Structure and stability of CN adlayers on Rh(110)

Article abstract of DOI:10.1016/S0039-6028(00)00520-3

The formation and stability of CN adlayers on Rh(110), formed by dissociative adsorption of C₂N₂ at 373 K, have been studied as a function of coverage and temperature by low-energy electron diffraction (LEED), X-ray photoelectron spectroscopy (XPS), and thermal desorption spectroscopy (TDS). Two different CN adsorption states have been distinguished by their different C 1s and N 1s XPS core-level binding energies. The CN-I state is exclusively occupied up to a surface coverage of 0.5 monolayers (ML), where a well-ordered c(2 x 2) LEED pattern is observed. The CN-II state becomes additionally populated at higher coverages from 0.5 ML to the saturation coverage of 0.87 ML. At CN saturation, a c(4 x 2) LEED structure is formed. Desorption of CN as molecular C₂N₂ occurs only for surface coverages >0.5 ML and appears to be mainly derived from the CN-II state. The onset of C-N bond rupture is indicated at ～450-550 K, depending on the CN coverage; the resulting N_ad desorbs at ～580 K from the more crowded surface and in the range of ～650-950 K, whereas C_ad remains at the Rh surface and cannot be desorbed thermally.

Full text of DOI:10.1016/S0039-6028(00)00520-3

Surface Science 461 (2000) 31–44
www.elsevier.nl/locate/susc
Structure and stability of CN adlayers on Rh(110)
*
Federica Bondino a, Alessandro Baraldi b, Giovanni Comelli a, Falko P. Netzer c,
a TASC-INFM Laboratory, Padriciano 99, I-34012 Trieste, Italy and Dipartimento di Fisica, Universita` di Trieste, via Valerio 2,
I-34127 Trieste, Italy
b Sincrotrone Trieste, S.S. 14 Km 163.5, 34012 Trieste, Italy
c Institut fur Experimentalphysik, Karl-Franzens-Universitat Graz, A-8010 Graz, Austria
¨
¨
Received 21 January 2000; accepted for publication 3 April 2000
Abstract
The formation and stability of CN adlayers on Rh(110), formed by dissociative adsorption of C N at 373 K,
2
2
have been studied as a function of coverage and temperature by low-energy electron diffraction (LEED), X-ray
photoelectron spectroscopy ( XPS), and thermal desorption spectroscopy (TDS). Two different CN adsorption states
have been distinguished by their different C 1s and N 1s XPS core-level binding energies. The CN-I state is exclusively
occupied up to a surface coverage of 0.5 monolayers (ML), where a well-ordered c(2×2) LEED pattern is observed.
The CN-II state becomes additionally populated at higher coverages from 0.5 ML to the saturation coverage of
0.87 ML. At CN saturation, a c(4×2) LEED structure is formed. Desorption of CN as molecular C N occurs only
2
2
for surface coverages >0.5 ML and appears to be mainly derived from the CN-II state. The onset of CMN bond
rupture is indicated at ~450–550 K, depending on the CN coverage; the resulting N desorbs at ~580 K from the
ad
more crowded surface and in the range of ~650–950 K, whereas C remains at the Rh surface and cannot be
ad
desorbed thermally. © 2000 Elsevier Science B.V. All rights reserved.
Keywords: Cyanogen; Low energy electron diffraction (LEED); Rhodium; X-ray photoelectron spectroscopy
1. Introduction
has been reported recently. The knowledge of the
adsorption properties of CN species on Rh surfaces
is therefore an important prerequisite to assess the
surface chemistry of these cyanide reaction inter-
mediates. Moreover, interest in CN surface chemis-
try on Rh derives from many large-scale industrial
processes, such as the Andrussow process for the
synthesis of HCN from CH , NH , and O over
Rh–Pt surfaces. The bonding and surface stability
of CN species also have a model character for the
surface interactions of organic and polymeric nit-
riles [4], where the CN functional groups provide
important centres for surface attachement.
Rhodium is an important metallic constituent
in catalytic automotive converters. One of the
reasons for using the expensive Rh in the three-
way catalysts of automotive exhausts is that it
appears to be particularly active for the removal
of NO, via the NO+CO reaction, converting it
to N and CO , or by reduction with hydro-
4
3
2
2
2
carbons. The formation of cyanide intermediates
in the catalytic reduction of NO by C H on
Rh(111) surfaces [1,2] and during the interaction
of CO and NO on ceria supported Rh surfaces [3]
2 4
The adsorption and decomposition of C N to
2 2
form CN on Rh(111) has been studied by
* Corresponding author. Fax: +43-316-3809816.
E-mail address: netzer@kfunigraz.ac.at (F.P. Netzer)
Solymosi and Bugyi [5] and Hwang et al. [6] using
0039-6028/00/$ - see front matter © 2000 Elsevier Science B.V. All rights reserved.
PII: S0039-6028 ( 00 ) 00520-3

