Heteroepitaxial growth of novel MoO3 nanostructures on Au(1 1 1)

Article abstract of DOI:10.1016/j.susc.2004.04.021

We have developed a synthetic procedure that yields novel nanocrystalline, islands of MoO₃ on Au(1 1 1). Careful control of the growth conditions yields monolayer islands with a rectangular unit cell that is aligned with the Au substrate. These structures are distinctly different than either bulk MoO₃ or than ramified, two-dimensional MoO₃ islands formed on Au(1 1 1) which were recently reported [J. Am. Chem. Soc. 125 (2003) 8059, Surf. Sci. 512 (2002) L353]. The atomic structure of these single-layer MoO₃ islands has been characterized by scanning tunnelling microscopy (STM), low energy electron diffraction (LEED), X-ray photoelectron spectroscopy (XPS), and Auger electron spectroscopy (AES). We discuss our synthetic method and important characteristics of the MoO₃ islands.

Full text of DOI:10.1016/j.susc.2004.04.021

Surface Science 559 (2004) L173–L179
www.elsevier.com/locate/susc
Surface Science Letters
Heteroepitaxial growth of novel MoO3 nanostructures
on Au(1 1 1)
a,b,
Monika M. Biener a,b, Cynthia M. Friend
*
a
Department of Chemistry, Harvard University, 12 Oxford Street, Cambridge, MA 02138, USA
b
Division of Engineering and Applied Sciences, Harvard University, 12 Oxford Street, Cambridge, MA 02138, USA
Received 10 February 2004; accepted for publication 16 April 2004
Available online 10 May 2004
Abstract
We have developed a synthetic procedure that yields novel nanocrystalline, islands of MoO3 on Au(1 1 1). Careful
control of the growth conditions yields monolayer islands with a rectangular unit cell that is aligned with the Au
substrate. These structures are distinctly different than either bulk MoO3 or than ramified, two-dimensional MoO3
islands formed on Au(1 1 1) which were recently reported [J. Am. Chem. Soc. 125 (2003) 8059, Surf. Sci. 512 (2002)
L353]. The atomic structure of these single-layer MoO3 islands has been characterized by scanning tunnelling
microscopy (STM), low energy electron diffraction (LEED), X-ray photoelectron spectroscopy (XPS), and Auger
electron spectroscopy (AES). We discuss our synthetic method and important characteristics of the MoO3 islands.
ꢀ 2004 Elsevier B.V. All rights reserved.
Keywords: Low energy electron diffraction (LEED); Scanning tunneling microscopy; Catalysis; Epitaxy; Growth; Surface chemical
reaction; Gold; Molybdenum oxides
1. Introduction
chemicals. Molybdenum trioxide is specifically
known to promote the partial oxidation of meth-
ane to formaldehyde [3–6].
Metal oxides are of considerable interest be-
cause of their electronic properties and their role
in several important technologies. One significant
application of metal oxides is in the area of het-
erogeneous catalysis, in particular conversion of
hydrocarbons to oxygenates that is of interest as
alternative fuels and as building blocks for other
Metal oxides have a wide range of chemical,
electronic and optical properties that depend on
stoichiometry and size scale. For example, bulk
MoO3 is a semiconductor with a bandgap of 2.8
eV; however, states are introduced in the gap when
point defects due to loss of oxygen are introduced
[7]. Catalytic activity is often attributed to the
presence of oxygen vacancies associated with edge
sites of the material [8].
* Corresponding author. Address: Department of Chemistry,
Harvard University, 12 Oxford Street, Cambridge, MA 02138,
USA. Tel.: +1-617-495-4052; fax: +1-617-496-8410.
Chemical, electronic, and optical properties of
materials can be influenced by changing the length
scale from macroscopic to nanoscale particles or
E-mail address: cfriend@deas.harvard.edu (C.M. Friend).
0039-6028/$ - see front matter ꢀ 2004 Elsevier B.V. All rights reserved.
doi:10.1016/j.susc.2004.04.021

L174
M.M. Biener, C.M. Friend / Surface Science 559 (2004) L173–L179
films. For example gold nanoparticles dispersed on
metal oxides are chemically active [2,9], whereas
bulk gold is inert. Molybdenum clusters on gold
exhibit a very low activity towards O2 and CO
[2,10], whereas bulk Mo dissociates CO and O2
below 300 K [11,12], providing another example.
