Surface structure determination of silica single layer on Mo(112) by LEED

Article abstract of DOI:10.1016/j.susc.2011.04.003

The atomic structure of silica single layer on a Mo(112) substrate was determined by means of low-energy electron diffraction analysis. The best-fit structure was consistent with findings of previous studies [Phys. Rev. Lett. 95 (2005) 076103 and Phys. Rev. Lett. 103 (2009) 017601]. The unit cell is c(2 × 2)-Si₂O₅ and is composed of a two-dimensional network of SiO₄ tetrahedrons. The tetrahedrons incline slightly to fit the silica network on the Mo(112) surface while maintaining the ideal Si-O bond length. Since there are no dangling bonds in the silica network, the surface is very stable even in the atmosphere.

Full text of DOI:10.1016/j.susc.2011.04.003

Surface Science 605 (2011) 1209–1213
Contents lists available at ScienceDirect
Surface Science
journal homepage: www.elsevier.com/ locate/susc
Surface structure determination of silica single layer on Mo(112) by LEED
Tetsuhiro Kinoshita, Seigi Mizuno ⁎
Department of Molecular and Material Sciences, Kyushu University, Kasuga, Fukuoka 816–8580, Japan
a r t i c l e i n f o
a b s t r a c t
Article history:
The atomic structure of silica single layer on a Mo(112) substrate was determined by means of low-energy
Received 10 March 2011
Accepted 1 April 2011
Available online 8 April 2011
electron diffraction analysis. The best-fit structure was consistent with findings of previous studies [Phys. Rev.
Lett. 95 (2005) 076103 and Phys. Rev. Lett. 103 (2009) 017601]. The unit cell is c(2×2)-Si
2 5
O and is composed
of a two-dimensional network of SiO tetrahedrons. The tetrahedrons incline slightly to fit the silica network
4
on the Mo(112) surface while maintaining the ideal Si–O bond length. Since there are no dangling bonds in
the silica network, the surface is very stable even in the atmosphere.
Keywords:
Low energy electron diffraction (LEED)
Molybdenum
©
2011 Elsevier B.V. All rights reserved.
Silicon oxides
1
. Introduction
density functional theory (DFT) calculation [16,17]. A similar network
model was also proposed by Giordano et al. [18,19]. Recently, Seifert
et al. solved the controversy between above two models using ion
beam triangulation method [20,21]. The data clearly provided
evidence for the network model. However, structural determination
using diffraction methods has not yet been done. Quantitative LEED
intensity analysis is one of the most powerful and reliable methods for
surface structure determination; it has been used to determine many
surface structures with good accuracy, e. g. 0.1 Å. In this study, we
examined the Mo(112)-c(2×2)-silica structure by means of LEED
analysis. The determined structure model is again consistent with the
network model proposed by Weissenrieder and Todorova et al.
Thin oxide films have received great attention for their numerous
applications to integrated circuits and as protective films against
corrosion, and as supports for metal nanoparticles in catalysts and
sensors. Their importance will be further increased with the
miniaturization of devices. The atomic-level structure of thin-film
oxides must be elucidated in order to determine their properties, and
these structures have been studied for their technological importance
and for the sake of fundamental sciences. Especially, thin-film oxides
on metal surfaces have been studied as support materials of catalytic
reactions [1]. Although silica layers were obtained on Mo(110) and
Mo(100) surfaces [2,3], they could not be used as model cases because
they were amorphous. On the other hand, Schroeder et al. obtained a
crystalline silica layer on a Mo(112) surface, and it would be ideal as a
support material for catalytic reactions [4–7]. Growth of crystalline
silica thin layers on Ni(111), Pd(100) and Ru(0001) has also been
reported [8–10], and many applications for these materials as model
support materials, devices, sensors and so on have been expected.