32
F. Bondino et al. / Surface Science 461 (2000) 31–44
Auger electron spectroscopy, LEED, and thermal
desorption spectroscopy (TDS). Both studies
agree that above room temperature, adsorbed
sured. The use of a channeltron, mounted behind
a 100 mm aperture, makes it possible to use an
electron beam current of less than 0.1 mA, thus
minimizing damage to the adsorbate layer under
investigation. The transfer width of the instrument
C N dissociates to form surface CN, which in
2 2
turn partly dissociates at a higher temperature
˚
into N and C , but partly also desorbs as
is better than 1000 A at an electron kinetic energy
of 85 eV.
ad
ad
C N for higher CN coverages. Below room tem-
2 2
perature, molecular C N adsorption has been
inferred indirectly [5,6]. No ordered LEED pattern
The sample, a Rh(110) disk, was mounted on
to the manipulator through two tungsten wires
and heated by direct current. The cleaning pro-
cedure involved repeated cycles of Ar+ bombard-
ment at room temperature and annealing to
1300 K followed by oxygen treatment at 1100 K
(10 min, 1×10−7 mbar). The sample was then
flashed to 1300 K and, while cooling, reduced in
hydrogen (5×10−7 mbar) in order to remove any
residual oxygen. Cyanogen was produced by ther-
mal decomposition of AgCN in a tube attached
to the gas line.
2 2
has been observed for CN on Rh(111) [5].
Here, we report a study of the structure and
surface stability of CN adlayers on Rh(110), using
LEED, X-ray photoelectron spectroscopy ( XPS),
and TDS. The CN surface species have been
obtained by dissociative adsorption of C N at
2 2
373 K. We have investigated the formation of the
CN adlayers as a function of coverage and have
observed the formation of several ordered CN
surface structures in LEED. The absolute CN
coverages have been determined by C 1s and N 1s
core-level XPS, using a well-ordered c(2×2)
LEED structure as the calibration point. The
competitive processes of CMN bond cleavage and
3. Results
associative C N desorption have been specified as
a function of temperature and coverage. Moreover,
plausible models of the ordered CN LEED struc-
tures and a consistent description of their evolution
with increasing surface population are presented.
Fig. 1 illustrates the formation of a CN adlayer
2 2
on Rh(110) after exposure to 20 L C N
2 2
[1 Langmuir (L)=1×10−6 Torr.s] at 100 K fol-
lowed by stepwise annealing to 375 K. Fig. 1a
displays a sequence of N 1s core-level spectra,
whereas Fig. 1b shows the corresponding C 1s XPS
spectra. After adsorption at 100 K, two spectral
structures are observed at 397.8 0.1 and 399.5 eV
for the N 1s and at 284.7 0.1 and 287 eV for the
C 1s spectra. The peaks at the higher binding
energies (BE) (399.5 and 287 eV for N 1s and
2. Experimental
The experiments were carried out in a standard
UHV chamber equipped with a rear-view LEED,
a 150 mm mean radius hemispherical electron
energy analyser with a multichannel detector, a
conventional laboratory X-ray source, a quadru-
pole mass spectrometer and facilities for gas inlet
and sample cleaning. The XPS spectra were taken
using the MgKa source operated at 800 W power
and with a fixed analyser pass energy of 20 eV.
The binding energies were calibrated using the
C 1s, respectively) are similar to the C N multi-
2 2
layer structures observed on Pd(110) [7] and are
therefore also associated here with a condensed
C N phase. The structures at 397.8 eV (N 1s) and
2 2
284.7 eV (C 1s) then derive from the molecular
C N monolayer. On heating, the condensed
2 2
C N desorbs, and at ~200 K, the monolayer is
2 2
left at the surface. Note that no ordered LEED
Rh 3p peak as a reference. Thermal desorption
experiments were performed with the sample in
pattern has been observed from the C N mono-
layer on Rh(110), in contrast to the c(2×2)
3/2
2 2
front of the mass spectrometer using a linear
temperature ramp of 3 K/s. The chamber is also
equipped with an Omicron SPA-LEED system,
which allows diffraction spot profiles to be mea-
LEED patterns observed on Pd(110) [8] and
Ni(110) surfaces [9]. Further annealing induces
dissociation of C N to adsorbed CN, which is
2 2
characterized by C 1s and N 1s binding energies

