Structural and compositional analysis of two ordered near-surface alloys produced by adsorption and thermal decomposition of Mo(CO)6 on Ni{ 1 1 1 }

Article abstract of DOI:10.1016/S0039-6028(02)02452-4

The adsorption and thermal decomposition of Mo(CO)₆ is reported on Ni{111}. At room temperature, Mo(CO)₆ has an extremely low sticking probability. At 450-550 K, Mo(CO)₆ adsorbs dissociatively via loss of one or more CO ligands to produce a surface intermediate stable up to about 675 K. Desorption of the remaining CO ligands leaves an essentially contaminant free NiMo surface. Two distinct NiMo near-surface alloys were synthesised by this route. These bimetallic surfaces were characterised by a combination of STM, LEED, AES, MEIS and TPD. In each case, the alloys were Ni rich and, in each case, are found to be very close to the structure of fcc nickel. The first alloy gives a (3 × 3) LEED pattern and has the composition Ni₈Mo. The second structure gives a (5 × 5) LEED pattern and has the composition Ni₂₄Mo.

Full text of DOI:10.1016/S0039-6028(02)02452-4

Surface Science 523 (2003) 12–20
www.elsevier.com/locate/susc
Structural and compositional analysis of two ordered
near-surface alloys produced by adsorption and
thermal decomposition of Mo(CO)6 on Ni{111}
a
T.E. Jones , T.C.Q. Noakes , P. Bailey , C.J. Baddeley
b
b
a,
*
a
School of Chemistry, University of St Andrews, North Haugh, St Andrews, Fife KY16 9ST, UK
b
MEIS facility, CLRC Daresbury Laboratory, Daresbury, Cheshire, WA4 4AD, UK
Received 5 September 2002; accepted for publication 17 October 2002
Abstract
The adsorption and thermal decomposition of Mo(CO)6 is reported on Ni{1 1 1}. At room temperature, Mo(CO)6
has an extremely low sticking probability. At 450–550 K, Mo(CO)6 adsorbs dissociatively via loss of one or more CO
ligands to produce a surface intermediate stable up to about 675 K. Desorption of the remaining CO ligands leaves an
essentially contaminant free NiMo surface. Two distinct NiMo near-surface alloys were synthesised by this route. These
bimetallic surfaces were characterised by a combination of STM, LEED, AES, MEIS and TPD. In each case, the alloys
were Ni rich and, in each case, are found to be very close to the structure of fcc nickel. The first alloy gives a (3 ꢀ 3)
LEED pattern and has the composition Ni8Mo. The second structure gives a (5 ꢀ 5) LEED pattern and has the
composition Ni24Mo.
Ó 2002 Elsevier Science B.V. All rights reserved.
Keywords: Scanning tunneling microscopy; Low energy electron diffraction (LEED); Medium energy ion scattering (MEIS); Catalysis;
Surface segregation; Nickel; Molybdenum; Alloys
1. Introduction
as model catalysts probing hydrodesulphurisation
processes. Several groups have investigated the
growth of Ni onto Mo{1 1 0} [5–10], but to our
knowledge, such an approach has never resulted in
the formation of a surface alloy. In the reverse
system, Mo on Ni{1 1 1}, intermixing has been
reported by Feinstein and Blanc [11]. In this case,
the surface alloy was created by clamping a
Ni{1 1 1} crystal into a cylinder of high purity Mo
foil with the Ni{1 1 1} face protruding by 1 mm.
The crystal was then heated radiatively to in-
duce Mo diffusion to the Ni{1 1 1} surface [11]. In
this early low energy electron diffraction (LEED)
study, two ordered alloy structures were reported
The Mo–Ni bimetallic system has attracted
much recent research as a result of its interesting
magnetic [1] and catalytic properties (e.g. hydro-
desulphurisation [2,3] and hydrodenitrogenation
[4]). In this paper, we are interested in forming
bimetallic NiMo surface alloys of varying com-
position in order to eventually use these surfaces
* Corresponding author. Tel.: +44-1334-467236; fax: +44-
1334-467285.
