previous studies of these reactions have included pilot plant
and laboratory scale reactor studies, detailed analyses of dis-
charged catalyst gauzes and more fundamental surface science
studies of the interaction of reactants, products and potential
intermediates with platinum surfaces.2h7 The detailed kinetics
of HCN synthesis over Pt and Rh foils have also been studied
at reduced pressures and the commercial process modelled by
the group of Schmidt in Minnesota.8h11 Despite this earlier
work, however, there is still considerable uncertainty over the
detailed role of Rh, the main synthesis pathways, the nature of
the active catalyst surface and the identity of key surface inter-
mediates.
In the approach adopted for the experiments described
below, we have examined the synthesis reaction on model
catalyst samples (Ñat polycrystalline foils) of pure Pt and a
PtÈ10%Rh alloy at relatively low pressure in a reaction cell
attached to a surface analysis system. The advantages of this
approach are that it permits the catalystÏs surface temperature
to be varied independently of the feed gas composition and
Ñow rate, and for the reaction to be studied under conditions
where the contribution of any homogenous processes and
mass transport limitations are minimized. The state of the
catalyst may also be accurately assessed, both prior to and
after reaction, by the application of surface analytical tech-
niques, and the catalyst may also be manipulated by the direct
additions of controlled levels of poisons and promoters.
Samples could be transferred directly from the main UHV
system into the high-pressure reaction cell using a magneti-
cally coupled translator. This reaction cell could be isolated
from the main chamber by a gate valve and separately
pumped by a liquid-nitrogen trapped di†usion pump; in
normal operation pumping was via a high-conductance valve
but for reaction studies pumping was normally limited to a
low conductance bypass with in-line regulating valve. The
reactor gas feed was pre-mixed in a special high vacuum
chamber connected to the main gas-handling line. The inlet
bleed valve from this chamber to the reaction cell and the
aforementioned pump regulation valve together provided
control over both the gas pressure and gas residence time in
the reactor. At the pressures employed in this study (typically
0.02È0.15 Torr) the reactor is best regarded as having charac-
teristics approaching those of a continuously stirred tank
reactor and the gas residence time has been estimated from
the half-life of reaction products in the cell when the reaction
is quenched by rapid cooling of the sample. The cell was also
connected through a Ðner Ñow control valve to a di†erentially
pumped quadrupole mass spectrometer and heating was
applied to maintain the whole of this gas inlet system at ca.
50 ¡C to prevent product condensation. Mass spectrometric
¤ E-mail: r.m.nix=qmw.ac.uk
J. Chem. Soc., Faraday T rans., 1997, 93(21), 3869È3878
3869

View Article Online
data were recorded in a multiple-ion monitoring mode in
which the sample temperature and the intensities of up to 15
mass fragments were monitored quasi-simultaneously. The
products that were routinely monitored in addition to HCN
were H , H O, COÈN , NO and CO ; no partial oxidation
carbon (CH ) conversion to the speciÐed product and the
internal calibration negates the e†ects of any experimental
drift in absolute intensities over time. The main drawback of
the use of mass spectrometric analysis is the inability to dis-
4
criminate between CO and N (given that the characteristic
2
2
2
2
2
products of any further reactions of HCN (such as urea or
formamide) were detected.
12, 14 and 16 fragment ion signals also overlap with those
from other species) and for this reason we do not report any
direct data on these particular products.
The model catalysts employed for this work consisted of
strips of platinum foil (Advent Research Materials: 0.015 mm
thick, [99.95%) and platinumÈrhodium foil (Johnson
Matthey: 90%PtÈ10%Rh, 0.0125 mm thick, [99.9%) cut to a
size of ca. 1 ] 2 cm2 and supported between Ta mounts on
gold-coated stainless-steel holders. The sample was heated
resistively by an ac supply and the temperature governed
using a Newtronic Micro 96 programmable controller. The
temperature was measured using a chromelÈalumel thermo-
couple, spot-welded directly onto the back of the sample. Care
was taken to ensure that the resistive heating did not inÑuence
the thermocouple reading and the stability of the thermocou-
ple reading over time was also veriÐed by using a large-area
photovoltaic detector mounted externally on a borosilicate
window to measure the intensity of the black-body radiation
emitted at set temperatures. This showed a drift of the sample
temperature measurement of no more than 5 ¡C over the six
months of operation.
The samples were initially cleaned by cycles of 500 eV
argon-ion bombardment (ca. 5 lA, 20 min) and vacuum
annealing at 700 ¡C, until only low levels of carbon and
oxygen impurities were apparent. The samples were Ðnally
cleaned by high-temperature oxygen treatment and, indeed,
this was routinely used as the Ðnal treatment prior to a cata-
lytic run. The exact procedure for cleaning in oxygen varied
according to the sample: platinum was heated in a dynamic
Ñow of oxygen ([10~7 Torr) at 700 ¡C for 10 min, cooled to
below 100 ¡C under oxygen and then reheated to 750 ¡C for 2
min in UHV to desorb any adsorbed oxygen. For the
platinumÈrhodium alloy, oxidation of the rhodium by oxygen
at moderate temperatures was a concern, so the catalyst was
heated in oxygen ([10~7 Torr) to 700 ¡C for 5 min and then
held at this temperature whilst the oxygen was evacuated for a
further 5 min in UHV.
Results
Polycrystalline Pt samples
The sample surface was routinely examined by XPS to check
its cleanliness before reaction and examined after reaction at
both room temperature and 250 ¡C. Post-reaction analysis
commonly showed very little surface nitrogen (a very weak
peak at ca. 399.5 ^ 0.5 eV; removed by heating to 250 ¡C),
oxygen, almost exclusively as CO, (giving an O 1s peak at ca.
532.6 eV; heating to 250 ¡C removed this peak and gave a 28
u desorption signal) and carbon as both CO (at 285.3 eV) and
an apparently elemental, less volatile form giving rise to a C
1s peak at ca. 284.3 eV. Re-exposure of the surface to CO at
low pressures after it had been heated to 250 ¡C did not,
however, generate the same levels of surface carbon and
oxygen seen after the sample was Ðrst removed from the
reactor; this suggests that the treatment at 250 ¡C itself
induces changes in the nature of the Pt surface and that
perhaps the reactive surface is unstable in the absence of
stabilising adsorbates.
In general, there was no obvious correlation between the
reaction conditions employed and the amount of elemental
carbon detected in the subsequent post-reaction analysis of
the surface, which would correspond to a near-monolayer
coverage on the freshly cleaned and annealed Pt surface. It
should be noted, however, that all samples were cooled from
the reaction temperature under the reactant gas mixture and
the surface being examined in the post-reaction analysis
cannot therefore be considered to be truly representative of
that existing under steady-state synthesis conditions.
Transient behaviour during reaction initiation. Fig. 1 shows a
typical set of raw data illustrating the initial stages of a reac-
tion run. The stabilisation of the mass readings is principally
dependent on stabilisation of the sample temperature. The
main exception to this is the 18 u peak, which takes slightly
longer to stabilise, indicating some degree of water retention
in the gas inlet system despite the applied heating; however,
the signals of all species had reliably stabilised well within the
standard run time of 5 min.