F. Bondino et al. / Surface Science 461 (2000) 31–44
33
Fig. 1. N 1s (a) and C 1s (b) XPS core-level spectra of the Rh(110) surface exposed to 20 L C N at 100 K and heated subsequently
2
2
to the indicated temperatures. The assignment of XPS peaks to various surface species is indicated.
of 284.3 and 397.4 eV, respectively. At room tem-
following C N dosing experiments have all been
carried out at 373 K, to ensure the formation of
pure CN adlayers.
2 2
perature (~300 K), the dissociation of C N into
2 2
CN appears to be still somewhat incomplete, as
also suggested by the TDS reported in the litera-
In Fig. 2, N 1s and C 1s XPS spectra are pre-
sented for increasing CN surface coverages on
Rh(110). The areas under the nitrogen XPS peaks
have been taken to derive the absolute CN surface
coverage, using the sharp c(2×2) LEED pattern
observed (see below) for calibration purposes, with
a coverage of 0.5 ML. The corresponding cover-
ages are indicated in Fig. 2, together with the two
major LEED structures observed. Up to a coverage
ture, where a so-called a C N desorption peak
2 2
between 310 K [5] and 380 K [6] has been associ-
ated with desorption from a molecular C N
2 2
adsorption state. The XPS spectra of Fig. 1
recorded after annealing to 373 K indicate, how-
ever, that at that temperature, all C N has dissoci-
2 2
ated into CN . Since the prime interest of this
ad
study focuses on the adsorption of CN, the

34
F. Bondino et al. / Surface Science 461 (2000) 31–44
Fig. 2. N 1s (a) and C 1s (b) XPS core-level spectra of CN on Rh(110) as a function of CN surface coverage.
of 0.5 ML, the XPS data can be fitted with a single
peak. For the analysis, the spectral profiles have
been represented by Doniac–Sunjic line shapes
components to the analysis of the higher coverage
spectra, located at 397.0 0.1 eV and 284.1
0.1 eV for N 1s and C 1s, respectively (CN-II).
The N 1s and C 1s XPS binding energies of all
CN-derived surface species observed on Rh(110)
are collected in Table 1.
(lorentzian width C =0.30 eV for C 1s and
L
0.32 eV for N 1s) convoluted by a gaussian
(gaussian width=0.95 eV, i.e., fixed to the experi-
mental resolution). The parameters obtained for
fitting at low coverages have then been applied to
the whole set of data. Thus, the c(2×2) LEED
structure contains a single CN adsorption state,
henceforth designated as CN-I, with N 1s and C 1s
binding energies of 397.4 0.1 and 284.3 0.1 eV,
respectively. Above 0.5 ML CN coverage, it was
not possible to fit the N 1s and C 1s data with a
single peak with fixed BE. Thus, we added second
Fig. 3 shows an uptake curve of CN on Rh(110)
in form of a plot of the CN coverage versus
C N exposure at 373 K. The figure contains the
2 2
total CN coverage and the coverage curves of the
CN-I and CN-II adsorption states as derived from
the two components of the curve-fitting analysis
of the N 1s spectra. The delayed population of the
CN-II state is clearly apparent from Fig. 3. The
plots obtained from the C 1s peaks (not shown)

