Physica E 38 (2007) 40–43
Growth of silicon nanowires on UV-structurable glass
using self-organized nucleation centres
K. Tonisch
a,Ã
, F. Weise
a
, M. Stubenrauch
a
, V. Cimalla
a
, G. Ecke
a
, F. Will
a
, H. Romanus
a
,
S. Mrotzek
a
, H. Hofmeister
b
, M. Hoffmann
a
,D.Hu
¨
lsenberg
a
, O. Ambacher
a
a

serve as nucleation centres. A succeeding heating step leads
to the formation of crystallized phases within the glass
matrix, whereby the shape and density of the crystals is
controlled by a self-organized process depending on the
process parameters. As we describe in this communication,
the growth of nanowires can be stimulated on the surface
of such partially crystallized glasses without adding a metal
catalyst.
Various methods have been reported for contacting
nanowires for electrical measurements. In the most
common approach, the nanowire is removed from its
growth substrate and placed on an additional sample by
suspension in solvents and spin-on deposition [1]. Metal
contacts can be defined by optical lithography [2], electron
beam lithography [1] or by focussed ion beam (FIB) [3].We
present the growth of nanowires without additional
catalyst and their chemical and electrical characterization.
For the latter, the nanowires were contacted by direct
writing platinum leads using a FIB equipment.
2. Experimental
The p hotostructurable glass fro m the LiO
2
–Al
2
O
3
–SiO
2
system wa s pretreated by a n exposure t o UV light and a
crystallization ste p at 570 1C. The surface of th e glass was

3
CH
3
C
5
H
4
Pt] into
predefined bridging leads.
3. Results
The substrate was found to be covered homogeneously
by a dense net of nanowires; ‘‘sea urchin’’-like bundles
grow only where large inhomogeneities occurred at the
surface (Fig. 1). The diameter of the nanowires varied from
50 to 300 nm, their length reached 10–20 mm. The photo-
structurable glass was exposed to UV light with an energy
flux of 60 J/cm
2
and tempered at 570 1C for several hours
prior to the nanowire growth, which leads to the formation
of Ag-clusters which init ialize the growth of lithium meta
silicate (LMS) crystals within the glass [4]. As a result of
this pretreatment, the glass partially crystallizes into
dendrite like LMS crystals within the glass matrix. The
Ag clusters and LMS crystals were analysed using
transmission electron microscopy (Fig. 2). Whereas the
Ag cluster itself is too small to initiate the vapour–liquid–
solid growth (VLS-growth) of such large nanowires, the
LMS crystals exhibit a size which seems to correspond to
the diameter of the nanowires. Since the eutectic tempera-

4.9 at% for silicon and oxygen, respectively. Though the
growth process leads to a pure silicon nanowire, a thin
layer of natural silicon oxide is formed soon after the
removal of the substrate from the growth chamber. No
ARTICLE IN PRESS
Fig. 1. Scanning electron microscopy (SEM) images of homogeneously grown nanowires (right) and sea urchin-like bundles (left).
K. Tonisch et al. / Physica E 38 (2007) 40–43 41
trace of the phosphor doping could be found due to the
measurement range of EDX analysis.
Fig. 4 shows a representative SEM image of a FIB-Pt
connected nanowire in a two-probe geometry. The nominal
Pt thickness was 500 nm, and the width of the leads was
4 mm. The FIB-deposition conditions must be caref ully
chosen in order to avoid damage or even destruction of
delicate structures by bombardment with massive Ga
+
ions. Therefore, the direct scanning of the nanowires with
the ion beam prior to the contact writing was reduced as
much as possible. We found no evidence for such damage.
On the contrary, a light sputtering at the contact areas on
the nanowires might even improve the contact resistance by
removing the native oxide. The conductivity of the
nanowires was calculated from their electrical resistance
which was determined by current–voltage measurements.
The linear I–V curves of several nanowires with a diameter
of 200–300 nm is displayed in Fig. 5. The different slope is
mainly caused by the different wire geometry resulting in a
different value of the electrical resistance.
Though the thickness of the SiO
2

4. Discussion and conclusions
Silicon nanowires were grown successfully on photo-
structurable glass without the need of an additional
catalyst and were analysed with regard to their chemical
composition and their conductivity. Though the LMS
crystals are a prominent candidate for initializing the
nanowire growth, an extended study of the first stages of
the growth process is necessary in order to gain a valid
model. Further research will also focus on a more detailed
characterization of the nanowires concerning their elec-
trical and structural properties. The contamination of the
substrate surface with carbon and gallium is expected to be
high due to low background pressure of the FIB system
and the necessity to use the Ga
+
-beam for visualizing,
which might lead to an incorporation of gallium. This
might result in an uncertainty of the conductivity
measurements due to possible leakage currents across the
substrate surface. Though the agreement of the specific
resistance of the nanowires with that of polycrystalline
silicon supports the correctness of the electrical measure-
ments, additional analysis with Auger electron spectro-
scopy or other surface-sensitive measurements are
necessary. Another way to eliminate the uncertainty of
the conductivity measurements is the undercutting of the
nanowires. Additionally, a freestanding structure will
enable a mechanical characterization concerning the
resonant frequency of such nanowire-based NEMS and
Young’s modulus of the nanowire itself.