The standard conditions employed for a reactor experiment,
to be assumed unless stated otherwise, were a total gas pres-
sure of 0.10 Torr, a gas residence time of 5.7 s and a catalyst
surface temperature of 600 ¡C. The standard reaction feed was
a mixture of CH ÈNH ÈO in a 1 : 1 : 0.5 mole ratio. The
4
3
2
heating was adjusted such that the sample would be brought
from room temperature to within 5 ¡C of the desired reaction
temperature in less than 25 s, with minimal overshoot.
Cooling of the sample to below 300 ¡C (where reaction had
ceased) occurred in a matter of seconds; the stabilisation of
the mass spectrometer readings was then dependent on the
gas residence time. In the absence of oxygen, steady-state syn-
thesis activity was not observed under the normal conditions
of temperature and pressure employed in this work; instead
an initial surge of products was followed by a gradual
reduction with time.
There is transient production of both NO and CO during
2
the initiation stages at levels which, although small, are still
noticeably higher than those observed at steady state. The
signals of a number of masses, but most noticeably hydrogen,
exhibit initial oscillations, but these are merely a direct reÑec-
tion of similar oscillations in the sample temperature as it
approaches its set-point value.
Mass spectrometer signals were shown to be linear with
reactor partial pressure over the experimental range using a
E†ect of gas residence time (1 : 1 : 0.5 CH –NH –O ). The
4
3
2
variation of HCN yield with gas residence time over platinum
at 600 ¡C is shown in Fig. 2; an initial near-linear rise in yield
with residence time begins to level o† after ca. 10 s. Prior to
this point, the extent of the synthesis reaction is clearly con-
trolled by the gas residence time and the measured steady-
state yield of HCN is principally determined by the kinetics of
the initial synthesis reactions, whereas at longer residence
times the HCN yield is beginning to be a†ected by either ther-
modynamic equilibrium limitations or by the onset of serial
reactions, such as hydrolysis, in which HCN is consumed. A
change in either the total pressure (a reduction to 0.02 Torr)
or in the temperature (to 700 ¡C) leads to a change in the per-
calibrated mixture of 5.03 ^ 0.05% CO in nitrogen (Air Pro-
2
ducts plc). The sensitivity of the detection system to CO ,
2
HCN and CH (all relative to N ) was determined using the
4
2
aforementioned CO mixture,
a
calibrated mixture of
2
0.98 ^ 0.03% HCN in nitrogen (MG Gas Products Ltd.) and
a 50 : 50 mixture of CH in N (prepared in situ). These data
were then used to determine the relative sensitivities of the
product gases HCN and CO relative to CH and the yields
quoted in this work are derived from comparison of the
4
2
2
4
product signal with the CH signal immediately prior to initi-
4
ation of the reaction; the quoted yields are thus based on
3870
J. Chem. Soc., Faraday T rans., 1997, V ol. 93

View Article Online
e†ect on the HCN yield, as would be expected for any hetero-
geneously catalysed process; we would note at this point that
the incident gas temperature will be substantially lower than
the surface temperature under all our operating conditions.
The variation of HCN yield with catalyst temperature for
platinum at 0.10 and 0.02 Torr is shown in Fig. 3 (together
with similar data for the platinumÈrhodium alloy at 0.10 Torr,
which is discussed below). There is no catalytic production of
HCN at 300 ¡C or below, but the yield then shows a steady
increase from 400 to 700 ¡C. The percentage HCN yield from
platinum is similar at the two pressures, indicating that HCN
formation is approximately Ðrst order with respect to total
pressure for a 1 : 1 : 0.5 mix of NH ÈCH ÈO under our con-
3
4
2
ditions. At the highest temperatures the activity apparently
starts to decrease at the higher pressure, but this is likely to be
due to the onset of further reactions in which HCN is con-
sumed. We would further note that, of the various other pro-
ducts, hydrogen shows a particularly strong dependence of
yield on substrate temperature.
Variation of the CH –NH –O feed ratios. In the industrial
4
3
2
process, a key role of oxygen is to provide, through com-
bustion, sufficient heat to maintain the otherwise endothermic
conversion of CH and NH into HCN and H . In the
4
3
2
Andrussow process, oxygen is mixed directly with these gases,
in the Degussa process the reaction gases are heated exter-
nally by the combustion of methane. In the experimental con-
ditions used in this study the catalyst was heated electrically
so, as mentioned previously, the oxygen concentration could
be varied without a†ecting the catalyst temperature. The
partial pressures of NH and CH were, however, kept equal
Fig. 1 Initial stages of reaction; raw mass spectral data for a stan-
dard run over Pt with a 1 : 1 : 0.5 CH ÈNH ÈO feed at 0.10 Torr
4
3
2
total pressure (the sample heating to 600 ¡C is initiated at t \ 60 s)
3
4
to each other, and the overall reaction pressure was main-
tained constant at 0.10 Torr. The yields of HCN and CO
(expressed as % conversion of methane), and the total % con-
version of methane are listed in Tables 1È3.
centage yield, but not in the basic form of the graph, nor to a
signiÐcant change in the residence time at which the HCN
yield begins to plateau. In view of these results, all further
studies of the synthesis kinetics used a constant (standard)
residence time of 5.7 s.
2
Increasing the relative concentration of oxygen by changing
the feed composition from the standard 1 : 1 : 0.5 mixture to
1 : 1 : 1 (CH ÈNH ÈO ), whilst maintaining the same total
4
3
2
E†ect of catalyst temperature (1 : 1 : 0.5 CH –NH –O ). A
4
3
2
change in the sample surface temperature has a signiÐcant
Fig. 3 Variation in the HCN yield from a 1 : 1 : 0.5 CH ÈNH ÈO
4
3
2
feed mixture with the catalyst temperature: (a) Pt sample (1 : 1 : 0.5
CH ÈNH ÈO , 0.02 Torr); (b) Pt sample (1 : 1 : 0.5 CH ÈNH ÈO ,
4 3 2 4 3 2
0.10 Torr); (c) PtÈ10%Rh sample (1 : 1 : 0.25 CH ÈNH ÈO , 0.10
Fig. 2 Variation of the HCN yield with gas residence time over the
Pt sample (1 : 1 : 0.5 CH ÈNH ÈO , 600 ¡C, 0.10 Torr)
4
3
2
Torr); (d) PtÈ10%Rh sample (1 : 1 : 0.5 CH ÈNH ÈO , 0.10 Torr)
4
3
2
4
3
2
J. Chem. Soc., Faraday T rans., 1997, V ol. 93
3871

View Article Online
Table 1 Steady-state yield (%) of HCN over platinum at di†erent
temperatures and gas compositions
impurities or to stabilise the catalytic surfaces at this lower
temperature and partial pressure.
In the complete absence of oxygen (anoxic conditions), a
1 : 1 ratio of CH and NH does not produce a signiÐcant
gas feed (CH ÈNH ÈO )
4
3
2
4
3
steady-state yield of HCN under our conditions at any tem-
perature up to 750 ¡C. Instead, there is an initial transient pro-
duction of HCN but the synthesis rate then gradually decays
to a near-zero value. The main products of this reaction
(albeit again only transiently) are H and N which are ini-
temperature/¡C
1 : 1 : 0
1 : 1 : 0.1
1 : 1 : 0.5
1 : 1 : 1
600
700
750
B
B
B
B
10.2(44)
10.5(33)
9.1(26)
0.0(0)
4.0(18)
7.2(22)
7.2(65)
8.2(37)
2
2
tially produced in very high yield, a feature which is reÑected
All at a total pressure of 0.10 Torr. HCN selectivity, given in brackets,
is based upon methane consumption. B: Decays with time.
in the NH consumption which temporarily approaches 50%
at 750 ¡C. In general, the decay in HCN yield is more pro-
3
nounced at higher temperatures and this is illustrated in the
typical proÐles of the HCN yield against time shown in Fig. 4.