E-mail address: cjb14@st-and.ac.uk (C.J. Baddeley).
0039-6028/02/$ - see front matter Ó 2002 Elsevier Science B.V. All rights reserved.
PII: S0039-6028(02)02452-4

T.E. Jones et al. / Surface Science 523 (2003) 12–20
13
giving rise to (5 ꢀ 5) and (4 ꢀ 4) LEED patterns
corresponding to surface compositions of 4% and
6.7% Mo on the assumption that the overlayer unit
cell contained one Mo atom per unit cell [11]. The
rationale that was used to derive this composition
was that the solubility of Mo in Ni at 650 °C was
less than 12% and at the time of writing there was
only limited evidence that anomalously high sol-
ubilities could be observed in the surface region
[11]. No consideration was given to the possibility
that Mo could form an overlayer on Ni giving rise
to a (5 ꢀ 5) LEED pattern.
tory, UK [18]. This facility comprises an ion
source and accelerator capable of producing Hþ or
Heþ ions at energies up to 400 keV (in our case 100
keV Heþ ions), a beamline for transport of the
ions with a well-defined energy (<0.1%) and low
angular divergence (<0.1°) and a multi-chamber
UHV end station. The end station consists of a
scattering chamber, a sample preparation and
characterisation chamber, a sample transfer sys-
tem and a fast entry load lock.
The scattering chamber contains a precision six-
axis goniometer to allow accurate alignment with
the beam and a toroidal electrostatic energy ana-
lyser with position sensitive detector to measure
the scattered ion intensities as functions of scat-
tering angle and ion energy. The detector system
produces two dimensional intensity maps of the
ion intensity over a 1.8% range of pass energy and
a scattering angle range of 24° with a resolution of
0.3% and 0.3° respectively. Full two-dimensional
data sets are accumulated by taking a series of
ÔtilesÕ which cover the required angle/energy range
and joining them together electronically to pro-
duce a single scattered ion intensity map. Whilst
two-dimensional data sets provide a complete
picture of the scattering behaviour it is normal to
process the data by integrating over a range of
angles or energies to produce one dimensional
plots. Angle versus scattered ion intensity plots
give structural information that can be used to
calculate parameters such as surface layer relax-
ation. Energy spectra can be used to obtain quan-
titative compositional information and a number
of these plots are presented in this paper. The
preparation chamber is used for sample cleaning
and characterisation (LEED/AES).
In this paper, we use scanning tunnelling mi-
croscopy (STM); temperature programmed desorp-
tion (TPD); LEED; Auger electron spectroscopy
(AES) and medium energy ion scattering (MEIS)
to investigate the growth of Mo on Ni{1 1 1} by
the adsorption and thermal decomposition of
Mo(CO)6. This approach has been used previously
to create Mo nanoparticles on Si{1 1 1}-(7 ꢀ 7)
[12]; Au{1 1 1} [13]; TiO2{1 1 0} [14]; Ru{0 0 0 1}
[15] and Ag{1 1 1} [16]. In addition, MnO nano-
particles have been grown on Pt{1 1 1} by thermal
decomposition of Mn2(CO)10 [17]. We show that
Mo(CO)6 undergoes dissociative adsorption at 450
K to produce a stable surface intermediate (prob-
ably Mo(CO)5). Above 675 K, further decompo-
sition occurs to produce predominantly Mo(ads)
and CO(g). We report the formation of two or-
dered near-surface alloys (one of which has not
been previously reported) on thermal processing of
the Mo containing adlayers.