F. Bondino et al. / Surface Science 461 (2000) 31–44
35
Fig. 3. CN uptake curve on Rh(110), in form of a plot of the CN coverage versus C N exposure at 373 K. The total CN coverage
2
2
and the coverage curves of the CN-I and CN-II adsorption states are given.
follow a similar behaviour, but the coverage ratio
between the CN-I and CN-II species is different.
We believe that the information obtained from the
N 1s area is more reliable, for the following
reasons: (1) the CN-I and CN-II components are
better separated in the N 1s region; (2) the cross-
section of the N 1s electrons is higher, yielding
better statistics; (3) in case the CN species are
tilted with the C end down, as found for CN on
Ni(110) [12], photoelectron diffraction effects are
expected to be more relevant for the C 1s peak.
Fig. 4 shows the evolution of LEED patterns
with increasing CN surface coverage, obtained
~0.2 L C N exposure, weak diffuse (1/2 1/2)
2 2
order reflexes first develop, which are elongated
along the [01] direction (Fig. 4a). These reflexes
then evolve into two distinct spots, thus forming
a c(2×2) structure with split spots along [01].
The spots move closer together with increasing
exposure (0.4–1 L), until a sharp c(2×2) pattern
is observed best after C N exposure of 1 L
2 2
(Fig. 4b). The spot profiles as a function of expo-
sure have been measured quantitatively using the
SPA-LEED apparatus (not shown), and the fully
developed c(2×2) structure has been used for
coverage calibration purposes, as described above.
For C N doses between 1 and 8 L, the (1/2 1/2)
after C N dosing on to Rh(110) at 373 K. After
2 2
2 2
order reflexes become split again, but this time
along the [10] direction (Fig. 4c). The separation
of the reflexes increases continuously up to the
maximum coverage of 0.87 ML, where a well-
ordered c(4×2) pattern is observed (Fig. 4d). On
annealing the sample to >375 K, this sequence of
structures can be observed in reverse order. The
c(2×2) pattern vanishes completely at ~823 K.
The temperature evolution and thermal stability
of the CN adlayers have been followed by TDS
and XPS. Let us consider first the TDS results in
Table 1
C 1s and N 1s core-level binding energies of C N - and
CN-derived surface species on Rh(110)
2
2
N 1s (eV
C 1s (eV)
C N multilayers
399.5
397.8
397.4
397.0
397.9
287.0
284.7
284.3
284.1
283.8
2
2
2
C N monolayer
2
CN-I
CN-II
N
, C
ad ad

36
F. Bondino et al. / Surface Science 461 (2000) 31–44
Fig. 4. LEED patterns of CN on Rh(110). (a) c(2×2) pattern with split spots along the [01] surface direction (corresponding to the
[001] crystal direction). (b) c(2×2) pattern. (c) c(2×2) pattern with split spots along the [10] surface direction (corresponding to
the [110] crystal direction). (d) c(4×2) pattern.
Fig. 5. Fig. 5 contains TDS spectra of atomic mass
units 12, 52, 14, and 28 (from bottom to top),
corresponding to C N (52, 12) and N (28, 14);
the desorption of undesired CO minority species,
picked up by the surface from the system ambient
at saturation coverage (see AMU 12 and 52
spectra). Note that no C N desorption occurs
2 2
from the c(2×2) CN overlayer.
2 2
2
The AMU 28 spectra show a more complicated
behaviour, with several overlapping desorption
peaks. We distinguish a weak feature at around
500 K for low coverages, a structure at 570 K for
high coverages, and a broad two-peaked desorp-
tion feature in the range of 700–950 K. The lower-
temperature peak of this latter structure is initially
located at ~660 K and shifts to 770 K at higher
or from the C N source, can be identified by
2 2
cross-examination of the AMU 14 and 28 spectra.
Molecular C N desorption is observed only for
2 2
CN surface coverages >0.5 ML in a single desorp-
tion peak between 473 and 623 K, with the peak
maximum shifting from 583 K initially to 543 K

F. Bondino et al. / Surface Science 461 (2000) 31–44
37
coverages. The second peak maximum is seen at
around 860–880 K. The signal of AMU 28 can in
principle be related to N and CO molecules. N
2
2
desorption is expected from the recombination of
adsorbed N atoms generated by the CMN bond
cleavage process. The AMU 14 spectra allow
desorbing N to be distinguished from CO. The
2
AMU 14 spectra show a good correspondence
with the AMU 28 spectra apart from the small
AMU 28 feature at ~500 K at low coverages,
which is thus revealed as due to minor CO desorp-
tion. N desorption is therefore identified at 570 K
2
for high CN coverages and in the broad range of
660–950 K.
The thermal behaviour of CN adlayers on
Rh(110) in XPS is illustrated in Fig. 6. Fig. 6a
contains a sequence of N 1s spectra, and Fig. 6b a
set of C 1s spectra, as a function of annealing to
the indicated temperatures of the initially CN
saturated surface (h =0.87 ML). At higher tem-
CN
peratures (>500 K), the XPS data could not be
fitted with the same parameters as those used for
the spectra after adsorption at 373 K: third spectral
components had to be introduced in the analysis,
with a N 1s BE of 397.9 eV and a C 1s BE of
283.8 eV. These components are associated with
atomic N and C , formed by the CN decomposi-
ad
ad
tion via CMN bond scission. Their binding ener-
gies are included in Table 1.
A detailed analysis of the XPS core-level spectra
of Fig. 6 in terms of the (de)population of different
CN derived surface species as a function of temper-
ature is given in Fig. 7a, for N 1s and Fig. 7b for
C 1s. Fig. 7a demonstrates that the total N 1s
signal decreases gradually from ~450 K on and
vanishes at ~900 K. In contrast, the total C 1s
signal intensity (Fig. 7b) is still about a half of its
initial value at that temperature. Both N 1s and
C 1s signals show that the CN-II state becomes
depopulated first, disappearing from the surface at
550–600 K. In conjunction with the TDS results
of Fig. 5, this indicates that the disappearance of
the CN-II species is correlated with the desorption
of C N . The CN-I-related N 1s and C 1s signals
remain approximately constant up to ~550–600 K
and then decrease gradually. Concomitantly, the
2 2
Fig. 5. Thermal desorption spectra of CN on Rh(110) display-
ing desorption curves of atomic mass units 12, 52, 14, and 28.
The surface coverages increase from bottom to top curves in
each panel. The corresponding LEED patterns are indicated.
Heating rate 3 K/s.
atomic adsorbate signals of N and C appear.
ad
ad
This signals the dissociation of the CN-I species