2. Experiment
Experiments were carried out in an ultrahigh vacuum (UHV)
chamber (the base pressure during experiments was better than
−
7
5×10 Pa) equipped with LEED optics (SPECTALEED, Omicron). A
single-crystal Mo(112) disk (10 mmϕ, 2 mm thick) was welded to a Ta
wire (0.7 mmϕ) for resistive heating. The sample was simultaneously
heated by electron bombardment from a tungsten filament placed
closed to the backside of the sample. The temperature was measured by
a thermo spot sensor (FTZ6, Japan Sensor). First, the sample was cleaned
Moreover, crystalline silica single layers were obtained on SiC(0001)
and SiC(0001) surfaces [11,12]. Their structures have been already
determined and they are robust even in air.
–
Crystalline silica thin layers on Mo(112) have a c(2×2) unit cell
−5
[
4], and their surface structures have been studied by several
by cycles of annealing in 1×10 Pa oxygen at 1200 K and flashing to
2200 K in UHV. A clear (1×1) LEED pattern was observed as reported in
Ref. 22. Secondly, the sample surface was oxidized for 10 min in
methods. Chen et al. studied the surface using low-energy electron
diffraction (LEED), high-resolution electron energy-loss spectroscopy
−6
(
HREELS) and scanning tunneling microscopy (STM), and proposed
5×10 Pa oxygen at 850 K. Finally, silicon was deposited on the
sample from a high-temperature Si wafer heated by direct resistive
heating. The silicate film was prepared by Si deposition for 5 min in
an isolated SiO model [13–15]. Weissenrieder and Todorva et al.
proposed a network model using STM, infrared reflection absorption
spectroscopy (IRAS), X-ray photoelectron spectroscopy (XPS) and
4
−5
5×10 Pa oxygen at 800 K, followed by annealing for 5 min at 1200 K
in UHV. A clear c(2×2) LEED pattern was observed after this treatment
as previously reported in Ref. 4.
The LEED spot intensities were measured by a digital charge-
coupled device camera with a computer-controlled data acquisition
⁎
Corresponding author.
E-mail address: mizuno.seigi.918@m.kyushu-u.ac.jp (S. Mizuno).
0039-6028/$ – see front matter © 2011 Elsevier B.V. All rights reserved.
doi:10.1016/j.susc.2011.04.003

1210
T. Kinoshita, S. Mizuno / Surface Science 605 (2011) 1209–1213
Table 1
Structural parameters obtained for the best-fit Mo(112)-(1×1) clean surface model.
¯
¯
¯
Atom (No.)
[112] (Å)
[110] (Å)
[111] (Å)
Mo(1)
Mo(2)
Mo(3)
Mo(4)
Mo(5)
Mo(6)
−1.10
0.01
1.36
2.58
3.90
5.17
± 0.02
± 0.03
± 0.03
± 0.03
± 0.04
± 0.03
2.23
0.00
2.23
0.00
2.23
0.00
0.91
-0.05
-0.97
0.83
-0.02
-0.97
± 0.07
± 0.05
± 0.06
± 0.06
± 0.07
± 0.07
by the dashed line. The unit cell is shown by a rectangle with side
–
lengths of 4.45 Å and 2.73 Å. The bulk layer distance toward the [112]
Fig. 1. LEED patterns of (a) Mo(112)-(1×1) clean surface and (b) Mo(112)-c(2×2)-silica
surface.
direction is 1.28 Å. The surface perpendicular stacking has a 6-layer
–
period with a 2-layer period toward the [110] direction and a 3-layer
–
period toward the [111] direction.
system [23]. For structural analysis, the intensity versus electron beam
energy spectra [I(E) curves] of the LEED spots were measured within
an incident energy range of 50–400 eV. The summation of energy
ranges of inequivalent I(E) curves, ΔE, were 1647 and 2934 eV for
clean Mo(112)-(1×1) and c(2×2)-silicate surfaces, respectively. A
Barbieri–Van Hove symmetrized automated tensor LEED package was
used to determine the atomic positions [24]. Thirteen phase shifts
were used to represent atomic scattering. The dumping of incident
electrons was represented by an imaginary part of the inner potential,
The structural parameters obtained from this study are shown in
Table 1. Pendry's reliability factor became 0.193 after relaxation of the
interlayer distances. We used seven inequivalent beams, and the total
energy range was 1647 eV. The calculated I(E) curves agreed with the
experimental I(E) curves very well as shown in Fig. 3. The Debye
temperatures were fitted to 150, 175, 250, 350, 450 and 450 K for each
Mo layer from the surface topmost layer to the 6th layer, respectively.
