A medium-energy ion scattering investigation of the structure and surface vibrations of two-dimensional YSi2 grown on Si(1 1 1)

Article abstract of DOI:10.1016/j.susc.2005.08.032

Medium-energy ion scattering has been used to determine the atomic structure of two-dimensional yttrium silicide on silicon (1 1 1). A full quantitative analysis of the atomic positions of the Si atoms in the top bilayer yields a model similar to that previously suggested in the literature with a Si1-Si2 vertical spacing of 0.80 ± 0.03 A?, but with the Si bilayer relaxed slightly further away from the Y layer (Si2-Y vertical spacing of 1.89 ± 0.02 A?). Observing the effects of the top bilayer vibrations yields a model with significant enhancements.

Full text of DOI:10.1016/j.susc.2005.08.032

Surface Science 598 (2005) 120–127
www.elsevier.com/locate/susc
A medium-energy ion scattering investigation
of the structure and surface vibrations of two-dimensional
YSi2 grown on Si(111)
a
a
b
b
a,
*
T.J. Wood , C. Bonet , T.C.Q. Noakes , P. Bailey , S.P. Tear
a
Department of Physics, University of York, Heslington, York YO10 5DD, UK
b
CCLRC Daresbury Laboratory, Daresbury, Warrington WA4 4AD, UK
Received 1 July 2005; accepted for publication 10 August 2005
Available online 21 October 2005
Abstract
Medium-energy ion scattering has been used to determine the atomic structure of two-dimensional yttrium silicide
on silicon (111). A full quantitative analysis of the atomic positions of the Si atoms in the top bilayer yields a model
˚
similar to that previously suggested in the literature with a Si1–Si2 vertical spacing of 0.80 0.03 A, but with the Si
˚
bilayer relaxed slightly further away from the Y layer (Si2–Y vertical spacing of 1.89 0.02 A). Observing the effects
of the top bilayer vibrations yields a model with significant enhancements.
ꢀ 2005 Elsevier B.V. All rights reserved.
Keywords: Medium energy ion scattering (MEIS); Ion–solid interactions, scattering, channeling; Surface relaxation and reconstruc-
tion; Surface structure, morphology, roughness, and topography; Metal–semiconductor interfaces; Silicon; Silicides; Yttrium
1. Introduction
to a complete understanding of the electronic
properties. The rare-earth (RE) silicide–silicon
interface has attracted much interest recently as
potential technological applications have been
identified [1]. When these silicides are formed on
the surface of n-type Si(111), it has been found
that they exhibit unusually low Schottky barrier
heights (0.3–0.4 eV) [2,3] and novel interface
reconstructions. A single monolayer (ML) deposit
of a ÔheavyÕ trivalent RE (e.g. Gd, Er, Dy and Ho)
onto clean Si(111)-7 · 7, followed by annealing
The properties of the metal–semiconductor
interface are of considerable interest due to its
technological and fundamental importance. In
particular, knowledge of the atomic arrangement
and quality of growth at the interface is essential
*
Corresponding author. Tel.: +44 1904 432202; fax: +44
0904 432335/432214.
E-mail address: spt1@york.ac.uk (S.P. Tear).
0039-6028/$ - see front matter ꢀ 2005 Elsevier B.V. All rights reserved.
doi:10.1016/j.susc.2005.08.032

T.J. Wood et al. / Surface Science 598 (2005) 120–127
121
results in the Ôtwo-dimensionalÕ (2D) RE silicide
(Fig. 1) [4–12]. This consists of a single atomic
layer of RE in T4 sites on the substrate Si, above
which lies a buckled bilayer of Si, similar to that
of the bulk, but with its buckling direction
reversed (type B). This structure has a p(1 · 1)
periodicity. Such reconstructions have been stud-
ied extensively by a variety of techniques, such as
Auger and photoelectron diffraction [5,7], surface
X-ray diffraction [8], low-energy electron diffrac-
tion (LEED) [9,10] and medium-energy ion scat-
tering (MEIS) [11–13].
LEED data and calculations, which for large val-
ues of the polar angle of the incident electron beam
showed a preference for the higher value of 0.79 A.
˚
This conclusion has subsequently been shown to
be consistent with the results of density functional
theory (DFT) calculations and angle-resolved pho-
toemission experiments [17].