However, it is important to note that the exact form of the
proÐle must be treated with caution because, at a set tem-
perature, it is dependent to some extent on the rate of heating
to that temperature: the faster the heating, the more rapid the
initial decay in yield. Nevertheless, it appears that the rate of
decay of HCN production is in general rather faster than that
of H and N . No transient production of HCN, H or N
pressure of 0.10 Torr, reduces the HCN yield and selectivity,
whilst increasing the production of CO , H O and, by impli-
cation, CO. The reduced HCN production is most marked at
the lowest temperature studied, 600 ¡C, where no HCN yield
is detected using the 1 : 1 : 1 feed. At higher temperatures the
production of HCN recovers, reaching almost 80% of the
2
2
amount produced by the standard mixture at 750 ¡C. CO
production at 600 ¡C is about eight times higher using the
2
2
2
2
2
more oxygen-rich mixture. This increase in the oxidation pro-
was evident when the same samples were re-tested without
cleaning, indicating that the decay in synthesis activity seen in
the Ðrst run cannot be simply reversed by an evacuation treat-
ment or by cooling and re-heating to reaction temperature.
Increasing the CH to NH ratio from 1 : 1 to 2 : 1 simply
ducts of CH as the oxidant concentration increases is
4
unsurprising, but both mixtures also show a reduction in CO
2
with increasing temperature. This decrease in CO yield is
2
likely to reÑect increasing selectivity to CO (in part, at least,
4
3
since the thermodynamics increasingly favour CO at these
higher temperatures) but we are unable to conÐrm this
directly owing to the aforementioned signal overlap in the
mass spectra. The trends observed in switching from the stan-
dard mixture to the 1 : 1 : 1 mixture were further enhanced
when 1 : 1 : 1.5 mixtures were employed.
exacerbated the e†ects reported above; a slightly smaller tran-
sient yield of HCN was again followed by a rapid decay in
activity to a near-zero limiting value. By contrast, increasing
the NH concentration by changing the CH to NH ratio
3
4
3
from 1 : 1 to 1 : 2 did not substantially a†ect the behaviour,
with the exception that an anomalously high 29 u signal was
evident in some experiments. In the experiments of concern,
the analysis system was modiÐed so that the mass spectro-
meter could sample gas from the reactor more directly. Com-
parison of the 29/28 u signal intensity ratio across a range of
such experiments indicates a deÐnite trend with the value
increasing from 2È5% in standard CH ÈNH ÈO (1 : 1 : 0.5)
The link between oxygen concentration and production of
CO also persists when the oxygen concentration is reduced,
2
in that a 1 : 1 : 0.1 ratio CH ÈNH ÈO mixture produces no
4
3
2
discernible CO . Relative to the standard 1 : 1 : 0.5 feed
mixture, the HCN yield from the 1 : 1 : 0.1 mixture is slightly
2
lower at high temperatures, although the selectivity to HCN is
noticeably higher. However, at 600 ¡C the HCN yield from
this more oxygen-lean mixture does not remain constant, but
shows a gradual decline after the Ðrst minute. This latter
result implies that the catalyst surface at 600 ¡C is either being
poisoned by a build-up of reaction by-products, which are not
now being removed sufficiently rapidly by oxidation, or that a
surface restructuring is occurring in the more reducing atmo-
sphere to give a non-catalytic surface. Thus the chemical
potential of oxygen must be insufficient either to oxidise the
4
3
2
runs, to around 10% in some 1 : 1 CH ÈNH runs and as high
as 24% during the high transient activity initially observed in
4
3
1 : 2 CH ÈNH runs. Analysis of the complete fragmentation
4
3
pattern suggests that the species responsible for this 29 u
Table 2 Steady-state yield (%) of CO over platinum at di†erent
2
temperatures and gas compositions
gas feed (CH ÈNH ÈO )
4
3
2
temperature/¡C
1 : 1 : 0
1 : 1 : 0.1
1 : 1 : 0.5
1 : 1 : 1
600
700
750
0.0
0.0
0.0
0.0
0.0
0.0
0.5
0.5
0.2
4.1
1.3
0.9
All at a total pressure of 0.10 Torr.
Table 3 Methane conversion (%) over platinum at di†erent tem-
peratures and gas compositions
gas feed (CH ÈNH ÈO )
4
3
2
temperature/¡C
1 : 1 : 0
1 : 1 : 0.1
1 : 1 : 0.5
1 : 1 : 1
600
700
750
B
B
B
B
23.2
32.0
34.7
11.1
22.2
33.3
11.1
22.2
Fig. 4 Transient HCN production from a 1 : 1 CH ÈNH feed
4
3
mixture over platinum at 0.10 Torr. (È È È È) 750, (ÈÈÈ) 700 and
(É É É É É É) 600 ¡C.
All at a pressure of 0.10 Torr. B: Decays with time.
3872 J. Chem. Soc., Faraday T rans., 1997, V ol. 93

View Article Online
signal might be H CxNH, but further work is required to
with time) are consumed at each reaction temperature. There
is no detectable water production (18 u) and whilst there is a
substantial 28 u signal, this can largely be attributed to nitro-
2
conÐrm that it is indeed this species which is responsible and
that it is a real product of the catalytic reaction over Pt rather
than an artifact of the modiÐed analysis system.
gen produced by NH decomposition rather than CO, and
3
there is a concomitant increase in the H signal.
2
The synthesis of HCN from CH ÈNO feeds (1 : 1 and 1 : 2)
4
Use of alternative oxidants and CH –NO feed mixtures.
4
was also brieÑy studied. At temperatures of 600È650 ¡C an
From the preceding section, it is clear that an oxidant of some
initial transient production of a low level of HCN was fol-
lowed by a slow decay to a very low level of activity. At higher
temperatures (700È750 ¡C) steady-state HCN production was
observed but the activity was typically a factor of 5È10 lower
than that seen from CH ÈNH ÈO (1 : 1 : 0.5) mixtures under
form is required to maintain HCN synthesis activity under
our accessible operating conditions of temperature and pres-
sure; in this section we therefore report brieÑy on the use of
the alternative reactant mixtures; (i) CH ÈNH ÈCO , and (ii)
4
3
2
CH ÈNO.
4
3
2
4
comparable conditions. In all cases the main products of the
At high temperatures we would anticipate CO could act as
2
reaction were COÈN and H O with only a very low pro-
a mild oxidant and could thus be capable of removing surface
2
2
duction of H evident. The reactions of CH ÈNO feed mix-
carbon by means of the reverse Boudouard reaction.