Hence, it is important to develop methods for
synthesizing nanoscopic materials.
In this work, we describe a procedure that
yields novel nanocrystalline, monolayer islands of
MoO3 on Au(1 1 1). The possibility of creating
well-defined, single layer MoO3 nanocrystals is a
first step towards the development of materials
with unusual chemical and physical properties.
Molybdenum nanoclusters were deposited on
Au(1 1 1) surfaces at 450 K via chemical vapor
deposition (CVD) of Mo(CO)6. Nanocrystalline
MoO3 was prepared by subsequent exposure to
NO2 at 450 K, followed by annealing at 600 K.
All STM images were collected at room tem-
perature using commercial Pt0:8Ir0:2 tips (molecular
imaging). Scan dimensions were calibrated by
imaging the unit cell of the Au(1 1 1) surface. The
sample bias voltage was set between +0.1 and +2.8
V. All images are unfiltered and only background
corrected.
3. Results and discussion
2. Experimental
Preferential nucleation of three-dimensional
molybdenum nanoclusters at step edges was ob-
served after dosing 4 L of Mo(CO)6 onto the
All experiments were performed in an ultrahigh
vacuum (UHV) system with a base pressure of
4 · 10ꢀ10 Torr. The system is equipped with a
‘‘beetle-type’’ STM and commercial instrumenta-
tion for AES and LEED, described elsewhere
[13,14]. The sample was radiatively heated via a
tungsten filament located behind the sample. The
temperature was monitored by a chromel/alumel
thermocouple affixed to the sample holder. To
account for the temperature gradient of the sample
holder versus the crystal a calibration was per-
formed using a thermocouple directly mounted to
the crystal.
p
Au(1 1 1)-(22 ꢂ 3) surface maintained at 450 K
(Fig. 1). The apparent height of the Mo clusters is
in the range of 0.3–0.7 nm and their average
diameter is ꢁ10 nm. No carbon or oxygen was
detected by AES. The herringbone reconstruction
persists on the terraces of the Au surface following
deposition of the Mo clusters (data not shown).
The preferred nucleation of Mo clusters along
the step edges is in contrast to the deposition of
The Au was cleaned by cycles of Arþ sputtering
(1000 eV, ꢁ5 lA) at 300 K, followed by annealing
to 700 K for 10 min and 600 K for 60 min. This
procedure was repeated several times until no
contaminants were detected using AES. Following
this procedure, a LEED pattern characteristic of
p
the Au(1 1 1)-(22 ꢂ 3) reconstruction [15] was
observed. The ‘‘herringbone’’ reconstruction was
also observed by STM.
Mo(CO)6 (Alfa Aesar) was initially purified by
freeze–pump–thaw cycles, and NO2 (Matheson,
anhydrous grade) was used as received. Both re-
agents were introduced to the sample by backfill-
ing the chamber to 1 · 10ꢀ7 Torr. Gas lines were
evacuated before each dose. All exposures are
given in uncorrected ion gauge readings using
units of Langmuir (1 L ¼ 10ꢀ6 Torr s).
Fig. 1. STM image showing Mo particles formed by exposure
p
of the Au(1 1 1)-(22 ꢂ 3) surface to 4 L of Mo(CO)6 at 450 K.