38
F. Bondino et al. / Surface Science 461 (2000) 31–44
Fig. 6. N 1s (a) and C 1s (b) XPS core-level spectra of the CN saturated Rh(110) surface as a function of annealing the surface to
the indicated temperatures.
and CMN bond rupture, which starts at ~550–
tion of CN adlayers. As a function of surface
coverage, two CN adsorption states have been
identified, which are distinguished by their respec-
tive core-level binding energies. It is of interest to
compare the two CN surface species on Rh(110)
with those found on Pd(110) and Ni(110), for
which photoelectron spectroscopy data have also
been reported in the literature. On Pd(110), only
one CN adsorption state has been detected with
C 1s and N 1s binding energies of 284.2 and
397.15 eV, respectively [7]. A previous angle-
resolved UV photoemission (ARUPS) study has
600 K. The N signal first increases to a broad
ad
maximum and then decreases again as a result of
the recombinative desorption, as indicated by the
TDS data. The C remains as a residue at the
ad
Rh(110) surface.
4. Discussion
The adsorption of C N on Rh(110) above
2 2
300 K leads to dissociative adsorption and forma-

F. Bondino et al. / Surface Science 461 (2000) 31–44
39
Fig. 7. Plot of the coverages of the various CN derived surface species on Rh(110) as a function of temperature. Panel (a) has been
derived from the N 1s XPS spectra of Fig. 6a, panel (b) from the C 1s spectra of Fig. 6b.
placed the CN species on Pd(110) into the grooves
of the (110) surface, with an orientation parallel
to the [110] direction [10]. The situation appears
to be more complex on Ni(110). ARUPS and
local density functional cluster calculations have
indicated that the CN species in the c(2×2) room-
temperature saturation structure on Ni(110) are
also flat lying in-groove species, which are oriented
parallel to the [110] azimuthal direction [11].
However, a recent NEXAFS and photoelectron
diffraction (Ph.D.) study of the Ni(110)
c(2×2)-CN surface has concluded that the CN is
flat-lying but approximately aligned in the [001]
azimuth [12]. The CN has still been positioned
within the grooves of the (110) surface, but in a
rather unusual asymmetric site: according to Ph.D.
CN lies approximately atop a second-layer Ni
atom such that the C atom has similar distances
to this second-layer Ni atom and to two top-layer
Ni atoms, thus implying essentially a threefold
coordination for the C atom. The N atom is then
bridge-bonded to two top-layer Ni atoms of the