The LEED work of Rogero et al. found that by
significantly enhancing the vibrations of the top Si
bilayer (by a factor of three in the case of the Si1
atoms), a much better fit could be obtained for
the LEED I–V curves. This is the first time such
large vibrations have been reported in the litera-
ture for the 2D RE silicide system. The MEIS
studies of Spence et al. [11,12] applied a smaller
Yttrium, whilst not possessing the 4f electrons
of the other rare earth elements, is considered to
be part of this group since it exhibits the same
chemical characteristics. Previous studies of Y on
Si(111) have shown that it forms reconstructions
very similar to those of the ÔheavyÕ RE silicides
[14–18]. Through the use of quantitative LEED
I–V and ab initio total energy calculations, Rogero
et al. [16] found it was not possible to distinguish
between two distinct buckling distances (0.33 and
enhancement to the bulk rms vibrational ampli-
pffiffi
tude by using the factor of 2 commonly associ-
ated with surface atoms. These significant
enhancements could explain why the normal inci-
dence LEED I–V could not distinguish between
the two models which were structurally very differ-
ent. Significant enhancements have also been
reported for the 3D Y silicide system [19].
˚
0.79 A) for the top bilayer in a 2D Y silicide using
normal incidence LEED I–V data alone, due to a
variety of local minima in the Pendry R-factor.
This was resolved through the use of off-normal
In contrast, MEIS yields real-space structural
information and the technique can be applied in
such a way as to focus on particular aspects of
Si1
Si2
RE
Si3
Si4
Si5
Si6
Side view
[112]
Top
view
RE
Substrate Si
Overlayer Si
Fig. 1. Structural model for the 2D RE silicides.

122
T.J. Wood et al. / Surface Science 598 (2005) 120–127
the surface structure, making it ideal for studying
the present class of systems. Due to the large mass
separation of the two elements, it is possible to
extract and analyse the RE scattering independent
of effects due to all the Si layers below [11–13,20]—
techniques such as LEED require the relaxations
of the Si3–Si6 layers to be considered in any struc-
tural determination, increasing the number of
parameters requiring optimisation [10,16]. Also,
since film quality is an important issue in the 2D
silicides, MEIS is of particular use since by analy-
sing the RE scattering yield only, the sensitivity to
disordered regions containing a very low RE con-
centration is reduced.
It is clear from Fig. 1 that the only atoms above
the RE in such systems are the Si1 and Si2 atoms.
Therefore these atoms alone will create blocking
features in the Y scattering yield at very specific
scattering angles, making it possible to determine
the positions of these layers [21]. It is also possible
to determine the vibrations of the atoms at the sur-
face, as the shape and depth of the blocking fea-
tures observed are governed by the magnitude of
the atomic vibrations. Thus MEIS is particularly
suited to solving the structure of the 2D Y silicide
as it offers an alternative technique which exclu-
sively determines the positions and vibrations of
the surface bilayer, without the added complica-
tions of effects from deeper layers.
obtained. Temperature measurements were made
using an infrared pyrometer. The heating cycles
were repeated until a sharp 7 · 7 LEED pattern
was observed.
The 2D silicide was prepared by depositing
approximately 1 ML (relative to the Si(111) sur-
face unit cell) of Y from a tungsten basket onto
the clean Si(111) surface and annealing to
ꢁ500 ꢁC for 10 min. The pressure was not allowed
to rise above 1 · 10ꢀ9 mbar during evaporation
and the coverage was determined using a quartz
microbalance and LEED. When a sharp p(1 · 1)
LEED pattern was observed, with no signs of
fractional order spots (which would be due the
presence of other coverage and temperature
dependant reconstructions on the surface), the
sample was transferred under UHV into the ion
scattering chamber.