2
4
tures were also characterised by a particularly high transient
C ] CO ] 2CO
production of CO during the Ðrst 10È20 s of the reaction and
2
2
a signiÐcant steady-state yield of CO throughout the remain-
It is also possible that the direct addition of CO (and
2
2
ing course of reaction.
through its action as an oxidising agent, also CO) might posi-
tively enhance the equilibrium HCN yield by its inÑuence on
the equilibrium positions of the various reactions in which
E†ect of surface pretreatments (1 : 1 : 0.5 CH –NH –O ).
4
3
2
The e†ect of the pre-existing surface condition of the sample
on the initial activity and subsequent evolution of the catalytic
behaviour was investigated in a series of experiments in which
the sample was Ðrst cleaned in the standard fashion and then
subjected to one of various pretreatments prior to exposure to
reaction conditions. The main pretreatments employed were:
(i) a brief argon-ion bombardment of 5 min (500 eV, 5 lA); (ii)
an extended argon-ion bombardment of 20 min, followed by a
250 ¡C anneal; (iii) exposure to two aliquots of 20 l of ethylene
with the sample held at 600 ¡C; (iv) exposure to 60 l of oxygen
at 323 ¡C; (v) an ammonia pretreatment in the reaction
chamber which involved heating the catalyst to 700 ¡C for 5
min in pure ammonia (0.05 Torr). The subsequent reactivity
tests all used the standard gas mixture (1 : 1 : 0.5
CH ÈNH ÈO ) at the standard temperature (600 ¡C) and pres-
CO are produced (including the direct partial oxidation of
x
methane and HCN hydrolysis); indeed, it has been previously
claimed that the yield of HCN may be increased by addition
of CO to the standard reaction mixture.12
2
Our results, obtained using a 1 : 1 : 1 mixture of CH , NH
4
3
and CO , are shown in Fig. 5 for a total pressure of 0.15
Torr; the partial pressures of NH and CH are thus the same
2
3
4
as for the measurements shown in Fig. 4. It is immediately
clear that this addition of carbon dioxide does not permit
steady state synthesis to be maintained, even at 750 ¡C (i.e. at
conditions under which a high HCN yield is obtained with a
5% oxygen component in the feed mixture); there is, however,
a slight improvement relative to a 1 : 1 CH ÈNH feed, in that
the rate of activity loss is noticeably reduced. Rather sur-
prisingly, the trend in behaviour with temperature is also dif-
ferent from that initially anticipated in that the addition of
CO to the 1 : 1 CH ÈNH mixture has only a very small
apparent e†ect at 750 ¡C, but does result in a much more pro-
nounced reduction in the rate of decay of synthesis activity at
lower temperatures. The mass spectra show that small
amounts of carbon dioxide (initially ca. 6È8%, but reducing
4
3
4
3
2
sure (0.10 Torr).
The e†ects of the Ðve di†erent pretreatments on the cata-
lytic behaviour are compared directly with the results of a
standard run in Fig. 6. It is immediately apparent from this
Ðgure that, whilst the pretreatments did have an inÑuence on
2
4
3
Fig. 6 Activity of
a Pt foil (standard conditions; 1 : 1 : 0.5
CH ÈNH ÈO , 600 ¡C, 0.10 Torr) subjected to the following pretreat-
4
3
2
Fig. 5 Transient HCN production from a 1 : 1 : 1 CH ÈNH ÈCO
ments: (a) brief ion bombardment, (b) extended ion bombardment and
250 ¡C vacuum anneal, (c) ammonia, (d) oxygen, (e) standard, (f) ethyl-
ene (see text for further details)
4
3
2
feed mixture over platinum at 0.15 Torr. (È È È È) 600, (É É É É É É) 700 and
(ÈÈÈ) 750 ¡C.
J. Chem. Soc., Faraday T rans., 1997, V ol. 93
3873

View Article Online
the initial characteristics, all gave yields that appear even-
tually to converge with those obtained after the standard
cleaning procedure.
normally employed for cleaning the platinum catalyst, but
whereas the standard oxygen cleaning treatment is followed
by heating in vacuo to remove adsorbed oxygen, this pro-
cedure was a way of restoring the adsorbed oxygen and gave
an O 1s XPS peak at 529.8 eV. Perhaps unsurprisingly, this
pretreatment gave results showing no di†erence within experi-
mental error from those of a standard run.
The initial HCN yield was accentuated after both argon ion
bombardment routines, albeit to di†erent extents, suggesting
that the surface roughening which this treatment will have
caused had a signiÐcant e†ect on the catalytic performance.
The highest HCN yield was produced after a brief (5 min)
bombardment period; the long (20 min) bombardment was
followed by heating to 250 ¡C for 5 min, and this vacuum
annealing apparently reduced the roughening e†ect by more
than was gained by prolonging the bombardment. The
increase in HCN yield in these experiments is not, however, a
Polycrystalline Pt–10%Rh samples
The composition of the freshly bombarded platinumÈrhodium
alloy surface was investigated by XPS using two methods: the
Ðrst was a comparison of the absolute platinum 4f
7@2
intensity from the alloy with that from the pure metal whilst
peak
result of substantially enhanced CH conversion but reÑects a
4
the second involved a comparison of the platinum 4f
and
signiÐcant increase in selectivity to HCN. Both the HCN yield
7@2
rhodium 3d
peaks, solely from alloy spectra. The Ðrst
and selectivity from the pretreated surfaces decreased over
time and tended towards that from the standard run, a behav-
iour which is entirely consistent with a gradual loss of surface
roughness.
5@2
method did not give particularly good reproducibility and was
susceptible to even slight amounts of surface contamination
and slight variation in the detector sensitivity over time, but
after correction for these factors this method gave results indi-
cating a platinum concentration in the alloy to be 91% or
more of that in pure platinum. The second method, involving
a direct comparison of two di†erent signals from the same
alloy, is not a†ected by such factors, but does require informa-
tion concerning the relative sensitivities of the two signals.
Published empirical sensitivity factors would suggest an absol-
ute concentration of 86% or less platinum.14 In view of the
errors inherent in both methods, it seems reasonable to accept
that the surface composition of the alloy is indeed close to the
bulk composition of 90.0% platinumÈ10.0% rhodium certiÐed
by the manufacturer.
Previous results from the anoxic runs (see above) have sug-
gested that carbon accumulation was responsible for loss of
catalytic activity. Pretreatment of the sample by twice expos-
ing the hot (600 ¡C) surface to 20 l of ethylene gave a heavy
carbon deposit characterised by an intense XPS C 1s peak at
284.5 eV; the XPS results therefore suggest a carbon deposit
similar in nature to those of post-reaction analyses, but the
peak is over Ðve times more intense. Despite all this initial
surface carbon, the e†ect on the HCN yield from the oxygen-
containing feed mixture was relatively mild. There was a slight
retardation of the onset of HCN synthesis but the yield never-
theless rapidly attained a value only slightly (ca. 20%) lower
than that seen in the standard run, and this di†erence then
lessened with time. Correspondingly, there was a very small
Irrespective of the absolute surface stoichiometry, we would
stress that the platinum to rhodium signal ratio observed for
the alloy immediately after ion bombardment was found to be
unchanged following annealing (to 750 ¡C) or cleaning in
oxygen (10~7 Torr, 700 ¡C). The only occasions when this
internal signal ratio varied was under reducing conditions,
and this e†ect could be reversed by reoxidation (this feature is
more fully discussed below). To quantify these changes in
surface stoichiometry under reaction conditions, the concen-
tration of each metal was determined from the integrated peak
areas of the Ðtted XPS data after subtraction of a Shirley
background.
rise in the yield of CO compared to the standard run (in
2
absolute terms, up 0.05%) but despite this the HCN selectivity
was very similar to that in the standard run, which indicates
CO production is also reduced. Overall, therefore, ethylene
pretreatment slightly inhibits both HCN synthesis and
methane oxidation. Post-reaction XPS analysis revealed that
over 80% of the initially deposited carbon had been burnt o†
during the single 5 min run and the Ðnal level of surface
carbon was comparable to that seen after any run employing
these reaction conditions.
