IOP PUBLISHING
NANOTECHNOLOGY
Nanotechnology 19 (2008) 455608 (7pp)
doi:10.1088/0957-4484/19/45/455608
Growth and process conditions of aligned
and patternable films of iron(III) oxide
nanowires by thermal oxidation of iron
P Hiralal1,5 , H E Unalan1 , K G U Wijayantha2 , A Kursumovic3,
D Jefferson4, J L MacManus-Driscoll3 and G A J Amaratunga1
1
Electrical Engineering Division, Department of Engineering, University of Cambridge,
9 J J Thomson Avenue, Cambridge CB3 0FA, UK
2
Department of Chemistry, University of Loughborough, Loughborough LE11 3TU, UK
3
Department of Materials Science and Metallurgy, University of Cambridge,
Pembroke Street, Cambridge CB2 3QZ, UK
4
Department of Chemistry, University of Cambridge, Lensfield Road,
Cambridge CB2 1EW, UK
E-mail: [email protected]
Received 9 July 2008, in final form 10 September 2008
Published 9 October 2008
Online at stacks.iop.org/Nano/19/455608
Abstract
A simple, catalyst-free growth method for vertically aligned, highly crystalline iron oxide
(α -Fe2 O3 ) wires and needles is reported. Wires are grown by the thermal oxidation of iron foils.
Growth properties are studied as a function of temperature, growth time and oxygen partial
pressure. The size, morphology and density of the nanostructures can be controlled by varying
growth temperature and time. Oxygen partial pressure shows no effect on the morphology of
resulting nanostructures, although the oxide thickness increases with oxygen partial pressure.
Additionally, by using sputtered iron films, the possibility of growth and patterning on a range
of different substrates is demonstrated. Growth conditions can be adapted to less tolerant
substrates by using lower temperatures and longer growth time. The results provide some
insight into the mechanism of growth.
(Some figures in this article are in colour only in the electronic version)
In particular, the use of vertically aligned nanowires and
nanotubes is currently being explored for field emission [2],
photovoltaics [3], batteries [4], light-emitting diodes [5] and
sensing [6] applications, amongst others.
Hematite (α -Fe2 O3 ), an n-type semiconductor, has a
bandgap of about 2.1 eV [7], is environmentally friendly,
nontoxic (in bulk form), corrosion-resistant and low cost. The
use of α -Fe2 O3 has been demonstrated as a photoanode for the
photoassisted electrolysis of water in tandem cells [8], an active
component of gas sensors [9], a photocatalyst [10], magnetic
coating for data storage devices [11] and as electrodes in
Li-ion batteries [4]. Lindgren et al showed that hematite
nanorod films yielded an enhanced photoconversion efficiency
as compared to thin films [12].
A range of methods exist for the fabrication of nanowire
arrays. Growth can be achieved by using precursors from
1. Introduction
A large number of quasi-one-dimensional (1D) semiconductor
materials are currently under extensive investigation (ZnO, Si,
GaN, In2 O3 and CuO, for instance) [1] on the promise of
enhanced electronic and optical properties for future nanoscale
electronics. 1D nanostructures provide not only a framework
for enhancing devices by increasing surface area, but also a
means by which to probe the fundamental quantum properties
of semiconductors. An enhanced performance compared to
their bulk counterparts has been shown in several cases. Even
though complete control of positioning and orientation of
individual wires may be some distance away, an ensemble of
aligned nanowires of controlled size on a substrate is sufficient
for the realization of certain electronic and photonic devices.
5 Author to whom any correspondence should be addressed.
0957-4484/08/455608+07$30.00
1
© 2008 IOP Publishing Ltd Printed in the UK
Nanotechnology 19 (2008) 455608
P Hiralal et al
the liquid phase (hydrothermal growth, electrodeposition),
from the gas phase (chemical vapor deposition) and from the
solid phase (thermal oxidation). In general, low temperature
methods (liquid phase) result in lower quality wires, with
lower aspect ratios. However, high temperature methods, such
as chemical vapor deposition (CVD), usually use catalysts
(normally Au) which may alter the properties of the wire
and render them incompatible with current CMOS processes.
There is also a limited number of substrates suited for the
high temperatures required (∼600–1100 ◦ C). To the best of
our knowledge, hematite nanowire growth from the gas phase
has not been reported, and reports from the liquid phase have
been limited. Growth from solid precursors, in which NWs are
formed as a result of the oxidation or sulfurization of metals at
high temperature, is less commonly explored, despite resulting
in the first observation of artificial 1D crystal growth [13].