2. Experimental
Three ultrahigh vacuum (UHV) systems were
used in this study. The first chamber has a base
pressure of ꢁ3 ꢀ 10ꢂ10 mbar and was built by PSP
Vacuum Technology Ltd. This chamber has fa-
cilities for argon ion sputtering; a quadrupole mass
spectrometer for TPD experiments and a four-grid
RFA for LEED and AES experiments. The second
chamber is an Omicron variable temperature-STM
(base pressure 1 ꢀ 10ꢂ10 mbar) which, in addition
to STM, has facilities for sample cleaning and
LEED/AES experiments. The third UHV system is
the CLRC MEIS facility at Daresbury Labora-
In each case, the Ni{1 1 1} sample was cleaned
by cycles of sputtering (Arþ, 1.5 keV) and an-
nealing to 1000 K until a sharp (1 ꢀ 1) LEED
pattern was obtained and no impurities were ob-
served by AES. Mo(CO)6 is a solid with a signifi-
cant vapour pressure at room temperature. It was
contained in a glass tube on a gas-line pumped by
a turbomolecular pump. The vapour was dosed via
a leak valve. In preliminary experiments, the va-
pour was monitored by a mass spectrometer which
revealed peaks at 28 (CO); 264 (Mo(CO)6); 236
(Mo(CO)5); 208 (Mo(CO)4) etc. However, in TPD

14
T.E. Jones et al. / Surface Science 523 (2003) 12–20
experiments, though each of these masses were
monitored, desorption was only ever observed for
mass 28. An approximately linear heating rate of
ꢁ2 K sꢂ1 was used in all TPD experiments. The
deposition of Mo was monitored by AES follow-
ing the attenuation of the Ni 848 eV Auger tran-
sitions and the growth of the Mo 180 and 234 eV
transitions.
3. Results and discussion
3.1. Temperature programmed desorption
The sticking probability of Mo(CO)6 on
Ni{1 1 1} appears to be extremely low at 300 K.
However, when the sample temperature was raised
to 450 K and above, significant evidence of ad-
sorption was detected. Fig. 1a and b show the TPD
spectra following increasing doses of Mo(CO)6 at
450 and 550 K respectively. In each case, CO de-
sorption is observed to occur in a broad peak
whose Tmax shifts progressively to higher temper-
atures with increasing coverage. Following a dose
of 30 L the peak is observed at around 725 K and
increases to 850 K for the 960 L dose at 450 K. It
can be seen that the onset of the desorption feature
occurs at approximately the same temperature for
each spectrum––namely ꢁ675 K. Similar behav-
iour was observed for Mo(CO)6 on Au{1 1 1} [13]
where the higher sticking probability at higher
sample temperature was ascribed to the existence
of a decomposition channel for Mo(CO)6. T he
existence of only one desorption feature whose
onset is at a considerably higher temperature than
the adsorption temperature implies that the initial
decomposition product is relatively thermally sta-
ble (probably Mo(CO)5) and that once decompo-
sition of this surface species begins, the loss of all
remaining CO molecules occurs without the for-
mation of any stable Mo(CO)n species (n ¼ 1–4).
Fig. 1. TPD spectra showing the desorption of mass 28 (CO)
from Ni{1 1 1} as a function of Mo(CO)6 dose at (a) 450 K and
(b) 550 K.
Mo(CO)6 at a sample temperature of 400 K. Then
we annealed the surface to 675 K for 5 min (i.e. to
just above the leading edge of the CO desorption).
We then repeated this procedure until a total dose
of 1200 L Mo(CO)6 had been reached. AES of the
surface showed the presence of significant amounts
of C and O as well as Mo and Ni. Annealing to 725
K resulted in the removal of all detectable oxygen.
Heating to 775 K led to the amount of carbon
remaining on the surface being below 0.1 ML.
LEED indicated that the as prepared surface
showed a (3 ꢀ 3) pattern with broad beams indi-
cating that the ordered domains are relatively
small. Annealing to 825 K caused a sharpening of
the fractional order beams. Subsequent annealing
3.2. Low energy electron diffraction/Auger electron
spectroscopy
On the basis of our TPD experiments, we chose
the following dosing procedure to build up Mo
overlayers on Ni{1 1 1}. First we dosed 200 L

T.E. Jones et al. / Surface Science 523 (2003) 12–20
15
to 875–925 K resulted in the disappearance of the
(3 ꢀ 3) pattern to be replaced by a beam in the (4/
5,0), (0,4/5) and all equivalent positions in the
diffraction pattern indicative of a (5 ꢀ 5) type pat-
tern. A similar (5 ꢀ 5) structure has been reported
for Mo on Ni{1 1 1} by Feinstein and Blanc [11].