Etch holes are indicated by arrows. The image was obtained at
300 K and corresponds to an area of 400 nm · 400 nm.

M.M. Biener, C.M. Friend / Surface Science 559 (2004) L173–L179
L175
Mo via physical vapor deposition (PVD), where
the clusters preferentially nucleate at the elbow
sites of the herringbone reconstruction as repor-
ted recently [16] and reproduced in our labora-
tory [17]. Our results are also different than
those reported previously for growth of Mo clus-
ters deposited using thermal decomposition of
Mo(CO)6, but using a different experimental con-
figuration [18]. This difference is attributed to the
strong sensitivity of the CVD process to temper-
ature and to CO induced mobility of Mo nano-
clusters [19].
sulfur coverages (<0.1 ML) completely inhibit the
decomposition of Mo(CO)6 on the Au(1 1 1) sur-
face, most likely by poisoning active sites. Specif-
ically, no Mo was detected after exposure to 4 L
Mo(CO)6 at 450 K by means of AES and STM. If
the Au surface were simply a source of thermal
energy, there would not be such a strong depen-
dence of the Mo deposition rate on the presence of
a small amount of impurity. It is known that the
most stable bonding sites for sulfur are located at
the step edges of Au(1 1 1) [21]. This indicates that
the active sites for Mo(CO)6 decomposition are
related to step edges of the Au surface. We were,
however, not able to identify the active sites by
STM.
The Mo nanoclusters probably also contain Au
based on changes in the surface morphology ob-
served in STM following CVD of the Mo. Spe-
cifically, the originally straight step edges become
irregular. Etch holes appear along the step edges
between neighbouring molybdenum clusters (Fig.
1). These observations are also consistent with the
appearance of nanoscopic holes as observed in our
previous work [19] when Mo is transported from
terraces to step edges by CO exposure.
Oxidation of the Mo nanoclusters using expo-
sure to NO2 at 450 K leads to spreading into a
two-dimensional layer and full oxidation to MoO3,
in qualitative agreement with previous studies
[1,2]. The predominant oxidation state was deter-
mined to be Mo6þ using X-ray photoelectron
spectroscopy (data not shown), as described in
detail elsewhere [22].
Nanocrystalline MoO3 islands that are one
layer high and ordered relative to the underlying
gold surface are formed using our iterative dosing
procedure described below (Figs. 2 and 3). The
structure of the observed single-layer MoO3 differs
from the bilayer structure of bulk MoO3 [23] and
can be explained in terms of an ordered, two-
dimensional array of interacting MoO3 entities
based on a combination of experimental data and
associated density functional theory calculations,
described elsewhere [24]. The apparent height of
the islands is ꢁ0.5 nm, in contrast to 1.38 nm as
the height of a bulk bilayer unit cell, which is
consistent with the single layer structure described
above.
The strong sensitivity of the CVD process to
surface temperature is illustrated by the fact that
the deposition rate decreases to an immeasurable
rate at a surface temperature of 400 K. Specifi-
cally, no Mo was detected by means of STM after
exposure of up to 100 L at 10ꢀ7 Torr of Mo(CO)6
to the surface maintained at 400 K. Furthermore,
no extraneous material or other changes were
observed. This decrease in deposition rate is con-
sistent with the fact that the CVD process is
thermally activated. At minimum, dissociation of
one Mo–CO bond is required to deposit Mo from
Mo(CO)6, which has an energy cost of ꢁ40 kcal/
mol [20]. Surprisingly, the amount of Mo observed
using STM after exposure of 4 L of Mo(CO)6 to
the Au surface maintained at 500 K was less than
20% of that observed at 450 K using the same flux
and integrated dose. This observation is consistent
with precursor-mediated decomposition: at higher
temperatures the surface lifetime of the Mo(CO)6
precursor decreases due to an increased desorption
rate. The lifetime, s, of the precursor would de-
crease by a factor of 5 upon increasing the surface
temperature from 450 to 500 K. This estimate as-
sumes a Mo(CO)6 desorption barrier of 14 kcal/
mol [18] and that the pre-exponential factor for
desorption, A, is the same at the two temperatures.
The ratio of lifetimes at the two temperatures is
calculated using the Frenkel equation: s ¼ 1=Aꢃ
expðEdes=RTsÞ, where Edes is the activation barrier
and Ts is the surface temperature. The pre-expo-
nential factors cancel in the ratio. This is in
agreement with our observation that the Mo cov-
erage decreased to ꢁ20%.
