Oxide templates for self-assembling arrays of metal nanoclusters

Article abstract of DOI:10.1016/j.susc.2006.07.059

We demonstrate that ordered arrays of metal clusters of a few nanometers in size can be fabricated by self-assembly on a range of oxide nanotemplates. Two systems were chosen, FeO/Pt(1 1 1) and Fe₃O₄, having in common a reconstructed close-packed oxygen surface layer. Deposition of various metal films results in the formation of nanosized clusters arranged according to the surface long-range order. Employing ab-initio calculations we discuss the observed mechanism of site-selective nucleation of the metal nanoclusters. Finally, we discuss the existence of a wide family oxide templates that can be used for self-assembled growth of nanoclusters governed by the same rules.

Full text of DOI:10.1016/j.susc.2006.07.059

Surface Science 600 (2006) L287–L290
www.elsevier.com/locate/susc
Surface Science Letters
Oxide templates for self-assembling arrays of metal nanoclusters
a,
b
b
b
b
*
N. Berdunov , G. Mariotto , K. Balakrishnan , S. Murphy , I.V. Shvets
a
School of Physics and Astronomy, Nottingham University NG7 2RD, UK
Centre for Research on Adaptive Nanostructures and Nanodevices, Physics Department, Trinity College, Dublin 2, Ireland
b
Received 31 May 2006; accepted for publication 28 July 2006
Available online 28 August 2006
Abstract
We demonstrate that ordered arrays of metal clusters of a few nanometers in size can be fabricated by self-assembly on a range of
oxide nanotemplates. Two systems were chosen, FeO/Pt(111) and Fe , having in common a reconstructed close-packed oxygen sur-
3 4
O
face layer. Deposition of various metal films results in the formation of nanosized clusters arranged according to the surface long-range
order. Employing ab-initio calculations we discuss the observed mechanism of site-selective nucleation of the metal nanoclusters. Finally,
we discuss the existence of a wide family oxide templates that can be used for self-assembled growth of nanoclusters governed by the
same rules.
ꢀ
2006 Elsevier B.V. All rights reserved.
Keywords: Self-assembly of nanocluster arrays; Patterned oxide surfaces; Nanotemplates
Nanometer scale structures, like nanodots, wires, crys-
tals, etc. have recently attracted a lot of interest because
of their potential for use in novel device applications for
magnetic recording, sensors, quantum computing, short-
wavelength optics and spin electronics. For example, the
magnetic anisotropy of nanostructured media can be tai-
lored by simply changing the shape and size of the nano-
structures it comprises, which can be used for new
magnetic storage media [1]. Electron confinement in such
nano-objects would also facilitate quantum-computing
applications. Achieving periodic ordering within a particu-
late media would provide coherent superposition of spin-
electron states at the nanometer scale, which is essential
for spin-transport applications [2]. To realise a device
based on such nano-objects it is essential to have a large
number of identical nano-objects acting in the same man-
ner, i.e. representing a collective behavior [3]. Thus, an
array of crystallites regularly arranged and of uniform
size/orientation is a desirable medium. We present two
examples of oxide nanotemplates for self-assembled epitax-
ial growth of nanocluster arrays [4–9], which demonstrate
the possibility of tuning the size and spatial order of metal
clusters. The following templates were investigated: FeO/
Pt(111) and Fe O (111). All experiments were performed
3
4
À10
in UHV conditions at a base pressure of 2 · 10
Torr.
Metal ultrathin films were deposited by e-beam evapora-
tion at room temperature using a quartz crystal balance
monitor for thickness control. LEED and STM measure-
ments were used to monitor the film quality. Constant cur-
rent STM scans were performed at room temperature.
1. Oxygen-terminated Fe O (111) nanotemplate
3 4
In this experiment, a Fe O (111) single crystal and a
3
4
1
00 nm Fe O (111) film grown on MgO(111) were used
3 4
(
for growth conditions see [10]). Both Fe O substrates were
3 4
À6
oxidised in 10 Torr O at 1000 K [11]. In the case of the
2
Fe O /MgO film, the annealing time was only 2 min, to pre-
3
4
vent significant diffusion of Mg to the surface. This results
in a 4.2 nm superstructure covering almost the entire sample
surface. As we have shown in our previous work [11] this
represents an oxygen-terminated surface, which recon-
structs under intrinsic stress creating a long-range order.
*
Corresponding author.
E-mail address: nikolai.berdunov@nottingham.ac.uk (N. Berdunov).
0
039-6028/$ - see front matter ꢀ 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.susc.2006.07.059