Reports on the growth of iron oxide ‘whiskers’ by thermal
oxidation of iron are few, and most date back to the 1950s
and 1960s [14]. It is believed that whiskers grow from the
tip while reactants (metal) diffuse via the surface from the
metal substrate. However, the exact mechanism of growth is far
from clear. More recently, with the advent of nanotechnology,
various groups have revisited the spontaneous growth of oxide
NWs by the thermal oxidation of metallic substrates (Zn [15],
Fe [16] and Cu [17]), although a detailed systematic study is
lacking.
In this work, a detailed parametric study of the growth of
hematite nanowires and needles from the simple oxidation of
iron under an oxygen atmosphere is reported. The effects of
growth temperature, time and O2 partial pressure are presented.
This approach leads to a dense array of α -Fe2 O3 wires with a
very high aspect ratio growing directly on the iron substrate.
Growth temperature can be tuned from 300–900 ◦ C, resulting
in wire densities from ∼1.6 to 200 μm−2 . Wires are grown
both from iron foils and from sputtered iron films deposited
on various substrates, which is more versatile for device
fabrication.
The method is simple, has no foreign elements involved
(e.g.catalysts), is patternable and very easily scalable.
Table 1. Summary of growth parameters presented on iron foil.
Temp. (◦ C)
(time = 10 h,
flow = 30 sccm,
O2 = 1%)
Time (h)
(flow = 30 sccm,
O2 = 1%,
temp = 530 ◦ C)
%O2
(flow = 30 sccm,
time = 10 h,
temp = 530 ◦ C)
175
300
450
620
1
4
10
1
15
30
45
60
100
system allowed to cool down to room temperature under argon.
The resulting foils had a deep red color which varies with O2
partial pressure and temperature. Table 1 above summarizes
the growth conditions tested.
For the growth of hematite NWs onto different materials,
iron was sputtered onto cleaned substrates using radiofrequency-magnetron sputtering at a power of 100 W until the
desired thickness was achieved. Oxide growth on the sputtered
films was then carried out in the same manner as above. A
similar procedure was conducted for patterned growth, where
the patterns were set photolithographically before sputtering.
For the lower temperature range (∼300 ◦ C) it was possible
to grow NWs by placing the sample on a hotplate, further
demonstrating the simplicity of this method.
3. Results
The morphology and size of the nanostructures was
investigated by field emission scanning electron microscopy
(FESEM) (Philips XL30, operated at 5 kV). Figure 1(a)
shows a micrograph of a typical growth, at a substrate
temperature of 620 ◦ C. Resulting nanostructures can be
divided into two types, thin long nanowires with lengths
∼8–10 μm and diameters under 50 nm (sometimes below
10 nm) and needle-shaped elongated platelets which are
wider at the base than the tip. High resolution transmission
electron microscopy studies (HRTEM) (JEOL 3011 running
at 300 kV) confirmed the hematite phase. The thinner
wires (∼<50 nm) and not the needles are shown to be a
bicrystal, split in the middle with a twin plane in the middle
along the entire longitudinal axis (figure 1(b)). A similar
bicrystalline structure has been observed before for the growth
of CuO wires by thermal oxidation [17]. Annealing twins are
known to form as a consequence of growth accidents during
recrystallization of deformed cubic close packed metals. Peak
positions of the Raman spectrum in figure 1(c) (Renishaw
In vivo Raman microscope with a laser line of 785 nm)
match perfectly with previous reports of hematite [7]. X-ray
diffraction (XRD) patterns were taken with a Philips PW1730
diffractometer (Cu Kα radiation 1.540 60 Å). Figure 1(d)
shows the resulting pattern. The peaks match the hematite
phase ( R 3̄c group), which crystallizes in a rhombohedral
crystal structure with lattice parameters a = b = 5.03 Å
and c = 13.73 Å. A strong preferential orientation is
observed in the (1 1 0) direction, in agreement with other
reports of thermal oxidation of iron [16]. The (1 1 0)
2. Growth of wires
The system for growth consists of a 1 inch (approx. 2.54 cm)
diameter ceramic tube furnace with a temperature controller
(carbolite-MTF12/38B), a gas flow control system with oxygen
and argon on the input and a bubbler at the output. Iron foils
(Advent, 0.25 mm thick, temper annealed) were cut into flat
∼5 × 5 mm2 , degreased and cleaned by ultrasonication in
acetone and isopropanol for 15 min. The foil was placed in
the tube furnace on a ceramic boat at 6 cm from the furnace
edge. NW growth was possible in a fairly large window of
temperature (300–700 ◦ C) and gas conditions from air to 1%
O2 . The furnace was flushed with 150 sccm of Ar for 10 min
to remove the air already present in the tube. The flow was
then adjusted to the desired setting, typically with a total flow
(argon + oxygen) of 50 sccm, and the furnace heated to the
desired temperature at a rate of 1600 ◦ C h−1 . Once the growth
was complete, the oxygen supply was switched off and the
2
Nanotechnology 19 (2008) 455608
P Hiralal et al
Figure 1. Structural and morphological characteristics of hematite NWs grown at 620 ◦ C. (a) SEM and (b) TEM images where the
bicrystalline nature and amorphous edges can be observed. (c) Raman spectrum and (d) XRD of the same sample. The phase is confirmed to
be hematite and there is a strong preferential orientation in the (1 1 0) direction.