As might be expected on the basis of the above results, pre-
exposure of the Pt surface to ethylene under milder condi-
tions, designed to give a near-saturation overlayer of surface
ethylidyne, had very little e†ect on the HCN activity, with the
exception that a slight retardation of the onset of HCN syn-
thesis was again evident.
The catalytic activity of the PtÈ10%Rh alloy towards HCN
and CO production, and methane conversion is detailed in
2
Tables 4È6, whilst the variation of activity with temperature
under standard conditions is compared directly with that for
pure platinum in Fig. 3. We would wish to emphasize,
however, that the activity of the PtÈ10%Rh surface (unlike the
pure Pt) is very susceptible to variations in the sample history,
i.e. the sample exhibits a relatively long-lasting memory e†ect
which can inÑuence the catalytic activity observed in any
given experimental run and can lead, for example, to hyster-
esis type e†ects in sequential measurements. The 1 : 1 : 0.5
data of Tables 4È6 and curve (c) of Fig. 3 were, for example,
recorded sequentially, in strict order of decreasing tem-
perature and the activities thus recorded at temperatures of
700 ¡C and below are signiÐcantly higher than would have
The platinum was pretreated with ammonia under condi-
tions (0.5 Torr, 700 ¡C) where the platinum can e†ectively
catalyse the decomposition reaction. However, during this
treatment the catalytic activity towards NH decomposition
3
dropped by ca. 80%, suggesting either that surface rearrange-
ment or site blocking was occurring (this reaction is known to
be very structure sensitive on Pt13). Subsequent XPS analysis
revealed at best only a slight accumulation of nitrogen on the
surface, with a very weak peak at ca. 400 eV, which would
apparently rule out stabilisation of a catalytically inactive
surface state by adsorbed nitrogen species, although the possi-
bility that the same e†ect might be induced by adsorbed
hydrogen cannot be excluded. Some adventitious CO adsorp-
tion was evident and it is possible that the CO masked other,
less abundant surface carbon species. The e†ect of this
pretreatment was to give an initial period of inhibition of
Table 4 Steady-state yield (%) of HCN over a 90% platinumÈ10%
rhodium alloy at di†erent temperatures and gas compositions
gas feed (CH ÈNH ÈO )
4
3
2
temperature/¡C
1 : 1 : 0
1 : 1 : 0.25
1 : 1 : 0.5
1 : 1 : 1
0.0(0)
600
700
750
B
B
B
2.6(31)
4.2(31)
4.8(36)
1.3(11)
3.2(26)
3.9(21)
HCN production but an increase in initial CO and NO for-
2
mation, followed by a rapid rise in HCN yield and a drop in
the oxidation products to the levels found in the standard run.
The mild oxidation pretreatment employed (exposure to
All at a total pressure of 0.10 Torr. HCN selectivity, given in brackets,
is based upon methane consumption. B: Decays with time.
10~7 Torr O for 10 min at 323 ¡C) was less severe than that
2
3874
J. Chem. Soc., Faraday T rans., 1997, V ol. 93

View Article Online
Table 5 Steady-state yield (%) of CO over a 90% platinum 10 %
ute change in concentration in the topmost surface layer. The
2
rhodium alloy at di†erent temperatures and gas compositions
manner in which the ratio of rhodium 3d
peak area to
5@2
platinum 4f peak area changes with reaction time at various
gas feed (CH ÈNH ÈO )
7@2
4
3
2
reaction temperatures is shown in Fig. 7, which clearly shows
how the surface composition tends to a new limiting value
which is essentially independent of the reaction temperature.
The decay of catalytic activity with a 1 : 1 mixture, as mea-
sured by the decrease of product peaks in the mass spectra
(both HCN at 27 and 26 u and N and H , from NH decom-
temperature/¡C
1 : 1 : 0
1 : 1 : 0.25
1 : 1 : 0.5
1 : 1 : 1
21
600
700
750
0.0
0.0
0.0
0.6
0.2
0.2
2.2
1.0
0.5
2
2
3
position, at 28 and 2 u), is virtually exponential over many
minutes and, furthermore, the rate of this decay is apparently
not dependent on the sample history. This indicates that the
process causing the decay in activity is pseudo-Ðrst order in
this region, and apparent rate constants for the decay in activ-
ity can therefore be derived without knowledge of the absolute
concentrations. If the rate of rhodium enrichment in the
surface region is also assumed to be Ðrst order then kinetic
data can also be derived for this segregation process for com-
parative purposes. Thus the rates of change of surface com-
position at the various reaction temperatures can be
determined and these values can be used to determine an acti-
vation energy for the surface enrichment. Arrhenius plots for
the decay in the rates of HCN synthesis and ammonia decom-
position, and for rhodium enrichment are shown in Fig. 8,
from which the corresponding activation energies are
76(^15), 79(^17) and 123(^19) kJ mol~1 respectively. The
similarity in values for the two catalytic reactions suggests
that the loss of catalytic activity is due to the same cause in
both cases. This is to be expected for such similar reactions.
The rhodium enrichment occurs with a completely di†erent
activation energy, showing that the loss of catalytic activity
cannot simply be related to the change in surface composition
of the catalyst.
All at a total pressure of 0.10 Torr.
been recorded in runs in which the reaction had previously
been operated at lower temperatures; this characteristic of the
reaction on PtÈ10%Rh is the origin of the apparently anom-
alous 600 ¡C data point in curve (d) where the data were
recorded after operation at a lower temperature. In summary,
we would note that treatment of the PtÈ10%Rh sample in the
reaction mixture at a high temperature enhances the activity
of the sample in subsequent measurements at lower tem-
peratures and that this e†ect can be quite long-lasting and
lead to some apparent irreproducibility in the activity mea-
surements.
Nevertheless, under all the conditions where a direct com-
parison has been made, the PtÈ10%Rh alloy sample gave a
signiÐcantly lower yield of HCN than did the pure platinum
sample. The yield (and selectivity) of HCN was higher when a
more oxygen lean mixture was employed, but still did not
approach that obtained over pure Pt. Furthermore, under the
more oxidising environment of a 1 : 1 : 1 CH ÈNH ÈO feed
4
3
2
at 700 ¡C the HCN production was below our detection levels,
whilst oxidation products of NH and CH were in abun-
3
4
dance, with large rises in the 44 (CO ), 30 (NO), 28 (CO) and
The e†ect of surface pretreatments on HCN yield for the
PtÈ10%Rh alloy have been studied in a similar fashion as for
pure Pt. Most notably, the argon-ion bombardment pretreat-
ments have similar e†ects in platinum and the alloy, although
for the PtÈ10%Rh alloy the longer ion bombardment pro-
cedure (with subsequent 250 ¡C anneal) gave the highest initial
activity. Our results for the chemical pretreatments are,
2
18 (H O) u signals. Having shown that the introduction of
2
rhodium promotes only oxidative catalysis under these more
oxygen-rich conditions, no further experiments were con-
ducted at a feed ratio of 1 : 1 : 1.