L288
N. Berdunov et al. / Surface Science 600 (2006) L287–L290
Fig. 1. (a) Series of STM images of Fe
3
O
4
(111) superstructured surface taken at different tunneling bias showing the inversion of tunneling contrast; (b)
˚
3 4 3 4
STM images of 2 A Fe film deposited on 100 nm film Fe O /MgO(111) and (c) 0.5 A Cr-film deposited on single crystal Fe O (111). Initially, both
˚
samples were annealed in oxygen atmosphere to form an oxygen-terminated surface reconstructed into a 4.2 nm superstructure. Clusters of 2 AL (b) and
AL (c) thickness form an ordered array with the superstructure periodicity; (inset-c shows a LEED picture of the surface after deposition).
1
The long-range order reveals a charge modulation along the
surface seen in bias-dependent STM images (Fig. 1(a)). As it
has been found in [12], this surface charge redistribution is
due to surface strain governing the interaction between oxy-
gen and iron. Thus the effective surface oxygen charge
changes following the surface strain pattern. One can there-
fore expect significant variations in adsorption properties
across the surface superstructure.
image it is possible to identify the vanadium clusters’ nucle-
ation sites. According to the FeO/Pt(111) model Fig. 2(d)
and STM simulations [14,15] the hcp sites appear as bright
protrusions (Fig. 3(b)), when the vanadium clusters are
positioned where the FeO layer forms fcc stacking se-
quence with the platinum surface, i.e. the darker regions
between the hcp sites.
We have performed DFT calculations to identify the
geometry of a single atomic layer of adsorbed metal film.
The CASTEP program [16] as a module of Materials Stu-
dio has been used in all our Density Functional Theory
(DFT) calculations. The local density GGA functional
with ultrasoft pseudopotentials was used to optimize the
surface geometry, calculate the electron density distribu-
tion and adsorption energy. A single unit cell vacuum slab
of FeO/Pt(111) similar to that used in [7] and a second
with the FeO film shifted to form an hcp stacking sequence
with the substrate have been used for adsorption energy
calculations. A plane wave cutoff of 340 eV and a
(6 · 6 · 1) grid of k-points were chosen. The calculation re-
sults suggest that in both cases the favourable adsorption
sites for Fe adatoms are on-top of oxygen. The calculated
Deposition of Fe and Cr films results in the formation of
regular arrays of nanoclusters on the magnetite surface
˚
˚
Fig. 1(b), (c). In the case of 0.5 A and 1 A film thickness,
the clusters’ height is one atomic layer (AL) [13], whereas
˚
a 2 A thick film results in 2 AL clusters formation, conserv-
ing the arrays regularity. LEED measurements (Fig. 1(c)
inset) suggest that the Fe and Cr clusters grow epitaxially
as a sharp LEED pattern is obtained.
2
. FeO/Pt(111) nanotemplate
The growth of a single FeO layer on Pt(111) results in a
coincidence structure (Moir e´ pattern) of 2.6 nm periodicity
˚
À6
[
14]. The FeO film was formed by depositing 1 A of Fe on a
˚
Pt(111) substrate and post oxidising it in 10 Torr O at
interlayer distance of 1.8 A between adsorbed Fe and sur-
2
1
000 K for 2 min similar to [14]. Fig. 2(a) shows Fe clusters
face oxygen is well in agreement with STM measured val-
ues of 1.5–1.8 A. Surprisingly, the difference in the
adsorption energy between fcc and hcp FeO/Pt(111) con-
figuration is only about 0.1 eV, which suggests that there
is probably another mechanism involved in the preferential
nucleation observed in the experiments.
˚
˚
nucleated on the surface after depositing 0.2 A Fe film. The
clusters are well confined and repeat the surface superlat-
tice. LEED patterns taken before and after deposition
(
Fig. 2(c)) confirm that low iron coverage does not destroy
FeO/Pt(111) superstructure.
The ‘site selective’ nucleation is demonstrated in
Fig. 2(b) corresponding to 0.05 A of vanadium deposited
on the superstructured FeO/Pt(111) surface. In the STM
In the cases presented above, the reason for the long-
range surface reconstruction is a structural instability:
either due to a non-uniform oxygen lattice as in magnetite
˚