40
F. Bondino et al. / Surface Science 461 (2000) 31–44
next close-packed surface atom row [12]. Very
recently, XPS and HREELS experiments on the
c(2×2) CN structure on Ni(110) [13,14] have
suggested, however, that two CN adspecies might
possibly be involved in this adlayer, with the same
C 1s BE of 283.3 eV, but with two N 1s BE compo-
nents of 396.2 and 396.5 eV, implying different
environments of the N atoms for the two adsorp-
tion states. Additional CN species in a disordered
form have been found on Ni(110) that display
higher binding energies than the CN species in the
c(2×2) structure and have been associated with
various CN-bonding modes in on-ridge configura-
tions [13].
signals at low CN coverages (not shown).
Molecular C N desorption is not observed for
2 2
h
<0.5. The N desorbs as N at ~600–900 K,
CN
ad
2
whereas C is not removed from the surface by
ad
heating. For h >0.5 ML, both dissociation and
CN
C N desorption occur, the latter in a single
2 2
desorption structure at 540 K at saturation cover-
age; the shift of the desorption peak from a higher
to lower temperature with increasing CN surface
coverage is as expected for second-order desorp-
tion kinetics. From the XPS results of Fig. 7, it
appears that the CN-II state is mostly involved in
the desorption process. However, the recombina-
tive desorption of CN at h >0.5 is competitive
CN
The present results on Rh(110) do not allow a
unique specification of the bonding sites of the
CN surface species. It is plausible, however, to
associate the CN-I state, which is exclusively popu-
lated in the c(2×2) structure, with in-groove CN
species, but the orientation and the exact bonding
sites must remain uncertain. In view of the close
similarity of the C 1s and N 1s binding energies
with those of the CN species on Pd(110), we
tentatively assign the CN-I state to a flat-lying
CN, probably oriented along the [110] direction,
i.e., parallel to the close-packed Rh rows. The
CN-II state, which becomes occupied for higher
coverages, beyond the c(2×2) structure up to the
c(4×2) structure, may then be associated with
in-groove species in a tilted configuration, perhaps
oriented along [001]. On-ridge CN species are
considered as less likely in view of the higher core-
level binding energies observed for these configu-
rations on Ni(110) [13]. In any case, it is likely
that various bonding configurations of CN exhibit
only small differences in total energy, as recently
reported by Yang and Whitten on the basis of
total energy calculations of CN on Ni(100) [15].
The two CN configurations as mentioned above
will be used in the following to construct the
models of the LEED structures, as discussed fur-
ther below.
with dissociation, and a redistribution of adsorp-
tion states near the desorption temperature may
be possible. For the more crowded CN surfaces
(h >0.5), a lower-temperature N desorption
CN
2
peak is observed at 580 K. This desorption state
is similar to N from the clean Rh(110) surface
2
[16]. It seems to reflect the desorption of N ,
ad
which has been generated in the first partial dissoci-
ation step of CN in competition with C N desorp-
2 2
tion, in the presence of CN (state I). The N
ad
2
desorption after dissociation of most of the CN
ad
at 700–950 K is at a considerably higher temper-
ature than for N adsorption on the clean Rh(110)
surface. It is instructive to compare the high-
temperature N desorption observed here with the
2
previously published TDS results from a mixed
C +N layer [17]. The N desorption traces are
ad
ad
2
very similar in these two cases. The high-temper-
ature N desorption has been related to the pres-
2
ence of C at the surface: the diffusion of N is
ad
ad
hindered by C , and therefore, the probability of
a diffusing N adatom meeting another N atom for
ad
recombination and desorption is reduced [17].
However, some reaction-limited desorption of N
via CN decomposition cannot be excluded to
2
contribute to the high-temperature desorption
structure (see Fig. 7).
The stability of surface CN with temperature
appears to be lower on Rh(110) than on Rh(111)
surfaces. Solymosi and Bugly [5] and Hwang et al.
The temperature stability of the CN adlayers
on Rh(110) is of some interest for the surface
chemistry of CN reaction intermediates. For cover-
ages <0.5 ML, CN dissociates completely. The
onset of dissociation is around 450 K, as derived
from the appearance of the C and N XPS
[6] have reported recombinative C N desorption
2 2
from Rh(111) in several desorption peaks up to
more than 800 K, and van Hardevelt et al. [18]
have observed that CN groups are stable in the
ad
ad

F. Bondino et al. / Surface Science 461 (2000) 31–44
41
Fig. 8. Schematic models of the c(2×2)-CN structure (a) and the c(2×2)-CN structure with reflexes split along the [01] surface
direction (b) (the corresponding LEED patterns are shown in Fig. 4b and a, respectively). The dashed lines in (b) indicate the domain
boundaries, which are responsible for the splitting of the LEED reflexes.
absence of surface hydrogen up to ~700 K. CN
decomposition-limited desorption of on
CMN dissociation occurs already at room temper-
ature [13]. Interestingly, however, Pd(110) has a
N
2
Rh(111) has been detected consistently by all three
groups at ~800–870 K [5,6,18]. This is consider-
ably higher in temperature than the observed CMN
bond breaking temperatures of ~450–550 K,
depending on coverage, on Rh(110).
Compared with the (110) surfaces of Pd and
Ni, the reactivity of Rh(110) for CMN bond
rupture is intermediate. Whereas no CMN bond
breaking occurs on Pd(110) — all CN can be
much greater ability to dissociate C N to CN
2 2
than Rh(110): on Pd(110), the C N dissociation
2 2
reaction is complete at 220 K [7], while on
Rh(110), temperatures between 300 and 370 K are
required for complete dissociation of C N adlay-
2 2
ers. Ni(110) again is very reactive, and C N
2 2
undergoes partial dissociation at 90 K [13].
CN on Rh(110) displays two well-ordered
LEED structures and interesting transitions
between them. Models for these structures will be
desorbed as C N [7,10] — on Ni(110), some
2 2