peak for Fe2 O3 matches with the (3 1 1) peak for Fe3 O4 .
However, any ambiguity is cleared by the rest of the peaks and
the Raman spectrum. Some peaks from the iron substrate are
also present.
be used as a first-order estimation of oxidation depth. Both
straight wires and tapered needles occur at all temperatures,
with the largest fraction of NWs occurring at around 400–
500 ◦ C. At 620 ◦ C the fraction of needles with a large base
increases considerably.
3.1. The effect of temperature
3.2. The effect of O2 partial pressure
Figure 2 shows the effect of temperature on the growth of NWs.
A series of samples were prepared, and heated for 10 h at
30 sccm flow of Ar with 1% O2 . Substrate temperature was
set to 175, 300, 450 and 620 ◦ C. Oxidation occurs readily at
175 ◦ C, but nanostructured features become noticeable only
at temperatures above 300 ◦ C. SEM micrographs also show
the strong effect of temperature on nanowire density and
size distribution. The histograms in figure 2 show the size
distributions of the nanowires. The diameter of the needles
is measured at half-height. The modal size increases from
25 nm at 300 ◦ C to 90 nm at 620 ◦ C. Both the average size
and density of the NWs increased with temperature. This is
further supported by normalized XRD spectra in the insets.
The relative sizes of the hematite peaks with respect to the iron
background increase, owing to the greater degree of oxidation
with temperature. Note that at 620 ◦ C all the hematite peaks
become more intense. This is particularly so for the (1 1 0)
peak, and hence the remaining peaks appear smaller as a result
of normalization. In fact, given that the relative increase in
intensity of the (1 1 0) peak is greater than all others, the
relative size of the (1 1 0) with respect to the other peaks can
A set of samples grown under different O2 partial pressures, at
a temperature of 530 ◦ C, is shown in figure 3. A growth time of
10 h and a flow rate of 35 sccm was used for all samples. SEM
micrographs show no obvious difference in morphology and
size. Size distribution plots show an approximately constant
distribution over the whole PO2 range, with a modal size of 15–
20 nm. However, the XRD patterns do show a varying degree
of oxidation up to 45% O2 , beyond which the difference is
negligible. Similar to the temperature study, the relative size
of the (1 1 0) hematite peak and the first Fe peak provides a
qualitative measure of the degree of oxidation as a result of
the penetration depth of x-rays. O2 partial pressure provides
another route to control the degree of oxidation.
3.3. The effect of growth time
Figure 4 shows the effect of growth time on the nanowires.
This set of samples was grown at 530 ◦ C and 1% oxygen at
30 sccm flow. In this case, instead of the furnace cooling
naturally, the samples were removed from the furnace at the
end of the growth period, i.e. at 1, 4 and 10 h. Growth starts
3
Nanotechnology 19 (2008) 455608
P Hiralal et al
Figure 2. XRD patterns, SEM images and size distribution histograms showing the effect of temperature on growth. An increase in density
and size of nanowires is observed with increasing temperature.
at an early stage. After 1 h, an array of triangular needles
can be observed, between 0.5 and 1 μm in length, presumably
nucleated on surface defects. As time progresses, the needles
appear to evolve to form thin, long wires, ∼5–15 μm in length
and a few nm in width, often keeping a needle-shaped base.
The FWHM of the (1 1 0) peak decreases from 0.236◦ at 1 h to
0.197◦ at 10 h, indicating either an increase in crystallite size
or an improvement of crystallinity.