The ordered alloy formed by this procedure was
assumed to consist of one Mo atom per unit cell
giving a surface composition of 4% Mo.
probed. Following the 650 K annealing treatment
(Fig. 3a), all Mo is located in a relatively intense
and narrow peak indicating that Mo is present in a
single surface layer. By contrast, annealing to
815 K results in a much broader Mo related fea-
ture indicating the presence of subsurface Mo (Fig.
3b). In the case of the 885 K annealed surface, the
Mo features are much less intense and much
broader. The relative intensities of the Mo peak
and the Ni peak give an indication of the amount
of Mo present in the near-surface region. The
width of the Mo rich layer can be used to infer
the distribution of Mo as a function of depth in the
sample by consideration of the stopping power
and path length. In the sample annealed to 650 K,
ꢃ1 ML of Mo is present and is located predomi-
nantly in the surface layer. Following the anneal to
815 K; Mo is present in significant quantities in the
top five layers with an average composition of 10%
Mo in each layer. By 885 K (Fig. 3c), the width of
the Mo feature corresponds to ꢃ15 layers with an
average composition of 2% Mo in each layer. The
integrated sum of Mo clearly decreases with in-
creasing annealing temperature. This can either be
caused by dissolution into the bulk or desorption.
As no Mo related desorption products were ob-
served, we must conclude that the former process
is responsible for the observed decrease. The sol-
ubility of Mo in Ni at 650 °C is about 12% so this
is certainly conceivable.
3.3. Scanning tunnelling microscopy
Fig. 2a shows an STM image following four
cycles of Mo(CO)6 dosing and annealing to 650 K.
The Mo containing islands are seen to have no
sharp boundaries and gaps are observed between
the relatively amorphous islands. The line profile
presented in Fig. 2a shows that the difference in
height between the islands is ꢁ0.2 nm––a value
typical for the height of a monolayer island. Fig.
2b shows that annealing to 750 K modifies the
adlayer, giving a much flatter appearance. This
sample preparation treatment corresponds to the
formation of the (3 ꢀ 3) LEED structure. In Fig.
2c, an atomically resolved image is shown of the
surface created by annealing the Mo overlayer to
925 K. The image shows features whose spacing
is, within experimental error, exactly that of
Ni{1 1 1}.
3.4. Medium energy ion scattering
We performed a more detailed layer by layer
compositional analysis of the surface for systems
(ii) and (iii) (analysis of the composition of the
surface in system (i) is made difficult by the pos-
sibility that the overlayer has a different struc-
ture to the underlying Ni{1 1 1}). In this case, we
adopted the approach first used by Deckers et al.
[19] in the analysis of Pt50Ni50{1 1 1} and then
modified by Baddeley et al. in the analysis of
Pd50Cu50{1 1 0} [20–22]. Using the incident and
exit beam directions shown in Fig. 4, we selectively
illuminated the top, top two and top three layers of
the surface. For an fcc {1 1 1} surface there is likely
to be a much smaller contribution from ions
scattered from deeper lying layers using these
‘‘double alignment’’ geometries than was observed
for an fcc {1 1 0} surface [20–22]. Using the VEGAS
programme [23] we estimated the contribution
MEIS experiments were carried out to investi-
gate (i) the Mo overlayer; (ii) the (3 ꢀ 3) structure
and (iii) the (5 ꢀ 5) structure. These surfaces were
created by effectively depositing 1.0 ML Mo and
annealing to 650, 815 and 885 K. Fig. 3 show plots
of scattered ion intensity versus ion energy for
each surface preparation. In each experiment the
beam was first aligned along a bulk crystallo-
ꢀꢀ
graphic direction (in this case the [110] incident
geometry) then deliberately rotated by 7° to achieve
a ‘‘random’’ incident geometry. In this case, we
take advantage of the fact that ions scattered from
a given element lose energy as a function of the
depth of the target atom beneath the surface. This
enables depth profiling and hence allows the dis-
tribution of Mo in the near-surface region to be

16
T.E. Jones et al. / Surface Science 523 (2003) 12–20
ꢁ
Fig. 2. (a) STM image (1000 ꢀ 1000 A) showing the growth of Mo overlayers by four cycles of 1800 L Mo(CO)6 doses and annealing
ꢁ
to 650 K. The inset shows the height variation along the arrow indicated on the STM image. (b) STM image (1300 ꢀ 1300 A) showing
the effect of annealing Mo overlayers to 720 K (inset shows the height variation along the arrow indicated on the STM image) and (c)
ꢁ
STM image (55 ꢀ 55 A) showing the effect of annealing Mo overlayers to 925 K.