42
F. Bondino et al. / Surface Science 461 (2000) 31–44
Fig. 9. Schematic models of the c(4×2)-CN structure (a) and of the c(2×2)-CN structure with reflexes split along the [10] surface
direction (b) (the corresponding LEED patterns are shown in Fig. 4d and c, respectively). The dashed lines in (b) represent the
domain boundaries.
presented in the last part of this discussion. As
mentioned above, it is emphasized that the CN
adsorption sites adopted for these models are
somewhat tentative and arbitrary. However, the
general character of the structures and the conclu-
sions drawn on their transitions remain valid,
irrespective of the true nature of the CN bonding
sites. The c(2×2) structure of CN on Rh(110)
structure, where we have placed the CN in the
grooves, parallel to [110] in a bridge-bonded site
[10]. The fully developed c(2×2) structure has
been used to calibrate the CN surface coverage
with
a
value of 0.5 ML, but already at
~0.40 0.02 ML, a c(2×2) structure is observed,
albeit with half-order spots split along the [01]
direction (Fig. 4a). The reason for this spot-split-
ting effect is the existence of domains boundaries
and antiphase relationships between them. The
splitting direction indicates that the domain walls
are oriented along the [10] direction, i.e., perpen-
dicular to the splitting direction. Careful examina-
tion of the separation of the split spots by SPA-
˚
has a domain width of ~200 A, as judged from
the full width at half maximum of the (1/2 1/2)
spots, measured in the SPA-LEED apparatus. This
indicates strong long-range interactions between
the CN molecules. c(2×2)-CN structures have
also been observed on Pd(110) and Ni(110) sur-
faces. Fig. 8a shows a schematic model of this
LEED gives a separation of 1/7 a . This leads to
1

F. Bondino et al. / Surface Science 461 (2000) 31–44
43
a model, as shown in Fig. 8b, where the c(2×2)
domains are separated by a domain wall and
shifted by one unit cell vector along [110] every
seventh row. The surface coverage of this model
is 0.42 ML, in close agreement with the XPS-
derived coverage value. The c(2×2) domains are
separated by so-called light domain walls, which
contain no adsorbate species. As the coverage
increases from 0.4 to 0.5 ML, the spot splitting
decreases due to the growth of c(2×2) domains,
and broad diffuse half-order reflexes develop, as a
result of a more random domain size distribution.
As the coverage of 0.5 ML is reached, the density
of domain walls is low, and the sharp c(2×2)
pattern of Fig. 4b is observed.
The c(4×2) structure contains two different
CN adsorbate configurations, as evidenced by
XPS. In our model, they are associated with
in-groove species parallel (CN-I) and perpendicu-
lar (CN-II) to the [110] direction (Fig. 9a). The
model shown in the figure yields a coverage of 7/8
(0.875) ML, in good agreement with XPS, and is
obtained by filling the c(2×4) structure with CN-I
species, four molecules per unit cell, and with
CN-II species, three molecules per unit cell. In this
way, adsorbate lines are formed with CN species
arranged with their axes alternatively parallel and
perpendicular to the [110] direction.
For coverages between 0.5 and 0.74 ML, LEED
patterns with (1/2 1/2) spot splittings along the
[10] direction have been observed (Fig. 4c). Again,
it is suggested that the spot splitting is related to
the existence of antiphase domains, with domain
boundaries perpendicular to [01]. The proposed
model is a phase between the c(2×2) and the
c(4×2) structures, with heavy domain walls in
this case, i.e., with the local surface coverage being
higher than 0.5 in the domain boundaries. The
model is presented in Fig. 9b. The idea is that the
intermolecular interaction between two adjacent
CN species in the domain wall induces a rotation
of the molecular axis to reduce repulsion effects,
thus creating a precursor arrangement to the
c(4×2) structure. The spot splitting increases with
increasing surface coverage until the c(4×2) struc-
ture emerges. In this way, the different models
(Figs. 8 and 9) provide a plausible description of
the CN surface structures on Rh(110) and of their
transformations, which incorporates consistently
both adsorption state and coverage information,
as obtained by XPS. A more quantitative descrip-
tion must await LEED intensity analysis, which is
presently underway.
5. Summary
We have studied the properties of CN adlayers
on Rh(110), formed by dissociative adsorption of
C N at 373 K, as a function of surface coverage
2 2
and temperature, using LEED, XPS and thermal
desorption spectroscopy (TDS). Two different CN
adsorption states have been detected by their
different C 1s and N 1s XPS binding energies:
CN-I is exclusively populated up to 0.5 ML and
forms a well-ordered c(2×2) surface structure.
The CN-II species occur additionally at higher CN
coverages from 0.5 ML up to the saturation cover-
age of 0.87 ML, where a c(4×2) LEED structure
is formed. Desorption of CN as molecular C N
2 2
is observed only for coverages >0.5, and is mainly
due to CN-II species. CMN bond rupture is shown
to begin at 450–550 K, depending on the CN
coverage. The resulting N desorbs at ~580 K
ad
from the more crowded surfaces and in the range
of 650–950 K, whereas C cannot be desorbed
ad
thermally. Schematic models for the ordered CN
surface structures on Rh(110) and for their trans-
itions are discussed.
Acknowledgements
We would like to thank Professor Renzo Rosei
for helpful discussions. Financial support from the
Italian MURST under the program ‘COFIN99’ is
acknowledged. F.P.N. acknowledges support from
the Austrian Science Foundation.
References
[1] R.M. van Hardevelt, A.J.G.W. Schmidt, J.W. Niemantsv-
erdriet, Catal. Lett. 41 (1996) 125.
[2] R.M. van Hardevelt, A.J.G.W. Schmidt, R.A. van Santen,