The ion scattering was conducted using 100 keV
H+ ions in two geometries. These geometries were
chosen to exploit the surface sensitivity of MEIS,
such that the ion yield from the Y layer would
show the blocking effects due to the Si1 and Si2
atoms above them. Thus the geometries used (inci-
ꢀ
ꢀ
dence/detection directions) were ½100ꢂ=½111ꢂ and
ꢀꢀ
½011ꢂ=½100ꢂ, as shown in Fig. 2. By choosing such
scattering angles, the mass separation is sufficiently
large that the angular-dependant ion yield from
each element can be extracted and analysed sepa-
rately. The absence of any contamination in the
sample (e.g. O and C) was established since it
would show up in the MEIS spectra. Data were
acquired for a total beam dose of ꢁ1016 ions cmꢀ2
and a beam size of 0.5 mm vertical and 1.0 mm
horizontal. This beam dose is sufficient to allow
good statistics, whilst minimising the level of
beam-induced damage [11–13,20].
What follows is further experimental evidence
which directly excludes the possibility of a bilayer
˚
separation of 0.33 A. A vibrational analysis has
also been performed, with the effects of varying this
parameter clearly showing a need for enhancement.
2. Experimental
Analysis of MEIS data is accomplished through
comparison of experimental data with Monte Car-
lo simulations of the ion scattering using postu-
lated models. The simulations were done using
XVegas, a multi-platform version of the Vegas
code originally developed by the FOM institute
[23,24]. The optimisation and comparison to
experimental data was achieved by the use of the
v2 R-factor [25,26]. For the analysis that follows,
the starting structures used were the final solution
proposed by Rogero et al. using off-normal LEED
All MEIS data were acquired at the UK MEIS
Facility, CCLRC Daresbury Laboratory. This
facility has been described in detail elsewhere
[22]. All experiments were conducted under ultra-
high vacuum (UHV), with typical base pressures
of 2 · 10ꢀ10 mbar. Clean Si(111) samples (cut
from a lightly doped n-type wafer) were prepared
by e-beam heating to ꢁ1200 ꢁC for 1 min, then
slowly cooled (<100 ꢁC minꢀ1) between 1000 ꢁC
and 600 ꢁC to ensure an ordered surface was

T.J. Wood et al. / Surface Science 598 (2005) 120–127
123
[100]
[100]
[011]
54.74
[111]
19.47
35.26
35.26
Incidence Direction
Detection Direction
Fig. 2. The shadowing/blocking geometries used in this study, for the ideal Si(111) surface.
I–V and the alternative local minimum structure
1.2
1
found in their normal incidence data with a buck-
γ
α
β
˚
ling separation of 0.33 A [16]. The enhanced vibra-
tions that they observed were also used as the
starting point and were assumed to be isotropic
for all atoms in the crystal.
0.8
0.6
0.4
0.2
3. Results and discussion
0.79 Å Buckling
0.33 Å Buckling
Experimental Data
3.1. Structural parameter optimisation
40
45
50
55
60
65
70
a
Scattering Angle / degrees
ꢀ
ꢀ
By using the ½100ꢂ=½111ꢂ geometry (which has
ꢀꢀ
better statistics compared to the ½011ꢂ=½100ꢂ
γ
β
α
geometry due to the lower scattering angles) MEIS
can be used directly to dismiss the possibility of a
˚
buckling separation, zSi1–Si2, of 0.33 A, as this
small value results in drastically different blocking
curves. Simulations of both possible structures
where the only difference is zSi1–Si2 (both structures
b
Y
Si
˚
Fig. 3. (a) The fit for a buckling separation of 0.79 and 0.33 A
and (b) a schematic of the atoms which cause each feature
observed.
˚
have a Si2–RE vertical distance of 1.85 A) are
shown in Fig. 3 along with a schematic of which
atoms cause each blocking feature. It is clear that
˚
˚
the lower value of 0.33 A is not possible since
0.79 A shows some subtle differences in dip posi-
˚
reducing zSi1–Si2 below 0.79 A, causes a large shift
tions. Presented below is a full structural optimisa-
tion of this model, using this structure as a starting
model, by varying the positions of both the Si1
and Si2 layers.
Because a large number of slightly different
models will be compared with experimental data,
it is imperative that the R-factor used is proven
to consistently select the best-fit model. The v2
R-factor is widely used in quantitative MEIS crys-
tallography and was initially used to select the
best-fit model for this system. However, it became
apparent that it was selecting models which were
in the position of the Si1 only feature (labelled b)
to angles below that observed in the experimental
data. It is also clear that the low angle feature,
labelled a, starts to split into two distinct dips,
which is not the case in the experimental data
where there is a very well defined single blocking
feature at ꢁ45ꢁ due to the alignment of the Si1
and Si2 atoms along this direction (see Fig. 3(b)).
