IOP PUBLISHING NANOTECHNOLOGY
Nanotechnology 19 (2008) 185601 (8pp) doi:10.1088/0957-4484/19/18/185601
The role of substrate surface alteration in
the fabrication of vertically aligned CdTe
nanowires
S Neretina
1,4
, R A Hughes
2,5
, G A Devenyi
1,2
, N V Sochinskii
3
,
J S Preston
1,2
and P Mascher
1
1
Department of Engineering Physics and Centre for Emerging Device Technologies,
McMaster University, Hamilton, ON, L8S 4L7, Canada
2
Brockhouse Institute for Materials Research, McMaster University, Hamilton,
ON, L8S 4M1, Canada
3
Instituto de Microelectronica de Madrid CNM–CSIC, Madrid, 28760, Spain
E-mail: hughesr@mcmaster.ca
Received 16 January 2008, in final form 27 February 2008
Published 1 April 2008
Online at stacks.iop.org/Nano/19/185601
Abstract
Previously we have described the deposition of vertically aligned wurtzite CdTe nanowires
derived from an unusual catalytically driven growth mode. This growth mode could only
proceed when the surface of the substrate was corrupted with an alcohol layer, although the role
of the corruption was not fully understood. Here, we present a study detailing the remarkable
role that this substrate surface alteration plays in the development of CdTe nanowires; it
dramatically improves the size uniformity and largely eliminates lateral growth. These effects
are demonstrated to arise from the altered surface’s ability to limit Ostwald ripening of the
catalytic seed material and by providing a surface unable to promote the epitaxial relationship
needed to sustain a lateral growth mode. The axial growth of the CdTe nanowires is found to be
exclusively driven through the direct impingement of adatoms onto the catalytic seeds leading
to a self-limiting wire height associated with the sublimation of material from the sidewall
facets. The work presented furthers the development of the mechanisms needed to promote high
quality substrate-based vertically aligned CdTe nanowires. With our present understanding of
the growth mechanism being a combination of selective area epitaxy and a catalytically driven
vapour–liquid–solid growth mode, these results also raise the intriguing possibility of
employing this growth mode in other material systems in an effort to produce superior
nanowires.
1. Introduction
The growth of semiconductor nanowires has now become com-
monplace with virtually all of the prominent semiconductors
now synthesized [1–3]. The vapour–liquid–solid (VLS) mech-
anism [4] has proved itself as the most versatile means of pro-
ducing vertically aligned substrate-based nanowires. In this
mechanism, gaseous reactants supersaturate a catalytic liquid
located on a solid substrate. Through the symmetry break-
ing offered by the liquid–solid interface it becomes possible to
4
Present address: Georgia Institute of Technology, School of Chemistry and
Biochemistry, 901 Atlantic Drive, Atlanta, GA 30332-0400, USA.
5
Author to whom any correspondence should be addressed.
nucleate a single crystal nanowire structure having an epitaxial
relationship with the underlying substrate. Growth will per-
sist as long as the dissolution rate of material into the catalyst
matches the extrusion of solid material at the liquid/solid inter-
face. This bottom-up approach gives rise to a one-dimensional
nanowire structure capped with a catalytic seed, where the size
of the seed determines the diameter of the nanowire structure.
Most of the commonly used thin film growth techniques
have been used to generate the vapour needed to initiate the
VLS process. Methods where the vapour flow is highly
directional have made it apparent that the adatoms can arrive at
the liquid/solid growth front either through direct impingement
on the catalytic seed or through a random walk up the nanowire
0957-4484/08/185601+08$30.00 © 2008 IOP Publishing Ltd Printed in the UK1

Nanotechnology 19 (2008) 185601 S Neretina et al
sidewalls from the substrate [5]. In the majority of reported
studies, the latter process is the dominant mechanism driving
nanowire growth. It is characterized by a nanowire height
distribution that is inversely proportional to the diameter of the
catalytic seed [6]. This is a simple consequence of the fact that
the collection area around the nanowire, from which it derives
material, is nearly independent of the nanowire diameter. Thus,
with the same number of atoms arriving at each nanowire,
small diameter nanowires will grow faster as they require less
volume to achieve a given height. The expected nanowire size
distribution has been modelled extensively for this scenario as
well as for the case of direct impingement [5–13].
An alternate route to the production of vertically aligned
substrate-based nanowires relies upon selective area epitaxy.
In this process, a suitable crystalline substrate is partially
covered with material that discourages epitaxy, typically an
amorphous layer. Then it is frequently possible to identify
growth conditions under which the semiconductor is able to
nucleate on the substrate, but not on the overlayer, leading to
the patterned growth of a film. The ability to achieve selective
area epitaxy has been promoted as a means of enabling self-
assembly [14]. Recently, it has also been promoted as a
means of producing semiconductor nanowires. For this case,
the overlayer would be deposited over the entire surface of a
substrate that is capable of sustaining semiconductor growth.