Interlayer distances calculated from Table 1 were 1.11, 1.35, 1.22, 1.32
and 1.27 Å from the surface and show a typical surface relaxation. This
was also consistent with previous LEED analyses (1.09, 1.31, 1.26 and
1.29 Å from surface) [22]. The total energy range used in the previous
V
0i, of −5.0 eV. Pendry's reliability factor (R
automated search algorithm [24,25]. The best agreement between
experimental and theoretical I(E) curves involved minimizing the R
Errors in the structural parameters were estimated from the variance
P
) was used to direct the
P
.
P
study was 991 eV and the obtained R was 0.205. The present study
had a larger total energy range (1647 eV) and a smaller R value
p
(0.193).
1/2
of the R
P
, ΔR=R
P
(8|V0i|/ΔE) [25].
3
. Results and discussion
3.2. Structural determination of the Mo(112)-c(2×2)-silica surface
3
.1. Structural determination of the clean Mo(112)-(1×1) surface
The LEED pattern of the silica surface is shown in Fig. 1(b). A sharp
c(2×2) pattern was observed as in previous studies after the sample
treatment of silicon deposition, oxidation and annealing [4,13,17]. We
have examined the isolated silica model, the network model and other
First we examined the structure of the clean Mo(112)-(1×1)
surface. The LEED pattern of the clean Mo(112) surface is shown in
Fig. 1(a). The structure had been determined by Kolthoff et al. using
LEED analysis and is an ideally truncated crystal surface [22]. The
structure model is shown in Fig. 2. There is one mirror plane displayed
Fig. 3. Experimental and theoretical I(E) curves of the best-fit Mo(112)-(1×1) clean
Fig. 2. Top and side views of the Mo(112)-(1×1) clean surface.
surface model.

T. Kinoshita, S. Mizuno / Surface Science 605 (2011) 1209–1213
1211
Fig. 4. Top and side views of examined structure models of the Mo(112)-c(2×2)-silica.
possible models by LEED analysis. The eleven examined Mo(112)-
c(2×2)-silica models are shown in Fig. 4. Every model has one mirror
plane indicated by a dashed line. Model 1 is the network model
proposed by Weissenrieder and Todorva et al. and models 2–6 are its
derivative models. Oxygen atoms sit on a bridge site, a top site or a
hollow site of the Mo interface layer for models 1, 2 and 3, respectively.
The topmost silicate networks sit on oxygen atoms, and the oxygen
atoms connect Mo and Si atoms with bent bonds. Models 4–6 are similar
to models 1–3, respectively, but the oxygen atoms bridge Mo and Si
atoms with almost straight bonds similar to the case of the silica layer on
by Chen et al., and models 8–11 are related models [13]. Model 8 and 10
have two or three SiO
4
units in their unit cells, respectively, and the SiO
4
units are connected with each other. Model 9 shows a dimer of two SiO
4
units sharing a corner oxygen. Model 11 resembles β quartz, in which
one SiO unit shares its four corner oxygen atoms with four neighboring
SiO units. We used 14 inequivalent beams for LEED analysis. The
obtained R values for the 11 models are shown in Table 2. Models 1 and
4 had small R values of 0.18 and 0.23, and other models had R values
greater than 0.5. Model 1 and model 4 have similar structures, and they
also have similar theoretical I(E) curves. However, their R values have
meaningful differences of 0.05 compared to the ΔR value of 0.02. The
4
4
P
P
P
P
4
an SiC(0001) surface [11]. Model 7 is the isolated SiO model proposed

1212
T. Kinoshita, S. Mizuno / Surface Science 605 (2011) 1209–1213
Table 2
Pendry's r-factors (R
shown in Fig. 4.
p
) for the models
Model
R
p
1
2
3
4
5
6
7
8
9
0.18
0.67
0.58
0.23
0.60
0.61
0.66
0.66
0.66
0.58
1
0
theoretical I(E) curves for model 1 were plotted with the experimental
I(E) curves in Fig. 5 and they have quite nice agreement. Therefore, we
concluded that model 1 is the correct model. It is consistent with the
previous study using ion beam triangulation method [20,21]. The
optimized best-fit model is shown in Fig. 6 and its structural parameters
are listed in Table 3. The optimized Debye temperatures were 300, 300,
2
25, 300, 350, 400 and 450 K for O(1,1′), O(2), Si(1,1′), O(3,3′), Mo
2 5
Fig. 6. Top and side views of the best-fit Mo(112)-c(2×2)-Si O surface (model 1).