The Au surface itself promotes the decomposi-
tion of Mo(CO)6 based on the fact that small

L176
M.M. Biener, C.M. Friend / Surface Science 559 (2004) L173–L179
Fig. 2. STM image showing nanocrystalline MoO3 islands on Au(1 1 1) after alternate exposure to 1 L Mo(CO)6 and 10 L NO2 at 450
K followed by annealing to 600 K for 1 min after every four cycles of dosing for a total of 16 cycles: (a) alignment of the island edges
(dotted lines) with the Æ1 1 )2æ directions of the substrate: a magnification of a domain wall is shown in the inset (28 nm · 18 nm); (b)
image is differentiated to enhance the contrast and show the orientation of the substrate. The Æ1 1 )2æ directions run parallel to the
herringbone dislocation lines; one of them is indicated in the figure. The total coverage, as estimated by AES, is ꢁ0.3 ML, however, the
local coverage is strongly dependent on the substrate step density. The image was obtained at 300 K and corresponds to an area of 120
nm · 120 nm.
The MoO3 islands nucleate preferentially along
step-edges, and exhibit well-defined edges with
kinks and domain walls (Fig. 2). The islands are
aligned with the Au(1 1 1) surface, based on the
observation of a well-ordered LEED pattern (see
Fig. 3c and discussion below). STM reveals that
the island edges run parallel to the Æ1 1 )2æ direc-
tions of the substrate, if the shape of the islands is
not dictated by the step edges of the substrate.
STM data further reveal a rectangular unit cell
with dimensions of 0.50 · 0.57 nm2 10%, consis-
tent with the observation of a well-defined c(2 · 4)
LEED pattern (Fig. 3). Note that the observation
of the c(4 · 2) LEED pattern, as shown in (Fig. 3c),
implies the existence of rotational domains.
Occasionally, domain walls (Fig. 2a, inset) were
observed by STM. The growth of the MoO3 is-
lands away from the step edge is a direct conse-
quence of the fact that the Mo clusters have
nucleated at the step edge prior to oxidation.
Nanocrystalline MoO3 islands with the same unit
cell form on terraces when Mo clusters, deposited
using physical vapor deposition, are oxidized [17].
The alignment of those islands with the Au sub-
strate is more pronounced, compared to the is-
lands shown in (Fig. 2), as the island shape is not
dictated by the Au step edges.
The crystallinity of the MoO3 islands depends
on the conditions used for oxidation due to kinetic
control of the oxidation and growth. Small Mo
clusters (diameter 6 5 nm) are required as starting
material in order to form crystalline structures.
Disordered molybdenum oxide particles, which
are only partially oxidized to Mo + 6 and not
completely spread into small 2D islands, form
upon oxidation of larger Mo clusters (Fig. 4). In
order to grow large MoO3 islands, repeated cycles
of Mo deposition in small amounts and sub-
sequent oxidation are used (Fig. 2). For example,
the MoO3 islands shown in Fig. 2 are formed
by alternate exposure of the surface to 1 L of
Mo(CO)6 and 10 L NO2 at 450 K followed by
annealing to 600 K for 1 min after every four cy-
cles of dosing for a total of 16 cycles. This proce-
dure yields well-ordered islands with an average
size of 600 nm2––large enough to observe a clear
LEED pattern.
In order to study the growth mechanism of the
MoO3 islands, Mo(CO)6 was dosed at 450 K to a
MoO3-covered Au(1 1 1) surface. STM images

M.M. Biener, C.M. Friend / Surface Science 559 (2004) L173–L179
L177
Fig. 3. Identification of the c(4 · 2) unit cell of the MoO3 islands on Au(1 1 1) prepared by alternate exposure of the surface to 1 L
Mo(CO)6 and 10 L NO2 at 450 K followed by annealing to 600 K for 1 min after every four cycles of dosing for a total of 16 cycles: (a)
high resolution STM image (50 nm · 60 nm) showing a rectangular unit cell with a ¼ 0:50 nm and b ¼ 0:57 nm 10%, obtained at 300
K; (b) schematic of a c(4 · 2) superlattice on a hexagonal substrate, corresponding to unit cell dimensions of a ¼ 0:499 nm, b ¼ 0:576
nm for a Au(1 1 1) surface with the closest-packed rows of the superlattice running in the [1 1 )2] direction; (c) LEED pattern obtained
at 300 K showing a c(4 · 2) structure (part of the picture is blocked out by the sample holder). Straight arrows indicate the basis of the
hexagonal unit cell of the Au substrate, dotted arrows indicate the basis of one rotational domain (basis of the other two domains not
shown) of the c(4 · 2) superlattice.