T.E. Jones et al. / Surface Science 523 (2003) 12–20
17
(top layer þ 13% second layer); in the two-layer
geometry 2.04 layers are illuminated (top
layer þ second layer þ 4% third layer) and in the
three-layer geometry the top three layers are all
illuminated with no contribution from deeper lying
layers. By integrating the number of scattered ions
at a particular scattering angle (e.g. in the one-
layer geometry the scattering angle in question is
90°) we obtain a plot of ion intensity versus ion
energy. Then, we measured the peak areas of the
Ni and Mo related features in this spectrum. Once
the peak areas are normalised to take into account
the ion scattering cross-section which varies as the
square of the atomic number, we have a direct
measure of the mole fraction of Ni and Mo in the
surface layer. By following the procedure for the
two-layer geometry, we can obtain a value for
the mole fraction of Ni and Mo in the top two
layers. By application of simultaneous equations
of the type reported in Refs. [20–22], we can cal-
culate the composition of the second layer and
similarly the third layer.
Fig. 3. MEIS data showing ion intensity versus ion energy for
ꢀꢀ
samples aligned along [110] incidence direction and then rotated
7° off this direction to achieve a ‘‘random’’ incidence direction.
These surfaces were created by effectively depositing 0.5–1.0 ML
Mo and annealing to (a) 650 K; (b) 815 K and (c) 885 K.
In system (ii), we found that the surface com-
position was 11:8 Æ 2:0 at.% Mo, the second layer
7:4 Æ 2:0 at.% Mo and the third layer 9:8 Æ 2:0
at.% Mo. The relatively consistent compositions of
the top three layers point to the formation of a
near-surface alloy of composition likely to be
Mo1Ni8. This is due to the fact that the (3 ꢀ 3)
periodicity of the unit cell implies that, if one Mo
atom is present per unit cell, then the coverage of
Mo is 1/9 ML. The accuracy of the MEIS data is
such that we can rule out unequivocally the possi-
bility of there being multiple Mo atoms per unit
cell. The error bars are chosen on the basis of in-
accuracies in our estimation of the layer illumina-
tion. We found in the Pd50Cu50{1 1 0} work [22]
that if the composition of the top and the second
layer was approximately equal, changes in the
calculated layer composition due to changes in the
estimated layer illumination were very small. By
contrast, in the case where the surface is heavily
enriched in one element and the second layer
heavily depleted in that element, the absolute value
of the calculated composition is strongly dependent
on the estimated layer illumination. This accounts
for our relatively high degree of confidence in the
layer compositions in the present case.
Fig. 4. Schematic diagram showing the incident ion trajectories
and analyser positions for the one-, two- and three-layer geom-
etries. The diagram shows a cross-section perpendicular to the
ꢀꢀ
ꢀꢀ
ꢀꢀꢀ
[112] azimuthal direction of the surface. h1i in ¼ ½110ꢄ;
ꢀꢀ
h1i out ¼ ½001ꢄ; h2i in ¼ ½332ꢄ; h2i out ¼ ½112ꢄ; h3i in ¼ ½111ꢄ;
h3i out ¼ ½118ꢄ.