(
1,1′), Mo(2,2′) and Mo(3,3′), respectively.
The diamond in the top view of Fig. 6 shows the c(2×2) unit cell. The
topmost layer in the unit cell consists of three oxygen atoms and one
silicon atom. Two oxygen atoms were equivalent (marked 1 and 1′) but
the other oxygen atom was inequivalent (marked 2). The oxygen atoms
bind with two equivalent silicon atoms (marked 1 and 1′) to form a two-
dimensional hexagonal network. The silicon atoms connect with the
second layer of oxygen atoms (marked 3 and 3'). As a result, each silicon
4
atom has four bonds, and the SiO units share the corner oxygen atoms.
The second-layer oxygen atoms are located on the bridge site of the
substrate Mo atoms (marked 1 and 1′). The structure of the Mo atoms is
similar to the structure of clean Mo(112)-(1×1), but the interlayer
distances are different. The first and second interlayer distances are 1.21
and 1.34 Å, and these are much closer to the bulk value (1.28 Å)
compared with the case of the clean surface (1.11 and 1.35 Å). It is
reasonable that the surface covered by the silica layer and the surface
relaxation becomes smaller than the clean surface.
2 5
3.3. Features of the Mo(112)-c(2×2)-Si O surface
The determined structure has no dangling bond in the silica layer
and is composed of Si . The bond lengths and bond angles of the
silica layer are listed in Table 4 and are calculated from the best-fit
structural parameters. Corresponding values obtained from an α-SiO
crystal [26] and MoO crystal [27], and previous results by first
2 5
O
2
2
principles calculation [17] are also shown. Our results are very similar
to the previous results by Todorova et al. The bond lengths are similar
to the typical values in the bulk crystal. These results suggest the high
stability of the Mo(112)-c(2×2)-Si O surface. Actually, we could
2 5
observe clear c(2×2) LEED patterns after we exposed the sample to
air without any treatment in the UHV chamber.
Table 3
Structural parameters obtained for the best-fit Mo(112)-c(2×2)-Si
O
2 5
surface (model 1).
¯
¯
¯
Atom (No.)
[112] (Å)
[110] (Å)
[111] (Å)
O(1)
O(1′)
O(2)
Si(1)
Si(1′)
O(3)
O(3′)
Mo(1)
Mo(1′)
Mo(2)
Mo(2′)
Mo(3)
Mo(3′)
−0.68
−0.68
−0.29
−0.03
−0.03
1.56
1.56
3.18
3.18
4.39
± 0.04
± 0.04
± 0.07
± 0.02
± 0.02
± 0.05
± 0.05
± 0.03
± 0.03
± 0.04
± 0.02
± 0.03
± 0.03
2.25
2.20
0.00
1.58
2.87
1.96
2.49
2.22
2.23
0.00
0.00
2.21
-2.21
± 0.12
± 0.12
−1.32
1.41
−0.01
0.01
2.74
0.04
2.77
−1.33
1.40
3.20
0.50
± 0.14
± 0.14
± 0.23
± 0.07
± 0.07
± 0.13
± 0.13
± 0.06
± 0.06
± 0.07
± 0.07
± 0.08
± 0.08
± 0.09
± 0.09
± 0.12
± 0.12
± 0.05
± 0.05
4.39
5.71
5.71
± 0.06
± 0.06
2.26
2.26
Fig. 5. Experimental and theoretical I(E) curves for the best-fit structural model of the
Mo(112)-c(2×2)-Si O surface.