obtained at 300 K reveal that Mo clusters nucleate
exclusively at Au step edges (Fig. 5). Mo de-
position at edges of the MoO3 islands was not
observed. This implies that the oxidation of
molybdenum clusters leads to mass transport from
the step edges to the islands. Mobile MoO3 species
are created at the step edges and diffuse along the
surface until they hit the edge of a MoO3 island.
Volatile (MoO3)n (n ¼ 3, 4, 5) species have been
previously detected by mass spectrometric studies
during sublimation of MoO3 bulk oxide [25]. The
compact shape of the islands indicates that the
aggregation of these mobile clusters is reversible
[26]. Detachment from the island or diffusion
along the edge of the islands is possible at the
chosen temperature (450 K) and influences island
shape and size. Larger crystalline islands grow at
the expense of smaller islands. Small islands have
a larger perimeter to area ratio than large islands
and are thus thermodynamically less favorable
(Ostwald ripening). In contrast, a fractal island
shape can be expected if the substrate temperature
is too low to activate detachment of atoms or
clusters from the islands and thus aggregation is
irreversible [26].
Gold as a noble metal is ideally suited as inert
support for the oxidation of metals since neither
O2 nor NO2 decompose under UHV conditions on

L178
M.M. Biener, C.M. Friend / Surface Science 559 (2004) L173–L179
facilitates the diffusion of the MoO3 species.
However, the interaction is strong enough to
influence the structure and orientation of the
MoO3 islands. As Mo has a higher surface free
energy (2.88 J mꢀ2) than gold (1.62 J mꢀ2) [30] the
formation of 3D molybdenum nanoclusters can be
expected. The spreading of the 3D metal clusters
during oxidation is consistent with the low free
surface energy of MoO3 (5–7 · 10ꢀ2 J mꢀ2) [31].
4. Conclusion
We are able to grow two-dimensional compact
MoO3 islands on Au(1 1 1) by carefully controlling
the growth kinetics. A c(4 · 2) unit cell was deter-
mined by STM and LEED. The location of the
deposited metal clusters dictates the initial nucle-
ation sites for the metal oxide islands. Repeated
cycles of Mo cluster deposition and oxidation in
small increments are necessary to create mobile
MoO3 species that are responsible for the mass
transport and thus for the island growth. The
possibility to create well-defined nanocrystalline
MoO3 islands will enable the study of interesting
reactions that are relevant in the field of catalysis
and novel sensors. Furthermore, we expect that
this methodology can be generalized to the growth
of other nanoscopic metal oxides.
Fig. 4. STM image showing partially oxidized Mo oxide islands
formed by exposure of 4 L of Mo(CO)6 and 40 L of NO2 at 450
K. The image was obtained at 300 K and corresponds to an
area of 200 nm · 200 nm.
Acknowledgements
We gratefully acknowledge support of this
work by the National Science Foundation under
support for the Harvard Nanoscale Science and
Engineering Center, grant no. PHY-011-7795, and
the Department of Energy, Basic Energy Sciences,
FG02-84-ER13289.
Fig. 5. STM image showing a MoO3-covered Au(1 1 1) surface
after exposure to 1 L of Mo(CO)6 at 450 K. The figure illus-
trates that Mo clusters (B) nucleate exclusively at Au step edges
and not at the edges of the MoO3 islands (A). The image was
obtained at 300 K and corresponds to an area of 300 nm · 300
nm.
References
Au(1 1 1) [27,28]. Selective oxidation of metal
clusters deposited on Au(1 1 1) is, thus, possible by
exposure to an oxidant. Recently the inertness of
Au(1 1 1) was utilized to grow V2O3 films by
evaporation of vanadium in an oxygen atmosphere
[29]. The weak interaction of Au with oxygen also
[1] Z. Song, T. Cai, Z. Chang, G. Liu, J.A. Rodriguez, J.