of each layer to the scattered ion signal. For our
model, we chose a Ni{1 1 1} lattice with a surface
layer relaxation of 5%, and an isotropic enhance-
ment of the vibrations of surface atoms compared
p
to the bulk of 2 which is a value typically used in
MEIS calculations. In the Pd50Cu50{1 1 0} case, a
value of 2 for the surface/bulk vibrational en-
hancement was found to achieve the best fit be-
tween experimental data and simulation. Brown
et al. [24] have since suggested that the additional
illumination of sub-surface atoms may be caused
by static displacement of atoms in the bimetallic
surface due to differences in atomic radii. The
output of the VEGAS simulation indicates that in
the one-layer geometry 1.13 layers are illuminated

18
T.E. Jones et al. / Surface Science 523 (2003) 12–20
In system (iii), the surface composition was
found to be 3:0 Æ 2:0 at.% Mo, the second layer
composition was 0:9 Æ 2:0 at.% Mo and the third
layer composition lower still. Using the same ar-
guments as were applied for system (ii) (i.e. to fit in
with the (5 ꢀ 5) LEED pattern observed in situ);
an alloy of composition Mo1Ni24 has been formed.
Once more, MEIS enables us to rule out the pos-
sibility of multiple occupancy of Mo atoms in the
unit cell. In the case of system (iii); it appears that
the only layer with stoichiometric Mo content is
the surface layer, with continuously decreasing Mo
content towards the measured layer thickness of 15
ML. This may reflect the formation of a surface
layer localised alloy or alternatively that our an-
nealing procedure was just above that required for
the onset of breakdown of the alloy.
Fig. 5a shows plots of ion intensity versus
scattering angle for each of the three prepared
surfaces. In this case, the beam was aligned to be
ꢀꢀꢀ
incident along the [111] direction of the Ni{1 1 1}
sample. The presented data were derived by inte-
grating the scattered ion intensity over the ion
energy region covered by the Mo related features
as a function of scattering angle. Also shown are
the corresponding data derived from the Ni peak
from clean Ni{1 1 1}. In each case, the ion energy
range covered by this procedure corresponds to
approximately the top six layers of the surface. It is
clear that the blocking curve for the Mo overlayer
(system (i)) is featureless. This is to be expected as
the overlayer contains only about 1 ML Mo and
hence is likely to exhibit little or no blocking fea-
tures. It confirms, however, that Mo is located at
the surface and that negligible intermixing has
occurred. In the case of systems (ii) and (iii), the
blocking curves closely resemble those of the clean
Ni{1 1 1} surface. The data of Figs. 3 and 5 indi-
cate that, in both cases, (a) significant intermixing
has occurred and (b) the bimetallic system has, to a
first approximation, adopted the Ni{1 1 1} struc-
ture. This aspect of the surface structure will be
considered below. In Fig. 5b, the analogous data
to those presented in Fig. 5a are shown for the Ni
derived surface features. In each case, the spectra
are similar to those of the Ni{1 1 1} clean surface.
In the 650 K annealed sample, the presence of a
monolayer of Mo would be anticipated to cause
Fig. 5. Ion intensity versus scattering angle data for samples
ꢀꢀꢀ
aligned along the [111] incidence direction as a function of
annealing tempeature. These data were acquired by summing
the ion intensity in the region of the (a) Mo related and (b) Ni
related scattering features. Also shown in each figure, for
comparison, is the equivalent data from the Ni related scat-
tering features of the clean Ni{1 1 1} surface.
additional blocking features in the surface Ni data.
We observed no such extra features. This probably
reflects the amorphous nature of the Mo over-
layer––in particular that the Mo overlayer is prob-
ably not metallic and consists of Mo(CO)n type
species presumably in poorly defined adsorption
sites. Any blocking of the Ni signal by such an
adlayer is not specific to certain scattering angles.