The fits shown in Fig. 3(a) were obtained by
simulating the models exactly as proposed by Rog-
ero et al. However, the fit for the accepted value of

124
T.J. Wood et al. / Surface Science 598 (2005) 120–127
close, but clearly not the best fit. This arose
due to the discrepancy in the magnitude of the
blocking feature at low scattering angles (a in
Fig. 3(a)). Since the v2 R-factor is based upon
the difference in absolute yield of the simulation
and experiment (as opposed to being sensitive to
positions of features as is the case for the Pendry
R-factor used in LEED), such a discrepancy can
be misleading for systems dominated by a few
strong blocking features, and in particular where
a low scattering angle feature is due to the align-
ment of two atoms. This is further exacerbated
by the high weighting applied to low scattering
angles due to the Rutherford cross-section. The
v2 R-factor was essentially minimising the discrep-
ancy in the size of this low angle blocking feature
by choosing models where the Si1 and Si2 atoms
are far out of alignment. However, it was possible
to obtain the fits shown in Fig. 4 through a simple
by-eye fitting which aims to get all dip minima
aligned.
1.2
1
0.8
0.6
0.4
0.2
Best Fit Simulation
Experimental Data
40
45
50
55
60
65
70
a
Scattering Angle / degrees
1.2
1
0.8
0.6
0.4
ꢀ
ꢀ
The fit for the ½100ꢂ=½111ꢂ geometry (Fig. 4(a))
˚
Best Fit Simulation
Experimental Data
is clearly an improvement over the 0.79 A buckle
model shown in Fig. 3(a), as the dips in the simu-
lations no longer lie slightly to the left of the
experimental data. Looking along the opposite
70
75
80
85
90
95
100
b
Scattering Angle / degrees
ꢀꢀ
azimuthal direction, the ½011ꢂ=½100ꢂ geometry
Fig. 4. Experimental data versus simulation for a by-eye fit of
ꢀ
ꢀ
the Si1 and Si2 positions. (a) The ½100ꢂ=½111ꢂ scattering
yields the fit shown in Fig. 4(b). Fitting two geom-
etries provides a self-consistency check since many
different atomic configurations could yield the cor-
rect blocking curve for a single geometry, but not
for two complementary geometries. However, it
ꢀꢀ
geometry. (b) The ½011ꢂ=½100ꢂ scattering geometry.
this low angle mismatch in depth is a consistent
problem. Recent measurements have shown that
the point where the beam hits the sample and the
entrance slit of the detector are vertically mis-
aligned by 0.25 mm. This combined with an
azimuthal misalignment of the detector of 0.5ꢁ
can be accommodated in the Vegas code which
improves the low angle fit. Other possibilities such
as non-isotropic vibrations, redistribution of
vibrations between Si1, Si2 and Y, and both long
and short range disorder have all been considered
as possible solutions but do not account for all
aspects of the discrepancy. The good match in
the low angle data, combined with the consistency
across all 2D silicide data obtained at this facility
gives strong evidence for these corrections to be
real. It is also important to note that these adjust-
ꢀꢀ
is clear that the fit in the ½011ꢂ=½100ꢂ geometry is
ꢀ
ꢀ
not as good as that for the ½100ꢂ=½111ꢂ geometry
when considering the size and shape of the block-
ing features. This is due to the poorer statistics
inherent with this geometry. These fits correspond
to a vertical buckling separation (Si1–Si2) of
˚
0.79 A which is almost the same as that observed
˚
in bulk Si (0.784 A), and a Si2–Y vertical distance
˚
of 1.89 A. It is this expansion of the Si2–Y distance
(and hence Si1–Y distance) which improves the fit.