44
F. Bondino et al. / Surface Science 461 (2000) 31–44
J.W. Niemantsverdriet, J. Vac. Sci. Technol. A 15 (1997)
1642.
[11] M.G. Ramsey, D. Steinmuller, F.P. Netzer, S. Kostlmeier,
¨
¨
J. Lauber, N. Rosch, Surf. Sci. 307–309 (1994) 82.
¨
[3] D.R. Mullins, L. Kundakovic, S.H. Overbury, to be
published.
[4] F.P. Netzer, M.G. Ramsey, Crit. Rev. Solid State Mater.
Sci. 17 (1992) 397.
[5] F. Solymosi, L. Bugyi, Surf. Sci. 147 (1984) 685.
[6] S.Y. Hwang, A.C.F. Kong, L.D. Schmidt, J. Phys. Chem.
93 (1989) 8327.
[7] A. Baraldi, S. Lizzit, M.G. Ramsey, F.P. Netzer, Surf. Sci.
416 (1998) 214.
[12] N.A. Booth, R. Davis, D.P. Woodruff, D. Chrysostomou,
T. McCabe, D.R. Lloyd, O. Schaff, V. Fernandez, S. Bau,
K.-M. Schindler, R. Lindsay, J.T. Hoeft, R. Terborg, P.
Baumgartel, A.M. Bradshaw, Surf. Sci. 416 (1998) 448.
¨
[13] R.I.R. Blyth, I. Kardinal, F.P. Netzer, M.G. Ramsey, D.
Chrysostomou, D.R. Lloyd, Surf. Sci. 415 (1998) 227.
[14] I. Kardinal, F.P. Netzer, M.G. Ramsey, Surf. Sci. 376
(1997) 229.
[15] H. Yang, J.L. Whitten, J. Chem. Phys. 107 (1997) 8518.
[16] M. Kiskinova, S. Lizzit, G. Comelli, G. Paolucci, R. Rosei,
Appl. Surf. Sci. 64 (1993) 185.
[8] M.G. Ramsey, G. Rosina, F.P. Netzer, H.B. Saalfeld, D.R.
Lloyd, Surf. Sci. 217 (1989) 140.
[9] M.G. Ramsey, D. Steinmuller, F.P. Netzer, M. Neuber, L.
¨
[17] S. Lizzit, G. Comelli, Ph. Hofmann, G. Paolucci, M. Kiski-
nova, R. Rosei, Surf. Sci. 276 (1992) 144.
Ackermann, J. Lauber, N. Rosch, Surf. Sci. 260 (1992)
¨
163.
[18] R.M. van Hardevelt, R.A. van Santen, J.W. Niemantsver-
driet, J. Phys. Chem. 101 (1997) 7901.
[10] M.G. Ramsey, G. Rosina, F.P. Netzer, H.B. Saalfeld, D.R.
Lloyd, Surf. Sci. 218 (1989) 317.