After reaction, the PtÈ10%Rh surface exhibited a slightly
higher surface concentration of N-containing species than was
found to be the case with pure Pt, with weak peaks evident at
399.4 (^0.2) and 397.3 (^0.2) eV. TPD measurements on PtÈ
10%Rh samples freshly transferred from the reactor, clearly
exhibited near-coincident HCN and H desorption in addition
2
to the major 28 u (CO) peak.
When the PtÈ10%Rh sample is operated under anoxic con-
ditions, with a 1 : 1 feed mixture of NH and CH , the cata-
3
4
lyst loses activity in the same manner as is seen for pure
platinum, but the very low ultimate yields of the two catalysts
are comparable (in contrast to the reduced yield from the
alloy in the presence of oxygen). Once again, the rate of loss of
catalytic activity is greater at higher temperatures. In this case,
however, another e†ect also manifests: the ratio of platinum
to rhodium in the surface region decreases with time of oper-
ation of the catalyst in the anoxic environment. These changes
have been followed by XPS, which enables the degree of
change within the sampled surface region to be clearly estab-
lished, although it does not permit determination of an absol-
Table 6 Methane conversion (%) over
a 90% platinumÈ10%
rhodium alloy at di†erent temperatures and gas compositions
gas feed (CH ÈNH ÈO )
4
3
2
temperature/¡C
1 : 1 : 0
1 : 1 : 0.25
1 : 1 : 0.5
1 : 1 : 1
22.9
600
700
750
B
B
B
8.4
13.5
13.5
12.1
12.5
18.9
Fig. 7 Change in the Rh 3d /Pt 4f XPS signal ratio with time of
5@2
7@2
exposure to a 1 : 1 CH ÈNH feed mixture at 0.10 Torr. (È È È È) 750,
4
3
All at a total pressure of 0.10 Torr.
(É É É É É É) 700, (È É È) 650 and (ÈÈÈ) 600 ¡C.
J. Chem. Soc., Faraday T rans., 1997, V ol. 93
3875

View Article Online
oxygen pretreatments have little e†ect (even transiently) on
the HCN synthesis reaction kinetics over PtÈ10%Rh.
Discussion
To facilitate the discussion, we will begin by outlining our
model for this reaction and then consider how the various
results described above are consistent with this model.
A generalised reaction scheme for the three-component feed
mixture is shown below.
Fig. 8 Arrhenius plots for the surface enrichment of rhodium and
associated reactions on a PtÈ10%Rh sample: (a) decay of HCN syn-
thesis activity; (b) decay of ammonia decomposition activity; (c)
surface enrichment of rhodium
Of the individual feed gases only NH exhibits a facile
decomposition reaction, whereas any two reagents react
readily as a binary mixture to give, principally, the products
indicated. These initial products may, in turn, act as reagents
for further conversion of components of the feed mixture (e.g.
3
however, complicated by the e†ects of sample history dis-
cussed above, and for this reason we will mention only the
most pertinent observations.
the reaction of NO and CH ), but there is little evidence in
4
our data for such reactions playing a signiÐcant role.
Little di†erence was found between a standard run and one
after ethylene pretreatment. The rate of recovery of catalytic
activity thus appears to be better than for platinum, even
though XPS reveals the initial level of surface carbon and the
reduction to a residual carbon level during the reaction to be
similar.
The NH oxidation reaction is particularly facile, with reac-
tion being initiated under our experimental conditions at tem-
peratures below 200 ¡C.17h19 The product distribution for this
reaction is controlled by the supply of surface oxygen and the
3
surface temperature, with NO formation favoured over N
2
formation at higher temperatures and higher O /NH ratios
2
3
NH pretreatment gave a change in the rhodium/platinum
in the feed mixture.17h20 When methane is added to the
3
ratio at the alloy surface, as found for the anoxic CH ÈNH
NH ÈO feed there is a stoichiometric deÐcit of oxygen and,
once methane oxidation is initiated at higher temperatures,
there is competition between NH and CH for the available
oxygen; a knock-on e†ect of this will be an increase in the
surface lifetime of NH fragments. When CH fragments gen-
erated by CH activation are present on the surface at low
4
3
3
2
mixtures. After such pretreatment, the initial HCN production
from the standard mixture is very low and only increases
slowly towards that observed after the normal sample
pretreatment. Although we have previously noted that there is
not a simple correspondence between Rh surface segregation
and activity decay, the slow increase in activity seen in this
instance does seem to be related to the gradual increase in
Pt : Rh ratio as it reverts back from that produced by the
3
4
x
x
4
concentrations, then they are readily scavenged by oxygen to
yield the undesired combustion products CO and CO .
2
However, as the extent of methane activation increases, and
NH pretreatment. However, the catalyst surface had still not
completely restored its concentration to that before pretreat-
consequently the surface CH population also increases, there
are several e†ects. (i) The steady-state surface oxygen coverage
3
x
ment, even after 10 min in the standard gas mixture at 600 ¡C.
will decrease; since adsorbed CH species bind strongly to the
Pt surface, thus blocking oxygen uptake, and the additional
hydrogen present further enhances the reactive removal of
oxygen. (ii) There is a corresponding reduction in the yields of
x
The initial production of NO (and to a lesser extent CO ) was
2
again more pronounced than in a standard run, as it was for
platinum after ammonia pretreatment. Pretreatment with a
1 : 1 mixture of NH and CH , which gave both surface
both NO and CO , whilst most of the oxygen is converted to
3
4
x
carbon deposits and a change in Pt : Rh ratio, gives results
H O. (iii) The probability of reaction between NH and CH
2
x
x
superimposable with those of NH pretreatment, thus con-
to give HCN (and H ) increases rapidly.
3
2
Ðrming restoration of the initial ratio and not removal of
It follows from the above that efficient activation of CH is
4
carbon deposits to be the signiÐcant factor in the improving
catalytic activity.