3.4. Growth and patterning of NWs on different substrates
So far, we have demonstrated that hematite wires can be grown
from iron foils at temperatures as low as ∼300 ◦ C. However,
the applications of this are fairly limited unless the wires can
be grown directly on other functional substrates in a controlled
fashion. In fact, by depositing a thin film of iron by rf
sputtering, and then oxidizing that film, it was possible to
grow wires on a range of substrates such as glass, Si, SiO2
4
Nanotechnology 19 (2008) 455608
P Hiralal et al
Fe
(a) 15%
Fe2O3
(1 1 0)
500 nm
(b) 30%
500 nm
a.u.
(c) 45%
500 nm
(d) 60%
Figure 4. SEM surface micrographs of NWs grown for different
times. Inset shows side view. Scale bar is 2 μm. The samples were
grown at 530 ◦ C under 1% O2 .
500 nm
300 ◦ C influence its conductivity. For more tolerant substrates
such as Si, differences in thermal expansion coefficients lead
to stresses which can crack the film. Figure 5(a) shows growth
on an SiO2 substrate at 325 ◦ C in air. At this low temperature,
growth was carried out for 24 h, due to the slower oxidation
rate. The process can be implemented on a simple hotplate in
air as well as in a furnace.
It is desirable to pattern the sputtered iron film not only
to give growth on selected locations but also to reduce thermal
stresses in the film for higher temperature growth. Figure 5(b)
shows a pattern of lines where wires have been grown on ITO.
Patterning was done by using a standard photolithographic
process. In brief, AZ5214 photoresist was spin-coated and
baked. After exposure to the desired pattern, the film was
developed and then the iron film sputtered. After sputtering,
lift-off was conducted in acetone, leaving an iron film with
the desired pattern. This film was subsequently oxidized to
produce the iron oxide nanowires.
The possibility of growth on different substrates opens
up possibilities for different applications such as photoelectrochemical cells, field emission devices, etc. The generalization
(e) 100%
500 nm
20
30
40
50
60
70
2θ
Figure 3. XRD patterns and SEM images showing the effect of %O2
concentration on nanowire growth.
and indium–tin oxide (ITO). It was found that a minimum iron
film thickness of around 400 nm is required for the oxidation
to result in nanowire growth. On these substrates, growth
at as low as possible a temperature is desirable for several
reasons. On ITO, for instance, temperatures much above
5
Nanotechnology 19 (2008) 455608
P Hiralal et al
Figure 5. SEM micrographs of growth from thin films sputtered on (a) SiO2 and (b) patterned on ITO.
of this process for different oxides also opens the possibility
of integration into a CMOS process as gold is not involved.
Hematite is a rarely used semiconductor due to poor carrier
mobility. However, it should be noted that mobility of hematite
is strongly anisotropic and has been reported to be up to four
orders of magnitude larger in the (1 1 0) crystal direction than
in others [18].
4. Discussion and conclusions
Results from a parametric study conducted on the thermal
oxidation of iron to produce hematite nanowires and
nanoneedles are reported. Variation in wire crystallinity and
morphology were studied as a function of growth conditions.
The roles of temperature, O2 partial pressure and growth time
were clarified. The possibility of starting with sputtered iron
and growing hematite nanowires on a wide range of substrates
which do not tolerate high temperatures was demonstrated, for
example, growth on ITO at as low as 325 ◦ C. The potential
of this simple and general process for the growth of oxide
nanowires and the degree of control that can be achieved is
shown.
The mechanism for growth cannot be vapor–liquid–solid
(VLS) or vapor–solid (VS) as the source of the iron is the
substrate itself. Iron has a melting point of 1538 ◦ C [19], so
it is unlikely that the iron melts to form the wire. It is believed
that the tip growth occurs by iron diffusion to the tip, possibly
via surface diffusion, and reaction with the O2 from the gas
phase. Some evidence for this is present in the growth at
618 ◦ C, where bulbs can be observed on the tips of the wires
(see figure 6). The occurrence of these bulbs is more obvious
the higher the growth temperature. A possible explanation for
the bulb-shaped tip can be obtained from evidence obtained by
Wang et al [20], who observed a higher Fe to O ratio in these
bulbs than the rest of the wire. This may be a mark of the end of
growth due to unbalanced mixing. Oxidation rate is expected
to decrease during cooling, resulting in a less oxidized part at
the active growth point. More so, this would be strong evidence
for tip growth.
Regarding the mechanism for the iron to reach the tip,
we hereby propose a model similar to the quasi-liquid layer
on the surface of ice, where even at temperatures well below
Figure 6. Schematic of the mechanism of growth, where Fe is
supplied from the substrate and O2 from the surroundings. Insets
show rhombohedral tip and amorphous edge of the wires, which
provide some hints on the growth mechanism.