Hrbek, J. Am. Chem. Soc. 125 (2003) 8059.
[2] Z.P. Chang, Z. Song, G. Liu, J.A. Rodriguez, J. Hrbek,
Surf. Sci. 512 (2002) L353.
[3] P.J. Spencer, United States Patent 4,607,127 (1986).
[4] N.D. Spencer, C.J. Pereira, AIChE J. 33 (1987) 1808.

M.M. Biener, C.M. Friend / Surface Science 559 (2004) L173–L179
L179
[5] N.D. Spencer, J. Catal. 109 (1988) 187.
[6] M.A. Banares, N.D. Spencer, M.D. Jones, I.E. Wachs, J.
Catal. 146 (1994) 204.
[19] A. Chan, B. Richter, C. Friend, M. Aziz, J. Hrbek, Z.
Song, Nano Letters (2003).
[20] K.E. Lewis, D.M. Golden, G.P. Smith, J. Am. Chem. Soc.
106 (1984) 3912.
[7] M. Chen, C.M. Friend, E. Kaxiras, JACS (2001).
[8] J. Haber, E. Lalik, Catal. Today 33 (1997) 119.
[9] M. Valden, X. Lai, D.W. Goodman, Science 281 (1998)
1647.
[21] C. Vericat, J.N. Andersen, M.E. Vela, R.C. Salvarezza,
J. Phys. Chem. B 104 (2000) 302.
[22] M.M. Biener, J. Biener, X. Deng, C.M. Friend, in press.
[23] E. Itoh, K. Hayakawa, S. Oishi, J. Phys.: Condens. Matter
13 (2001) 6853.
[10] J.A. Rodriguez, J. Dvorak, T. Jirsak, J. Hrbek, Surf. Sci.
490 (2001) 315.
[11] A.J. Jaworowski, M. Smedh, M. Borg, A. Sandell, A.
Beutler, S.L. Sorensen, E. Lundgreen, J.N. Andersen, Surf.
Sci. 492 (2001) 185.
[24] S.Y. Quek, M.M. Biener, J. Biener, C.M. Friend, E.
Kaxiras, in press.
[25] G. Demaria, R.P. Burns, J. Drowart, M.G. Inghram,
J. Chem. Phys. 32 (1960) 1373.
[12] G.H. Smudde, P.C. Stair, Surf. Sci. 317 (1994) 65.
[13] P.G. Clark, C.M. Friend, J. Chem. Phys. 111 (1999) 6991.
[14] M. Chen, P.G. Clark, T. Mueller, C.M. Friend, E. Kaxiras,
Phys. Rev. B 60 (1999) 11783.
[26] A.-L. Barabasi, H.E. Stanley, Fractal Concepts in Sur-
face Growth, Cambridge University Press, Cambridge,
1995.
[15] M.V.R. Murty, T. Curcic, A. Judy, B.H. Cooper, Phys.
Rev. B 60 (1999) 16956.
[27] J. Wang, E. Koehl, J. Phys. Chem. A 102 (1998) 8573.
[28] N.D.S. Canning, D. Outka, R.J. Madix, Surf. Sci. 141
(1984) 240.
[16] S. Helveg, J.V. Lauritsen, E. Laegsgaard, I. Stensgaard,
J.K. Norskov, B.S. Clausen, H. Topsoe, F. Besenbacher,
Phys. Rev. Lett. 84 (2000) 951.
[29] A.C. Dupuis, M. Abu Haija, B. Richter, H. Kuhlenbeck,
H.J. Freund, Surf. Sci. 539 (2003) 99.
[17] M.M. Biener, J. Biener, R. Schalek, C.M. Friend, in press.
[18] Z. Song, T.H. Cai, J.A. Rodriguez, J. Hrbek, A.S.Y. Chan,
C.M. Friend, J. Phys. Chem. B 107 (2003) 1036.
[30] L.Z. Mezey, J. Giber, Jpn. J. Appl. Phys. 21 (1982)
1569.
[31] C.B. Wang, C.T.H.E. Wachs, Langmuir 15 (1999) 1223.