2 5

T. Kinoshita, S. Mizuno / Surface Science 605 (2011) 1209–1213
1213
Table 4
4. Conclusion
Summary of bond lengths and bond angles.
This study
Ref. [17]
α-SiO
2
[26]
MoO
2
[27]
We have reinvestigated the structure of a clean Mo(112) surface
and silica single layer on a Mo(112) surface. The clean Mo(112)
surface has a crystal truncated structure with large surface relaxation.
Si(1)–O(1) (Å)
Si(1)–O(1′) (Å)
Si(1)–O(2) (Å)
Si(1)–O(3) (Å)
Mo(1)–O(3) (Å)
Si(1)–O(1)–Si(1′)
Si(1)–O(2)–Si(1′)
1.63
1.66
1.60
1.64
2.14
134°
162°
1.64
1.64
1.62
1.65
2.11
133°
163°
1.61
1.61
1.61
1.61
The silica single layer has a two-dimensional network of SiO
tetrahedrons with a c(2×2)-Si unit cell. The tetrahedrons incline
4
2 5
O
2.01
slightly to fit the Mo(112) substrate and have ideal Si–O bond lengths.
The silica layer has no dangling bonds and this makes the surface very
stable.
144°
144°
Acknowledgements
Similar network structures of silica or silicon oxynitride layers
were reported on 6H–SiC (0001) and 6H–SiC(0001) substrates
11,12]. Their structural parameters have been determined by LEED
analysis. They are also terminated by silica single layer networks. The
–
The authors thank Prof. M.-S. Chen for suggestions about sample
preparation. This work was supported by a Grant-in-Aid for Scientific
Research (KAKENHI 20340077) from the Ministry of Education,
Culture, Sports, Science and Technology of Japan.
[
Si–O bond lengths are 1.61–1.63 Å and the Si–O–Si bond angles are
1
41–146° and 180°. Though the bond lengths are quite similar to the
case of the Mo surface, the bond angles are different. The topmost
silica layer of the SiC substrate has a hexagonal unit cell, and the base
References
[
1] H.-J. Freund, Surf. Sci. 500 (2002) 271.
of the SiO
on SiC are
5
4
tetrahedron is parallel to the surface. The unit cells of silica
pffiffiffi pffiffiffi
ð Þ
R30° and hexagonal with a lattice constant of
4
.35 Å. On the other hand, the SiO tetrahedron on the Mo substrate
[2] X. Xu, D.W. Goodman, Surf. Sci. 282 (1993) 323.
[3] J.W. He, X. Xu, J.S. Corneille, D.W. Goodman, Surf. Sci. 279 (1992) 119.
[4] T. Schroeder, M. Adelt, B. Richter, M. Naschitzki, M. Bäumer, H.-J. Freund, Surf. Rev.
Lett. 7 (2000) 7.
3 ×
3
–
–
inclines toward the [110] or [110] direction. The O(1,1′) atoms come
[
5] T. Schroeder, A. Hammoudeh, M. Pykavy, N. Magg, M. Adelt, M. Bäumer, H.-J.
Freund, Solid State Electron. 45 (2001) 1471.
out to the vacuum and the O(2) atoms go down to the bulk. Since the
O(3,3′) atoms adsorb on the bridge site of the Mo(112), the SiO
4
[6] T. Schroeder, J.B. Giorgi, M. Bäumer, H.-J. Freund, Phys. Rev. B 66 (2002) 165422.
[
7] T. Schroeder, M. Adelt, B. Richter, M. Naschitzki, M. Bäumer, H.-J. Freund,
Microelectr. Rel. 40 (2002) 841.
tetrahedron is inclined and the Si–O–Si bond angles are changed from
the typical values. The unit cell of silica on Mo is quasi-hexagonal with
a lattice constant of 5.23 Å. While the size of silica on Mo is smaller
than that of silica on SiC, the Si–O bond lengths are almost equal due
to the incline of the tetrahedrons. Only the Mo(112) surface is suitable
for the growth of silica among the Mo crystal surfaces. Other low-
index Mo surfaces do not have suitable cells for the growth of silica
layers. Very recently, the growth of a double-layer sheet silica model
on an Ru(0001) surface was reported [10]. The lattice constant of the
hexagonal Ru(0001)-(2×2) unit was 5.42 Å, and the size was suitable
for the growth of similar silica network structures.
