pringer-Verlag 2003
ABSTRACT
Photoconducting properties of In
2
O
3
nanowires were studied. Devices
based on individual In
2
O
3
nanowires showed a substantial increase in conductance of
up to four orders of magnitude upon exposure to UV light. Such devices also exhibited
short response times and signicant shifts in the threshold gate voltage. The sensitivity
to UV of different wavelengths was studied and compared. We have further demon-
strated the use of UV light as a gas cleanser for In
2
O
3
nanowire chemical sensors,
leading to a recovery time as short as 80 s.
PACS
78.67
1
Introduction
One-dimensional nanostruc-
tures, such as nanowires and nano-
tubes, are the most promising materi-
als for solid-state sensors due to their
ultrahigh surface-to-volume ratios [1,
2]. In addition to chemical and bio-
logical sensing properties, their pho-
todetection, or so-called optical switch-
ing properties [3], have also attracted
much attention because an optical gat-
ing can work as an alternative to elec-
trical gating for nanowire devices used
in memory storage and logic circuits,
where the performance critically re-
lies on binary switching. Moreover, it
has been found that the photosensi-
tive characteristics of nanowires can
greatly enhance the performance of
some nanowire gas sensors by improv-
ing their sensitivities, as well as short-
ening the recovery time [4]. Light-
enhanced gas sensing properties have
been reported for SnO
2
nanoribbon sen-
sors [4] and SnO
2
/In
2
O
3
thin-lm sen-
sors [5]. In sharp contrast, the photode-
tection and light-enhanced gas sens-
ing properties of In
2
O
3
nanowires have
u Fax: +1-213/740-8677, E-mail: chongwuz@usc.edu
been rarely reported, presumably due to
a lack of high-quality In
2
O
3
nanowires
in the past. Based on our recent suc-
cess regarding the diameter-controlled
synthesis of single-crystalline In
2
O
3
nanowires [6], we have studied the
UV response of devices consisting of
individual nanowires. These devices
showed sensitivities as high as 10
4
to
UV light illumination, accompanied by
substantial shifts in the gate threshold
voltage. Differential response to UV
illumination at different wavelengths
has also been observed. We have fur-
ther demonstrated the use of UV light
as a gas cleanser for In
2
O
3
nanowire
chemical sensors after exposure to di-
luted NO
2
gas, leading to a recovery
time as short as 80 s.
2
Experimental
In
2
O
3
nanowires were syn-
thesized via the vaporliquidsolid
mechanism, and details of the process
can be found in our previous publica-
tion [6]. After the synthesis, scanning
electron microscopy (SEM) and trans-
mission electron microscopy (TEM)
examination revealed that these nano-
wires had well-controlled diameters of
around 10 nm and lengths exceeding
3
µm.Ahigh-resolutionTEM(HRTEM)
image is shown in Fig. 1a. The lat-
tice spacing along the [110] growth
direction
(0.72 nm) is in good agree-
ment with the lattice constant for In
2
O
3
(1.01 nm). The as-synthesized nano-
wires were deposited from a suspension
in isopropyl alcohol onto a degenerately
doped silicon wafer covered with 500-
nm SiO
2
. Photolithography and Ti
/Au
deposition were performed to pattern
the drain and source electrodes to con-
tact both ends of the individual wires.
The Si substrate was used as a back
gate in our electronic measurements.
Figure 1b shows the SEM image of
a device fabricated in this way, where
an In
2
O
3
nanowire bridging the source
and the drain electrodes can be clearly
seen. A double-wavelength (365 nm and
254 nm
) UV lamp with a power dens-
FIGURE 1
a HRTEM image of an In
2
O
3
nanowire showing the [110] growth direction. The
lattice spacing is consistent with the lattice con-
stant for In
2
O
3
, which is 10
.1 Å. b SEM image of
the nanowire device used in our measurements 164
Applied Physics A Materials Science & Processing
ity of approximately 3 mW
/cm
2
was
xed approximately 1 cm away from
the sample. The photoresponse of our
devices was carried out under practi-
cal conditions the devices were kept
in air, at room temperature and under
indoor incandescent light during the
measurements.