vital to HCN production. Indeed, we believe that the extent
and nature of CH activation is the single most crucial factor
4
Oxygen pretreatment of the alloy was carried out under
various conditions, including treatment at 327 ¡C in 10~7 Torr
of oxygen (under which conditions sub-surface incorporation
and/or surface oxides formation are known to occur on
rhodium15,16) which gave a substantially more intense O 1s
XPS peak at 529.8 eV on the alloy surface than was pre-
viously found for pure Pt. No surface oxide is apparent in the
XPS data at the end of the runs, however; the only oxygen
peak being that due to CO at 532.5 eV in the room-
temperature spectrum, whilst both the platinumÈrhodium
ratio and the accumulation of carbon are the same as found in
a standard run. The activities of such pre-treated samples have
been observed to be both slightly lower and slightly higher
than found for a standard preparation but we attribute this
inconsistency to the sample history and conclude that mild
in obtaining high HCN yields, since this produces the high
steady-state concentration of surface CH species necessary to
x
give HCN in high yield and selectivity. We assert that this is
greatly facilitated by a speciÐc state of the Pt surface, which
can normally only be achieved by the presence of an oxygen
component in the feed mixture when operating the catalyst at
the nominal temperatures that we have employed. We further
believe that this state corresponds to a particular type of
surface structure involving Pt atoms of low coordination (or,
conceivably, a certain degree of Pt atom mobility). Such a
state would be expected to exhibit a very high surface free
energy, but it could nevertheless be attained and sustained
either by (i) the highly exothermic reactions of oxygen with
surface hydrogen, nitrogen and carbon, or (ii) very high tem-
peratures resulting from external heat input (as, for example,
3876
J. Chem. Soc., Faraday T rans., 1997, V ol. 93

View Article Online
in the Degussa process). The evidence from our work which
supports the presence of surface restructuring and its role in
generating atomic sites with low coordination during HCN
synthesis is presented below, in the discussion of the separate
sets of experiments, but we would note that the role of surface
defects and coordinatively highly unsaturated metal atoms in
the activation of alkanes on precious metal catalysts has also
been emphasized by several other workers.21,22
In our standard runs, the catalyst surface just prior to initi-
ation of reaction is either that produced by the standard
cleaning procedure (of which an anneal in oxygen is an inte-
gral part) or that produced in a prior reaction run. The struc-
ture of both these surfaces is compatible with efficient
methane activation or, at the very least, rapidly becomes so.
As seen in Fig. 1 there is, therefore, a very brief period in
with which the catalyst regains most of its activity suggests
that the carbon can be quickly (reactively) removed from par-
ticular sites to generate the necessary reactive surface centres.
The remaining more stable carbon deposits have less e†ect on
the catalytic reaction and their slow removal generates little in
the way of further active surface.
The observed lag in the initiation of HCN synthesis after
NH pretreatment is, at Ðrst sight, inconsistent with the
3
absence of any signiÐcant change in the surface composition.
The e†ect is not, however, directly associated with the
ammonia but with the e†ect on the surface structure of the
extended annealing in a non-oxidising environment (since
similar characteristics have since been seen after heating in an
inert environment). We believe that this extended annealing
destroys the nucleation sites at which CH activation is initi-
4
which oxidation chemistry (especially NH oxidation) pre-
ated; i.e. the treatment has an opposite e†ect to that of ion
3
dominates before the surface concentration of C fragments
bombardment, and a signiÐcant period of reaction involving
builds up to the level at which HCN synthesis occurs with
concurrent generation of mainly the ammonia decomposition
products (H , N ) and water. At termination, there is again a
brief period in which oxidation chemistry predominates as the
sample rapidly cools. The CO generated in this process stabil-
ises the “dispersedÏ Pt which characterises the active surface;
removal of this CO by heating in vacuum causes a collapse of
this active surface structure such that the desorption process is
essentially irreversible, as previously noted.
In the absence of oxygen, the initial HCN activity is quite
high (since there are no competing oxidation reactions) but
cannot be maintained at these temperatures (see, e.g. Fig. 4).
The active surface sites are then either e†ectively removed by
annealing or become decorated with inactive carbon deposits
which are stable at these temperatures in the absence of
oxygen. The remaining Pt terraces are essentially inactive for
HCN synthesis or susceptible to deactivation by growth of
unreactive (graphitic?) carbon overlayers. At the temperatures
exothermic oxidation processes (in which CO and, especially,
NO is produced) is required before the surface can re-
structure sufficiently to cause signiÐcant methane activation
2
2
2
and thus build up the requisite concentration of surface CH
species for signiÐcant HCN synthesis to occur.
x
It is clear from the data presented in Fig. 3 and a compari-
son of the data of Tables 4È6 with those of Tables 1È3 that
the addition of rhodium to the platinum is detrimental to
HCN synthesis under our reaction conditions. More speciÐ-
cally, at a rhodium content of 10%, the HCN yield is substan-
tially reduced, and the alloy surface is less than a third as
active as platinum at an equivalent temperature. Owing to the
limits of the apparatus, we were unable to test whether this
reduction persists at temperatures higher than those investi-
gated here; we note, however, that previous work on PtÈRh
alloys indicates that there is a signiÐcant increase in Pt surface
segregation when such alloys are annealed in vacuum at tem-
peratures above 1000 K24 and, if the same segregation were to
be observed under HCN synthesis conditions, then the di†er-
ence in activity between pure Pt and the PtÈ10%Rh alloy
under commercial synthesis conditions might be less marked
than we observe.
employed, the addition of CO to the feed and the consequent
2
CO decomposition does not yield a sufficiently high surface
2
oxygen activity to prevent this deactivation occurring (Fig. 5).
The fact that a reasonable HCN activity is transiently
observed in the complete absence of oxygen also indicates that
The further reduction in HCN activity seen after enrich-
whilst surface oxygen species (O, OH) may facilitate CH (and
ment of the surface with Rh (by pretreatment in pure NH or
reaction under anoxic conditions) indicates that the inhibiting
4
indeed NH ) activation by assisting the dissociative adsorp-
3
3
tion, the presence of such species is certainly not essential for
methane conversion at these temperatures.
e†ect of Rh is more pronounced still at higher surface concen-
trations. The driving force for this rhodium segregation on the
alloy surface is thought to be the stabilisation of surface
rhodium atoms by adsorbed hydrogen.25 In this work, such a
condition will clearly exist during both the anoxic runs and
the catalytic decomposition of pure ammonia since ammonia
readily decomposes to generate hydrogen which is less readily
removed when oxygen is absent.
The rapid regeneration of HCN activity on PtÈ10%Rh after
pretreatment in ethylene suggests that the blocked sites on the
alloy catalyst are more easily cleaned with oxygen than they
are in pure platinum, which may have relevance to the
resilience of such catalysts to the presence of higher hydrocar-
bons in the feedstock, but this needs much more investigation.
The key question which remains to be addressed is the
cause of the reduced HCN activity in the presence of Rh. This
could be attributed to the greater resistance of the alloy to the
surface restructuring which we have proposed produces the
The results obtained following pretreatments of the Pt
sample (Fig. 6) provide clear support for the proposed model
of low coordination Pt atoms being most active in HCN syn-
thesis. Ion bombardment leads to an enhanced initial HCN
yield. Such an increase in yield might be attributable to a
simple increase in the geometric area of platinum surface, but
this is not consistent with the observation that the increase in
yield is principally associated with an increase in selectivity to
HCN. Instead, we attribute these e†ects to the artiÐcal rough-
ening of the surface induced by the high energy ion bombard-
ment, i.e. to an increase in the concentration of low
coordination sites which we have previously commented as
being important to CH activation. The subsequent gradual
4
decline in yield (and selectivity) is then attributable to a
gradual re-equilibration of the surface structure to that which
is more stable under dynamic synthesis conditions, whilst the
di†erence between the top two data curves in Fig. 6 illustrates
once again how efficient vacuum annealing is in removing the
surface roughening which leads to high HCN yields. It is also
worth noting that the e†ects that we see with ion bombard-
ment are very similar in many respects to those reported when
Pt-monolith catalysts (and PtÈRh gauzes) are transiently
heated to higher temperatures in an oxygen rich mixture.23
Carbon deposition by pretreatment in pure ethylene is only
partially e†ective in poisoning HCN synthesis. There is a very
slight delay in the initiation of the reaction but the efficiency
high HCN yields on the pure Pt surface. In NH oxidation,
3
the greater resistance of the alloy to surface reconstruction has
been demonstrated in a direct comparison of the rate of
etching of Pt and PtÈ10%Rh crystals placed in a high pressure
reactor;26 however, this does relate to conditions which are
far more oxidising than those pertaining to HCN synthesis
and could be associated with the volatility of higher oxidation
state Pt compounds. At lower temperatures and in a less oxi-
dising environment, a greater resistance to reconstruction
might be related to the lower surface (and bulk) atom mobility
J. Chem. Soc., Faraday T rans., 1997, V ol. 93
3877

View Article Online
in the alloy that is to be predicted on the basis of its higher
melting point and greater mechanical strength; certainly such
a resistance to change is evident in our experiments by the
way in which the sample history inÑuences the results from
the PtÈ10%Rh alloy, which indicates that the surface equili-
brates much more slowly to new reaction conditions than is
the case for pure Pt. However, if di†erences in the observed
extent of reconstruction were purely due to slower kinetics of
reconstruction then the HCN activity of the alloy might ulti-
mately more closely approach that of Pt, whereas the limiting
value implied by the slow decline in activity that we see after
pretreatment by ion bombardment suggests that this is deÐ-
nitely not the case; i.e. reduced restructuring under reaction
conditions must be intrinsic to the nature of the alloy.