However, a simple by-eye fit is not a satisfac-
tory solution as it is too subjective. Resolving the
low angle discrepancy should enable the v2 R-fac-
tor to function correctly. By looking at other
MEIS data sets on the 2D RE silicides obtained
at Daresbury Laboratory [11–13], it is clear that

T.J. Wood et al. / Surface Science 598 (2005) 120–127
125
v2 R-factor. Both the positions of the Si1 and Si2
atoms and thermal vibrations have been varied
to obtain the fits shown in Fig. 5. The structural
parameters obtained are summarised in Table 1
and the vibrations are discussed in the following
section. The buckling of the surface bilayer is
found to be in agreement with the LEED I–V
result of Rogero et al., with a vertical separation
1.4
1.3
1.2
1.1
1
0.9
0.8
0.7
0.6
0.5
˚
of 0.80 0.03 A, whilst the distance between this
bilayer and the Y layer is found to be expanded
Best Fit Simulation
Experimental Data
˚
slightly (1.89 0.02 A)—though they agree within
the limits of the errors. These values are typical of
those observed for other REs [5–8,10–13].
40
45
50
55
60
65
70
a
Scattering Angle / degrees
It is interesting to draw comparisons between
the results presented here for Y and those deter-
mined for dysprosium silicide on Si(111) [10].
Despite the great difference in atomic number,
and hence mass, the 2D silicides of both these RE
elements have almost identical structural parame-
ters. Both Y and Dy have bulk silicides of the
AlB2 type which present an almost perfect lattice
match to the surface of Si(111) (0% and ꢀ0.36%
respectively [27]). This suggests that the exact
atomic arrangement of the 2D silicide is not gov-
erned by the atomic mass of the RE element being
used, but by the size of the RE present and hence
the strain induced relative to the substrate. This
will be discussed in detail elsewhere [13].
1.05
1
0.95
0.9
0.85
0.8
0.75
0.7
Best Fit Simulation
Experimental Data
75
80
85
90
95
100
b
Scattering Angle / degrees
Fig. 5. Experimental data versus simulation for the best fit
structural model, as determined by the v2 R-factor. (a) The
ꢀ
ꢀ
ꢀꢀ
½100ꢂ=½111ꢂ scattering geometry. (b) The ½011ꢂ=½100ꢂ scattering
3.2. Enhanced vibrations
geometry.
Vibrations play a significant role in the tech-
nique of MEIS by determining the shape of the
blocking features observed. In the case of single
atom blocking features, harmonic vibrations will
not affect the position of the feature, but do alter
ments do not cause any shift in the scattering angle
of the features observed.
Using the modified Vegas codes [13], it has been
possible to do a full optimisation based upon the
Table 1
Structural parameters obtained for Si(111)(1 · 1)-Y using MEIS
Parameter
Si(111)1 · 1-Y
Si(111)1 · 1-Dy
This work
LEED result from Ref. [16]
MEIS result from Ref. [12]
˚
˚
˚
˚
˚
z/A
d/A
z/A
z/A
d/A
Si1–Si2
Si2–RE
0.80 0.03
1.89 0.02
2.35 0.01
2.91 0.01
0.79 0.04
1.85 0.04
0.81 0.03
1.87 0.02
2.26 0.01
2.90 0.01
Layer spacings and bond lengths are represented by z and d respectively. For comparison the LEED result [16] for Si(111)(1 · 1)-Y
and the MEIS result [12] for Si(111)(1 · 1)-Dy are also shown (Dy being the rare earth that in a bulk silicide has the RE–Si bond
closest in length to the bulk Y–Si bond). Errors are as determined by the v2 R-factor [25].

126
T.J. Wood et al. / Surface Science 598 (2005) 120–127
the magnitude and shape of it. The magnitude of
the vibrational enhancements proposed by Rogero
et al. have never been suggested for the 2D silicides
before, so it was particularly valuable to use MEIS
to seek confirmation of this effect. Since the LEED
study showed only the Si bilayer atoms to be
enhanced with respect to their bulk values, only
these atoms were varied in the present study,
which seems a reasonable assumption as any
enhancement would be expected to be at the sur-
face of the reconstruction. Thus the Y vibrational
amplitude was set to the bulk metal value of
1.6
1.4
1.2
1
α
β
γ
0.8
0.6
0.4
0.06 Å
0.11 Å
0.15 Å
0.20 Å
Experimental Data
40
45
50
55 60
65
70
˚
a
Scattering Angle / degrees
0.08 A (calculated from the Debye temperature
of 280 K [28]).