ꢁ
The differences in metallic radii of Mo (2.72 A)
ꢁ
and Ni (2.49 A) would lead one to predict that

T.E. Jones et al. / Surface Science 523 (2003) 12–20
19
changes in the lattice parameter would be observed
on going from Ni to a NiMo alloy which may lead
to the incommensurate growth of the alloy on the
pure metal. This has been observed, for example,
in the Pt{1 1 1}/Pt5Ce [25] system, to give rise to
catalyst than more traditional methods such as co-
precipitation and wet-impregnation. Further ex-
periments will be undertaken to investigate these
possibilities.
ꢂ
the appearance of Moire structures. Such incom-
4. Conclusions
mensurate growth would result in profound
changes to the surface blocking features of both
Mo and Ni. We see no evidence for this type of
growth in the present case either in the MEIS or
STM data. Since the two alloys are each strongly
Ni rich, any change in lattice parameter of analo-
gous bulk alloys may be anticipated to be small.
However, it may be possible to detect changes in
interlayer spacing as shifts in the positions of
blocking features in the MEIS data derived from
the Mo and Ni surface peaks. Using the data
presented in Fig. 5, we found that, in the 4% Mo
alloy, the measured shift was 0.15°, while in the
11% Mo alloy, the measured shift (averaged over
all blocking dips) was 0.45°. It should be noted
that the shifts are barely significant in comparison
to the resolution of the MEIS instrument (ꢁ0.3°).
Surprisingly, each shift corresponds to a contrac-
tion in the interlayer spacing compared to
Ni{1 1 1}. However, similar behaviour has been
reported for amorphous NiMo alloys by Cocco
et al. [26] prepared by milling pure metal powders.
The measured shift of 0.45° in blocking features
relative to the bulk corresponds to a change in
(i) The sticking probability of Mo(CO)6 on
Ni{1 1 1} is negligible at 300 K, but increases
significantly by 450–550 K.
(ii) Cycles of Mo(CO)6 at 375 K and annealing to
650–700 K resulted in the formation of a Mo
overlayer.
(iii) Annealing of the overlayer to 810 K resulted
in the formation of a (3 ꢀ 3) near-surface
alloy. Annealing further to 880 K resulted in
the formation of a (5 ꢀ 5) near-surface alloy
with extensive dissolution of Mo into the bulk
of the Ni{1 1 1} sample. In each case, the sur-
faces were essentially free of adsorbed carbon
and oxygen.
(iv) Both near-surface alloys appear to consist of
Mo incorporated directly into the Ni lattice.
MEIS compositional analysis confirms that
in each case, one Mo atom is present per ad-
layer unit cell giving compositions of 11.1%
Mo for the (3 ꢀ 3) alloy and 4% Mo for the
(5 ꢀ 5) alloy.
ꢁ
interlayer spacing of <0.05 A.
Acknowledgements
It is interesting that we did not observe the
(4 ꢀ 4) LEED pattern reported by Feinstein and
Blanc [11]. This may reflect the radically different
preparation procedures employed in the two stud-
ies. The change in annealing temperature required
to convert the (3 ꢀ 3) structure into the (5 ꢀ 5)
structure is only 70 K. Therefore, future experi-
ments could concentrate on this temperature win-
dow in an attempt to elucidate whether the (4 ꢀ 4)
structure could also be observed using our prepa-
ration procedures.
TEJ acknowledges the Engineering and Physical
Sciences Research Council for the award of a
Ph.D. studentship. We gratefully acknowledge the
EPSRC for the award of 10 days MEIS facility
time at CLRC Daresbury Laboratory (MAP 6-06).
We thank Professor Neville Richardson for access
to his STM chamber and Dr. Qiao Chen for help
with setting up the STM experiments.
We hope that this method for creating bimetallic
surfaces can be extended to the synthesis of bi-
metallic catalysts. Thermal decomposition of mo-
lybdenum hexacarbonyl onto Ni nanoparticle
surfaces may be expected to produce rather more
uniform bimetallic surfaces throughout a given
References
[1] J.W. Cable, M.R. Khan, G.P. Felcher, J.K. Schuller, Phys.
Rev. B 34 (1986) 1643.
[2] A. Calafat, J. Laine, A. Lopez-Agudo, J.M. Palacios, J.