[
8] K. Manisha, Y. Murata, Appl. Phys. Lett. 80 (2002) 1921.
[9] Z. Zhang, Z. Jiang, Y. Yao, D. Tan, Q. Fu, X. Bao, Thin Solid Films 516 (2008) 3741.
10] D. Löffler, J.J. Uhlrich, M. Baron, B. Yang, X. Yu, L. Lichtenstein, L. Heinke, C. Büchner, M.
Heyde, S. Shaikhutdinov, H.-J. Freund, R. Włodarczyk, M. Sierka, J. Sauer, Phys. Rev.
Lett. 105 (2010) 146104.
[11] J. Bernhardt, J. Schardt, U. Starke, K. Heinz, Appl. Phys. Lett. 74 (1999) 1084.
12] T. Shirasawa, K. Hayashi, S. Mizuno, S. Tanaka, K. Nakatsuji, F. Komori, H.
Tochihara, Phys. Rev. Lett. 98 (2007) 136105.
13] M.-S. Chen, A.K. Santra, D.W. Goodman, Phys. Rev. B 69 (2004) 155404.
[14] S. Wendt, E. Ozensoy, T. Wei, M. Frerichs, Y. Cai, M.-S. Chen, D.W. Goodman, Phys.
Rev. B 72 (2005) 115409.
15] M.-S. Chen, D.W. Goodman, Surf. Sci. 600 (2006) L255.
16] J. Weissenrieder, S. Kaya, J.-L. Lu, H.-J. Gao, S. Shaikhutdinov, H.-J. Freund, M.
Sierka, T.K. Todorova, J. Sauer, Phys. Rev. Lett. 95 (2005) 076103.
[17] T.K. Todorova, M. Sierka, J. Wessenrieder, J.-L. Lu, H.-J. Gao, S. Shaikhutdinov, H.-J.
Freund, Phys. Rev. B 73 (2006) 165414.
18] L. Giordano, D. Ricci, G. Pacchioni, P. Ugliengo, Surf. Sci. 584 (2005) 225.
19] D. Ricci, G. Pacchioni, Phys. Rev. B 69 (2004) 161307 (R).
[20] J. Seifert, D. Blauth, H. Winter, Phys. Rev. Lett. 103 (2009) 017601.
21] J. Seifert, H. Winter, Surf. Sci. 603 (2009) L109.
22] D. Kolthoff, H. Pfnur, A.G. Fedorus, V. Koval, A.G. Naumovets, Surf. Sci. 439 (1999)224.
23] S. Mizuno, H. Tochihara, A. Barbieri, M.A. Van Hove, Phys. Rev. B 52 (1995) R11658.
[24] M.A. Van Hove, W. Moritz, H. Over, P.J. Rous, A. Wander, A. Barbieri, N. Materer, U.
Strarke, G.A. Somorjai, Surf. Sci. Rep. 19 (1993) 191.
25] J.B. Pendry, J. Phys. C 13 (1980) 937.
26] L. Levien, C.T. Prewitt, D.J. Weidner, A. Mineral. 65 (1980) 920.
[
[
[
[
[
The STM images in the previous studies show chain-like features
––
toward the [111] directions [15,17]. This is very consistent with our
model. The SiO tetrahedrons incline slightly, and the O(1,1′) atoms
4
[
[
protrude further than the O(2) atoms. As a result, the STM images
might appear as stripes rather than a honeycomb. In the high-
resolution STM images by Chen et al., single or double spots appear in
the unit cell depending on the sample bias voltages [15]. The double
spots are difficult to explain by the hexagonal silica layer. However,
we have two equivalent O(1,1′) atoms and one inequivalent O(2)
atom in the unit cell, and it is likely to have electronic states to be
showing two spots in the unit cell.
[
[
[
[
[
[27] D.B. Rogers, R.D. Shannon, A.W. Sleight, J.L. Gillson, Inorg. Chem. 8 (1969) 841.