To study the effects of surface vibrations on the
blocking curves, it has been possible to fit the
1.6
1.4
1.2
1
α
β
γ
ꢀ
ꢀ
experimental data for the ½100ꢂ=½111ꢂ geometry
to simulations of a variety of vibrational ampli-
tudes for the best-fit structure determined in the
previous section. This geometry has not only been
chosen because of the better statistics, but since it
clearly exhibits two single atom dips due to the Si1
and Si2 atoms and a dip that results from the
alignment of these two atoms—varying vibrations
for such a situation would result in possible move-
ment of the dip if the atoms are not perfectly
aligned. Fig. 6 shows the results of this study.
The effects of varying the vibrations proves to
be dramatic in the case of the single atom dips.
Fig. 6(a) shows the effect of varying the Si1 vibra-
tion which primarily affects blocking feature b. It
is clear that it is just the size and width of the
feature which varies—it remains at the same
scattering angle for all values of the vibrational
amplitude. An optimal fit of this feature, can be
obtained for a rms vibrational amplitude of
0.8
0.6
0.4
0.2
0.06 Å
0.09 Å
0.12 Å
0.15 Å
Experimental Data
40
45
50
55 60
65
70
b
Scattering Angle / degrees
Fig. 6. The effect of vibrations in the Si bilayer on the shape of
the blocking curves. (a) The Si1 rms vibrational amplitudes.
(b) The Si2 rms vibrational amplitudes.
Table 2
Optimal vibrational amplitudes obtained for Si(111)(1 · 1)-Y
using MEIS
˚
Atom
r/A
This work
LEED result from Ref. [16]
˚
0.15 A. Values which tend towards the bulk value
Si1
Si2
0.15 0.02
0.12 0.02
0.17
0.13
˚
of 0.075 A (as calculated from the Debye tempera-
ture of 543 K [29]) result in this dip being far too
deep. A similar analysis of feature c, which is
due to the position of the Si2 layer, shows an
The isotropic one-dimensional rms vibrational amplitude is
represented by r. The LEED result [16] is given for comparison.
pffiffi
˚
optimal fit for 0.12 A (Fig. 6(b)). These values
ments over the factor of 2 commonly expected
are summarised in Table 2 and are in reasonable
agreement with those obtained in the LEED
I–V analysis of Rogero et al. [16], adding further
evidence to support the presence of significant
enhancements of vibrations at the surface—both
the Si1 and Si2 atoms display substantial enhance-
for surface atoms. The errors shown were deter-
mined through the use of the v2 R-factor [25,26].
These values were used in the structural fitting as
shown in Fig. 5. The small differences in the vibra-
tional amplitudes as determined by MEIS and
LEED could be due to either a difference in how

T.J. Wood et al. / Surface Science 598 (2005) 120–127
127
[3] H. Norde, J. de Sousa Pires, F.M. dÕHeule, F. Pesavento, S.
Petersson, P.A. Tove, Appl. Phys. Lett. 38 (1981) 865.
[4] P. Paki, U. Kafader, P. Wetzel, C. Pirri, J.C. Peruchetti, D.
Bolmont, G. Gewinner, Phys. Rev. B 45 (1992) 8490.
[5] P. Wetzel, C. Pirri, P. Paki, D. Bolmont, G. Gewinner,
Phys. Rev. B 47 (1993) 3677.
[6] L. Stauffer, A. Mharchi, C. Pirri, P. Wetzel, D. Bolmont,
G. Gewinner, Phys. Rev. B 47 (1993) 10555.
[7] M.H. Tulier, P. Wetzel, C. Pirri, D. Bolmont, G. Gewin-
ner, Phys. Rev. B 50 (1994) 2333.
[8] M. Lohmeier, W.J. Huisman, G. ter Horst, P.M. Zagwijn,
E. Vlieg, C.L. Nicklin, T.S. Turner, Phys. Rev. B 54 (1996)
2004.
[9] H. Kitayama, S.P. Tear, D.J. Spence, T. Urano, Surf. Sci.
482 (2001) 1481.
[10] C. Bonet, D.J. Spence, S.P. Tear, Surf. Sci. 504 (2002) 183.
[11] D.J. Spence, S.P. Tear, T.C.Q. Noakes, P. Bailey, Phys.