N. Berdunov et al. / Surface Science 600 (2006) L287–L290
L289
˚
Fig. 2. (a) STM image of 0.2 A Fe film grown on FeO/Pt(111) template. Clusters of 1 AL thickness nucleate following the long-range surface
superstructure of 2.6 nm periodicity. The image is presented in differentiated contrast. 2D Fourier transformation image shows preferred clusters’
˚
periodicity of 2.6 ± 0.6 nm. (b) Demonstration of ‘site selective’ nucleation of vanadium clusters (white spots) after depositing a 0.05 A vanadium film. The
smaller corrugation seen on the image is the surface superstructure. Three hollow arrows point out the hcp sites of FeO/Pt(111) and the solid arrows
indicate the vanadium clusters nucleated at fcc sites. Unit cell of FeO/Pt(111) superstructure is indicated. (c) LEED pictures before (left) and after (right)
˚
depositing a 0.2 A Fe film. (d) The model of FeO/Pt(111) coincidence structure [13], comprising a FeO bi-layer on the Pt(111) surface (top-view and side-
view). Two sites of FeO forming an fcc and an hcp stacking sequence with Pt(111) surface are indicated.
[
11], or caused by a film’s mismatch with the substrate and
binding energy than stacking sequence variations. Fig. 3
shows that charge transfer between the surface oxygen
and the iron in the layer below depends on the lateral stress
[12,18]. This also explains surface charge modulation
across the surface superstructure observed in our bias-
dependent STM images Fig. 1(a) and also in Ref. [19].
Therefore, we conclude that the lateral charge re-allocation
caused by the Fe–O bond distortion results in a modulation
of the surface binding energy within the surface superlat-
tice, which gives rise to the preferential nucleation of the
metal clusters. Moreover, in contrast to the easy oxidized
metals used in our experiments, gold clusters preferential
nucleation can only be observed at low temperature due
to lesser adhesion as has been demonstrated in [19].
A close-packed oxygen termination is a common feature
of type three polar surfaces for (111) and (0001) orienta-
tions of oxide crystals [17], so the other oxides of this fam-
ily may display similar reconstructions. Among the
experimental reports on long-range reconstructions of oxy-
gen-terminated oxide surfaces we have identified a number
of systems which fall into our nanotemplate family
surface polarity [17]. In some cases, the distortion in the
oxygen lattice is evident from STM data [7,11,15], while
in others the existence of the strain in the surface layer
can be deduced from the local variations in the stacking se-
quence. In particular, fcc and hcp local structures, which
can be identified in the FeO/Pt(111) coincidence structure
(
Fig. 2(b)), relax differently resulting in a lateral strain
modulation. Thus, geometry optimization of the single
FeO layer on Pt(111) gives us two stable structures with
different in-plane lattice constants of 3.1 A and 2.8 A.
The first value is for FeO single layer relaxed on a fcc site
and the second on an hcp site Fig. 2(b).
To understand the effect of the strain/distortion in the
top surface layer we have performed DFT calculations
for the FeO(111) laterally distorted structure. The calcu-
lated values of adsorption energy for the Fe adlayer gives
1
(
energy difference between fcc and hcp FeO/Pt(111). This
suggests that the local distortion has a greater effect on
˚
˚
1
eV stronger adsorption in the case of stretched FeO
˚
3.1 A), which is much stronger than the 0.1 eV adsorption
[
8,11,14,15,20–29]. Thus, FeO grown on Rh(111) [15]
forms a coincidence structure similar to FeO/Pt(111) but
smaller periodicity of 1.4 nm due to the large film/substrate
mismatch. In some cases, growth of a native oxide by oxi-
dizing a metal substrate results in the formation of similar
1
DFT calculations settings: GGA-PBE functional, 8 · 8 · 1 k-point set,
˚
3
00 eV energy cutoff; two vacuum slabs constructed of 7 A thick FeO with
in-plane lattice constants of (2.8 A · 2.8 A) and (3.1 A · 3.1 A) and 10 A
of vacuum; full geometry optimization performed for fixed unit cell.
˚
˚
˚
˚
˚