3
Results and discussion
As shown in Fig. 2a, the
currentvoltage (IV ) curve of the
nanowire device taken before the expo-
sure to UV light indicates a relatively
high resistance, where a differential
conductance of only 3
.6 × 10
1
nS
is
obtained for V
= 0 V. The asymme-
try in the IV curve is a result of the
local gating effect, as reported previ-
ously [7]. Upon exposure to UV of
365-nm wavelength, the device conduc-
tance increased immediately and sta-
bilized after tens of seconds, leading
to an IV curve with a linear conduc-
tance around 1
.7 × 10
3
nS
, shown in
Fig. 2a. The device became even more
conductive after the device was ex-
posed to UV of 254-nm wavelength,
where a linear IV curve with a con-
-1000
-500
0
500
1000
I(
n
A
)
-1.0
-0.5
0.0
0.5
1.0
V (V)
before UV
illumination
= 365 nm
= 254 nm
400
300
200
100
0
I(
n
A
)
-40
-30
-20
-10
0
10
V
g
(V)
= 254
nm
15
10
5
0
I(
n
A
)
-10
0
10
V
g
(V)
= 365 nm
before UV
a
b
FIGURE 2
IV curves (a) and IV
g
curves (b) recorded before UV light illu-
mination and after exposure to UV light
at wavelengths of 365 nm and 254 nm, re-
spectively
ductance of 5
.0 × 10
3
nS
was observed,
indicating a sensitivity of up to 10
4
.
We have also measured the current vs.
gate voltage
(IV
g
) characteristics of
the device, with the sourcedrain bias
xed at 0
.32 V, under various condi-
tions, shown in Fig. 2b. In addition to
the apparent change in the conduc-
tance, both the threshold gate volt-
age and the slope of the IV
g
curves
were signicantly modulated by the
UV light. The threshold gate voltage
shifted from
+5 V for the initial unex-
posed state to
2 V and 25 V after
exposure to 365-nm and 254-nm UV
light, respectively. The change in the
carrier concentrations can be estimated
from the formula
n = CV
t
/L, where
C
and L represent the capacitance
and length of the nanowire and
V
t
is the shift in the threshold gate volt-
age. The capacitance can be estimated
to be C
= 8.16 × 10
5
pF
for a 2-
µm-
long nanowire [6]. It is thus found that
the carrier concentration was enhanced
by 1
.79 nm
1
and 7
.65 nm
1
for the
365-nm and 254-nm UV light, cor-
responding to threshold-voltage shifts
of
7 V and 30 V, respectively. The
carrier mobility of the nanowire can
be estimated from the IV
g
charac-
teristics with the relation d I
/dV
g
=
µ(C/L
2
)V. The different slopes of the
three IV
g
curves in Fig. 2b give mobil-
ity values of 0
.85 cm
2
/V s, 6.1 cm
2
/V s
and 68 cm
2
/V s, for the device without
UV, with 365-nm UV and 254-nm UV,
respectively. The above-mentioned an-
alysis reveals an evident enhancement
of both the carrier concentration and
the mobility upon UV illumination, thus
leading to the pronounced sensitivities
we have observed.
To further explore the photoresponse
properties, time-domain measurements
were also carried out to evaluate the
response speed of the nanowire. Fig-
ure 3a shows the current measured
over time, where the 254-nm UV light
was switched on and off repeatedly
and a total of ve cycles of data were
recorded. We kept a constant drain
source voltage of 0
.3 V and zero gate
bias during the measurement. The con-
ductance of the device was observed
to increase rapidly upon UV exposure
and decrease gradually once the UV was
turned off. The device exhibited good
reversibility between the high- and the
low-conductivity states, as little drift
was observed for both the on state and
the off state after repeated exposure.
The response time, which is dened
as the time duration for a conductance
change by one order of magnitude, was
obtained to be approximately 10 s from
this gure. Figure 3b shows the time-
domain response of the device after it
was exposed to 365-nm and 254-nm
UV light sequentially. While 365-nm
UV light brought the current to 33 nA,
signicantly more conduction
(290 nA)
was observed upon exposure to 254-nm
UV light. This indicates that good se-
lectivity between UV light of certain
wavelengths can be readily achieved
with our devices.
In
2
O
3
nanowires are known to have
a direct bandgap of 3
.75 eV at k = 0 and
an indirect energy gap of 2
.62 V [8].
This explains the selectivity of our
nanowire to those two different wave-
lengths we used. For the 254-nm UV
light, each photon has an energy of
4
.9 eV, sufciently large to excite elec-
trons directly from the valence band of
the In
2
O
3
nanowires to the conduction
band at k
= 0. In contrast, for the UV
light with a peak wavelength at 365 nm,
the majority of the photons have ener- ZHANG
et al. Ultraviolet photodetection properties of indium oxide nanowires
165
Time (s)
Time (s)
FIGURE 3
a Reversible switching of
the nanowire between low- and high-
conductivity states when the 254-nm UV
light was turned on and off. b Photore-
sponse of the In
2
O
3
nanowire to sequen-
tial UV illumination at wavelengths of
365 nm and 254 nm
gies below the In
2
O
3
direct bandgap and
hence the direct transition is prohibited.
Though the indirect-forbidden transi-
tions still exist with assisting phonons
( 0.07 eV), their contribution to the
conduction is limited due to the very
small probability of simultaneous ab-
sorption of a photon and phonon [8].
However, there also exist some pho-
tons with higher energies due to the
nite spectrum width, which can gener-
ate electronhole pai