0 ◦ C, a thin surface layer of water is always present [21]. In
curved surfaces, the energy of dissociation may be lower than
that in the bulk, allowing for more rapid diffusion. Increasing
temperature would hence increase the mobility of the surface
layer and allow for faster growth. This hypothesis is further
supported by the fact that many wires are wider at the base than
the tip. However, the bicrystalline nature of the wires and the
observation that most form at the tip of small needle-shaped
blades opens further questions on the growth mechanism,
which seems to be general for many metal oxides.
The thermal oxidation mechanism of metals has been
studied a few times in the last 40–50 years. It is known
that, under typical conditions of NW growth a layered oxide
scale is usually formed on the surface of iron, consisting of
hematite as an outer layer, magnetite as an intermediate layer
and wustite as a layer in direct contact with the iron substrate.
A model proposed by Gulbransen and coworkers [22] assumes
that an internal diffusion of iron atoms and/or ions along axial
screw dislocations or twins combined with surface diffusion in
grain boundaries of magnetite are responsible for the growth.
Although the issue of the mode of transportation of material to
6
Nanotechnology 19 (2008) 455608
P Hiralal et al
the growing site is important, a bigger insight into the growth
mechanism may be provided by clarifying the nature of the
sites of growth (the positions at which the wires grow) and that
of the growing sites (the action occurring at the tip from where
the wire grows). Unfortunately no data is available concerning
these, so the growth mechanism is speculative. More recent
studies [15–17] have recapped this method of growth for 1D
crystals, proposed for functional structures in nanotechnology.
This structure may serve as a template for other nanowires,
like Fe3 O4 for novel magnetic devices. Wires of width <10 nm
provide ideal candidates for probing 1D quantum phenomena
and further studies are in progress.
[6] Cui Y, Wei Q, Park H and Lieber C M 2001 Science 293 1289
[7] Cornell R M and Schwertmann U 2003 The Iron Oxides:
Structure, Properties, Reactions, Occurrences and Uses
(New York: Wiley–VCH)
[8] Gratzel M 2001 Nature 414 338
[9] Siroky K, Jiresova J and Hudec L 1994 Thin Solid Films
245 211
[10] Jang J S, Kim H G, Reddy V R, Bae S W, Ji S M and Lee J S
2005 J. Catal. 231 213
[11] Tepper T, Ilievski F, Ross C A, Zaman T R, Ram R J, Yung S Y
and Stadler B 2003 J. Appl. Phys. 93 6948
[12] Lindgren T, Wang H, Beermann N, Vayssieres L,
Hagfeldt A and Lindquist S E 2002 Sol. Energy Mater. Sol.
Cells 71 231
[13] Givargizov E I 1987 Highly Anisotropic Crystals
(Dordrecht: Reidel)
[14] Pfefferkorn G 1953 Naturwissenschaften 40 551
[15] Ren S et al 2007 Mater. Lett. 61 666
[16] Wen X et al 2005 J. Phys. Chem. B 109 215
Fu Y Y, Wang R M, Xu J, Chen J, Yan Y, Narlikar A V and
Zhang H 2003 Chem. Phys. Lett. 379 373
[17] Jiang X et al 2002 Nano Lett. 2 1333
[18] Kay A et al 2006 J. Am. Chem. Soc. 128 15714
[19] Lide D R 1995 Handbook of Chemistry and Physics 76th edn
(Boca Raton, FL: CRC Press)
[20] Wang R et al 2005 J. Phys. Chem. B 109 12245
[21] Lied A, Dosch H and Bilgram J H 1994 Phys. Rev. Lett.
72 3554
[22] Tallman R L and Gulbransen E A 1967 J. Electrochem. Soc.
114 1227
Tallman R L and Gulbrandsen E A 1968 Nature 218 1046
Acknowledgment
The authors would like to acknowledge Varun Gupta at Rutgers
University for Raman spectroscopy measurements.
References
[1] Fan H J, Werner P and Zacharias M 2006 Small 2 700
Li Y, Qian F, Xiang J and Lieber C M 2006 Mater. Today 9 18
[2] Okada M et al 2006 Japan. J. Appl. Phys. 45 5624
[3] Law M et al 2005 Nat. Mater. 4 455
[4] Chen J, Xu L, Li W and Gou X 2005 Adv. Mater. 17 582
[5] Bao J et al 2006 Nano Lett. 6 1719
7