L290
N. Berdunov et al. / Surface Science 600 (2006) L287–L290
Fig. 3. Cross-sectional view of the FeO vacuum slab and charge density difference distribution for stretched (a) and compressed (b) FeO, corresponding to
˚
˚
in-plane lattice constants of 3.1 A and 2.8 A, respectively.
surface superstructure as been demonstrated in the case of
[10] K. Balakrishnan, S.K. Arora, I.V. Shvets, J. Phys. Condens. Matter
6 (2004) 5387.
11] N. Berdunov, S. Murphy, G. Mariotto, I.V. Shvets, PRB 70 (2004)
85404.
12] N. Berdunov, S. Murphy, G. Mariotto, I.V. Shvets, PRL 93 (2004)
057201.
1
RhO /Rh(111) and been exploited to form Fe-clusters ar-
2
[
[
rays [15]. In all these examples the superstructure periodic-
ity ranges from 1.4 nm to 4.5 nm. The future expansion of
this list will give us the freedom of tailoring the nanoclus-
ters in size and array spacing. Moreover, the different
spin/electron transport properties of the substrates provide
enormous potential for manufacturing high-performance
magnetic and spin-transport materials and also for molec-
ular engineering.
0
[13] N. Berdunov, G. Mariotto, S. Murphy, K. Balakrishnan, I.V. Shvets,
PRB 71 (2005) 113406.
[
[
14] H.C. Galloway, P. Sautet, M. Salmeron, PRB 54 (1996) R11145.
15] N. Berdunov, G. Mariotto, I.V. Shvets, to be published, see also
http://www.nottingham.ac.uk/~ppznb/FeO_on_Rh.html;
www.nottingham.ac.uk/~ppznb/others.html.
http://
[
16] M.D. Segall, P.J.D. Lindan, M.J. Probert, C.J. Pickard, P.J. Hasnip,
S.J. Clark, M.C. Payne, J. Phys. Condens. Matter 14 (2002) 2717.
17] C. Noguera, J. Phys.: Condens. Matter 12 (2000) R367.
18] M. Todorova, K. Reuter, M. Scheffler, PRB 71 (2005) 195403.
19] E.D.L. Rienks, N. Nilius, H.-P. Rust, H.-J. Freund, PRB 71 (2005)
241404;
Acknowledgement
[
[
[
This work was supported by CRANN and Science
Foundation of Ireland (SFI) under contract 00/PI.1/C042.
N. Nilius, E.D.L. Rienks, H.-P. Rust, H.-J. Freund, PRL 95 (2005)
0
66101.
References
[
[
20] N.G. Condon, F.M. Leibsle, T. Parker, A.R. Lennie, D.J. Vaughan,
G. Thorton, PRB 55 (1997) 15885.
21] N.G. Condon, F.M. Leibsle, A.R. Lennie, P.W. Murray, D.J.
Vaughan, G. Thorton, PRL 75 (1995) 1961.
22] C. Barth, M. Reichling, Nature 414 (2001) 54.
23] R.A. Fellows, A.R. Lennie, A.W. Munz, D.J. Vaughan, G. Thornton,
American Mineralogist 84 (1999) 1384.
24] J. Schoiswohl, M. Sock, S. Esk, S. Surnev, M.G. Ramsey, F.P.
Netzer, G. Kresse, PRB 69 (2004) 155403.
25] G. Ketteler, W. Ranke, PRB 66 (2002) 033405.
26] A. Mannig, Z. Zhao, D. Rosenthal, K. Christmann, H. Hoster, H.
Rauscher, R.J. Behm, Surf. Sci. 576 (2005) 29.
27] F. Sedona, G.A. Rizzi, S. Agnoli, F.X. Llabr e´ s i Xamena, A.
Papageorgiou, D. Ostermann, M. Sambi, P. Finetti, K. Schierbaum,
G. Granozzi, J. Phys. Chem. B 109 (2005) 24411.
28] J.V. Barth, G. Costantini, K. Kern, Nature 437 (2005) 671.
29] J. Gustafson, A. Mikkelsen, M. Borg, E. Lundgren, L. Kohler, G.
Kresse, M. Schmid, P. Vagra, J. Yuhara, X. Torrelles, C. Quiros, J.N.
Andersen, PRL 92 (2004) 126102.
[
[
1] R. Skomski, J. Phys.: Condens. Matter 15 (2003) R841.
2] I. Zutic, J. Fabian, S. Das Sarma, Rev. Mod. Physics 76 (2004) 323;
C.T. Black, C.B. Murray, R.L. Sandstrom, S.H. Sun, Science 290
[
[
(2000) 1131.
[
[
[
[
3] A. Courty, A. Mermet, P.A. Albouy, E. Duval, M.P. Pileni, Nature
Mat. 4 (2005) 395.
4] H. Brune, M. Giovannini, K. Bromann, K. Kern, Nature 394 (1998)
[
451.
[
[
5] V. Repain, J.M. Berroir, S. Rousset, J. Lecoeur, Surf. Sci. 447 (2000)
L152.
6] M.H. Pan, H. Liu, J.Z. Wang, J.F. Jia, Q.K. Xue, J.L. Li, S. Qin,
U.M. Mirsaidov, X.R. Wang, J.T. Markert, Z.Y. Zhang, C.K. Shih,
Nano Lett. 5 (2005) 87.
[
[
[
[
7] S.K. Shaikhutdinov, R. Meyer, D. Lahav, M. Baumer, T. Kluner,
H.J. Freund, PRL 91 (2003) 076102.
8] S. Degen, C. Becker, K. Wandelt, Faraday Discussions 125 (2004)
[
[
3
43.
9] W.-C. Lin, C.-C. Kuo, M.-F. Luo, K.-J. Song, M.-T. Lin, APL 86
2005) 043105.
(