Rev. B 61 (2000) 5707.
[12] D.J. Spence, T.C.Q. Noakes, P. Bailey, S.P. Tear, Surf. Sci.
512 (2002) 61.
[13] C. Bonet, I.M. Scott, D.J. Spence, T.J. Wood, T.C.Q.
Noakes, P. Bailey, S.P. Tear, Phys. Rev. B 72 (2005)
165407.
[14] R. Baptist, S. Ferrer, G. Grenet, H.C. Poon, Phys. Rev.
Lett. 64 (3) (1990) 311.
static disorder influences the data for each tech-
nique, or the presence of anisotropy in the vibra-
tions of these atoms.
4. Conclusions
The technique of MEIS has been exploited to
determine the atomic positions and vibrations of
the surface Si atoms in the 2D yttrium silicide. A
quantitative optimisation of the positions of the
Si atoms in the top bilayer yields a structure where
the buckling separation (Si1–Si2) is found to be
very similar to the values obtained for other RE
silicides. When compared to the other 2D RE sili-
cides, this layer spacing in the top bilayer has the
closest match to that of bulk Si, providing an
opportunity to study an unperturbed bulk-like Si
bilayer at the surface of a material. However, the
distance between this bilayer and the Y layer is
found to be slightly expanded over the value found
in a LEED I–V study, though it does still lie within
the error limits. It has also been determined inde-
pendently that there is a significant enhancement
in the surface bilayer vibrations as observed by
Rogero et al. [16].
[15] C. Polop, C. Rogero, J.L. Sacedo´n, J.A. Mart´ın-Gago,
Surf. Sci. 482–485 (2001) 1337.
´
[16] C. Rogero, C. Polop, L. Magaud, J.L. Sacedon, P.L. de
´
´
Andres, J.A. Martın-Gago, Phys. Rev. B 66 (2002) 235421.
[17] C. Koitzsch, M. Bovet, M.G. Garnier, P. Aebi, C. Rogero,
´
J.A. Martın-Gago, Surf. Sci. 566–568 (2004) 1047.
´
´
[18] C. Rogero, C. Koitzsch, M.E. Gonzalez, P. Aebi, J. Cerda,
´
J.A. Martın-Gago, Phys. Rev. B 69 (2004) 045312.
Acknowledgements
´
[19] C. Rogero, P.L. de Andres, J.A. Martın-Gago, Phys. Rev.
B 71 (2005) 165306.
[20] D.J. Spence, T.C.Q. Noakes, P. Bailey, S.P. Tear, Phys.
Rev. B 62 (2000) 5016.
[21] J.F. van der Veen, Surf. Sci. Rep. 5 (1985) 199.
[22] P. Bailey, T.C.Q. Noakes, D.P. Woodruff, Surf. Sci. 426
(1999) 358.
[23] R.M. Tromp, J.F.V. der Veen, Surf. Sci. 133 (1983) 159.
[24] J.W.M. Frenken, J.F. van der Veen, R.M. Tromp, Nucl.
Instr. Meth. B 17 (1986) 334.
[25] T.C.Q. Noakes, P. Bailey, D.P. Woodruff, Nucl. Instr.
Meth. B 136–138 (1998) 1125.
The authors would like to acknowledge the
Engineering and Physical Science Research Coun-
cil, for funding this research. Paul Quinn and the
FOM institute are thanked for supplying the XVe-
gas and Vegas simulation codes. The assistance
and technical support of Kevin Connell and Mark
Pendleton are also much appreciated.
[26] D. Brown, T.C.Q. Noakes, D.P. Woodruff, P. Bailey, Y. le
Goaziou, J. Phys.: Condens. Mat. 11 (1999) 1889.
[27] J.A. Knapp, S.T. Picraux, Appl. Phys. Lett. 48 (1986) 466.
[28] C. Kittel, Introduction to Solid State Physics, 7th ed., John
Wiley and Sons, New York, 1996.
References
[1] F.P. Netzer, J. Phys.: Condens. Mat. 7 (1995) 991.
[2] K.N. Tu, R.D. Thompson, B.Y. Tsaur, Appl. Phys. Lett.
38 (1981) 626.
[29] D.S. Gemmel, Rev. Mod. Phys. 46 (1974) 129.