Notably, Cowans et al.6 also comment brieÑy on an apparent,
inverse correlation between local Rh concentrations and the
extent of localised surface restructuring in gauze catalysts used
for HCN synthesis.
tions play an important role in supporting an active surface
structure
4. a PtÈ10%Rh alloy surface gives lower HCN yields and
selectivity, and an increase in oxidation products when com-
pared to pure Pt
5. reactions on the PtÈ10%Rh alloy surface are further
inÑuenced by Rh surface segregation (under anoxic
conditions) and memory e†ects (whereby the e†ects of pre-
vious sample treatments persist for some time under the
testing conditions).
Financial assistance from the EPSRC (GR/K 23218) is grate-
fully acknowledged. We also wish to thank ICI Acrylics plc
for technical and material support; in particular, Dr. S. A.
Axon and S. J. Frank for useful discussions on the industrial
aspects of the Andrussow process.
We have also observed an increase in oxidation catalysis on
the PtÈ10%Rh surface relative to that seen for pure Pt. Within
our model this is a direct consequence of a less efficient activa-
tion of methane and would thus follow automatically if the
alloy surface exhibits less reconstruction. However, it may also
reÑect the greater affinity of Rh for oxygen which will tend to
increase the steady state concentration of surface oxidants and
References
1
C. N. SatterÐeld, Heterogeneous Catalysis in Practice, McGraw-
Hill, New York, 2nd edn., 1991.
2
B. Y. K. Pan and R. G. Roth, Ind. Eng. Chem. Process Res. Dev.,
1968, 7, 53.
3
4
5
6
B. Y. K. Pan, J. Catal., 1971, 21, 27.
L. D. Schmidt and D. Luss, J. Catal., 1971, 22, 269.
D. R. Anderson, J. Catal., 1988, 113, 475.
B. A. Cowans, K. A. Jurman, W. N. Delgass, Y. Z. Li, R. Reifen-
berger and T. A. Koch, J. Catal., 1990, 125, 501.
C. L. Levoguer and R. M. Nix, J. Chem. Soc., Faraday T rans.,
1996, 92, 4799, and references therein.
D. Hasenberg and L. D. Schmidt, J. Catal., 1985, 91, 116.
D. Hasenberg and L. D. Schmidt, J. Catal., 1986, 97, 156.
thus lead to enhanced yields of CO and NO at the expense of
x
HCN selectivity.
Finally, we shall brieÑy consider how the results reported
here may relate to features of the commercial Andrussow
process. With regard to the comparison of Pt and PtÈ10%Rh,
our data indicates a clear beneÐt in using a pure Pt catalyst.
This neglects any consideration of the mechanical beneÐts acc-
ruing from alloying but, nevertheless, our conclusion supports
the observation of Hickman et al.27 who, despite working
under very di†erent conditions, also noted that the addition of
Rh reduces the HCN yield and selectivity.
7
8
9
10 D. Hasenberg and L. D. Schmidt, J. Catal., 1987, 104, 441.
11 N. Waletzko and L. D. Schmidt, AIChE J., 1988, 34, 1146.
12 V. K-K. Wang, US Pat. 4,107278 (Aug. 15, 1978)
13 D. G. Loffler and L. D. Schmidt, Surf. Sci., 1976, 59, 195.
14 Practical Surface Analysis, ed. D. Briggs and M. P. Seah, 1990
Wiley, 1990, vol. 1.
15 A. N. Salanov and V. I. Savchenko, Surf. Sci., 1993, 296, 393.
16 K. A. Peterlinz and S. J. Sibener, J. Phys. Chem., 1995, 99, 2817.
17 S. Meyer and R. M. Nix, unpublished results.
18 T. Pignet and L. D. Schmidt, Chem. Eng. Sci., 1974, 29, 1123; J.
Catal., 1975, 40, 212.
The commercial process is characterised by substantial
macroscopic restructuring of the gauzes and also by an induc-
tion period prior to optimal HCN production after the install-
ation of new gauze packs; although unproven we believe these
phenomena are related to the importance of microscopic
surface restructuring that we have seen evidence for in this
work.
19 C. W. Nutt and S. Kapur, Nature (L ondon), 1968, 220, 697; 1969,
224, 169.
20 J. M. Bradley, A. Hopkinson and D. A. King, J. Phys. Chem.,
1995, 99, 17032.
21 E. J. Kelly, W. R. Patterson and J. J. Rooney, J. Mol. Catal.,
1994, 89, 19.
Conclusions
22 D. F. Johnson and W. H. Weinberg, J. Chem. Phys., 1994, 101,
Studies of HCN synthesis on pure Pt and Pt-alloy surfaces, at
total pressures of 0.02È0.15 Torr and temperatures of up to
780 ¡C, indicate that:
1. efficient methane activation is a crucial factor in obtain-
ing HCN in high yield and with good selectivity,
2. the structure of the catalyst surface has a major inÑuence
on the reaction; in particular, low coordination Pt sites
appear to be important in methane activation and HCN syn-
thesis,
6289.
23 S. S. Bharadwaj and L. D. Schmidt, Ind. Eng. Chem. Res., 1996,
35, 1524.
24 A. D. van Langeveld and J. W. Niemantsverdriet, Surf. Sci., 1986,
178, 880.
25 N. Savargaonkar, B. C. Khanra, M. Pruski and T. S. King, J.
Catal., 1996, 162, 277.
26 J. Pielaszek, Plat. Metals Rev., 1984, 28(3), 109.
27 D. A. Hickman, M. Hu† and L. D. Schmidt, Ind. Eng. Chem.
Res., 1993, 32, 809.
3. anoxic conditions do not give a sustainable HCN yield,
suggesting surface oxygen atoms or exothermic oxidising reac-
Paper 7/04184E; Received 16th June, 1997
3878
J. Chem. Soc., Faraday T rans., 1997, V ol. 93