Then, through the use of electron beam lithography, nanometre
scale openings are created through the film exposing the
substrate below. Through the process of selective area
epitaxy, semiconductor deposition would occur only where the
substrate was exposed. Over time, isolated nanowire structures
emerge from the openings, with the surrounding film inhibiting
lateral growth. This catalyst-free method has been used to
fabricate ordered arrays of GaAs [15], InGaAs [16], InP [17]
and ZnO [18] nanowires.
Previously we demonstrated the production of substrate-
based vertically aligned CdTe nanowires derived from a newly
developed catalytically driven process that utilizes the VLS
mechanism [19]. Like many semiconducting nanowires,
the one-dimensional CdTe structures were highly faceted
and shared an epitaxial relationship with the underlying
substrate. There were, however, a number of unique properties
that differentiated these structures from other semiconductor
systems, both in terms of the nanowire attributes and the
growth mode used to fabricate them.
The produced CdTe nanowires showed a remarkably high
degree of size uniformity with nearly constant heights and
diameters over large substrate areas. Their diameter was
also constant over the length of the nanowire, showing no
indication of tapering. Growth dynamics, however, limited
the wire height to approximately 300 nm, making the term
nanorod a more apt description of the structures. Of
significance was the fact that the nanowires were isolated,
with a complete absence of the two-dimensional planar growth
that is often simultaneously observed in other semiconducting
systems [6, 20]. This was quite fortuitous as the nanowires
grew at a slow rate. From a structural perspective the nanowires
were unusual in the sense that they formed in the wurtzite phase
instead of the zinc blende crystal structure normally associated
with the bulk material.
The most unusual quality relating to these nanowires was
the growth pathway needed to initiate their formation. In order
to effectively nucleate nanowires it was not only necessary to
find an appropriate catalytic seed material, but also to carry
out a substrate surface alteration. The procedure began by first
exposing a pristine (0001) sapphire surface to a low vapour
pressure alcohol (either polyvinyl alcohol or terpineol). A thin
layer of bismuth was then deposited on the corrupted surface
using pulsed laser deposition (PLD). When heated, the bismuth
layer dewetted yielding nanoscale catalytic seeds which led
to the effective nucleation of CdTe nanowires. Removal of
either the alcohol or bismuth from the process resulted in a
bare substrate.
The interplay between the bismuth and alcohol was not
entirely clear in the previous experiments, but it was clear that
the presence of the alcohol stabilized the initial bismuth seeds
prior to growth. In the absence of the alcohol corruption, the
bismuth seeds evaporated as the substrate was heated to the
CdTe nanowire growth temperature. It was concluded that
the alcohol, or the products of its decomposition, persisted
at the nanowire growth temperature and reduced the mobility
of atoms travelling along the surface. As a result, once a
bismuth dot formed it became more difficult for it to deteriorate
through the lateral dispersal of individual bismuth atoms. Thus
it was possible, through this surface corruption, to increase the
bismuth seed’s ability to withstand the elevated temperatures
required for nanowire formation. It was also observed that
the stability of the seed particles was greatly enhanced once
they were exposed to the flux of cadmium and tellurium
atoms needed to initiate nanowire growth. Based on the
hypothesis that this stabilization arose from an evolution in
the seed’s composition after it was exposed to tellurium, we
have now grown CdTe nanowires where the catalytic material
was derived from bismuth-telluride (Bi
2
Te
3
). The increased
stability of Bi
2
Te
3
rendered the alcohol pre-treatment of the
substrate’s surface unnecessary. Given that the steady-state
composition of the seed is likely the same for both scenarios,
we are now able to provide a comparison of wires grown on
a pristine versus a corrupted surface. This comparison shows
that the surface corruption plays a far more intricate role than
first thought; it is this role that will be the primary focus of this
report.
2. Experimental procedures
The experimental results and procedures presented here will
focus on those nanowires derived from Bi
2
Te
3
catalytic seeds
deposited on pristine (0001) sapphire substrates. The results
obtained will then be compared to the CdTe nanowires,
described elsewhere [19], obtained using bismuth catalytic
seeds deposited on alcohol-altered (0001) sapphire substrates.
The Bi
2
Te
3
seeds were prepared using the PLD process
(GSI Lumonics IPEX-848 excimer laser, λ = 248 nm, laser
energy density = 2 J cm
−2
, laser spot size = 1.2 × 1.2 mm
2
).
The target used was prepared in-house from commercially
available Bi
2
Te
3
pieces (99.999% purity). These pieces were
melted in a cylindrical graphite mould that was machined to
sizes able to yield a 1 inch target weighing approximately 10 g.
2