Catalysis 162 (1996) 20.

20
T.E. Jones et al. / Surface Science 523 (2003) 12–20
[3] J.A. Rodriguez, J. Dvorak, T. Jirsak, S.Y. Li, J. Hrbek,
A.T. Capitano, A.M. Gabelnick, J.L. Gland, J. Phys.
Chem. B. 103 (1999) 8310.
[15] H.H. Huang, C.S. Sreekanth, C.S. Seet, X. Jiang, G.Q. Xu,
Surf. Sci. 365 (1996) 769.
[16] S.K. So, W. Ho, J. Chem. Phys. 95 (1991) 656.
[17] G.A. Rizzi, R. Zanoni, S. Di Siro, L. Perriello, G.
Granozzi, Surf. Sci. 462 (2000) 187.
[4] W.R.A.M. Robinson, J.N.M. van Gestel, T.I. Koranyi, S.
Eijsbouts, A.M. van der Kraan, J.A.R. van Veen, V.H.J. de
Beer, J. Catalysis 161 (1996) 539.
[18] P. Bailey, T.C.Q. Noakes, D.P. Woodruff, Surf. Sci. 426
(1999) 358.
[5] H. Tsunematu, Y. Gotoh, Jpn. J. Appl. Phys. 38 (1999) 2899.
[6] H. Tsunematu, H. Aita, M. Tanaka, A. Suzuki, S. Sato, Y.
Gotoh, J. Electron Spectr. Related Phenomena 103 (1999)
281.
[19] S. Deckers, F.H.P.M. Habraken, W.F. van der Weg,
A.W.D. van der Gon, B. Pluis, J.F. van der Veen, R.
Baudoing, Phys. Rev. B 42 (1990) 3253.
[7] J.H. van der Merwe, E. Bauer, D.L. Tonsing, P.M. Stoop,
Phys. Rev. B. 49 (1994) 2137.
[20] C.J. Baddeley, L.H. Bloxham, S.C. Laroze, R. Raval,
T.C.Q. Noakes, P. Bailey, Surf. Sci. 435 (1999) 827.
[21] T.C.Q. Noakes, P. Bailey, S.C. Laroze, R. Raval, C.J.
Baddeley, Surf. Int. Anal. 30 (2000) 81.
[8] J.W. He, W.K. Kuhn, D.W. Goodman, Surf. Sci. 292
(1993) 248.
[9] R.A. Campbell, J.A. Rodriguez, D.W. Goodman, Surf.
Sci. 256 (1991) 272.
[22] C.J. Baddeley, L.H. Bloxham, S.C. Laroze, R. Raval,
T.C.Q. Noakes, P. Bailey, J. Phys. Chem. B 105 (2001)
2766.
[10] J.W. He, W.K. Kuhn, D.W. Goodman, J. Am. Chem. Soc.
113 (1991) 6416.
[23] J.W.M. Frenken, R.M. Tromp, J.F. van der Veen, Nucl.
Instr. Meth. B 17 (1986) 334.
[11] L.G. Feinstein, E. Blanc, Surf. Sci. 18 (1969) 350.
[12] U.R. Schoffel, H. Rauscher, R.J. Behm, J. Appl. Phys. 91
(2002) 285.
[24] D. Brown, P.D. Quinn, D.P. Woodruff, T.C.Q. Noakes, P.
Bailey, Surf. Sci. 497 (2002) 1.
[13] J.A. Rodriguez, J. Dvorak, T. Jirsak, J. Hrbek, Surf. Sci.
490 (2001) 315.
[25] C.J. Baddeley, A.W. Stephenson, C. Hardacre, M.S.
Tikhov, R.M. Lambert, Phys. Rev. B 56 (1997) 12589.
[26] G. Cocco, S. Enzo, N.T. Barrett, K.J. Roberts, Phys. Rev.
B 45 (1992) 7066.
[14] J. Evans, B.E. Hayden, G. Lu, J. Chem. Soc. Faraday
Trans. 92 (1996) 4733.